xiaoxue_admin
/
dbx
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
dbx/crates/dbx-core/src/db/postgres.rs

7548 lines
318 KiB
Rust
Raw Blame History

use chrono::{DateTime, Local, NaiveDate, NaiveDateTime, NaiveTime};
use deadpool_postgres::{ManagerConfig, Pool, PoolError, RecyclingMethod, Runtime};
use futures::{SinkExt, StreamExt};
use percent_encoding::percent_decode_str;
use rust_decimal::Decimal;
use rustls::client::danger::{HandshakeSignatureValid, ServerCertVerified, ServerCertVerifier};
use rustls::client::verify_server_cert_signed_by_trust_anchor;
use rustls::crypto::{verify_tls12_signature, verify_tls13_signature, CryptoProvider};
use rustls::pki_types::{CertificateDer, PrivateKeyDer, ServerName, UnixTime};
use rustls::server::ParsedCertificate;
use sqlparser::ast::Statement;
use sqlparser::dialect::PostgreSqlDialect;
use sqlparser::parser::Parser;
use std::collections::{BTreeSet, HashMap};
use std::fs::File;
use std::future::Future;
use std::io::BufReader;
use std::pin::Pin;
use std::str::FromStr;
use std::sync::atomic::AtomicBool;
use std::sync::{Arc, Mutex, OnceLock, Weak};
use std::time::{Duration, Instant};
use tokio::task::JoinHandle;
use tokio_postgres::config::SslMode;
use tokio_postgres::tls::{MakeTlsConnect, TlsConnect};
use tokio_postgres::types::{FromSql, Kind, Type};
use tokio_postgres::{AsyncMessage, NoTls, Row, SimpleQueryMessage, Socket};
use tokio_util::sync::CancellationToken;
use super::file_validator::validate_file_path;
use crate::query::{await_stream_with_progress_timeout, DbOperationBudget, StreamProgressClock};
use crate::sql::starts_with_executable_sql_keyword;
use crate::types::{
ColumnInfo, CompletionAssistantCandidate, CompletionAssistantCandidateKind, CompletionAssistantMatchMode,
CompletionAssistantObjectKind, CompletionAssistantRequest, CompletionAssistantResponse, DatabaseInfo,
DatabaseStorageInfo, ExtensionInfo, ForeignKeyInfo, FunctionInfo, IndexInfo, ObjectInfo, ObjectStatistics,
OwnerInfo, QueryMessage, QueryResult, RuleInfo, SchemaInfo, SequenceInfo, SpatialColumnBuilder, TableInfo,
TriggerInfo,
};
pub(crate) const GAUSSDB_COMPATIBILITY_SQL: &str =
"SELECT datcompatibility FROM pg_catalog.pg_database WHERE datname = current_database()";
pub async fn gaussdb_identifier_quote(pool: &Pool) -> Option<String> {
let timeout = super::connection_timeout();
let client = checkout_postgres_client(pool, None, timeout).await.ok()?;
let row = tokio::time::timeout(timeout, client.query_opt(GAUSSDB_COMPATIBILITY_SQL, &[])).await.ok()?.ok()??;
let compatibility_mode = row.try_get::<_, String>(0).ok()?;
gaussdb_identifier_quote_for_compatibility_mode(&compatibility_mode).map(str::to_string)
}
pub(crate) fn gaussdb_identifier_quote_for_compatibility_mode(compatibility_mode: &str) -> Option<&'static str> {
match compatibility_mode.trim().to_ascii_uppercase().as_str() {
"M" | "B" | "MYSQL" => Some("`"),
"A" | "PG" | "ORA" | "POSTGRESQL" => Some("\""),
_ => None,
}
}
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct PostgresTablePrivilegeInfo {
pub grantor: String,
pub grantee: String,
pub privilege_type: String,
pub is_grantable: bool,
pub column_name: Option<String>,
}
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct PostgresTableAccessInfo {
pub owner: String,
pub owner_default_privileges: Vec<String>,
pub privileges: Vec<PostgresTablePrivilegeInfo>,
}
#[derive(Debug, Clone, Default, PartialEq, Eq)]
pub struct PostgresTablePartitionInfo {
pub is_partition: bool,
pub parent_schema: Option<String>,
pub parent_table: Option<String>,
pub bound: Option<String>,
pub key: Option<String>,
}
#[derive(Debug, Clone, Default, PartialEq, Eq)]
pub struct PostgresTablePartitionLocalObjects {
pub has_primary_key: bool,
pub foreign_keys: BTreeSet<String>,
pub indexes: BTreeSet<String>,
}
fn pg_temporal_to_json_value(row: &Row, idx: usize) -> Option<serde_json::Value> {
if let Ok(v) = row.try_get::<_, DateTime<Local>>(idx) {
return Some(serde_json::Value::String(format_pg_timestamptz(v)));
}
if let Ok(v) = row.try_get::<_, NaiveDateTime>(idx) {
return Some(serde_json::Value::String(v.to_string()));
}
if let Ok(v) = row.try_get::<_, NaiveDate>(idx) {
return Some(serde_json::Value::String(v.to_string()));
}
if let Ok(v) = row.try_get::<_, NaiveTime>(idx) {
return Some(serde_json::Value::String(v.to_string()));
}
None
}
struct PgSystemU32(u32);
impl<'a> FromSql<'a> for PgSystemU32 {
fn from_sql(_: &Type, raw: &'a [u8]) -> Result<Self, Box<dyn std::error::Error + Sync + Send>> {
let bytes: [u8; 4] = raw.try_into().map_err(|_| "expected 4 bytes for PostgreSQL system u32")?;
Ok(Self(u32::from_be_bytes(bytes)))
}
fn accepts(ty: &Type) -> bool {
matches!(*ty, Type::XID | Type::CID)
}
}
/// A `FromSql` adapter that accepts any PostgreSQL type and reads its raw
/// bytes as a UTF-8 string. This is used as a last-resort fallback to handle
/// custom types (enums, domains, etc.) that tokio_postgres cannot map to
/// built-in Rust types in the binary protocol.
struct PgAnyString(String);
impl<'a> FromSql<'a> for PgAnyString {
fn from_sql(_: &Type, raw: &'a [u8]) -> Result<Self, Box<dyn std::error::Error + Sync + Send>> {
std::str::from_utf8(raw)
.map(|s| PgAnyString(s.to_string()))
.map_err(|e| Box::new(e) as Box<dyn std::error::Error + Sync + Send>)
}
fn accepts(_: &Type) -> bool {
true
}
}
/// A `FromSql` adapter that accepts any PostgreSQL type and returns the raw
/// bytes unchanged. Used to decode custom types like pgvector whose binary
/// format we handle ourselves.
struct PgRawBytes(Vec<u8>);
impl<'a> FromSql<'a> for PgRawBytes {
fn from_sql(_: &Type, raw: &'a [u8]) -> Result<Self, Box<dyn std::error::Error + Sync + Send>> {
Ok(PgRawBytes(raw.to_vec()))
}
fn accepts(_: &Type) -> bool {
true
}
}
#[derive(Clone, Copy, Debug, PartialEq)]
struct PgPoint {
x: f64,
y: f64,
}
impl<'a> FromSql<'a> for PgPoint {
fn from_sql(_: &Type, raw: &'a [u8]) -> Result<Self, Box<dyn std::error::Error + Sync + Send>> {
decode_pg_point_bytes(raw).ok_or_else(|| "expected 16 bytes for PostgreSQL point".into())
}
fn accepts(ty: &Type) -> bool {
*ty == Type::POINT
}
}
fn decode_pg_point_bytes(raw: &[u8]) -> Option<PgPoint> {
let raw: [u8; 16] = raw.try_into().ok()?;
Some(PgPoint {
x: f64::from_be_bytes(raw[0..8].try_into().ok()?),
y: f64::from_be_bytes(raw[8..16].try_into().ok()?),
})
}
fn format_pg_float(value: f64) -> String {
if value == f64::INFINITY {
"Infinity".to_string()
} else if value == f64::NEG_INFINITY {
"-Infinity".to_string()
} else {
value.to_string()
}
}
fn format_pg_point(point: PgPoint) -> String {
format!("({},{})", format_pg_float(point.x), format_pg_float(point.y))
}
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
struct PgInterval {
microseconds: i64,
days: i32,
months: i32,
}
impl<'a> FromSql<'a> for PgInterval {
fn from_sql(_: &Type, raw: &'a [u8]) -> Result<Self, Box<dyn std::error::Error + Sync + Send>> {
decode_pg_interval_bytes(raw).ok_or_else(|| "expected 16 bytes for PostgreSQL interval".into())
}
fn accepts(ty: &Type) -> bool {
*ty == Type::INTERVAL
}
}
fn decode_pg_interval_bytes(raw: &[u8]) -> Option<PgInterval> {
let raw: [u8; 16] = raw.try_into().ok()?;
Some(PgInterval {
microseconds: i64::from_be_bytes(raw[0..8].try_into().ok()?),
days: i32::from_be_bytes(raw[8..12].try_into().ok()?),
months: i32::from_be_bytes(raw[12..16].try_into().ok()?),
})
}
fn push_pg_interval_component(parts: &mut Vec<String>, value: i64, singular: &str, plural: &str) {
if value == 0 {
return;
}
let unit = if value.abs() == 1 { singular } else { plural };
parts.push(format!("{value} {unit}"));
}
fn format_pg_interval_time(microseconds: i64) -> String {
let signed_microseconds = i128::from(microseconds);
let sign = if signed_microseconds < 0 { "-" } else { "" };
let absolute_microseconds = signed_microseconds.abs();
let hours = absolute_microseconds / 3_600_000_000;
let minutes = absolute_microseconds / 60_000_000 % 60;
let seconds = absolute_microseconds / 1_000_000 % 60;
let fraction = absolute_microseconds % 1_000_000;
let mut formatted = format!("{sign}{hours:02}:{minutes:02}:{seconds:02}");
if fraction != 0 {
let fraction = format!("{fraction:06}");
formatted.push('.');
formatted.push_str(fraction.trim_end_matches('0'));
}
formatted
}
fn format_pg_interval(interval: PgInterval) -> String {
let total_months = i64::from(interval.months);
let years = total_months / 12;
let months = total_months % 12;
let mut parts = Vec::with_capacity(4);
push_pg_interval_component(&mut parts, years, "year", "years");
push_pg_interval_component(&mut parts, months, "mon", "mons");
push_pg_interval_component(&mut parts, i64::from(interval.days), "day", "days");
parts.push(format_pg_interval_time(interval.microseconds));
parts.join(" ")
}
struct PgDateRange(String);
impl<'a> FromSql<'a> for PgDateRange {
fn from_sql(_: &Type, raw: &'a [u8]) -> Result<Self, Box<dyn std::error::Error + Sync + Send>> {
decode_pg_daterange_bytes(raw).map(Self).ok_or_else(|| "invalid PostgreSQL daterange binary value".into())
}
fn accepts(ty: &Type) -> bool {
*ty == Type::DATE_RANGE
}
}
const PG_RANGE_EMPTY: u8 = 0b0000_0001;
const PG_RANGE_LOWER_INCLUSIVE: u8 = 0b0000_0010;
const PG_RANGE_UPPER_INCLUSIVE: u8 = 0b0000_0100;
const PG_RANGE_LOWER_UNBOUNDED: u8 = 0b0000_1000;
const PG_RANGE_UPPER_UNBOUNDED: u8 = 0b0001_0000;
const PG_RANGE_KNOWN_FLAGS: u8 = PG_RANGE_EMPTY
| PG_RANGE_LOWER_INCLUSIVE
| PG_RANGE_UPPER_INCLUSIVE
| PG_RANGE_LOWER_UNBOUNDED
| PG_RANGE_UPPER_UNBOUNDED;
fn decode_pg_date_bound(raw: &[u8], cursor: &mut usize, unbounded: bool) -> Option<String> {
if unbounded {
return Some(String::new());
}
let length = read_i32_be(raw, cursor)?;
if length != 4 {
return None;
}
let days = read_i32_be(raw, cursor)?;
match days {
i32::MIN => Some("-infinity".to_string()),
i32::MAX => Some("infinity".to_string()),
days => NaiveDate::from_ymd_opt(2000, 1, 1)?
.checked_add_signed(chrono::Duration::days(i64::from(days)))
.map(|date| date.to_string()),
}
}
fn decode_pg_daterange_bytes(raw: &[u8]) -> Option<String> {
let (&flags, _) = raw.split_first()?;
if flags & !PG_RANGE_KNOWN_FLAGS != 0 {
return None;
}
if flags == PG_RANGE_EMPTY {
return (raw.len() == 1).then(|| "empty".to_string());
}
if flags & PG_RANGE_EMPTY != 0 {
return None;
}
let lower_unbounded = flags & PG_RANGE_LOWER_UNBOUNDED != 0;
let upper_unbounded = flags & PG_RANGE_UPPER_UNBOUNDED != 0;
if (lower_unbounded && flags & PG_RANGE_LOWER_INCLUSIVE != 0)
|| (upper_unbounded && flags & PG_RANGE_UPPER_INCLUSIVE != 0)
{
return None;
}
let mut cursor = 1;
let lower = decode_pg_date_bound(raw, &mut cursor, lower_unbounded)?;
let upper = decode_pg_date_bound(raw, &mut cursor, upper_unbounded)?;
if cursor != raw.len() {
return None;
}
let lower_delimiter = if flags & PG_RANGE_LOWER_INCLUSIVE != 0 { '[' } else { '(' };
let upper_delimiter = if flags & PG_RANGE_UPPER_INCLUSIVE != 0 { ']' } else { ')' };
Some(format!("{lower_delimiter}{lower},{upper}{upper_delimiter}"))
}
/// Decode pgvector binary format into a Vec<f32>.
///
/// pgvector binary layout (big-endian):
/// - 2 bytes: dimensions (uint16)
/// - 2 bytes: unused (padding)
/// - N*4 bytes: IEEE 754 f32 values
fn decode_pgvector_bytes(raw: &[u8]) -> Option<Vec<f32>> {
if raw.len() < 4 {
return None;
}
let dims = u16::from_be_bytes([raw[0], raw[1]]) as usize;
let expected_len = 4 + dims * 4;
if raw.len() != expected_len {
return None;
}
let floats: Vec<f32> =
raw[4..].chunks_exact(4).map(|chunk| f32::from_be_bytes([chunk[0], chunk[1], chunk[2], chunk[3]])).collect();
Some(floats)
}
fn pg_u32_number(v: u32) -> serde_json::Value {
serde_json::Value::Number(serde_json::Number::from(v))
}
fn pg_system_u32_to_json(row: &Row, idx: usize) -> Option<serde_json::Value> {
if let Ok(v) = row.try_get::<_, u32>(idx) {
return Some(pg_u32_number(v));
}
row.try_get::<_, PgSystemU32>(idx).ok().map(|v| pg_u32_number(v.0))
}
fn pg_optional_array_to_json<T>(
values: Vec<Option<T>>,
map_value: impl Fn(T) -> serde_json::Value,
) -> serde_json::Value {
serde_json::Value::Array(
values.into_iter().map(|value| value.map(&map_value).unwrap_or(serde_json::Value::Null)).collect(),
)
}
fn pg_json_array_values_to_json(values: Vec<Option<serde_json::Value>>) -> serde_json::Value {
pg_optional_array_to_json(values, |value| serde_json::Value::String(value.to_string()))
}
fn pg_float_number(v: f64) -> serde_json::Value {
serde_json::Number::from_f64(v).map(serde_json::Value::Number).unwrap_or(serde_json::Value::Null)
}
fn decode_pg_network_address_bytes(raw: &[u8], force_cidr_output: bool) -> Option<String> {
let family = *raw.first()?;
let bits = *raw.get(1)?;
let is_cidr = *raw.get(2)? != 0;
let addr_len = *raw.get(3)? as usize;
let addr = raw.get(4..)?;
if addr.len() != addr_len {
return None;
}
let (address, host_bits) = match (family, addr_len) {
(2, 4) => {
let bytes: [u8; 4] = addr.try_into().ok()?;
(std::net::IpAddr::V4(std::net::Ipv4Addr::from(bytes)).to_string(), 32)
}
(3, 16) => {
let bytes: [u8; 16] = addr.try_into().ok()?;
(std::net::IpAddr::V6(std::net::Ipv6Addr::from(bytes)).to_string(), 128)
}
_ => return None,
};
if bits > host_bits {
return None;
}
if force_cidr_output || is_cidr || bits != host_bits {
Some(format!("{address}/{bits}"))
} else {
Some(address)
}
}
fn decode_pg_macaddr_bytes(raw: &[u8]) -> Option<String> {
if !matches!(raw.len(), 6 | 8) {
return None;
}
Some(raw.iter().map(|byte| format!("{byte:02x}")).collect::<Vec<_>>().join(":"))
}
fn decode_pg_bit_string_bytes(raw: &[u8]) -> Option<String> {
let mut cursor = 0;
let bit_len = read_i32_be(raw, &mut cursor)?;
if bit_len < 0 {
return None;
}
let bit_len = bit_len as usize;
let data = raw.get(cursor..)?;
if data.len() != bit_len.div_ceil(8) {
return None;
}
let mut bits = String::with_capacity(bit_len);
for index in 0..bit_len {
let byte = data[index / 8];
let bit = (byte >> (7 - (index % 8))) & 1;
bits.push(if bit == 1 { '1' } else { '0' });
}
Some(bits)
}
fn pg_network_address_to_json_value(row: &Row, idx: usize, force_cidr_output: bool) -> Option<serde_json::Value> {
row.try_get::<_, PgRawBytes>(idx)
.ok()
.and_then(|raw| decode_pg_network_address_bytes(&raw.0, force_cidr_output))
.map(serde_json::Value::String)
}
fn pg_macaddr_to_json_value(row: &Row, idx: usize) -> Option<serde_json::Value> {
row.try_get::<_, PgRawBytes>(idx)
.ok()
.and_then(|raw| decode_pg_macaddr_bytes(&raw.0))
.map(serde_json::Value::String)
}
fn pg_bit_string_to_json_value(row: &Row, idx: usize) -> Option<serde_json::Value> {
row.try_get::<_, PgRawBytes>(idx)
.ok()
.and_then(|raw| decode_pg_bit_string_bytes(&raw.0))
.map(serde_json::Value::String)
}
fn pg_network_address_array_to_json_value(row: &Row, idx: usize, force_cidr_output: bool) -> Option<serde_json::Value> {
row.try_get::<_, Vec<Option<PgRawBytes>>>(idx).ok().map(|values| {
pg_optional_array_to_json(values, |raw| {
decode_pg_network_address_bytes(&raw.0, force_cidr_output)
.map(serde_json::Value::String)
.unwrap_or_else(|| super::binary_value_to_json(&raw.0))
})
})
}
fn pg_macaddr_array_to_json_value(row: &Row, idx: usize) -> Option<serde_json::Value> {
row.try_get::<_, Vec<Option<PgRawBytes>>>(idx).ok().map(|values| {
pg_optional_array_to_json(values, |raw| {
decode_pg_macaddr_bytes(&raw.0)
.map(serde_json::Value::String)
.unwrap_or_else(|| super::binary_value_to_json(&raw.0))
})
})
}
fn pg_bit_string_array_to_json_value(row: &Row, idx: usize) -> Option<serde_json::Value> {
row.try_get::<_, Vec<Option<PgRawBytes>>>(idx).ok().map(|values| {
pg_optional_array_to_json(values, |raw| {
decode_pg_bit_string_bytes(&raw.0)
.map(serde_json::Value::String)
.unwrap_or_else(|| super::binary_value_to_json(&raw.0))
})
})
}
fn pg_array_to_json_value(row: &Row, idx: usize) -> Option<serde_json::Value> {
if let Ok(values) = row.try_get::<_, Vec<Option<serde_json::Value>>>(idx) {
return Some(pg_json_array_values_to_json(values));
}
if let Ok(values) = row.try_get::<_, Vec<Option<String>>>(idx) {
return Some(pg_optional_array_to_json(values, serde_json::Value::String));
}
if let Ok(values) = row.try_get::<_, Vec<Option<bool>>>(idx) {
return Some(pg_optional_array_to_json(values, serde_json::Value::Bool));
}
if let Ok(values) = row.try_get::<_, Vec<Option<Decimal>>>(idx) {
return Some(pg_optional_array_to_json(values, |v| serde_json::Value::String(v.to_string())));
}
if let Ok(values) = row.try_get::<_, Vec<Option<uuid::Uuid>>>(idx) {
return Some(pg_optional_array_to_json(values, |v| serde_json::Value::String(v.to_string())));
}
if let Ok(values) = row.try_get::<_, Vec<Option<DateTime<Local>>>>(idx) {
return Some(pg_optional_array_to_json(values, |v| serde_json::Value::String(format_pg_timestamptz(v))));
}
if let Ok(values) = row.try_get::<_, Vec<Option<NaiveDateTime>>>(idx) {
return Some(pg_optional_array_to_json(values, |v| serde_json::Value::String(v.to_string())));
}
if let Ok(values) = row.try_get::<_, Vec<Option<NaiveDate>>>(idx) {
return Some(pg_optional_array_to_json(values, |v| serde_json::Value::String(v.to_string())));
}
if let Ok(values) = row.try_get::<_, Vec<Option<NaiveTime>>>(idx) {
return Some(pg_optional_array_to_json(values, |v| serde_json::Value::String(v.to_string())));
}
if let Ok(values) = row.try_get::<_, Vec<Option<u32>>>(idx) {
return Some(pg_optional_array_to_json(values, pg_u32_number));
}
if let Ok(values) = row.try_get::<_, Vec<Option<i8>>>(idx) {
return Some(pg_optional_array_to_json(values, |v| serde_json::Value::Number(v.into())));
}
if let Ok(values) = row.try_get::<_, Vec<Option<i16>>>(idx) {
return Some(pg_optional_array_to_json(values, |v| serde_json::Value::Number(v.into())));
}
if let Ok(values) = row.try_get::<_, Vec<Option<i32>>>(idx) {
return Some(pg_optional_array_to_json(values, |v| serde_json::Value::Number(v.into())));
}
if let Ok(values) = row.try_get::<_, Vec<Option<i64>>>(idx) {
return Some(pg_optional_array_to_json(values, super::safe_i64_to_json));
}
if let Ok(values) = row.try_get::<_, Vec<Option<f32>>>(idx) {
return Some(pg_optional_array_to_json(values, |v| pg_float_number(v as f64)));
}
if let Ok(values) = row.try_get::<_, Vec<Option<f64>>>(idx) {
return Some(pg_optional_array_to_json(values, pg_float_number));
}
if let Ok(values) = row.try_get::<_, Vec<Option<PgAnyString>>>(idx) {
return Some(pg_optional_array_to_json(values, |v| serde_json::Value::String(v.0)));
}
None
}
fn format_pg_timestamptz(value: DateTime<Local>) -> String {
value.to_rfc3339()
}
/// 时间类型解码失败后的回退目标，与原 if 链中时间分支之后的匹配顺序一一对应。
#[derive(Clone, Copy, PartialEq, Eq, Debug)]
pub(crate) enum PgTemporalFallback {
/// 数组类型名（下划线开头）落到通用数组解码。
GenericArray,
/// `VECTOR(...)` 形式的类型名落到 pgvector 解码。
Vector,
/// 其余落到通用试探链。
Probe,
}
/// 每列一次的类型分类结果，避免在逐单元格路径上重复 `to_uppercase` 与字符串比较链。
#[derive(Clone, Copy, PartialEq, Eq, Debug)]
pub(crate) enum PgColType {
Bytea,
Json,
Bool,
Point,
Interval,
DateRange,
Temporal { fallback: PgTemporalFallback },
Numeric,
Uuid,
Inet { cidr: bool },
MacAddr,
BitString,
TsVector,
SystemU32,
InetArray { cidr: bool },
MacAddrArray,
BitStringArray,
GenericArray,
Vector,
Geometry,
Other,
}
fn pg_scalar_type_requires_text_protocol(oid: u32, col_type: PgColType) -> bool {
Type::from_oid(oid).is_none() && !matches!(col_type, PgColType::Vector | PgColType::Geometry)
}
fn pg_type_requires_text_protocol(pg_type: &Type, col_type: PgColType) -> bool {
if pg_type.oid() == Type::RECORD.oid() || pg_type.oid() == Type::RECORD_ARRAY.oid() {
return true;
}
match pg_type.kind() {
Kind::Enum(_) => false,
Kind::Array(element_type) => Type::from_oid(element_type.oid()).is_none(),
Kind::Simple => pg_scalar_type_requires_text_protocol(pg_type.oid(), col_type),
_ => Type::from_oid(pg_type.oid()).is_none(),
}
}
pub(crate) fn classify_pg_type(type_name: &str) -> PgColType {
let upper = type_name.to_uppercase();
if upper == "BYTEA" {
return PgColType::Bytea;
}
if upper == "JSON" || upper == "JSONB" {
return PgColType::Json;
}
if upper == "BOOL" {
return PgColType::Bool;
}
if upper == "POINT" {
return PgColType::Point;
}
if upper == "INTERVAL" {
return PgColType::Interval;
}
if upper == "DATERANGE" {
return PgColType::DateRange;
}
if upper.contains("TIMESTAMP")
|| upper == "DATE"
|| upper == "TIME"
|| upper == "TIMETZ"
|| upper.contains("INTERVAL")
{
let fallback = if upper.starts_with('_') {
PgTemporalFallback::GenericArray
} else if upper.starts_with("VECTOR(") {
PgTemporalFallback::Vector
} else {
PgTemporalFallback::Probe
};
return PgColType::Temporal { fallback };
}
if upper == "NUMERIC" || upper == "DECIMAL" || upper == "MONEY" {
return PgColType::Numeric;
}
if upper == "UUID" {
return PgColType::Uuid;
}
if matches!(upper.as_str(), "INET" | "CIDR") {
return PgColType::Inet { cidr: upper == "CIDR" };
}
if matches!(upper.as_str(), "MACADDR" | "MACADDR8") {
return PgColType::MacAddr;
}
if matches!(upper.as_str(), "BIT" | "VARBIT") {
return PgColType::BitString;
}
if upper == "TSVECTOR" {
return PgColType::TsVector;
}
if matches!(upper.as_str(), "OID" | "XID" | "CID") {
return PgColType::SystemU32;
}
if matches!(upper.as_str(), "_INET" | "_CIDR") {
return PgColType::InetArray { cidr: upper == "_CIDR" };
}
if matches!(upper.as_str(), "_MACADDR" | "_MACADDR8") {
return PgColType::MacAddrArray;
}
if matches!(upper.as_str(), "_BIT" | "_VARBIT") {
return PgColType::BitStringArray;
}
if upper.starts_with('_') {
return PgColType::GenericArray;
}
if upper == "VECTOR" || upper.starts_with("VECTOR(") {
return PgColType::Vector;
}
if upper == "GEOMETRY" || upper == "GEOGRAPHY" {
return PgColType::Geometry;
}
PgColType::Other
}
pub(crate) fn classify_pg_column_types(column_types: &[String]) -> Vec<PgColType> {
column_types.iter().map(|type_name| classify_pg_type(type_name)).collect()
}
pub(crate) fn pg_value_to_json_classified(row: &Row, idx: usize, col_type: PgColType) -> serde_json::Value {
match col_type {
PgColType::Bytea => row
.try_get::<_, Vec<u8>>(idx)
.map(|bytes| super::binary_value_to_json(&bytes))
.unwrap_or(serde_json::Value::Null),
PgColType::Json => {
if let Ok(v) = row.try_get::<_, serde_json::Value>(idx) {
return serde_json::Value::String(v.to_string());
}
if let Ok(v) = row.try_get::<_, String>(idx) {
return serde_json::Value::String(v);
}
serde_json::Value::Null
}
PgColType::Bool => pg_bool_value_to_json(row, idx),
PgColType::Point => row
.try_get::<_, PgPoint>(idx)
.map(|point| serde_json::Value::String(format_pg_point(point)))
.unwrap_or(serde_json::Value::Null),
PgColType::Interval => row
.try_get::<_, PgInterval>(idx)
.map(|interval| serde_json::Value::String(format_pg_interval(interval)))
.unwrap_or_else(|_| pg_fallback_value_to_json(row, idx)),
PgColType::DateRange => row
.try_get::<_, PgDateRange>(idx)
.map(|range| serde_json::Value::String(range.0))
.unwrap_or_else(|_| pg_fallback_value_to_json(row, idx)),
PgColType::Temporal { fallback } => {
if let Some(v) = pg_temporal_to_json_value(row, idx) {
return v;
}
match fallback {
PgTemporalFallback::GenericArray => pg_array_to_json_value(row, idx).unwrap_or(serde_json::Value::Null),
PgTemporalFallback::Vector => pg_vector_value_to_json(row, idx),
PgTemporalFallback::Probe => pg_fallback_value_to_json(row, idx),
}
}
PgColType::Numeric => row
.try_get::<_, Decimal>(idx)
.map(|v: Decimal| serde_json::Value::String(v.to_string()))
.unwrap_or(serde_json::Value::Null),
PgColType::Uuid => row
.try_get::<_, uuid::Uuid>(idx)
.map(|v| serde_json::Value::String(v.to_string()))
.unwrap_or(serde_json::Value::Null),
PgColType::Inet { cidr } => pg_network_address_to_json_value(row, idx, cidr).unwrap_or(serde_json::Value::Null),
PgColType::MacAddr => pg_macaddr_to_json_value(row, idx).unwrap_or(serde_json::Value::Null),
PgColType::BitString => pg_bit_string_to_json_value(row, idx).unwrap_or(serde_json::Value::Null),
PgColType::TsVector => row
.try_get::<_, PgRawBytes>(idx)
.ok()
.and_then(|raw| decode_tsvector_bytes(&raw.0))
.map(serde_json::Value::String)
.unwrap_or(serde_json::Value::Null),
PgColType::SystemU32 => pg_system_u32_to_json(row, idx).unwrap_or(serde_json::Value::Null),
PgColType::InetArray { cidr } => {
pg_network_address_array_to_json_value(row, idx, cidr).unwrap_or(serde_json::Value::Null)
}
PgColType::MacAddrArray => pg_macaddr_array_to_json_value(row, idx).unwrap_or(serde_json::Value::Null),
PgColType::BitStringArray => pg_bit_string_array_to_json_value(row, idx).unwrap_or(serde_json::Value::Null),
PgColType::GenericArray => pg_array_to_json_value(row, idx).unwrap_or(serde_json::Value::Null),
PgColType::Vector => pg_vector_value_to_json(row, idx),
PgColType::Geometry => {
if let Ok(PgRawBytes(raw)) = row.try_get::<_, PgRawBytes>(idx) {
return super::wkb::wkb_to_wkt(&raw)
.map(serde_json::Value::String)
.unwrap_or_else(|| super::binary_value_to_json(&raw));
}
serde_json::Value::Null
}
PgColType::Other => pg_fallback_value_to_json(row, idx),
}
}
fn pg_value_to_json_with_srid(row: &Row, idx: usize, col_type: PgColType) -> (serde_json::Value, Option<u32>) {
if col_type != PgColType::Geometry {
return (pg_value_to_json_classified(row, idx, col_type), None);
}
if let Ok(PgRawBytes(raw)) = row.try_get::<_, PgRawBytes>(idx) {
return match super::wkb::decode_wkb_geometry(&raw) {
Some(geometry) => (serde_json::Value::String(geometry.wkt), geometry.srid),
None => (super::binary_value_to_json(&raw), None),
};
}
(serde_json::Value::Null, None)
}
fn pg_text_fallback_value(value: &str, col_type: Option<PgColType>) -> (serde_json::Value, Option<u32>) {
let (value, srid, _) = pg_text_fallback_value_with_spatial(value, col_type);
(value, srid)
}
fn pg_text_fallback_value_with_spatial(
value: &str,
col_type: Option<PgColType>,
) -> (serde_json::Value, Option<u32>, bool) {
match col_type {
Some(PgColType::Geometry) => decode_pg_text_wkb(value)
.map(|geometry| (serde_json::Value::String(geometry.wkt), geometry.srid, true))
.or_else(|| split_pg_ewkt(value, false).map(|(value, srid)| (value, srid, true)))
.unwrap_or_else(|| (serde_json::Value::String(value.to_string()), None, true)),
Some(_) => (serde_json::Value::String(value.to_string()), None, false),
None => decode_pg_text_wkb(value)
.map(|geometry| (serde_json::Value::String(geometry.wkt), geometry.srid, true))
.or_else(|| split_pg_ewkt(value, true).map(|(value, srid)| (value, srid, true)))
.unwrap_or_else(|| (serde_json::Value::String(value.to_string()), None, false)),
}
}
fn decode_pg_text_wkb(value: &str) -> Option<super::wkb::DecodedGeometry> {
let trimmed = value.trim();
let hex = trimmed
.strip_prefix("0x")
.or_else(|| trimmed.strip_prefix("0X"))
.or_else(|| trimmed.strip_prefix("\\x"))
.or_else(|| trimmed.strip_prefix("\\X"))
.unwrap_or(trimmed);
if hex.len() < 10 || !hex.len().is_multiple_of(2) || !matches!(&hex[..2], "00" | "01") || !hex.is_ascii() {
return None;
}
let bytes = hex
.as_bytes()
.chunks_exact(2)
.map(|pair| {
let text = std::str::from_utf8(pair).ok()?;
u8::from_str_radix(text, 16).ok()
})
.collect::<Option<Vec<_>>>()?;
super::wkb::decode_wkb_geometry(&bytes)
}
fn split_pg_ewkt(value: &str, require_recognizable_wkt: bool) -> Option<(serde_json::Value, Option<u32>)> {
let rest = value.strip_prefix("SRID=")?;
let (srid, wkt) = rest.split_once(';')?;
let srid = srid.parse::<i64>().ok().and_then(|value| u32::try_from(value).ok())?;
if require_recognizable_wkt && !is_recognizable_wkt(wkt) {
return None;
}
Some((serde_json::Value::String(wkt.to_string()), (srid != 0).then_some(srid)))
}
fn is_recognizable_wkt(value: &str) -> bool {
let trimmed = value.trim_start();
const TYPES: [&str; 7] =
["POINT", "LINESTRING", "POLYGON", "MULTIPOINT", "MULTILINESTRING", "MULTIPOLYGON", "GEOMETRYCOLLECTION"];
TYPES.iter().any(|geometry_type| {
trimmed
.get(..geometry_type.len())
.filter(|prefix| prefix.eq_ignore_ascii_case(geometry_type))
.and_then(|_| trimmed.as_bytes().get(geometry_type.len()))
.is_some_and(|next| next.is_ascii_whitespace() || *next == b'(')
})
}
/// Serialize a pgvector `vector` component with f32 shortest round-trip decimal text.
///
/// Casting through `f64` (or fixed fractional rounding) either expands binary noise or
/// truncates remaining single-precision digits; formatting via `f32` display keeps the
/// full float4 value that pgvector stores.
fn pg_vector_element_number(v: f32) -> serde_json::Value {
v.to_string().parse().map(serde_json::Value::Number).unwrap_or(serde_json::Value::Null)
}
fn pg_vector_value_to_json(row: &Row, idx: usize) -> serde_json::Value {
if let Ok(PgRawBytes(raw)) = row.try_get::<_, PgRawBytes>(idx) {
if let Some(floats) = decode_pgvector_bytes(&raw) {
return serde_json::Value::Array(floats.into_iter().map(pg_vector_element_number).collect());
}
}
serde_json::Value::Null
}
fn pg_fallback_value_to_json(row: &Row, idx: usize) -> serde_json::Value {
row.try_get::<_, String>(idx)
.map(serde_json::Value::String)
.or_else(|e| pg_system_u32_to_json(row, idx).ok_or(e))
.or_else(|_| row.try_get::<_, i64>(idx).map(super::safe_i64_to_json))
.or_else(|_| row.try_get::<_, i32>(idx).map(|v| serde_json::Value::Number(v.into())))
.or_else(|_| row.try_get::<_, i16>(idx).map(|v| serde_json::Value::Number(v.into())))
.or_else(|_| row.try_get::<_, i8>(idx).map(|v| serde_json::Value::Number(v.into())))
.or_else(|e| pg_array_to_json_value(row, idx).ok_or(e))
.or_else(|_| {
row.try_get::<_, f64>(idx).map(|v| {
serde_json::Number::from_f64(v).map(serde_json::Value::Number).unwrap_or(serde_json::Value::Null)
})
})
.or_else(|_| {
row.try_get::<_, f32>(idx).map(|v| {
serde_json::Number::from_f64((v as f64 * 1_000_000.0).round() / 1_000_000.0)
.map(serde_json::Value::Number)
.unwrap_or(serde_json::Value::Null)
})
})
.or_else(|_| row.try_get::<_, bool>(idx).map(serde_json::Value::Bool))
.or_else(|_| row.try_get::<_, uuid::Uuid>(idx).map(|v| serde_json::Value::String(v.to_string())))
.or_else(|e| pg_temporal_to_json_value(row, idx).ok_or(e))
.or_else(|_| row.try_get::<_, Vec<u8>>(idx).map(|bytes| super::binary_value_to_json(&bytes)))
.or_else(|_| row.try_get::<_, PgAnyString>(idx).map(|v| serde_json::Value::String(v.0)))
.or_else(|_| row.try_get::<_, PgRawBytes>(idx).map(|v| super::binary_value_to_json(&v.0)))
.unwrap_or(serde_json::Value::Null)
}
fn decode_tsvector_bytes(raw: &[u8]) -> Option<String> {
let mut cursor = 0;
let count = read_i32_be(raw, &mut cursor)?;
if count < 0 {
return None;
}
let mut entries = Vec::with_capacity(count as usize);
for _ in 0..count {
let start = cursor;
while cursor < raw.len() && raw[cursor] != 0 {
cursor += 1;
}
if cursor >= raw.len() {
return None;
}
let lexeme = std::str::from_utf8(&raw[start..cursor]).ok()?;
cursor += 1;
let position_count = read_u16_be(raw, &mut cursor)? as usize;
let mut positions = Vec::with_capacity(position_count);
for _ in 0..position_count {
let encoded = read_u16_be(raw, &mut cursor)?;
let position = encoded & 0x3fff;
let weight = match encoded >> 14 {
3 => "A",
2 => "B",
1 => "C",
_ => "",
};
positions.push(format!("{position}{weight}"));
}
let mut entry = format!("'{}'", escape_tsvector_lexeme(lexeme));
if !positions.is_empty() {
entry.push(':');
entry.push_str(&positions.join(","));
}
entries.push(entry);
}
if cursor == raw.len() {
Some(entries.join(" "))
} else {
None
}
}
fn read_i32_be(raw: &[u8], cursor: &mut usize) -> Option<i32> {
let bytes: [u8; 4] = raw.get(*cursor..*cursor + 4)?.try_into().ok()?;
*cursor += 4;
Some(i32::from_be_bytes(bytes))
}
fn read_u16_be(raw: &[u8], cursor: &mut usize) -> Option<u16> {
let bytes: [u8; 2] = raw.get(*cursor..*cursor + 2)?.try_into().ok()?;
*cursor += 2;
Some(u16::from_be_bytes(bytes))
}
fn escape_tsvector_lexeme(value: &str) -> String {
value.replace('\\', "\\\\").replace('\'', "''")
}
fn pg_error_to_string(err: tokio_postgres::Error) -> String {
err.as_db_error().map(ToString::to_string).unwrap_or_else(|| err.to_string())
}
fn pg_db_error_to_string(err: &tokio_postgres::error::DbError) -> String {
format!("{err} (SQLSTATE {})", err.code().code())
}
fn pg_error_from_sources(err: &(dyn std::error::Error + 'static)) -> Option<String> {
let mut current = Some(err);
while let Some(source) = current {
if let Some(pg_error) = source.downcast_ref::<tokio_postgres::Error>() {
if let Some(db_error) = pg_error.as_db_error() {
return Some(pg_db_error_to_string(db_error));
}
}
if let Some(db_error) = source.downcast_ref::<tokio_postgres::error::DbError>() {
return Some(pg_db_error_to_string(db_error));
}
current = source.source();
}
None
}
fn error_with_sources_to_string(err: &(dyn std::error::Error + 'static)) -> String {
let mut messages = vec![err.to_string()];
let mut current = err.source();
while let Some(source) = current {
let message = source.to_string();
if !messages.iter().any(|existing| existing == &message) {
messages.push(message);
}
current = source.source();
}
messages.join(": ")
}
fn pg_pool_error_to_string(err: PoolError) -> String {
pg_error_from_sources(&err).unwrap_or_else(|| error_with_sources_to_string(&err))
}
fn should_retry_postgres_text_query(err: &tokio_postgres::Error) -> bool {
let message = err.as_db_error().map(ToString::to_string).unwrap_or_else(|| err.to_string()).to_ascii_lowercase();
should_retry_postgres_text_query_message(&message)
}
fn should_retry_postgres_text_query_message(message: &str) -> bool {
message.contains("no binary output function")
|| message.contains("no binary send function")
|| message.contains("cannot display a value of type")
}
fn should_retry_postgres_stale_cache(err: &tokio_postgres::Error) -> bool {
if let Some(db_error) = err.as_db_error() {
return should_retry_postgres_stale_cache_fields(
Some(db_error.code().code()),
db_error.routine(),
db_error.message(),
);
}
should_retry_postgres_stale_cache_fields(None, None, &err.to_string())
}
fn should_retry_postgres_stale_cache_fields(sqlstate: Option<&str>, routine: Option<&str>, message: &str) -> bool {
let structured_match = sqlstate == Some("0A000")
&& routine.is_some_and(|routine| routine.eq_ignore_ascii_case("RevalidateCachedQuery"));
structured_match || message.to_ascii_lowercase().contains("cached plan must not change result type")
}
async fn postgres_query_cached(
client: &deadpool_postgres::Client,
sql: &str,
params: &[&(dyn tokio_postgres::types::ToSql + Sync)],
) -> Result<Vec<Row>, tokio_postgres::Error> {
let stmt = client.prepare_cached(sql).await?;
match client.query(&stmt, params).await {
Ok(rows) => Ok(rows),
Err(err) if should_retry_postgres_stale_cache(&err) => {
// Metadata queries can be cached while a table/view definition is
// changed from another session. Evict and retry once with fresh
// statement/type metadata instead of surfacing PostgreSQL's stale
// cached-plan error to the UI.
log::warn!("[postgres][metadata:stale_cache] evicting cached statement: {}", pg_error_to_string(err));
client.statement_cache.remove(sql, &[]);
client.clear_type_cache();
let stmt = client.prepare_cached(sql).await?;
client.query(&stmt, params).await
}
Err(err) => Err(err),
}
}
async fn postgres_query_one_cached(
client: &deadpool_postgres::Client,
sql: &str,
params: &[&(dyn tokio_postgres::types::ToSql + Sync)],
) -> Result<Row, tokio_postgres::Error> {
let stmt = client.prepare_cached(sql).await?;
match client.query_one(&stmt, params).await {
Ok(row) => Ok(row),
Err(err) if should_retry_postgres_stale_cache(&err) => {
// Same stale-cache protection as postgres_query_cached, for scalar
// catalog probes such as pg_proc feature detection.
log::warn!("[postgres][metadata:stale_cache] evicting cached statement: {}", pg_error_to_string(err));
client.statement_cache.remove(sql, &[]);
client.clear_type_cache();
let stmt = client.prepare_cached(sql).await?;
client.query_one(&stmt, params).await
}
Err(err) => Err(err),
}
}
enum PreparedSelectOutcome {
Complete(Box<QueryResult>),
TextFallback { column_types: Vec<String>, unsupported_type: String },
}
struct PreparedSelectMetadata {
columns: Vec<String>,
column_types: Vec<String>,
column_classes: Vec<PgColType>,
unsupported_type: Option<String>,
}
fn prepared_select_metadata(stmt: &tokio_postgres::Statement) -> PreparedSelectMetadata {
let columns: Vec<String> = stmt.columns().iter().map(|c| c.name().to_string()).collect();
let column_types: Vec<String> = stmt.columns().iter().map(|c| c.type_().name().to_string()).collect();
let column_classes = classify_pg_column_types(&column_types);
let unsupported_type = stmt.columns().iter().zip(&column_classes).find_map(|(column, col_type)| {
let pg_type = column.type_();
pg_type_requires_text_protocol(pg_type, *col_type).then(|| pg_type.name().to_string())
});
PreparedSelectMetadata { columns, column_types, column_classes, unsupported_type }
}
async fn prepare_select_with_metadata(
client: &deadpool_postgres::Client,
sql: &str,
) -> Result<(tokio_postgres::Statement, PreparedSelectMetadata), tokio_postgres::Error> {
let mut stmt = client.prepare_cached(sql).await?;
let mut metadata = prepared_select_metadata(&stmt);
if metadata.unsupported_type.is_some() {
stmt = client.prepare(sql).await?;
metadata = prepared_select_metadata(&stmt);
}
Ok((stmt, metadata))
}
async fn execute_select_prepared(
client: &deadpool_postgres::Client,
sql: &str,
start: Instant,
row_limit: usize,
progress_clock: Option<&StreamProgressClock>,
) -> Result<PreparedSelectOutcome, tokio_postgres::Error> {
let prepared_start = Instant::now();
let (stmt, metadata) = prepare_select_with_metadata(client, sql).await?;
if let Some(progress_clock) = progress_clock {
progress_clock.mark();
}
log::info!(
"[postgres][select:prepare_cached:done] elapsed_ms={} total_ms={}",
prepared_start.elapsed().as_millis(),
start.elapsed().as_millis()
);
let PreparedSelectMetadata { columns, column_types, column_classes, unsupported_type } = metadata;
if let Some(unsupported_type) = unsupported_type {
return Ok(PreparedSelectOutcome::TextFallback { column_types, unsupported_type });
}
let params: Vec<&(dyn tokio_postgres::types::ToSql + Sync)> = Vec::new();
let query_start = Instant::now();
let stream = client.query_raw(&stmt, params).await?;
if let Some(progress_clock) = progress_clock {
progress_clock.mark();
}
log::info!(
"[postgres][select:query_raw:done] elapsed_ms={} total_ms={} column_count={}",
query_start.elapsed().as_millis(),
start.elapsed().as_millis(),
columns.len()
);
tokio::pin!(stream);
let mut result_rows: Vec<Vec<serde_json::Value>> = Vec::new();
let mut spatial_values: Vec<Vec<Option<u32>>> = Vec::new();
let mut spatial_columns = SpatialColumnBuilder::new(
column_classes
.iter()
.enumerate()
.filter_map(|(index, col_type)| (*col_type == PgColType::Geometry).then_some(index)),
);
let mut truncated = false;
let rows_start = Instant::now();
while let Some(row_result) = stream.next().await {
if let Some(progress_clock) = progress_clock {
progress_clock.mark();
}
if result_rows.len() >= row_limit {
truncated = true;
break;
}
let row = row_result?;
let mut values = Vec::with_capacity(row.columns().len());
let mut row_srids = vec![None; row.columns().len()];
for (i, row_srid) in row_srids.iter_mut().enumerate() {
let col_type = column_classes.get(i).copied().unwrap_or(PgColType::Other);
let (value, srid) = pg_value_to_json_with_srid(&row, i, col_type);
if col_type == PgColType::Geometry {
spatial_columns.observe(i, srid);
*row_srid = srid;
}
values.push(value);
}
result_rows.push(values);
spatial_values.push(row_srids);
}
log::info!(
"[postgres][select:rows:done] elapsed_ms={} total_ms={} row_count={} truncated={}",
rows_start.elapsed().as_millis(),
start.elapsed().as_millis(),
result_rows.len(),
truncated
);
let (spatial_columns, spatial_values) = spatial_columns.finish_with_values(spatial_values);
Ok(PreparedSelectOutcome::Complete(Box::new(QueryResult {
columns,
column_types,
column_sortables: Vec::new(),
spatial_columns,
spatial_values,
rows: result_rows,
affected_rows: 0,
execution_time_ms: start.elapsed().as_millis(),
truncated,
session_id: None,
has_more: false,
elasticsearch_raw_body: None,
messages: Vec::new(),
})))
}
fn matching_pg_text_column_types(columns: &[String], prepared: Option<Vec<String>>) -> Vec<String> {
prepared.filter(|types| types.len() == columns.len()).unwrap_or_default()
}
async fn execute_select_text(
client: &deadpool_postgres::Client,
sql: &str,
start: Instant,
row_limit: usize,
prepared_column_types: Option<Vec<String>>,
progress_clock: Option<&StreamProgressClock>,
) -> Result<QueryResult, String> {
let stream = client.simple_query_raw(sql).await.map_err(pg_error_to_string)?;
if let Some(progress_clock) = progress_clock {
progress_clock.mark();
}
tokio::pin!(stream);
let mut columns: Vec<String> = Vec::new();
let mut result_rows: Vec<Vec<serde_json::Value>> = Vec::new();
let mut spatial_values: Vec<Vec<Option<u32>>> = Vec::new();
let column_classes =
prepared_column_types.as_ref().map(|types| classify_pg_column_types(types)).unwrap_or_default();
let mut spatial_columns = SpatialColumnBuilder::new(
column_classes
.iter()
.enumerate()
.filter_map(|(index, col_type)| (*col_type == PgColType::Geometry).then_some(index)),
);
let mut truncated = false;
while let Some(message) = stream.next().await {
if let Some(progress_clock) = progress_clock {
progress_clock.mark();
}
match message {
Ok(SimpleQueryMessage::RowDescription(cols)) => {
columns = cols.iter().map(|c| c.name().to_string()).collect();
}
Ok(SimpleQueryMessage::Row(row)) => {
if columns.is_empty() {
columns = row.columns().iter().map(|c| c.name().to_string()).collect();
}
if result_rows.len() >= row_limit {
truncated = true;
break;
}
let mut values = Vec::with_capacity(row.len());
let mut row_srids = vec![None; row.len()];
for (i, row_srid) in row_srids.iter_mut().enumerate() {
match row.try_get(i).map_err(pg_error_to_string)? {
Some(value) => {
let (decoded, srid, is_spatial) =
pg_text_fallback_value_with_spatial(value, column_classes.get(i).copied());
values.push(decoded);
if is_spatial {
spatial_columns.observe(i, srid);
*row_srid = srid;
}
}
None => {
values.push(serde_json::Value::Null);
}
}
}
result_rows.push(values);
spatial_values.push(row_srids);
}
Err(_) if result_rows.len() >= row_limit => {
truncated = true;
break;
}
Err(err) => return Err(pg_error_to_string(err)),
Ok(SimpleQueryMessage::CommandComplete(_)) => {}
Ok(_) => {}
}
}
let (spatial_columns, spatial_values) = spatial_columns.finish_with_values(spatial_values);
Ok(QueryResult {
column_types: matching_pg_text_column_types(&columns, prepared_column_types),
columns,
column_sortables: Vec::new(),
spatial_columns,
spatial_values,
rows: result_rows,
affected_rows: 0,
execution_time_ms: start.elapsed().as_millis(),
truncated,
session_id: None,
has_more: false,
elasticsearch_raw_body: None,
messages: Vec::new(),
})
}
async fn finish_prepared_select(
client: &deadpool_postgres::Client,
sql: &str,
start: Instant,
row_limit: usize,
outcome: PreparedSelectOutcome,
progress_clock: Option<&StreamProgressClock>,
) -> Result<QueryResult, String> {
match outcome {
PreparedSelectOutcome::Complete(result) => Ok(*result),
PreparedSelectOutcome::TextFallback { column_types, unsupported_type } => {
log::info!(
"[postgres][select:text_fallback] unsupported_type={} switching_to=simple_query",
unsupported_type
);
execute_select_text(client, sql, start, row_limit, Some(column_types), progress_clock).await
}
}
}
pub(crate) async fn execute_select_query(
client: &deadpool_postgres::Client,
sql: &str,
start: Instant,
row_limit: usize,
) -> Result<QueryResult, String> {
execute_select_query_with_progress(client, sql, start, row_limit, None).await
}
async fn execute_select_query_with_progress(
client: &deadpool_postgres::Client,
sql: &str,
start: Instant,
row_limit: usize,
progress_clock: Option<&StreamProgressClock>,
) -> Result<QueryResult, String> {
match execute_select_prepared(client, sql, start, row_limit, progress_clock).await {
Ok(outcome) => finish_prepared_select(client, sql, start, row_limit, outcome, progress_clock).await,
Err(err) if should_retry_postgres_stale_cache(&err) => {
// The cached prepared statement is stale (e.g. the view or table
// schema changed since the statement was prepared). Evict the
// stale entry and retry with a fresh server-side prepare.
log::warn!("[postgres][select:stale_cache] evicting cached statement: {}", pg_error_to_string(err));
client.statement_cache.remove(sql, &[]);
match execute_select_prepared(client, sql, start, row_limit, progress_clock).await {
Ok(outcome) => finish_prepared_select(client, sql, start, row_limit, outcome, progress_clock).await,
Err(err) if should_retry_postgres_text_query(&err) => {
execute_select_text(client, sql, start, row_limit, None, progress_clock).await
}
Err(err) => Err(pg_error_to_string(err)),
}
}
Err(err) if should_retry_postgres_text_query(&err) => {
execute_select_text(client, sql, start, row_limit, None, progress_clock).await
}
Err(err) => Err(pg_error_to_string(err)),
}
}
pub enum PostgresQueryStreamItem {
Columns { columns: Vec<String>, column_types: Vec<String> },
Row(Vec<serde_json::Value>),
}
enum PostgresQueryStreamError {
Postgres { err: tokio_postgres::Error, emitted: bool },
TextFallback { column_types: Vec<String>, unsupported_type: String },
Export(String),
}
impl PostgresQueryStreamError {
fn into_string(self) -> String {
match self {
Self::Postgres { err, .. } => pg_error_to_string(err),
Self::TextFallback { unsupported_type, .. } => {
format!("PostgreSQL type {unsupported_type} requires text protocol")
}
Self::Export(err) => err,
}
}
}
async fn stream_select_query_prepared(
client: &deadpool_postgres::Client,
sql: &str,
row_limit: Option<usize>,
on_item: &mut impl FnMut(PostgresQueryStreamItem) -> Result<(), String>,
) -> Result<u64, PostgresQueryStreamError> {
let (stmt, metadata) = prepare_select_with_metadata(client, sql)
.await
.map_err(|err| PostgresQueryStreamError::Postgres { err, emitted: false })?;
let PreparedSelectMetadata { columns, column_types, column_classes, unsupported_type } = metadata;
if let Some(unsupported_type) = unsupported_type {
return Err(PostgresQueryStreamError::TextFallback { column_types, unsupported_type });
}
let params: Vec<&(dyn tokio_postgres::types::ToSql + Sync)> = Vec::new();
let stream = client
.query_raw(&stmt, params)
.await
.map_err(|err| PostgresQueryStreamError::Postgres { err, emitted: false })?;
tokio::pin!(stream);
let mut rows_streamed = 0_u64;
let mut columns_emitted = false;
while let Some(row_result) = stream.next().await {
if row_limit.is_some_and(|limit| rows_streamed as usize >= limit) {
break;
}
let row = row_result
.map_err(|err| PostgresQueryStreamError::Postgres { err, emitted: columns_emitted || rows_streamed > 0 })?;
if !columns_emitted {
on_item(PostgresQueryStreamItem::Columns { columns: columns.clone(), column_types: column_types.clone() })
.map_err(PostgresQueryStreamError::Export)?;
columns_emitted = true;
}
let values = (0..row.columns().len())
.map(|i| pg_value_to_json_classified(&row, i, column_classes.get(i).copied().unwrap_or(PgColType::Other)))
.collect();
on_item(PostgresQueryStreamItem::Row(values)).map_err(PostgresQueryStreamError::Export)?;
rows_streamed += 1;
}
if !columns_emitted {
on_item(PostgresQueryStreamItem::Columns { columns, column_types })
.map_err(PostgresQueryStreamError::Export)?;
}
Ok(rows_streamed)
}
async fn stream_select_query_text(
client: &deadpool_postgres::Client,
sql: &str,
row_limit: Option<usize>,
prepared_column_types: Option<Vec<String>>,
on_item: &mut impl FnMut(PostgresQueryStreamItem) -> Result<(), String>,
) -> Result<u64, String> {
let stream = client.simple_query_raw(sql).await.map_err(pg_error_to_string)?;
tokio::pin!(stream);
let mut columns: Vec<String> = Vec::new();
let column_classes =
prepared_column_types.as_ref().map(|types| classify_pg_column_types(types)).unwrap_or_default();
let mut rows_streamed = 0_u64;
while let Some(message) = stream.next().await {
match message.map_err(pg_error_to_string)? {
SimpleQueryMessage::RowDescription(cols) => {
columns = cols.iter().map(|c| c.name().to_string()).collect();
let column_types = matching_pg_text_column_types(&columns, prepared_column_types.clone());
on_item(PostgresQueryStreamItem::Columns { columns: columns.clone(), column_types })?;
}
SimpleQueryMessage::Row(row) => {
if row_limit.is_some_and(|limit| rows_streamed as usize >= limit) {
break;
}
if columns.is_empty() {
columns = row.columns().iter().map(|c| c.name().to_string()).collect();
let column_types = matching_pg_text_column_types(&columns, prepared_column_types.clone());
on_item(PostgresQueryStreamItem::Columns { columns: columns.clone(), column_types })?;
}
let mut values = Vec::with_capacity(row.len());
for i in 0..row.len() {
values.push(match row.try_get(i).map_err(pg_error_to_string)? {
Some(value) => pg_text_fallback_value(value, column_classes.get(i).copied()).0,
None => serde_json::Value::Null,
});
}
on_item(PostgresQueryStreamItem::Row(values))?;
rows_streamed += 1;
}
SimpleQueryMessage::CommandComplete(_) => {}
_ => {}
}
}
Ok(rows_streamed)
}
pub(crate) async fn stream_select_query_inner(
client: &deadpool_postgres::Client,
sql: &str,
row_limit: Option<usize>,
on_item: &mut impl FnMut(PostgresQueryStreamItem) -> Result<(), String>,
) -> Result<u64, String> {
match stream_select_query_prepared(client, sql, row_limit, on_item).await {
Ok(rows) => Ok(rows),
Err(PostgresQueryStreamError::TextFallback { column_types, unsupported_type }) => {
log::info!(
"[postgres][stream:text_fallback] unsupported_type={} switching_to=simple_query",
unsupported_type
);
stream_select_query_text(client, sql, row_limit, Some(column_types), on_item).await
}
Err(PostgresQueryStreamError::Postgres { err, emitted: false }) if should_retry_postgres_stale_cache(&err) => {
// The cached prepared statement can become stale after schema changes.
// Evict and retry once, matching the normal query execution path.
log::warn!("[postgres][stream:stale_cache] evicting cached statement: {}", pg_error_to_string(err));
client.statement_cache.remove(sql, &[]);
match stream_select_query_prepared(client, sql, row_limit, on_item).await {
Ok(rows) => Ok(rows),
Err(PostgresQueryStreamError::Postgres { err, emitted: false })
if should_retry_postgres_text_query(&err) =>
{
stream_select_query_text(client, sql, row_limit, None, on_item).await
}
Err(PostgresQueryStreamError::TextFallback { column_types, unsupported_type }) => {
log::info!(
"[postgres][stream:text_fallback] unsupported_type={} switching_to=simple_query",
unsupported_type
);
stream_select_query_text(client, sql, row_limit, Some(column_types), on_item).await
}
Err(err) => Err(err.into_string()),
}
}
Err(PostgresQueryStreamError::Postgres { err, emitted: false }) if should_retry_postgres_text_query(&err) => {
stream_select_query_text(client, sql, row_limit, None, on_item).await
}
Err(err) => Err(err.into_string()),
}
}
pub async fn stream_query_rows(
pool: &Pool,
sql: &str,
max_rows: Option<usize>,
cancelled: &AtomicBool,
mut on_row: impl FnMut(&[serde_json::Value]) -> Result<(), String>,
) -> Result<u64, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
match stream_query_rows_on_client(&client, sql, max_rows, cancelled, &mut on_row).await {
Ok(rows) => Ok(rows),
Err(error) if should_retry_postgres_text_query_message(&error.to_ascii_lowercase()) => {
stream_query_rows_text_on_client(&client, sql, max_rows, cancelled, None, &mut on_row).await
}
Err(error) => Err(error),
}
}
async fn stream_query_rows_on_client(
client: &deadpool_postgres::Client,
sql: &str,
max_rows: Option<usize>,
cancelled: &AtomicBool,
on_row: &mut impl FnMut(&[serde_json::Value]) -> Result<(), String>,
) -> Result<u64, String> {
let (stmt, metadata) = prepare_select_with_metadata(client, sql).await.map_err(pg_error_to_string)?;
let PreparedSelectMetadata { column_classes, unsupported_type, .. } = metadata;
if let Some(unsupported_type) = unsupported_type {
log::info!(
"[postgres][row_stream:text_fallback] unsupported_type={} switching_to=simple_query",
unsupported_type
);
return stream_query_rows_text_on_client(client, sql, max_rows, cancelled, Some(&column_classes), on_row).await;
}
let params: Vec<&(dyn tokio_postgres::types::ToSql + Sync)> = Vec::new();
let stream = client.query_raw(&stmt, params).await.map_err(pg_error_to_string)?;
tokio::pin!(stream);
let row_limit = max_rows.unwrap_or(usize::MAX);
let mut rows_exported = 0_u64;
while let Some(row_result) = stream.next().await {
if cancelled.load(std::sync::atomic::Ordering::SeqCst) {
return Err(crate::query::canceled_error());
}
if rows_exported as usize >= row_limit {
break;
}
let row = row_result.map_err(pg_error_to_string)?;
let values: Vec<serde_json::Value> = (0..row.columns().len())
.map(|i| pg_value_to_json_classified(&row, i, column_classes.get(i).copied().unwrap_or(PgColType::Other)))
.collect();
on_row(&values)?;
rows_exported += 1;
}
Ok(rows_exported)
}
async fn stream_query_rows_text_on_client(
client: &deadpool_postgres::Client,
sql: &str,
max_rows: Option<usize>,
cancelled: &AtomicBool,
column_classes: Option<&[PgColType]>,
on_row: &mut impl FnMut(&[serde_json::Value]) -> Result<(), String>,
) -> Result<u64, String> {
let stream = client.simple_query_raw(sql).await.map_err(pg_error_to_string)?;
tokio::pin!(stream);
let row_limit = max_rows.unwrap_or(usize::MAX);
let mut rows_exported = 0_u64;
while let Some(message) = stream.next().await {
if cancelled.load(std::sync::atomic::Ordering::SeqCst) {
return Err(crate::query::canceled_error());
}
if rows_exported as usize >= row_limit {
break;
}
let message = message.map_err(pg_error_to_string)?;
if let SimpleQueryMessage::Row(row) = message {
let mut values = Vec::with_capacity(row.len());
for i in 0..row.len() {
values.push(match row.try_get(i).map_err(pg_error_to_string)? {
Some(value) => {
pg_text_fallback_value(value, column_classes.and_then(|classes| classes.get(i)).copied()).0
}
None => serde_json::Value::Null,
});
}
on_row(&values)?;
rows_exported += 1;
}
}
Ok(rows_exported)
}
pub async fn connect(url: &str, fallback_timeout: Duration) -> Result<Pool, String> {
#[cfg(all(windows, target_vendor = "win7"))]
{
connect_with_optional_local_timezone(url, fallback_timeout, None).await
}
#[cfg(not(all(windows, target_vendor = "win7")))]
{
let timezone = iana_time_zone::get_timezone().unwrap_or_else(|_| "UTC".to_string());
connect_with_local_timezone(url, fallback_timeout, &timezone).await
}
}
async fn connect_with_local_timezone(url: &str, fallback_timeout: Duration, timezone: &str) -> Result<Pool, String> {
connect_with_optional_local_timezone(url, fallback_timeout, Some(timezone)).await
}
/// Identity of one physical backend connection for notice attribution:
/// (server address, server port, backend PID). The PID alone is not unique
/// across different servers, so the server's own address/port disambiguate
/// (`inet_server_addr()` is NULL for Unix sockets, hence the fallback).
type PostgresConnectionKey = (String, i32, i32);
const POSTGRES_CONNECTION_IDENTITY_SQL: &str = "SELECT pg_backend_pid(), \
COALESCE(host(inet_server_addr()), 'unix'), \
COALESCE(inet_server_port(), current_setting('port')::integer)";
/// Notice buffers for live connections, keyed by connection identity. Entries
/// are weak so they disappear once the pooled connection (and its driver
/// task) is dropped.
type PostgresNoticeBuffers = HashMap<PostgresConnectionKey, Weak<Mutex<Vec<QueryMessage>>>>;
fn postgres_notice_buffers() -> &'static Mutex<PostgresNoticeBuffers> {
static BUFFERS: OnceLock<Mutex<PostgresNoticeBuffers>> = OnceLock::new();
BUFFERS.get_or_init(|| Mutex::new(HashMap::new()))
}
/// Connection identity cache, keyed by the per-physical-connection statement
/// cache pointer (same identity pattern as `postgres_single_schema_clients`).
/// The `Weak` guards against pointer reuse: a new connection whose
/// `StatementCache` lands on a freed entry's address fails the `ptr_eq` check
/// and is treated as a miss. The value is `None` when the identity query
/// failed, so it is issued at most once per physical connection — it must
/// never be retried from `drain_postgres_notices`, which can run inside the
/// read-only transaction used for EXPLAIN, where a failing query would abort
/// the user's statement.
type PostgresClientKeys = HashMap<usize, (Weak<deadpool_postgres::StatementCache>, Option<PostgresConnectionKey>)>;
fn postgres_client_keys() -> &'static Mutex<PostgresClientKeys> {
static KEYS: OnceLock<Mutex<PostgresClientKeys>> = OnceLock::new();
KEYS.get_or_init(|| Mutex::new(HashMap::new()))
}
/// Establishes connections like deadpool's `ConfigConnectImpl`, but drives
/// each connection with a task that captures `NoticeResponse` messages
/// (`RAISE NOTICE`/`WARNING`, etc.) into a per-backend buffer instead of
/// discarding them. Query execution drains the buffer so notices are
/// attached to the `QueryResult` of the statement that raised them.
struct NoticeCapturingConnect<T>
where
T: MakeTlsConnect<Socket> + Clone + Sync + Send + 'static,
T::Stream: Sync + Send,
T::TlsConnect: Sync + Send,
<T::TlsConnect as TlsConnect<Socket>>::Future: Send,
{
tls: T,
}
impl<T> deadpool_postgres::Connect for NoticeCapturingConnect<T>
where
T: MakeTlsConnect<Socket> + Clone + Sync + Send + 'static,
T::Stream: Sync + Send,
T::TlsConnect: Sync + Send,
<T::TlsConnect as TlsConnect<Socket>>::Future: Send,
{
fn connect(
&self,
pg_config: &tokio_postgres::Config,
) -> Pin<
Box<dyn Future<Output = Result<(tokio_postgres::Client, JoinHandle<()>), tokio_postgres::Error>> + Send + '_>,
> {
let tls = self.tls.clone();
let pg_config = pg_config.clone();
Box::pin(async move {
let (client, mut connection) = pg_config.connect(tls).await?;
// No query can complete before the connection is being driven, so
// the notice buffer is handed to the driver task through a slot
// that is filled once the backend PID is known.
let notice_buffer = Arc::new(Mutex::new(None::<Arc<Mutex<Vec<QueryMessage>>>>));
let task_buffer = Arc::clone(&notice_buffer);
let conn_task = tokio::spawn(async move {
loop {
match std::future::poll_fn(|cx| connection.poll_message(cx)).await {
Some(Ok(AsyncMessage::Notice(error))) => {
let message = QueryMessage {
severity: error.severity().to_string(),
message: error.message().to_string(),
code: Some(error.code().code().to_string()),
detail: error.detail().map(str::to_string),
hint: error.hint().map(str::to_string),
};
let buffer = task_buffer.lock().unwrap_or_else(|poisoned| poisoned.into_inner()).clone();
match buffer {
Some(buffer) => {
buffer.lock().unwrap_or_else(|poisoned| poisoned.into_inner()).push(message);
}
None => {
log::info!("[postgres][notice] {}: {}", message.severity, message.message);
}
}
}
// LISTEN/NOTIFY messages are not surfaced anywhere.
Some(Ok(_)) => {}
Some(Err(err)) => {
log::warn!("[postgres] connection driver error: {err}");
break;
}
None => break,
}
}
});
// Best-effort: without the connection identity, notices cannot be
// attributed to query results on this connection and are logged
// by the driver task instead. Never fail the connection over this.
if let Ok(row) = client.query_one(POSTGRES_CONNECTION_IDENTITY_SQL, &[]).await {
let key: PostgresConnectionKey = (row.get(1), row.get(2), row.get(0));
let buffer = Arc::new(Mutex::new(Vec::new()));
let mut buffers = postgres_notice_buffers().lock().unwrap_or_else(|poisoned| poisoned.into_inner());
buffers.retain(|_, weak| weak.strong_count() > 0);
buffers.insert(key, Arc::downgrade(&buffer));
drop(buffers);
*notice_buffer.lock().unwrap_or_else(|poisoned| poisoned.into_inner()) = Some(buffer);
}
Ok((client, conn_task))
})
}
}
/// Cached identity lookup. `Some(Some(key))` = resolved, `Some(None)` =
/// identity query failed before (do not retry), `None` = never seen (or the
/// entry belonged to a dropped connection whose address was reused).
fn cached_postgres_client_key(client: &deadpool_postgres::Client) -> Option<Option<PostgresConnectionKey>> {
let cache_key = Arc::as_ptr(&client.statement_cache) as usize;
let mut keys = postgres_client_keys().lock().unwrap_or_else(|poisoned| poisoned.into_inner());
match keys.get(&cache_key) {
Some((cached, key)) if cached.upgrade().is_some_and(|cached| Arc::ptr_eq(&cached, &client.statement_cache)) => {
Some(key.clone())
}
Some(_) => {
keys.remove(&cache_key);
None
}
None => None,
}
}
/// Best-effort identity resolution. Failures are cached as `None` so the
/// identity query runs at most once per physical connection.
async fn resolve_postgres_client_key(client: &deadpool_postgres::Client) -> Option<PostgresConnectionKey> {
if let Some(key) = cached_postgres_client_key(client) {
return key;
}
let key: Option<PostgresConnectionKey> = client
.query_one(POSTGRES_CONNECTION_IDENTITY_SQL, &[])
.await
.ok()
.map(|row| (row.get(1), row.get(2), row.get(0)));
let mut keys = postgres_client_keys().lock().unwrap_or_else(|poisoned| poisoned.into_inner());
keys.retain(|_, (cached, _)| cached.strong_count() > 0);
let cache_key = Arc::as_ptr(&client.statement_cache) as usize;
keys.insert(cache_key, (Arc::downgrade(&client.statement_cache), key.clone()));
key
}
async fn postgres_client_key(client: &deadpool_postgres::Client) -> Option<PostgresConnectionKey> {
resolve_postgres_client_key(client).await
}
fn take_notices_for_key(key: &PostgresConnectionKey) -> Vec<QueryMessage> {
let buffer = {
let mut buffers = postgres_notice_buffers().lock().unwrap_or_else(|poisoned| poisoned.into_inner());
buffers.retain(|_, weak| weak.strong_count() > 0);
buffers.get(key).and_then(Weak::upgrade)
};
let Some(buffer) = buffer else {
return Vec::new();
};
let notices = std::mem::take(&mut *buffer.lock().unwrap_or_else(|poisoned| poisoned.into_inner()));
notices
}
async fn drain_postgres_notices(client: &deadpool_postgres::Client) -> Vec<QueryMessage> {
match postgres_client_key(client).await {
Some(key) => take_notices_for_key(&key),
None => Vec::new(),
}
}
async fn connect_with_optional_local_timezone(
url: &str,
fallback_timeout: Duration,
timezone: Option<&str>,
) -> Result<Pool, String> {
let url_with_keepalive = inject_postgres_keepalive_params(url);
let postgres_url = postgres_connection_url(&url_with_keepalive)?;
let _ = rustls::crypto::aws_lc_rs::default_provider().install_default();
let timeout = super::parse_connect_timeout_with_fallback(url, fallback_timeout);
let (pool, client) = super::with_connection_timeout("PostgreSQL", timeout, async {
let pg_config = tokio_postgres::Config::from_str(&postgres_url.url)
.map_err(|e| format!("Invalid PostgreSQL connection URL: {e}"))?;
// Fast recycling only checks whether the connection is already closed
// instead of issuing a validation query on every checkout, saving one
// round-trip per query. Connections that went stale without being
// observed are caught when the query runs and recovered by the
// executor's ReconnectAndRetry path (see pool_error_action / do_execute
// in query.rs).
let mgr_config = ManagerConfig { recycling_method: RecyclingMethod::Fast };
let tls_config = postgres_tls_config(
&pg_config,
&postgres_url.ssl_files,
postgres_url.accepts_invalid_certs,
postgres_url.verifies_hostname,
)?;
let mgr = deadpool_postgres::Manager::from_connect(
pg_config.clone(),
NoticeCapturingConnect { tls: tokio_postgres_rustls::MakeRustlsConnect::new(tls_config) },
mgr_config,
);
let pool = Pool::builder(mgr)
.max_size(10)
.runtime(Runtime::Tokio1)
.wait_timeout(Some(timeout))
.create_timeout(Some(timeout))
.recycle_timeout(Some(timeout))
.build()
.map_err(|e| format!("Failed to create PostgreSQL pool: {e}"))?;
// Verify connectivity. Explicit connection options are handled by
// PostgreSQL during startup and must remain strict.
let client =
pool.get().await.map_err(|e| format!("PostgreSQL connection failed: {}", pg_pool_error_to_string(e)))?;
Ok((pool, client))
})
.await?;
// Creating the physical connection and applying session defaults are two
// sequential network phases. Give each phase the configured connection
// timeout instead of sharing one deadline that can expire during the
// optional SET timezone round-trip on higher-latency tunnels.
if !pg_url_has_timezone_setting(url) {
if let Some(timezone) = timezone {
postgres_session_setup_with_timeout(timeout, set_automatic_postgres_timezone(&client, timezone)).await?;
}
}
drop(client);
Ok(pool)
}
async fn postgres_session_setup_with_timeout<T, F>(timeout: Duration, future: F) -> Result<T, String>
where
F: Future<Output = Result<T, String>>,
{
super::with_connection_timeout("PostgreSQL session setup", timeout, future).await
}
async fn set_automatic_postgres_timezone(client: &deadpool_postgres::Client, timezone: &str) -> Result<(), String> {
let candidates = postgres_timezone_candidates(timezone);
for (index, candidate) in candidates.iter().enumerate() {
let sql = format!("SET timezone = '{}'", candidate.replace('\'', "''"));
match client.execute(&sql, &[]).await {
Ok(_) => {
if *candidate != timezone {
log::warn!(
"PostgreSQL does not recognize local timezone '{timezone}'; using compatible alias '{candidate}'"
);
}
return Ok(());
}
Err(error) if postgres_timezone_error_is_nonfatal(&error) => {
let detail = pg_error_to_string(error);
if index + 1 == candidates.len() {
// A connected server may have older tzdata or only partial PostgreSQL compatibility.
// Keep its session default rather than making optional local display alignment fatal.
log::warn!(
"PostgreSQL connected, but automatic local timezone '{timezone}' was rejected; \
keeping the server default timezone: {detail}"
);
return Ok(());
}
}
Err(error) => {
return Err(format!("PostgreSQL SET timezone failed after connecting: {}", pg_error_to_string(error)));
}
}
}
Ok(())
}
fn postgres_timezone_error_is_nonfatal(error: &tokio_postgres::Error) -> bool {
let Some(db_error) = error.as_db_error() else {
return false;
};
// SET failures reported as ordinary SQL errors are optional session setup.
// FATAL/PANIC responses mean the connection itself is not safe to return.
!matches!(
db_error.parsed_severity(),
Some(tokio_postgres::error::Severity::Fatal | tokio_postgres::error::Severity::Panic)
) && !matches!(db_error.severity().to_ascii_uppercase().as_str(), "FATAL" | "PANIC")
}
fn postgres_timezone_candidates(timezone: &str) -> Vec<&str> {
let legacy_alias = match timezone {
"Asia/Saigon" => Some("Asia/Ho_Chi_Minh"),
"Asia/Ho_Chi_Minh" => Some("Asia/Saigon"),
"Europe/Kyiv" => Some("Europe/Kiev"),
"Europe/Kiev" => Some("Europe/Kyiv"),
"Asia/Calcutta" => Some("Asia/Kolkata"),
"Asia/Kolkata" => Some("Asia/Calcutta"),
_ => None,
};
std::iter::once(timezone).chain(legacy_alias).collect()
}
#[derive(Debug, Default, Clone, PartialEq, Eq)]
pub struct PostgresSslFiles {
pub sslcert: Option<String>,
pub sslkey: Option<String>,
pub sslrootcert: Option<String>,
}
/// TLS context info, used to reconstruct the TLS connector when cancelling a query.
#[derive(Debug, Clone)]
pub struct PostgresCancelContext {
pub ssl_files: PostgresSslFiles,
pub accepts_invalid_certs: bool,
pub verifies_hostname: bool,
pub ssl_mode: SslMode,
}
/// Build a TLS cancel context from the connection URL.
/// Returns None if URL parsing fails or sslmode=disable (no TLS cancel needed).
pub fn build_postgres_cancel_context(url: &str) -> Option<PostgresCancelContext> {
let postgres_url = postgres_connection_url(url).ok()?;
let pg_config = tokio_postgres::Config::from_str(&postgres_url.url).ok()?;
if pg_config.get_ssl_mode() == SslMode::Disable {
return None;
}
Some(PostgresCancelContext {
ssl_files: postgres_url.ssl_files,
accepts_invalid_certs: postgres_url.accepts_invalid_certs,
verifies_hostname: postgres_url.verifies_hostname,
ssl_mode: pg_config.get_ssl_mode(),
})
}
/// Reconstruct a TLS connector from the cancel context, used for TLS connection cancellation.
fn make_rustls_connect_from_context(
ctx: &PostgresCancelContext,
) -> Result<tokio_postgres_rustls::MakeRustlsConnect, String> {
// Build a minimal pg_config solely for ssl_mode determination
let mut pg_config = tokio_postgres::Config::new();
pg_config.ssl_mode(ctx.ssl_mode);
let tls_config = postgres_tls_config(&pg_config, &ctx.ssl_files, ctx.accepts_invalid_certs, ctx.verifies_hostname)?;
Ok(tokio_postgres_rustls::MakeRustlsConnect::new(tls_config))
}
#[derive(Debug, Clone, PartialEq, Eq)]
struct PostgresConnectionUrl {
url: String,
ssl_files: PostgresSslFiles,
accepts_invalid_certs: bool,
verifies_hostname: bool,
}
/// Inject TCP keepalive parameters into the PostgreSQL URL (only when the user has not explicitly specified them).
/// Default parameters shorten half-open connection detection time, suitable for desktop/VPN/NAT environments.
fn inject_postgres_keepalive_params(url: &str) -> String {
let (base, fragment) = url.split_once('#').map_or((url, ""), |(base, fragment)| (base, fragment));
let query = base.split('?').nth(1);
let has_keepalives = query
.map(|q| q.split('&').any(|p| p.split('=').next().is_some_and(|k| k.eq_ignore_ascii_case("keepalives"))))
.unwrap_or(false);
if has_keepalives {
return url.to_string(); // User has explicitly configured keepalive
}
let separator = if base.contains('?') { "&" } else { "?" };
let injected =
format!("{base}{separator}keepalives=1&keepalives_idle=30&keepalives_interval=10&keepalives_retries=3");
if fragment.is_empty() {
injected
} else {
format!("{injected}#{fragment}")
}
}
fn postgres_connection_url(url: &str) -> Result<PostgresConnectionUrl, String> {
let Some(query_start) = url.find('?') else {
let pg_config =
tokio_postgres::Config::from_str(url).map_err(|e| format!("Invalid PostgreSQL connection URL: {e}"))?;
return Ok(PostgresConnectionUrl {
url: url.to_string(),
ssl_files: PostgresSslFiles::default(),
accepts_invalid_certs: postgres_sslmode_accepts_invalid_certs(pg_config.get_ssl_mode()),
verifies_hostname: false,
});
};
let prefix = &url[..query_start];
let suffix = &url[query_start + 1..];
let (query_string, fragment) = suffix.split_once('#').map_or((suffix, ""), |(query, fragment)| (query, fragment));
let mut ssl_files = PostgresSslFiles::default();
let mut kept_params = Vec::new();
let mut accepts_invalid_certs = true;
let mut verifies_hostname = false;
for param in query_string.split('&') {
if param.is_empty() {
continue;
}
let Some((key, value)) = param.split_once('=') else {
kept_params.push(param.to_string());
continue;
};
if key.eq_ignore_ascii_case("sslcert")
|| key.eq_ignore_ascii_case("sslkey")
|| key.eq_ignore_ascii_case("sslrootcert")
{
let decoded = percent_decode_str(value)
.decode_utf8()
.map_err(|_| format!("Invalid URL encoding in {key}"))?
.into_owned();
validate_file_path(&decoded, |_| false).map_err(|e| format!("{key}: {e}"))?;
if key.eq_ignore_ascii_case("sslcert") {
ssl_files.sslcert = Some(decoded);
} else if key.eq_ignore_ascii_case("sslkey") {
ssl_files.sslkey = Some(decoded);
} else {
ssl_files.sslrootcert = Some(decoded);
}
} else if key.eq_ignore_ascii_case("channel_binding") {
// channel_binding=require fails when the server does not offer
// SCRAM-SHA-256-PLUS (e.g. Neon). Normalize require→prefer so
// channel binding is used when available but does not cause a
// hard failure when the server doesn't support it.
match value.to_ascii_lowercase().as_str() {
"require" => kept_params.push("channel_binding=prefer".to_string()),
_ => kept_params.push(param.to_string()),
}
} else if key.eq_ignore_ascii_case("sslmode") {
match value.to_ascii_lowercase().as_str() {
"verify-ca" => {
accepts_invalid_certs = false;
kept_params.push("sslmode=require".to_string());
}
"verify-full" | "verify_identity" | "verify-identity" => {
accepts_invalid_certs = false;
verifies_hostname = true;
kept_params.push("sslmode=require".to_string());
}
"disable" => {
accepts_invalid_certs = false;
kept_params.push(param.to_string());
}
"prefer" | "require" => {
accepts_invalid_certs = true;
kept_params.push(param.to_string());
}
_ => kept_params.push(param.to_string()),
}
} else {
kept_params.push(param.to_string());
}
}
let mut sanitized_url = prefix.to_string();
if !kept_params.is_empty() {
sanitized_url.push('?');
sanitized_url.push_str(&kept_params.join("&"));
}
if !fragment.is_empty() {
sanitized_url.push('#');
sanitized_url.push_str(fragment);
}
Ok(PostgresConnectionUrl { url: sanitized_url, ssl_files, accepts_invalid_certs, verifies_hostname })
}
fn postgres_tls_config(
pg_config: &tokio_postgres::Config,
ssl_files: &PostgresSslFiles,
accepts_invalid_certs: bool,
verifies_hostname: bool,
) -> Result<rustls::ClientConfig, String> {
if pg_config.get_ssl_mode() != SslMode::Disable && accepts_invalid_certs {
let provider = Arc::new(rustls::crypto::aws_lc_rs::default_provider());
let builder = rustls::ClientConfig::builder()
.dangerous()
.with_custom_certificate_verifier(Arc::new(NoPostgresCertVerification { provider }));
return postgres_tls_client_auth(builder, ssl_files);
}
let root_store = postgres_root_cert_store(ssl_files)?;
let builder = if verifies_hostname {
rustls::ClientConfig::builder().with_root_certificates(root_store)
} else {
let provider = Arc::new(rustls::crypto::aws_lc_rs::default_provider());
rustls::ClientConfig::builder().dangerous().with_custom_certificate_verifier(Arc::new(
PostgresCaOnlyCertVerification { provider, roots: Arc::new(root_store) },
))
};
postgres_tls_client_auth(builder, ssl_files)
}
fn postgres_root_cert_store(ssl_files: &PostgresSslFiles) -> Result<rustls::RootCertStore, String> {
let mut root_store = rustls::RootCertStore::empty();
if let Some(path) = ssl_files.sslrootcert.as_deref() {
let certs = read_postgres_pem_certs("sslrootcert", path)?;
let (valid_count, _) = root_store.add_parsable_certificates(certs);
if valid_count == 0 {
return Err(format!("sslrootcert: no valid CA certificates found in {path}"));
}
} else {
root_store.extend(webpki_roots::TLS_SERVER_ROOTS.iter().cloned());
}
Ok(root_store)
}
fn postgres_tls_client_auth(
builder: rustls::ConfigBuilder<rustls::ClientConfig, rustls::client::WantsClientCert>,
ssl_files: &PostgresSslFiles,
) -> Result<rustls::ClientConfig, String> {
match (ssl_files.sslcert.as_deref(), ssl_files.sslkey.as_deref()) {
(Some(cert_path), Some(key_path)) => {
let certs = read_postgres_pem_certs("sslcert", cert_path)?;
if certs.is_empty() {
return Err(format!("sslcert: no certificates found in {cert_path}"));
}
let private_key = read_postgres_private_key(key_path)?;
builder
.with_client_auth_cert(certs, private_key)
.map_err(|e| format!("PostgreSQL client certificate/key mismatch or invalid key: {e}"))
}
(Some(_), None) => Err("PostgreSQL sslcert requires sslkey".to_string()),
(None, Some(_)) => Err("PostgreSQL sslkey requires sslcert".to_string()),
(None, None) => Ok(builder.with_no_client_auth()),
}
}
fn read_postgres_pem_certs(label: &str, path: &str) -> Result<Vec<CertificateDer<'static>>, String> {
let file = File::open(path).map_err(|e| format!("{label}: failed to open {path}: {e}"))?;
let mut reader = BufReader::new(file);
rustls_pemfile::certs(&mut reader)
.collect::<Result<Vec<_>, _>>()
.map_err(|e| format!("{label}: failed to read PEM certificates from {path}: {e}"))
}
fn read_postgres_private_key(path: &str) -> Result<PrivateKeyDer<'static>, String> {
let file = File::open(path).map_err(|e| format!("sslkey: failed to open {path}: {e}"))?;
let mut reader = BufReader::new(file);
rustls_pemfile::private_key(&mut reader)
.map_err(|e| format!("sslkey: failed to read PEM private key from {path}: {e}"))?
.ok_or_else(|| format!("sslkey: no private key found in {path}"))
}
fn postgres_sslmode_accepts_invalid_certs(ssl_mode: SslMode) -> bool {
matches!(ssl_mode, SslMode::Prefer | SslMode::Require)
}
#[derive(Debug)]
struct NoPostgresCertVerification {
provider: Arc<CryptoProvider>,
}
impl ServerCertVerifier for NoPostgresCertVerification {
fn verify_server_cert(
&self,
_end_entity: &CertificateDer<'_>,
_intermediates: &[CertificateDer<'_>],
_server_name: &ServerName<'_>,
_ocsp_response: &[u8],
_now: UnixTime,
) -> Result<ServerCertVerified, rustls::Error> {
Ok(ServerCertVerified::assertion())
}
fn verify_tls12_signature(
&self,
_message: &[u8],
cert: &CertificateDer<'_>,
_dss: &rustls::DigitallySignedStruct,
) -> Result<HandshakeSignatureValid, rustls::Error> {
self.accept_tls_signature_for_unverified_cert(cert)
}
fn verify_tls13_signature(
&self,
_message: &[u8],
cert: &CertificateDer<'_>,
_dss: &rustls::DigitallySignedStruct,
) -> Result<HandshakeSignatureValid, rustls::Error> {
self.accept_tls_signature_for_unverified_cert(cert)
}
fn supported_verify_schemes(&self) -> Vec<rustls::SignatureScheme> {
self.provider.signature_verification_algorithms.supported_schemes()
}
}
impl NoPostgresCertVerification {
fn accept_tls_signature_for_unverified_cert(
&self,
_cert: &CertificateDer<'_>,
) -> Result<HandshakeSignatureValid, rustls::Error> {
// PostgreSQL sslmode=prefer/require does not authenticate the server certificate.
// Avoid rustls' default signature helpers here because they parse the certificate
// before chain verification and reject legacy server certificates that libpq/JDBC
// still accept in these non-verifying modes.
Ok(HandshakeSignatureValid::assertion())
}
}
#[derive(Debug)]
struct PostgresCaOnlyCertVerification {
provider: Arc<CryptoProvider>,
roots: Arc<rustls::RootCertStore>,
}
impl ServerCertVerifier for PostgresCaOnlyCertVerification {
fn verify_server_cert(
&self,
end_entity: &CertificateDer<'_>,
intermediates: &[CertificateDer<'_>],
_server_name: &ServerName<'_>,
_ocsp_response: &[u8],
now: UnixTime,
) -> Result<ServerCertVerified, rustls::Error> {
let cert = ParsedCertificate::try_from(end_entity)?;
verify_server_cert_signed_by_trust_anchor(
&cert,
&self.roots,
intermediates,
now,
self.provider.signature_verification_algorithms.all,
)?;
Ok(ServerCertVerified::assertion())
}
fn verify_tls12_signature(
&self,
message: &[u8],
cert: &CertificateDer<'_>,
dss: &rustls::DigitallySignedStruct,
) -> Result<HandshakeSignatureValid, rustls::Error> {
verify_tls12_signature(message, cert, dss, &self.provider.signature_verification_algorithms)
}
fn verify_tls13_signature(
&self,
message: &[u8],
cert: &CertificateDer<'_>,
dss: &rustls::DigitallySignedStruct,
) -> Result<HandshakeSignatureValid, rustls::Error> {
verify_tls13_signature(message, cert, dss, &self.provider.signature_verification_algorithms)
}
fn supported_verify_schemes(&self) -> Vec<rustls::SignatureScheme> {
self.provider.signature_verification_algorithms.supported_schemes()
}
}
/// Check whether the user's connection URL already specifies a timezone via
/// the `options` parameter so we don't overwrite it with the local timezone.
fn pg_url_has_timezone_setting(url: &str) -> bool {
let Some(query) = url.split_once('?').map(|(_, query)| query.split('#').next().unwrap_or(query)) else {
return false;
};
query.split('&').any(|parameter| {
let (raw_key, raw_value) = parameter.split_once('=').unwrap_or((parameter, ""));
let key = percent_decode_str(raw_key).decode_utf8_lossy();
if !key.eq_ignore_ascii_case("options") {
return false;
}
let options = percent_decode_str(raw_value).decode_utf8_lossy().to_ascii_lowercase();
options.split_ascii_whitespace().any(|token| {
let option = token.trim_start_matches('-');
option.starts_with("timezone=") || option.starts_with("time_zone=")
})
})
}
#[cfg(test)]
fn validate_postgres_ssl_paths(url: &str) -> Result<(), String> {
postgres_connection_url(url).map(|_| ())
}
fn list_databases_sql() -> &'static str {
"SELECT datname FROM pg_database \
WHERE datallowconn = true \
ORDER BY datname"
}
fn database_storage_sql() -> &'static str {
"SELECT d.datname, \
CASE \
WHEN has_database_privilege(d.datname, 'CONNECT') \
OR COALESCE(( \
SELECT pg_has_role(current_user, r.oid, 'MEMBER') \
FROM pg_roles r \
WHERE r.rolname = 'pg_read_all_stats' \
), false) \
THEN pg_database_size(d.oid) \
ELSE NULL \
END AS size_bytes \
FROM pg_database d \
WHERE d.datallowconn = true \
AND d.datname = ANY($1::text[]) \
ORDER BY d.datname"
}
pub async fn list_databases(pool: &Pool) -> Result<Vec<DatabaseInfo>, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let rows = postgres_query_cached(&client, list_databases_sql(), &[]).await.map_err(|e| e.to_string())?;
Ok(rows.iter().map(|row| DatabaseInfo { name: pg_row_try_string(row, 0) }).collect())
}
pub async fn list_database_storage(pool: &Pool, database_names: &[String]) -> Result<Vec<DatabaseStorageInfo>, String> {
if database_names.is_empty() {
return Ok(Vec::new());
}
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let rows =
postgres_query_cached(&client, database_storage_sql(), &[&database_names]).await.map_err(|e| e.to_string())?;
Ok(rows
.iter()
.map(|row| DatabaseStorageInfo {
name: pg_row_try_string(row, 0),
size_bytes: row.try_get::<_, Option<i64>>(1).ok().flatten(),
})
.collect())
}
pub async fn list_tables(pool: &Pool, schema: &str) -> Result<Vec<TableInfo>, String> {
list_tables_filtered(pool, schema, None, None, None).await
}
pub async fn list_tables_filtered(
pool: &Pool,
schema: &str,
filter: Option<&str>,
limit: Option<usize>,
offset: Option<usize>,
) -> Result<Vec<TableInfo>, String> {
let schema = if schema.is_empty() { "public" } else { schema };
let filter = filter.unwrap_or("").trim();
let filter_pattern = like_contains_pattern(filter);
let fuzzy_filter_pattern =
if crate::sql::fuzzy_filter_enabled(filter) { like_fuzzy_pattern(filter) } else { String::new() };
let limit_param = limit.and_then(|value| i64::try_from(value).ok());
let offset_param = offset.and_then(|value| i64::try_from(value).ok()).unwrap_or(0);
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let sql = postgres_tables_sql(limit_param, offset_param);
let params: &[(&(dyn tokio_postgres::types::ToSql + Sync), Type)] =
&[(&schema, Type::VARCHAR), (&filter_pattern, Type::VARCHAR), (&fuzzy_filter_pattern, Type::VARCHAR)];
// The pagination literals make this SQL vary by page. Use an unnamed typed
// query so each load stays one round trip without growing the statement cache.
let rows = client.query_typed(&sql, params).await.map_err(|e| e.to_string())?;
Ok(rows
.iter()
.map(|row| TableInfo {
name: pg_row_try_string(row, 0),
table_type: pg_row_try_string(row, 1),
comment: row.try_get::<_, Option<String>>(2).ok().flatten().filter(|s| !s.is_empty()),
parent_schema: row.try_get::<_, Option<String>>(3).ok().flatten().filter(|s| !s.is_empty()),
parent_name: row.try_get::<_, Option<String>>(4).ok().flatten().filter(|s| !s.is_empty()),
})
.collect())
}
pub async fn completion_assistant_search(
pool: &Pool,
request: &CompletionAssistantRequest,
) -> Result<CompletionAssistantResponse, String> {
let schema = request.schema.as_deref().or(request.parent_schema.as_deref());
let routine_schema = schema.unwrap_or("public");
let limit = request.max_results.unwrap_or(100).clamp(1, 1000);
let kinds = if request.object_kinds.is_empty() {
vec![CompletionAssistantObjectKind::Table, CompletionAssistantObjectKind::View]
} else {
request.object_kinds.clone()
};
let pattern = postgres_completion_like_pattern(&request.mask, request.match_mode.as_ref());
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let mut candidates = Vec::new();
if kinds.iter().any(|kind| matches!(kind, CompletionAssistantObjectKind::Schema)) {
for row in postgres_query_cached(
&client,
"SELECT nspname FROM pg_catalog.pg_namespace \
WHERE nspname NOT LIKE 'pg_%' AND nspname <> 'information_schema' \
AND ($1 = '%%' OR nspname ILIKE $1 ESCAPE '~') \
ORDER BY nspname LIMIT $2",
&[&pattern, &(limit as i64)],
)
.await
.map_err(|e| e.to_string())?
{
let schema_name: String = pg_row_try_string(&row, 0);
candidates.push(CompletionAssistantCandidate {
name: schema_name.clone(),
kind: CompletionAssistantCandidateKind::Schema,
database: Some(request.database.clone()),
schema: Some(schema_name),
parent_schema: None,
parent_name: None,
comment: None,
data_type: None,
signature: None,
});
}
}
if candidates.len() < limit && kinds.iter().any(CompletionAssistantObjectKind::is_table_like) {
let relkinds = postgres_completion_relkinds(&kinds);
let rows = postgres_query_cached(
&client,
postgres_completion_tables_sql(),
&[&schema, &pattern, &relkinds, &((limit - candidates.len()) as i64)],
)
.await
.map_err(|e| e.to_string())?;
for row in rows {
let table_type: String = pg_row_try_string(&row, 2);
candidates.push(CompletionAssistantCandidate {
name: pg_row_try_string(&row, 0),
kind: if table_type.contains("VIEW") {
CompletionAssistantCandidateKind::View
} else {
CompletionAssistantCandidateKind::Table
},
database: Some(request.database.clone()),
schema: Some(pg_row_try_string(&row, 1)),
parent_schema: row.try_get::<_, Option<String>>(4).ok().flatten(),
parent_name: row.try_get::<_, Option<String>>(5).ok().flatten(),
comment: row.try_get::<_, Option<String>>(3).ok().flatten(),
data_type: None,
signature: None,
});
}
}
if candidates.len() < limit && kinds.iter().any(CompletionAssistantObjectKind::is_routine_like) {
let prokinds = postgres_completion_prokinds(&kinds);
let rows = postgres_query_cached(
&client,
postgres_completion_routines_sql(),
&[&routine_schema, &pattern, &prokinds, &((limit - candidates.len()) as i64)],
)
.await
.map_err(|e| e.to_string())?;
for row in rows {
let routine_type: String = pg_row_try_string(&row, 2);
candidates.push(CompletionAssistantCandidate {
name: pg_row_try_string(&row, 0),
kind: if routine_type == "PROCEDURE" {
CompletionAssistantCandidateKind::Procedure
} else {
CompletionAssistantCandidateKind::Function
},
database: Some(request.database.clone()),
schema: Some(pg_row_try_string(&row, 1)),
parent_schema: None,
parent_name: None,
comment: row.try_get::<_, Option<String>>(3).ok().flatten(),
data_type: row.try_get::<_, Option<String>>(4).ok().flatten(),
signature: row.try_get::<_, Option<String>>(5).ok().flatten(),
});
}
}
if candidates.len() < limit && kinds.iter().any(|kind| matches!(kind, CompletionAssistantObjectKind::Column)) {
let table = request.parent_name.as_deref().unwrap_or("");
if !table.is_empty() {
// Unqualified PostgreSQL objects resolve through search_path, so column
// metadata must use the same visible relation instead of assuming public.
let resolved_schema = match schema {
Some(schema) => Some(schema.to_string()),
None => postgres_query_cached(&client, postgres_visible_table_schema_sql(), &[&table])
.await
.map_err(|e| e.to_string())?
.first()
.map(|row| pg_row_try_string(row, 0)),
};
let Some(resolved_schema) = resolved_schema else {
return Ok(CompletionAssistantResponse { incomplete: false, candidates, fallback_used: false });
};
let rows = postgres_query_cached(
&client,
postgres_completion_columns_sql(),
&[&resolved_schema, &table, &pattern, &((limit - candidates.len()) as i64)],
)
.await
.map_err(|e| e.to_string())?;
for row in rows {
candidates.push(CompletionAssistantCandidate {
name: pg_row_try_string(&row, 0),
kind: CompletionAssistantCandidateKind::Column,
database: Some(request.database.clone()),
schema: Some(resolved_schema.clone()),
parent_schema: Some(resolved_schema.clone()),
parent_name: Some(table.to_string()),
comment: row.try_get::<_, Option<String>>(2).ok().flatten(),
data_type: Some(pg_row_try_string(&row, 1)),
signature: None,
});
}
}
}
Ok(CompletionAssistantResponse { incomplete: candidates.len() >= limit, candidates, fallback_used: false })
}
fn postgres_completion_tables_sql() -> &'static str {
"SELECT c.relname, n.nspname, \
CASE c.relkind WHEN 'v' THEN 'VIEW' WHEN 'm' THEN 'VIEW' ELSE 'TABLE' END AS table_type, \
obj_description(c.oid) AS table_comment, \
CASE WHEN pc.relkind = 'p' THEN pn.nspname ELSE NULL END AS parent_schema, \
CASE WHEN pc.relkind = 'p' THEN pc.relname ELSE NULL END AS parent_name \
FROM pg_catalog.pg_class c \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
LEFT JOIN pg_catalog.pg_inherits i ON i.inhrelid = c.oid \
LEFT JOIN pg_catalog.pg_class pc ON pc.oid = i.inhparent \
LEFT JOIN pg_catalog.pg_namespace pn ON pn.oid = pc.relnamespace \
WHERE ($1::text IS NOT NULL AND n.nspname = $1 \
OR $1::text IS NULL AND pg_catalog.pg_table_is_visible(c.oid)) \
AND c.relkind::text = ANY($3::text[]) \
AND ($2 = '%%' OR c.relname ILIKE $2 ESCAPE '~') \
ORDER BY c.relname LIMIT $4"
}
fn postgres_completion_routines_sql() -> &'static str {
"SELECT p.proname, n.nspname, CASE p.prokind WHEN 'p' THEN 'PROCEDURE' ELSE 'FUNCTION' END, \
obj_description(p.oid) AS routine_comment, COALESCE(pg_get_function_result(p.oid), '') AS data_type, \
pg_get_function_identity_arguments(p.oid) AS signature \
FROM pg_catalog.pg_proc p \
JOIN pg_catalog.pg_namespace n ON n.oid = p.pronamespace \
WHERE n.nspname = $1 AND p.prokind::text = ANY($3::text[]) \
AND ($2 = '%%' OR p.proname ILIKE $2 ESCAPE '~') \
ORDER BY p.proname LIMIT $4"
}
fn postgres_completion_columns_sql() -> &'static str {
"SELECT a.attname, pg_catalog.format_type(a.atttypid, a.atttypmod), col_description(c.oid, a.attnum) \
FROM pg_catalog.pg_attribute a \
JOIN pg_catalog.pg_class c ON c.oid = a.attrelid \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
WHERE n.nspname = $1 AND c.relname = $2 AND a.attnum > 0 AND NOT a.attisdropped \
AND ($3 = '%%' OR a.attname ILIKE $3 ESCAPE '~') \
ORDER BY a.attnum LIMIT $4"
}
fn postgres_visible_table_schema_sql() -> &'static str {
"SELECT n.nspname FROM pg_catalog.pg_class c \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
WHERE c.relname = $1 AND pg_catalog.pg_table_is_visible(c.oid) \
LIMIT 1"
}
fn postgres_completion_relkinds(kinds: &[CompletionAssistantObjectKind]) -> Vec<String> {
let mut relkinds = Vec::new();
if kinds.iter().any(|kind| matches!(kind, CompletionAssistantObjectKind::Table)) {
relkinds.extend(["r", "p", "f"].into_iter().map(str::to_string));
}
if kinds.iter().any(|kind| matches!(kind, CompletionAssistantObjectKind::View)) {
relkinds.extend(["v", "m"].into_iter().map(str::to_string));
}
relkinds
}
fn postgres_completion_prokinds(kinds: &[CompletionAssistantObjectKind]) -> Vec<String> {
let mut prokinds = Vec::new();
if kinds
.iter()
.any(|kind| matches!(kind, CompletionAssistantObjectKind::Procedure | CompletionAssistantObjectKind::Routine))
{
prokinds.push("p".to_string());
}
if kinds
.iter()
.any(|kind| matches!(kind, CompletionAssistantObjectKind::Function | CompletionAssistantObjectKind::Routine))
{
prokinds.push("f".to_string());
}
prokinds
}
fn postgres_completion_like_pattern(value: &str, mode: Option<&CompletionAssistantMatchMode>) -> String {
if value.trim().is_empty() || value == "%" {
return "%%".to_string();
}
let escaped = value.trim().replace('~', "~~").replace('%', "~%").replace('_', "~_");
match mode.unwrap_or(&CompletionAssistantMatchMode::Prefix) {
CompletionAssistantMatchMode::Prefix => format!("{escaped}%"),
CompletionAssistantMatchMode::Contains => format!("%{escaped}%"),
}
}
pub async fn get_table_comment(pool: &Pool, schema: &str, table: &str) -> Result<Option<String>, String> {
let schema = if schema.is_empty() { "public" } else { schema };
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let rows = postgres_query_cached(&client, postgres_table_comment_sql(), &[&schema, &table])
.await
.map_err(|e| e.to_string())?;
Ok(rows.first().and_then(|row| row.try_get::<_, Option<String>>(0).ok().flatten()).filter(|s| !s.is_empty()))
}
pub async fn get_table_partition_info(
pool: &Pool,
schema: &str,
table: &str,
) -> Result<PostgresTablePartitionInfo, String> {
let schema = if schema.is_empty() { "public" } else { schema };
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let relation_rows = postgres_query_cached(&client, postgres_table_partition_relation_sql(), &[&schema, &table])
.await
.map_err(|e| e.to_string())?;
let Some(relation) = relation_rows.first() else {
return Ok(PostgresTablePartitionInfo::default());
};
let relkind = relation.try_get::<_, String>(0).unwrap_or_default();
let is_partition = relation.try_get::<_, bool>(1).unwrap_or(false);
if relkind != "p" && !is_partition {
return Ok(PostgresTablePartitionInfo::default());
}
let rows = postgres_query_cached(&client, postgres_table_partition_info_sql(), &[&schema, &table])
.await
.map_err(|e| e.to_string())?;
let Some(row) = rows.first() else {
return Ok(PostgresTablePartitionInfo { is_partition, ..Default::default() });
};
Ok(PostgresTablePartitionInfo {
is_partition,
parent_schema: row.try_get::<_, Option<String>>(0).ok().flatten().filter(|value| !value.is_empty()),
parent_table: row.try_get::<_, Option<String>>(1).ok().flatten().filter(|value| !value.is_empty()),
bound: row.try_get::<_, Option<String>>(2).ok().flatten().filter(|value| !value.is_empty()),
key: row.try_get::<_, Option<String>>(3).ok().flatten().filter(|value| !value.is_empty()),
})
}
pub async fn get_table_partition_key(pool: &Pool, schema: &str, table: &str) -> Result<Option<String>, String> {
Ok(get_table_partition_info(pool, schema, table).await?.key)
}
pub async fn get_table_partition_local_objects(
pool: &Pool,
schema: &str,
table: &str,
) -> Result<PostgresTablePartitionLocalObjects, String> {
let schema = if schema.is_empty() { "public" } else { schema };
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let rows = postgres_query_cached(&client, postgres_table_partition_local_objects_sql(), &[&schema, &table])
.await
.map_err(|e| e.to_string())?;
let mut result = PostgresTablePartitionLocalObjects::default();
for row in rows {
let object_kind = row.try_get::<_, String>(0).unwrap_or_default();
let object_name = row.try_get::<_, String>(1).unwrap_or_default();
let object_type = row.try_get::<_, Option<String>>(2).ok().flatten().unwrap_or_default();
match object_kind.as_str() {
"constraint" if object_type == "p" => result.has_primary_key = true,
"constraint" if object_type == "f" && !object_name.is_empty() => {
result.foreign_keys.insert(object_name);
}
"index" if !object_name.is_empty() => {
result.indexes.insert(object_name);
}
_ => {}
}
}
Ok(result)
}
fn postgres_table_partition_relation_sql() -> &'static str {
"SELECT c.relkind::text, \
COALESCE((pg_catalog.row_to_json(c)->>'relispartition')::boolean, false) AS is_partition \
FROM pg_catalog.pg_class c \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
WHERE n.nspname = $1 AND c.relname = $2 AND c.relkind IN ('r','p') \
LIMIT 1"
}
fn postgres_table_partition_info_sql() -> &'static str {
"SELECT CASE WHEN c.relispartition THEN pn.nspname ELSE NULL END AS parent_schema, \
CASE WHEN c.relispartition THEN pc.relname ELSE NULL END AS parent_table, \
CASE WHEN c.relispartition THEN pg_catalog.pg_get_expr(c.relpartbound, c.oid, true) ELSE NULL END AS partition_bound, \
CASE WHEN c.relkind = 'p' THEN pg_catalog.pg_get_partkeydef(c.oid) ELSE NULL END AS partition_key \
FROM pg_catalog.pg_class c \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
LEFT JOIN pg_catalog.pg_inherits i ON i.inhrelid = c.oid \
LEFT JOIN pg_catalog.pg_class pc ON pc.oid = i.inhparent \
LEFT JOIN pg_catalog.pg_namespace pn ON pn.oid = pc.relnamespace \
WHERE n.nspname = $1 AND c.relname = $2 AND c.relkind IN ('r','p') \
ORDER BY i.inhseqno NULLS LAST \
LIMIT 1"
}
fn postgres_table_partition_local_objects_sql() -> &'static str {
"SELECT 'constraint'::text AS object_kind, con.conname AS object_name, con.contype::text AS object_type \
FROM pg_catalog.pg_constraint con \
JOIN pg_catalog.pg_class c ON c.oid = con.conrelid \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
WHERE n.nspname = $1 AND c.relname = $2 AND con.contype IN ('p','f') \
AND COALESCE(NULLIF(pg_catalog.row_to_json(con)->>'conparentid', '')::oid, 0) = 0 \
UNION ALL \
SELECT 'index'::text AS object_kind, idx.relname AS object_name, NULL::text AS object_type \
FROM pg_catalog.pg_index ix \
JOIN pg_catalog.pg_class c ON c.oid = ix.indrelid \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
JOIN pg_catalog.pg_class idx ON idx.oid = ix.indexrelid \
WHERE n.nspname = $1 AND c.relname = $2 \
AND NOT EXISTS (SELECT 1 FROM pg_catalog.pg_inherits i WHERE i.inhrelid = idx.oid) \
ORDER BY object_kind, object_name"
}
fn postgres_table_comment_sql() -> &'static str {
"SELECT obj_description(c.oid) AS table_comment \
FROM pg_catalog.pg_class c \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
WHERE n.nspname = $1 AND c.relname = $2 AND c.relkind IN ('r','m','f','p') \
LIMIT 1"
}
fn postgres_tables_sql(limit: Option<i64>, offset: i64) -> String {
// PostgreSQL-compatible servers do not agree on the inferred wire types
// or accepted expression grammar for LIMIT/OFFSET parameters. These values
// originate as usize and are converted to non-negative i64 literals.
// Omitting an explicit ESCAPE keeps compatibility with servers that expose
// only two-argument ILIKE; bound patterns use the default backslash escape.
let pagination = match limit {
Some(limit) => format!("LIMIT {limit} OFFSET {offset}"),
None => format!("OFFSET {offset}"),
};
format!(
"SELECT c.relname AS table_name, \
CASE c.relkind WHEN 'r' THEN 'BASE TABLE' WHEN 'v' THEN 'VIEW' \
WHEN 'm' THEN 'MATERIALIZED_VIEW' WHEN 'f' THEN 'FOREIGN TABLE' \
WHEN 'p' THEN 'BASE TABLE' END AS table_type, \
obj_description(c.oid) AS table_comment, \
CASE WHEN pc.relkind = 'p' THEN pn.nspname ELSE NULL END AS parent_schema, \
CASE WHEN pc.relkind = 'p' THEN pc.relname ELSE NULL END AS parent_name \
FROM pg_catalog.pg_class c \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
LEFT JOIN pg_catalog.pg_inherits i ON i.inhrelid = c.oid \
LEFT JOIN pg_catalog.pg_class pc ON pc.oid = i.inhparent \
LEFT JOIN pg_catalog.pg_namespace pn ON pn.oid = pc.relnamespace \
WHERE n.nspname = $1 AND c.relkind IN ('r','v','m','f','p') \
AND ($2 = '%%' OR c.relname ILIKE $2 OR ($3 <> '' AND c.relname ILIKE $3)) \
ORDER BY CASE WHEN pc.relkind = 'p' THEN 1 ELSE 0 END, c.relname \
{pagination}"
)
}
fn like_contains_pattern(value: &str) -> String {
if value.is_empty() {
return "%%".to_string();
}
let mut pattern = String::with_capacity(value.len() + 2);
pattern.push('%');
for ch in value.chars() {
if ch == '\\' || ch == '%' || ch == '_' {
pattern.push('\\');
}
pattern.push(ch);
}
pattern.push('%');
pattern
}
fn like_fuzzy_pattern(value: &str) -> String {
crate::sql::fuzzy_like_pattern_with_escape(value, |value| {
let mut escaped = String::with_capacity(value.len() + 1);
for ch in value.chars() {
if ch == '\\' || ch == '%' || ch == '_' {
escaped.push('\\');
}
escaped.push(ch);
}
escaped
})
}
fn list_object_relations_sql(include_timestamps: bool) -> &'static str {
if include_timestamps {
return "SELECT c.relname AS object_name, \
CASE c.relkind \
WHEN 'v' THEN 'VIEW' \
WHEN 'm' THEN 'MATERIALIZED_VIEW' \
WHEN 'S' THEN 'SEQUENCE' \
ELSE 'TABLE' \
END AS object_type, \
obj_description(c.oid) AS object_comment, \
stat.creation::text AS created_at, \
COALESCE( \
CASE WHEN current_setting('track_commit_timestamp', true) = 'on' \
THEN pg_xact_commit_timestamp(c.xmin)::text END, \
stat.modification::text \
) AS updated_at, \
CASE WHEN pc.relkind = 'p' THEN pn.nspname ELSE NULL END AS parent_schema, \
CASE WHEN pc.relkind = 'p' THEN pc.relname ELSE NULL END AS parent_name, \
NULL::text AS signature, \
CASE c.relkind WHEN 'v' THEN 1 WHEN 'm' THEN 1 WHEN 'S' THEN 4 ELSE 0 END AS sort_order \
FROM pg_catalog.pg_class c \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
LEFT JOIN pg_catalog.pg_inherits i ON i.inhrelid = c.oid \
LEFT JOIN pg_catalog.pg_class pc ON pc.oid = i.inhparent \
LEFT JOIN pg_catalog.pg_namespace pn ON pn.oid = pc.relnamespace \
LEFT JOIN LATERAL pg_stat_file( \
CASE WHEN c.relkind IN ('r','m','f','p') THEN pg_relation_filepath(c.oid) END, true \
) stat ON true \
WHERE n.nspname = $1 AND c.relkind IN ('r','v','m','f','p','S')";
}
"SELECT c.relname AS object_name, \
CASE c.relkind \
WHEN 'v' THEN 'VIEW' \
WHEN 'm' THEN 'MATERIALIZED_VIEW' \
WHEN 'S' THEN 'SEQUENCE' \
ELSE 'TABLE' \
END AS object_type, \
obj_description(c.oid) AS object_comment, \
NULL::text AS created_at, \
NULL::text AS updated_at, \
CASE WHEN pc.relkind = 'p' THEN pn.nspname ELSE NULL END AS parent_schema, \
CASE WHEN pc.relkind = 'p' THEN pc.relname ELSE NULL END AS parent_name, \
NULL::text AS signature, \
CASE c.relkind WHEN 'v' THEN 1 WHEN 'm' THEN 1 WHEN 'S' THEN 4 ELSE 0 END AS sort_order \
FROM pg_catalog.pg_class c \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
LEFT JOIN pg_catalog.pg_inherits i ON i.inhrelid = c.oid \
LEFT JOIN pg_catalog.pg_class pc ON pc.oid = i.inhparent \
LEFT JOIN pg_catalog.pg_namespace pn ON pn.oid = pc.relnamespace \
WHERE n.nspname = $1 AND c.relkind IN ('r','v','m','f','p','S')"
}
fn list_object_routines_sql(include_timestamps: bool, has_proc_prokind: bool, has_proc_prosp: bool) -> &'static str {
if has_proc_prokind && has_proc_prosp {
if include_timestamps {
return "SELECT p.proname AS object_name, \
CASE WHEN p.prokind = 'p' OR p.prosp THEN 'PROCEDURE' ELSE 'FUNCTION' END AS object_type, \
obj_description(p.oid) AS object_comment, \
NULL::text AS created_at, \
CASE WHEN current_setting('track_commit_timestamp', true) = 'on' \
THEN pg_xact_commit_timestamp(p.xmin)::text END AS updated_at, \
NULL::text AS parent_schema, \
NULL::text AS parent_name, \
pg_get_function_identity_arguments(p.oid) AS signature, \
CASE WHEN p.prokind = 'p' OR p.prosp THEN 2 ELSE 3 END AS sort_order \
FROM pg_catalog.pg_proc p \
JOIN pg_catalog.pg_namespace n ON n.oid = p.pronamespace \
WHERE n.nspname = $1 AND (p.prokind IN ('p','f') OR p.prosp)";
}
return "SELECT p.proname AS object_name, \
CASE WHEN p.prokind = 'p' OR p.prosp THEN 'PROCEDURE' ELSE 'FUNCTION' END AS object_type, \
obj_description(p.oid) AS object_comment, \
NULL::text AS created_at, \
NULL::text AS updated_at, \
NULL::text AS parent_schema, \
NULL::text AS parent_name, \
pg_get_function_identity_arguments(p.oid) AS signature, \
CASE WHEN p.prokind = 'p' OR p.prosp THEN 2 ELSE 3 END AS sort_order \
FROM pg_catalog.pg_proc p \
JOIN pg_catalog.pg_namespace n ON n.oid = p.pronamespace \
WHERE n.nspname = $1 AND (p.prokind IN ('p','f') OR p.prosp)";
}
if has_proc_prokind {
if include_timestamps {
return "SELECT p.proname AS object_name, \
CASE p.prokind WHEN 'p' THEN 'PROCEDURE' ELSE 'FUNCTION' END AS object_type, \
obj_description(p.oid) AS object_comment, \
NULL::text AS created_at, \
CASE WHEN current_setting('track_commit_timestamp', true) = 'on' \
THEN pg_xact_commit_timestamp(p.xmin)::text END AS updated_at, \
NULL::text AS parent_schema, \
NULL::text AS parent_name, \
pg_get_function_identity_arguments(p.oid) AS signature, \
CASE p.prokind WHEN 'p' THEN 2 ELSE 3 END AS sort_order \
FROM pg_catalog.pg_proc p \
JOIN pg_catalog.pg_namespace n ON n.oid = p.pronamespace \
WHERE n.nspname = $1 AND p.prokind IN ('p','f')";
}
return "SELECT p.proname AS object_name, \
CASE p.prokind WHEN 'p' THEN 'PROCEDURE' ELSE 'FUNCTION' END AS object_type, \
obj_description(p.oid) AS object_comment, \
NULL::text AS created_at, \
NULL::text AS updated_at, \
NULL::text AS parent_schema, \
NULL::text AS parent_name, \
pg_get_function_identity_arguments(p.oid) AS signature, \
CASE p.prokind WHEN 'p' THEN 2 ELSE 3 END AS sort_order \
FROM pg_catalog.pg_proc p \
JOIN pg_catalog.pg_namespace n ON n.oid = p.pronamespace \
WHERE n.nspname = $1 AND p.prokind IN ('p','f')";
}
if has_proc_prosp {
if include_timestamps {
return "SELECT p.proname AS object_name, \
CASE WHEN p.prosp THEN 'PROCEDURE' ELSE 'FUNCTION' END AS object_type, \
obj_description(p.oid) AS object_comment, \
NULL::text AS created_at, \
CASE WHEN current_setting('track_commit_timestamp', true) = 'on' \
THEN pg_xact_commit_timestamp(p.xmin)::text END AS updated_at, \
NULL::text AS parent_schema, \
NULL::text AS parent_name, \
pg_get_function_identity_arguments(p.oid) AS signature, \
CASE WHEN p.prosp THEN 2 ELSE 3 END AS sort_order \
FROM pg_catalog.pg_proc p \
JOIN pg_catalog.pg_namespace n ON n.oid = p.pronamespace \
WHERE n.nspname = $1 AND NOT p.proisagg AND NOT p.proiswindow";
}
return "SELECT p.proname AS object_name, \
CASE WHEN p.prosp THEN 'PROCEDURE' ELSE 'FUNCTION' END AS object_type, \
obj_description(p.oid) AS object_comment, \
NULL::text AS created_at, \
NULL::text AS updated_at, \
NULL::text AS parent_schema, \
NULL::text AS parent_name, \
pg_get_function_identity_arguments(p.oid) AS signature, \
CASE WHEN p.prosp THEN 2 ELSE 3 END AS sort_order \
FROM pg_catalog.pg_proc p \
JOIN pg_catalog.pg_namespace n ON n.oid = p.pronamespace \
WHERE n.nspname = $1 AND NOT p.proisagg AND NOT p.proiswindow";
}
if include_timestamps {
return "SELECT p.proname AS object_name, \
'FUNCTION' AS object_type, \
obj_description(p.oid) AS object_comment, \
NULL::text AS created_at, \
CASE WHEN current_setting('track_commit_timestamp', true) = 'on' \
THEN pg_xact_commit_timestamp(p.xmin)::text END AS updated_at, \
NULL::text AS parent_schema, \
NULL::text AS parent_name, \
pg_get_function_identity_arguments(p.oid) AS signature, \
3 AS sort_order \
FROM pg_catalog.pg_proc p \
JOIN pg_catalog.pg_namespace n ON n.oid = p.pronamespace \
WHERE n.nspname = $1 AND NOT p.proisagg AND NOT p.proiswindow";
}
"SELECT p.proname AS object_name, \
'FUNCTION' AS object_type, \
obj_description(p.oid) AS object_comment, \
NULL::text AS created_at, \
NULL::text AS updated_at, \
NULL::text AS parent_schema, \
NULL::text AS parent_name, \
pg_get_function_identity_arguments(p.oid) AS signature, \
3 AS sort_order \
FROM pg_catalog.pg_proc p \
JOIN pg_catalog.pg_namespace n ON n.oid = p.pronamespace \
WHERE n.nspname = $1 AND NOT p.proisagg AND NOT p.proiswindow"
}
fn list_objects_sql(
include_timestamps: bool,
has_proc_prokind: bool,
has_proc_prosp: bool,
has_function_identity_arguments: bool,
) -> String {
let sql = format!(
"{} UNION ALL {} ORDER BY sort_order, object_name",
list_object_relations_sql(include_timestamps),
list_object_routines_sql(include_timestamps, has_proc_prokind, has_proc_prosp)
);
if has_function_identity_arguments {
sql
} else {
// Redshift and older PostgreSQL-compatible servers may only expose the
// older formatter. It includes argument names but still distinguishes
// overloads instead of making the whole schema browser unavailable.
sql.replace("pg_get_function_identity_arguments(p.oid)", "pg_get_function_arguments(p.oid)")
}
}
fn postgres_has_function_identity_arguments_sql() -> &'static str {
"SELECT EXISTS ( \
SELECT 1 \
FROM pg_catalog.pg_proc p \
JOIN pg_catalog.pg_namespace n ON n.oid = p.pronamespace \
WHERE n.nspname = 'pg_catalog' \
AND p.proname = 'pg_get_function_identity_arguments' \
)"
}
async fn postgres_has_function_identity_arguments(client: &deadpool_postgres::Client) -> Result<bool, String> {
let row = postgres_query_one_cached(client, postgres_has_function_identity_arguments_sql(), &[])
.await
.map_err(|e| e.to_string())?;
Ok(pg_row_try_bool(&row, 0).unwrap_or(false))
}
fn postgres_proc_has_prokind_sql() -> &'static str {
"SELECT EXISTS ( \
SELECT 1 \
FROM pg_catalog.pg_attribute \
WHERE attrelid = 'pg_catalog.pg_proc'::regclass \
AND attname = 'prokind' \
AND NOT attisdropped \
)"
}
async fn postgres_proc_has_prokind(client: &deadpool_postgres::Client) -> Result<bool, String> {
let row =
postgres_query_one_cached(client, postgres_proc_has_prokind_sql(), &[]).await.map_err(|e| e.to_string())?;
Ok(pg_row_try_bool(&row, 0).unwrap_or(false))
}
fn postgres_proc_has_prosp_sql() -> &'static str {
"SELECT EXISTS ( \
SELECT 1 \
FROM pg_catalog.pg_attribute \
WHERE attrelid = 'pg_catalog.pg_proc'::regclass \
AND attname = 'prosp' \
AND NOT attisdropped \
)"
}
async fn postgres_proc_has_prosp(client: &deadpool_postgres::Client) -> Result<bool, String> {
let row = postgres_query_one_cached(client, postgres_proc_has_prosp_sql(), &[]).await.map_err(|e| e.to_string())?;
Ok(pg_row_try_bool(&row, 0).unwrap_or(false))
}
async fn list_objects_rows(
client: &deadpool_postgres::Client,
schema: &str,
include_timestamps: bool,
has_proc_prokind: bool,
has_proc_prosp: bool,
has_function_identity_arguments: bool,
) -> Result<Vec<Row>, String> {
let sql = list_objects_sql(include_timestamps, has_proc_prokind, has_proc_prosp, has_function_identity_arguments);
postgres_query_cached(client, &sql, &[&schema]).await.map_err(|e| e.to_string())
}
pub async fn list_objects(pool: &Pool, schema: &str) -> Result<Vec<ObjectInfo>, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let has_proc_prokind = postgres_proc_has_prokind(&client).await?;
// Some GaussDB-compatible catalogs expose prosp alongside, or instead of,
// PostgreSQL 11's prokind. Treat prosp as an extra procedure signal.
let has_proc_prosp = postgres_proc_has_prosp(&client).await?;
let has_function_identity_arguments = postgres_has_function_identity_arguments(&client).await?;
let rows = match list_objects_rows(
&client,
schema,
true,
has_proc_prokind,
has_proc_prosp,
has_function_identity_arguments,
)
.await
{
Ok(rows) => rows,
Err(primary_error) => {
log::debug!("[postgres][list_objects:timestamp-fallback] primary_error={}", primary_error);
match list_objects_rows(
&client,
schema,
false,
has_proc_prokind,
has_proc_prosp,
has_function_identity_arguments,
)
.await
{
Ok(rows) => rows,
Err(fallback_error) => {
return Err(format!("{primary_error}; timestamp fallback failed: {fallback_error}"));
}
}
}
};
Ok(rows
.iter()
.map(|row| ObjectInfo {
name: pg_row_try_string(row, 0),
object_type: pg_row_try_string(row, 1),
schema: Some(schema.to_string()),
valid: None,
comment: row.try_get::<_, Option<String>>(2).ok().flatten().filter(|s| !s.is_empty()),
created_at: row.try_get::<_, Option<String>>(3).ok().flatten().filter(|s| !s.is_empty()),
updated_at: row.try_get::<_, Option<String>>(4).ok().flatten().filter(|s| !s.is_empty()),
parent_schema: row.try_get::<_, Option<String>>(5).ok().flatten().filter(|s| !s.is_empty()),
parent_name: row.try_get::<_, Option<String>>(6).ok().flatten().filter(|s| !s.is_empty()),
signature: row.try_get::<_, Option<String>>(7).ok().flatten(),
})
.collect())
}
pub async fn list_object_statistics(pool: &Pool, schema: &str) -> Result<Vec<ObjectStatistics>, String> {
let schema = if schema.is_empty() { "public" } else { schema };
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let rows = postgres_query_cached(
&client,
"SELECT c.relname, \
GREATEST(c.reltuples, 0)::bigint AS estimated_rows, \
pg_catalog.pg_total_relation_size(c.oid)::bigint AS total_bytes \
FROM pg_catalog.pg_class c \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
WHERE n.nspname = $1 AND c.relkind IN ('r','m','f','p') \
ORDER BY c.relname",
&[&schema],
)
.await
.map_err(|e| e.to_string())?;
Ok(rows
.iter()
.map(|row| ObjectStatistics {
name: pg_row_try_string(row, 0),
schema: Some(schema.to_string()),
estimated_rows: row.try_get::<_, i64>(1).ok(),
total_bytes: row.try_get::<_, i64>(2).ok(),
})
.collect())
}
pub async fn list_schemas(pool: &Pool) -> Result<Vec<String>, String> {
list_schemas_with_system(pool, false).await
}
pub async fn list_schema_infos(pool: &Pool) -> Result<Vec<SchemaInfo>, String> {
list_schema_infos_with_system(pool, false).await
}
pub async fn list_schemas_with_system(pool: &Pool, show_system_schemas: bool) -> Result<Vec<String>, String> {
Ok(list_schema_infos_with_system(pool, show_system_schemas).await?.into_iter().map(|schema| schema.name).collect())
}
const POSTGRES_SCHEMA_INFOS_SQL: &str = "SELECT n.nspname AS schema_name, d.description AS schema_comment \
FROM pg_catalog.pg_namespace n \
LEFT JOIN pg_catalog.pg_description d \
ON d.objoid = n.oid \
AND d.objsubid = 0 \
AND d.classoid = 'pg_namespace'::regclass \
ORDER BY n.nspname";
const POSTGRES_SCHEMA_INFOS_HIDE_SYSTEM_SQL: &str = "SELECT n.nspname AS schema_name, d.description AS schema_comment \
FROM pg_catalog.pg_namespace n \
LEFT JOIN pg_catalog.pg_description d \
ON d.objoid = n.oid \
AND d.objsubid = 0 \
AND d.classoid = 'pg_namespace'::regclass \
WHERE n.nspname NOT IN ('information_schema', 'pg_catalog', 'pg_toast') \
AND n.nspname NOT LIKE 'pg_toast_temp_%' \
AND n.nspname NOT LIKE 'pg_temp_%' \
ORDER BY n.nspname";
fn postgres_schema_infos_sql(show_system_schemas: bool) -> &'static str {
if show_system_schemas {
POSTGRES_SCHEMA_INFOS_SQL
} else {
POSTGRES_SCHEMA_INFOS_HIDE_SYSTEM_SQL
}
}
pub async fn list_schema_infos_with_system(pool: &Pool, show_system_schemas: bool) -> Result<Vec<SchemaInfo>, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let rows = postgres_query_cached(&client, postgres_schema_infos_sql(show_system_schemas), &[])
.await
.map_err(|e| e.to_string())?;
Ok(rows
.iter()
.map(|row| SchemaInfo {
name: pg_row_try_string(row, 0),
comment: row.try_get::<_, Option<String>>(1).ok().flatten(),
})
.collect())
}
const POSTGRES_COLUMNS_SQL: &str = "SELECT a.attname AS column_name, \
format_type(a.atttypid, a.atttypmod) AS full_type, \
COALESCE(c.is_nullable = 'YES', NOT a.attnotnull) AS is_nullable, \
CASE WHEN a.attgenerated <> '' THEN NULL ELSE pg_get_expr(ad.adbin, ad.adrelid) END AS column_default, \
EXISTS ( \
SELECT 1 FROM pg_constraint co \
JOIN pg_index i ON i.indrelid = co.conrelid AND co.conindid = i.indexrelid \
WHERE co.conrelid = a.attrelid AND co.contype = 'p' \
AND a.attnum = ANY(i.indkey) \
) AS is_pk, \
col_description(a.attrelid, a.attnum) AS column_comment, \
CASE a.attidentity \
WHEN 'd' THEN 'generated by default as identity' || CASE WHEN pseq.seqstart IS NOT NULL THEN format(' (start with %s increment by %s)', pseq.seqstart, pseq.seqincrement) ELSE '' END \
WHEN 'a' THEN 'generated always as identity' || CASE WHEN pseq.seqstart IS NOT NULL THEN format(' (start with %s increment by %s)', pseq.seqstart, pseq.seqincrement) ELSE '' END \
ELSE CASE a.attgenerated \
WHEN 's' THEN 'generated always as (' || pg_get_expr(ad.adbin, ad.adrelid) || ') stored' \
WHEN 'v' THEN 'generated always as (' || pg_get_expr(ad.adbin, ad.adrelid) || ') virtual' \
ELSE NULL \
END \
END AS column_extra, \
CASE WHEN t.typname = 'numeric' AND a.atttypmod > 0 \
THEN ((a.atttypmod - 4) >> 16) & 65535 ELSE NULL END AS numeric_precision, \
CASE WHEN t.typname = 'numeric' AND a.atttypmod > 0 \
THEN (a.atttypmod - 4) & 65535 ELSE NULL END AS numeric_scale, \
CASE WHEN t.typname IN ('varchar', 'bpchar') AND a.atttypmod > 0 \
THEN a.atttypmod - 4 ELSE NULL END AS character_maximum_length, \
CASE WHEN enum_t.oid IS NULL THEN NULL \
ELSE COALESCE((SELECT array_to_json(array_agg(e.enumlabel ORDER BY e.enumsortorder))::text \
FROM pg_enum e WHERE e.enumtypid = enum_t.oid), '[]') END AS enum_values \
FROM pg_attribute a \
JOIN pg_type t ON t.oid = a.atttypid \
LEFT JOIN pg_type enum_t ON enum_t.oid = CASE WHEN t.typtype = 'd' THEN t.typbasetype WHEN t.typtype = 'e' THEN t.oid ELSE NULL END AND enum_t.typtype = 'e' \
LEFT JOIN pg_attrdef ad ON ad.adrelid = a.attrelid AND ad.adnum = a.attnum \
LEFT JOIN pg_depend dep ON dep.refobjid = a.attrelid AND dep.refobjsubid = a.attnum AND dep.deptype = 'i' \
LEFT JOIN pg_sequence pseq ON pseq.seqrelid = dep.objid \
LEFT JOIN information_schema.columns c \
ON c.table_schema = $1 AND c.table_name = $2 AND c.column_name = a.attname \
WHERE a.attrelid = (quote_ident($1) || '.' || quote_ident($2))::regclass \
AND a.attnum > 0 AND NOT a.attisdropped \
ORDER BY a.attnum";
const POSTGRES_COLUMNS_COMPAT_SQL: &str = "SELECT a.attname AS column_name, \
format_type(a.atttypid, a.atttypmod) AS full_type, \
COALESCE(c.is_nullable = 'YES', NOT a.attnotnull) AS is_nullable, \
pg_get_expr(ad.adbin, ad.adrelid) AS column_default, \
EXISTS ( \
SELECT 1 FROM pg_constraint co \
JOIN pg_index i ON i.indrelid = co.conrelid AND co.conindid = i.indexrelid \
WHERE co.conrelid = a.attrelid AND co.contype = 'p' \
AND a.attnum = ANY(i.indkey) \
) AS is_pk, \
col_description(a.attrelid, a.attnum) AS column_comment, \
NULL::text AS column_extra, \
CASE WHEN t.typname = 'numeric' AND a.atttypmod > 0 \
THEN ((a.atttypmod - 4) >> 16) & 65535 ELSE NULL END AS numeric_precision, \
CASE WHEN t.typname = 'numeric' AND a.atttypmod > 0 \
THEN (a.atttypmod - 4) & 65535 ELSE NULL END AS numeric_scale, \
CASE WHEN t.typname IN ('varchar', 'bpchar') AND a.atttypmod > 0 \
THEN a.atttypmod - 4 ELSE NULL END AS character_maximum_length, \
NULL::text AS enum_values \
FROM pg_attribute a \
JOIN pg_type t ON t.oid = a.atttypid \
LEFT JOIN pg_attrdef ad ON ad.adrelid = a.attrelid AND ad.adnum = a.attnum \
LEFT JOIN information_schema.columns c \
ON c.table_schema = $1 AND c.table_name = $2 AND c.column_name = a.attname \
WHERE a.attrelid = (quote_ident($1) || '.' || quote_ident($2))::regclass \
AND a.attnum > 0 AND NOT a.attisdropped \
ORDER BY a.attnum";
const POSTGRES_COLUMNS_INFORMATION_SCHEMA_SQL: &str = "SELECT c.column_name, \
CASE WHEN c.data_type = 'USER-DEFINED' THEN c.udt_name ELSE c.data_type END AS full_type, \
c.is_nullable = 'YES' AS is_nullable, \
c.column_default, \
EXISTS ( \
SELECT 1 FROM information_schema.table_constraints tc \
JOIN information_schema.key_column_usage kcu \
ON kcu.constraint_catalog = tc.constraint_catalog \
AND kcu.constraint_schema = tc.constraint_schema \
AND kcu.constraint_name = tc.constraint_name \
AND kcu.table_schema = tc.table_schema \
AND kcu.table_name = tc.table_name \
WHERE tc.constraint_type = 'PRIMARY KEY' \
AND tc.table_schema = c.table_schema \
AND tc.table_name = c.table_name \
AND kcu.column_name = c.column_name \
) AS is_pk, \
NULL::text AS column_comment, \
NULL::text AS column_extra, \
CAST(c.numeric_precision AS int) AS numeric_precision, \
CAST(c.numeric_scale AS int) AS numeric_scale, \
CAST(c.character_maximum_length AS int) AS character_maximum_length, \
NULL::text AS enum_values \
FROM information_schema.columns c \
WHERE c.table_schema = $1 AND c.table_name = $2 \
ORDER BY c.ordinal_position";
fn parse_enum_values_from_row(row: &Row, index: usize) -> Option<Vec<String>> {
let raw = row.try_get::<_, Option<String>>(index).ok().flatten()?;
serde_json::from_str::<Vec<String>>(&raw).ok()
}
/// Decode a boolean column to JSON, tolerating databases (e.g. GaussDB) that
/// encode booleans as the ASCII bytes `t` (0x74) / `f` (0x66) in the binary
/// protocol instead of the standard PostgreSQL 0x00 / 0x01.
fn pg_bool_value_to_json(row: &Row, idx: usize) -> serde_json::Value {
if let Some(v) = pg_row_try_bool(row, idx) {
return serde_json::Value::Bool(v);
}
serde_json::Value::Null
}
/// Map raw boolean bytes to a Rust `bool`.
///
/// Standard PostgreSQL binary uses `[0x00]` / `[0x01]`; GaussDB sends the ASCII
/// text representation `[b't']` / `[b'f']` instead.
fn decode_bool_bytes(raw: &[u8]) -> Option<bool> {
match raw {
[0x00] => Some(false),
[0x01] => Some(true),
[b't'] | [b'T'] => Some(true),
[b'f'] | [b'F'] => Some(false),
_ => None,
}
}
fn decode_bool_candidates(raw: Option<&[u8]>, standard: Option<bool>) -> Option<bool> {
raw.and_then(decode_bool_bytes).or(standard)
}
/// Read a boolean column from a PostgreSQL row, tolerating databases that
/// encode booleans as integers (0/1) or text ('t'/'f') instead of the standard
/// `bool` OID. Returns `None` when the column is NULL or truly unreadable.
fn pg_row_try_bool(row: &Row, idx: usize) -> Option<bool> {
// GaussDB encodes boolean as ASCII 't' (0x74) / 'f' (0x66) in binary.
let raw = row.try_get::<_, PgRawBytes>(idx).ok();
let standard = row.try_get::<_, bool>(idx).ok();
if let Some(v) = decode_bool_candidates(raw.as_ref().map(|value| value.0.as_slice()), standard) {
return Some(v);
}
if let Ok(v) = row.try_get::<_, i32>(idx) {
return Some(v != 0);
}
if let Ok(v) = row.try_get::<_, i16>(idx) {
return Some(v != 0);
}
if let Ok(Some(v)) = row.try_get::<_, Option<String>>(idx) {
match v.as_str() {
"t" | "true" | "1" | "yes" | "YES" => return Some(true),
"f" | "false" | "0" | "no" | "NO" => return Some(false),
_ => return None,
}
}
None
}
/// Read a String column from a PostgreSQL row, tolerating databases that
/// return text as other types. Falls back to i64/i32/i16/bool formatting.
fn pg_row_try_string(row: &Row, idx: usize) -> String {
if let Ok(v) = row.try_get::<_, String>(idx) {
return v;
}
if let Ok(v) = row.try_get::<_, i64>(idx) {
return v.to_string();
}
if let Ok(v) = row.try_get::<_, i32>(idx) {
return v.to_string();
}
if let Ok(v) = row.try_get::<_, i16>(idx) {
return v.to_string();
}
if let Some(v) = pg_row_try_bool(row, idx) {
return v.to_string();
}
String::new()
}
fn column_info_from_row(row: &Row) -> ColumnInfo {
let full_type = row.try_get::<_, Option<String>>(1).ok().flatten().unwrap_or_default();
ColumnInfo {
name: pg_row_try_string(row, 0),
data_type: full_type,
is_nullable: pg_row_try_bool(row, 2).unwrap_or(true),
column_default: row.try_get::<_, Option<String>>(3).ok().flatten(),
is_primary_key: pg_row_try_bool(row, 4).unwrap_or(false),
extra: row.try_get::<_, Option<String>>(6).ok().flatten(),
comment: row.try_get::<_, Option<String>>(5).ok().flatten(),
numeric_precision: row.try_get::<_, Option<i32>>(7).ok().flatten(),
numeric_scale: row.try_get::<_, Option<i32>>(8).ok().flatten(),
character_maximum_length: row.try_get::<_, Option<i32>>(9).ok().flatten(),
enum_values: parse_enum_values_from_row(row, 10),
..Default::default()
}
}
async fn get_columns_with_sql(
client: &deadpool_postgres::Client,
sql: &str,
schema: &str,
table: &str,
) -> Result<Vec<ColumnInfo>, tokio_postgres::Error> {
let rows = postgres_query_cached(client, sql, &[&schema, &table]).await?;
Ok(rows.iter().map(column_info_from_row).collect())
}
pub async fn get_columns(pool: &Pool, schema: &str, table: &str) -> Result<Vec<ColumnInfo>, String> {
let schema = if schema.is_empty() { "public" } else { schema };
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
match get_columns_with_sql(&client, POSTGRES_COLUMNS_SQL, schema, table).await {
Ok(columns) => Ok(columns),
Err(primary_error) => match get_columns_with_sql(&client, POSTGRES_COLUMNS_COMPAT_SQL, schema, table).await {
Ok(columns) => Ok(columns),
Err(fallback_error) => {
let primary_message = pg_error_to_string(primary_error);
let fallback_message = pg_error_to_string(fallback_error);
match get_columns_with_sql(&client, POSTGRES_COLUMNS_INFORMATION_SCHEMA_SQL, schema, table).await {
Ok(columns) => Ok(columns),
Err(information_schema_error) => {
let information_schema_message = pg_error_to_string(information_schema_error);
log::debug!(
"[postgres][get_columns:compat-failed] primary_error={} fallback_error={} information_schema_error={}",
primary_message,
fallback_message,
information_schema_message
);
Err(information_schema_message)
}
}
}
},
}
}
fn pg_quote_literal(value: &str) -> String {
format!("'{}'", value.replace('\'', "''"))
}
fn redshift_columns_sql(schema: &str, table: &str) -> String {
format!(
"SELECT c.column_name, \
CASE WHEN c.data_type = 'USER-DEFINED' THEN c.udt_name ELSE c.data_type END AS full_type, \
c.is_nullable, \
c.column_default, \
CAST(c.numeric_precision AS varchar) AS numeric_precision, \
CAST(c.numeric_scale AS varchar) AS numeric_scale, \
CAST(c.character_maximum_length AS varchar) AS character_maximum_length \
FROM information_schema.columns c \
WHERE c.table_schema = {} AND c.table_name = {} \
ORDER BY c.ordinal_position",
pg_quote_literal(schema),
pg_quote_literal(table)
)
}
fn query_result_text(row: &[serde_json::Value], index: usize) -> Option<String> {
row.get(index).and_then(|value| match value {
serde_json::Value::String(value) => Some(value.clone()),
serde_json::Value::Number(value) => Some(value.to_string()),
serde_json::Value::Bool(value) => Some(value.to_string()),
serde_json::Value::Null | serde_json::Value::Array(_) | serde_json::Value::Object(_) => None,
})
}
fn query_result_i32(row: &[serde_json::Value], index: usize) -> Option<i32> {
query_result_text(row, index)?.parse().ok()
}
fn redshift_columns_from_query_result(result: QueryResult) -> Vec<ColumnInfo> {
result
.rows
.into_iter()
.filter_map(|row| {
Some(ColumnInfo {
name: query_result_text(&row, 0)?,
data_type: query_result_text(&row, 1).unwrap_or_default(),
is_nullable: query_result_text(&row, 2).is_none_or(|value| value.eq_ignore_ascii_case("YES")),
column_default: query_result_text(&row, 3),
is_primary_key: false,
extra: None,
comment: None,
numeric_precision: query_result_i32(&row, 4),
numeric_scale: query_result_i32(&row, 5),
character_maximum_length: query_result_i32(&row, 6),
enum_values: None,
..Default::default()
})
})
.collect()
}
pub async fn get_redshift_columns(pool: &Pool, schema: &str, table: &str) -> Result<Vec<ColumnInfo>, String> {
let schema = if schema.is_empty() { "public" } else { schema };
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let result = execute_select_text(
&client,
&redshift_columns_sql(schema, table),
Instant::now(),
crate::query::MAX_ROWS,
None,
None,
)
.await?;
Ok(redshift_columns_from_query_result(result))
}
pub(crate) fn pg_quote_ident(ident: &str) -> String {
format!("\"{}\"", ident.replace('"', "\"\""))
}
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub(crate) enum PostgresSearchPathContext {
Query,
Transaction,
LocalTransaction,
LocalQueryTransaction,
}
pub(crate) fn postgres_set_search_path_sql(schema: &str, context: PostgresSearchPathContext) -> String {
let (scope, suffix) = match context {
// Ordinary queries and exports historically fall back to public for
// extensions and helper functions after checking the selected schema.
PostgresSearchPathContext::Query => ("", ", pg_catalog, public"),
PostgresSearchPathContext::Transaction => ("", ", pg_catalog"),
PostgresSearchPathContext::LocalTransaction => (" LOCAL", ", pg_catalog"),
PostgresSearchPathContext::LocalQueryTransaction => (" LOCAL", ", pg_catalog, public"),
};
// PostgreSQL otherwise searches pg_catalog before every explicit path item.
format!("SET{scope} search_path TO {}{suffix}", pg_quote_ident(schema))
}
fn postgres_set_single_schema_search_path_sql(schema: &str, context: PostgresSearchPathContext) -> String {
let scope = if context == PostgresSearchPathContext::LocalTransaction { " LOCAL" } else { "" };
format!("SET{scope} search_path TO {}", pg_quote_ident(schema))
}
fn postgres_requires_single_schema_search_path(error: &str) -> bool {
error.to_ascii_lowercase().contains("does not support search_path with multiple names")
}
fn postgres_single_schema_clients() -> &'static Mutex<HashMap<usize, Weak<deadpool_postgres::StatementCache>>> {
static CLIENTS: OnceLock<Mutex<HashMap<usize, Weak<deadpool_postgres::StatementCache>>>> = OnceLock::new();
CLIENTS.get_or_init(|| Mutex::new(HashMap::new()))
}
fn postgres_client_uses_single_schema_search_path(client: &deadpool_postgres::Client) -> bool {
let statement_cache = &client.statement_cache;
let key = Arc::as_ptr(statement_cache) as usize;
let mut clients = postgres_single_schema_clients().lock().unwrap_or_else(|poisoned| poisoned.into_inner());
match clients.get(&key).and_then(Weak::upgrade) {
Some(cached) if Arc::ptr_eq(&cached, statement_cache) => true,
_ => {
clients.remove(&key);
false
}
}
}
fn mark_postgres_client_single_schema_search_path(client: &deadpool_postgres::Client) {
let statement_cache = &client.statement_cache;
let key = Arc::as_ptr(statement_cache) as usize;
let mut clients = postgres_single_schema_clients().lock().unwrap_or_else(|poisoned| poisoned.into_inner());
clients.retain(|_, cached| cached.strong_count() > 0);
clients.insert(key, Arc::downgrade(statement_cache));
}
pub(crate) async fn set_postgres_search_path(
client: &deadpool_postgres::Client,
schema: &str,
context: PostgresSearchPathContext,
timeout_duration: Duration,
) -> Result<u64, String> {
if postgres_client_uses_single_schema_search_path(client) {
return execute_postgres_infra_statement(
client,
&postgres_set_single_schema_search_path_sql(schema, context),
timeout_duration,
"schema.set",
)
.await;
}
let primary_sql = postgres_set_search_path_sql(schema, context);
match execute_postgres_infra_statement(client, &primary_sql, timeout_duration, "schema.set").await {
Ok(affected) => Ok(affected),
Err(primary_error) if postgres_requires_single_schema_search_path(&primary_error) => {
mark_postgres_client_single_schema_search_path(client);
log::info!("[postgres][schema.set:single-schema-fallback] schema={schema}");
execute_postgres_infra_statement(
client,
&postgres_set_single_schema_search_path_sql(schema, context),
timeout_duration,
"schema.set",
)
.await
.map_err(|fallback_error| {
format!("{primary_error}; single-schema search_path fallback failed: {fallback_error}")
})
}
Err(error) => Err(error),
}
}
fn query_result_row_limit(max_rows: Option<usize>) -> usize {
max_rows.unwrap_or(crate::query::MAX_ROWS).max(1)
}
/// Returns whether PostgreSQL should execute this statement through the row
/// retrieval path. DML without `RETURNING` needs `execute` for its command
/// tag/affected-row count, while DML with `RETURNING` produces a result set.
pub(crate) fn postgres_statement_returns_rows(sql: &str) -> bool {
if starts_with_executable_sql_keyword(sql, &["SELECT", "SHOW", "EXPLAIN", "WITH", "TABLE"]) {
return true;
}
let Ok(statements) = Parser::parse_sql(&PostgreSqlDialect {}, sql) else {
return false;
};
let [statement] = statements.as_slice() else {
return false;
};
match statement {
Statement::Insert(insert) => insert.returning.is_some(),
Statement::Update(update) => update.returning.is_some(),
Statement::Delete(delete) => delete.returning.is_some(),
Statement::Merge(merge) => merge.output.is_some(),
_ => false,
}
}
pub async fn execute_query(pool: &Pool, sql: &str) -> Result<QueryResult, String> {
execute_query_with_max_rows(pool, sql, None).await
}
pub async fn execute_query_with_max_rows(
pool: &Pool,
sql: &str,
max_rows: Option<usize>,
) -> Result<QueryResult, String> {
let start = Instant::now();
let row_limit = query_result_row_limit(max_rows);
if postgres_statement_returns_rows(sql) {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
execute_select_query(&client, sql, start, row_limit).await
} else {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let affected = client.execute(sql, &[]).await.map_err(pg_error_to_string)?;
clear_postgres_caches_after_ddl(pool, Some(&client), sql);
Ok(QueryResult {
columns: vec![],
column_types: Vec::new(),
column_sortables: Vec::new(),
spatial_columns: vec![],
spatial_values: vec![],
rows: vec![],
affected_rows: affected,
execution_time_ms: start.elapsed().as_millis(),
truncated: false,
session_id: None,
has_more: false,
elasticsearch_raw_body: None,
messages: Vec::new(),
})
}
}
pub async fn execute_query_with_max_rows_and_cancel(
pool: &Pool,
sql: &str,
max_rows: Option<usize>,
cancel_token: Option<CancellationToken>,
budget: DbOperationBudget,
cancel_context: Option<PostgresCancelContext>,
prefer_text_protocol: bool,
) -> Result<QueryResult, String> {
let client = checkout_postgres_client(pool, cancel_token.as_ref(), budget.checkout_timeout).await?;
execute_postgres_user_query(
&client,
sql,
max_rows,
cancel_token,
budget.query_timeout,
budget.cancel_timeout,
cancel_context,
prefer_text_protocol,
)
.await
}
fn postgres_read_only_transaction_setup(schema: Option<&str>) -> Vec<(String, &'static str)> {
let mut statements = vec![("BEGIN READ ONLY".to_string(), "explain_analyze.begin")];
if let Some(schema) = schema.map(str::trim).filter(|schema| !schema.is_empty()) {
statements.push((
postgres_set_search_path_sql(schema, PostgresSearchPathContext::LocalQueryTransaction),
"explain_analyze.schema",
));
}
statements
}
fn postgres_read_only_transaction_cleanup_error(error: String) -> String {
format!("PostgreSQL read-only transaction cleanup failed: {error}")
}
fn merge_postgres_operation_and_rollback_result<T>(
operation_result: Result<T, String>,
rollback_result: Result<(), String>,
) -> Result<T, String> {
match (operation_result, rollback_result) {
(Ok(result), Ok(())) => Ok(result),
(Err(operation_error), Ok(())) => Err(operation_error),
(Ok(_), Err(rollback_error)) => Err(postgres_read_only_transaction_cleanup_error(rollback_error)),
(Err(operation_error), Err(rollback_error)) => {
Err(format!("{operation_error}; {}", postgres_read_only_transaction_cleanup_error(rollback_error)))
}
}
}
async fn run_postgres_operation_with_rollback<T, Operation, OperationFuture, Rollback, RollbackFuture>(
operation: Operation,
rollback: Rollback,
) -> Result<T, String>
where
Operation: FnOnce() -> OperationFuture,
OperationFuture: Future<Output = Result<T, String>>,
Rollback: FnOnce() -> RollbackFuture,
RollbackFuture: Future<Output = Result<(), String>>,
{
let operation_result = operation().await;
let rollback_result = rollback().await;
merge_postgres_operation_and_rollback_result(operation_result, rollback_result)
}
pub async fn execute_query_in_read_only_transaction_with_rollback(
pool: &Pool,
schema: Option<&str>,
sql: &str,
max_rows: Option<usize>,
cancel_token: Option<CancellationToken>,
budget: DbOperationBudget,
cancel_context: Option<PostgresCancelContext>,
) -> Result<QueryResult, String> {
let client = checkout_postgres_client(pool, cancel_token.as_ref(), budget.checkout_timeout).await?;
let setup = postgres_read_only_transaction_setup(schema);
run_postgres_operation_with_rollback(
|| async {
for (statement, stage) in setup {
execute_postgres_infra_statement(&client, &statement, budget.recycle_timeout, stage).await?;
}
execute_postgres_user_query(
&client,
sql,
max_rows,
cancel_token,
budget.query_timeout,
budget.cancel_timeout,
cancel_context,
false,
)
.await
},
|| async {
execute_postgres_infra_statement(&client, "ROLLBACK", budget.cleanup_timeout, "explain_analyze.rollback")
.await
.map(|_| ())
},
)
.await
}
pub async fn stream_select_query_with_cancel(
pool: &Pool,
schema: Option<&str>,
setup_sql: &[String],
sql: &str,
max_rows: Option<usize>,
cancel_token: Option<CancellationToken>,
budget: DbOperationBudget,
cancel_context: Option<PostgresCancelContext>,
on_item: impl FnMut(PostgresQueryStreamItem) -> Result<(), String>,
) -> Result<u64, String> {
let start = Instant::now();
let client = checkout_postgres_client(pool, cancel_token.as_ref(), budget.checkout_timeout).await?;
let mut on_item = on_item;
let row_limit = max_rows.map(|limit| limit.max(1));
let schema = schema.map(str::trim).filter(|schema| !schema.is_empty());
let schema_was_set = schema.is_some_and(|_| !is_transaction_recovery_statement(sql));
if let Some(schema) = schema.filter(|_| schema_was_set) {
// Match normal query execution: export may reference unqualified names
// in the active schema, so the streaming path must use the same search_path.
set_postgres_search_path(&client, schema, PostgresSearchPathContext::Query, budget.recycle_timeout).await?;
}
let setup_transaction_started = !setup_sql.is_empty();
if setup_transaction_started {
execute_postgres_infra_statement(&client, "BEGIN", budget.recycle_timeout, "export_setup.begin").await?;
}
let query_timeout = budget.query_timeout;
let timeout_error =
format!("Query timed out after {} seconds", query_timeout.map_or(0, |timeout| timeout.as_secs()));
let setup_result = async {
for setup_statement in setup_sql {
wait_postgres_query(
client.cancel_token(),
cancel_context.clone(),
cancel_token.clone(),
query_timeout,
budget.cancel_timeout,
async {
client.batch_execute(setup_statement).await.map_err(pg_error_to_string)?;
Ok(())
},
)
.await?;
}
Ok(())
}
.await;
let result = match setup_result {
Ok(()) => {
let pg_cancel_token = client.cancel_token();
let progress_clock = Arc::new(StreamProgressClock::new());
let progress_clock_for_stream = progress_clock.clone();
let mut on_stream_item = |item| {
on_item(item)?;
progress_clock_for_stream.mark();
Ok(())
};
let result = await_stream_with_progress_timeout(
stream_select_query_inner(&client, sql, row_limit, &mut on_stream_item),
query_timeout,
progress_clock,
cancel_token.as_ref(),
timeout_error.clone(),
)
.await;
if result.as_ref().is_err_and(|error| error == &timeout_error || error == crate::query::QUERY_CANCELED) {
cancel_postgres_query(pg_cancel_token, cancel_context.as_ref(), budget.cancel_timeout).await;
}
result
}
Err(error) => Err(error),
};
let result = if setup_transaction_started {
let rollback_result =
execute_postgres_infra_statement(&client, "ROLLBACK", budget.cleanup_timeout, "export_setup.rollback")
.await;
match (result, rollback_result) {
(Ok(rows), Ok(_)) => Ok(rows),
(Err(query_err), Ok(_)) => Err(query_err),
(Ok(_), Err(rollback_err)) => Err(rollback_err),
(Err(query_err), Err(rollback_err)) => Err(format!("{query_err}; {rollback_err}")),
}
} else {
result
};
if schema_was_set {
let reset_result = reset_postgres_search_path(&client, budget.cleanup_timeout, start).await;
match (result, reset_result) {
(Ok(rows), Ok(())) => Ok(rows),
(Err(query_err), Ok(())) => Err(query_err),
(Ok(_), Err(reset_err)) => Err(reset_err),
(Err(query_err), Err(reset_err)) => Err(format!("{query_err}; {reset_err}")),
}
} else {
result
}
}
pub async fn execute_query_with_schema(pool: &Pool, schema: &str, sql: &str) -> Result<QueryResult, String> {
execute_query_with_schema_and_max_rows(pool, schema, sql, None).await
}
pub async fn execute_query_with_schema_and_max_rows(
pool: &Pool,
schema: &str,
sql: &str,
max_rows: Option<usize>,
) -> Result<QueryResult, String> {
let start = Instant::now();
let checkout_start = Instant::now();
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
log::info!(
"[postgres][execute_with_schema:pool:done] elapsed_ms={} total_ms={} schema={}",
checkout_start.elapsed().as_millis(),
start.elapsed().as_millis(),
schema
);
if is_transaction_recovery_statement(sql) {
log::info!(
"[postgres][execute_with_schema:skip-search-path] total_ms={} reason=transaction-recovery",
start.elapsed().as_millis()
);
return execute_query_with_max_rows_inner(&client, sql, max_rows, false, None).await;
}
let set_schema_start = Instant::now();
set_postgres_search_path(&client, schema, PostgresSearchPathContext::Query, super::connection_timeout()).await?;
log::info!(
"[postgres][execute_with_schema:set-search-path:done] elapsed_ms={} total_ms={}",
set_schema_start.elapsed().as_millis(),
start.elapsed().as_millis()
);
let query_start = Instant::now();
let result = execute_query_with_max_rows_inner(&client, sql, max_rows, false, None).await;
if result.is_ok() {
clear_postgres_caches_after_ddl(pool, Some(&client), sql);
}
log::info!(
"[postgres][execute_with_schema:query:done] elapsed_ms={} total_ms={} ok={}",
query_start.elapsed().as_millis(),
start.elapsed().as_millis(),
result.is_ok()
);
let reset_result = reset_postgres_search_path(&client, super::connection_timeout(), start).await;
merge_postgres_query_and_reset_result(result, reset_result)
}
pub async fn execute_query_with_schema_and_max_rows_and_cancel(
pool: &Pool,
schema: &str,
sql: &str,
max_rows: Option<usize>,
cancel_token: Option<CancellationToken>,
budget: DbOperationBudget,
cancel_context: Option<PostgresCancelContext>,
prefer_text_protocol: bool,
) -> Result<QueryResult, String> {
let start = Instant::now();
let checkout_start = Instant::now();
let client = checkout_postgres_client(pool, cancel_token.as_ref(), budget.checkout_timeout).await?;
log::info!(
"[postgres][execute_with_schema:pool:done] elapsed_ms={} total_ms={} schema={}",
checkout_start.elapsed().as_millis(),
start.elapsed().as_millis(),
schema
);
if is_transaction_recovery_statement(sql) {
log::info!(
"[postgres][execute_with_schema:skip-search-path] total_ms={} reason=transaction-recovery",
start.elapsed().as_millis()
);
return execute_postgres_user_query(
&client,
sql,
max_rows,
cancel_token,
budget.query_timeout,
budget.cancel_timeout,
cancel_context,
prefer_text_protocol,
)
.await;
}
let set_schema_start = Instant::now();
set_postgres_search_path(&client, schema, PostgresSearchPathContext::Query, budget.recycle_timeout).await?;
log::info!(
"[postgres][execute_with_schema:set-search-path:done] elapsed_ms={} total_ms={}",
set_schema_start.elapsed().as_millis(),
start.elapsed().as_millis()
);
let query_start = Instant::now();
let result = execute_postgres_user_query(
&client,
sql,
max_rows,
cancel_token,
budget.query_timeout,
budget.cancel_timeout,
cancel_context,
prefer_text_protocol,
)
.await;
if result.is_ok() {
clear_postgres_caches_after_ddl(pool, Some(&client), sql);
}
log::info!(
"[postgres][execute_with_schema:query:done] elapsed_ms={} total_ms={} ok={}",
query_start.elapsed().as_millis(),
start.elapsed().as_millis(),
result.is_ok()
);
let reset_result = reset_postgres_search_path(&client, budget.cleanup_timeout, start).await;
merge_postgres_query_and_reset_result(result, reset_result)
}
async fn reset_postgres_search_path(
client: &deadpool_postgres::Client,
timeout_duration: Duration,
start: Instant,
) -> Result<(), String> {
let reset_start = Instant::now();
match execute_postgres_infra_statement(client, "RESET search_path", timeout_duration, "schema.reset").await {
Ok(_) => {
log::info!(
"[postgres][execute_with_schema:reset-search-path:done] elapsed_ms={} total_ms={}",
reset_start.elapsed().as_millis(),
start.elapsed().as_millis()
);
Ok(())
}
Err(err) => {
log::warn!(
"[postgres][execute_with_schema:reset-search-path:error] elapsed_ms={} total_ms={} error={}",
reset_start.elapsed().as_millis(),
start.elapsed().as_millis(),
err
);
Err(postgres_schema_reset_cleanup_error(err))
}
}
}
fn merge_postgres_query_and_reset_result(
query_result: Result<QueryResult, String>,
reset_result: Result<(), String>,
) -> Result<QueryResult, String> {
match (query_result, reset_result) {
(Ok(result), Ok(())) => Ok(result),
(Err(query_err), Ok(())) => Err(query_err),
(Ok(_), Err(reset_err)) => Err(reset_err),
(Err(query_err), Err(reset_err)) => Err(format!("{query_err}; {reset_err}")),
}
}
fn postgres_schema_reset_cleanup_error(err: String) -> String {
format!("PostgreSQL schema.reset cleanup failed: {err}")
}
pub(crate) async fn execute_postgres_infra_statement(
client: &deadpool_postgres::Client,
sql: &str,
timeout_duration: Duration,
stage: &str,
) -> Result<u64, String> {
tokio::time::timeout(timeout_duration, client.execute(sql, &[]))
.await
.map_err(|_| format!("PostgreSQL {stage} timed out after {} seconds", timeout_duration.as_secs()))?
.map_err(pg_error_to_string)
}
pub(crate) async fn wait_postgres_operation<T, F>(
pg_cancel_token: tokio_postgres::CancelToken,
cancel_context: Option<PostgresCancelContext>,
timeout_duration: Option<Duration>,
cancel_timeout: Duration,
future: F,
) -> Result<T, String>
where
F: Future<Output = Result<T, String>>,
{
wait_postgres_query(pg_cancel_token, cancel_context, None, timeout_duration, cancel_timeout, future).await
}
async fn wait_postgres_query<T, F>(
pg_cancel_token: tokio_postgres::CancelToken,
cancel_context: Option<PostgresCancelContext>,
cancel_token: Option<CancellationToken>,
timeout_duration: Option<Duration>,
cancel_timeout: Duration,
future: F,
) -> Result<T, String>
where
F: Future<Output = Result<T, String>>,
{
match (cancel_token, timeout_duration) {
(Some(token), Some(duration)) => {
tokio::select! {
biased;
_ = token.cancelled() => {
cancel_postgres_query(pg_cancel_token, cancel_context.as_ref(), cancel_timeout).await;
Err(crate::query::canceled_error())
}
result = tokio::time::timeout(duration, future) => match result {
Ok(result) => result,
Err(_) => {
cancel_postgres_query(pg_cancel_token, cancel_context.as_ref(), cancel_timeout).await;
Err(format!("Query timed out after {} seconds", duration.as_secs()))
}
},
}
}
(None, Some(duration)) => match tokio::time::timeout(duration, future).await {
Ok(result) => result,
Err(_) => {
cancel_postgres_query(pg_cancel_token, cancel_context.as_ref(), cancel_timeout).await;
Err(format!("Query timed out after {} seconds", duration.as_secs()))
}
},
(Some(token), None) => {
tokio::select! {
biased;
_ = token.cancelled() => {
cancel_postgres_query(pg_cancel_token, cancel_context.as_ref(), cancel_timeout).await;
Err(crate::query::canceled_error())
}
result = future => result,
}
}
(None, None) => future.await,
}
}
async fn execute_postgres_user_query(
client: &deadpool_postgres::Client,
sql: &str,
max_rows: Option<usize>,
cancel_token: Option<CancellationToken>,
timeout_duration: Option<Duration>,
cancel_timeout: Duration,
cancel_context: Option<PostgresCancelContext>,
prefer_text_protocol: bool,
) -> Result<QueryResult, String> {
let pg_cancel_token = client.cancel_token();
// Commands do not expose incremental results, so keep their timeout as a
// wall-clock deadline. Row-returning queries may spend longer transferring
// a bounded result through a slow tunnel; for those, the same setting is an
// inactivity budget that resets only after real PostgreSQL progress.
if !postgres_statement_returns_rows(sql) {
return wait_postgres_query(
pg_cancel_token,
cancel_context,
cancel_token,
timeout_duration,
cancel_timeout,
execute_query_with_max_rows_inner(client, sql, max_rows, prefer_text_protocol, None),
)
.await;
}
let timeout_error =
format!("Query timed out after {} seconds", timeout_duration.map_or(0, |timeout| timeout.as_secs()));
let progress_clock = Arc::new(StreamProgressClock::new());
let result = await_stream_with_progress_timeout(
execute_query_with_max_rows_inner(client, sql, max_rows, prefer_text_protocol, Some(progress_clock.clone())),
timeout_duration,
progress_clock,
cancel_token.as_ref(),
timeout_error.clone(),
)
.await;
if result.as_ref().is_err_and(|error| error == &timeout_error || error == crate::query::QUERY_CANCELED) {
cancel_postgres_query(pg_cancel_token, cancel_context.as_ref(), cancel_timeout).await;
}
result
}
/// PostgreSQL pool checkout with timeout and cancel token support.
/// When the checkout phase is stuck, the cancel token can terminate the wait early.
/// The timeout error message includes "checkout timed out" to ensure is_connection_error can classify it correctly.
pub async fn checkout_postgres_client(
pool: &Pool,
cancel_token: Option<&CancellationToken>,
checkout_timeout: Duration,
) -> Result<deadpool_postgres::Object, String> {
let checkout_timeout = effective_postgres_checkout_timeout(pool, checkout_timeout);
let start = Instant::now();
let get_future = async {
tokio::time::timeout(checkout_timeout, pool.get())
.await
.map_err(|_| {
let elapsed = start.elapsed().as_millis();
log::warn!(
"[db:pool.checkout:error] elapsed_ms={} timeout_ms={} error=checkout timed out",
elapsed,
checkout_timeout.as_millis()
);
format!("PostgreSQL connection pool checkout timed out ({}s)", checkout_timeout.as_secs())
})?
.map_err(|e| {
let elapsed = start.elapsed().as_millis();
let err = pg_pool_error_to_string(e);
log::warn!(
"[db:pool.checkout:error] elapsed_ms={} timeout_ms={} error={}",
elapsed,
checkout_timeout.as_millis(),
err
);
format!("PostgreSQL connection pool checkout failed: {err}")
})
};
let result = match cancel_token {
Some(token) => tokio::select! {
biased;
_ = token.cancelled() => {
log::info!(
"[db:pool.checkout:cancelled] elapsed_ms={} timeout_ms={}",
start.elapsed().as_millis(),
checkout_timeout.as_millis()
);
return Err(crate::query::canceled_error());
}
result = get_future => result,
},
None => get_future.await,
};
if let Ok(client) = &result {
log::debug!(
"[db:pool.checkout:done] elapsed_ms={} timeout_ms={}",
start.elapsed().as_millis(),
checkout_timeout.as_millis()
);
// Resolve the notice-attribution identity here, outside of any
// transaction: a lazy first lookup from `drain_postgres_notices`
// could run inside the read-only EXPLAIN transaction, where a
// failing identity query would abort the user's statement. Cached
// after the first checkout of each physical connection.
let _ = resolve_postgres_client_key(client).await;
}
result
}
fn effective_postgres_checkout_timeout(pool: &Pool, requested_timeout: Duration) -> Duration {
let configured = pool.timeouts();
[configured.wait, configured.create, configured.recycle]
.into_iter()
.flatten()
.fold(requested_timeout, std::cmp::max)
}
async fn cancel_postgres_query(
pg_cancel_token: tokio_postgres::CancelToken,
cancel_context: Option<&PostgresCancelContext>,
cancel_timeout: Duration,
) {
let cancel_timeout = postgres_cancel_attempt_timeout(cancel_timeout, cancel_context);
if let Some(ctx) = cancel_context {
match make_rustls_connect_from_context(ctx) {
Ok(tls) => match tokio::time::timeout(cancel_timeout, pg_cancel_token.cancel_query(tls)).await {
Ok(Ok(())) => return,
Ok(Err(err)) => {
log::warn!("Failed to send PostgreSQL TLS cancel request: {err}");
return;
}
Err(_) => {
log::warn!("Timed out sending PostgreSQL TLS cancel request ({}s)", cancel_timeout.as_secs());
return;
}
},
Err(err) => {
log::warn!("Failed to build TLS connector for cancel: {err}; falling back to NoTls cancel");
}
}
}
match tokio::time::timeout(cancel_timeout, pg_cancel_token.cancel_query(NoTls)).await {
Ok(Ok(())) => {}
Ok(Err(err)) => log::warn!("Failed to send PostgreSQL cancel request: {err}"),
Err(_) => log::warn!("Timed out sending PostgreSQL cancel request ({}s)", cancel_timeout.as_secs()),
}
}
fn postgres_cancel_attempt_timeout(
cancel_timeout: Duration,
_cancel_context: Option<&PostgresCancelContext>,
) -> Duration {
cancel_timeout
}
fn is_transaction_recovery_statement(sql: &str) -> bool {
starts_with_executable_sql_keyword(sql, &["ROLLBACK", "ABORT", "COMMIT", "END"])
}
async fn execute_query_with_max_rows_inner(
client: &deadpool_postgres::Client,
sql: &str,
max_rows: Option<usize>,
prefer_text_protocol: bool,
progress_clock: Option<Arc<StreamProgressClock>>,
) -> Result<QueryResult, String> {
let start = Instant::now();
let row_limit = query_result_row_limit(max_rows);
// Discard stale notices from infrastructure statements (e.g. the timezone
// SET issued at connect time) so only messages raised by this statement
// are attached to its result.
let _ = drain_postgres_notices(client).await;
let result = if postgres_statement_returns_rows(sql) {
if prefer_text_protocol {
execute_select_text(client, sql, start, row_limit, None, progress_clock.as_deref()).await
} else {
execute_select_query_with_progress(client, sql, start, row_limit, progress_clock.as_deref()).await
}
} else {
client.execute(sql, &[]).await.map_err(pg_error_to_string).map(|affected| QueryResult {
columns: vec![],
column_types: Vec::new(),
column_sortables: Vec::new(),
spatial_columns: vec![],
spatial_values: vec![],
rows: vec![],
affected_rows: affected,
execution_time_ms: start.elapsed().as_millis(),
truncated: false,
session_id: None,
has_more: false,
elasticsearch_raw_body: None,
messages: Vec::new(),
})
};
match result {
Ok(mut result) => {
result.messages = drain_postgres_notices(client).await;
Ok(result)
}
Err(error) => {
// Drop notices so an errored statement's messages cannot leak
// into the next query on this pooled connection.
let _ = drain_postgres_notices(client).await;
Err(error)
}
}
}
const POSTGRES_INDEXES_SQL: &str = "SELECT i.relname AS index_name, \
array_agg(COALESCE(a.attname, pg_get_indexdef(ix.indexrelid, k.n::int, true)) ORDER BY k.n) AS columns, \
ix.indisunique AS is_unique, \
ix.indisprimary AS is_primary, \
pg_get_expr(ix.indpred, ix.indrelid) AS filter_expr, \
am.amname AS index_type, \
ix.indnkeyatts AS nkeyatts, \
ix.indkey AS indkey, \
obj_description(i.oid, 'pg_class') AS index_comment \
FROM pg_index ix \
JOIN pg_class t ON t.oid = ix.indrelid \
JOIN pg_class i ON i.oid = ix.indexrelid \
JOIN pg_namespace n ON n.oid = t.relnamespace \
JOIN pg_am am ON am.oid = i.relam \
JOIN LATERAL unnest(ix.indkey) WITH ORDINALITY AS k(attnum, n) ON true \
LEFT JOIN pg_attribute a ON a.attrelid = t.oid AND a.attnum = k.attnum AND k.attnum > 0 \
WHERE n.nspname = $1 AND t.relname = $2 \
GROUP BY i.relname, i.oid, ix.indisunique, ix.indisprimary, ix.indpred, ix.indrelid, am.amname, ix.indnkeyatts, ix.indkey \
ORDER BY i.relname";
const POSTGRES_INDEXES_COMPAT_SQL: &str = "SELECT i.relname AS index_name, \
ARRAY( \
SELECT COALESCE(a.attname, pg_get_indexdef(ix.indexrelid, pos.n, true)) \
FROM generate_series(1, array_length(string_to_array(ix.indkey::text, ' '), 1)) AS pos(n) \
LEFT JOIN pg_attribute a \
ON a.attrelid = t.oid \
AND a.attnum = (string_to_array(ix.indkey::text, ' '))[pos.n]::int2 \
AND a.attnum > 0 \
ORDER BY pos.n \
) AS columns, \
ix.indisunique AS is_unique, \
ix.indisprimary AS is_primary, \
pg_get_expr(ix.indpred, ix.indrelid) AS filter_expr, \
am.amname AS index_type, \
NULL::smallint AS nkeyatts, \
ix.indkey AS indkey, \
obj_description(i.oid, 'pg_class') AS index_comment \
FROM pg_index ix \
JOIN pg_class t ON t.oid = ix.indrelid \
JOIN pg_class i ON i.oid = ix.indexrelid \
JOIN pg_namespace n ON n.oid = t.relnamespace \
JOIN pg_am am ON am.oid = i.relam \
WHERE n.nspname = $1 AND t.relname = $2 \
ORDER BY i.relname";
const POSTGRES_OWNERS_SQL: &str =
"SELECT n.nspname, c.relname, c.relkind::text AS relkind, pg_get_userbyid(c.relowner) \
FROM pg_class c \
JOIN pg_namespace n ON n.oid = c.relnamespace \
WHERE n.nspname = $1 \
AND c.relkind IN ('r', 'v', 'm', 'S', 'f', 'p')";
const POSTGRES_TABLE_OWNER_SQL: &str = "SELECT pg_get_userbyid(c.relowner)::text, \
ARRAY(SELECT default_acl.privilege_type::text \
FROM pg_catalog.aclexplode(pg_catalog.acldefault('r', c.relowner)) default_acl \
WHERE default_acl.grantee = c.relowner \
ORDER BY default_acl.privilege_type) \
FROM pg_catalog.pg_class c \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
WHERE n.nspname = $1 AND c.relname = $2 AND c.relkind IN ('r', 'p') \
ORDER BY c.oid LIMIT 1";
const POSTGRES_TABLE_ACL_PRIVILEGES_SQL: &str =
"SELECT CASE WHEN acl.grantee = 0 THEN 'PUBLIC' ELSE grantee.rolname END::text, \
acl.privilege_type::text, acl.is_grantable, pg_get_userbyid(acl.grantor)::text \
FROM pg_catalog.pg_class c \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
JOIN LATERAL pg_catalog.aclexplode(COALESCE(c.relacl, pg_catalog.acldefault('r', c.relowner))) acl ON true \
LEFT JOIN pg_catalog.pg_roles grantee ON grantee.oid = acl.grantee \
WHERE n.nspname = $1 AND c.relname = $2 AND c.relkind IN ('r', 'p') \
ORDER BY 4, 1, 2, 3";
const POSTGRES_COLUMN_ACL_PRIVILEGES_SQL: &str =
"SELECT CASE WHEN acl.grantee = 0 THEN 'PUBLIC' ELSE grantee.rolname END::text, \
acl.privilege_type::text, acl.is_grantable, a.attname::text, pg_get_userbyid(acl.grantor)::text \
FROM pg_catalog.pg_class c \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
JOIN pg_catalog.pg_attribute a ON a.attrelid = c.oid AND a.attnum > 0 AND NOT a.attisdropped \
JOIN LATERAL pg_catalog.aclexplode(a.attacl) acl ON true \
LEFT JOIN pg_catalog.pg_roles grantee ON grantee.oid = acl.grantee \
WHERE n.nspname = $1 AND c.relname = $2 AND c.relkind IN ('r', 'p') \
ORDER BY 5, 1, 2, 3, 4";
fn postgres_owner_object_type(relkind: &str) -> &str {
match relkind {
"r" => "TABLE",
"v" => "VIEW",
"m" => "MATERIALIZED_VIEW",
"S" => "SEQUENCE",
"f" => "FOREIGN TABLE",
"p" => "PARTITIONED TABLE",
"I" => "PARTITIONED INDEX",
_ => relkind,
}
}
async fn list_indexes_with_sql(
client: &deadpool_postgres::Client,
sql: &str,
schema: &str,
table: &str,
) -> Result<Vec<IndexInfo>, tokio_postgres::Error> {
let rows = postgres_query_cached(client, sql, &[&schema, &table]).await?;
Ok(rows
.iter()
.map(|row| {
let all_cols: Vec<String> = row.try_get::<_, Vec<String>>(1).unwrap_or_default();
let nkeyatts = row.try_get::<_, Option<i16>>(6).ok().flatten().unwrap_or(all_cols.len() as i16) as usize;
let split_at = nkeyatts.min(all_cols.len());
let key_cols = all_cols[..split_at].to_vec();
let included = if split_at < all_cols.len() { all_cols[split_at..].to_vec() } else { vec![] };
IndexInfo {
name: pg_row_try_string(row, 0),
columns: key_cols,
is_unique: pg_row_try_bool(row, 2).unwrap_or(false),
is_primary: pg_row_try_bool(row, 3).unwrap_or(false),
filter: row.try_get::<_, Option<String>>(4).ok().flatten(),
index_type: row.try_get::<_, Option<String>>(5).ok().flatten(),
included_columns: if included.is_empty() { None } else { Some(included) },
comment: row.try_get::<_, Option<String>>(8).ok().flatten(),
}
})
.collect())
}
pub async fn list_indexes(pool: &Pool, schema: &str, table: &str) -> Result<Vec<IndexInfo>, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
match list_indexes_with_sql(&client, POSTGRES_INDEXES_SQL, schema, table).await {
Ok(indexes) => Ok(indexes),
Err(primary_error) => match list_indexes_with_sql(&client, POSTGRES_INDEXES_COMPAT_SQL, schema, table).await {
Ok(indexes) => Ok(indexes),
Err(fallback_error) => {
let primary_message = pg_error_to_string(primary_error);
let fallback_message = pg_error_to_string(fallback_error);
log::debug!(
"[postgres][list_indexes:compat-failed] primary_error={} fallback_error={}",
primary_message,
fallback_message
);
Err(fallback_message)
}
},
}
}
fn postgres_foreign_keys_sql() -> &'static str {
"SELECT fk.constraint_name, fk.column_name, \
pk.table_schema AS ref_schema, pk.table_name AS ref_table, pk.column_name AS ref_column, \
rc.update_rule AS on_update, rc.delete_rule AS on_delete \
FROM information_schema.table_constraints tc \
JOIN information_schema.key_column_usage fk \
ON fk.constraint_name = tc.constraint_name \
AND fk.constraint_schema = tc.constraint_schema \
AND fk.table_schema = tc.table_schema \
AND fk.table_name = tc.table_name \
JOIN information_schema.referential_constraints rc \
ON rc.constraint_name = tc.constraint_name \
AND rc.constraint_schema = tc.constraint_schema \
JOIN information_schema.key_column_usage pk \
ON pk.constraint_name = rc.unique_constraint_name \
AND pk.constraint_schema = rc.unique_constraint_schema \
AND pk.ordinal_position = fk.position_in_unique_constraint \
WHERE tc.constraint_type = 'FOREIGN KEY' \
AND fk.table_schema = $1 AND fk.table_name = $2 \
ORDER BY fk.constraint_name, fk.ordinal_position"
}
fn postgres_foreign_key_action(value: String) -> Option<String> {
let value = value.trim();
if value.is_empty() {
None
} else {
Some(value.to_string())
}
}
pub async fn list_foreign_keys(pool: &Pool, schema: &str, table: &str) -> Result<Vec<ForeignKeyInfo>, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let rows = postgres_query_cached(&client, postgres_foreign_keys_sql(), &[&schema, &table])
.await
.map_err(|e| e.to_string())?;
Ok(rows
.iter()
.map(|row| ForeignKeyInfo {
name: pg_row_try_string(row, 0),
column: pg_row_try_string(row, 1),
ref_schema: Some(pg_row_try_string(row, 2)),
ref_table: pg_row_try_string(row, 3),
ref_column: pg_row_try_string(row, 4),
on_update: postgres_foreign_key_action(pg_row_try_string(row, 5)),
on_delete: postgres_foreign_key_action(pg_row_try_string(row, 6)),
})
.collect())
}
fn postgres_table_dependencies_sql() -> &'static str {
"SELECT DISTINCT child.relname AS table_name, parent.relname AS ref_table \
FROM pg_catalog.pg_constraint con \
JOIN pg_catalog.pg_class child ON child.oid = con.conrelid \
JOIN pg_catalog.pg_namespace child_schema ON child_schema.oid = child.relnamespace \
JOIN pg_catalog.pg_class parent ON parent.oid = con.confrelid \
JOIN pg_catalog.pg_namespace parent_schema ON parent_schema.oid = parent.relnamespace \
WHERE con.contype = 'f' \
AND child_schema.nspname = $1 \
AND parent_schema.nspname = $1 \
ORDER BY child.relname, parent.relname"
}
/// Fetch all same-schema table dependencies in one round trip. Whole-database
/// exports use this instead of issuing one information_schema query per table.
pub async fn list_table_dependencies(pool: &Pool, schema: &str) -> Result<Vec<(String, String)>, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let rows = postgres_query_cached(&client, postgres_table_dependencies_sql(), &[&schema])
.await
.map_err(|e| e.to_string())?;
Ok(rows.iter().map(|row| (pg_row_try_string(row, 0), pg_row_try_string(row, 1))).collect())
}
pub async fn list_triggers(pool: &Pool, schema: &str, table: &str) -> Result<Vec<TriggerInfo>, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let rows = postgres_query_cached(
&client,
"SELECT trigger_name, event_manipulation, action_timing \
FROM information_schema.triggers \
WHERE trigger_schema = $1 AND event_object_table = $2 \
ORDER BY trigger_name",
&[&schema, &table],
)
.await
.map_err(|e| e.to_string())?;
Ok(rows
.iter()
.map(|row| TriggerInfo {
name: pg_row_try_string(row, 0),
event: pg_row_try_string(row, 1),
timing: pg_row_try_string(row, 2),
statement: None,
})
.collect())
}
pub async fn list_trigger_definitions(pool: &Pool, schema: &str, table: &str) -> Result<Vec<String>, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let rows = postgres_query_cached(&client, postgres_trigger_definitions_sql(), &[&schema, &table])
.await
.map_err(|e| e.to_string())?;
Ok(rows.iter().map(|row| pg_row_try_string(row, 0)).filter(|definition| !definition.trim().is_empty()).collect())
}
fn postgres_trigger_definitions_sql() -> &'static str {
"SELECT pg_catalog.pg_get_triggerdef(t.oid, true) AS trigger_definition \
FROM pg_catalog.pg_trigger t \
JOIN pg_catalog.pg_class c ON c.oid = t.tgrelid \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
WHERE n.nspname = $1 AND c.relname = $2 AND NOT t.tgisinternal \
ORDER BY t.tgname, t.oid"
}
fn postgres_functions_sql(has_proc_prokind: bool) -> &'static str {
if has_proc_prokind {
return "SELECT p.proname, \
CASE p.prokind WHEN 'f' THEN 'FUNCTION' WHEN 'p' THEN 'PROCEDURE' ELSE 'FUNCTION' END, \
COALESCE(pg_get_function_result(p.oid), ''), \
pg_get_functiondef(p.oid), \
COALESCE(pg_get_function_arguments(p.oid), '') \
FROM pg_proc p \
JOIN pg_namespace n ON n.oid = p.pronamespace \
WHERE n.nspname = $1 AND p.prokind IN ('f', 'p') \
ORDER BY p.proname";
}
// PostgreSQL 10 and older do not have pg_proc.prokind; procedures were
// introduced with prokind, so the legacy path can only return functions.
"SELECT p.proname, \
'FUNCTION', \
COALESCE(pg_get_function_result(p.oid), ''), \
pg_get_functiondef(p.oid), \
COALESCE(pg_get_function_arguments(p.oid), '') \
FROM pg_proc p \
JOIN pg_namespace n ON n.oid = p.pronamespace \
WHERE n.nspname = $1 AND NOT p.proisagg AND NOT p.proiswindow \
ORDER BY p.proname"
}
pub async fn list_functions(pool: &Pool, schema: &str) -> Result<Vec<FunctionInfo>, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
// Use pg_proc + pg_get_functiondef() instead of information_schema.routines
// for reliable function definition retrieval (information_schema.routines.routine_definition
// is NULL for non-SQL functions like plpgsql)
let has_proc_prokind = postgres_proc_has_prokind(&client).await?;
let rows = postgres_query_cached(&client, postgres_functions_sql(has_proc_prokind), &[&schema])
.await
.map_err(|e| e.to_string())?;
Ok(rows
.iter()
.map(|row| {
let def: String = pg_row_try_string(row, 3);
// Remove schema qualification from CREATE FUNCTION statement
// to avoid false differences when comparing across schemas.
// Handle both "schema.name" and schema.name formats.
let normalized_def = def
.replace(&format!("CREATE OR REPLACE FUNCTION \"{}\".", schema), "CREATE OR REPLACE FUNCTION ")
.replace(&format!("CREATE OR REPLACE FUNCTION {}.", schema), "CREATE OR REPLACE FUNCTION ");
FunctionInfo {
name: pg_row_try_string(row, 0),
function_type: pg_row_try_string(row, 1),
data_type: pg_row_try_string(row, 2),
definition: normalized_def,
arguments: pg_row_try_string(row, 4),
}
})
.collect())
}
fn postgres_sequences_sql() -> &'static str {
"SELECT c.relname, \
COALESCE(format_type(s.seqtypid, NULL), 'bigint'), \
COALESCE(s.seqstart::text, '1'), \
COALESCE(s.seqmin::text, '1'), \
COALESCE(s.seqmax::text, '9223372036854775807'), \
COALESCE(s.seqincrement::text, '1'), \
CASE WHEN s.seqcycle THEN 'YES' ELSE 'NO' END \
FROM pg_class c \
JOIN pg_namespace n ON n.oid = c.relnamespace \
LEFT JOIN pg_sequence s ON s.seqrelid = c.oid \
WHERE c.relkind = 'S' AND n.nspname = $1 \
ORDER BY c.relname"
}
fn opengauss_sequences_sql() -> &'static str {
"SELECT s.sequence_name, \
COALESCE(s.data_type::text, 'bigint'), \
COALESCE(s.start_value::text, '1'), \
COALESCE(s.minimum_value::text, '1'), \
COALESCE(s.maximum_value::text, '9223372036854775807'), \
COALESCE(s.increment::text, '1'), \
COALESCE(s.cycle_option::text, 'NO') \
FROM information_schema.sequences s \
JOIN pg_namespace n ON n.nspname = s.sequence_schema \
JOIN pg_class c ON c.relnamespace = n.oid AND c.relname = s.sequence_name \
WHERE s.sequence_schema = $1 AND c.relkind IN ('S','L','z','Z') \
ORDER BY s.sequence_name"
}
fn postgres_sequence_last_values_sql() -> &'static str {
"SELECT c.relname, pg_sequence_last_value(c.oid)::text \
FROM pg_class c \
JOIN pg_namespace n ON n.oid = c.relnamespace \
WHERE c.relkind = 'S' AND n.nspname = $1"
}
fn opengauss_sequence_last_values_sql() -> &'static str {
"SELECT c.relname, (pg_sequence_last_value(c.oid)).last_value::text \
FROM pg_class c \
JOIN pg_namespace n ON n.oid = c.relnamespace \
WHERE c.relkind IN ('S','L','z','Z') AND n.nspname = $1"
}
async fn list_sequences_with_sql(
pool: &Pool,
schema: &str,
with_last_values: bool,
metadata_sql: &str,
last_values_sql: &str,
) -> Result<Vec<SequenceInfo>, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let rows = postgres_query_cached(&client, metadata_sql, &[&schema]).await.map_err(|e| e.to_string())?;
let mut sequences: Vec<SequenceInfo> = rows
.iter()
.map(|row| SequenceInfo {
name: pg_row_try_string(row, 0),
data_type: pg_row_try_string(row, 1),
start_value: pg_row_try_string(row, 2),
min_value: pg_row_try_string(row, 3),
max_value: pg_row_try_string(row, 4),
increment: pg_row_try_string(row, 5),
cycle: pg_row_try_string(row, 6) == "YES",
last_value: None,
})
.collect();
if with_last_values {
if let Ok(rows) = postgres_query_cached(&client, last_values_sql, &[&schema]).await {
for row in rows {
let name: String = pg_row_try_string(&row, 0);
if let Ok(Some(value)) = row.try_get::<_, Option<String>>(1) {
if let Some(seq) = sequences.iter_mut().find(|s| s.name == name) {
seq.last_value = Some(value);
}
}
}
}
}
Ok(sequences)
}
pub async fn list_sequences(pool: &Pool, schema: &str, with_last_values: bool) -> Result<Vec<SequenceInfo>, String> {
// PostgreSQL 10+ stores sequence properties in pg_sequence.
list_sequences_with_sql(
pool,
schema,
with_last_values,
postgres_sequences_sql(),
postgres_sequence_last_values_sql(),
)
.await
}
pub async fn list_opengauss_sequences(
pool: &Pool,
schema: &str,
with_last_values: bool,
) -> Result<Vec<SequenceInfo>, String> {
// openGauss does not expose PostgreSQL 10's pg_sequence catalog. Its
// information_schema view contains the portable sequence properties, while
// pg_sequence_last_value returns a record rather than a scalar.
list_sequences_with_sql(
pool,
schema,
with_last_values,
opengauss_sequences_sql(),
opengauss_sequence_last_values_sql(),
)
.await
}
pub async fn list_rules(pool: &Pool, schema: &str) -> Result<Vec<RuleInfo>, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let rows = postgres_query_cached(
&client,
"SELECT schemaname, tablename, rulename, definition \
FROM pg_rules \
WHERE schemaname = $1 \
ORDER BY rulename",
&[&schema],
)
.await
.map_err(|e| e.to_string())?;
Ok(rows
.iter()
.map(|row| RuleInfo {
name: pg_row_try_string(row, 2),
table_name: pg_row_try_string(row, 1),
definition: pg_row_try_string(row, 3),
})
.collect())
}
pub async fn list_extensions(pool: &Pool, schema: Option<&str>) -> Result<Vec<ExtensionInfo>, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let rows = if let Some(schema) = schema.filter(|value| !value.is_empty()) {
postgres_query_cached(
&client,
"SELECT e.extname, COALESCE(e.extversion, '') AS extversion, d.description, n.nspname \
FROM pg_catalog.pg_extension e \
JOIN pg_catalog.pg_namespace n ON n.oid = e.extnamespace \
LEFT JOIN pg_catalog.pg_description d ON d.objoid = e.oid AND d.classoid = 'pg_extension'::regclass \
WHERE n.nspname = $1 \
ORDER BY e.extname",
&[&schema],
)
.await
} else {
postgres_query_cached(
&client,
"SELECT e.extname, COALESCE(e.extversion, '') AS extversion, d.description, n.nspname \
FROM pg_catalog.pg_extension e \
JOIN pg_catalog.pg_namespace n ON n.oid = e.extnamespace \
LEFT JOIN pg_catalog.pg_description d ON d.objoid = e.oid AND d.classoid = 'pg_extension'::regclass \
ORDER BY n.nspname, e.extname",
&[],
)
.await
}
.map_err(|e| e.to_string())?;
Ok(rows
.iter()
.map(|row| ExtensionInfo {
name: pg_row_try_string(row, 0),
version: pg_row_try_string(row, 1),
comment: row.try_get::<_, Option<String>>(2).ok().flatten().filter(|s| !s.is_empty()),
schema: row.try_get::<_, Option<String>>(3).ok().flatten().filter(|s| !s.is_empty()),
})
.collect())
}
fn list_extension_member_objects_sql() -> &'static str {
"SELECT 'RELATION'::text AS object_kind, c.relname, ''::text AS signature \
FROM pg_catalog.pg_class c \
JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace \
WHERE n.nspname = $1 \
AND EXISTS ( \
SELECT 1 FROM pg_catalog.pg_depend d \
WHERE d.classid = 'pg_catalog.pg_class'::regclass \
AND d.objid = c.oid \
AND d.refclassid = 'pg_catalog.pg_extension'::regclass \
AND d.deptype = 'e' \
) \
UNION ALL \
SELECT 'FUNCTION'::text, p.proname, pg_get_function_identity_arguments(p.oid) \
FROM pg_catalog.pg_proc p \
JOIN pg_catalog.pg_namespace n ON n.oid = p.pronamespace \
WHERE n.nspname = $1 \
AND EXISTS ( \
SELECT 1 FROM pg_catalog.pg_depend d \
WHERE d.classid = 'pg_catalog.pg_proc'::regclass \
AND d.objid = p.oid \
AND d.refclassid = 'pg_catalog.pg_extension'::regclass \
AND d.deptype = 'e' \
)"
}
pub async fn list_extension_member_objects(pool: &Pool, schema: &str) -> Result<Vec<(String, String, String)>, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let rows = match postgres_query_cached(&client, list_extension_member_objects_sql(), &[&schema]).await {
Ok(rows) => rows,
Err(primary_error) => {
// PostgreSQL-compatible servers before the identity-argument
// formatter can still be filtered using their legacy formatter.
let fallback_sql = list_extension_member_objects_sql()
.replace("pg_get_function_identity_arguments(p.oid)", "pg_get_function_arguments(p.oid)");
postgres_query_cached(&client, &fallback_sql, &[&schema])
.await
.map_err(|fallback_error| format!("{primary_error}; legacy fallback failed: {fallback_error}"))?
}
};
Ok(rows
.iter()
.map(|row| (pg_row_try_string(row, 0), pg_row_try_string(row, 1), pg_row_try_string(row, 2)))
.collect())
}
pub async fn list_available_extensions(pool: &Pool) -> Result<Vec<ExtensionInfo>, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let rows = postgres_query_cached(
&client,
"SELECT name, default_version, comment \
FROM pg_catalog.pg_available_extensions \
WHERE installed_version IS NULL \
ORDER BY name",
&[],
)
.await
.map_err(|e| e.to_string())?;
Ok(rows
.iter()
.map(|row| ExtensionInfo {
name: pg_row_try_string(row, 0),
version: pg_row_try_string(row, 1),
comment: row.try_get::<_, Option<String>>(2).ok().flatten().filter(|s| !s.is_empty()),
schema: None,
})
.collect())
}
pub async fn list_owners(pool: &Pool, schema: &str) -> Result<Vec<OwnerInfo>, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let rows = postgres_query_cached(&client, POSTGRES_OWNERS_SQL, &[&schema]).await.map_err(|e| e.to_string())?;
Ok(rows
.iter()
.map(|row| {
let relkind: String = pg_row_try_string(row, 2);
OwnerInfo {
object_name: pg_row_try_string(row, 1),
object_type: postgres_owner_object_type(&relkind).to_string(),
owner: pg_row_try_string(row, 3),
}
})
.collect())
}
pub async fn get_table_access(pool: &Pool, schema: &str, table: &str) -> Result<PostgresTableAccessInfo, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let params: [&(dyn tokio_postgres::types::ToSql + Sync); 2] = [&schema, &table];
let owner_rows =
postgres_query_cached(&client, POSTGRES_TABLE_OWNER_SQL, &params).await.map_err(pg_error_to_string)?;
let owner_row = owner_rows.first().ok_or_else(|| "Table owner not found".to_string())?;
let owner = pg_row_try_string(owner_row, 0);
if owner.is_empty() {
return Err("Table owner not found".to_string());
}
let owner_default_privileges = owner_row.try_get::<_, Vec<String>>(1).unwrap_or_default();
if owner_default_privileges.is_empty() {
return Err("Table owner default privileges are unavailable".to_string());
}
let (table_privileges, column_privileges) = tokio::try_join!(
postgres_query_cached(&client, POSTGRES_TABLE_ACL_PRIVILEGES_SQL, &params),
postgres_query_cached(&client, POSTGRES_COLUMN_ACL_PRIVILEGES_SQL, &params),
)
.map_err(pg_error_to_string)?;
let privileges = table_privileges
.iter()
.map(|row| PostgresTablePrivilegeInfo {
grantor: pg_row_try_string(row, 3),
grantee: pg_row_try_string(row, 0),
privilege_type: pg_row_try_string(row, 1),
is_grantable: pg_row_try_bool(row, 2).unwrap_or(false),
column_name: None,
})
.chain(column_privileges.iter().map(|row| PostgresTablePrivilegeInfo {
grantor: pg_row_try_string(row, 4),
grantee: pg_row_try_string(row, 0),
privilege_type: pg_row_try_string(row, 1),
is_grantable: pg_row_try_bool(row, 2).unwrap_or(false),
column_name: Some(pg_row_try_string(row, 3)),
}))
.collect::<Vec<_>>();
if privileges.iter().any(|privilege| {
privilege.grantor.is_empty() || privilege.grantee.is_empty() || privilege.privilege_type.is_empty()
}) {
return Err("Table ACL metadata is incomplete".to_string());
}
Ok(PostgresTableAccessInfo { owner, owner_default_privileges, privileges })
}
/// Execute multiple SQL statements in a single round-trip using batch_execute.
/// Best for DDL scripts where per-statement affected-row counts are not needed.
pub async fn execute_batch(pool: &Pool, statements: &[String]) -> Result<(), String> {
let combined = statements.iter().map(|s| s.trim()).filter(|s| !s.is_empty()).collect::<Vec<_>>().join(";\n");
if combined.is_empty() {
return Ok(());
}
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
client.batch_execute(&combined).await.map_err(pg_error_to_string)?;
clear_postgres_caches_after_ddl(pool, Some(&client), &combined);
Ok(())
}
pub async fn terminate_current_user_database_backends(pool: &Pool, database: &str) -> Result<u64, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
client
.execute(
"SELECT pg_terminate_backend(pid) \
FROM pg_stat_activity \
WHERE datname = $1 \
AND pid <> pg_backend_pid() \
AND usename = current_user",
&[&database],
)
.await
.map_err(pg_error_to_string)
}
fn clear_postgres_caches_after_ddl(pool: &Pool, client: Option<&deadpool_postgres::Client>, sql: &str) {
if !invalidates_postgres_statement_cache(sql) {
return;
}
pool.manager().statement_caches.clear();
if let Some(client) = client {
client.clear_type_cache();
}
}
fn invalidates_postgres_statement_cache(sql: &str) -> bool {
let trimmed = sql.trim_start();
starts_with_executable_sql_keyword(
trimmed,
&["ALTER", "CREATE", "DROP", "TRUNCATE", "COMMENT", "REINDEX", "VACUUM"],
)
}
/// Export data via COPY TO STDOUT. `sql` must be a complete COPY statement, e.g.
/// `COPY table (col1, col2) TO STDOUT (FORMAT CSV, HEADER)`.
/// Returns the raw COPY output bytes.
pub async fn copy_out(pool: &Pool, sql: &str) -> Result<Vec<u8>, String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let stream = client.copy_out(sql).await.map_err(pg_error_to_string)?;
tokio::pin!(stream);
let mut result = Vec::new();
while let Some(chunk) = stream.next().await {
result.extend_from_slice(&chunk.map_err(pg_error_to_string)?);
}
Ok(result)
}
/// Import data via COPY FROM STDIN. `sql` must be a complete COPY statement, e.g.
/// `COPY table (col1, col2) FROM STDIN (FORMAT CSV)`.
/// `data` is the raw input in the format specified by the COPY command.
pub async fn copy_in(pool: &Pool, sql: &str, data: &[u8]) -> Result<(), String> {
let client = checkout_postgres_client(pool, None, super::connection_timeout()).await?;
let sink = client.copy_in::<str, bytes::Bytes>(sql).await.map_err(pg_error_to_string)?;
let mut sink = Box::pin(sink);
sink.as_mut().send(bytes::Bytes::copy_from_slice(data)).await.map_err(pg_error_to_string)?;
sink.as_mut().close().await.map_err(pg_error_to_string)
}
#[cfg(test)]
mod tests {
use super::*;
use std::cell::Cell;
use std::process::Command;
use std::time::Instant;
use tokio_postgres::types::FromSql;
fn pg_array_binary(element_oid: u32, elements: &[Option<Vec<u8>>]) -> Vec<u8> {
let mut raw = Vec::new();
raw.extend_from_slice(&1_i32.to_be_bytes());
raw.extend_from_slice(&i32::from(elements.iter().any(Option::is_none)).to_be_bytes());
raw.extend_from_slice(&element_oid.to_be_bytes());
raw.extend_from_slice(&(elements.len() as i32).to_be_bytes());
raw.extend_from_slice(&1_i32.to_be_bytes());
for element in elements {
match element {
Some(bytes) => {
raw.extend_from_slice(&(bytes.len() as i32).to_be_bytes());
raw.extend_from_slice(bytes);
}
None => raw.extend_from_slice(&(-1_i32).to_be_bytes()),
}
}
raw
}
fn pg_jsonb_binary(value: &[u8]) -> Vec<u8> {
let mut raw = Vec::with_capacity(value.len() + 1);
raw.push(1);
raw.extend_from_slice(value);
raw
}
#[test]
fn gaussdb_compatibility_mode_selects_identifier_quote() {
for mode in ["M", "B", "mysql", " MYSQL "] {
assert_eq!(gaussdb_identifier_quote_for_compatibility_mode(mode), Some("`"));
}
for mode in ["A", "PG", "ora", " PostgreSQL "] {
assert_eq!(gaussdb_identifier_quote_for_compatibility_mode(mode), Some("\""));
}
assert_eq!(gaussdb_identifier_quote_for_compatibility_mode("C"), None);
assert_eq!(gaussdb_identifier_quote_for_compatibility_mode(""), None);
}
fn test_query_message(message: &str) -> QueryMessage {
QueryMessage {
severity: "NOTICE".to_string(),
message: message.to_string(),
code: Some("00000".to_string()),
detail: None,
hint: None,
}
}
#[test]
fn take_notices_for_key_returns_buffered_notices_and_empties_buffer() {
let key = ("test-host".to_string(), 9_000_001, 9_000_001);
let buffer = Arc::new(Mutex::new(vec![test_query_message("first"), test_query_message("second")]));
postgres_notice_buffers()
.lock()
.unwrap_or_else(|poisoned| poisoned.into_inner())
.insert(key.clone(), Arc::downgrade(&buffer));
let notices = take_notices_for_key(&key);
assert_eq!(notices.len(), 2);
assert_eq!(notices[0].message, "first");
assert_eq!(notices[1].message, "second");
assert_eq!(notices[0].severity, "NOTICE");
assert_eq!(notices[0].code.as_deref(), Some("00000"));
// The buffer was drained but stays registered while the connection lives.
assert!(take_notices_for_key(&key).is_empty());
buffer.lock().unwrap_or_else(|poisoned| poisoned.into_inner()).push(test_query_message("third"));
let notices = take_notices_for_key(&key);
assert_eq!(notices.len(), 1);
assert_eq!(notices[0].message, "third");
}
#[test]
fn take_notices_for_key_prunes_dead_buffers_and_misses_return_empty() {
let live_key = ("test-host".to_string(), 9_000_002, 9_000_002);
let dead_key = ("test-host".to_string(), 9_000_003, 9_000_003);
let live = Arc::new(Mutex::new(vec![test_query_message("live")]));
let dead = Arc::new(Mutex::new(vec![test_query_message("dead")]));
{
let mut buffers = postgres_notice_buffers().lock().unwrap_or_else(|poisoned| poisoned.into_inner());
buffers.insert(live_key.clone(), Arc::downgrade(&live));
buffers.insert(dead_key.clone(), Arc::downgrade(&dead));
}
drop(dead);
assert!(take_notices_for_key(&("test-host".to_string(), 9_000_004, 9_000_004)).is_empty());
assert!(take_notices_for_key(&dead_key).is_empty());
let buffers = postgres_notice_buffers().lock().unwrap_or_else(|poisoned| poisoned.into_inner());
assert!(!buffers.contains_key(&dead_key));
assert!(buffers.contains_key(&live_key));
drop(buffers);
let notices = take_notices_for_key(&live_key);
assert_eq!(notices.len(), 1);
assert_eq!(notices[0].message, "live");
}
#[test]
fn take_notices_for_key_distinguishes_same_pid_on_different_servers() {
// Backend PIDs collide across servers; the (address, port, pid) key
// keeps notice attribution separate.
let key_a = ("server-a".to_string(), 5432, 42);
let key_b = ("server-b".to_string(), 5432, 42);
let buffer_a = Arc::new(Mutex::new(vec![test_query_message("from-a")]));
let buffer_b = Arc::new(Mutex::new(vec![test_query_message("from-b")]));
{
let mut buffers = postgres_notice_buffers().lock().unwrap_or_else(|poisoned| poisoned.into_inner());
buffers.insert(key_a.clone(), Arc::downgrade(&buffer_a));
buffers.insert(key_b.clone(), Arc::downgrade(&buffer_b));
}
let notices_a = take_notices_for_key(&key_a);
let notices_b = take_notices_for_key(&key_b);
assert_eq!(notices_a.len(), 1);
assert_eq!(notices_a[0].message, "from-a");
assert_eq!(notices_b.len(), 1);
assert_eq!(notices_b[0].message, "from-b");
}
#[test]
fn postgres_json_arrays_decode_elements_without_jsonb_version_bytes() {
let json_raw = pg_array_binary(
Type::JSON.oid(),
&[
Some(br#"{"kind":"json"}"#.to_vec()),
Some(br#""text""#.to_vec()),
Some(br#"[1,true,null]"#.to_vec()),
None,
],
);
let jsonb_raw = pg_array_binary(
Type::JSONB.oid(),
&[
Some(pg_jsonb_binary(br#"{"port":10031,"type":"admin_web"}"#)),
Some(pg_jsonb_binary(br#""quoted""#)),
Some(pg_jsonb_binary(br#"[2,false,{"nested":true}]"#)),
None,
],
);
let json_values = Vec::<Option<serde_json::Value>>::from_sql(&Type::JSON_ARRAY, &json_raw).unwrap();
let jsonb_values = Vec::<Option<serde_json::Value>>::from_sql(&Type::JSONB_ARRAY, &jsonb_raw).unwrap();
assert_eq!(
pg_json_array_values_to_json(json_values),
serde_json::json!([r#"{"kind":"json"}"#, r#""text""#, "[1,true,null]", null])
);
let decoded = pg_json_array_values_to_json(jsonb_values);
assert_eq!(
decoded,
serde_json::json!([
r#"{"port":10031,"type":"admin_web"}"#,
r#""quoted""#,
r#"[2,false,{"nested":true}]"#,
null
])
);
assert!(!decoded.to_string().contains('\u{1}'));
}
#[test]
fn postgres_json_array_decoder_is_limited_to_json_element_types() {
assert!(Vec::<Option<serde_json::Value>>::accepts(&Type::JSON_ARRAY));
assert!(Vec::<Option<serde_json::Value>>::accepts(&Type::JSONB_ARRAY));
assert!(!Vec::<Option<serde_json::Value>>::accepts(&Type::TEXT_ARRAY));
assert!(!Vec::<Option<serde_json::Value>>::accepts(&Type::INT4_ARRAY));
}
#[test]
fn postgres_explain_analyze_uses_read_only_transaction_local_schema() {
assert_eq!(
postgres_read_only_transaction_setup(Some("sales")),
vec![
("BEGIN READ ONLY".to_string(), "explain_analyze.begin"),
("SET LOCAL search_path TO \"sales\", pg_catalog, public".to_string(), "explain_analyze.schema"),
]
);
assert_eq!(
postgres_read_only_transaction_setup(None),
vec![("BEGIN READ ONLY".to_string(), "explain_analyze.begin")]
);
}
#[tokio::test]
async fn postgres_explain_analyze_rolls_back_after_success_and_query_failures() {
let rollback_calls = Cell::new(0);
let result = run_postgres_operation_with_rollback(
|| async { Ok::<_, String>(7) },
|| async {
rollback_calls.set(rollback_calls.get() + 1);
Ok(())
},
)
.await;
assert_eq!(result, Ok(7));
assert_eq!(rollback_calls.get(), 1);
for operation_error in ["query failed", crate::query::QUERY_CANCELED, "Query timed out after 30 seconds"] {
let rollback_calls = Cell::new(0);
let result = run_postgres_operation_with_rollback(
|| async { Err::<(), _>(operation_error.to_string()) },
|| async {
rollback_calls.set(rollback_calls.get() + 1);
Ok(())
},
)
.await;
assert_eq!(result, Err(operation_error.to_string()));
assert_eq!(rollback_calls.get(), 1);
}
}
#[tokio::test]
async fn postgres_explain_analyze_marks_rollback_failure_as_pool_pollution() {
let result = run_postgres_operation_with_rollback(
|| async { Err::<(), _>("query failed".to_string()) },
|| async { Err("PostgreSQL explain_analyze.rollback timed out after 3 seconds".to_string()) },
)
.await;
assert_eq!(
result,
Err("query failed; PostgreSQL read-only transaction cleanup failed: PostgreSQL explain_analyze.rollback timed out after 3 seconds".to_string())
);
}
fn pg_interval_bytes(microseconds: i64, days: i32, months: i32) -> [u8; 16] {
let mut raw = [0_u8; 16];
raw[0..8].copy_from_slice(&microseconds.to_be_bytes());
raw[8..12].copy_from_slice(&days.to_be_bytes());
raw[12..16].copy_from_slice(&months.to_be_bytes());
raw
}
#[test]
fn postgres_interval_binary_decodes_and_formats_components() {
let microseconds = 4 * 3_600_000_000 + 5 * 60_000_000 + 6 * 1_000_000 + 123_456;
let interval = PgInterval::from_sql(&Type::INTERVAL, &pg_interval_bytes(microseconds, 3, 14)).unwrap();
assert_eq!(interval, PgInterval { microseconds, days: 3, months: 14 });
assert_eq!(format_pg_interval(interval), "1 year 2 mons 3 days 04:05:06.123456");
}
#[test]
fn postgres_interval_formats_negative_mixed_and_zero_values() {
assert_eq!(
format_pg_interval(PgInterval { microseconds: -3_723_450_000, days: -2, months: -13 }),
"-1 year -1 mon -2 days -01:02:03.45"
);
assert_eq!(
format_pg_interval(PgInterval { microseconds: -1, days: 2, months: -1 }),
"-1 mon 2 days -00:00:00.000001"
);
assert_eq!(format_pg_interval(PgInterval { microseconds: 0, days: 0, months: 0 }), "00:00:00");
}
#[test]
fn postgres_interval_formats_now_minus_xact_start_shape() {
let elapsed = PgInterval { microseconds: 123_450_000, days: 0, months: 0 };
assert_eq!(format_pg_interval(elapsed), "00:02:03.45");
}
#[test]
fn postgres_interval_rejects_invalid_binary_and_keeps_binary_protocol() {
assert!(PgInterval::from_sql(&Type::INTERVAL, &[0; 15]).is_err());
assert_eq!(classify_pg_type("interval"), PgColType::Interval);
assert_eq!(classify_pg_type("_interval"), PgColType::Temporal { fallback: PgTemporalFallback::GenericArray });
assert!(!pg_type_requires_text_protocol(&Type::INTERVAL, PgColType::Interval));
}
fn pg_date_range_bytes(flags: u8, lower: Option<i32>, upper: Option<i32>) -> Vec<u8> {
let mut raw = vec![flags];
for days in [lower, upper].into_iter().flatten() {
raw.extend_from_slice(&4_i32.to_be_bytes());
raw.extend_from_slice(&days.to_be_bytes());
}
raw
}
#[test]
fn postgres_daterange_binary_decodes_reported_value() {
let raw = pg_date_range_bytes(PG_RANGE_LOWER_INCLUSIVE, Some(9_163), Some(9_168));
let range = PgDateRange::from_sql(&Type::DATE_RANGE, &raw).unwrap();
assert_eq!(range.0, "[2025-02-01,2025-02-06)");
}
#[test]
fn postgres_daterange_binary_handles_empty_unbounded_and_infinite_bounds() {
assert_eq!(PgDateRange::from_sql(&Type::DATE_RANGE, &[PG_RANGE_EMPTY]).unwrap().0, "empty");
assert_eq!(
PgDateRange::from_sql(
&Type::DATE_RANGE,
&pg_date_range_bytes(PG_RANGE_UPPER_UNBOUNDED | PG_RANGE_LOWER_INCLUSIVE, Some(9_163), None),
)
.unwrap()
.0,
"[2025-02-01,)"
);
assert_eq!(
PgDateRange::from_sql(
&Type::DATE_RANGE,
&pg_date_range_bytes(PG_RANGE_LOWER_UNBOUNDED, None, Some(9_168)),
)
.unwrap()
.0,
"(,2025-02-06)"
);
assert_eq!(
PgDateRange::from_sql(
&Type::DATE_RANGE,
&pg_date_range_bytes(PG_RANGE_LOWER_INCLUSIVE, Some(i32::MIN), Some(i32::MAX)),
)
.unwrap()
.0,
"[-infinity,infinity)"
);
}
#[test]
fn postgres_daterange_rejects_malformed_binary_and_keeps_binary_protocol() {
assert!(PgDateRange::from_sql(&Type::DATE_RANGE, &[]).is_err());
assert!(PgDateRange::from_sql(&Type::DATE_RANGE, &[PG_RANGE_EMPTY, 0]).is_err());
assert!(PgDateRange::from_sql(
&Type::DATE_RANGE,
&pg_date_range_bytes(PG_RANGE_LOWER_UNBOUNDED | PG_RANGE_LOWER_INCLUSIVE, None, Some(9_168)),
)
.is_err());
assert!(PgDateRange::from_sql(&Type::DATE_RANGE, &[PG_RANGE_LOWER_INCLUSIVE, 0, 0, 0, 3, 0, 0, 0]).is_err());
assert_eq!(classify_pg_type("daterange"), PgColType::DateRange);
assert_eq!(classify_pg_type("_daterange"), PgColType::GenericArray);
assert!(!pg_type_requires_text_protocol(&Type::DATE_RANGE, PgColType::DateRange));
}
#[test]
fn postgres_custom_other_type_requires_text_protocol() {
assert!(pg_scalar_type_requires_text_protocol(8_880, PgColType::Other));
assert!(pg_scalar_type_requires_text_protocol(98_765, PgColType::Other));
assert!(pg_scalar_type_requires_text_protocol(98_765, PgColType::GenericArray));
}
#[test]
fn postgres_record_types_require_text_protocol() {
assert!(pg_type_requires_text_protocol(&Type::RECORD, PgColType::Other));
assert!(pg_type_requires_text_protocol(&Type::RECORD_ARRAY, PgColType::GenericArray));
let dynamic_record =
Type::new("record".to_string(), Type::RECORD.oid(), Kind::Simple, "pg_catalog".to_string());
let dynamic_record_array =
Type::new("_record".to_string(), Type::RECORD_ARRAY.oid(), Kind::Simple, "pg_catalog".to_string());
assert!(pg_type_requires_text_protocol(&dynamic_record, PgColType::Other));
assert!(pg_type_requires_text_protocol(&dynamic_record_array, PgColType::GenericArray));
}
#[test]
fn postgres_enum_keeps_binary_protocol() {
let enum_type = Type::new(
"withdraw_btc_status".to_string(),
98_765,
Kind::Enum(vec!["pending".to_string(), "completed".to_string()]),
"risk".to_string(),
);
let enum_array =
Type::new("_withdraw_btc_status".to_string(), 98_766, Kind::Array(enum_type.clone()), "risk".to_string());
assert!(!pg_type_requires_text_protocol(&enum_type, PgColType::Other));
assert!(pg_type_requires_text_protocol(&enum_array, PgColType::GenericArray));
}
#[test]
fn postgres_builtin_or_supported_type_keeps_binary_protocol() {
assert!(!pg_type_requires_text_protocol(&Type::INT4, PgColType::Other));
assert!(!pg_type_requires_text_protocol(&Type::VARCHAR, PgColType::Other));
assert!(!pg_type_requires_text_protocol(&Type::INT4_ARRAY, PgColType::GenericArray));
assert!(!pg_scalar_type_requires_text_protocol(98_765, PgColType::Vector));
assert!(!pg_scalar_type_requires_text_protocol(98_765, PgColType::Geometry));
}
#[test]
fn postgres_compatible_builtin_names_with_unknown_low_oids_require_text_protocol() {
let compatible_date = Type::new("date".to_string(), 8_881, Kind::Simple, "pg_catalog".to_string());
let compatible_tinyint = Type::new("tinyint".to_string(), 8_882, Kind::Simple, "pg_catalog".to_string());
assert!(Type::from_oid(compatible_date.oid()).is_none());
assert!(Type::from_oid(compatible_tinyint.oid()).is_none());
assert!(pg_type_requires_text_protocol(
&compatible_date,
PgColType::Temporal { fallback: PgTemporalFallback::Probe }
));
assert!(pg_type_requires_text_protocol(&compatible_tinyint, PgColType::Other));
}
#[test]
fn postgres_query_uses_text_when_any_output_type_is_unsupported() {
let columns =
[(Type::INT4.oid(), PgColType::Other), (98_765, PgColType::Other), (Type::TEXT.oid(), PgColType::Other)];
assert!(columns.into_iter().any(|(oid, col_type)| pg_scalar_type_requires_text_protocol(oid, col_type)));
}
#[test]
fn postgres_text_fallback_keeps_matching_prepared_column_types() {
let columns = vec!["payload".to_string(), "id".to_string()];
let types = vec!["payload_type".to_string(), "int4".to_string()];
assert_eq!(matching_pg_text_column_types(&columns, Some(types.clone())), types);
}
#[test]
fn postgres_text_fallback_discards_misaligned_column_types() {
let columns = vec!["payload".to_string(), "id".to_string()];
let types = vec!["payload_type".to_string()];
assert!(matching_pg_text_column_types(&columns, Some(types)).is_empty());
assert!(matching_pg_text_column_types(&columns, None).is_empty());
}
#[test]
fn postgres_query_search_path_preserves_public_after_catalog() {
assert_eq!(
postgres_set_search_path_sql("application", PostgresSearchPathContext::Query),
"SET search_path TO \"application\", pg_catalog, public"
);
}
#[test]
fn postgres_transaction_search_paths_prioritize_selected_schema() {
assert_eq!(
postgres_set_search_path_sql("application", PostgresSearchPathContext::Transaction),
"SET search_path TO \"application\", pg_catalog"
);
assert_eq!(
postgres_set_search_path_sql("application", PostgresSearchPathContext::LocalTransaction),
"SET LOCAL search_path TO \"application\", pg_catalog"
);
assert_eq!(
postgres_set_search_path_sql("application", PostgresSearchPathContext::LocalQueryTransaction),
"SET LOCAL search_path TO \"application\", pg_catalog, public"
);
}
#[test]
fn postgres_search_path_safely_quotes_selected_schema() {
assert_eq!(
postgres_set_search_path_sql("tenant\"; RESET search_path; --", PostgresSearchPathContext::Query,),
"SET search_path TO \"tenant\"\"; RESET search_path; --\", pg_catalog, public"
);
}
#[test]
fn postgres_single_schema_search_path_preserves_scope_and_quoting() {
assert_eq!(
postgres_set_single_schema_search_path_sql("application", PostgresSearchPathContext::Query),
"SET search_path TO \"application\""
);
assert_eq!(
postgres_set_single_schema_search_path_sql("application", PostgresSearchPathContext::Transaction),
"SET search_path TO \"application\""
);
assert_eq!(
postgres_set_single_schema_search_path_sql(
"tenant\"; RESET search_path; --",
PostgresSearchPathContext::LocalTransaction,
),
"SET LOCAL search_path TO \"tenant\"\"; RESET search_path; --\""
);
}
#[test]
fn postgres_single_schema_fallback_only_matches_compatible_server_error() {
assert!(postgres_requires_single_schema_search_path(
"ERROR: Hologres does not support search_path with multiple names: admaterial."
));
assert!(!postgres_requires_single_schema_search_path("ERROR: permission denied for schema admaterial"));
}
#[test]
fn postgres_statement_returns_rows_for_returning_dml_only() {
for sql in [
"INSERT INTO users (id) VALUES (1) RETURNING id",
"UPDATE users SET name = 'Ada' RETURNING id, name",
"DELETE FROM users WHERE id = 1 RETURNING id",
"MERGE INTO users AS target USING updates AS source ON target.id = source.id WHEN MATCHED THEN UPDATE SET name = source.name RETURNING target.id",
"WITH removed AS (DELETE FROM users WHERE id = 1 RETURNING id) SELECT * FROM removed",
] {
assert!(postgres_statement_returns_rows(sql), "expected result rows for: {sql}");
}
for sql in [
"INSERT INTO users (id) VALUES (1)",
"UPDATE users SET name = 'Ada'",
"DELETE FROM users WHERE id = 1",
"MERGE INTO users AS target USING updates AS source ON target.id = source.id WHEN MATCHED THEN UPDATE SET name = source.name",
"INSERT INTO users (note) VALUES ('RETURNING is text')",
] {
assert!(!postgres_statement_returns_rows(sql), "expected command result for: {sql}");
}
}
#[test]
fn database_list_does_not_collect_storage_usage() {
assert!(list_databases_sql().contains("pg_database"));
assert!(!list_databases_sql().contains("pg_database_size"));
}
#[test]
fn database_storage_is_scoped_and_permission_guarded() {
let sql = database_storage_sql();
assert!(sql.contains("d.datname = ANY($1::text[])"));
assert!(sql.contains("has_database_privilege"));
assert!(sql.contains("pg_read_all_stats"));
assert!(sql.contains("pg_database_size"));
assert!(sql.contains("ELSE NULL"));
}
#[test]
fn classify_pg_type_covers_all_dispatch_branches() {
assert_eq!(classify_pg_type("bytea"), PgColType::Bytea);
assert_eq!(classify_pg_type("json"), PgColType::Json);
assert_eq!(classify_pg_type("JSONB"), PgColType::Json);
assert_eq!(classify_pg_type("bool"), PgColType::Bool);
assert_eq!(classify_pg_type("point"), PgColType::Point);
assert_eq!(classify_pg_type("timestamp"), PgColType::Temporal { fallback: PgTemporalFallback::Probe });
assert_eq!(classify_pg_type("timestamptz"), PgColType::Temporal { fallback: PgTemporalFallback::Probe });
assert_eq!(classify_pg_type("date"), PgColType::Temporal { fallback: PgTemporalFallback::Probe });
assert_eq!(classify_pg_type("time"), PgColType::Temporal { fallback: PgTemporalFallback::Probe });
assert_eq!(classify_pg_type("timetz"), PgColType::Temporal { fallback: PgTemporalFallback::Probe });
assert_eq!(classify_pg_type("interval"), PgColType::Interval);
// 时间数组类型名在原实现中先进时间分支、解码失败后落到通用数组分支
assert_eq!(classify_pg_type("_timestamp"), PgColType::Temporal { fallback: PgTemporalFallback::GenericArray });
assert_eq!(classify_pg_type("_interval"), PgColType::Temporal { fallback: PgTemporalFallback::GenericArray });
// 同时命中时间关键字与 VECTOR( 前缀的类型名，原实现时间解码失败后走 vector 分支
assert_eq!(classify_pg_type("vector(timestamp)"), PgColType::Temporal { fallback: PgTemporalFallback::Vector });
assert_eq!(classify_pg_type("numeric"), PgColType::Numeric);
assert_eq!(classify_pg_type("money"), PgColType::Numeric);
assert_eq!(classify_pg_type("uuid"), PgColType::Uuid);
assert_eq!(classify_pg_type("inet"), PgColType::Inet { cidr: false });
assert_eq!(classify_pg_type("cidr"), PgColType::Inet { cidr: true });
assert_eq!(classify_pg_type("macaddr"), PgColType::MacAddr);
assert_eq!(classify_pg_type("macaddr8"), PgColType::MacAddr);
assert_eq!(classify_pg_type("bit"), PgColType::BitString);
assert_eq!(classify_pg_type("varbit"), PgColType::BitString);
assert_eq!(classify_pg_type("tsvector"), PgColType::TsVector);
assert_eq!(classify_pg_type("oid"), PgColType::SystemU32);
assert_eq!(classify_pg_type("xid"), PgColType::SystemU32);
assert_eq!(classify_pg_type("_inet"), PgColType::InetArray { cidr: false });
assert_eq!(classify_pg_type("_cidr"), PgColType::InetArray { cidr: true });
assert_eq!(classify_pg_type("_macaddr"), PgColType::MacAddrArray);
assert_eq!(classify_pg_type("_bit"), PgColType::BitStringArray);
assert_eq!(classify_pg_type("_varbit"), PgColType::BitStringArray);
assert_eq!(classify_pg_type("_int4"), PgColType::GenericArray);
assert_eq!(classify_pg_type("_time"), PgColType::GenericArray);
assert_eq!(classify_pg_type("vector"), PgColType::Vector);
assert_eq!(classify_pg_type("vector(3)"), PgColType::Vector);
assert_eq!(classify_pg_type("geometry"), PgColType::Geometry);
assert_eq!(classify_pg_type("geography"), PgColType::Geometry);
assert_eq!(classify_pg_type("int4"), PgColType::Other);
assert_eq!(classify_pg_type("varchar"), PgColType::Other);
assert_eq!(classify_pg_type(""), PgColType::Other);
}
#[test]
fn postgres_text_spatial_value_separates_srid_from_wkt() {
assert_eq!(
pg_text_fallback_value("SRID=4326;POINT(1 2)", Some(PgColType::Geometry)),
(serde_json::json!("POINT(1 2)"), Some(4326))
);
assert_eq!(
pg_text_fallback_value("SRID=0;POINT(1 2)", Some(PgColType::Geometry)),
(serde_json::json!("POINT(1 2)"), None)
);
assert_eq!(
pg_text_fallback_value("POINT(1 2)", Some(PgColType::Geometry)),
(serde_json::json!("POINT(1 2)"), None)
);
}
#[test]
fn postgres_text_spatial_value_decodes_hex_ewkb() {
let ewkb = "0101000020E6100000000000000000F03F0000000000000040";
for value in [ewkb.to_string(), format!("0x{ewkb}"), format!("\\x{ewkb}")] {
assert_eq!(
pg_text_fallback_value(&value, Some(PgColType::Geometry)),
(serde_json::json!("POINT(1 2)"), Some(4326))
);
assert_eq!(pg_text_fallback_value(&value, None), (serde_json::json!("POINT(1 2)"), Some(4326)));
}
let srid_zero = "010100002000000000000000000000F03F0000000000000040";
assert_eq!(pg_text_fallback_value(srid_zero, None), (serde_json::json!("POINT(1 2)"), None));
}
#[test]
fn postgres_text_fallback_does_not_reinterpret_ordinary_text() {
for value in ["SRID=abc;POINT(1 2)", "SRID=4326;not geometry", "0101-not-hex", "POINTLESS"] {
assert_eq!(pg_text_fallback_value(value, None), (serde_json::json!(value), None));
}
assert_eq!(pg_text_fallback_value("SRID=4326;point(1 2)", None), (serde_json::json!("point(1 2)"), Some(4326)));
}
struct DockerPostgres {
name: String,
port: u16,
}
impl DockerPostgres {
fn url(&self) -> String {
format!("postgres://postgres:postgres@127.0.0.1:{}/postgres?sslmode=disable", self.port)
}
}
impl Drop for DockerPostgres {
fn drop(&mut self) {
let _ = Command::new("docker").args(["rm", "-f", &self.name]).status();
}
}
fn docker_ready() -> bool {
Command::new("docker")
.args(["version", "--format", "{{.Server.Version}}"])
.output()
.map(|output| output.status.success())
.unwrap_or(false)
}
async fn start_docker_postgres() -> Option<DockerPostgres> {
if !docker_ready() {
eprintln!("skipping docker-backed postgres test because Docker is unavailable");
return None;
}
let port = portpicker::pick_unused_port().expect("pick unused postgres port");
let container = DockerPostgres { name: format!("dbx-postgres-enum-{}", uuid::Uuid::new_v4()), port };
let status = Command::new("docker")
.args([
"run",
"-d",
"--rm",
"--name",
&container.name,
"-e",
"POSTGRES_PASSWORD=postgres",
"-e",
"POSTGRES_USER=postgres",
"-e",
"POSTGRES_DB=postgres",
"-p",
&format!("{port}:5432"),
"postgres:16-alpine",
])
.status()
.expect("start docker postgres");
assert!(status.success(), "docker run postgres container should succeed");
let deadline = Instant::now() + Duration::from_secs(60);
loop {
match connect(&container.url(), Duration::from_secs(2)).await {
Ok(pool) => {
drop(pool);
return Some(container);
}
Err(_) if Instant::now() < deadline => tokio::time::sleep(Duration::from_millis(500)).await,
Err(error) => panic!("docker postgres did not become ready: {error}"),
}
}
}
#[tokio::test]
async fn postgres_dml_returning_preserves_result_rows() {
let Some(container) = start_docker_postgres().await else {
return;
};
let pool = connect(&container.url(), Duration::from_secs(5)).await.expect("connect postgres");
execute_query(&pool, "CREATE TABLE dml_returning (id integer PRIMARY KEY, name text NOT NULL)")
.await
.expect("create table");
let inserted = execute_query(&pool, "INSERT INTO dml_returning VALUES (1, 'alice'), (2, 'bob')")
.await
.expect("insert rows");
let updated = execute_query(&pool, "UPDATE dml_returning SET name = 'bob-updated' WHERE id = 2")
.await
.expect("update rows");
let deleted = execute_query(&pool, "DELETE FROM dml_returning WHERE id = 1").await.expect("delete row");
let unmatched =
execute_query(&pool, "UPDATE dml_returning SET name = name WHERE id = 999").await.expect("update no rows");
assert_eq!(inserted.affected_rows, 2);
assert_eq!(updated.affected_rows, 1);
assert_eq!(deleted.affected_rows, 1);
assert_eq!(unmatched.affected_rows, 0);
let insert_returning = execute_query(&pool, "INSERT INTO dml_returning VALUES (3, 'carol') RETURNING id, name")
.await
.expect("insert returning");
let update_returning =
execute_query(&pool, "UPDATE dml_returning SET name = 'carol-updated' WHERE id = 3 RETURNING id, name")
.await
.expect("update returning");
let delete_returning = execute_query(&pool, "DELETE FROM dml_returning WHERE id = 3 RETURNING id, name")
.await
.expect("delete returning");
let empty_returning =
execute_query(&pool, "UPDATE dml_returning SET name = name WHERE id = 999 RETURNING id, name")
.await
.expect("empty returning");
for result in [&insert_returning, &update_returning, &delete_returning] {
assert_eq!(result.columns, vec!["id", "name"]);
assert_eq!(result.rows.len(), 1);
}
assert_eq!(insert_returning.rows[0], vec![serde_json::json!(3), serde_json::json!("carol")]);
assert_eq!(update_returning.rows[0], vec![serde_json::json!(3), serde_json::json!("carol-updated")]);
assert_eq!(delete_returning.rows[0], vec![serde_json::json!(3), serde_json::json!("carol-updated")]);
assert_eq!(empty_returning.columns, vec!["id", "name"]);
assert!(empty_returning.rows.is_empty());
}
async fn assert_postgres_18(pool: &Pool) {
let version = execute_query(pool, "SHOW server_version_num").await.expect("query PostgreSQL version");
let version_num = version.rows[0][0]
.as_str()
.expect("server_version_num should be text")
.parse::<u32>()
.expect("server_version_num should be numeric");
assert!((180_000..190_000).contains(&version_num), "expected PostgreSQL 18, got {version_num}");
}
#[tokio::test]
#[ignore = "requires DBX_TEST_POSTGRES_URL pointing at a writable PostgreSQL 18 database"]
async fn postgres_custom_composite_result_uses_server_text_output() {
let url = std::env::var("DBX_TEST_POSTGRES_URL").expect("DBX_TEST_POSTGRES_URL");
let pool = connect(&url, Duration::from_secs(5)).await.expect("connect postgres");
assert_postgres_18(&pool).await;
let schema = format!("dbx_custom_text_{}", uuid::Uuid::new_v4().simple());
let schema_ident = pg_quote_ident(&schema);
let payload_type = format!("{schema_ident}.payload");
execute_query(&pool, &format!("CREATE SCHEMA {schema_ident}")).await.expect("create schema");
let exercise = async {
execute_query(&pool, &format!("CREATE TYPE {payload_type} AS (id integer, label text)")).await?;
let custom =
execute_query(&pool, &format!("SELECT ROW(7, 'alpha')::{payload_type} AS payload, 42::int4 AS id"))
.await?;
let builtin = execute_query(&pool, "SELECT 42::int4 AS id").await?;
Ok::<_, String>((custom, builtin))
}
.await;
let cleanup = execute_query(&pool, &format!("DROP SCHEMA {schema_ident} CASCADE")).await;
cleanup.expect("drop schema");
let (custom, builtin) = exercise.expect("exercise custom composite fallback");
assert_eq!(custom.columns, vec!["payload", "id"]);
assert_eq!(custom.column_types, vec!["payload", "int4"]);
assert_eq!(custom.rows[0][0], serde_json::Value::String("(7,alpha)".to_string()));
assert_eq!(custom.rows[0][1], serde_json::Value::String("42".to_string()));
assert!(!custom.rows[0][0].as_str().unwrap().chars().any(char::is_control));
assert_eq!(builtin.column_types, vec!["int4"]);
assert_eq!(builtin.rows[0][0], serde_json::Value::Number(42.into()));
}
#[tokio::test]
#[ignore = "requires DBX_TEST_POSTGRES_URL pointing at a writable PostgreSQL 18 database"]
async fn postgres_custom_type_arrays_and_exports_use_server_text_output() {
let url = std::env::var("DBX_TEST_POSTGRES_URL").expect("DBX_TEST_POSTGRES_URL");
let pool = connect(&url, Duration::from_secs(5)).await.expect("connect postgres");
assert_postgres_18(&pool).await;
let schema = format!("dbx_custom_array_{}", uuid::Uuid::new_v4().simple());
let schema_ident = pg_quote_ident(&schema);
let payload_type = format!("{schema_ident}.payload");
let mood_type = format!("{schema_ident}.mood");
let score_type = format!("{schema_ident}.positive_int");
let underscore_scalar_type = format!("{schema_ident}._hidden");
let vector_named_enum_type = format!("{schema_ident}.vector");
let table = format!("{schema_ident}.custom_arrays");
let select_sql = format!("SELECT payloads, moods, scores FROM {table}");
execute_query(&pool, &format!("CREATE SCHEMA {schema_ident}")).await.expect("create schema");
let exercise = async {
execute_query(&pool, &format!("CREATE TYPE {payload_type} AS (id integer, label text)")).await?;
execute_query(&pool, &format!("CREATE TYPE {mood_type} AS ENUM ('ready', 'done')")).await?;
execute_query(&pool, &format!("CREATE DOMAIN {score_type} AS integer CHECK (VALUE > 0)")).await?;
execute_query(&pool, &format!("CREATE TYPE {underscore_scalar_type} AS ENUM ('secret')")).await?;
execute_query(&pool, &format!("CREATE TYPE {vector_named_enum_type} AS ENUM ('label')")).await?;
execute_query(
&pool,
&format!(
"CREATE TABLE {table} (payloads {payload_type}[], moods {mood_type}[], scores {score_type}[])"
),
)
.await?;
execute_query(
&pool,
&format!(
"INSERT INTO {table} VALUES \
(ARRAY[ROW(7, 'alpha')::{payload_type}], ARRAY['ready'::{mood_type}], ARRAY[7::{score_type}])"
),
)
.await?;
let query = execute_query(&pool, &select_sql).await?;
let underscore_scalar =
execute_query(&pool, &format!("SELECT 'secret'::{underscore_scalar_type} AS hidden")).await?;
let vector_named_enum =
execute_query(&pool, &format!("SELECT 'label'::{vector_named_enum_type} AS label")).await?;
let client = checkout_postgres_client(&pool, None, Duration::from_secs(5)).await?;
let mut query_export_rows = Vec::new();
stream_select_query_inner(&client, &select_sql, None, &mut |item| {
if let PostgresQueryStreamItem::Row(row) = item {
query_export_rows.push(row);
}
Ok(())
})
.await?;
let cancelled = AtomicBool::new(false);
let mut table_export_rows = Vec::new();
stream_query_rows_on_client(&client, &select_sql, None, &cancelled, &mut |row| {
table_export_rows.push(row.to_vec());
Ok(())
})
.await?;
drop(client);
Ok::<_, String>((query, underscore_scalar, vector_named_enum, query_export_rows, table_export_rows))
}
.await;
let cleanup = execute_query(&pool, &format!("DROP SCHEMA {schema_ident} CASCADE")).await;
cleanup.expect("drop schema");
let (query, underscore_scalar, vector_named_enum, query_export_rows, table_export_rows) =
exercise.expect("exercise custom array fallbacks");
let expected = vec![
serde_json::Value::String(r#"{"(7,alpha)"}"#.to_string()),
serde_json::Value::String("{ready}".to_string()),
serde_json::Value::String("{7}".to_string()),
];
assert_eq!(query.rows, vec![expected.clone()]);
assert_eq!(underscore_scalar.rows, vec![vec![serde_json::Value::String("secret".to_string())]]);
assert_eq!(vector_named_enum.rows, vec![vec![serde_json::Value::String("label".to_string())]]);
assert_eq!(query_export_rows, vec![expected.clone()]);
assert_eq!(table_export_rows, vec![expected]);
}
#[tokio::test]
#[ignore = "requires DBX_TEST_POSTGRES_URL pointing at a writable PostgreSQL 18 database"]
async fn postgres_custom_type_fallback_refreshes_stale_cached_metadata() {
let url = std::env::var("DBX_TEST_POSTGRES_URL").expect("DBX_TEST_POSTGRES_URL");
let pool_a = connect(&url, Duration::from_secs(5)).await.expect("connect postgres pool A");
let pool_b = connect(&url, Duration::from_secs(5)).await.expect("connect postgres pool B");
assert_postgres_18(&pool_a).await;
let schema = format!("dbx_custom_stale_{}", uuid::Uuid::new_v4().simple());
let schema_ident = pg_quote_ident(&schema);
let payload_type = format!("{schema_ident}.payload");
let view = format!("{schema_ident}.cached_payload");
let view_sql = format!("SELECT payload FROM {view}");
execute_query(&pool_a, &format!("CREATE SCHEMA {schema_ident}")).await.expect("create schema");
let exercise = async {
execute_query(&pool_a, &format!("CREATE TYPE {payload_type} AS (id integer, label text)")).await?;
execute_query(&pool_a, &format!("CREATE VIEW {view} AS SELECT ROW(7, 'alpha')::{payload_type} AS payload"))
.await?;
let custom = execute_query(&pool_a, &view_sql).await?;
execute_query(&pool_b, &format!("DROP VIEW {view}")).await?;
execute_query(&pool_b, &format!("CREATE VIEW {view} AS SELECT 42::int4 AS payload")).await?;
let builtin = execute_query(&pool_a, &view_sql).await?;
Ok::<_, String>((custom, builtin))
}
.await;
let cleanup = execute_query(&pool_a, &format!("DROP SCHEMA {schema_ident} CASCADE")).await;
cleanup.expect("drop schema");
let (custom, builtin) = exercise.expect("exercise stale cached custom metadata");
assert_eq!(custom.column_types, vec!["payload"]);
assert_eq!(custom.rows[0][0], serde_json::Value::String("(7,alpha)".to_string()));
assert_eq!(builtin.column_types, vec!["int4"]);
assert_eq!(builtin.rows[0][0], serde_json::Value::Number(42.into()));
}
#[tokio::test]
#[ignore = "requires DBX_TEST_POSTGRES_URL pointing at a writable PostgreSQL 18 database"]
async fn postgres_text_fallback_stops_before_late_row_error_at_limit() {
let url = std::env::var("DBX_TEST_POSTGRES_URL").expect("DBX_TEST_POSTGRES_URL");
let pool = connect(&url, Duration::from_secs(5)).await.expect("connect postgres");
assert_postgres_18(&pool).await;
let schema = format!("dbx_custom_limit_{}", uuid::Uuid::new_v4().simple());
let schema_ident = pg_quote_ident(&schema);
let payload_type = format!("{schema_ident}.payload");
let fail_after_two = format!("{schema_ident}.fail_after_two");
execute_query(&pool, &format!("CREATE SCHEMA {schema_ident}")).await.expect("create schema");
let exercise = async {
execute_query(&pool, &format!("CREATE TYPE {payload_type} AS (id integer)")).await?;
execute_query(
&pool,
&format!(
"CREATE FUNCTION {fail_after_two}(i integer) RETURNS integer LANGUAGE plpgsql AS $$ \
BEGIN IF i >= 2 THEN RAISE EXCEPTION 'late row failure'; END IF; RETURN i; END $$"
),
)
.await?;
let client = checkout_postgres_client(&pool, None, Duration::from_secs(5)).await?;
let custom_sql = format!(
"SELECT ROW({fail_after_two}(i))::{payload_type} AS payload \
FROM generate_series(1, 2) AS series(i)"
);
let limited = execute_select_query(&client, &custom_sql, Instant::now(), 1).await;
let cancelled = AtomicBool::new(false);
let mut exported_rows = Vec::new();
let exported = stream_query_rows_on_client(&client, &custom_sql, Some(1), &cancelled, &mut |row| {
exported_rows.push(row.to_vec());
Ok(())
})
.await;
let recovery = execute_select_query(&client, "SELECT 1::int4 AS value", Instant::now(), 1).await;
drop(client);
Ok::<_, String>((limited, exported, exported_rows, recovery))
}
.await;
let cleanup = execute_query(&pool, &format!("DROP SCHEMA {schema_ident} CASCADE")).await;
cleanup.expect("drop schema");
let (limited, exported, exported_rows, recovery) = exercise.expect("set up late row error query");
let limited = limited.expect("query should stop before late row error");
let exported = exported.expect("streamed export should stop before late row error");
let recovery = recovery.expect("connection should remain reusable");
assert_eq!(limited.column_types, vec!["payload"]);
assert_eq!(limited.rows, vec![vec![serde_json::Value::String("(1)".to_string())]]);
assert!(limited.truncated);
assert_eq!(exported, 1);
assert_eq!(exported_rows, vec![vec![serde_json::Value::String("(1)".to_string())]]);
assert_eq!(recovery.column_types, vec!["int4"]);
assert_eq!(recovery.rows[0][0], serde_json::Value::Number(1.into()));
}
fn state_enum_values(columns: &[ColumnInfo]) -> Option<Vec<String>> {
columns.iter().find(|column| column.name == "state").and_then(|column| column.enum_values.clone())
}
// --- pg_quote_ident ---
#[test]
fn pg_system_u32_decodes_catalog_integer_types() {
let raw = 42_u32.to_be_bytes();
assert_eq!(u32::from_sql(&Type::OID, &raw).unwrap(), 42);
assert_eq!(PgSystemU32::from_sql(&Type::XID, &raw).unwrap().0, 42);
assert_eq!(PgSystemU32::from_sql(&Type::CID, &raw).unwrap().0, 42);
assert!(u32::accepts(&Type::OID));
assert!(PgSystemU32::accepts(&Type::XID));
assert!(PgSystemU32::accepts(&Type::CID));
assert!(!PgSystemU32::accepts(&Type::OID));
assert!(!PgSystemU32::accepts(&Type::INT4));
}
#[test]
fn pg_point_decodes_binary_coordinates_as_text() {
let mut raw = Vec::new();
raw.extend_from_slice(&(-194.0_f64).to_be_bytes());
raw.extend_from_slice(&53.0_f64.to_be_bytes());
let point = PgPoint::from_sql(&Type::POINT, &raw).unwrap();
assert_eq!(point, PgPoint { x: -194.0, y: 53.0 });
assert_eq!(format_pg_point(point), "(-194,53)");
assert_eq!(format_pg_point(PgPoint { x: f64::NEG_INFINITY, y: f64::INFINITY }), "(-Infinity,Infinity)");
assert!(PgPoint::from_sql(&Type::POINT, &[0; 15]).is_err());
assert!(PgPoint::accepts(&Type::POINT));
assert!(!PgPoint::accepts(&Type::BYTEA));
}
#[test]
fn pg_any_string_accepts_all_types_and_decodes_utf8() {
// Accepts any type — built-in, custom enum OIDs, domains, etc.
assert!(PgAnyString::accepts(&Type::TEXT));
assert!(PgAnyString::accepts(&Type::INT4));
assert!(PgAnyString::accepts(&Type::UNKNOWN));
assert!(PgAnyString::accepts(&Type::OID));
assert!(PgAnyString::accepts(&Type::BOOL));
let label = PgAnyString::from_sql(&Type::UNKNOWN, b"pending").unwrap();
assert_eq!(label.0, "pending");
let label = PgAnyString::from_sql(&Type::UNKNOWN, b"hello world").unwrap();
assert_eq!(label.0, "hello world");
// Non-UTF-8 bytes should fail gracefully
assert!(PgAnyString::from_sql(&Type::UNKNOWN, &[0xFF, 0xFE, 0xFD]).is_err());
}
#[test]
fn pg_raw_bytes_accepts_all_types_and_preserves_binary_payloads() {
assert!(PgRawBytes::accepts(&Type::TEXT));
assert!(PgRawBytes::accepts(&Type::UNKNOWN));
assert!(PgRawBytes::accepts(&Type::OID));
let raw = PgRawBytes::from_sql(&Type::UNKNOWN, &[0x01, 0xAB, 0xFF]).unwrap();
assert_eq!(raw.0, vec![0x01, 0xAB, 0xFF]);
}
#[test]
fn decode_bool_bytes_handles_standard_and_gaussdb_encodings() {
// Standard PostgreSQL binary boolean: 0x00 / 0x01
assert_eq!(decode_bool_bytes(&[0x00]), Some(false));
assert_eq!(decode_bool_bytes(&[0x01]), Some(true));
// GaussDB binary boolean: ASCII 't' (0x74) / 'f' (0x66)
assert_eq!(decode_bool_bytes(&[0x74]), Some(true));
assert_eq!(decode_bool_bytes(&[0x66]), Some(false));
assert_eq!(decode_bool_bytes(b"t"), Some(true));
assert_eq!(decode_bool_bytes(b"f"), Some(false));
assert_eq!(decode_bool_bytes(b"T"), Some(true));
assert_eq!(decode_bool_bytes(b"F"), Some(false));
// Unrecognized encodings return None
assert_eq!(decode_bool_bytes(&[0x02]), None);
assert_eq!(decode_bool_bytes(&[0x74, 0x66]), None);
assert_eq!(decode_bool_bytes(&[]), None);
}
#[test]
fn raw_gaussdb_boolean_takes_precedence_over_standard_decoder() {
assert_eq!(decode_bool_candidates(Some(b"f"), Some(true)), Some(false));
assert_eq!(decode_bool_candidates(Some(b"t"), Some(true)), Some(true));
assert_eq!(decode_bool_candidates(Some(&[0x00]), Some(true)), Some(false));
assert_eq!(decode_bool_candidates(Some(&[0x01]), Some(false)), Some(true));
assert_eq!(decode_bool_candidates(Some(&[0x02]), Some(false)), Some(false));
}
#[test]
fn postgres_foreign_keys_sql_selects_referential_actions() {
let sql = postgres_foreign_keys_sql();
assert!(sql.contains("rc.update_rule AS on_update"));
assert!(sql.contains("rc.delete_rule AS on_delete"));
assert!(sql.contains("information_schema.referential_constraints rc"));
}
#[test]
fn postgres_table_dependencies_sql_batches_schema_foreign_keys() {
let sql = postgres_table_dependencies_sql();
assert!(sql.contains("pg_catalog.pg_constraint"));
assert!(sql.contains("con.contype = 'f'"));
assert!(sql.contains("child_schema.nspname = $1"));
assert!(sql.contains("parent_schema.nspname = $1"));
assert!(!sql.contains("information_schema"));
}
#[test]
fn postgres_foreign_key_action_keeps_non_empty_action() {
assert_eq!(postgres_foreign_key_action("CASCADE".to_string()), Some("CASCADE".to_string()));
assert_eq!(postgres_foreign_key_action(" SET NULL ".to_string()), Some("SET NULL".to_string()));
assert_eq!(postgres_foreign_key_action("".to_string()), None);
assert_eq!(postgres_foreign_key_action(" ".to_string()), None);
}
#[test]
fn decodes_tsvector_binary_output() {
let raw = [
0, 0, 0, 2, b'b', b'a', b'c', b'k', b'\\', b's', b'l', b'a', b's', b'h', 0, 0, 1, 0x80, 0x03, b'o', b'\'',
b'c', b'l', b'o', b'c', b'k', 0, 0, 2, 0, 1, 0xc0, 0x02,
];
assert_eq!(decode_tsvector_bytes(&raw).as_deref(), Some("'back\\\\slash':3B 'o''clock':1,2A"));
}
fn encode_pgvector_bytes(values: &[f32]) -> Vec<u8> {
let dims = u16::try_from(values.len()).expect("dim fits u16");
let mut raw = Vec::with_capacity(4 + values.len() * 4);
raw.extend_from_slice(&dims.to_be_bytes());
raw.extend_from_slice(&0u16.to_be_bytes());
for value in values {
raw.extend_from_slice(&value.to_be_bytes());
}
raw
}
#[test]
fn decodes_pgvector_binary_output() {
let values = [0.1f32, -2.5f32, 1.2345679e-5f32];
let decoded = decode_pgvector_bytes(&encode_pgvector_bytes(&values)).expect("decode vector");
assert_eq!(decoded, values);
}
#[test]
fn pgvector_element_number_round_trips_full_f32_precision() {
let values = [0.1f32, 0.12345679f32, 1.2345679e-5f32, -0.00012345679f32, 1.2345678f32, 1e20f32];
for value in values {
let json = pg_vector_element_number(value);
let text = json.to_string();
let restored: f32 = text.parse().expect("json number parses as f32");
let rounded_six = ((value as f64 * 1_000_000.0).round() / 1_000_000.0) as f32;
// Display text must recover the exact stored float4 bits.
assert_eq!(restored, value, "lost f32 precision for {value} -> {text}");
// Fixed 6-decimal rounding is what caused #3931; reject that path when it differs.
if rounded_six != value {
assert_ne!(restored, rounded_six, "still clamped to 6 decimals for {value}");
}
}
}
#[test]
fn pgvector_binary_to_json_preserves_component_precision() {
let values = [0.12345679f32, 1.2345679e-5f32, -2.5f32];
let decoded = decode_pgvector_bytes(&encode_pgvector_bytes(&values)).expect("decode vector");
let json = serde_json::Value::Array(decoded.into_iter().map(pg_vector_element_number).collect());
let arr = json.as_array().expect("vector json array");
assert_eq!(arr.len(), values.len());
for (component, expected) in arr.iter().zip(values) {
let restored: f32 = component.to_string().parse().expect("component parses as f32");
assert_eq!(restored, expected);
}
}
fn decode_hex(hex: &str) -> Vec<u8> {
assert_eq!(hex.len() % 2, 0, "hex input must have an even number of chars");
(0..hex.len()).step_by(2).map(|idx| u8::from_str_radix(&hex[idx..idx + 2], 16).unwrap()).collect()
}
#[test]
fn decodes_postgres_inet_binary_output() {
assert_eq!(
decode_pg_network_address_bytes(&decode_hex("02200004c0a8010a"), false).as_deref(),
Some("192.168.1.10")
);
assert_eq!(
decode_pg_network_address_bytes(&decode_hex("0310001020010db8abcd00120000000000000001"), false).as_deref(),
Some("2001:db8:abcd:12::1/16")
);
assert_eq!(
decode_pg_network_address_bytes(&decode_hex("0340001020010db8abcd00120000000000000001"), false).as_deref(),
Some("2001:db8:abcd:12::1/64")
);
}
#[test]
fn decodes_postgres_cidr_binary_output() {
assert_eq!(
decode_pg_network_address_bytes(&decode_hex("02180104c0a80100"), true).as_deref(),
Some("192.168.1.0/24")
);
assert_eq!(
decode_pg_network_address_bytes(&decode_hex("02200104c0a8010a"), true).as_deref(),
Some("192.168.1.10/32")
);
assert_eq!(
decode_pg_network_address_bytes(&decode_hex("0380011000000000000000000000000000000001"), true).as_deref(),
Some("::1/128")
);
}
#[test]
fn rejects_invalid_postgres_network_binary_output() {
assert_eq!(decode_pg_network_address_bytes(&[], false), None);
assert_eq!(decode_pg_network_address_bytes(&decode_hex("04200004c0a8010a"), false), None);
assert_eq!(decode_pg_network_address_bytes(&decode_hex("02210004c0a8010a"), false), None);
assert_eq!(decode_pg_network_address_bytes(&decode_hex("02200004c0a801"), false), None);
}
#[test]
fn decodes_postgres_macaddr_binary_output() {
assert_eq!(decode_pg_macaddr_bytes(&decode_hex("08002b010203")).as_deref(), Some("08:00:2b:01:02:03"));
assert_eq!(
decode_pg_macaddr_bytes(&decode_hex("08002bfffe010203")).as_deref(),
Some("08:00:2b:ff:fe:01:02:03")
);
assert_eq!(decode_pg_macaddr_bytes(&decode_hex("08002b")), None);
}
#[test]
fn decodes_postgres_bit_string_binary_output() {
assert_eq!(decode_pg_bit_string_bytes(&decode_hex("00000005a8")).as_deref(), Some("10101"));
assert_eq!(decode_pg_bit_string_bytes(&decode_hex("00000009a880")).as_deref(), Some("101010001"));
assert_eq!(decode_pg_bit_string_bytes(&decode_hex("00000000")).as_deref(), Some(""));
assert_eq!(decode_pg_bit_string_bytes(&decode_hex("00000005a8ff")), None);
assert_eq!(decode_pg_bit_string_bytes(&decode_hex("ffffffff")), None);
}
#[test]
fn ewkb_point_with_srid_formats_as_wkt() {
let raw = decode_hex("0101000020E6100000C520B07268195D404E62105839F44340");
assert_eq!(super::super::wkb::wkb_to_wkt(&raw), Some("POINT(116.397 39.908)".to_string()));
}
#[test]
fn ewkb_multi_polygon_formats_as_wkt() {
let raw = decode_hex(
"0106000020E610000002000000010300000001000000050000000000000000005D4000000000000044400000000000405D4000000000000044400000000000405D4000000000008044400000000000005D4000000000008044400000000000005D400000000000004440010300000001000000050000000000000000805D4000000000008043400000000000C05D4000000000008043400000000000C05D4000000000000044400000000000805D4000000000000044400000000000805D400000000000804340",
);
assert_eq!(
super::super::wkb::wkb_to_wkt(&raw),
Some(
"MULTIPOLYGON(((116 40,117 40,117 41,116 41,116 40)),((118 39,119 39,119 40,118 40,118 39)))"
.to_string()
)
);
}
#[test]
fn ewkb_geometry_collection_formats_as_wkt() {
let raw = decode_hex(
"0107000020E61000000200000001010000000000000000005D4000000000000044400102000000020000000000000000405D4000000000008044400000000000805D400000000000004540",
);
assert_eq!(
super::super::wkb::wkb_to_wkt(&raw),
Some("GEOMETRYCOLLECTION(POINT(116 40),LINESTRING(117 41,118 42))".to_string())
);
}
#[test]
fn pg_optional_array_to_json_preserves_text_values_and_nulls() {
let value = pg_optional_array_to_json(
vec![Some("productManager".to_string()), None, Some("projectOwner".to_string())],
serde_json::Value::String,
);
assert_eq!(value, serde_json::json!(["productManager", null, "projectOwner"]));
}
#[test]
fn pg_quote_ident_plain_identifier() {
assert_eq!(pg_quote_ident("public"), "\"public\"");
}
#[test]
fn pg_quote_ident_escapes_double_quotes() {
assert_eq!(pg_quote_ident("my\"schema"), "\"my\"\"schema\"");
}
#[test]
fn pg_quote_ident_empty_string() {
assert_eq!(pg_quote_ident(""), "\"\"");
}
#[test]
fn pg_quote_ident_special_chars() {
// PostgreSQL allows many special chars in quoted identifiers
let ident = "my schema with spaces";
assert_eq!(pg_quote_ident(ident), "\"my schema with spaces\"");
}
#[test]
fn pg_quote_ident_injection_attempt() {
// A malicious schema name that tries to break out of quotes
let malicious = r#"public"; DROP TABLE users; --"#;
let escaped = pg_quote_ident(malicious);
// Double quotes should be doubled, not breaking out
assert_eq!(escaped, r#""public""; DROP TABLE users; --""#);
assert!(escaped.matches('"').count().is_multiple_of(2), "quote count should be even");
}
#[test]
fn postgres_table_access_reads_complete_catalog_acls() {
assert!(POSTGRES_TABLE_OWNER_SQL.contains("acldefault('r', c.relowner)"));
assert!(
POSTGRES_TABLE_ACL_PRIVILEGES_SQL.contains("COALESCE(c.relacl, pg_catalog.acldefault('r', c.relowner))")
);
assert!(POSTGRES_COLUMN_ACL_PRIVILEGES_SQL.contains("aclexplode(a.attacl)"));
assert!(POSTGRES_TABLE_ACL_PRIVILEGES_SQL.contains("acl.grantee = 0 THEN 'PUBLIC'"));
assert!(POSTGRES_COLUMN_ACL_PRIVILEGES_SQL.contains("acl.grantee = 0 THEN 'PUBLIC'"));
assert!(POSTGRES_TABLE_ACL_PRIVILEGES_SQL.contains("pg_get_userbyid(acl.grantor)"));
assert!(POSTGRES_COLUMN_ACL_PRIVILEGES_SQL.contains("pg_get_userbyid(acl.grantor)"));
}
// --- query_result_row_limit ---
#[test]
fn row_limit_uses_max_rows_when_present() {
assert_eq!(query_result_row_limit(Some(50)), 50);
}
#[test]
fn row_limit_falls_back_to_default() {
let default = crate::query::MAX_ROWS;
assert_eq!(query_result_row_limit(None), default);
}
#[test]
fn row_limit_clamps_zero_to_one() {
assert_eq!(query_result_row_limit(Some(0)), 1);
}
#[test]
fn row_limit_allows_max_rows_override() {
assert_eq!(query_result_row_limit(Some(5)), 5);
}
#[test]
fn timestamptz_display_preserves_local_offset() {
let text = format_pg_timestamptz(Local::now());
assert!(!text.ends_with("+00:00") || Local::now().offset().local_minus_utc() == 0);
}
// --- validate_postgres_ssl_paths ---
#[test]
fn ssl_validation_passes_for_clean_url() {
assert!(validate_postgres_ssl_paths("postgres://localhost/db").is_ok());
}
#[test]
fn ssl_validation_passes_for_url_without_query() {
assert!(validate_postgres_ssl_paths("host=localhost dbname=test").is_ok());
}
#[test]
fn ssl_validation_passes_for_irrelevant_params() {
assert!(validate_postgres_ssl_paths("postgres://localhost/db?sslmode=require&connect_timeout=10").is_ok());
}
#[test]
fn ssl_validation_rejects_nonexistent_sslcert_path() {
let result = validate_postgres_ssl_paths("postgres://localhost/db?sslcert=/nonexistent/path/cert.pem");
assert!(result.is_err());
assert!(result.unwrap_err().contains("sslcert"), "error should mention sslcert");
}
#[test]
fn ssl_validation_rejects_nonexistent_sslkey_path() {
let result = validate_postgres_ssl_paths("postgres://localhost/db?sslkey=/nonexistent/path/key.pem");
assert!(result.is_err());
assert!(result.unwrap_err().contains("sslkey"), "error should mention sslkey");
}
#[test]
fn ssl_validation_rejects_nonexistent_sslrootcert_path() {
let result = validate_postgres_ssl_paths("postgres://localhost/db?sslrootcert=/nonexistent/path/root.crt");
assert!(result.is_err());
assert!(result.unwrap_err().contains("sslrootcert"), "error should mention sslrootcert");
}
#[test]
fn ssl_validation_rejects_path_traversal_in_sslcert() {
let result = validate_postgres_ssl_paths("postgres://localhost/db?sslcert=../../../etc/passwd");
assert!(result.is_err());
}
#[test]
fn ssl_validation_handles_url_encoded_ssl_param() {
// %2F = '/', so sslcert=%2Ftmp%2Fcert.pem means sslcert=/tmp/cert.pem
let result = validate_postgres_ssl_paths("postgres://localhost/db?sslcert=%2Fnonexistent%2Fcert.pem");
assert!(result.is_err());
}
#[test]
fn ssl_validation_handles_multiple_params() {
let result =
validate_postgres_ssl_paths("postgres://localhost/db?sslmode=require&sslcert=/nonexistent/cert.pem");
assert!(result.is_err());
}
#[test]
fn postgres_connection_url_strips_ssl_file_params_before_driver_parse() {
let dir = std::env::temp_dir();
let cert = dir.join(format!("dbx-postgres-cert-{}.pem", std::process::id()));
let key = dir.join(format!("dbx-postgres-key-{}.pem", std::process::id()));
let root = dir.join(format!("dbx-postgres-root-{}.pem", std::process::id()));
std::fs::write(&cert, "not a real cert").unwrap();
std::fs::write(&key, "not a real key").unwrap();
std::fs::write(&root, "not a real root").unwrap();
let url = format!(
"postgres://localhost/db?sslmode=verify-full&sslcert={}&sslkey={}&sslrootcert={}&application_name=dbx",
cert.display(),
key.display(),
root.display()
);
let parsed = postgres_connection_url(&url).unwrap();
assert_eq!(parsed.url, "postgres://localhost/db?sslmode=require&application_name=dbx");
assert_eq!(parsed.ssl_files.sslcert.as_deref(), Some(cert.to_str().unwrap()));
assert_eq!(parsed.ssl_files.sslkey.as_deref(), Some(key.to_str().unwrap()));
assert_eq!(parsed.ssl_files.sslrootcert.as_deref(), Some(root.to_str().unwrap()));
assert!(!parsed.accepts_invalid_certs);
assert!(parsed.verifies_hostname);
tokio_postgres::Config::from_str(&parsed.url).unwrap();
let _ = std::fs::remove_file(cert);
let _ = std::fs::remove_file(key);
let _ = std::fs::remove_file(root);
}
#[test]
fn postgres_connection_url_keeps_verify_ca_ca_only_semantics() {
let parsed = postgres_connection_url("postgres://localhost/db?sslmode=verify-ca").unwrap();
assert_eq!(parsed.url, "postgres://localhost/db?sslmode=require");
assert!(!parsed.accepts_invalid_certs);
assert!(!parsed.verifies_hostname);
}
#[test]
fn postgres_connection_url_normalizes_channel_binding_require_to_prefer() {
let parsed =
postgres_connection_url("postgres://localhost/db?sslmode=require&channel_binding=require").unwrap();
assert_eq!(parsed.url, "postgres://localhost/db?sslmode=require&channel_binding=prefer");
// The sanitized URL must be parseable by the driver
tokio_postgres::Config::from_str(&parsed.url).unwrap();
}
#[test]
fn postgres_connection_url_keeps_channel_binding_prefer() {
let parsed = postgres_connection_url("postgres://localhost/db?channel_binding=prefer").unwrap();
assert_eq!(parsed.url, "postgres://localhost/db?channel_binding=prefer");
tokio_postgres::Config::from_str(&parsed.url).unwrap();
}
#[test]
fn postgres_tls_rejects_unpaired_client_cert_and_key() {
let _ = rustls::crypto::aws_lc_rs::default_provider().install_default();
let pg_config = tokio_postgres::Config::from_str("postgres://localhost/db?sslmode=require").unwrap();
let ssl_files =
PostgresSslFiles { sslcert: Some("/tmp/client.crt".to_string()), sslkey: None, sslrootcert: None };
let error = match postgres_tls_config(&pg_config, &ssl_files, true, false) {
Ok(_) => panic!("expected missing sslkey to fail"),
Err(error) => error,
};
assert!(error.contains("sslkey"));
}
#[test]
fn postgres_accept_all_tls_signature_does_not_parse_unverified_cert() {
let verifier = NoPostgresCertVerification { provider: Arc::new(rustls::crypto::aws_lc_rs::default_provider()) };
let malformed_cert = CertificateDer::from(vec![0x30, 0x03, 0x02, 0x01, 0x00]);
assert!(verifier.accept_tls_signature_for_unverified_cert(&malformed_cert).is_ok());
}
#[test]
fn inject_postgres_keepalive_params_preserves_url_fragment() {
let url = "postgres://localhost/app?sslmode=require#read-only";
assert_eq!(
inject_postgres_keepalive_params(url),
"postgres://localhost/app?sslmode=require&keepalives=1&keepalives_idle=30&keepalives_interval=10&keepalives_retries=3#read-only"
);
}
#[test]
fn postgres_cancel_attempt_timeout_is_single_budget() {
assert_eq!(postgres_cancel_attempt_timeout(Duration::from_secs(5), None), Duration::from_secs(5));
assert_eq!(
postgres_cancel_attempt_timeout(
Duration::from_secs(5),
Some(&PostgresCancelContext {
ssl_files: PostgresSslFiles::default(),
accepts_invalid_certs: true,
verifies_hostname: false,
ssl_mode: SslMode::Require,
})
),
Duration::from_secs(5)
);
}
#[test]
fn postgres_checkout_timeout_respects_pool_configuration() {
let manager = deadpool_postgres::Manager::new(tokio_postgres::Config::new(), NoTls);
let pool = Pool::builder(manager)
.runtime(Runtime::Tokio1)
.wait_timeout(Some(Duration::from_secs(10)))
.create_timeout(Some(Duration::from_secs(10)))
.recycle_timeout(Some(Duration::from_secs(10)))
.build()
.expect("build postgres pool");
assert_eq!(effective_postgres_checkout_timeout(&pool, Duration::from_secs(5)), Duration::from_secs(10));
assert_eq!(effective_postgres_checkout_timeout(&pool, Duration::from_secs(15)), Duration::from_secs(15));
}
#[tokio::test]
async fn postgres_session_setup_reports_its_own_timeout_stage() {
let error =
postgres_session_setup_with_timeout(Duration::from_millis(1), std::future::pending::<Result<(), String>>())
.await
.expect_err("pending session setup should time out");
assert!(error.starts_with("PostgreSQL session setup connection timed out"), "{error}");
}
#[tokio::test]
async fn postgres_row_query_timeout_tracks_inactivity_instead_of_total_duration() {
let progress_clock = Arc::new(StreamProgressClock::new());
let progress_clock_for_query = progress_clock.clone();
let future = async {
for _ in 0..4 {
tokio::time::sleep(Duration::from_millis(40)).await;
progress_clock_for_query.mark();
}
Ok::<_, String>(())
};
let result = await_stream_with_progress_timeout(
future,
Some(Duration::from_millis(100)),
progress_clock,
None,
"query inactivity timeout".to_string(),
)
.await;
assert_eq!(result, Ok(()));
}
#[tokio::test]
async fn postgres_row_query_timeout_still_fires_after_no_progress() {
let progress_clock = Arc::new(StreamProgressClock::new());
let error = await_stream_with_progress_timeout(
std::future::pending::<Result<(), String>>(),
Some(Duration::from_millis(10)),
progress_clock,
None,
"query inactivity timeout".to_string(),
)
.await
.expect_err("a stalled query should time out");
assert_eq!(error, "query inactivity timeout");
}
#[test]
fn postgres_cancel_context_omits_disabled_ssl_mode() {
assert!(build_postgres_cancel_context("postgres://localhost/app?sslmode=disable").is_none());
}
#[test]
fn postgres_tls_accepts_invalid_certs_for_require_sslmode() {
let pg_config = tokio_postgres::Config::from_str("postgres://localhost/db?sslmode=require").unwrap();
assert!(postgres_sslmode_accepts_invalid_certs(pg_config.get_ssl_mode()));
}
#[test]
fn postgres_tls_accepts_invalid_certs_for_default_prefer_sslmode() {
let pg_config = tokio_postgres::Config::from_str("postgres://localhost/db").unwrap();
assert!(postgres_sslmode_accepts_invalid_certs(pg_config.get_ssl_mode()));
}
#[test]
fn postgres_tls_keeps_verification_off_only_when_ssl_is_disabled() {
let pg_config = tokio_postgres::Config::from_str("postgres://localhost/db?sslmode=disable").unwrap();
assert!(!postgres_sslmode_accepts_invalid_certs(pg_config.get_ssl_mode()));
}
// --- SQL generation ---
#[test]
fn postgres_tables_sql_contains_expected_columns() {
let sql = postgres_tables_sql(Some(500), 0);
assert!(sql.contains("table_name"));
assert!(sql.contains("table_type"));
assert!(sql.contains("table_comment"));
assert!(sql.contains("pg_catalog.pg_inherits"));
assert!(sql.contains("parent_schema"));
assert!(sql.contains("parent_name"));
assert!(sql.contains("pc.relkind = 'p'"));
assert!(sql.contains("$1"));
assert!(sql.contains("BASE TABLE"));
assert!(sql.contains("VIEW"));
assert!(sql.contains("MATERIALIZED_VIEW"));
assert!(sql.contains("FOREIGN TABLE"));
}
#[test]
fn postgres_table_comment_sql_targets_single_table() {
let sql = postgres_table_comment_sql();
assert!(sql.contains("obj_description(c.oid)"));
assert!(sql.contains("n.nspname = $1"));
assert!(sql.contains("c.relname = $2"));
assert!(sql.contains("LIMIT 1"));
assert!(!sql.contains("ORDER BY"));
}
#[test]
fn postgres_table_partition_sql_tracks_parents_bounds_and_local_objects() {
let relation_sql = postgres_table_partition_relation_sql();
let info_sql = postgres_table_partition_info_sql();
let local_objects_sql = postgres_table_partition_local_objects_sql();
assert!(relation_sql.contains("row_to_json(c)->>'relispartition'"));
assert!(relation_sql.contains("c.relkind IN ('r','p')"));
assert!(info_sql.contains("pg_catalog.pg_get_expr(c.relpartbound, c.oid, true)"));
assert!(info_sql.contains("pg_catalog.pg_get_partkeydef(c.oid)"));
assert!(info_sql.contains("pg_catalog.pg_inherits"));
assert!(info_sql.contains("parent_schema"));
assert!(info_sql.contains("parent_table"));
assert!(local_objects_sql.contains("row_to_json(con)->>'conparentid'"));
assert!(local_objects_sql.contains("con.contype IN ('p','f')"));
assert!(local_objects_sql.contains("i.inhrelid = idx.oid"));
}
#[test]
fn postgres_column_metadata_reads_identity_extra() {
assert!(POSTGRES_COLUMNS_SQL.contains("a.attidentity"));
assert!(POSTGRES_COLUMNS_SQL.contains("pg_sequence"));
assert!(POSTGRES_COLUMNS_SQL.contains("generated by default as identity"));
assert!(POSTGRES_COLUMNS_SQL.contains("generated always as identity"));
assert!(POSTGRES_COLUMNS_SQL.contains("COALESCE(c.is_nullable = 'YES', NOT a.attnotnull)"));
assert!(POSTGRES_COLUMNS_SQL.contains("LEFT JOIN information_schema.columns"));
assert!(POSTGRES_COLUMNS_SQL.contains("pg_enum"));
assert!(POSTGRES_COLUMNS_SQL.contains("AS enum_values"));
}
#[test]
fn postgres_column_metadata_has_opengauss_compatible_fallback() {
assert!(!POSTGRES_COLUMNS_COMPAT_SQL.contains("a.attidentity"));
assert!(!POSTGRES_COLUMNS_COMPAT_SQL.contains("pg_sequence"));
assert!(POSTGRES_COLUMNS_COMPAT_SQL.contains("NULL::text AS column_extra"));
assert!(POSTGRES_COLUMNS_COMPAT_SQL.contains("col_description"));
assert!(POSTGRES_COLUMNS_COMPAT_SQL.contains("COALESCE(c.is_nullable = 'YES', NOT a.attnotnull)"));
assert!(POSTGRES_COLUMNS_COMPAT_SQL.contains("LEFT JOIN information_schema.columns"));
assert!(POSTGRES_COLUMNS_COMPAT_SQL.contains("NULL::text AS enum_values"));
assert!(!POSTGRES_COLUMNS_COMPAT_SQL.contains("pg_enum"));
}
#[test]
fn postgres_column_metadata_has_information_schema_fallback() {
assert!(POSTGRES_COLUMNS_INFORMATION_SCHEMA_SQL.contains("information_schema.columns"));
assert!(POSTGRES_COLUMNS_INFORMATION_SCHEMA_SQL.contains("information_schema.table_constraints"));
assert!(POSTGRES_COLUMNS_INFORMATION_SCHEMA_SQL.contains("information_schema.key_column_usage"));
assert!(POSTGRES_COLUMNS_INFORMATION_SCHEMA_SQL.contains("NULL::text AS enum_values"));
assert!(!POSTGRES_COLUMNS_INFORMATION_SCHEMA_SQL.contains("pg_attribute"));
assert!(!POSTGRES_COLUMNS_INFORMATION_SCHEMA_SQL.contains("regclass"));
}
#[test]
fn opengauss_sequence_metadata_uses_compatible_information_schema_view() {
let sql = opengauss_sequences_sql();
assert!(sql.contains("information_schema.sequences"));
assert!(sql.contains("s.sequence_schema = $1"));
assert!(sql.contains("c.relkind IN ('S','L','z','Z')"));
assert!(sql.contains("sequence_name"));
assert!(sql.contains("start_value"));
assert!(sql.contains("minimum_value"));
assert!(sql.contains("maximum_value"));
assert!(sql.contains("increment"));
assert!(sql.contains("cycle_option"));
assert!(!sql.contains("pg_sequence s"));
}
#[test]
fn opengauss_sequence_last_values_extract_record_field_as_text() {
let sql = opengauss_sequence_last_values_sql();
assert!(sql.contains("(pg_sequence_last_value(c.oid)).last_value::text"));
assert!(sql.contains("c.relkind IN ('S','L','z','Z')"));
assert!(sql.contains("n.nspname = $1"));
}
#[test]
fn postgres_sequence_last_values_are_read_as_text() {
assert!(postgres_sequence_last_values_sql().contains("pg_sequence_last_value(c.oid)::text"));
}
#[test]
fn extension_member_query_filters_only_owned_relations_and_routines() {
let sql = list_extension_member_objects_sql();
assert!(sql.contains("d.classid = 'pg_catalog.pg_class'::regclass"));
assert!(sql.contains("d.classid = 'pg_catalog.pg_proc'::regclass"));
assert!(sql.contains("d.refclassid = 'pg_catalog.pg_extension'::regclass"));
assert!(sql.contains("d.deptype = 'e'"));
assert!(sql.contains("pg_get_function_identity_arguments(p.oid)"));
assert!(!sql.contains("d.deptype = 'x'"));
}
#[tokio::test]
async fn postgres_column_metadata_query_returns_enum_values_against_real_postgres() {
let Some(container) = start_docker_postgres().await else {
return;
};
let pool = connect(&container.url(), Duration::from_secs(5)).await.expect("connect postgres");
let schema = format!("dbx_enum_meta_{}", std::process::id());
let schema_ident = format!("\"{}\"", schema.replace('\"', "\"\""));
let table = format!("{schema_ident}.orders");
let type_ident = format!("{schema_ident}.\"status\"");
execute_query(&pool, &format!("CREATE SCHEMA {schema_ident}")).await.expect("create schema");
execute_query(&pool, &format!("CREATE TYPE {type_ident} AS ENUM ('pending', 'active', 'archived')"))
.await
.expect("create enum type");
execute_query(&pool, &format!("CREATE TABLE {table} (id integer PRIMARY KEY, state {type_ident} NOT NULL)"))
.await
.expect("create table");
let client =
checkout_postgres_client(&pool, None, crate::db::connection_timeout()).await.expect("checkout client");
let columns =
get_columns_with_sql(&client, POSTGRES_COLUMNS_SQL, &schema, "orders").await.expect("primary columns");
assert_eq!(
state_enum_values(&columns),
Some(vec!["pending".to_string(), "active".to_string(), "archived".to_string()])
);
}
#[tokio::test]
#[ignore = "requires DBX_TEST_POSTGRES_URL pointing at a writable PostgreSQL database"]
async fn postgres_column_metadata_decode_type_mismatch_uses_fallbacks() {
let url = std::env::var("DBX_TEST_POSTGRES_URL").expect("DBX_TEST_POSTGRES_URL");
let pool = connect(&url, std::time::Duration::from_secs(5)).await.expect("connect postgres");
let client =
checkout_postgres_client(&pool, None, std::time::Duration::from_secs(5)).await.expect("checkout postgres");
let row = client
.query_one(
"SELECT \
1::int4 AS column_name, \
'text'::text AS full_type, \
'YES'::text AS is_nullable, \
NULL::text AS column_default, \
1::int4 AS is_pk, \
NULL::text AS column_comment, \
NULL::text AS column_extra, \
NULL::int4 AS numeric_precision, \
NULL::int4 AS numeric_scale, \
NULL::int4 AS character_maximum_length",
&[],
)
.await
.expect("query mismatched metadata row");
let info = column_info_from_row(&row);
// int4 column_name should be converted to string "1" instead of panicking
assert_eq!(info.name, "1");
// text 'YES' is not a standard bool, pg_row_try_bool falls back to string match
assert!(info.is_nullable);
// int4 1 should be interpreted as true for is_primary_key
assert!(info.is_primary_key);
}
#[tokio::test]
#[ignore = "requires DBX_TEST_POSTGRES_URL pointing at a writable PostgreSQL database"]
async fn postgres_partition_metadata_renders_replayable_children_and_subpartitions() {
let url = std::env::var("DBX_TEST_POSTGRES_URL").expect("DBX_TEST_POSTGRES_URL");
let pool = connect(&url, Duration::from_secs(5)).await.expect("connect postgres");
let schema = format!("dbx_partition_meta_{}", std::process::id());
let schema_ident = pg_quote_ident(&schema);
let replay_schema = format!("{schema}_replay");
let replay_schema_ident = pg_quote_ident(&replay_schema);
let parent = format!("{schema_ident}.parent");
let child = format!("{schema_ident}.child");
let default_child = format!("{schema_ident}.child_default");
let subpartition = format!("{schema_ident}.subpartition");
let inherited_parent = format!("{schema_ident}.inherited_parent");
let inherited_child = format!("{schema_ident}.inherited_child");
execute_query(&pool, &format!("CREATE SCHEMA {schema_ident}")).await.expect("create schema");
let client = pool.get().await.expect("get postgres client");
client
.batch_execute(&format!(
"CREATE TABLE {parent} (id integer, payload text) PARTITION BY RANGE (id); \
CREATE TABLE {child} PARTITION OF {parent} (PRIMARY KEY (id)) FOR VALUES FROM (1) TO (10); \
CREATE INDEX child_payload_idx ON {child} (payload); \
CREATE TABLE {default_child} PARTITION OF {parent} DEFAULT; \
CREATE TABLE {subpartition} PARTITION OF {parent} FOR VALUES FROM (10) TO (20) PARTITION BY HASH (payload); \
CREATE TABLE {inherited_parent} (id integer, bucket integer, PRIMARY KEY (id, bucket)) PARTITION BY RANGE (bucket); \
CREATE TABLE {inherited_child} PARTITION OF {inherited_parent} FOR VALUES FROM (1) TO (10)"
))
.await
.expect("create partitioned tables");
let parent_info = get_table_partition_info(&pool, &schema, "parent").await.expect("parent metadata");
let child_info = get_table_partition_info(&pool, &schema, "child").await.expect("child metadata");
let default_info = get_table_partition_info(&pool, &schema, "child_default").await.expect("default metadata");
let subpartition_info =
get_table_partition_info(&pool, &schema, "subpartition").await.expect("subpartition metadata");
let child_local_objects =
get_table_partition_local_objects(&pool, &schema, "child").await.expect("child local objects");
let inherited_child_local_objects = get_table_partition_local_objects(&pool, &schema, "inherited_child")
.await
.expect("inherited child local objects");
let parent_ddl = crate::schema::pg_ddl(&pool, &schema, "parent").await.expect("parent ddl");
let child_ddl = crate::schema::pg_ddl(&pool, &schema, "child").await.expect("child ddl");
let default_ddl = crate::schema::pg_ddl(&pool, &schema, "child_default").await.expect("default ddl");
let subpartition_ddl = crate::schema::pg_ddl(&pool, &schema, "subpartition").await.expect("subpartition ddl");
let inherited_parent_ddl =
crate::schema::pg_ddl(&pool, &schema, "inherited_parent").await.expect("inherited parent ddl");
let inherited_child_ddl =
crate::schema::pg_ddl(&pool, &schema, "inherited_child").await.expect("inherited child ddl");
execute_query(&pool, &format!("CREATE SCHEMA {replay_schema_ident}")).await.expect("create replay schema");
for ddl in
[&parent_ddl, &child_ddl, &default_ddl, &subpartition_ddl, &inherited_parent_ddl, &inherited_child_ddl]
{
client
.batch_execute(&ddl.replace(&schema_ident, &replay_schema_ident))
.await
.unwrap_or_else(|error| panic!("replay partition ddl failed: {error}; ddl: {ddl}"));
}
client
.batch_execute(&format!("DROP SCHEMA {schema_ident} CASCADE; DROP SCHEMA {replay_schema_ident} CASCADE"))
.await
.expect("drop schemas");
assert_eq!(parent_info.key.as_deref(), Some("RANGE (id)"));
assert!(!parent_info.is_partition);
assert_eq!(child_info.parent_schema.as_deref(), Some(schema.as_str()));
assert_eq!(child_info.parent_table.as_deref(), Some("parent"));
assert_eq!(child_info.bound.as_deref(), Some("FOR VALUES FROM (1) TO (10)"));
assert!(child_local_objects.has_primary_key);
assert!(child_local_objects.indexes.contains("child_payload_idx"));
assert!(!inherited_child_local_objects.has_primary_key);
assert!(inherited_child_local_objects.indexes.is_empty());
assert_eq!(default_info.bound.as_deref(), Some("DEFAULT"));
assert_eq!(subpartition_info.key.as_deref(), Some("HASH (payload)"));
assert!(parent_ddl.contains(") PARTITION BY RANGE (id);"), "ddl: {parent_ddl}");
assert!(child_ddl.contains("CREATE TABLE"), "ddl: {child_ddl}");
assert!(child_ddl.contains("PARTITION OF"), "ddl: {child_ddl}");
assert!(child_ddl.contains("PRIMARY KEY (\"id\")"), "ddl: {child_ddl}");
assert!(child_ddl.contains("FOR VALUES FROM (1) TO (10);"), "ddl: {child_ddl}");
assert!(child_ddl.contains("CREATE INDEX \"child_payload_idx\""), "ddl: {child_ddl}");
assert!(default_ddl.contains(" DEFAULT;"), "ddl: {default_ddl}");
assert!(
subpartition_ddl.contains("FOR VALUES FROM (10) TO (20) PARTITION BY HASH (payload);"),
"ddl: {subpartition_ddl}"
);
assert!(!inherited_child_ddl.contains("PRIMARY KEY"), "ddl: {inherited_child_ddl}");
assert!(!inherited_child_ddl.contains("CREATE INDEX"), "ddl: {inherited_child_ddl}");
}
#[tokio::test]
#[ignore = "requires DBX_TEST_POSTGRES_URL pointing at a writable PostgreSQL database"]
async fn postgres_schema_context_prioritizes_selected_schema_and_cleans_up() {
let url = std::env::var("DBX_TEST_POSTGRES_URL").expect("DBX_TEST_POSTGRES_URL");
let pool = connect(&url, Duration::from_secs(5)).await.expect("connect postgres");
let suffix = format!("{}_{}", std::process::id(), uuid::Uuid::new_v4().simple());
let schema = format!("dbx_issue_830_\"{suffix}");
let schema_ident = pg_quote_ident(&schema);
let helper = format!("dbx_issue_830_public_{suffix}");
let helper_ident = pg_quote_ident(&helper);
let initial_path = execute_query(&pool, "SHOW search_path").await.expect("read initial search_path");
let initial_path_value = initial_path.rows[0][0].as_str().expect("search_path string").to_string();
let client = pool.get().await.expect("get setup client");
client
.batch_execute(&format!(
"CREATE SCHEMA {schema_ident}; \
CREATE TABLE {schema_ident}.pg_settings(marker text); \
INSERT INTO {schema_ident}.pg_settings VALUES ('selected-schema'); \
CREATE FUNCTION public.{helper_ident}() RETURNS text \
LANGUAGE SQL IMMUTABLE AS $$ SELECT 'public-fallback'::text $$"
))
.await
.expect("create search_path fixtures");
drop(client);
let query_sql = format!("SELECT marker, {helper_ident}() AS helper FROM pg_settings");
let ordinary_result = execute_query_with_schema(&pool, &schema, &query_sql).await;
let path_after_ordinary = execute_query(&pool, "SHOW search_path").await;
let mut streamed_rows = Vec::new();
let streaming_result = stream_select_query_with_cancel(
&pool,
Some(&schema),
&[],
&query_sql,
None,
None,
DbOperationBudget::with_defaults(),
None,
|item| {
if let PostgresQueryStreamItem::Row(row) = item {
streamed_rows.push(row);
}
Ok(())
},
)
.await;
let path_after_streaming = execute_query(&pool, "SHOW search_path").await;
let transaction_cleanup = async {
let client = pool.get().await.map_err(|error| error.to_string())?;
client
.execute(&postgres_set_search_path_sql(&schema, PostgresSearchPathContext::Transaction), &[])
.await
.map_err(pg_error_to_string)?;
let selected: String = client
.query_one("SELECT marker FROM pg_settings", &[])
.await
.map_err(pg_error_to_string)?
.try_get(0)
.map_err(pg_error_to_string)?;
client.execute("RESET search_path", &[]).await.map_err(pg_error_to_string)?;
let after_reset: String = client
.query_one("SHOW search_path", &[])
.await
.map_err(pg_error_to_string)?
.try_get(0)
.map_err(pg_error_to_string)?;
client.execute("BEGIN", &[]).await.map_err(pg_error_to_string)?;
client
.execute(&postgres_set_search_path_sql(&schema, PostgresSearchPathContext::LocalTransaction), &[])
.await
.map_err(pg_error_to_string)?;
let local_selected: String = client
.query_one("SELECT marker FROM pg_settings", &[])
.await
.map_err(pg_error_to_string)?
.try_get(0)
.map_err(pg_error_to_string)?;
client.execute("COMMIT", &[]).await.map_err(pg_error_to_string)?;
let after_commit: String = client
.query_one("SHOW search_path", &[])
.await
.map_err(pg_error_to_string)?
.try_get(0)
.map_err(pg_error_to_string)?;
Ok::<_, String>((selected, after_reset, local_selected, after_commit))
}
.await;
let cleanup_client = pool.get().await.expect("get cleanup client");
cleanup_client
.batch_execute(&format!("DROP FUNCTION public.{helper_ident}(); DROP SCHEMA {schema_ident} CASCADE"))
.await
.expect("clean search_path fixtures");
let ordinary = ordinary_result.expect("ordinary schema query");
assert_eq!(
ordinary.rows,
vec![vec![serde_json::json!("selected-schema"), serde_json::json!("public-fallback")]]
);
assert_eq!(path_after_ordinary.expect("path after ordinary query").rows, initial_path.rows);
assert_eq!(streaming_result.expect("streaming schema query"), 1);
assert_eq!(
streamed_rows,
vec![vec![serde_json::json!("selected-schema"), serde_json::json!("public-fallback")]]
);
assert_eq!(path_after_streaming.expect("path after streaming query").rows, initial_path.rows);
let (selected, after_reset, local_selected, after_commit) = transaction_cleanup.expect("transaction cleanup");
assert_eq!(selected, "selected-schema");
assert_eq!(local_selected, "selected-schema");
assert_eq!(after_reset, initial_path_value);
assert_eq!(after_commit, initial_path_value);
}
#[test]
fn postgres_index_metadata_has_legacy_catalog_fallback() {
assert!(POSTGRES_INDEXES_SQL.contains("ix.indnkeyatts"));
assert!(!POSTGRES_INDEXES_COMPAT_SQL.contains("ix.indnkeyatts"));
assert!(POSTGRES_INDEXES_COMPAT_SQL.contains("NULL::smallint AS nkeyatts"));
assert!(!POSTGRES_INDEXES_COMPAT_SQL.contains("LATERAL"));
assert!(!POSTGRES_INDEXES_COMPAT_SQL.contains("WITH ORDINALITY"));
assert!(POSTGRES_INDEXES_COMPAT_SQL.contains("generate_series"));
assert!(POSTGRES_INDEXES_COMPAT_SQL.contains("string_to_array(ix.indkey::text, ' ')"));
}
#[test]
fn postgres_owner_metadata_casts_relkind_to_text() {
assert!(POSTGRES_OWNERS_SQL.contains("c.relkind::text AS relkind"));
assert!(POSTGRES_OWNERS_SQL.contains("c.relkind IN ('r', 'v', 'm', 'S', 'f', 'p')"));
}
#[test]
fn postgres_owner_object_type_maps_relkind_codes() {
assert_eq!(postgres_owner_object_type("r"), "TABLE");
assert_eq!(postgres_owner_object_type("v"), "VIEW");
assert_eq!(postgres_owner_object_type("m"), "MATERIALIZED_VIEW");
assert_eq!(postgres_owner_object_type("S"), "SEQUENCE");
assert_eq!(postgres_owner_object_type("f"), "FOREIGN TABLE");
assert_eq!(postgres_owner_object_type("p"), "PARTITIONED TABLE");
assert_eq!(postgres_owner_object_type("?"), "?");
}
#[test]
fn list_objects_sql_includes_routines() {
let sql = list_objects_sql(true, true, false, true);
assert!(sql.contains("pg_catalog.pg_class"));
assert!(sql.contains("pg_catalog.pg_proc"));
assert!(sql.contains("pg_catalog.pg_inherits"));
assert!(sql.contains("parent_schema"));
assert!(sql.contains("parent_name"));
assert!(sql.contains("NULL::text AS signature"));
assert!(sql.contains("pg_get_function_identity_arguments(p.oid) AS signature"));
assert!(sql.contains("pc.relkind = 'p'"));
assert!(sql.contains("pg_stat_file"));
assert!(sql.contains("pg_xact_commit_timestamp"));
assert!(sql.contains("'PROCEDURE'"));
assert!(sql.contains("'FUNCTION'"));
}
#[test]
fn list_objects_sql_without_timestamps_omits_stat_file() {
let sql = list_objects_sql(false, true, false, true);
assert!(!sql.contains("pg_stat_file"));
assert!(sql.contains("NULL::text AS created_at"));
assert!(sql.contains("NULL::text AS updated_at"));
}
#[test]
fn redshift_compatible_list_objects_sql_uses_legacy_argument_formatter() {
let sql = list_objects_sql(false, false, false, false);
assert!(sql.contains("pg_get_function_arguments(p.oid) AS signature"));
assert!(!sql.contains("pg_get_function_identity_arguments"));
}
#[test]
fn redshift_columns_sql_uses_simple_information_schema_metadata() {
let sql = redshift_columns_sql("tenant's", "orders");
assert!(sql.contains("FROM information_schema.columns c"));
assert!(sql.contains("c.table_schema = 'tenant''s'"));
assert!(sql.contains("c.table_name = 'orders'"));
assert!(!sql.contains("pg_attribute"));
assert!(!sql.contains("pg_index"));
assert!(!sql.contains('$'));
}
#[test]
fn redshift_columns_from_text_result_preserves_basic_metadata() {
let result = QueryResult {
columns: vec![
"column_name".to_string(),
"full_type".to_string(),
"is_nullable".to_string(),
"column_default".to_string(),
"numeric_precision".to_string(),
"numeric_scale".to_string(),
"character_maximum_length".to_string(),
],
column_types: Vec::new(),
column_sortables: Vec::new(),
spatial_columns: Vec::new(),
spatial_values: Vec::new(),
rows: vec![vec![
serde_json::json!("amount"),
serde_json::json!("numeric"),
serde_json::json!("NO"),
serde_json::Value::Null,
serde_json::json!(18),
serde_json::json!(2),
serde_json::Value::Null,
]],
affected_rows: 0,
execution_time_ms: 0,
truncated: false,
session_id: None,
has_more: false,
elasticsearch_raw_body: None,
messages: Vec::new(),
};
let columns = redshift_columns_from_query_result(result);
assert_eq!(columns.len(), 1);
assert_eq!(columns[0].name, "amount");
assert_eq!(columns[0].data_type, "numeric");
assert!(!columns[0].is_nullable);
assert_eq!(columns[0].numeric_precision, Some(18));
assert_eq!(columns[0].numeric_scale, Some(2));
assert_eq!(columns[0].character_maximum_length, None);
assert!(!columns[0].is_primary_key);
}
#[test]
fn function_identity_arguments_probe_uses_pg_proc() {
let sql = postgres_has_function_identity_arguments_sql();
assert!(sql.contains("pg_catalog.pg_proc"));
assert!(sql.contains("n.nspname = 'pg_catalog'"));
assert!(sql.contains("p.proname = 'pg_get_function_identity_arguments'"));
}
#[test]
fn both_list_objects_sql_variants_use_parameter() {
assert!(list_objects_sql(true, true, true, true).contains("$1"));
assert!(list_objects_sql(false, true, true, true).contains("$1"));
assert!(list_objects_sql(true, true, false, true).contains("$1"));
assert!(list_objects_sql(false, true, false, true).contains("$1"));
assert!(list_objects_sql(true, false, true, true).contains("$1"));
assert!(list_objects_sql(false, false, true, true).contains("$1"));
assert!(list_objects_sql(true, false, false, true).contains("$1"));
assert!(list_objects_sql(false, false, false, true).contains("$1"));
}
#[test]
fn both_list_objects_sql_variants_include_pg_proc() {
assert!(list_objects_sql(true, true, true, true).contains("pg_catalog.pg_proc"));
assert!(list_objects_sql(false, true, true, true).contains("pg_catalog.pg_proc"));
assert!(list_objects_sql(true, true, false, true).contains("pg_catalog.pg_proc"));
assert!(list_objects_sql(false, true, false, true).contains("pg_catalog.pg_proc"));
assert!(list_objects_sql(true, false, true, true).contains("pg_catalog.pg_proc"));
assert!(list_objects_sql(false, false, true, true).contains("pg_catalog.pg_proc"));
assert!(list_objects_sql(true, false, false, true).contains("pg_catalog.pg_proc"));
assert!(list_objects_sql(false, false, false, true).contains("pg_catalog.pg_proc"));
}
#[test]
fn legacy_list_objects_sql_avoids_pg11_proc_kind_column() {
let sql = list_objects_sql(true, false, false, true);
assert!(!sql.contains("p.prokind"));
assert!(!sql.contains("p.prosp"));
assert!(sql.contains("NOT p.proisagg"));
assert!(sql.contains("NOT p.proiswindow"));
assert!(sql.contains("pg_get_function_identity_arguments(p.oid) AS signature"));
assert!(sql.contains("'FUNCTION' AS object_type"));
assert!(!sql.contains("'PROCEDURE'"));
}
#[test]
fn gaussdb_compatible_list_objects_sql_uses_prosp_when_prokind_is_missing() {
let sql = list_objects_sql(true, false, true, true);
assert!(!sql.contains("p.prokind"));
assert!(sql.contains("CASE WHEN p.prosp THEN 'PROCEDURE' ELSE 'FUNCTION' END AS object_type"));
assert!(sql.contains("CASE WHEN p.prosp THEN 2 ELSE 3 END AS sort_order"));
assert!(sql.contains("NOT p.proisagg"));
assert!(sql.contains("NOT p.proiswindow"));
assert!(sql.contains("pg_get_function_identity_arguments(p.oid) AS signature"));
}
#[test]
fn gaussdb_compatible_list_objects_sql_uses_prosp_with_prokind_when_available() {
let sql = list_objects_sql(true, true, true, true);
assert!(
sql.contains("CASE WHEN p.prokind = 'p' OR p.prosp THEN 'PROCEDURE' ELSE 'FUNCTION' END AS object_type")
);
assert!(sql.contains("CASE WHEN p.prokind = 'p' OR p.prosp THEN 2 ELSE 3 END AS sort_order"));
assert!(sql.contains("p.prokind IN ('p','f') OR p.prosp"));
assert!(sql.contains("pg_get_function_identity_arguments(p.oid) AS signature"));
}
#[test]
fn postgres_functions_sql_uses_proc_kind_when_available() {
let sql = postgres_functions_sql(true);
assert!(sql.contains("p.prokind IN ('f', 'p')"));
assert!(sql.contains("WHEN 'p' THEN 'PROCEDURE'"));
assert!(!sql.contains("p.proisagg"));
assert!(!sql.contains("p.proiswindow"));
}
#[test]
fn legacy_postgres_functions_sql_avoids_proc_kind_column() {
let sql = postgres_functions_sql(false);
assert!(!sql.contains("p.prokind"));
assert!(sql.contains("NOT p.proisagg"));
assert!(sql.contains("NOT p.proiswindow"));
assert!(sql.contains("'FUNCTION'"));
assert!(!sql.contains("'PROCEDURE'"));
}
#[test]
fn postgres_proc_has_prokind_sql_checks_catalog_attribute() {
let sql = postgres_proc_has_prokind_sql();
assert!(sql.contains("pg_catalog.pg_attribute"));
assert!(sql.contains("'pg_catalog.pg_proc'::regclass"));
assert!(sql.contains("attname = 'prokind'"));
}
#[test]
fn postgres_proc_has_prosp_sql_checks_catalog_attribute() {
let sql = postgres_proc_has_prosp_sql();
assert!(sql.contains("pg_catalog.pg_attribute"));
assert!(sql.contains("'pg_catalog.pg_proc'::regclass"));
assert!(sql.contains("attname = 'prosp'"));
}
#[test]
fn postgres_trigger_definitions_sql_excludes_internal_triggers() {
let sql = postgres_trigger_definitions_sql();
assert!(sql.contains("pg_catalog.pg_get_triggerdef(t.oid, true) AS trigger_definition"));
assert!(sql.contains("NOT t.tgisinternal"));
assert!(sql.contains("ORDER BY t.tgname, t.oid"));
}
#[test]
fn transaction_recovery_statement_detection_matches_common_postgres_commands() {
assert!(is_transaction_recovery_statement("ROLLBACK"));
assert!(is_transaction_recovery_statement("rollback work"));
assert!(is_transaction_recovery_statement("ABORT TRANSACTION"));
assert!(is_transaction_recovery_statement("commit"));
assert!(is_transaction_recovery_statement("END"));
}
#[test]
fn transaction_recovery_statement_detection_ignores_regular_queries() {
assert!(!is_transaction_recovery_statement("SELECT 1"));
assert!(!is_transaction_recovery_statement("BEGIN"));
assert!(!is_transaction_recovery_statement("UPDATE users SET name = 'dbx'"));
}
#[test]
fn postgres_ddl_detection_covers_schema_changing_statements() {
assert!(invalidates_postgres_statement_cache("ALTER TABLE users ADD COLUMN email text"));
assert!(invalidates_postgres_statement_cache(" CREATE INDEX idx_users_email ON users(email)"));
assert!(invalidates_postgres_statement_cache("COMMENT ON COLUMN users.email IS 'Email'"));
assert!(invalidates_postgres_statement_cache("DROP TABLE users"));
assert!(invalidates_postgres_statement_cache("TRUNCATE users"));
assert!(invalidates_postgres_statement_cache("REINDEX TABLE users"));
assert!(invalidates_postgres_statement_cache("VACUUM users"));
}
#[test]
fn postgres_ddl_detection_ignores_regular_dml_and_selects() {
assert!(!invalidates_postgres_statement_cache("SELECT * FROM users"));
assert!(!invalidates_postgres_statement_cache("UPDATE users SET name = 'Ada'"));
assert!(!invalidates_postgres_statement_cache("INSERT INTO users(name) VALUES ('Ada')"));
assert!(!invalidates_postgres_statement_cache("DELETE FROM users WHERE id = 1"));
}
#[test]
fn postgres_stale_cache_retry_uses_structured_fields_for_localized_errors() {
assert!(should_retry_postgres_stale_cache_fields(
Some("0A000"),
Some("RevalidateCachedQuery"),
"已缓冲的计划不能改变结果类型",
));
assert!(should_retry_postgres_stale_cache_fields(None, None, "cached plan must not change result type",));
}
#[test]
fn postgres_stale_cache_retry_rejects_other_feature_errors() {
assert!(!should_retry_postgres_stale_cache_fields(Some("0A000"), Some("CheckFeatureSupport"), "不支持该功能",));
assert!(!should_retry_postgres_stale_cache_fields(
Some("23505"),
Some("RevalidateCachedQuery"),
"duplicate key value violates unique constraint",
));
}
// --- execute_batch ---
#[tokio::test]
async fn execute_batch_empty_statements_returns_ok() {
// Empty input should not error or try to connect
// We can't test with a real pool, but we can verify the empty-early-return logic
// by testing that an empty Vec doesn't need a pool reference
let statements: Vec<String> = vec![];
// This test validates the early return logic at code review level
// Actual execution requires a pool; we just verify the empty path exists
assert!(statements.is_empty());
}
#[tokio::test]
async fn execute_batch_whitespace_only_is_filtered() {
let statements = [" ".to_string(), "\t\n".to_string(), "".to_string()];
let combined = statements.iter().map(|s| s.trim()).filter(|s| !s.is_empty()).collect::<Vec<_>>().join(";\n");
assert!(combined.is_empty());
}
#[test]
fn execute_batch_joins_with_semicolons() {
let statements = ["SELECT 1".to_string(), "SELECT 2".to_string()];
let combined = statements.iter().map(|s| s.trim()).filter(|s| !s.is_empty()).collect::<Vec<_>>().join(";\n");
assert_eq!(combined, "SELECT 1;\nSELECT 2");
}
// --- SET timezone escaping ---
#[test]
fn timezone_single_quotes_are_doubled() {
let tz = "UTC";
let escaped = tz.replace('\'', "''");
assert_eq!(escaped, "UTC");
}
#[test]
fn timezone_with_quote_is_escaped() {
let tz = "Some'Zone";
let escaped = tz.replace('\'', "''");
assert_eq!(escaped, "Some''Zone");
}
// --- pg_url_has_timezone_setting ---
#[test]
fn url_without_timezone_returns_false() {
assert!(!pg_url_has_timezone_setting("postgres://localhost/db"));
assert!(!pg_url_has_timezone_setting("postgres://localhost/db?sslmode=require"));
}
#[test]
fn url_with_options_timezone_returns_true() {
assert!(pg_url_has_timezone_setting("postgres://localhost/db?options=-c timezone=Asia/Shanghai"));
}
#[test]
fn url_with_url_encoded_timezone_returns_true() {
assert!(pg_url_has_timezone_setting("postgres://localhost/db?options=-c%20timezone%3DUTC"));
}
#[test]
fn url_with_uppercase_timezone_returns_true() {
assert!(pg_url_has_timezone_setting("postgres://localhost/db?options=--TimeZone=UTC"));
}
#[test]
fn unrelated_timezone_text_is_not_treated_as_explicit() {
assert!(!pg_url_has_timezone_setting("postgres://localhost/db?timezone=UTC"));
assert!(!pg_url_has_timezone_setting(
"postgres://localhost/db?application_name=timezone%3DUTC&options=-c%20search_path%3Dpublic"
));
}
#[test]
fn postgres_timezone_candidates_include_known_tzdata_aliases() {
assert_eq!(postgres_timezone_candidates("Europe/Kyiv"), vec!["Europe/Kyiv", "Europe/Kiev"]);
assert_eq!(postgres_timezone_candidates("Asia/Kolkata"), vec!["Asia/Kolkata", "Asia/Calcutta"]);
assert_eq!(postgres_timezone_candidates("America/New_York"), vec!["America/New_York"]);
}
#[tokio::test]
#[ignore = "requires DBX_TEST_POSTGRES_URL pointing at a PostgreSQL database"]
async fn automatic_invalid_timezone_keeps_connected_server_default() {
let url = std::env::var("DBX_TEST_POSTGRES_URL").expect("DBX_TEST_POSTGRES_URL");
let pool = connect_with_local_timezone(&url, Duration::from_secs(5), "Invalid/DBX_Timezone")
.await
.expect("automatic local timezone rejection must not reject a valid connection");
let client = pool.get().await.expect("checkout postgres");
let timezone: String = client.query_one("SHOW timezone", &[]).await.unwrap().get(0);
assert_ne!(timezone, "Invalid/DBX_Timezone");
}
#[tokio::test]
#[ignore = "requires DBX_TEST_POSTGRES_URL pointing at a PostgreSQL database"]
async fn explicit_timezone_remains_strict_and_overrides_local_timezone() {
let url = std::env::var("DBX_TEST_POSTGRES_URL").expect("DBX_TEST_POSTGRES_URL");
let separator = if url.contains('?') { '&' } else { '?' };
let explicit_url = format!("{url}{separator}options=-c%20TimeZone%3DAsia%2FShanghai");
let pool = connect_with_local_timezone(&explicit_url, Duration::from_secs(5), "UTC")
.await
.expect("valid explicit timezone");
let client = pool.get().await.expect("checkout postgres");
let timezone: String = client.query_one("SHOW timezone", &[]).await.unwrap().get(0);
assert_eq!(timezone, "Asia/Shanghai");
let invalid_url = format!("{url}{separator}options=-c%20TimeZone%3DInvalid%2FDBX_Timezone");
let error = connect_with_local_timezone(&invalid_url, Duration::from_secs(5), "UTC")
.await
.expect_err("invalid explicit timezone must remain a connection error");
assert!(error.contains("Invalid/DBX_Timezone") || error.contains("time zone"), "{error}");
}
#[tokio::test]
#[ignore = "requires DBX_TEST_POSTGRES_URL pointing at a PostgreSQL database"]
async fn valid_automatic_timezone_is_applied_normally() {
let url = std::env::var("DBX_TEST_POSTGRES_URL").expect("DBX_TEST_POSTGRES_URL");
let pool =
connect_with_local_timezone(&url, Duration::from_secs(5), "UTC").await.expect("valid automatic timezone");
let client = pool.get().await.expect("checkout postgres");
let timezone: String = client.query_one("SHOW timezone", &[]).await.unwrap().get(0);
assert_eq!(timezone, "UTC");
}
#[tokio::test]
#[ignore = "requires DBX_TEST_POSTGRES_URL pointing at a writable PostgreSQL-compatible database"]
async fn list_tables_filtered_supports_risingwave_pagination_and_filtering() {
let url = std::env::var("DBX_TEST_POSTGRES_URL").expect("DBX_TEST_POSTGRES_URL");
let pool = connect_with_local_timezone(&url, Duration::from_secs(5), "UTC")
.await
.expect("connect PostgreSQL-compatible database");
let client = pool.get().await.expect("checkout postgres");
client
.batch_execute(
r#"DROP TABLE IF EXISTS public."dbx_issue_5584_live_page_a";
DROP TABLE IF EXISTS public."dbx_issue_5584_live_page_b";
DROP TABLE IF EXISTS public."dbx_issue_5584_live_order_100%";
DROP TABLE IF EXISTS public."dbx_issue_5584_live_back\slash";
DROP TABLE IF EXISTS public."dbx_issue_5584_live_system_users";
CREATE TABLE public."dbx_issue_5584_live_page_a"(id int);
CREATE TABLE public."dbx_issue_5584_live_page_b"(id int);
CREATE TABLE public."dbx_issue_5584_live_order_100%"(id int);
CREATE TABLE public."dbx_issue_5584_live_back\slash"(id int);
CREATE TABLE public."dbx_issue_5584_live_system_users"(id int);"#,
)
.await
.expect("create live table fixtures");
drop(client);
let all_tables = list_tables(&pool, "public").await.expect("expand complete table list");
let first_page = list_tables_filtered(&pool, "public", Some("dbx_issue_5584_live_page_"), Some(1), Some(0))
.await
.expect("list first table page");
let second_page = list_tables_filtered(&pool, "public", Some("dbx_issue_5584_live_page_"), Some(1), Some(1))
.await
.expect("list second table page");
let wildcard_match =
list_tables_filtered(&pool, "public", Some("dbx_issue_5584_live_order_100%"), Some(10), Some(0))
.await
.expect("filter table with wildcard characters");
let backslash_match =
list_tables_filtered(&pool, "public", Some(r"dbx_issue_5584_live_back\slash"), Some(10), Some(0))
.await
.expect("filter table with backslash");
let fuzzy_match = list_tables_filtered(&pool, "public", Some("i5584su"), Some(10), Some(0))
.await
.expect("fuzzy filter table name");
assert!(all_tables.iter().any(|table| table.name == "dbx_issue_5584_live_page_a"));
assert!(all_tables.iter().any(|table| table.name == "dbx_issue_5584_live_page_b"));
assert_eq!(
first_page.iter().map(|table| table.name.as_str()).collect::<Vec<_>>(),
["dbx_issue_5584_live_page_a"]
);
assert_eq!(
second_page.iter().map(|table| table.name.as_str()).collect::<Vec<_>>(),
["dbx_issue_5584_live_page_b"]
);
assert_eq!(
wildcard_match.iter().map(|table| table.name.as_str()).collect::<Vec<_>>(),
["dbx_issue_5584_live_order_100%"]
);
assert_eq!(
backslash_match.iter().map(|table| table.name.as_str()).collect::<Vec<_>>(),
[r"dbx_issue_5584_live_back\slash"]
);
assert_eq!(
fuzzy_match.iter().map(|table| table.name.as_str()).collect::<Vec<_>>(),
["dbx_issue_5584_live_system_users"]
);
let client = pool.get().await.expect("checkout postgres for cleanup");
client
.batch_execute(
r#"DROP TABLE IF EXISTS public."dbx_issue_5584_live_page_a";
DROP TABLE IF EXISTS public."dbx_issue_5584_live_page_b";
DROP TABLE IF EXISTS public."dbx_issue_5584_live_order_100%";
DROP TABLE IF EXISTS public."dbx_issue_5584_live_back\slash";
DROP TABLE IF EXISTS public."dbx_issue_5584_live_system_users";"#,
)
.await
.expect("drop live table fixtures");
}
#[test]
fn like_contains_pattern_escapes_wildcards() {
assert_eq!(like_contains_pattern(""), "%%");
assert_eq!(like_contains_pattern("order_100%"), "%order\\_100\\%%");
assert_eq!(like_contains_pattern("tilde~name"), "%tilde~name%");
assert_eq!(like_contains_pattern(r"foo\bar"), r"%foo\\bar%");
}
#[test]
fn like_fuzzy_pattern_escapes_wildcards() {
assert_eq!(like_fuzzy_pattern(""), "%%");
assert_eq!(like_fuzzy_pattern("sysu"), "%s%y%s%u%");
assert_eq!(like_fuzzy_pattern("user_%"), "%u%s%e%r%\\_%\\%%");
assert_eq!(like_fuzzy_pattern("tilde~name"), "%t%i%l%d%e%~%n%a%m%e%");
}
#[test]
fn postgres_tables_sql_uses_literal_pagination_without_escape_clause() {
let paged_sql = postgres_tables_sql(Some(500), 200);
assert!(paged_sql.contains("ILIKE $2 OR"));
assert!(paged_sql.contains("$3 <> ''"));
assert!(paged_sql.contains("ILIKE $3"));
assert!(!paged_sql.contains("ESCAPE"));
assert!(paged_sql.contains("ORDER BY CASE WHEN pc.relkind = 'p' THEN 1 ELSE 0 END, c.relname"));
assert!(paged_sql.contains("LIMIT 500 OFFSET 200"));
assert!(!paged_sql.contains("$4"));
assert!(!paged_sql.contains("$5"));
let unbounded_sql = postgres_tables_sql(None, 0);
assert!(unbounded_sql.ends_with("OFFSET 0"));
assert!(!unbounded_sql.contains("LIMIT"));
}
#[test]
fn postgres_completion_like_pattern_uses_prefix_by_default() {
assert_eq!(postgres_completion_like_pattern("Temp", Some(&CompletionAssistantMatchMode::Prefix)), "Temp%");
assert_eq!(postgres_completion_like_pattern("Temp", Some(&CompletionAssistantMatchMode::Contains)), "%Temp%");
assert_eq!(
postgres_completion_like_pattern("order_100%", Some(&CompletionAssistantMatchMode::Prefix)),
"order~_100~%%"
);
}
#[test]
fn postgres_completion_sql_filters_before_limit() {
assert!(postgres_completion_tables_sql().contains("c.relname ILIKE $2 ESCAPE '~'"));
assert!(postgres_completion_tables_sql().contains("pg_catalog.pg_table_is_visible(c.oid)"));
assert!(postgres_completion_tables_sql().contains("c.relkind::text = ANY($3::text[])"));
assert!(postgres_completion_tables_sql().contains("ORDER BY c.relname LIMIT $4"));
assert!(postgres_completion_routines_sql().contains("p.proname ILIKE $2 ESCAPE '~'"));
assert!(postgres_completion_routines_sql().contains("p.prokind::text = ANY($3::text[])"));
assert!(postgres_completion_routines_sql().contains("pg_get_function_identity_arguments(p.oid) AS signature"));
assert!(postgres_completion_routines_sql().contains("ORDER BY p.proname LIMIT $4"));
assert!(postgres_completion_columns_sql().contains("a.attname ILIKE $3 ESCAPE '~'"));
assert!(postgres_visible_table_schema_sql().contains("pg_catalog.pg_table_is_visible(c.oid)"));
}
#[test]
fn postgres_schema_info_sql_only_filters_system_schemas_when_disabled() {
let hidden_sql = postgres_schema_infos_sql(false);
assert!(hidden_sql.contains("information_schema"));
assert!(hidden_sql.contains("pg_temp_%"));
let visible_sql = postgres_schema_infos_sql(true);
assert!(!visible_sql.contains("NOT IN"));
assert!(!visible_sql.contains("NOT LIKE"));
}
}
Reference in New Issue View Git Blame Copy Permalink