#!/usr/bin/env python3
"""
os_sense.py — 织忆 Soulful OS 无感感知
============================
轻量感知牧尘当前在做什么项目，不侵入隐私。

用法：
  python3 os_sense.py          # 可读摘要
  python3 os_sense.py --json   # JSON 输出
"""

import json, os, sys, re
from datetime import datetime
from pathlib import Path
from collections import Counter

HOME = Path.home()
HERMES = HOME / ".hermes"
MC = HOME / "mc"

STOPWORDS = {"织忆", "小唯", "hermes", "python", "文件", "修改", "更新",
             "了", "的", "在", "是", "我", "你", "他", "她", "它",
             "这个", "那个", "什么", "怎么", "为什么", "和", "与",
             "或", "但", "如果", "因为", "所以", "虽然", "不过",
             "还是", "可以", "会", "能", "有", "没有", "不", "别"}

CONFIG_FILES = [
    HERMES / "SOUL.md",
    HERMES / "AGENTS.md",
    HERMES / "config.yaml",
]


def get_recent_files(n=10):
    """最近修改的 md 文件"""
    md_files = list((MC).rglob("*.md")) if MC.exists() else []
    md_files = [(f, os.path.getmtime(f)) for f in md_files]
    md_files.sort(key=lambda x: x[1], reverse=True)
    return [{"path": str(f.relative_to(HOME)), "mtime": datetime.fromtimestamp(m).isoformat()}
            for f, m in md_files[:n]]


def get_project_keywords(n=5):
    """从最近 Wiki 概念文件中提取项目关键词"""
    concepts = MC / "小唯" / "07-Wiki" / "concepts"
    if not concepts.exists():
        return []
    files = list(concepts.glob("*.md"))
    files.sort(key=lambda f: os.path.getmtime(f), reverse=True)
    keywords = []
    for f in files[:n]:
        name = f.stem
        # 去掉日期前缀
        name = re.sub(r'^\d+-', '', name)
        # 提取有意义的词
        for part in re.split(r'[-_]', name):
            if len(part) > 1 and part not in STOPWORDS:
                keywords.append(part)
    return list(set(keywords))[:5]


def get_journal_keywords(n=50):
    """从 journal 统计高频词"""
    journal_path = HERMES / "daemon" / "journal.jsonl"
    if not journal_path.exists():
        return [], []
    words = []
    try:
        with open(journal_path) as f:
            lines = f.readlines()
        recent = lines[-n:] if len(lines) > n else lines
        for line in recent:
            try:
                entry = json.loads(line)
                text = entry.get("summary", "") + " " + entry.get("action_taken", "")
                # 简单分词（英文）+ 中文词提取（\w 不匹配中文）
                tokens = re.findall(r'[\w]{2,}', text.lower())
                chinese_words = re.findall(r'[\u4e00-\u9fff]{2,}', text)
                tokens = [t for t in tokens if t not in STOPWORDS and len(t) > 1]
                chinese_words = [w for w in chinese_words if w not in STOPWORDS]
                words.extend(tokens + chinese_words)
            except:
                pass
    except:
        pass
    if not words:
        return [], []
    counter = Counter(words)
    top = counter.most_common(10)
    return [w for w, _ in top], top


def check_config_change():
    """检查是否有重大配置变更"""
    results = []
    for f in CONFIG_FILES:
        if f.exists():
            mtime = os.path.getmtime(f)
            results.append({"file": f.name, "mtime": datetime.fromtimestamp(mtime).isoformat()})
    return results


def generate_summary(projects, keywords, recent_files):
    """生成自然语言摘要"""
    parts = []
    if keywords:
        parts.append("、".join(keywords[:3]))
    if recent_files:
        latest = recent_files[0]
        # 从路径提取项目名
        path_parts = Path(latest["path"]).parts
        if len(path_parts) >= 2:
            parts.append(f"最近在看 {path_parts[-2]}")
    if not parts:
        return "牧尘最近没有明显的工作模式变化"
    return f"牧尘这周/这段时间在关注：{', '.join(parts)}"


def sense():
    recent_files = get_recent_files(10)
    project_keywords = get_project_keywords(5)
    journal_keywords, journal_freq = get_journal_keywords(50)
    config_files = check_config_change()
    summary = generate_summary(project_keywords, journal_keywords, recent_files)

    # 判断是否有新方向（config 改变）
    config_change = len(config_files) > 0

    return {
        "projects": project_keywords,
        "recent_files": recent_files[:5],
        "keywords": journal_keywords[:8],
        "config_change": config_change,
        "config_files": config_files,
        "summary": summary,
        "sensed_at": datetime.now().isoformat(),
    }


def print_human(data):
    print(f"🖥️  OS 感知报告 — {data['sensed_at'][:10]}")
    print()
    if data["recent_files"]:
        print("📁 最近在看的文件：")
        for f in data["recent_files"][:5]:
            print(f"   {f['path']}")
        print()
    if data["keywords"]:
        print(f"🔍 工作关键词：{' '.join(data['keywords'][:8])}")
        print()
    if data["projects"]:
        print(f"📂 项目方向：{' '.join(data['projects'])}")
        print()
    if data["config_change"]:
        print("⚙️  配置有变更：")
        for cf in data["config_files"]:
            print(f"   {cf['file']}（{cf['mtime'][:10]}）")
        print()
    print(f"💭 {data['summary']}")


if __name__ == "__main__":
    data = sense()
    if "--json" in sys.argv:
        print(json.dumps(data, ensure_ascii=False, indent=2))
    else:
        print_human(data)