"""导出微信朋友圈动态（SnsTimeLine 表）

输出目录: <output_base_dir>/<display_name>/SNS/<yyyyMMddHHmmss000>.json
媒体文件: <output_base_dir>/<display_name>/SNS/<yyyyMMddHHmmss000>_<n>.<ext>
汇总文件: <output_base_dir>/<display_name>/SNS/timeline.json
时间线:   <output_base_dir>/<display_name>/SNS/timeline.html
"""
import bisect
import os
import sys
import json
import sqlite3
import struct
import re
import xml.etree.ElementTree as ET
from datetime import datetime
from urllib.request import urlopen, Request
from urllib.error import URLError

if sys.platform == "win32":
    sys.stdout.reconfigure(encoding="utf-8", errors="replace")

from config import load_config
from decode_image import aligned_aes_block_size

# 朋友圈 XML 来源是不可信输入 (他人朋友圈的 content), 必须挡 XXE。
# 跟 mcp_server._XML_UNSAFE_RE 保持同一过滤模式; max_len 比 mcp_server 宽松
# (朋友圈 timeline XML 含媒体列表 + 评论, 实测可达几十KB; 给 200K 余量)。
_SNS_XML_UNSAFE_RE = re.compile(r'<!DOCTYPE|<!ENTITY', re.IGNORECASE)
_SNS_XML_MAX_LEN = 200_000

_cfg = load_config()
DECRYPTED_DIR = _cfg["decrypted_dir"]
SNS_DB_PATH = os.path.join(DECRYPTED_DIR, "sns", "sns.db")
CONTACT_DB_PATH = os.path.join(DECRYPTED_DIR, "contact", "contact.db")
OUTPUT_DIR = _cfg["output_base_dir"]

# 图片缓存 / 解密相关配置
IMAGE_AES_KEY = _cfg.get("image_aes_key")
IMAGE_XOR_KEY = _cfg.get("image_xor_key", 0x88)
XWECHAT_CACHE_DIR = _cfg.get("xwechat_cache_dir", "")
SNS_CACHE_DIR = _cfg.get("sns_cache_dir", "")

# 联系人筛选（与 export_messages.py 一致）
_CONTACT_FILTER = None
_filter_raw = os.environ.get("WECHAT_EXPORT_CONTACTS", "").strip()
if _filter_raw:
    _CONTACT_FILTER = set(_filter_raw.split(","))
    print(f"朋友圈联系人筛选: {len(_CONTACT_FILTER)} 个")

# ── 媒体下载 ─────────────────────────────────────────────────────────────────

_DOWNLOAD_TIMEOUT = 10  # 秒

# ── 本地缓存图片解密 & 匹配 ──────────────────────────────────────────────────

_V2_MAGIC = b'\x07\x08V2\x08\x07'
_V1_MAGIC = b'\x07\x08V1\x08\x07'
_IMAGE_MAGICS = {
    'jpg': [0xFF, 0xD8, 0xFF],
    'png': [0x89, 0x50, 0x4E, 0x47],
    'gif': [0x47, 0x49, 0x46, 0x38],
    'webp': [0x52, 0x49, 0x46, 0x46],
}
_TIME_WINDOW = 72 * 3600  # 72 小时


def _decrypt_sns_dat(dat_path):
    """解密 SNS 缓存 .dat 文件，返回 bytes 或 None"""
    try:
        with open(dat_path, 'rb') as f:
            data = f.read()
    except OSError:
        return None
    if len(data) < 15:
        return None

    head6 = data[:6]

    # V2 / V1 格式（xwechat cache）
    if head6 in (_V2_MAGIC, _V1_MAGIC):
        aes_key = None
        if head6 == _V1_MAGIC:
            aes_key = b'cfcd208495d565ef'
        elif IMAGE_AES_KEY:
            aes_key = IMAGE_AES_KEY.encode('ascii')[:16] if isinstance(IMAGE_AES_KEY, str) else IMAGE_AES_KEY[:16]
        if not aes_key or len(aes_key) < 16:
            return None
        try:
            from Crypto.Cipher import AES
            from Crypto.Util import Padding
            aes_size, xor_size = struct.unpack_from('<LL', data, 6)
            aligned = aligned_aes_block_size(aes_size)
            offset = 15
            if offset + aligned > len(data):
                return None
            cipher = AES.new(aes_key[:16], AES.MODE_ECB)
            dec_aes = Padding.unpad(cipher.decrypt(data[offset:offset + aligned]), AES.block_size)
            offset += aligned
            raw_end = len(data) - xor_size
            raw_data = data[offset:raw_end] if offset < raw_end else b''
            xor_data = data[raw_end:]
            xor_key = IMAGE_XOR_KEY if isinstance(IMAGE_XOR_KEY, int) else 0x88
            dec_xor = bytes(b ^ xor_key for b in xor_data)
            return dec_aes + raw_data + dec_xor
        except Exception:
            return None

    # 旧 XOR 格式（FileStorage Sns Cache）
    for magic in _IMAGE_MAGICS.values():
        key = data[0] ^ magic[0]
        if all(i < len(data) and (data[i] ^ key) == magic[i] for i in range(len(magic))):
            return bytes(b ^ key for b in data)

    return None


def _detect_format(header):
    """检测解密后数据的图片格式，返回扩展名"""
    if header[:3] == b'\xff\xd8\xff':
        return 'jpg'
    if header[:4] == b'\x89PNG':
        return 'png'
    if header[:3] == b'GIF':
        return 'gif'
    if header[:4] == b'RIFF' and len(header) >= 12 and header[8:12] == b'WEBP':
        return 'webp'
    return 'bin'


def _image_size_from_bytes(data):
    """从解密后的图片数据提取 (width, height)，失败返回 (0, 0)"""
    if not data or len(data) < 24:
        return 0, 0

    # PNG: IHDR 位于字节 16-24
    if data[:4] == b'\x89PNG':
        w = struct.unpack('>I', data[16:20])[0]
        h = struct.unpack('>I', data[20:24])[0]
        return w, h

    # JPEG: 查找 SOF 标记
    if data[:2] == b'\xff\xd8':
        i = 2
        while i < len(data) - 9:
            if data[i] != 0xFF:
                break
            marker = data[i + 1]
            if 0xC0 <= marker <= 0xCF and marker not in (0xC4, 0xC8, 0xCC):
                h = struct.unpack('>H', data[i + 5:i + 7])[0]
                w = struct.unpack('>H', data[i + 7:i + 9])[0]
                return w, h
            if i + 3 >= len(data):
                break
            seg_len = struct.unpack('>H', data[i + 2:i + 4])[0]
            i += 2 + seg_len
        return 0, 0

    # WEBP VP8
    if data[:4] == b'RIFF' and len(data) >= 30 and data[8:12] == b'WEBP':
        if data[12:16] == b'VP8 ':
            w = struct.unpack('<H', data[26:28])[0] & 0x3FFF
            h = struct.unpack('<H', data[28:30])[0] & 0x3FFF
            return w, h

    return 0, 0


def _build_sns_cache_index():
    """扫描 SNS 缓存目录，预解密文件头提取元数据

    返回按 mtime 排序的索引:
    [(mtime, path, est_dec_size, fmt, width, height), ...]
    """
    raw_paths = []  # 先收集所有路径

    # 1. xwechat cache: <cache_dir>/YYYY-MM/Sns/Img/<2hex>/<30hex>
    if XWECHAT_CACHE_DIR and os.path.isdir(XWECHAT_CACHE_DIR):
        for month_dir in os.listdir(XWECHAT_CACHE_DIR):
            sns_img = os.path.join(XWECHAT_CACHE_DIR, month_dir, "Sns", "Img")
            if not os.path.isdir(sns_img):
                continue
            for sub in os.listdir(sns_img):
                sub_path = os.path.join(sns_img, sub)
                if not os.path.isdir(sub_path):
                    continue
                for fname in os.listdir(sub_path):
                    fp = os.path.join(sub_path, fname)
                    if os.path.isfile(fp):
                        raw_paths.append(fp)

    # 2. FileStorage Sns Cache: <sns_cache_dir>/YYYY-MM/<hash>
    if SNS_CACHE_DIR and os.path.isdir(SNS_CACHE_DIR):
        for month_dir in os.listdir(SNS_CACHE_DIR):
            month_path = os.path.join(SNS_CACHE_DIR, month_dir)
            if not os.path.isdir(month_path):
                continue
            for fname in os.listdir(month_path):
                if fname.endswith('_t'):  # 跳过缩略图
                    continue
                fp = os.path.join(month_path, fname)
                if os.path.isfile(fp):
                    raw_paths.append(fp)

    if not raw_paths:
        return []

    print(f"  预读取 {len(raw_paths)} 个缓存文件元数据...")

    # 准备 AES key（避免在循环内重复构造）
    aes_key = None
    if IMAGE_AES_KEY:
        aes_key = IMAGE_AES_KEY.encode('ascii')[:16] if isinstance(IMAGE_AES_KEY, str) else IMAGE_AES_KEY[:16]

    entries = []
    for path in raw_paths:
        try:
            fsize = os.path.getsize(path)
            mtime = os.path.getmtime(path)
            if fsize < 15:
                continue

            with open(path, 'rb') as f:
                data = f.read(min(fsize, 4096))

            head6 = data[:6]
            dec_header = None
            est_dec_size = fsize

            if head6 in (_V2_MAGIC, _V1_MAGIC):
                # V2/V1: 解密 AES 部分获取文件头
                k = b'cfcd208495d565ef' if head6 == _V1_MAGIC else aes_key
                if not k or len(k) < 16:
                    continue
                try:
                    from Crypto.Cipher import AES as _AES
                    aes_size, xor_size = struct.unpack_from('<LL', data, 6)
                    aligned = aligned_aes_block_size(aes_size)
                    est_dec_size = fsize - 15 - (aligned - aes_size)
                    available = min(aligned, len(data) - 15)
                    # 按 16 字节块对齐（ECB 可逐块解密）
                    usable = (available // 16) * 16
                    if usable < 16:
                        continue
                    cipher = _AES.new(k[:16], _AES.MODE_ECB)
                    dec_header = cipher.decrypt(data[15:15 + usable])
                except Exception:
                    continue
            else:
                # XOR 格式
                for magic in _IMAGE_MAGICS.values():
                    key = data[0] ^ magic[0]
                    if all(i < len(data) and (data[i] ^ key) == magic[i] for i in range(len(magic))):
                        dec_header = bytes(b ^ key for b in data[:4096])
                        est_dec_size = fsize
                        break

            if dec_header is None:
                continue

            fmt = _detect_format(dec_header[:16])
            if fmt == 'bin':
                continue

            w, h = _image_size_from_bytes(dec_header)
            entries.append((mtime, path, est_dec_size, fmt, w, h))
        except OSError:
            continue

    entries.sort(key=lambda x: x[0])
    return entries


def _match_cache_images(create_time, media_list, index, index_mtimes):
    """为一条动态的所有媒体项匹配本地缓存图片（无需解密，仅查元数据索引）

    返回: [(matched_path, fmt), ...] 与 media_list 等长，未匹配为 (None, None)
    """
    results = []
    if not index or not media_list:
        return [(None, None)] * len(media_list)

    t_low = create_time - _TIME_WINDOW
    t_high = create_time + _TIME_WINDOW
    lo = bisect.bisect_left(index_mtimes, t_low)
    hi = bisect.bisect_right(index_mtimes, t_high)

    # 如果时间窗口为空（xwechat cache mtime 异常），扩大到全部
    if lo >= hi:
        lo, hi = 0, len(index)

    used_paths = set()

    for media in media_list:
        mtype = media.get("type", "")
        if mtype not in ("2", ""):
            results.append((None, None))
            continue

        want_w = int(media.get("width") or 0)
        want_h = int(media.get("height") or 0)
        want_size = int(media.get("total_size") or 0)

        candidates = []  # (score, path, fmt)

        for i in range(lo, hi):
            mtime_i, path_i, dec_size_i, fmt_i, w_i, h_i = index[i]
            if path_i in used_paths:
                continue

            # 尺寸匹配
            if want_w > 0 and want_h > 0 and w_i > 0 and h_i > 0:
                if w_i != want_w or h_i != want_h:
                    continue

            # 大小匹配
            if want_size > 0:
                if dec_size_i > want_size * 3 or dec_size_i < want_size * 0.3:
                    continue

            size_diff = abs(dec_size_i - want_size) if want_size > 0 else 0
            time_diff = abs(mtime_i - create_time)
            candidates.append((size_diff, time_diff, path_i, fmt_i))

        if candidates:
            candidates.sort(key=lambda x: (x[0], x[1]))
            best = candidates[0]
            used_paths.add(best[2])
            results.append((best[2], best[3]))
        else:
            results.append((None, None))

    return results

# ContentObject type 含义（已知）
_CONTENT_TYPES = {
    1: "图文",
    2: "纯文本",
    3: "链接",
    5: "视频链接",
    7: "位置",
    15: "视频",
    28: "短视频",
    30: "音乐",
    34: "笔记",
    42: "小程序",
    54: "直播",
}


def _try_download_media(url, save_path):
    """尝试下载媒体文件，返回 True/False

    微信朋友圈 shmmsns.qpic.cn 图片需要携带 Referer 和 User-Agent。
    注意: URL 返回的数据可能是加密的（enc_idx=1 的情况），
    解密算法尚未公开，此时下载的文件无法直接查看。
    如果下载失败返回 False，后续可替换为更复杂的下载逻辑。
    """
    if not url or not url.startswith("http"):
        return False
    try:
        req = Request(url, headers={
            "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
                          "AppleWebKit/537.36 (KHTML, like Gecko) "
                          "Chrome/120.0.0.0 Safari/537.36",
            "Referer": "https://weixin.qq.com/",
        })
        with urlopen(req, timeout=_DOWNLOAD_TIMEOUT) as resp:
            if resp.status != 200:
                return False
            data = resp.read()
            if len(data) < 100:
                return False
            # 检测格式
            if data[:3] == b'\xff\xd8\xff':
                ext = '.jpg'
            elif data[:4] == b'\x89PNG':
                ext = '.png'
            elif data[:4] == b'GIF8':
                ext = '.gif'
            elif data[:4] == b'RIFF' and data[8:12] == b'WEBP':
                ext = '.webp'
            else:
                ext = '.bin'
            if not os.path.splitext(save_path)[1]:
                save_path += ext
            with open(save_path, 'wb') as f:
                f.write(data)
            return True
    except (URLError, OSError, Exception):
        return False


def _parse_media_list(timeline_obj):
    """解析 TimelineObject 中的 mediaList，返回 media 信息列表"""
    medias = []
    for media_el in timeline_obj.findall('.//media'):
        media_type = media_el.findtext('type', '')
        sub_type = media_el.findtext('sub_type', '')
        vid_duration = media_el.findtext('videoDuration', '0')

        thumb_el = media_el.find('thumb')
        url_el = media_el.find('url')
        size_el = media_el.find('size')

        info = {
            "type": media_type,
            "sub_type": sub_type,
            "video_duration": vid_duration,
        }

        if thumb_el is not None:
            info["thumb_url"] = thumb_el.text or ""
            info["thumb_key"] = thumb_el.get("key", "")
            info["thumb_token"] = thumb_el.get("token", "")

        if url_el is not None:
            info["url"] = url_el.text or ""
            info["url_md5"] = url_el.get("md5", "")
            info["url_key"] = url_el.get("key", "")
            info["url_token"] = url_el.get("token", "")

        if size_el is not None:
            info["width"] = size_el.get("width", "")
            info["height"] = size_el.get("height", "")
            info["total_size"] = size_el.get("totalSize", "")

        medias.append(info)
    return medias


def _parse_timeline_xml(content_xml):
    """解析 SnsTimeLine 的 Content XML，返回结构化数据"""
    if not content_xml:
        return None
    if len(content_xml) > _SNS_XML_MAX_LEN:
        return None
    if _SNS_XML_UNSAFE_RE.search(content_xml):
        # XXE 防护: 拒绝 DOCTYPE/ENTITY,避免恶意朋友圈 XML 通过 entity expansion
        # 或外部实体引用执行 SSRF/读取本地文件
        return None
    try:
        root = ET.fromstring(content_xml)
    except ET.ParseError:
        return None

    tl = root.find('.//TimelineObject')
    if tl is None:
        return None

    create_time_str = tl.findtext('createTime', '0')
    try:
        create_time = int(create_time_str)
    except ValueError:
        create_time = 0

    content_type = tl.findtext('.//ContentObject/type', '0')
    try:
        content_type_int = int(content_type)
    except ValueError:
        content_type_int = 0

    # 解析位置
    loc_el = tl.find('.//location')
    location = None
    if loc_el is not None:
        lat = loc_el.get('latitude', '0')
        lon = loc_el.get('longitude', '0')
        if lat != '0' or lon != '0':
            location = {
                "latitude": lat,
                "longitude": lon,
                "poi_name": loc_el.get("poiName", ""),
            }

    return {
        "id": tl.findtext('id', ''),
        "username": tl.findtext('username', ''),
        "create_time": create_time,
        "create_time_str": datetime.fromtimestamp(create_time).strftime("%Y-%m-%d %H:%M:%S") if create_time else "",
        "content_desc": tl.findtext('contentDesc', ''),
        "content_type": content_type_int,
        "content_type_name": _CONTENT_TYPES.get(content_type_int, f"未知({content_type_int})"),
        "nickname": root.findtext('.//LocalExtraInfo/nickname', ''),
        "is_private": tl.findtext('private', '0') == '1',
        "location": location,
        "media": _parse_media_list(tl),
    }


def _load_comments(conn):
    """加载 SnsMessage_tmp3 评论/点赞，按 feed_id 分组"""
    comments = {}
    try:
        rows = conn.execute(
            "SELECT feed_id, create_time, type, from_username, from_nickname,"
            " to_username, to_nickname, content"
            " FROM SnsMessage_tmp3 ORDER BY create_time"
        ).fetchall()
        for feed_id, ctime, ctype, from_u, from_n, to_u, to_n, content in rows:
            if feed_id not in comments:
                comments[feed_id] = []
            comments[feed_id].append({
                "create_time": ctime,
                "create_time_str": datetime.fromtimestamp(ctime).strftime("%Y-%m-%d %H:%M:%S") if ctime else "",
                "type": ctype,  # 1=点赞, 2=评论
                "type_name": "点赞" if ctype == 1 else "评论" if ctype == 2 else f"未知({ctype})",
                "from_username": from_u or "",
                "from_nickname": from_n or "",
                "to_username": to_u or "",
                "to_nickname": to_n or "",
                "content": content or "",
            })
    except Exception as e:
        print(f"读取评论数据失败: {e}")
    return comments


def _safe_dirname(name: str) -> str:
    """清理文件夹名中的非法字符"""
    for ch in r'\/:*?"<>|':
        name = name.replace(ch, "_")
    return name.strip() or "unknown"


def _load_contact_map():
    """从 contact.db 加载 {username: display_name}"""
    cmap = {}
    if not os.path.exists(CONTACT_DB_PATH):
        return cmap
    try:
        conn = sqlite3.connect(CONTACT_DB_PATH)
        for uname, remark, nick_name in conn.execute(
            "SELECT username, remark, nick_name FROM contact"
        ):
            dname = remark or nick_name or uname
            cmap[uname] = _safe_dirname(dname)
        conn.close()
    except Exception as e:
        print(f"读取联系人数据库失败: {e}")
    return cmap


def _timestamp_filename(unix_ts):
    """Unix 时间戳 → yyyyMMddHHmmss000 文件名（毫秒部分为 000）"""
    if not unix_ts:
        return "00000000000000000"
    dt = datetime.fromtimestamp(unix_ts)
    return dt.strftime("%Y%m%d%H%M%S") + "000"


def _html_escape(text):
    """简单 HTML 转义"""
    return (text or "").replace("&", "&amp;").replace("<", "&lt;").replace(">", "&gt;").replace('"', "&quot;")


def _generate_timeline_html(display_name, posts, sns_dir, image_files):
    """生成朋友圈时间线 HTML

    Args:
        display_name: 联系人显示名
        posts: 按时间倒序排列的动态列表
        sns_dir: SNS 输出目录
        image_files: {final_name: [(rel_path, ext), ...]} 每条动态的图片文件列表
    """
    html_path = os.path.join(sns_dir, "timeline.html")
    parts = [f"""<!DOCTYPE html>
<html lang="zh-CN">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>{_html_escape(display_name)} - 朋友圈</title>
<style>
body {{ font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, sans-serif; max-width: 800px; margin: 0 auto; padding: 20px; background: #f5f5f5; color: #333; }}
h1 {{ text-align: center; color: #07c160; border-bottom: 2px solid #07c160; padding-bottom: 10px; }}
.stats {{ text-align: center; color: #888; margin-bottom: 30px; font-size: 14px; }}
.post {{ background: #fff; border-radius: 10px; padding: 16px 20px; margin-bottom: 16px; box-shadow: 0 1px 3px rgba(0,0,0,0.08); }}
.post-time {{ font-size: 12px; color: #999; margin-bottom: 8px; }}
.post-type {{ display: inline-block; font-size: 11px; background: #e8f5e9; color: #2e7d32; padding: 1px 6px; border-radius: 3px; margin-left: 8px; }}
.post-text {{ margin: 8px 0; white-space: pre-wrap; word-break: break-word; line-height: 1.6; }}
.post-images {{ display: flex; flex-wrap: wrap; gap: 6px; margin: 10px 0; }}
.post-images img {{ max-width: 240px; max-height: 240px; border-radius: 6px; object-fit: cover; cursor: pointer; }}
.post-images img:hover {{ opacity: 0.85; }}
.post-location {{ font-size: 12px; color: #1a73e8; margin: 4px 0; }}
.comments {{ margin-top: 10px; padding-top: 8px; border-top: 1px solid #f0f0f0; }}
.comment {{ font-size: 13px; color: #555; margin: 4px 0; line-height: 1.5; }}
.comment-name {{ color: #576b95; font-weight: 500; }}
.comment-like {{ color: #e64a19; }}
.private-tag {{ font-size: 11px; background: #fff3e0; color: #e65100; padding: 1px 6px; border-radius: 3px; margin-left: 6px; }}
</style>
</head>
<body>
<h1>{_html_escape(display_name)} 的朋友圈</h1>
<div class="stats">共 {len(posts)} 条动态</div>
"""]

    for post in posts:
        final_name = post.get("_final_name", "")
        time_str = _html_escape(post.get("create_time_str", ""))
        type_name = _html_escape(post.get("content_type_name", ""))
        text = _html_escape(post.get("content_desc", ""))
        is_private = post.get("is_private", False)

        parts.append('<div class="post">')
        parts.append(f'<div class="post-time">{time_str}<span class="post-type">{type_name}</span>')
        if is_private:
            parts.append('<span class="private-tag">仅自己可见</span>')
        parts.append('</div>')

        if text:
            parts.append(f'<div class="post-text">{text}</div>')

        # 图片
        imgs = image_files.get(final_name, [])
        if imgs:
            parts.append('<div class="post-images">')
            for rel_path, _ in imgs:
                parts.append(f'<img src="{_html_escape(rel_path)}" loading="lazy" onclick="window.open(this.src)">')
            parts.append('</div>')

        # 位置
        loc = post.get("location")
        if loc and loc.get("poi_name"):
            parts.append(f'<div class="post-location">📍 {_html_escape(loc["poi_name"])}</div>')

        # 评论
        comments = post.get("comments", [])
        if comments:
            parts.append('<div class="comments">')
            for c in comments:
                if c.get("type") == 1:
                    parts.append(f'<div class="comment comment-like">❤️ <span class="comment-name">{_html_escape(c["from_nickname"])}</span></div>')
                else:
                    to_part = ""
                    if c.get("to_nickname"):
                        to_part = f' 回复 <span class="comment-name">{_html_escape(c["to_nickname"])}</span>'
                    parts.append(f'<div class="comment"><span class="comment-name">{_html_escape(c["from_nickname"])}</span>{to_part}: {_html_escape(c.get("content", ""))}</div>')
            parts.append('</div>')

        parts.append('</div>')

    parts.append('</body></html>')

    with open(html_path, 'w', encoding='utf-8') as f:
        f.write('\n'.join(parts))
    return html_path


def export_sns_timeline():
    """导出朋友圈动态主函数"""
    if not os.path.exists(SNS_DB_PATH):
        print(f"朋友圈数据库不存在: {SNS_DB_PATH}")
        print("请先运行「解密数据库」")
        return

    # 加载联系人
    contact_map = _load_contact_map()
    print(f"联系人: {len(contact_map)} 条")

    conn = sqlite3.connect(SNS_DB_PATH)

    # 加载评论
    print("加载评论数据...")
    comments_map = _load_comments(conn)
    print(f"评论/点赞: {sum(len(v) for v in comments_map.values())} 条")

    # 读取所有动态
    print("读取朋友圈动态...")
    rows = conn.execute(
        "SELECT tid, user_name, content FROM SnsTimeLine WHERE content IS NOT NULL"
    ).fetchall()
    conn.close()

    print(f"共 {len(rows)} 条动态")
    if not rows:
        return

    # 是否尝试下载媒体
    try_download = os.environ.get("WECHAT_SNS_DOWNLOAD_MEDIA", "0").strip() == "1"

    # ── 构建缓存索引 ─────────────────────────────────────────────────────
    print("扫描 SNS 图片缓存...")
    cache_index = _build_sns_cache_index()
    index_mtimes = [e[0] for e in cache_index]
    print(f"缓存索引: {len(cache_index)} 个有效图片文件")

    # ── 按 user_name 分组 ─────────────────────────────────────────────────
    user_posts: dict[str, list] = {}  # user_name -> [post, ...]
    user_nicknames: dict[str, str] = {}  # user_name -> nickname (从 XML 提取)
    skipped = 0

    for tid, user_name, content_xml in rows:
        if not content_xml:
            continue
        if _CONTACT_FILTER and user_name not in _CONTACT_FILTER:
            skipped += 1
            continue

        post = _parse_timeline_xml(content_xml)
        if not post:
            continue

        post["tid"] = tid
        post["db_user_name"] = user_name or ""
        post["comments"] = comments_map.get(tid, [])

        key = user_name or "unknown"
        if key not in user_posts:
            user_posts[key] = []
        user_posts[key].append(post)

        # 记录 nickname（取第一个非空的）
        nick = post.get("nickname", "")
        if nick and key not in user_nicknames:
            user_nicknames[key] = nick

    if skipped:
        print(f"筛选跳过: {skipped} 条")

    # ── 按联系人输出 ──────────────────────────────────────────────────────
    total_posts = 0
    cache_match_ok = 0
    cache_match_fail = 0
    media_download_ok = 0
    media_download_fail = 0

    for user_name, posts in user_posts.items():
        dname = contact_map.get(user_name) or _safe_dirname(
            user_nicknames.get(user_name) or user_name
        )
        sns_dir = os.path.join(OUTPUT_DIR, dname, "SNS")
        os.makedirs(sns_dir, exist_ok=True)

        # 用 set 处理同一秒多条动态的文件名冲突
        used_names = set()
        # image_files: {final_name: [(rel_path, ext), ...]} 用于 HTML 生成
        image_files: dict[str, list] = {}

        posts.sort(key=lambda p: p.get("create_time", 0))

        for post in posts:
            ts_name = _timestamp_filename(post.get("create_time"))
            # 处理冲突: 递增末尾毫秒
            final_name = ts_name
            counter = 1
            while final_name in used_names:
                final_name = ts_name[:-3] + f"{counter:03d}"
                counter += 1
            used_names.add(final_name)
            post["_final_name"] = final_name

            # ── 缓存图片匹配 ─────────────────────────────────────────
            media_list = post.get("media", [])
            if media_list and cache_index:
                matches = _match_cache_images(
                    post.get("create_time", 0), media_list,
                    cache_index, index_mtimes,
                )
                for i, (matched_path, fmt) in enumerate(matches):
                    if matched_path is not None:
                        dec_bytes = _decrypt_sns_dat(matched_path)
                        if dec_bytes:
                            ext = _detect_format(dec_bytes[:16])
                            img_name = f"{final_name}_{i}.{ext}"
                            img_path = os.path.join(sns_dir, img_name)
                            with open(img_path, 'wb') as f:
                                f.write(dec_bytes)
                            if final_name not in image_files:
                                image_files[final_name] = []
                            image_files[final_name].append((img_name, ext))
                            cache_match_ok += 1
                            continue
                        cache_match_fail += 1
                    else:
                        cache_match_fail += 1

            # ── 网络下载（仅对缓存未匹配的媒体尝试） ─────────────────
            if try_download and media_list:
                existing_imgs = image_files.get(final_name, [])
                existing_indices = {int(p.rsplit('_', 1)[1].split('.')[0]) for p, _ in existing_imgs} if existing_imgs else set()
                for i, media in enumerate(media_list):
                    if i in existing_indices:
                        continue
                    media_url = media.get("url", "") or media.get("thumb_url", "")
                    if not media_url:
                        continue
                    save_name = os.path.join(sns_dir, f"{final_name}_{i}")
                    if _try_download_media(media_url, save_name):
                        # 下载成功后更新 image_files
                        for cand_ext in ('.jpg', '.png', '.gif', '.webp', '.bin'):
                            if os.path.exists(save_name + cand_ext):
                                if final_name not in image_files:
                                    image_files[final_name] = []
                                image_files[final_name].append((f"{final_name}_{i}{cand_ext}", cand_ext[1:]))
                                break
                        media_download_ok += 1
                    else:
                        media_download_fail += 1

            # 保存 JSON（去掉内部字段）
            post_out = {k: v for k, v in post.items() if not k.startswith("_")}
            post_file = os.path.join(sns_dir, f"{final_name}.json")
            with open(post_file, 'w', encoding='utf-8') as f:
                json.dump(post_out, f, ensure_ascii=False, indent=2)

            total_posts += 1

        # 每个联系人的汇总 JSON
        posts.sort(key=lambda p: p.get("create_time", 0), reverse=True)
        summary_posts = [{k: v for k, v in p.items() if not k.startswith("_")} for p in posts]
        summary_path = os.path.join(sns_dir, "timeline.json")
        with open(summary_path, 'w', encoding='utf-8') as f:
            json.dump({
                "user_name": user_name,
                "display_name": dname,
                "export_time": datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
                "total_posts": len(posts),
                "posts": summary_posts,
            }, f, ensure_ascii=False, indent=2)

        # 生成 HTML 时间线
        _generate_timeline_html(dname, posts, sns_dir, image_files)

        print(f"  {dname}: {len(posts)} 条动态")

    print(f"\n完成: {len(user_posts)} 个联系人, 共 {total_posts} 条动态")
    if cache_index:
        print(f"缓存匹配: 成功 {cache_match_ok}, 失败 {cache_match_fail}")
    if try_download:
        print(f"媒体下载: 成功 {media_download_ok}, 失败 {media_download_fail}")
    print(f"输出目录: {os.path.abspath(OUTPUT_DIR)}")


if __name__ == "__main__":
    export_sns_timeline()