weflow-cli 1.8.2 → 1.10.0

package/scripts/contact_schema.py

@@ -0,0 +1,339 @@
+# -*- coding: utf-8 -*-
+"""解 `contact.db` / `chat_room.ext_buffer` 里**已经验证过**的那几个字段（只读）。
+
+为什么只解"验证过的"：顶层 `#3`/`#4` 的语义在本机实测里被三次证伪——不是
+`chatroom_member.member_id` 的最大/最小值、不是 `chat_room.id`、也不在消息库
+`message_*.db` 的任何 id 列里。语义没定就不给它编名字：本脚本把它们原样放进
+`unrecognized`，谁也别照着一个猜出来的名字写逻辑。结构笔记见
+`docs/CONTACT_DB_SCHEMA.md`，取舍见 D-067 / D-068。
+
+已验证、因而输出的字段：
+
+    RoomData
+      #1 (重复) 成员 —— 每个成员：1=userName  2=displayName  3=status  4=inviter
+      #5 (重复) = 成员里 **status 位 11 置位**的那批 id（本机 77/77 与该集合完全相等；
+                 它**不是**"额外的参与者"——早前那么叫是错的，见 DECISIONS D-070）
+
+联系人模式（`--contacts`）：读 `contact.extra_buffer` —— 同样只输出**已验证**的字段：
+
+    #5  (ld)     国家/地区代码（ISO 3166-1 alpha-2）—— 本机 25/25 落在码表、众数 CN 占 94%
+    #13 (varint) = 该行 `biz_info.type`（账号服务类型）—— 本机 687/687 逐行相等
+    #41 (varint) = 该行资料的"最近更新时间"（epoch 秒；0 表示显式未设置，按缺省处理）
+
+**故意不输出** `#4`/`#9`：它们已定性为"该账号自身资料文本的汇集"（含主体名/菜单按钮名），
+是**内容不是字段名**，打进 CLI 输出等于把账号资料漏出去 —— 它们只以长度出现在 `unrecognized` 里。
+其余字段未定名，一律原样进 `unrecognized`（键就是字段号）。取舍见 D-068 / D-082 / D-084 / D-088。
+
+用法：
+    python scripts/contact_schema.py --json
+    python scripts/contact_schema.py --limit 5            # 只看前 5 个群
+    python scripts/contact_schema.py --room <群 username 或 id>
+    python scripts/contact_schema.py --contacts --limit 5 # 改看联系人（#5 地区码 / #13 服务类型 / #41 更新时间）
+"""
+import argparse
+import json
+import os
+import re
+import sys
+import time
+
+sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+# 手跑时 stdout 是本机编码（Windows 是 GBK）：中文会乱码、非 GBK 字符直接崩掉整次输出
+sys.stdout.reconfigure(encoding='utf-8', errors='replace')
+
+
+def varint(buf, i):
+    val = 0
+    shift = 0
+    while True:
+        if i >= len(buf):
+            raise ValueError('varint 越界')
+        b = buf[i]
+        i += 1
+        val |= (b & 0x7F) << shift
+        if not (b & 0x80):
+            return val, i
+        shift += 7
+        if shift > 63:
+            raise ValueError('varint 过长')
+
+
+def walk(buf):
+    """走一遍 wire format，返回 [(字段号, wire type, 值)]。
+
+    **不认识的 wire type 一律抛错**：宁可整条报"解不动"，也不把错位的字节当成字段值
+    ——那正是"看着像"的来源。
+    """
+    i = 0
+    out = []
+    while i < len(buf):
+        tag, i = varint(buf, i)
+        fno, wt = tag >> 3, tag & 7
+        if fno == 0 or wt in (3, 4):
+            raise ValueError('字段号/wire type 异常（fno=%d wt=%d）' % (fno, wt))
+        if wt == 0:
+            v, i = varint(buf, i)
+            out.append((fno, wt, v))
+        elif wt == 2:
+            ln, i = varint(buf, i)
+            if i + ln > len(buf):
+                raise ValueError('长度越界')
+            out.append((fno, wt, buf[i:i + ln]))
+            i += ln
+        elif wt == 5:
+            if i + 4 > len(buf):
+                raise ValueError('fixed32 越界')
+            out.append((fno, wt, buf[i:i + 4]))
+            i += 4
+        elif wt == 1:
+            if i + 8 > len(buf):
+                raise ValueError('fixed64 越界')
+            out.append((fno, wt, buf[i:i + 8]))
+            i += 8
+        else:
+            raise ValueError('未知 wire type %d' % wt)
+    return out
+
+
+def _text(raw):
+    if raw is None:
+        return None
+    try:
+        return raw.decode('utf-8')
+    except UnicodeDecodeError:
+        return None  # 不是 UTF-8 就别硬当字符串用（返回 None，调用方按未知处理）
+
+
+def decode_room(blob):
+    """把一个 ext_buffer 解成 {members, statusBit11Ids, unrecognized}；解不动就抛错。"""
+    members = []
+    extra_ids = []
+    unrecognized = {}
+    for fno, wt, value in walk(blob):
+        if fno == 1 and wt == 2:
+            inner = walk(value)
+            # **键固定**：缺 #2 的成员也要有 displayName 这个键（值为 None），
+            # 否则 JSON 里时有时无，下游每次都得先问"键在不在"——那种形状迟早被漏判成"没名字"。
+            member = {'userName': None, 'displayName': None, 'status': None, 'inviter': None}
+            for m_fno, m_wt, m_val in inner:
+                if m_fno in (1, 2, 4) and m_wt == 2:
+                    key = {1: 'userName', 2: 'displayName', 4: 'inviter'}[m_fno]
+                    member[key] = _text(m_val)
+                elif m_fno == 3 and m_wt == 0:
+                    member['status'] = m_val
+                else:
+                    member.setdefault('unrecognized', {})['%d' % m_fno] = (
+                        len(m_val) if m_wt == 2 else m_val)
+            members.append(member)
+        elif fno == 5 and wt == 2:
+            extra_ids.append(_text(value))
+        else:
+            # 不猜：原样放进来，键就是字段号
+            unrecognized['%d' % fno] = (len(value) if wt == 2 else value)
+    return {'members': members, 'statusBit11Ids': extra_ids, 'unrecognized': unrecognized}
+
+
+# `contact.extra_buffer` 的 `#5` 只在它是"恰好两个大写 ASCII 字母"时才当地区码输出。
+REGION_RE = re.compile(r'^[A-Z]{2}$')
+
+
+def decode_contact(blob):
+    """解 `contact.extra_buffer` 里**已经验证过**的那几个字段；解不动就抛错。
+
+    只解"验证过的"（D-068 / D-082 / D-084），三个：
+
+      #5  (ld) 国家/地区代码（ISO 3166-1 alpha-2）—— 本机 25/25 落在码表、众数 CN 占 94%
+      #13 (varint) = 该行 `biz_info.type`（账号服务类型）—— 本机 687/687 逐行相等
+      #41 (varint) epoch 秒 = 该行资料的"最近更新时间"—— 0 表示显式未设置，按缺省处理
+
+    **故意不输出的**（说清楚，免得后人以为是漏了）：
+
+      - `#4` / `#9`：已定性为"该账号自身资料文本的汇集"（含主体名/菜单按钮名），它是**内容不是字段名** ——
+        塞进 CLI 输出等于把账号资料漏出去。它们在 `unrecognized` 里**只留长度**。
+      - 顶层 `#1`/`#2` 那一族（`contact.extra_buffer` 里 44 行的 OpenIM 小形状）：只报 `kind='openim'`，
+        其余按键原样放进 `unrecognized`。那两键在文档里是 openim 库的外键，对"读联系人"没用。
+      - 其余 30 来个字段：**语义未定，一律原样进 `unrecognized`**，键就是字段号（同 `decode_room`）。
+    """
+    fields = {}
+    for fno, wt, value in walk(blob):
+        fields.setdefault(fno, (wt, value))
+    if not fields:
+        return {'kind': None, 'region': None, 'bizType': None, 'updatedAt': None,
+                'unrecognized': {}}
+    raw = {'%d' % fno: (len(v) if wt == 2 else v) for fno, (wt, v) in fields.items()}
+    # 小形状 = **字段号全 <= 9**（大 proto 一定有 #2..#38，不可能全 <= 9）。
+    # ⚠️ 别再要求"必须有 #1"：那族里有一行没有 #1，加上这个条件会把它错判成联系人
+    #    （全量核对时抓到：kind 分布变成 944/43，文档是 940/44 + 2 个残行）。
+    if max(fields) <= 9:
+        return {'kind': 'openim', 'region': None, 'bizType': None, 'updatedAt': None,
+                'unrecognized': raw}
+
+    out = {'kind': 'contact', 'region': None, 'bizType': None, 'updatedAt': None,
+           'unrecognized': raw}
+    wt5, v5 = fields.get(5, (None, None))
+    if wt5 == 2 and isinstance(v5, bytes):
+        text = _text(v5)
+        if text and REGION_RE.match(text):
+            out['region'] = text
+    wt13, v13 = fields.get(13, (None, None))
+    if wt13 == 0:
+        out['bizType'] = v13
+    wt41, v41 = fields.get(41, (None, None))
+    if wt41 == 0 and v41:  # 0 = 显式"未设置"，不是 1970 年
+        out['updatedAt'] = v41
+    return out
+
+
+def load_rows(args):
+    from _utils import get_db_config  # noqa: E402
+    from nt_decrypt import require_sqlcipher  # noqa: E402
+
+    cfg = get_db_config()
+    db = args.db or cfg['contact_db']
+    key = args.key or cfg['contact_key']
+    salt = args.salt or cfg['contact_salt']
+    if not db or not os.path.isfile(db):
+        raise SystemExit('找不到 contact 库：%s' % (db or '(未配置)'))
+    if not key or not salt:
+        raise SystemExit('缺少库密钥（先跑 weflow-cli init，或显式给 --key/--salt）')
+    conn = require_sqlcipher().connect(db)
+    cur = conn.cursor()
+    cur.execute('PRAGMA key = "x\'%s%s\'";' % (key, salt))
+    sql = ('SELECT id, username, owner, ext_buffer FROM chat_room '
+           'WHERE ext_buffer IS NOT NULL AND length(ext_buffer) > 0')
+    params = []
+    if args.room:
+        sql += ' AND (username = ? OR id = ?)'
+        params = [args.room, args.room]
+    sql += ' ORDER BY id LIMIT ?'
+    params.append(args.limit)
+    rows = cur.execute(sql, params).fetchall()
+    conn.close()
+    return rows
+
+
+def load_contacts(args):
+    from _utils import get_db_config  # noqa: E402
+    from nt_decrypt import require_sqlcipher  # noqa: E402
+
+    cfg = get_db_config()
+    db = args.db or cfg['contact_db']
+    key = args.key or cfg['contact_key']
+    salt = args.salt or cfg['contact_salt']
+    if not db or not os.path.isfile(db):
+        raise SystemExit('找不到 contact 库：%s' % (db or '(未配置)'))
+    if not key or not salt:
+        raise SystemExit('缺少库密钥（先跑 weflow-cli init，或显式给 --key/--salt）')
+    conn = require_sqlcipher().connect(db)
+    cur = conn.cursor()
+    cur.execute('PRAGMA key = "x\'%s%s\'";' % (key, salt))
+    sql = ('SELECT id, username, local_type, extra_buffer FROM contact '
+           'WHERE extra_buffer IS NOT NULL AND length(extra_buffer) > 0')
+    params = []
+    if args.room:
+        sql += ' AND (username = ? OR id = ?)'
+        params = [args.room, args.room]
+    sql += ' ORDER BY id LIMIT ?'
+    params.append(args.limit)
+    rows = cur.execute(sql, params).fetchall()
+    conn.close()
+    return rows
+
+
+def main(argv=None):
+    ap = argparse.ArgumentParser(
+        description='读 contact.db 的 ext_buffer（只读，不联网、不调模型）：默认读**群**，--contacts 改读**联系人**')
+    ap.add_argument('--limit', type=int, default=20, help='最多看几个（群 / 联系人，默认 20）')
+    ap.add_argument('--room', help='只看某个（群 / 联系人 的 username 或 id）')
+    ap.add_argument('--contacts', action='store_true',
+                    help='改读联系人的 extra_buffer（#5 地区码 / #13 服务类型 / #41 资料更新时间），而不是群')
+    ap.add_argument('--db', help='contact.db 路径（默认取配置）')
+    ap.add_argument('--key', help='库密钥（默认取配置）')
+    ap.add_argument('--salt', help='库盐（默认取配置）')
+    ap.add_argument('--json', action='store_true', help='输出机器可读结果')
+    args = ap.parse_args(argv)
+
+    try:
+        rows = load_contacts(args) if args.contacts else load_rows(args)
+    except SystemExit:
+        raise
+    except Exception as exc:
+        if args.json:
+            print(json.dumps({'success': False, 'error': str(exc)[:200]}, ensure_ascii=False))
+        else:
+            print('读库失败：%s' % str(exc)[:200], file=sys.stderr)
+        return 1
+
+    if args.contacts:
+        contacts = []
+        failures = []
+        for rid, username, local_type, blob in rows:
+            try:
+                decoded = decode_contact(bytes(blob))
+            except Exception as exc:
+                # 同群那条：解不动就说解不动，不返回一条"什么都没有"的联系人
+                failures.append({'id': rid, 'username': username, 'error': str(exc)[:80]})
+                continue
+            contacts.append({'id': rid, 'username': username, 'localType': local_type,
+                             'kind': decoded['kind'], 'region': decoded['region'],
+                             'bizType': decoded['bizType'], 'updatedAt': decoded['updatedAt'],
+                             'unrecognized': decoded['unrecognized']})
+        if args.json:
+            print(json.dumps({'success': True, 'action': 'contact-schema',
+                              'contacts': contacts, 'failures': failures,
+                              'readsLocalData': True, 'invokesAI': False,
+                              'sendsNothing': True}, ensure_ascii=False))
+        else:
+            print('看了 %d 个联系人；解不动 %d 个' % (len(contacts), len(failures)))
+            print('（只输出已验证字段：#5 地区码 / #13 服务类型 / #41 资料更新时间；'
+                  '#4/#9 是账号资料文本，属内容，不打印，只在"未识别"里留长度）')
+            for c in contacts:
+                when = time.strftime('%Y-%m-%d', time.gmtime(c['updatedAt'])) if c['updatedAt'] else '-'
+                print('  %-28s local_type=%-3s kind=%-8s 地区=%-3s 服务类型=%-4s 更新=%-11s 未识别 %d 个'
+                      % (c['username'], c['localType'], c['kind'], c['region'] or '-',
+                         c['bizType'] if c['bizType'] is not None else '-', when,
+                         len(c['unrecognized'])))
+            for f in failures:
+                print('解不动: id=%s %s' % (f['id'], f['error']), file=sys.stderr)
+        return 0
+
+    rooms = []
+    failures = []
+    for rid, username, owner, blob in rows:
+        try:
+            decoded = decode_room(bytes(blob))
+        except Exception as exc:
+            # **解不动就说解不动**，不返回空成员表冒充"这个群没人"
+            failures.append({'id': rid, 'username': username, 'error': str(exc)[:80]})
+            continue
+        rooms.append({'id': rid, 'username': username, 'owner': owner,
+                      'memberCount': len(decoded['members']),
+                      'members': decoded['members'],
+                      'statusBit11Ids': decoded['statusBit11Ids'],
+                      'unrecognized': decoded['unrecognized']})
+
+    if args.json:
+        print(json.dumps({'success': True, 'action': 'contact-schema', 'rooms': rooms,
+                          'failures': failures, 'readsLocalData': True,
+                          'invokesAI': False, 'sendsNothing': True}, ensure_ascii=False))
+    else:
+        print('看了 %d 个群；解不动 %d 个' % (len(rooms), len(failures)))
+        for room in rooms:
+            print('\n群 %s（id=%s）成员 %d 个；未识别字段 %s' % (
+                room['username'], room['id'], room['memberCount'],
+                ','.join(sorted(room['unrecognized'])) or '无'))
+            for m in room['members'][:5]:
+                print('   %-28s %-12s status=%-8s inviter=%s' % (
+                    m.get('userName'), m.get('displayName'), m.get('status'), m.get('inviter')))
+            if room['statusBit11Ids']:
+                print('   #5（status 位 11 的成员）: %s'
+                      % ', '.join(x or '?' for x in room['statusBit11Ids'][:5]))
+        for f in failures:
+            print('解不动: id=%s %s' % (f['id'], f['error']), file=sys.stderr)
+    return 0
+
+
+if __name__ == '__main__':
+    sys.exit(main())

package/scripts/data/wechat_emoji.json

@@ -0,0 +1,225 @@
+{
+  "_comment": "WeChat face codes -> Unicode emoji. WeChat ships its own artwork (private-use code points such as U+E414 plus smileys/*.png), so several entries here are the closest standard equivalent rather than an exact match. Unknown codes are left untouched by the converters.",
+  "_aliases_note": "Chinese face names as they appear in WeChat message text, e.g. [害羞].",
+
+  "微笑": "😊",
+  "撇嘴": "😖",
+  "色": "😍",
+  "发呆": "😳",
+  "得意": "😎",
+  "流泪": "😭",
+  "害羞": "😳",
+  "闭嘴": "🤐",
+  "睡": "😴",
+  "大哭": "😭",
+  "尴尬": "😅",
+  "发怒": "😡",
+  "调皮": "😜",
+  "呲牙": "😁",
+  "惊讶": "😲",
+  "难过": "😔",
+  "酷": "😎",
+  "冷汗": "😰",
+  "抓狂": "😫",
+  "吐": "🤮",
+  "偷笑": "🤭",
+  "可爱": "😊",
+  "白眼": "🙄",
+  "傲慢": "😤",
+  "饥饿": "🤤",
+  "困": "😪",
+  "惊恐": "😱",
+  "流汗": "😓",
+  "憨笑": "😄",
+  "大兵": "😎",
+  "奋斗": "😤",
+  "咒骂": "🤬",
+  "疑问": "❓",
+  "嘘": "🤫",
+  "晕": "😵",
+  "折磨": "😩",
+  "衰": "😞",
+  "骷髅": "💀",
+  "敲打": "🔨",
+  "再见": "👋",
+  "擦汗": "😅",
+  "抠鼻": "🤧",
+  "鼓掌": "👏",
+  "糗大了": "😳",
+  "坏笑": "😏",
+  "左哼哼": "😤",
+  "右哼哼": "😤",
+  "哈欠": "🥱",
+  "鄙视": "🖕",
+  "委屈": "🥺",
+  "快哭了": "😢",
+  "阴险": "😏",
+  "亲亲": "😘",
+  "吓": "😱",
+  "可怜": "🥺",
+  "菜刀": "🔪",
+  "西瓜": "🍉",
+  "啤酒": "🍺",
+  "篮球": "🏀",
+  "乒乓": "🏓",
+  "咖啡": "☕",
+  "饭": "🍚",
+  "猪头": "🐷",
+  "玫瑰": "🌹",
+  "凋谢": "🥀",
+  "示爱": "❤️",
+  "爱心": "❤️",
+  "心碎": "💔",
+  "蛋糕": "🎂",
+  "闪电": "⚡",
+  "炸弹": "💣",
+  "刀": "🔪",
+  "足球": "⚽",
+  "瓢虫": "🐞",
+  "便便": "💩",
+  "月亮": "🌙",
+  "太阳": "☀️",
+  "礼物": "🎁",
+  "拥抱": "🤗",
+  "强": "👍",
+  "弱": "👎",
+  "握手": "🤝",
+  "胜利": "✌️",
+  "抱拳": "🙏",
+  "勾引": "💅",
+  "拳头": "👊",
+  "差劲": "👎",
+  "爱你": "🤟",
+  "NO": "🙅",
+  "OK": "👌",
+  "爱情": "💑",
+  "飞吻": "😘",
+  "跳跳": "🕺",
+  "发抖": "😨",
+  "怄火": "😠",
+  "转圈": "🌀",
+  "磕头": "🙇",
+  "回头": "🔙",
+  "跳绳": "🏃",
+  "挥手": "👋",
+  "激动": "😆",
+  "街舞": "🕺",
+  "献吻": "😘",
+  "左太极": "☯️",
+  "右太极": "☯️",
+
+  "捂脸": "🤦",
+  "奸笑": "😏",
+  "机智": "🤓",
+  "皱眉": "😟",
+  "耶": "✌️",
+  "吃瓜": "🍉",
+  "加油": "💪",
+  "汗": "😅",
+  "天啊": "😱",
+  "社会社会": "🤝",
+  "哇": "😲",
+  "666": "👍",
+  "让我看看": "👀",
+  "叹气": "😮‍💨",
+  "苦涩": "😖",
+  "裂开": "🤯",
+  "翻白眼": "🙄",
+  "破涕为笑": "😂",
+  "无语": "😑",
+  "失望": "😞",
+  "恐惧": "😱",
+  "庆祝": "🎉",
+  "红包": "🧧",
+  "福": "🧧",
+  "發": "🀄",
+  "烟花": "🎆",
+  "爆竹": "🧨",
+  "合十": "🙏",
+
+  "旺柴": "🐶",
+  "囧": "😅",
+  "嘴唇": "💋",
+  "好的": "👌",
+  "嘿哈": "😆",
+  "打脸": "🤦",
+  "愉快": "😊",
+  "笑脸": "😊",
+  "生病": "🤒",
+  "脸红": "😊",
+  "悠闲": "😌",
+  "Emm": "🤔",
+  "開红包": "🧧",
+  "开红包": "🧧",
+  "Salute": "🫡",
+
+  "Smile": "😊",
+  "Grimace": "😖",
+  "Drool": "😍",
+  "Scowl": "😳",
+  "CoolGuy": "😎",
+  "Sob": "😭",
+  "Shy": "😳",
+  "Silent": "🤐",
+  "Sleep": "😴",
+  "Cry": "😭",
+  "Awkward": "😅",
+  "Angry": "😡",
+  "Tongue": "😜",
+  "Grin": "😁",
+  "Surprise": "😲",
+  "Frown": "😔",
+  "Ruthless": "😎",
+  "Blush": "😓",
+  "Scream": "😫",
+  "Puke": "🤮",
+  "Chuckle": "🤭",
+  "Joyful": "😄",
+  "Slight": "😁",
+  "Smug": "😤",
+  "Hungry": "🤤",
+  "Drowsy": "😪",
+  "Panic": "😱",
+  "Sweat": "😓",
+  "Laugh": "😄",
+  "Commando": "😎",
+  "Determined": "😤",
+  "Scold": "🤬",
+  "Shocked": "❓",
+  "Shhh": "🤫",
+  "Dizzy": "😵",
+  "Tormented": "😩",
+  "Toasted": "😞",
+  "Skull": "💀",
+  "Hammer": "🔨",
+  "Wave": "👋",
+  "Speechless": "😅",
+  "NosePick": "🤧",
+  "Clap": "👏",
+  "Shame": "😳",
+  "Trick": "😏",
+  "Yawn": "🥱",
+  "Pooh-pooh": "🙄",
+  "Shrunken": "🥺",
+  "TearingUp": "😢",
+  "Sly": "😏",
+  "Kiss": "😘",
+  "Wrath": "😠",
+  "Whimper": "🥺",
+  "Cleaver": "🔪",
+  "Watermelon": "🍉",
+  "Beer": "🍺",
+  "Basketball": "🏀",
+  "PingPong": "🏓",
+  "Coffee": "☕",
+  "Rice": "🍚",
+  "Pig": "🐷",
+  "Rose": "🌹",
+  "Wilt": "🥀",
+  "Lips": "❤️",
+  "Heart": "❤️",
+  "BrokenHeart": "💔",
+  "Cake": "🎂",
+  "Lightning": "⚡",
+  "ThumbsUp": "👍"
+}

package/scripts/export_chat_html.py

@@ -2379,7 +2379,9 @@ def main():
     # Voice: load the payloads so a cached transcript can be shown. Anything
     # not yet transcribed simply renders as `[语音 N″]`.
     if _WECHAT_VOICE:
-        VOICE_STATE['cache'] = wechat_voice.TranscriptCache(os.path.join(args.out, '.voice-cache'))
+        # 缓存的家由 `wechat_voice` 定义：以前这里按 `--out` 拼，于是同一段音频在不同
+        # 输出目录下会被各转写一遍，而卡片那条线读的又是另一处（2026-10-01）。
+        VOICE_STATE['cache'] = wechat_voice.TranscriptCache(wechat_voice.DEFAULT_CACHE_DIR)
         try:
             VOICE_STATE['map'] = wechat_voice.load_voice_map(
                 args.db, args.key, args.salt, args.passphrase, args.talker)

package/scripts/fav_notes.py

@@ -99,7 +99,10 @@ def classify_source(item: dict) -> str:
 def favorite_text(item: dict, fetch=None):
     """一条收藏的正文：本地文本优先，没有（或太短）就抓链接。返回 (正文, 来源说明)。
 
-    收藏分两种：笔记/文本类**自带内容**（`desc`/`content`），文章类只有 `link`。
+    收藏分两种：笔记/文本类**自带内容**（`desc`/`content`），文章类**基本**只有 `link`——
+    "基本"是量出来的：本机 1,234 篇文章里有 **7 条**的网页描述（`weburlitem/pagedesc`）超过
+    `MIN_ARTICLE_CHARS`，会被当成"自带正文"（那是**摘要不是正文**）。门槛 200 字挡住了短噪声
+    （例如本机 20 条 `type=1` 的 `desc` 只有 49 字、且是路径式字符串 ⇒ 全部落 none）。
     抓取复用日报那条线的 `fetch_article_cached`——不重写第二份（那套微信 UA 绕 WAF、
     gzip、重试是踩出来的），而且**带缓存**：日报抓过的文章这里零网络。