fix(chat): 修复导出 ZIP 写入冲突并支持按消息类型导出

- 修复导出时 ZIP entry 写句柄冲突（messages 先写临时文件再写入 ZIP）\n- 新增 message_types 过滤，进度/统计按所选类型计数\n- manifest.json 记录 messageTypes，避免导出含义歧义

src/wechat_decrypt_tool/chat_export_service.py

@@ -1,10 +1,10 @@
 from __future__ import annotations
 
 import heapq
-import io
 import json
 import re
 import sqlite3
+import tempfile
 import threading
 import time
 import uuid
@@ -87,6 +87,57 @@ def _is_md5(s: str) -> bool:
     return bool(re.fullmatch(r"(?i)[0-9a-f]{32}", str(s or "").strip()))
 
 
+def _normalize_render_type_key(value: Any) -> str:
+    v = str(value or "").strip()
+    if not v:
+        return ""
+    if v == "redPacket":
+        return "redpacket"
+    lower = v.lower()
+    if lower in {"redpacket", "red_packet", "red-packet", "redenvelope", "red_envelope"}:
+        return "redpacket"
+    return lower
+
+
+def _render_types_to_local_types(render_types: set[str]) -> Optional[set[int]]:
+    rt = {str(x or "").strip() for x in (render_types or set())}
+    rt = {x for x in rt if x}
+    if not rt:
+        return None
+
+    out: set[int] = set()
+    for k in rt:
+        if k == "text":
+            out.add(1)
+        elif k == "image":
+            out.add(3)
+        elif k == "voice":
+            out.add(34)
+        elif k == "video":
+            out.update({43, 62})
+        elif k == "emoji":
+            out.add(47)
+        elif k == "voip":
+            out.add(50)
+        elif k == "system":
+            out.update({10000, 266287972401})
+        elif k == "quote":
+            out.add(244813135921)
+            out.add(49)  # Some quote messages are embedded as appmsg (local_type=49).
+        elif k in {"link", "file", "transfer", "redpacket"}:
+            out.add(49)
+        else:
+            # Unknown type: cannot safely prefilter by local_type.
+            return None
+    return out
+
+
+def _should_estimate_by_local_type(render_types: set[str]) -> bool:
+    # Only estimate counts when every requested type maps 1:1 to local_type.
+    # App messages (local_type=49) are heterogeneous and cannot be counted accurately without parsing.
+    return not bool(render_types & {"link", "file", "transfer", "redpacket", "quote"})
+
+
 @dataclass
 class ExportProgress:
     conversations_total: int = 0
@@ -183,6 +234,7 @@ class ChatExportManager:
         include_official: bool,
         include_media: bool,
         media_kinds: list[MediaKind],
+        message_types: list[str],
         allow_process_key_extract: bool,
         privacy_mode: bool,
         file_name: Optional[str],
@@ -204,6 +256,7 @@ class ChatExportManager:
                 "includeOfficial": bool(include_official),
                 "includeMedia": bool(include_media),
                 "mediaKinds": media_kinds,
+                "messageTypes": list(dict.fromkeys([str(t or "").strip() for t in (message_types or []) if str(t or "").strip()])),
                 "allowProcessKeyExtract": bool(allow_process_key_extract),
                 "privacyMode": bool(privacy_mode),
                 "fileName": str(file_name or "").strip(),
@@ -267,6 +320,18 @@ class ChatExportManager:
         st = int(opts.get("startTime") or 0) or None
         et = int(opts.get("endTime") or 0) or None
 
+        message_types_raw = opts.get("messageTypes") or []
+        want_types: Optional[set[str]] = None
+        if message_types_raw:
+            parts = [_normalize_render_type_key(x) for x in message_types_raw]
+            want = {p for p in parts if p}
+            if want:
+                want_types = want
+
+        local_types = _render_types_to_local_types(want_types) if want_types else None
+        can_estimate = (want_types is None) or _should_estimate_by_local_type(want_types)
+        estimate_local_types = local_types if (want_types and can_estimate) else None
+
         target_usernames = _resolve_export_targets(
             account_dir=account_dir,
             scope=scope,
@@ -391,11 +456,15 @@ class ChatExportManager:
                         job.progress.current_conversation_messages_total = 0
 
                     try:
+                        if not can_estimate:
+                            estimated_total = 0
+                        else:
                             estimated_total = _estimate_conversation_message_count(
                                 account_dir=account_dir,
                                 conv_username=conv_username,
                                 start_time=st,
                                 end_time=et,
+                                local_types=estimate_local_types,
                             )
                     except Exception:
                         estimated_total = 0
@@ -430,6 +499,8 @@ class ChatExportManager:
                             conv_is_group=conv_is_group,
                             start_time=st,
                             end_time=et,
+                            want_types=want_types,
+                            local_types=local_types,
                             resource_conn=resource_conn,
                             resource_chat_id=chat_id,
                             head_image_conn=head_image_conn,
@@ -456,6 +527,8 @@ class ChatExportManager:
                             conv_is_group=conv_is_group,
                             start_time=st,
                             end_time=et,
+                            want_types=want_types,
+                            local_types=local_types,
                             resource_conn=resource_conn,
                             resource_chat_id=chat_id,
                             head_image_conn=head_image_conn,
@@ -496,6 +569,7 @@ class ChatExportManager:
                     "filters": {
                         "startTime": st,
                         "endTime": et,
+                        "messageTypes": sorted(want_types) if want_types else None,
                         "includeHidden": include_hidden,
                         "includeOfficial": include_official,
                     },
@@ -612,6 +686,7 @@ def _estimate_conversation_message_count(
     conv_username: str,
     start_time: Optional[int],
     end_time: Optional[int],
+    local_types: Optional[set[int]] = None,
 ) -> int:
     total = 0
     for db_path in _iter_message_db_paths(account_dir):
@@ -623,6 +698,12 @@ def _estimate_conversation_message_count(
             quoted = _quote_ident(table)
             where = []
             params: list[Any] = []
+            if local_types:
+                lt = sorted({int(x) for x in local_types if int(x) != 0})
+                if lt:
+                    placeholders = ",".join(["?"] * len(lt))
+                    where.append(f"local_type IN ({placeholders})")
+                    params.extend(lt)
             if start_time is not None:
                 where.append("create_time >= ?")
                 params.append(int(start_time))
@@ -658,6 +739,7 @@ def _iter_rows_for_conversation(
     conv_username: str,
     start_time: Optional[int],
     end_time: Optional[int],
+    local_types: Optional[set[int]] = None,
 ) -> Iterable[_Row]:
     db_paths = _iter_message_db_paths(account_dir)
     if not db_paths:
@@ -691,6 +773,12 @@ def _iter_rows_for_conversation(
             quoted = _quote_ident(table_name)
             where = []
             params: list[Any] = []
+            if local_types:
+                lt = sorted({int(x) for x in local_types if int(x) != 0})
+                if lt:
+                    placeholders = ",".join(["?"] * len(lt))
+                    where.append(f"m.local_type IN ({placeholders})")
+                    params.extend(lt)
             if start_time is not None:
                 where.append("m.create_time >= ?")
                 params.append(int(start_time))
@@ -1100,6 +1188,8 @@ def _write_conversation_json(
     conv_is_group: bool,
     start_time: Optional[int],
     end_time: Optional[int],
+    want_types: Optional[set[str]],
+    local_types: Optional[set[int]],
     resource_conn: Optional[sqlite3.Connection],
     resource_chat_id: Optional[int],
     head_image_conn: Optional[sqlite3.Connection],
@@ -1118,8 +1208,11 @@ def _write_conversation_json(
     arcname = f"{conv_dir}/messages.json"
     exported = 0
 
-    with zf.open(arcname, "w") as fp:
+    # NOTE: Do not keep an entry handle opened while also writing other entries (avatars/media).
-        tw = io.TextIOWrapper(fp, encoding="utf-8", newline="\n")
+    # zipfile forbids interleaving writes; stream to a temp file then add it to zip at the end.
+    with tempfile.TemporaryDirectory(prefix="wechat_chat_export_") as tmp_dir:
+        tmp_path = Path(tmp_dir) / "messages.json"
+        with open(tmp_path, "w", encoding="utf-8", newline="\n") as tw:
             tw.write("{\n")
             tw.write("  \"schemaVersion\": 1,\n")
             tw.write(f"  \"exportedAt\": {json.dumps(_now_iso(), ensure_ascii=False)},\n")
@@ -1140,7 +1233,11 @@ def _write_conversation_json(
             tw.write(
                 "  \"filters\": "
                 + json.dumps(
-                {"startTime": int(start_time) if start_time else None, "endTime": int(end_time) if end_time else None},
+                    {
+                        "startTime": int(start_time) if start_time else None,
+                        "endTime": int(end_time) if end_time else None,
+                        "messageTypes": sorted(want_types) if want_types else None,
+                    },
                     ensure_ascii=False,
                 )
                 + ",\n"
@@ -1149,12 +1246,15 @@ def _write_conversation_json(
 
             sender_alias_map: dict[str, int] = {}
             first = True
+            scanned = 0
             for row in _iter_rows_for_conversation(
                 account_dir=account_dir,
                 conv_username=conv_username,
                 start_time=start_time,
                 end_time=end_time,
+                local_types=local_types,
             ):
+                scanned += 1
                 msg = _parse_message_for_export(
                     row=row,
                     conv_username=conv_username,
@@ -1162,6 +1262,13 @@ def _write_conversation_json(
                     resource_conn=resource_conn,
                     resource_chat_id=resource_chat_id,
                 )
+                if want_types:
+                    rt_key = _normalize_render_type_key(msg.get("renderType"))
+                    if rt_key not in want_types:
+                        if scanned % 500 == 0 and job.cancel_requested:
+                            raise _JobCancelled()
+                        continue
+
                 su = str(msg.get("senderUsername") or "").strip()
                 if privacy_mode:
                     _privacy_scrub_message(msg, conv_is_group=conv_is_group, sender_alias_map=sender_alias_map)
@@ -1203,12 +1310,14 @@ def _write_conversation_json(
                     job.progress.messages_exported += 1
                     job.progress.current_conversation_messages_exported = exported
 
-            if exported % 200 == 0 and job.cancel_requested:
+                if scanned % 500 == 0 and job.cancel_requested:
                     raise _JobCancelled()
 
             tw.write("\n  ]\n")
             tw.write("}\n")
             tw.flush()
+
+        zf.write(str(tmp_path), arcname)
     return exported
 
 
@@ -1223,6 +1332,8 @@ def _write_conversation_txt(
     conv_is_group: bool,
     start_time: Optional[int],
     end_time: Optional[int],
+    want_types: Optional[set[str]],
+    local_types: Optional[set[int]],
     resource_conn: Optional[sqlite3.Connection],
     resource_chat_id: Optional[int],
     head_image_conn: Optional[sqlite3.Connection],
@@ -1241,8 +1352,10 @@ def _write_conversation_txt(
     arcname = f"{conv_dir}/messages.txt"
     exported = 0
 
-    with zf.open(arcname, "w") as fp:
+    # Same as JSON: write to temp file first to avoid zip interleaving writes.
-        tw = io.TextIOWrapper(fp, encoding="utf-8", newline="\n")
+    with tempfile.TemporaryDirectory(prefix="wechat_chat_export_") as tmp_dir:
+        tmp_path = Path(tmp_dir) / "messages.txt"
+        with open(tmp_path, "w", encoding="utf-8", newline="\n") as tw:
             if privacy_mode:
                 tw.write("会话: 已隐藏\n")
                 tw.write("账号: hidden\n")
@@ -1255,16 +1368,21 @@ def _write_conversation_txt(
                 st = _format_ts(int(start_time)) if start_time else "不限"
                 et = _format_ts(int(end_time)) if end_time else "不限"
                 tw.write(f"时间范围: {st} ~ {et}\n")
+            if want_types:
+                tw.write(f"消息类型: {', '.join(sorted(want_types))}\n")
             tw.write(f"导出时间: {_now_iso()}\n")
             tw.write("\n")
 
             sender_alias_map: dict[str, int] = {}
+            scanned = 0
             for row in _iter_rows_for_conversation(
                 account_dir=account_dir,
                 conv_username=conv_username,
                 start_time=start_time,
                 end_time=end_time,
+                local_types=local_types,
             ):
+                scanned += 1
                 msg = _parse_message_for_export(
                     row=row,
                     conv_username=conv_username,
@@ -1272,6 +1390,13 @@ def _write_conversation_txt(
                     resource_conn=resource_conn,
                     resource_chat_id=resource_chat_id,
                 )
+                if want_types:
+                    rt_key = _normalize_render_type_key(msg.get("renderType"))
+                    if rt_key not in want_types:
+                        if scanned % 500 == 0 and job.cancel_requested:
+                            raise _JobCancelled()
+                        continue
+
                 su = str(msg.get("senderUsername") or "").strip()
                 if privacy_mode:
                     _privacy_scrub_message(msg, conv_is_group=conv_is_group, sender_alias_map=sender_alias_map)
@@ -1310,10 +1435,12 @@ def _write_conversation_txt(
                     job.progress.messages_exported += 1
                     job.progress.current_conversation_messages_exported = exported
 
-            if exported % 200 == 0 and job.cancel_requested:
+                if scanned % 500 == 0 and job.cancel_requested:
                     raise _JobCancelled()
 
             tw.flush()
+
+        zf.write(str(tmp_path), arcname)
     return exported
 
 

src/wechat_decrypt_tool/routers/chat_export.py

@@ -15,6 +15,7 @@ router = APIRouter(route_class=PathFixRoute)
 ExportFormat = Literal["json", "txt"]
 ExportScope = Literal["selected", "all", "groups", "singles"]
 MediaKind = Literal["image", "emoji", "video", "video_thumb", "voice", "file"]
+MessageType = Literal["text", "image", "emoji", "video", "voice", "file", "link", "transfer", "redPacket", "system", "quote", "voip"]
 
 
 class ChatExportCreateRequest(BaseModel):
@@ -31,6 +32,10 @@ class ChatExportCreateRequest(BaseModel):
         default_factory=lambda: ["image", "emoji", "video", "video_thumb", "voice", "file"],
         description="打包的媒体类型",
     )
+    message_types: list[MessageType] = Field(
+        default_factory=list,
+        description="导出消息类型（renderType）过滤：为空=导出全部消息；可多选（如仅 voice / 仅 transfer / 仅 redPacket 等）",
+    )
     allow_process_key_extract: bool = Field(
         False,
         description="预留字段：本项目不从微信进程提取媒体密钥，请使用 wx_key 获取并保存/批量解密",
@@ -55,6 +60,7 @@ async def create_chat_export(req: ChatExportCreateRequest):
         include_official=req.include_official,
         include_media=req.include_media,
         media_kinds=req.media_kinds,
+        message_types=req.message_types,
         allow_process_key_extract=req.allow_process_key_extract,
         privacy_mode=req.privacy_mode,
         file_name=req.file_name,