first commit

wxbot_framework/text.py

@@ -0,0 +1,107 @@
+"""纯 Python 的文本会话解析；不依赖 Windows 或原生 DLL。"""
+from __future__ import annotations
+
+import re
+
+WXID = re.compile(r"^[A-Za-z0-9_.@-]{1,127}$")
+SYSTEM_PEERS = frozenset({"weixin", "fmessage", "newsapp", "floatbottle"})
+GROUP_CONTROL_COMMANDS = frozenset({"/help", "成语接龙", "合并接龙", "退出接龙"})
+FOUR_CHINESE = re.compile(r"^[\u3400-\u4dbf\u4e00-\u9fff]{4}$")
+
+
+def own_group_command(text):
+    value = text.strip()
+    return bool(value in GROUP_CONTROL_COMMANDS or FOUR_CHINESE.fullmatch(value) or
+                value.split(maxsplit=1)[:1] in (["/绑定"], ["/总结"], ["/总结群聊"], ["/切换人格"], ["/清空"]))
+
+
+def validate_target(target):
+    if not isinstance(target, str) or not WXID.fullmatch(target):
+        raise ValueError("目标应为微信 ID 或群 ID，长度最多 127 个 ASCII 字符")
+    return target
+
+
+def validate_mentions(room, members):
+    validate_target(room)
+    if not room.endswith("@chatroom"):
+        raise ValueError("艾特消息的目标必须是群 ID")
+    if not isinstance(members, (list, tuple)) or not 1 <= len(members) <= 8:
+        raise ValueError("at_list 应包含 1–8 个成员微信 ID")
+    from .account import validate_account
+    result = tuple(validate_account(member) for member in members)
+    if len(set(result)) != len(result):
+        raise ValueError("at_list 不能重复")
+    return result
+
+
+class TextRouter:
+    """先识别本账号，再区分会话与作者，避免回复自己同步过来的消息。"""
+
+    def __init__(self, *, include_self_groups=False):
+        self.include_self_groups = include_self_groups
+        self.self_wxid = ""
+        self.unidentified = 0
+        self.unparsed_groups = 0
+        self.ignored_self = 0
+
+    def set_account(self, account):
+        from .account import validate_account
+        validate_account(account)
+        if self.self_wxid and self.self_wxid != account:
+            raise RuntimeError("当前登录账号发生变化，需要关闭旧连接后重新识别")
+        self.self_wxid = account
+
+    def convert(self, sender, receiver, text, *, sequence=0, thread_id=0):
+        validate_target(sender)
+        validate_target(receiver)
+        filehelper = "filehelper" in (sender, receiver)
+        incoming_group = sender.endswith("@chatroom")
+        # 文件传输助手的另一端，以及收到群消息的接收方，都是本账号。
+        identified = (receiver if sender == "filehelper" else sender) if filehelper else (
+            receiver if incoming_group else "")
+        if identified:
+            validate_target(identified)
+            if identified == "filehelper" or identified.endswith("@chatroom"):
+                return None
+            self.set_account(identified)
+        if not self.self_wxid:
+            self.unidentified += 1
+            return None
+
+        is_self = sender == self.self_wxid
+        if filehelper:
+            chat, author, room, body = "filehelper", sender, "", text
+        elif incoming_group:
+            if receiver != self.self_wxid:
+                return None
+            author, separator, body = text.partition(":\n")
+            # Sync AddMsg 的群文本包含“成员 ID:\n正文”；不猜测缺失的作者。
+            if not separator or not WXID.fullmatch(author) or author.endswith("@chatroom"):
+                self.unparsed_groups += 1
+                return None
+            chat = room = sender
+            is_self = author == self.self_wxid
+        elif receiver.endswith("@chatroom"):
+            if not is_self:
+                return None
+            chat = room = receiver
+            author, body = sender, text
+        elif receiver == self.self_wxid:
+            chat, author, room, body = sender, sender, "", text
+        elif is_self:
+            chat, author, room, body = receiver, sender, "", text
+        else:
+            return None
+        own_group_input = bool(room and own_group_command(body))
+        if not filehelper and is_self and not own_group_input and not (room and self.include_self_groups):
+            self.ignored_self += 1
+            return None
+        if chat.startswith("gh_") or chat in SYSTEM_PEERS:
+            return None
+        return {"msg": body, "content": body, "from_wxid": author if room else chat,
+                "room_wxid": room, "to_wxid": receiver, "wx_type": 1,
+                "chat_id": chat, "sender_id": author, "is_self": is_self,
+                "self_wxid": self.self_wxid, "native_sender": sender,
+                "native_receiver": receiver, "sequence": sequence,
+                "dispatch_to_plugins": not (room and is_self and not own_group_input),
+                "native_thread_id": thread_id}