fix(import): avoid deduplicating WeFlow local IDs

packages/parser-native/src/weflow.rs

@@ -49,17 +49,17 @@ fn convert_message_type(type_str: Option<&str>) -> u32 {
     }
 }
 
-/// JS `String(x)` semantics for the localId field.
-fn js_string(value: Option<&Value>) -> String {
-    match value {
-        None => "undefined".to_string(),
-        Some(Value::Null) => "null".to_string(),
-        Some(Value::String(s)) => s.clone(),
-        Some(Value::Bool(b)) => b.to_string(),
-        Some(Value::Number(n)) => n.to_string(),
-        // Arrays/objects are pathological for an ID field; compact JSON is a
-        // reasonable stand-in for JS coercion here.
-        Some(other) => serde_json::to_string(other).unwrap_or_default(),
+fn normalize_platform_message_id(value: Option<&Value>) -> Option<String> {
+    let raw = match value {
+        Some(Value::String(value)) => value.clone(),
+        Some(Value::Number(value)) => value.to_string(),
+        _ => return None,
+    };
+    let id = js_trim(&raw);
+    if id.is_empty() || id == "0" {
+        None
+    } else {
+        Some(id.to_string())
     }
 }
 
@@ -272,8 +272,7 @@ pub fn parse_weflow(
             };
 
             messages.push(NativeMessage {
-                // WeFlow always emits an id string (String(localId), JS semantics).
-                platform_message_id: Some(js_string(obj.get("localId"))),
+                platform_message_id: normalize_platform_message_id(obj.get("platformMessageId")),
                 sender_platform_id: sender.to_string(),
                 sender_account_name: account_name.to_string(),
                 sender_group_nickname: None,
@@ -330,7 +329,7 @@ mod tests {
       "session": {"wxid": "room@chatroom", "nickname": "群名", "displayName": "群显示名", "type": "群聊", "avatar": ""},
       "avatars": {"wxid_a": "data:image/jpeg;base64,AAA", "room@chatroom": "data:image/jpeg;base64,ROOM", "wxid_empty": ""},
       "messages": [
-        {"localId": 1, "createTime": 100, "type": "文本消息", "content": " hello ", "isSend": 0, "senderUsername": "wxid_a", "senderDisplayName": "Alice", "senderAvatarKey": "wxid_a"},
+        {"localId": 1, "platformMessageId": "server-1", "createTime": 100, "type": "文本消息", "content": " hello ", "isSend": 0, "senderUsername": "wxid_a", "senderDisplayName": "Alice", "senderAvatarKey": "wxid_a"},
         {"localId": 2, "createTime": 101, "type": "图片消息", "content": "[图片]", "isSend": 1, "senderUsername": "wxid_b", "senderDisplayName": "Bob"},
         {"localId": 3, "createTime": 102, "type": "系统消息", "content": "sys", "isSend": null, "senderUsername": "room@chatroom", "senderDisplayName": "群"},
         {"localId": 4, "createTime": 103, "type": "未知类型(88)", "content": null, "isSend": 0, "senderUsername": "wxid_a", "senderDisplayName": "Alice2"}
@@ -349,7 +348,11 @@ mod tests {
         assert_eq!(out.messages.len(), 3);
         assert_eq!(out.messages[0].content.as_deref(), Some("hello"));
         assert_eq!(out.messages[0].message_type, 0);
-        assert_eq!(out.messages[0].platform_message_id.as_deref(), Some("1"));
+        assert_eq!(
+            out.messages[0].platform_message_id.as_deref(),
+            Some("server-1")
+        );
+        assert_eq!(out.messages[1].platform_message_id, None);
         assert_eq!(out.messages[1].message_type, 1);
         assert_eq!(out.messages[2].message_type, 99);
         assert_eq!(out.messages[2].content, None);
@@ -386,11 +389,8 @@ mod tests {
         assert_eq!(out.messages[1].sender_account_name, "wxid_a");
         // Whitespace-only content is normalized to null.
         assert_eq!(out.messages[1].content, None);
-        // localId missing -> String(undefined).
-        assert_eq!(
-            out.messages[0].platform_message_id.as_deref(),
-            Some("undefined")
-        );
+        // Missing stable platform IDs remain absent so import can use content fingerprinting.
+        assert_eq!(out.messages[0].platform_message_id, None);
     }
 
     #[test]
@@ -466,12 +466,29 @@ mod tests {
     }
 
     #[test]
-    fn js_string_semantics_for_local_id() {
+    fn normalizes_stable_platform_message_id() {
         let n: Value = serde_json::from_str("3.5").unwrap();
-        assert_eq!(js_string(Some(&n)), "3.5");
-        assert_eq!(js_string(Some(&Value::Null)), "null");
-        assert_eq!(js_string(None), "undefined");
-        assert_eq!(js_string(Some(&Value::String("abc".into()))), "abc");
-        assert_eq!(js_string(Some(&Value::Bool(true))), "true");
+        assert_eq!(
+            normalize_platform_message_id(Some(&n)).as_deref(),
+            Some("3.5")
+        );
+        assert_eq!(
+            normalize_platform_message_id(Some(&Value::String(" server-1 ".into()))).as_deref(),
+            Some("server-1")
+        );
+        assert_eq!(
+            normalize_platform_message_id(Some(&Value::String("0".into()))),
+            None
+        );
+        assert_eq!(
+            normalize_platform_message_id(Some(&Value::Number(0.into()))),
+            None
+        );
+        assert_eq!(normalize_platform_message_id(Some(&Value::Null)), None);
+        assert_eq!(normalize_platform_message_id(None), None);
+        assert_eq!(
+            normalize_platform_message_id(Some(&Value::Bool(true))),
+            None
+        );
     }
 }

packages/parser/src/browser/weflow.test.ts

@@ -42,4 +42,46 @@ describe('WeFlow browser-safe JSON parser', () => {
     )
     assert.equal(progress.at(-1)?.progress, 1)
   })
+
+  it('does not treat export-local IDs as stable message IDs', async () => {
+    const content = JSON.stringify({
+      weflow: { version: '1.0.0' },
+      session: { wxid: 'group@chatroom', displayName: 'Project Team', type: '群聊' },
+      messages: [
+        {
+          localId: 1,
+          createTime: 1704164645,
+          type: '文本消息',
+          content: 'old message',
+          senderUsername: 'member-1',
+          senderDisplayName: 'Member 1',
+        },
+        {
+          localId: 1,
+          platformMessageId: 'server-2',
+          createTime: 1767225600,
+          type: '文本消息',
+          content: 'new message',
+          senderUsername: 'member-2',
+          senderDisplayName: 'Member 2',
+        },
+        {
+          localId: 2,
+          platformMessageId: '0',
+          createTime: 1767225601,
+          type: '文本消息',
+          content: 'message without a stable ID',
+          senderUsername: 'member-2',
+          senderDisplayName: 'Member 2',
+        },
+      ],
+    })
+
+    const result = await parseWeFlowJson(content, 'project.json')
+
+    assert.deepEqual(
+      result.messages.map((message) => message.platformMessageId),
+      [undefined, 'server-2', undefined]
+    )
+  })
 })

packages/parser/src/browser/weflow.ts

@@ -89,8 +89,9 @@ export async function parseWeFlowJson(
       members.set(sender, { platformId: sender, accountName, avatar })
     }
 
+    const platformMessageId = normalizePlatformMessageId(message.platformMessageId)
     messages.push({
-      platformMessageId: String(message.localId),
+      platformMessageId,
       senderPlatformId: sender,
       senderAccountName: accountName,
       senderGroupNickname: undefined,
@@ -170,6 +171,12 @@ function normalizeContent(value: unknown): string | null {
   return content || null
 }
 
+function normalizePlatformMessageId(value: unknown): string | undefined {
+  if (typeof value !== 'string' && typeof value !== 'number') return undefined
+  const id = String(value).trim()
+  return id && id !== '0' ? id : undefined
+}
+
 function nameFromFilename(fileName: string): string {
   const name = fileName
     .split(/[\\/]/)

packages/parser/src/formats/weflow.ts

@@ -83,6 +83,7 @@ interface WeFlowSession {
 
 interface WeFlowMessage {
   localId: number
+  platformMessageId?: string | number
   createTime: number // Unix 时间戳（秒）
   formattedTime: string
   type: string // 中文消息类型
@@ -95,6 +96,12 @@ interface WeFlowMessage {
   source: string
 }
 
+function normalizePlatformMessageId(value: unknown): string | undefined {
+  if (typeof value !== 'string' && typeof value !== 'number') return undefined
+  const id = String(value).trim()
+  return id && id !== '0' ? id : undefined
+}
+
 // ==================== 消息类型映射 ====================
 
 /**
@@ -527,8 +534,10 @@ async function* parseWeFlow(options: ParseOptions): AsyncGenerator<ParseEvent, v
         content = rawContent.trim() || null
       }
 
+      const platformMessageId = normalizePlatformMessageId(msg.platformMessageId)
+
       return {
-        platformMessageId: String(msg.localId), // 消息的平台原始 ID（用于回复关联查询）
+        platformMessageId,
         senderPlatformId: platformId,
         senderAccountName: accountName,
         // WeFlow 格式没有单独的群昵称字段

packages/parser/src/native/weflow-native.test.ts

@@ -82,6 +82,7 @@ const FULL_FIXTURE = {
   messages: [
     {
       localId: 1,
+      platformMessageId: 'server-1',
       createTime: 1704164645,
       formattedTime: '',
       type: '文本消息',
@@ -151,7 +152,7 @@ const FULL_FIXTURE = {
       senderDisplayName: '',
       senderAvatarKey: 'wxid_keyed',
     },
-    // Numeric localId with fraction, emoji content, unicode whitespace trim
+    // Export-local ID must not be used when no stable platform ID is present.
     {
       localId: 10.5,
       createTime: 1704164750,
@@ -161,7 +162,7 @@ const FULL_FIXTURE = {
       senderUsername: 'wxid_owner',
       senderDisplayName: 'Me2',
     },
-    // Missing localId -> String(undefined); missing content -> null
+    // Missing localId and platformMessageId -> no stable ID; missing content -> null
     {
       createTime: 1704164760,
       type: '红包卡片',
@@ -192,8 +193,9 @@ describe('weflow native parser parity', { skip: !nativeAvailable() && 'native mo
     assert.equal(nativeResult.messages[2].timestamp, null)
     assert.equal(nativeResult.messages[3].content, '{"title":"引用","items":[1,true,null],"nested":{"a":"b"}}')
     assert.equal(nativeResult.messages[4].content, null)
-    assert.equal(nativeResult.messages[5].platformMessageId, '10.5')
-    assert.equal(nativeResult.messages[6].platformMessageId, 'undefined')
+    assert.equal(nativeResult.messages[0].platformMessageId, 'server-1')
+    assert.equal(nativeResult.messages[5].platformMessageId, undefined)
+    assert.equal(nativeResult.messages[6].platformMessageId, undefined)
     const bob = nativeResult.members.find((m) => m.platformId === 'wxid_bob')
     assert.equal(bob?.avatar, 'data:image/jpeg;base64,KEYED')
     assert.equal(bob?.accountName, 'wxid_bob')