fix: record generated wiki page versions (#17931)

api/apps/services/dataset_api_service.py

@@ -3174,7 +3174,7 @@ async def update_wiki_page(
     content_before = ""
     try:
         res = settings.docStoreConn.search(
-            select_fields=["id", "content_with_weight"],
+            select_fields=["id", "md_with_weight", "content_with_weight"],
             highlight_fields=[],
             condition={
                 "compile_kwd": [WIKI_PAGE_COMPILE_KWD],
@@ -3190,11 +3190,11 @@ async def update_wiki_page(
         )
         field_map = settings.docStoreConn.get_fields(
             res,
-            ["id", "content_with_weight"],
+            ["id", "md_with_weight", "content_with_weight"],
         )
         if field_map:
             row_id, row = next(iter(field_map.items()))
-            content_before = row.get("content_with_weight") or ""
+            content_before = row.get("md_with_weight") or row.get("content_with_weight") or ""
     except Exception:
         logging.exception(
             "update_wiki_page: lookup failed for kb=%s slug=%s",
@@ -3215,6 +3215,7 @@ async def update_wiki_page(
         ok = settings.docStoreConn.update(
             {"id": row_id},
             {
+                "md_with_weight": rendered,
                 "content_with_weight": rendered,
                 "summary_with_weight": summary,
                 "outlinks_kwd": list(outlinks),
@@ -3233,6 +3234,17 @@ async def update_wiki_page(
     if not ok:
         return True, None
 
+    refresh_idx = getattr(settings.docStoreConn, "refresh_idx", None)
+    if callable(refresh_idx):
+        try:
+            await thread_pool_exec(refresh_idx, index_nm)
+        except Exception:
+            logging.exception(
+                "update_wiki_page: index refresh failed for kb=%s slug=%s",
+                dataset_id,
+                full_slug,
+            )
+
     # Record a file_commit row on every real change. ``record_page_edit``
     # returns None for empty-diff saves, which we silently swallow.
     try:
@@ -3824,4 +3836,18 @@ async def clear_wiki(dataset_id: str, tenant_id: str):
             )
             deleted[kwd] = False
 
+    from api.db.services.file_commit_service import FileCommitService
+
+    if all(result is not False for result in deleted.values()):
+        try:
+            deleted["file_commit_history"] = FileCommitService.delete_all_page_history(dataset_id)
+        except Exception:
+            logging.exception(
+                "clear_wiki: failed to delete page version history for kb=%s",
+                dataset_id,
+            )
+            deleted["file_commit_history"] = False
+    else:
+        deleted["file_commit_history"] = False
+
     return True, {"deleted": deleted}

api/db/services/file_commit_service.py

@@ -801,6 +801,49 @@ class FileCommitService(CommonService):
 
         return commit_id
 
+    @classmethod
+    @DB.connection_context()
+    def delete_page_history(cls, kb_id: str, page_type: str, slug: str) -> int:
+        """Delete all stored versions for one Wiki page.
+
+        Wiki versions are represented by a ``FileCommit`` plus its single
+        ``FileCommitItem``.  They are not workspace commits, so removing the
+        page must remove both rows instead of leaving an orphaned history
+        that can reappear when the same slug is generated again.
+        """
+        file_id = _wiki_file_id(kb_id, slug)
+        commit_ids = [
+            row.commit_id
+            for row in FileCommitItem.select(FileCommitItem.commit_id).where((FileCommitItem.file_id == file_id) & (FileCommitItem.slug_kwd == slug) & (FileCommitItem.page_type_kwd == page_type))
+        ]
+        if not commit_ids:
+            return 0
+
+        with DB.atomic():
+            FileCommitItem.delete().where(FileCommitItem.commit_id.in_(commit_ids)).execute()
+            deleted = FileCommit.delete().where((FileCommit.folder_id == kb_id) & FileCommit.id.in_(commit_ids)).execute()
+        return deleted
+
+    @classmethod
+    @DB.connection_context()
+    def delete_all_page_history(cls, kb_id: str) -> int:
+        """Delete all Wiki page versions belonging to a knowledge base."""
+        commit_ids = [
+            row.commit_id
+            for row in (
+                FileCommitItem.select(FileCommitItem.commit_id)
+                .join(FileCommit, on=(FileCommit.id == FileCommitItem.commit_id))
+                .where((FileCommit.folder_id == kb_id) & FileCommitItem.slug_kwd.is_null(False) & FileCommitItem.page_type_kwd.is_null(False))
+            )
+        ]
+        if not commit_ids:
+            return 0
+
+        with DB.atomic():
+            FileCommitItem.delete().where(FileCommitItem.commit_id.in_(commit_ids)).execute()
+            deleted = FileCommit.delete().where((FileCommit.folder_id == kb_id) & FileCommit.id.in_(commit_ids)).execute()
+        return deleted
+
     @classmethod
     @DB.connection_context()
     def list_page_commits(