Minor tweats (#11271)

### What problem does this PR solve?

As title.

### Type of change

- [x] Refactoring

---------

Signed-off-by: Jin Hai <haijin.chn@gmail.com>

api/db/services/connector_service.py

@@ -70,7 +70,7 @@ class ConnectorService(CommonService):
     def rebuild(cls, kb_id:str, connector_id: str, tenant_id:str):
         e, conn = cls.get_by_id(connector_id)
         if not e:
-            return
+            return None
         SyncLogsService.filter_delete([SyncLogs.connector_id==connector_id, SyncLogs.kb_id==kb_id])
         docs = DocumentService.query(source_type=f"{conn.source}/{conn.id}", kb_id=kb_id)
         err = FileService.delete_docs([d.id for d in docs], tenant_id)
@@ -125,11 +125,11 @@ class SyncLogsService(CommonService):
             )
 
         query = query.distinct().order_by(cls.model.update_time.desc())
-        totbal = query.count()
+        total = query.count()
         if page_number:
             query = query.paginate(page_number, items_per_page)
 
-        return list(query.dicts()), totbal
+        return list(query.dicts()), total
 
     @classmethod
     def start(cls, id, connector_id):

api/db/services/dialog_service.py

@@ -342,7 +342,7 @@ def chat(dialog, messages, stream=True, **kwargs):
     if not dialog.kb_ids and not dialog.prompt_config.get("tavily_api_key"):
         for ans in chat_solo(dialog, messages, stream):
             yield ans
-        return
+        return None
 
     chat_start_ts = timer()
 
@@ -386,7 +386,7 @@ def chat(dialog, messages, stream=True, **kwargs):
         ans = use_sql(questions[-1], field_map, dialog.tenant_id, chat_mdl, prompt_config.get("quote", True), dialog.kb_ids)
         if ans:
             yield ans
-            return
+            return None
 
     for p in prompt_config["parameters"]:
         if p["key"] == "knowledge":
@@ -617,6 +617,8 @@ def chat(dialog, messages, stream=True, **kwargs):
         res["audio_binary"] = tts(tts_mdl, answer)
         yield res
 
+    return None
+
 
 def use_sql(question, field_map, tenant_id, chat_mdl, quota=True, kb_ids=None):
     sys_prompt = """
@@ -745,7 +747,7 @@ Please write the SQL, only SQL, without any other explanations or text.
 
 def tts(tts_mdl, text):
     if not tts_mdl or not text:
-        return
+        return None
     bin = b""
     for chunk in tts_mdl.tts(text):
         bin += chunk

api/db/services/document_service.py

@@ -113,7 +113,7 @@ class DocumentService(CommonService):
     def check_doc_health(cls, tenant_id: str, filename):
         import os
         MAX_FILE_NUM_PER_USER = int(os.environ.get("MAX_FILE_NUM_PER_USER", 0))
-        if MAX_FILE_NUM_PER_USER > 0 and DocumentService.get_doc_count(tenant_id) >= MAX_FILE_NUM_PER_USER:
+        if 0 < MAX_FILE_NUM_PER_USER <= DocumentService.get_doc_count(tenant_id):
             raise RuntimeError("Exceed the maximum file number of a free user!")
         if len(filename.encode("utf-8")) > FILE_NAME_LEN_LIMIT:
             raise RuntimeError("Exceed the maximum length of file name!")
@@ -464,7 +464,7 @@ class DocumentService(CommonService):
             cls.model.id == doc_id, Knowledgebase.status == StatusEnum.VALID.value)
         docs = docs.dicts()
         if not docs:
-            return
+            return None
         return docs[0]["tenant_id"]
 
     @classmethod
@@ -473,7 +473,7 @@ class DocumentService(CommonService):
         docs = cls.model.select(cls.model.kb_id).where(cls.model.id == doc_id)
         docs = docs.dicts()
         if not docs:
-            return
+            return None
         return docs[0]["kb_id"]
 
     @classmethod
@@ -486,7 +486,7 @@ class DocumentService(CommonService):
             cls.model.name == name, Knowledgebase.status == StatusEnum.VALID.value)
         docs = docs.dicts()
         if not docs:
-            return
+            return None
         return docs[0]["tenant_id"]
 
     @classmethod
@@ -533,7 +533,7 @@ class DocumentService(CommonService):
             cls.model.id == doc_id, Knowledgebase.status == StatusEnum.VALID.value)
         docs = docs.dicts()
         if not docs:
-            return
+            return None
         return docs[0]["embd_id"]
 
     @classmethod
@@ -569,7 +569,7 @@ class DocumentService(CommonService):
             .where(cls.model.name == doc_name)
         doc_id = doc_id.dicts()
         if not doc_id:
-            return
+            return None
         return doc_id[0]["id"]
 
     @classmethod
@@ -715,7 +715,7 @@ class DocumentService(CommonService):
                     prg = 1
                     status = TaskStatus.DONE.value
 
-                # only for special task and parsed docs and unfinised
+                # only for special task and parsed docs and unfinished
                 freeze_progress = special_task_running and doc_progress >= 1 and not finished
                 msg = "\n".join(sorted(msg))
                 info = {
@@ -974,13 +974,13 @@ def doc_upload_and_parse(conversation_id, file_objs, user_id):
 
     def embedding(doc_id, cnts, batch_size=16):
         nonlocal embd_mdl, chunk_counts, token_counts
-        vects = []
+        vectors = []
         for i in range(0, len(cnts), batch_size):
             vts, c = embd_mdl.encode(cnts[i: i + batch_size])
-            vects.extend(vts.tolist())
+            vectors.extend(vts.tolist())
             chunk_counts[doc_id] += len(cnts[i:i + batch_size])
             token_counts[doc_id] += c
-        return vects
+        return vectors
 
     idxnm = search.index_name(kb.tenant_id)
     try_create_idx = True
@@ -1011,15 +1011,15 @@ def doc_upload_and_parse(conversation_id, file_objs, user_id):
             except Exception:
                 logging.exception("Mind map generation error")
 
-        vects = embedding(doc_id, [c["content_with_weight"] for c in cks])
-        assert len(cks) == len(vects)
+        vectors = embedding(doc_id, [c["content_with_weight"] for c in cks])
+        assert len(cks) == len(vectors)
         for i, d in enumerate(cks):
-            v = vects[i]
+            v = vectors[i]
             d["q_%d_vec" % len(v)] = v
         for b in range(0, len(cks), es_bulk_size):
             if try_create_idx:
                 if not settings.docStoreConn.indexExist(idxnm, kb_id):
-                    settings.docStoreConn.createIdx(idxnm, kb_id, len(vects[0]))
+                    settings.docStoreConn.createIdx(idxnm, kb_id, len(vectors[0]))
                 try_create_idx = False
             settings.docStoreConn.insert(cks[b:b + es_bulk_size], idxnm, kb_id)
 

api/db/services/knowledgebase_service.py

@@ -424,6 +424,7 @@ class KnowledgebaseService(CommonService):
 
         # Default parser_config (align with kb_app.create) — do not accept external overrides
         payload["parser_config"] = get_parser_config(parser_id, kwargs.get("parser_config"))
+
         return payload