feat(avatar): show multi-file knowledge upload progress

digital-avatar-app/backend/services/knowledge_vectorizer.py

@@ -74,11 +74,19 @@ class KnowledgeVectorizer:
             if not stored_name or not os.path.isfile(path):
                 raise FileNotFoundError("原文件不可用，请重新上传")
 
+            self._set_progress(db, doc, "extracting", 8)
             text = embeddings.extract_text(path, f".{doc.file_type}")
+            self._set_progress(db, doc, "chunking", 22)
             chunks = embeddings.chunk_text(text)
             if not chunks:
                 raise ValueError("文档没有可建立索引的文字内容")
-            vectors = embeddings.embed(chunks)
+            self._set_progress(db, doc, "embedding", 30)
+
+            def embedding_progress(done: int, total: int):
+                percent = 30 + int((done / max(1, total)) * 65)
+                self._set_progress(db, doc, "embedding", min(percent, 95))
+
+            vectors = embeddings.embed(chunks, on_progress=embedding_progress)
             if len(vectors) != len(chunks):
                 raise ValueError("向量服务返回数量与文档分段不一致")
 
@@ -103,6 +111,8 @@ class KnowledgeVectorizer:
             doc.vectorized_at = datetime.now(timezone.utc)
             doc.status = "ready"
             doc.error_message = ""
+            doc.index_stage = "ready"
+            doc.index_progress = 100
             db.commit()
             logger.info("Knowledge document %s indexed with %s chunks", doc.id, len(chunks))
         except Exception as exc:
@@ -116,10 +126,18 @@ class KnowledgeVectorizer:
                 failed_doc.chunk_count = 0
                 failed_doc.vectorized_at = None
                 failed_doc.error_message = str(exc)[:300] or "建立知识索引失败"
+                failed_doc.index_stage = "failed"
+                failed_doc.index_progress = 0
                 db.commit()
             logger.exception("Knowledge vectorization failed for %s: %s", doc_id, exc)
         finally:
             db.close()
 
+    @staticmethod
+    def _set_progress(db, doc, stage: str, progress: int):
+        doc.index_stage = stage
+        doc.index_progress = progress
+        db.commit()
+
 
 knowledge_vectorizer = KnowledgeVectorizer()