add internet reference in playground chat stream

Author: yuan.wang

src/memos/api/handlers/chat_handler.py

@@ -394,7 +394,12 @@ def generate_chat_response() -> Generator[str, None, None]:
 
                     # Prepare reference data
                     reference = prepare_reference_data(filtered_memories)
+                    # get internet reference
+                    internet_reference = self._get_internet_reference(
+                        search_response.data.get("text_mem")[0]["memories"]
+                    )
                     yield f"data: {json.dumps({'type': 'reference', 'data': reference})}\n\n"
+                    yield f"data: {json.dumps({'type': 'internet_reference', 'data': internet_reference})}\n\n"
 
                     # Prepare preference markdown string
                     if chat_req.include_preference:
@@ -535,6 +540,23 @@ def generate_chat_response() -> Generator[str, None, None]:
             self.logger.error(f"Failed to start chat stream: {traceback.format_exc()}")
             raise HTTPException(status_code=500, detail=str(traceback.format_exc())) from err
 
+    def _get_internet_reference(
+        self, search_response: list[dict[str, any]]
+    ) -> list[dict[str, any]]:
+        """Get internet reference from search response."""
+        unique_set = set()
+        result = []
+
+        for item in search_response:
+            meta = item.get("metadata", {})
+            if meta.get("source") == "web" and meta.get("internet_info"):
+                info = meta.get("internet_info")
+                key = json.dumps(info, sort_keys=True)
+                if key not in unique_set:
+                    unique_set.add(key)
+                    result.append(info)
+        return result
+
     def _build_pref_md_string_for_playground(self, pref_mem_list: list[any]) -> str:
         """Build preference markdown string for playground."""
         explicit = []

src/memos/memories/textual/tree_text_memory/retrieve/bochasearch.py

@@ -200,9 +200,11 @@ def _process_result(
         """Process one Bocha search result into TextualMemoryItem."""
         title = result.get("name", "")
         content = result.get("summary", "") or result.get("snippet", "")
-        summary = result.get("snippet", "")
+        summary = result.get("summary", "") or result.get("snippet", "")
         url = result.get("url", "")
         publish_time = result.get("datePublished", "")
+        site_name = result.get("siteName", "")
+        site_icon = result.get("siteIcon")
 
         if publish_time:
             try:
@@ -229,5 +231,12 @@ def _process_result(
             read_item_i.metadata.memory_type = "OuterMemory"
             read_item_i.metadata.sources = [SourceMessage(type="web", url=url)] if url else []
             read_item_i.metadata.visibility = "public"
+            read_item_i.metadata.internet_info = {
+                "title": title,
+                "url": url,
+                "site_name": site_name,
+                "site_icon": site_icon,
+                "summary": summary,
+            }
             memory_items.append(read_item_i)
         return memory_items

src/memos/memories/textual/tree_text_memory/retrieve/searcher.py

@@ -445,7 +445,7 @@ def _retrieve_from_internet(
         user_id: str | None = None,
     ):
         """Retrieve and rerank from Internet source"""
-        if not self.internet_retriever or mode == "fast":
+        if not self.internet_retriever:
             logger.info(f"[PATH-C] '{query}' Skipped (no retriever, fast mode)")
             return []
         if memory_type not in ["All"]: