fix: preserve URL-only papers and normalize sender IDs

fastinsight/parse.py

@@ -320,12 +320,13 @@ def cleaned_text(self) -> str:
 
 
 def _sender_id(sender: Dict[str, Any]) -> str:
-    sender_id = str(sender.get("id") or sender.get("sender_id") or "").strip()
-    if sender_id:
-        return sender_id
-    nested = sender.get("sender_id")
-    if isinstance(nested, dict):
-        return str(nested.get("open_id") or nested.get("user_id") or "").strip()
+    for value in (sender.get("id"), sender.get("sender_id")):
+        if isinstance(value, str) and value.strip():
+            return value.strip()
+        if isinstance(value, dict):
+            nested = value.get("open_id") or value.get("user_id")
+            if isinstance(nested, str) and nested.strip():
+                return nested.strip()
     return ""
 
 

scripts/auto_pipeline.py

@@ -31,9 +31,13 @@ def _text(value):
 
 
 def _verification_query(paper, title):
-    """Combine identifiers with the best available title or content search query."""
+    """Keep supplied locator evidence while adding the best search text."""
     search_query = _text(paper.get("search_query"))
-    pieces = [search_query or _text(title)]
+    url = _text(paper.get("url"))
+    search_text = search_query or _text(title)
+    pieces = [url]
+    if search_text and search_text != url:
+        pieces.append(search_text)
     doi = _text(paper.get("doi"))
     arxiv_id = _text(paper.get("arxiv_id"))
     if doi:
@@ -178,7 +182,7 @@ def enrich_paper(paper):
     search_query = _text(paper.get("search_query"))
     content_query = _text(paper.get("text"))[:500]
     paper = _normalize_paper(paper)
-    query = search_query or source_title or content_query
+    query = search_query or source_title or content_query or _text(paper.get("url"))
     if not query:
         return paper
     with tempfile.TemporaryDirectory(prefix="fastinsight-verify-") as temp_dir:

tests/test_parse.py

@@ -18,6 +18,18 @@
 
 
 class ParseTests(unittest.TestCase):
+    def test_nested_sender_id_is_normalized_without_dict_repr(self) -> None:
+        parsed = parse_event(
+            {
+                "event": {
+                    "sender": {"sender_id": {"open_id": "ou_nested", "user_id": "u_nested"}},
+                    "message": {"content": '{"text":"hello"}'},
+                }
+            }
+        )
+        self.assertEqual(parsed.sender_id, "ou_nested")
+        self.assertNotIn("{", parsed.sender_id)
+
     def test_command_with_mention(self) -> None:
         parsed = parse_feishu_message(
             {

tests/test_pipeline.py

@@ -58,6 +58,21 @@ def test_search_query_is_prioritized_for_verification(self) -> None:
         )
         self.assertEqual(query, "FirmAgent firmware fuzzing")
 
+    def test_verification_query_keeps_existing_url(self) -> None:
+        query = _verification_query(
+            {"url": "https://arxiv.org/abs/1706.03762", "title": "A description"},
+            "A description",
+        )
+        self.assertEqual(query, "https://arxiv.org/abs/1706.03762 A description")
+
+    def test_url_only_paper_reaches_verifier(self) -> None:
+        with patch("scripts.auto_pipeline._run_json", return_value=({}, 0)) as run:
+            from scripts.auto_pipeline import enrich_paper
+
+            enrich_paper({"url": "https://example.test/paper"})
+        command = run.call_args.args[0]
+        self.assertEqual(command[command.index("--query") + 1], "https://example.test/paper")
+
 
 if __name__ == "__main__":
     unittest.main()