Add page id to extra_info (run-llama#542)

Add page id to extra_info

llama_hub/confluence/base.py

@@ -289,7 +289,9 @@ def process_page(self, page, include_attachments, text_maker):
             attachment_texts
         )
         return Document(
-            text=text, doc_id=page["id"], extra_info={"title": page["title"]}
+            text=text,
+            doc_id=page["id"],
+            extra_info={"title": page["title"], "page_id": page["id"]},
         )
 
     def process_attachment(self, page_id):

tests/tests_confluence/test_confluence_reader.py

@@ -146,9 +146,9 @@ def test_confluence_reader_load_data_by_page_ids(self, mock_confluence):
         assert len(documents) == 2
         assert all(isinstance(doc, Document) for doc in documents)
         assert documents[0].doc_id == "123"
-        assert documents[0].extra_info == {"title": "Page 123"}
+        assert documents[0].extra_info == {"title": "Page 123", "page_id": "123"}
         assert documents[1].doc_id == "456"
-        assert documents[1].extra_info == {"title": "Page 456"}
+        assert documents[1].extra_info == {"title": "Page 456", "page_id": "456"}
 
         assert mock_confluence.get_page_by_id.call_count == 2
 
@@ -205,9 +205,9 @@ def test_confluence_reader_load_data_by_space_id(self, mock_confluence):
         assert len(documents) == 2
         assert all(isinstance(doc, Document) for doc in documents)
         assert documents[0].doc_id == "123"
-        assert documents[0].extra_info == {"title": "Page 123"}
+        assert documents[0].extra_info == {"title": "Page 123", "page_id": "123"}
         assert documents[1].doc_id == "456"
-        assert documents[1].extra_info == {"title": "Page 456"}
+        assert documents[1].extra_info == {"title": "Page 456", "page_id": "456"}
 
         assert mock_confluence.get_page_by_id.call_count == 0
         assert mock_confluence.get_all_pages_by_label.call_count == 0
@@ -256,9 +256,9 @@ def test_confluence_reader_load_data_by_space_id_pagination(self, mock_confluenc
         assert len(documents) == 2
         assert all(isinstance(doc, Document) for doc in documents)
         assert documents[0].doc_id == "123"
-        assert documents[0].extra_info == {"title": "Page 123"}
+        assert documents[0].extra_info == {"title": "Page 123", "page_id": "123"}
         assert documents[1].doc_id == "456"
-        assert documents[1].extra_info == {"title": "Page 456"}
+        assert documents[1].extra_info == {"title": "Page 456", "page_id": "456"}
 
         assert mock_confluence.get_page_by_id.call_count == 0
         assert mock_confluence.get_all_pages_by_label.call_count == 0