Merge branch 'main' into collapse-create-base-text-units

graphrag/index/workflows/v1/subflows/create_base_documents.py

@@ -59,19 +59,18 @@ def create_base_documents(
 
     rejoined = docs_with_text_units.merge(
         source,
-        left_on="id",
-        right_on="id",
+        on="id",
         how="right",
     )
     rejoined.rename(columns={"text": "raw_content"}, inplace=True)
     rejoined["id"] = rejoined["id"].astype(str)
 
     # attribute columns are converted to strings and then collapsed into a single json object
-    if document_attribute_columns is not None and len(document_attribute_columns) > 0:
+    if document_attribute_columns:
         for column in document_attribute_columns:
             rejoined[column] = rejoined[column].astype(str)
         rejoined["attributes"] = rejoined[document_attribute_columns].apply(
-            lambda row: ({**row}),
+            lambda row: {**row},
             axis=1,
         )
         rejoined.drop(columns=document_attribute_columns, inplace=True)

tests/verbs/util.py

@@ -73,12 +73,10 @@ def compare_outputs(
     """Compare the actual and expected dataframes, optionally specifying columns to compare.
     This uses assert_series_equal since we are sometimes intentionally omitting columns from the actual output."""
     cols = expected.columns if columns is None else columns
-    try:
-        assert len(actual) == len(expected)
-    except AssertionError:
-        print("Expected:", cols)
-        print("Actual:", actual.columns)
-        raise
+
+    assert len(actual) == len(
+        expected
+    ), f"Expected: {len(expected)}, Actual: {len(actual)}"
 
     for column in cols:
         assert column in actual.columns