Add parquet input support (fixes #40)

.github/workflows/python-ci.yml

@@ -54,4 +54,8 @@ jobs:
 
       - name: benchmark-qed - Check
         run: |
-          uv run poe check
+          uv run poe check
+
+      - name: benchmark-qed - Test
+        run: |
+          uv run poe test

.gitignore

@@ -171,7 +171,7 @@ cython_debug/
 #  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
 #  and can be added to the global gitignore or merged into this file.  For a more nuclear
 #  option (not recommended) you can uncomment the following to ignore the entire idea folder.
-#.idea/
+.idea/
 
 # Ruff stuff:
 .ruff_cache/

.semversioner/next-release/minor-20251217232258224287.json

@@ -0,0 +1,4 @@
+{
+  "type": "minor",
+  "description": "Support parquet inputs"
+}

benchmark_qed/autod/io/document.py

@@ -1,4 +1,5 @@
-# Copyright (c) 2025 Microsoft Corporation.
+# Copyright (c) Microsoft Corporation.
+# Licensed under the MIT License.
 """Load input files into Document objects."""
 
 import datetime
@@ -106,17 +107,16 @@ def load_text_dir(
     return documents
 
 
-def load_csv_doc(
-    file_path: str,
-    encoding: str = defs.FILE_ENCODING,
+def _load_docs_from_dataframe(
+    data_df: pd.DataFrame,
+    input_type: InputDataType,
+    title: str,
     text_tag: str = defs.TEXT_COLUMN,
     metadata_tags: list[str] | None = None,
     max_text_length: int | None = None,
 ) -> list[Document]:
-    """Load a CSV file and return a Document object."""
-    data_df = pd.read_csv(file_path, encoding=encoding)
-
     documents: list[Document] = []
+
     for index, row in enumerate(data_df.itertuples()):
         text = getattr(row, text_tag, "")
         if max_text_length is not None:
@@ -127,6 +127,7 @@ def load_csv_doc(
             for tag in metadata_tags:
                 if tag in data_df.columns:
                     metadata[tag] = getattr(row, tag)
+
         if "date_created" not in metadata:
             metadata["date_created"] = datetime.datetime.now(
                 tz=datetime.UTC
@@ -136,15 +137,33 @@ def load_csv_doc(
             Document(
                 id=str(uuid4()),
                 short_id=str(index),
-                title=str(file_path.replace(".csv", "")),
-                type="csv",
+                title=title,
+                type=str(input_type),
                 text=text,
                 attributes=metadata,
             )
         )
     return documents
 
 
+def load_csv_doc(
+    file_path: str,
+    encoding: str = defs.FILE_ENCODING,
+    text_tag: str = defs.TEXT_COLUMN,
+    metadata_tags: list[str] | None = None,
+    max_text_length: int | None = None,
+) -> list[Document]:
+    """Load a CSV file and return a Document object."""
+    return _load_docs_from_dataframe(
+        data_df=pd.read_csv(file_path, encoding=encoding),
+        input_type=InputDataType.CSV,
+        title=str(file_path.replace(".csv", "")),
+        text_tag=text_tag,
+        metadata_tags=metadata_tags,
+        max_text_length=max_text_length,
+    )
+
+
 def load_csv_dir(
     dir_path: str,
     encoding: str = defs.FILE_ENCODING,
@@ -171,6 +190,47 @@ def load_csv_dir(
     return documents
 
 
+def load_parquet_doc(
+    file_path: str,
+    text_tag: str = defs.TEXT_COLUMN,
+    metadata_tags: list[str] | None = None,
+    max_text_length: int | None = None,
+) -> list[Document]:
+    """Load Documents from a parquet file."""
+    return _load_docs_from_dataframe(
+        data_df=pd.read_parquet(file_path),
+        input_type=InputDataType.PARQUET,
+        title=str(file_path.replace(".parquet", "")),
+        text_tag=text_tag,
+        metadata_tags=metadata_tags,
+        max_text_length=max_text_length,
+    )
+
+
+def load_parquet_dir(
+    dir_path: str,
+    text_tag: str = defs.TEXT_COLUMN,
+    metadata_tags: list[str] | None = None,
+    max_text_length: int | None = None,
+) -> list[Document]:
+    """Load a directory of parquet files and return a list of Document objects."""
+    documents: list[Document] = []
+    for file_path in Path(dir_path).rglob("*.parquet"):
+        documents.extend(
+            load_parquet_doc(
+                file_path=str(file_path),
+                text_tag=text_tag,
+                metadata_tags=metadata_tags,
+                max_text_length=max_text_length,
+            )
+        )
+
+    for index, document in enumerate(documents):
+        document.short_id = str(index)
+
+    return documents
+
+
 def create_documents(
     input_path: str,
     input_type: InputDataType | str = InputDataType.JSON,
@@ -205,6 +265,13 @@ def create_documents(
                     metadata_tags=metadata_tags,
                     max_text_length=max_text_length,
                 )
+            case InputDataType.PARQUET:
+                documents = load_parquet_dir(
+                    dir_path=str(input_path),
+                    text_tag=text_tag,
+                    metadata_tags=metadata_tags,
+                    max_text_length=max_text_length,
+                )
             case _:
                 msg = f"Unsupported input type: {input_type}"
                 raise ValueError(msg)
@@ -236,6 +303,13 @@ def create_documents(
                     metadata_tags=metadata_tags,
                     max_text_length=max_text_length,
                 )
+            case InputDataType.PARQUET:
+                documents = load_parquet_doc(
+                    file_path=str(input_path),
+                    text_tag=text_tag,
+                    metadata_tags=metadata_tags,
+                    max_text_length=max_text_length,
+                )
             case _:
                 msg = f"Unsupported input type: {input_type}"
                 raise ValueError(msg)
@@ -254,18 +328,19 @@ def load_documents(
     """Read documents from a dataframe using pre-converted records."""
     records = df.to_dict("records")
 
+    def _get_attributes(row: dict) -> dict[str, Any]:
+        attributes = row.get("attributes", {})
+        selected_attributes = attributes_cols or []
+        return {attr: attributes.get(attr, None) for attr in selected_attributes}
+
     return [
         Document(
             id=row.get(id_col, str(uuid4())),
             short_id=row.get(short_id_col, str(index)),
             title=row.get(title_col, ""),
             type=row.get(type_col, ""),
             text=row.get(text_col, ""),
-            attributes=(
-                {col: row.get(col) for col in attributes_cols}
-                if attributes_cols
-                else {}
-            ),
+            attributes=_get_attributes(row),
         )
         for index, row in enumerate(records)
     ]

benchmark_qed/autod/io/enums.py

@@ -10,3 +10,4 @@ class InputDataType(StrEnum):
     JSON = "json"
     CSV = "csv"
     TEXT = "text"
+    PARQUET = "parquet"

pyproject.toml

@@ -84,7 +84,7 @@ test = "pytest tests"
 serve_docs = "mkdocs serve"
 build_docs = "mkdocs build"
 
-_test_with_coverage = 'coverage run --source=benchmark_qed -m pytest tests/unit'
+_test_with_coverage = 'coverage run --source=benchmark_qed -m pytest tests'
 _coverage_report = 'coverage report --fail-under=100 --show-missing --omit="benchmark_qed/doc_gen/__main__.py"'
 _generate_coverage_xml = 'coverage xml --omit="benchmark_qed/doc_gen/__main__.py"'
 _generate_coverage_html = 'coverage html --omit="benchmark_qed/doc_gen/__main__.py"'
@@ -120,3 +120,6 @@ sequence = [
 [tool.pyright]
 include = ["benchmark_qed", "tests"]
 exclude = ["**/__pycache__"]
+
+[pytest]
+tmp_path_retention_policy = "failed"

ruff.toml

@@ -95,3 +95,6 @@ builtins-ignorelist = ["input", "id", "bytes"]
 
 [lint.pydocstyle]
 convention = "numpy"
+
+[lint.flake8-copyright]
+notice-rgx = "(?i)Copyright \\(C\\) (\\d{4} )?Microsoft Corporation"

tests/autod/__init__.py

@@ -0,0 +1,2 @@
+# Copyright (c) Microsoft Corporation.
+# Licensed under the MIT License.

tests/autod/io/__init__.py

@@ -0,0 +1,2 @@
+# Copyright (c) Microsoft Corporation.
+# Licensed under the MIT License.