apache · andygrove · Jul 18, 2024 · Jul 9, 2024 · Jul 9, 2024 · Jul 9, 2024
diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
@@ -89,6 +89,10 @@ jobs:
           name: python-wheel-license
           path: .
 
+      # To remove once https://github.com/MaterializeInc/rust-protobuf-native/issues/20 is resolved
+      - name: Install gtest
+        uses: MarkusJx/googletest-installer@v1.1
+
       - name: Install Protoc
         uses: arduino/setup-protoc@v1
         with:

diff --git a/.github/workflows/test.yaml b/.github/workflows/test.yaml
@@ -55,6 +55,10 @@ jobs:
           version: '3.20.2'
           repo-token: ${{ secrets.GITHUB_TOKEN }}
 
+      # To remove once https://github.com/MaterializeInc/rust-protobuf-native/issues/20 is resolved
+      - name: Install gtest
+        uses: MarkusJx/googletest-installer@v1.1
+
       - name: Setup Python
         uses: actions/setup-python@v5
         with:

diff --git a/benchmarks/db-benchmark/join-datafusion.py b/benchmarks/db-benchmark/join-datafusion.py
@@ -74,7 +74,8 @@ def ans_shape(batches):
 ctx = df.SessionContext()
 print(ctx)
 
-# TODO we should be applying projections to these table reads to crete relations of different sizes
+# TODO we should be applying projections to these table reads to create relations
+# of different sizes
 
 x_data = pacsv.read_csv(
     src_jn_x, convert_options=pacsv.ConvertOptions(auto_dict_encode=True)

diff --git a/conda/recipes/meta.yaml b/conda/recipes/meta.yaml
@@ -51,6 +51,7 @@ requirements:
   run:
     - python
     - pyarrow >=11.0.0
+    - typing_extensions
 
 test:
   imports:

diff --git a/docs/source/api/functions.rst b/docs/source/api/functions.rst
@@ -24,4 +24,4 @@ Functions
 .. autosummary::
    :toctree: ../generated/
 
-   functions.functions
+   functions
diff --git a/docs/source/conf.py b/docs/source/conf.py
@@ -15,6 +15,8 @@
 # specific language governing permissions and limitations
 # under the License.
 
+"""Documenation generation."""
+
 # Configuration file for the Sphinx documentation builder.
 #
 # This file only contains a selection of the most common options. For a full
@@ -78,6 +80,25 @@
 
 autosummary_generate = True
 
+
+def autodoc_skip_member(app, what, name, obj, skip, options):
+    exclude_functions = "__init__"
+    exclude_classes = ("Expr", "DataFrame")
+
+    class_name = ""
+    if hasattr(obj, "__qualname__"):
+        if obj.__qualname__ is not None:
+            class_name = obj.__qualname__.split(".")[0]
+
+    should_exclude = name in exclude_functions and class_name in exclude_classes
+
+    return True if should_exclude else None
+
+
+def setup(app):
+    app.connect("autodoc-skip-member", autodoc_skip_member)
+
+
 # -- Options for HTML output -------------------------------------------------
 
 # The theme to use for HTML and HTML Help pages.  See the documentation for

diff --git a/examples/substrait.py b/examples/substrait.py
@@ -18,16 +18,13 @@
 from datafusion import SessionContext
 from datafusion import substrait as ss
 
-
 # Create a DataFusion context
 ctx = SessionContext()
 
 # Register table with context
 ctx.register_csv("aggregate_test_data", "./testing/data/csv/aggregate_test_100.csv")
 
-substrait_plan = ss.substrait.serde.serialize_to_plan(
-    "SELECT * FROM aggregate_test_data", ctx
-)
+substrait_plan = ss.Serde.serialize_to_plan("SELECT * FROM aggregate_test_data", ctx)
 # type(substrait_plan) -> <class 'datafusion.substrait.plan'>
 
 # Encode it to bytes
@@ -38,17 +35,15 @@
 # Alternative serialization approaches
 # type(substrait_bytes) -> <class 'bytes'>, at this point the bytes can be distributed to file, network, etc safely
 # where they could subsequently be deserialized on the receiving end.
-substrait_bytes = ss.substrait.serde.serialize_bytes(
-    "SELECT * FROM aggregate_test_data", ctx
-)
+substrait_bytes = ss.Serde.serialize_bytes("SELECT * FROM aggregate_test_data", ctx)
 
 # Imagine here bytes would be read from network, file, etc ... for example brevity this is omitted and variable is simply reused
 # type(substrait_plan) -> <class 'datafusion.substrait.plan'>
-substrait_plan = ss.substrait.serde.deserialize_bytes(substrait_bytes)
+substrait_plan = ss.Serde.deserialize_bytes(substrait_bytes)
 
 # type(df_logical_plan) -> <class 'substrait.LogicalPlan'>
-df_logical_plan = ss.substrait.consumer.from_substrait_plan(ctx, substrait_plan)
+df_logical_plan = ss.Consumer.from_substrait_plan(ctx, substrait_plan)
 
 # Back to Substrait Plan just for demonstration purposes
 # type(substrait_plan) -> <class 'datafusion.substrait.plan'>
-substrait_plan = ss.substrait.producer.to_substrait_plan(df_logical_plan)
+substrait_plan = ss.Producer.to_substrait_plan(df_logical_plan)
diff --git a/examples/tpch/_tests.py b/examples/tpch/_tests.py
@@ -21,6 +21,7 @@
 from datafusion import col, lit, functions as F
 from util import get_answer_file
 
+
 def df_selection(col_name, col_type):
     if col_type == pa.float64() or isinstance(col_type, pa.Decimal128Type):
         return F.round(col(col_name), lit(2)).alias(col_name)
@@ -29,14 +30,16 @@ def df_selection(col_name, col_type):
     else:
         return col(col_name)
 
+
 def load_schema(col_name, col_type):
     if col_type == pa.int64() or col_type == pa.int32():
         return col_name, pa.string()
     elif isinstance(col_type, pa.Decimal128Type):
         return col_name, pa.float64()
     else:
         return col_name, col_type
-
+
+
 def expected_selection(col_name, col_type):
     if col_type == pa.int64() or col_type == pa.int32():
         return F.trim(col(col_name)).cast(col_type).alias(col_name)
@@ -45,20 +48,23 @@ def expected_selection(col_name, col_type):
     else:
         return col(col_name)
 
+
 def selections_and_schema(original_schema):
-    columns = [ (c, original_schema.field(c).type) for c in original_schema.names ]
+    columns = [(c, original_schema.field(c).type) for c in original_schema.names]
 
-    df_selections = [ df_selection(c, t) for (c, t) in columns]
-    expected_schema = [ load_schema(c, t) for (c, t) in columns]
-    expected_selections = [ expected_selection(c, t) for (c, t) in columns]
+    df_selections = [df_selection(c, t) for (c, t) in columns]
+    expected_schema = [load_schema(c, t) for (c, t) in columns]
+    expected_selections = [expected_selection(c, t) for (c, t) in columns]
 
     return (df_selections, expected_schema, expected_selections)
 
+
 def check_q17(df):
     raw_value = float(df.collect()[0]["avg_yearly"][0].as_py())
     value = round(raw_value, 2)
     assert abs(value - 348406.05) < 0.001
 
+
 @pytest.mark.parametrize(
     ("query_code", "answer_file"),
     [
@@ -72,9 +78,7 @@ def check_q17(df):
         ("q08_market_share", "q8"),
         ("q09_product_type_profit_measure", "q9"),
         ("q10_returned_item_reporting", "q10"),
-        pytest.param(
-            "q11_important_stock_identification", "q11", 
-        ),
+        ("q11_important_stock_identification", "q11"),
         ("q12_ship_mode_order_priority", "q12"),
         ("q13_customer_distribution", "q13"),
         ("q14_promotion_effect", "q14"),
@@ -92,18 +96,26 @@ def test_tpch_query_vs_answer_file(query_code: str, answer_file: str):
     module = import_module(query_code)
     df = module.df
 
-    # Treat q17 as a special case. The answer file does not match the spec. Running at
-    # scale factor 1, we have manually verified this result does match the expected value.
+    # Treat q17 as a special case. The answer file does not match the spec.
+    # Running at scale factor 1, we have manually verified this result does
+    # match the expected value.
     if answer_file == "q17":
         return check_q17(df)
 
-    (df_selections, expected_schema, expected_selections) = selections_and_schema(df.schema())
+    (df_selections, expected_schema, expected_selections) = selections_and_schema(
+        df.schema()
+    )
 
     df = df.select(*df_selections)
 
     read_schema = pa.schema(expected_schema)
 
-    df_expected = module.ctx.read_csv(get_answer_file(answer_file), schema=read_schema, delimiter="|", file_extension=".out")
+    df_expected = module.ctx.read_csv(
+        get_answer_file(answer_file),
+        schema=read_schema,
+        delimiter="|",
+        file_extension=".out",
+    )
 
     df_expected = df_expected.select(*expected_selections)
 

diff --git a/examples/tpch/convert_data_to_parquet.py b/examples/tpch/convert_data_to_parquet.py
@@ -117,15 +117,14 @@
 
 curr_dir = os.path.dirname(os.path.abspath(__file__))
 for filename, curr_schema in all_schemas.items():
-
     # For convenience, go ahead and convert the schema column names to lowercase
     curr_schema = [(s[0].lower(), s[1]) for s in curr_schema]
 
     # Pre-collect the output columns so we can ignore the null field we add
     # in to handle the trailing | in the file
     output_cols = [r[0] for r in curr_schema]
 
-    curr_schema = [ pyarrow.field(r[0], r[1], nullable=False) for r in curr_schema]
+    curr_schema = [pyarrow.field(r[0], r[1], nullable=False) for r in curr_schema]
 
     # Trailing | requires extra field for in processing
     curr_schema.append(("some_null", pyarrow.null()))

diff --git a/examples/tpch/q08_market_share.py b/examples/tpch/q08_market_share.py
@@ -47,7 +47,9 @@
 
 ctx = SessionContext()
 
-df_part = ctx.read_parquet(get_data_path("part.parquet")).select_columns("p_partkey", "p_type")
+df_part = ctx.read_parquet(get_data_path("part.parquet")).select_columns(
+    "p_partkey", "p_type"
+)
 df_supplier = ctx.read_parquet(get_data_path("supplier.parquet")).select_columns(
     "s_suppkey", "s_nationkey"
 )

diff --git a/examples/tpch/q09_product_type_profit_measure.py b/examples/tpch/q09_product_type_profit_measure.py
@@ -39,7 +39,9 @@
 
 ctx = SessionContext()
 
-df_part = ctx.read_parquet(get_data_path("part.parquet")).select_columns("p_partkey", "p_name")
+df_part = ctx.read_parquet(get_data_path("part.parquet")).select_columns(
+    "p_partkey", "p_name"
+)
 df_supplier = ctx.read_parquet(get_data_path("supplier.parquet")).select_columns(
     "s_suppkey", "s_nationkey"
 )

diff --git a/examples/tpch/q13_customer_distribution.py b/examples/tpch/q13_customer_distribution.py
@@ -41,7 +41,9 @@
 df_orders = ctx.read_parquet(get_data_path("orders.parquet")).select_columns(
     "o_custkey", "o_comment"
 )
-df_customer = ctx.read_parquet(get_data_path("customer.parquet")).select_columns("c_custkey")
+df_customer = ctx.read_parquet(get_data_path("customer.parquet")).select_columns(
+    "c_custkey"
+)
 
 # Use a regex to remove special cases
 df_orders = df_orders.filter(

diff --git a/examples/tpch/q14_promotion_effect.py b/examples/tpch/q14_promotion_effect.py
@@ -44,7 +44,9 @@
 df_lineitem = ctx.read_parquet(get_data_path("lineitem.parquet")).select_columns(
     "l_partkey", "l_shipdate", "l_extendedprice", "l_discount"
 )
-df_part = ctx.read_parquet(get_data_path("part.parquet")).select_columns("p_partkey", "p_type")
+df_part = ctx.read_parquet(get_data_path("part.parquet")).select_columns(
+    "p_partkey", "p_type"
+)
 
 
 # Check part type begins with PROMO

diff --git a/examples/tpch/q16_part_supplier_relationship.py b/examples/tpch/q16_part_supplier_relationship.py
@@ -62,7 +62,8 @@
 # Select the parts we are interested in
 df_part = df_part.filter(col("p_brand") != lit(BRAND))
 df_part = df_part.filter(
-    F.substring(col("p_type"), lit(0), lit(len(TYPE_TO_IGNORE) + 1)) != lit(TYPE_TO_IGNORE)
+    F.substring(col("p_type"), 
D96B
lit(0), lit(len(TYPE_TO_IGNORE) + 1))
+    != lit(TYPE_TO_IGNORE)
 )
 
 # Python conversion of integer to literal casts it to int64 but the data for

diff --git a/examples/tpch/q17_small_quantity_order.py b/examples/tpch/q17_small_quantity_order.py
@@ -56,7 +56,13 @@
 # Find the average quantity
 window_frame = WindowFrame("rows", None, None)
 df = df.with_column(
-    "avg_quantity", F.window("avg", [col("l_quantity")], window_frame=window_frame, partition_by=[col("l_partkey")])
+    "avg_quantity",
+    F.window(
+        "avg",
+        [col("l_quantity")],
+        window_frame=window_frame,
+        partition_by=[col("l_partkey")],
+    ),
 )
 
 df = df.filter(col("l_quantity") < lit(0.2) * col("avg_quantity"))

diff --git a/examples/tpch/q20_potential_part_promotion.py b/examples/tpch/q20_potential_part_promotion.py
@@ -40,7 +40,9 @@
 
 ctx = SessionContext()
 
-df_part = ctx.read_parquet(get_data_path("part.parquet")).select_columns("p_partkey", "p_name")
+df_part = ctx.read_parquet(get_data_path("part.parquet")).select_columns(
+    "p_partkey", "p_name"
+)
 df_lineitem = ctx.read_parquet(get_data_path("lineitem.parquet")).select_columns(
     "l_shipdate", "l_partkey", "l_suppkey", "l_quantity"
 )

diff --git a/examples/tpch/q22_global_sales_opportunity.py b/examples/tpch/q22_global_sales_opportunity.py
@@ -38,7 +38,9 @@
 df_customer = ctx.read_parquet(get_data_path("customer.parquet")).select_columns(
     "c_phone", "c_acctbal", "c_custkey"
 )
-df_orders = ctx.read_parquet(get_data_path("orders.parquet")).select_columns("o_custkey")
+df_orders = ctx.read_parquet(get_data_path("orders.parquet")).select_columns(
+    "o_custkey"
+)
 
 # The nation code is a two digit number, but we need to convert it to a string literal
 nation_codes = F.make_array(*[lit(str(n)) for n in NATION_CODES])

diff --git a/examples/tpch/util.py b/examples/tpch/util.py
@@ -20,14 +20,17 @@
 """
 
 import os
-from pathlib import Path
+
 
 def get_data_path(filename: str) -> str:
     path = os.path.dirname(os.path.abspath(__file__))
 
     return os.path.join(path, "data", filename)
 
+
 def get_answer_file(answer_file: str) -> str:
     path = os.path.dirname(os.path.abspath(__file__))
 
-    return os.path.join(path, "../../benchmarks/tpch/data/answers", f"{answer_file}.out")
+    return os.path.join(
+        path, "../../benchmarks/tpch/data/answers", f"{answer_file}.out"
+    )
diff --git a/pyproject.toml b/pyproject.toml
@@ -64,3 +64,21 @@ exclude = [".github/**", "ci/**", ".asf.yaml"]
 # Require Cargo.lock is up to date
 locked = true
 features = ["substrait"]
+
+# Enable docstring linting using the google style guide
+[tool.ruff.lint]
+select = ["E4", "E7", "E9", "F", "D", "W"]
+
+[tool.ruff.lint.pydocstyle]
+convention = "google"
+
+[tool.ruff.lint.pycodestyle]
+max-doc-length = 88
+
+# Disable docstring checking for these directories
+[tool.ruff.lint.per-file-ignores]
+"python/datafusion/tests/*" = ["D"]
+"examples/*" = ["D", "W505"]
+"dev/*" = ["D"]
+"benchmarks/*" = ["D", "F"]
+"docs/*" = ["D"]
-Original file line number
+Diff line change
@@ Expand Up / @@ -51,6 +51,7 @@ requirements: @@
       run:
         - python
         - pyarrow >=11.0.0
+        - typing_extensions
     test:
       imports:
@@ Expand Down @@