Yicong-Huang commented on code in PR #54084: URL: https://github.com/apache/spark/pull/54084#discussion_r2760656779
########## python/pyspark/testing/goldenutils.py: ########## @@ -0,0 +1,497 @@ +# +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# + +from typing import Any, Callable, Iterable, Optional, TYPE_CHECKING +import concurrent.futures +import inspect +import os +import re +import time + +from pyspark.testing.utils import have_pandas, have_numpy + +if have_pandas: + import pandas as pd +if have_numpy: + import numpy as np + +if TYPE_CHECKING: + from pyspark.sql.types import DataType + + +class GoldenFileTestMixin: + """ + Mixin class providing utilities for golden file based testing. + + Golden files are CSV files that store expected test results. This mixin provides: + - Timezone setup/teardown for deterministic results + - Golden file read/write with SPARK_GENERATE_GOLDEN_FILES env var support + - Result string cleaning utilities + + To regenerate golden files, set SPARK_GENERATE_GOLDEN_FILES=1 before running tests. + """ + + _tz_prev: Optional[str] = None + + @classmethod + def setUpClass(cls) -> None: + """Setup test class with timezone configuration.""" + super().setUpClass() + cls.setup_timezone() + + @classmethod + def tearDownClass(cls) -> None: + """Teardown test class and restore timezone.""" + cls.teardown_timezone() + super().tearDownClass() + + @classmethod + def setup_timezone(cls, tz: str = "America/Los_Angeles") -> None: + """ + Setup timezone for deterministic test results. + Synchronizes timezone between Python and Java. + """ + cls._tz_prev = os.environ.get("TZ", None) + os.environ["TZ"] = tz + time.tzset() + + cls.sc.environment["TZ"] = tz + cls.spark.conf.set("spark.sql.session.timeZone", tz) + + @classmethod + def teardown_timezone(cls) -> None: + """Restore original timezone.""" + if "TZ" in os.environ: + del os.environ["TZ"] + if cls._tz_prev is not None: + os.environ["TZ"] = cls._tz_prev + time.tzset() + + @staticmethod + def is_generating_golden() -> bool: + """Check if we are generating golden files (vs testing against them).""" + return os.environ.get("SPARK_GENERATE_GOLDEN_FILES", "0") == "1" + + @staticmethod + def load_golden_csv(golden_csv: str, use_index: bool = True) -> "pd.DataFrame": + """ + Load golden file from CSV. + + Parameters + ---------- + golden_csv : str + Path to the golden CSV file. + use_index : bool + If True, use first column as index (for matrix format). + If False, don't use index (for row list format). + + Returns + ------- + pd.DataFrame + The loaded golden data with string dtype. + """ + return pd.read_csv( + golden_csv, + sep="\t", + index_col=0 if use_index else None, + dtype="str", + na_filter=False, + engine="python", + ) + + @staticmethod + def save_golden(df: "pd.DataFrame", golden_csv: str, golden_md: Optional[str] = None) -> None: + """ + Save DataFrame as golden file (CSV and optionally Markdown). + + Parameters + ---------- + df : pd.DataFrame + The DataFrame to save. + golden_csv : str + Path to save the CSV file. + golden_md : str, optional + Path to save the Markdown file. Requires tabulate package. + """ + df.to_csv(golden_csv, sep="\t", header=True, index=True) + + if golden_md is not None: + try: + df.to_markdown(golden_md, index=True, tablefmt="github") + except Exception as e: + import warnings + + warnings.warn( + f"Failed to write markdown file {golden_md}: {e}. " + "Install 'tabulate' package to generate markdown files." + ) + + @staticmethod + def repr_spark_type(spark_type: "DataType") -> str: + """Convert Spark type to string representation.""" + return spark_type.simpleString() + + @classmethod + def repr_value( + cls, + value: Any, + max_len: int = 32, + type_override: Optional[str] = None, + ) -> str: + """ + Convert Python value to string representation for golden file. + + Format: "value_str@type_info" + - For numpy.ndarray: includes dtype, e.g., "[1 2]@ndarray[int64]" + - For pandas.DataFrame: includes schema, e.g., "{...}@DataFrame[_1 int64]" + - For list: includes element types, e.g., "[1, 2]@List[int]" + - For other types: uses type name, e.g., "True@bool" + + Java object hash codes are normalized (e.g., @69420149 -> @<hash>) + for deterministic test results. + + Parameters + ---------- + value : Any + The Python value to represent. + max_len : int, default 32 + Maximum length for the value string portion. + type_override : str, optional + If provided, use this as the type string instead of auto-detecting. + + Returns + ------- + str + String representation in format "value@type". + """ + # Get value string representation + if have_pandas and isinstance(value, pd.DataFrame): + v_str = value.to_json() + elif have_pandas and isinstance(value, pd.Series): + v_str = str(value.tolist()) + else: + v_str = str(value) + + # Get type string + type_str = type_override if type_override is not None else cls.repr_type(value) + + # Clean up: replace newlines, normalize Java hash codes, then truncate + v_str = v_str.replace("\n", " ") + v_str = re.sub(r"@[a-fA-F0-9]+", "@<hash>", v_str) + v_str = v_str[:max_len] + return f"{v_str}@{type_str}" + + @classmethod + def repr_type(cls, value: Any) -> str: + """ + Get the type representation string for a value (recursively for containers). + + Parameters + ---------- + value : Any + The value to get type representation for. + + Returns + ------- + str + Type string, e.g., "int", "list[int | NoneType]", "DataFrame[col1 int64]". + """ + return cls._repr_element_type(value) + + @classmethod + def _repr_element_type(cls, elem: Any) -> str: + """ + Recursively get the type representation for an element. + + For containers (list, dict, tuple, DataFrame), inspects nested element types. + """ + if elem is None: + return "NoneType" + elif have_pandas and isinstance(elem, pd.DataFrame): + schema = ", ".join([f"{col} {dtype.name}" for col, dtype in elem.dtypes.items()]) + return f"DataFrame[{schema}]" + elif have_pandas and isinstance(elem, pd.Series): + return f"Series[{elem.dtype.name}]" + elif have_numpy and isinstance(elem, np.ndarray): + return f"ndarray[{elem.dtype.name}]" + elif isinstance(elem, list): + if len(elem) == 0: + return "list" + inner = cls._repr_container(elem, container=None) + return f"list[{inner}]" + elif isinstance(elem, dict): + if len(elem) == 0: + return "dict" + key_str = cls._repr_container(list(elem.keys()), container=None) + val_str = cls._repr_container(list(elem.values()), container=None) + return f"dict[{key_str}, {val_str}]" + elif isinstance(elem, tuple): + if len(elem) == 0: + return "tuple" + inner = cls._repr_container(list(elem), container=None) + return f"tuple[{inner}]" + else: + return type(elem).__name__ + + @classmethod + def repr_pandas_type(cls, data: "pd.Series") -> str: + """ + Get the type representation for a pandas Series, with element type inspection. + + For object dtype, inspects actual element types recursively. + + Parameters + ---------- + data : pd.Series + The pandas Series to get type representation for. + + Returns + ------- + str + Type string, e.g., "int64", "list[int | NoneType]", "dict[str, int]". + """ + if not hasattr(data, "dtype"): + return type(data).__name__ + + dtype_str = str(data.dtype) + # For object dtype, inspect actual element types recursively + if dtype_str == "object" and len(data) > 0: + return cls._repr_container(list(data), container=None) + return dtype_str + + @staticmethod + def _join_type_strings(type_strs: list, container: str = None) -> str: + """ + Join a list of type strings into a formatted type string. + + Parameters + ---------- + type_strs : list + List of type name strings (e.g., ['int', 'str', 'NoneType']). + container : str, optional + Container type name (e.g., "list", "Series"). If None, returns just the type string. + + Returns + ------- + str + Formatted type string with NoneType at the end, e.g., "int | str | NoneType". + """ + unique_types = set(type_strs) + # Sort with NoneType at the end + has_none = "NoneType" in unique_types + other_types = sorted(t for t in unique_types if t != "NoneType") + if has_none: + other_types.append("NoneType") + + if len(other_types) == 0: + type_str = "" + elif len(other_types) == 1: + type_str = other_types[0] + else: + type_str = " | ".join(other_types) + + if container is None: + return type_str or "object" + elif type_str: + return f"{container}[{type_str}]" + else: + return container + + @classmethod + def _repr_container(cls, values: list, container: str = None) -> str: + """ + Format a list of values into a container type string. + + Parameters + ---------- + values : list + List of values to get type representations for. + container : str, optional + Container type name (e.g., "list", "Series"). If None, returns just the type string. + + Returns + ------- + str + Formatted type string, e.g., "list[int]" or "Series[int64 | object]". + NoneType is always placed at the end if present. + """ + type_strs = [cls._repr_element_type(v) for v in values] + return cls._join_type_strings(type_strs, container) + + @staticmethod + def clean_result(result: str) -> str: + """Clean result string by removing newlines and extra whitespace.""" + return result.replace("\n", " ").replace("\r", " ").replace("\t", " ") + + def _compare_or_generate_golden( + self, + golden_file: str, + row_keys: list[str], + column_names: list[str], + results: list[tuple[str, str, str, Optional[str]]], + index_name: str = "Source Value \\ Target Type", + ) -> None: + """ + Compare test results against golden file, or generate new golden file. + + This method provides a common framework for matrix-style golden file tests. + Each test produces a matrix where rows are source values and columns are + target types (or other categories). + + Parameters + ---------- + golden_file : str + Full base path of golden file (without extension). Will generate both + .csv and .md files. E.g., "/path/to/golden_test_name" + row_keys : list[str] + Index values (row labels) for the DataFrame. + column_names : list[str] + Column names for the DataFrame. + results : list[tuple[str, str, str, Optional[str]]] + List of (row_key, column_name, cell_value, error_or_none) tuples. + Each tuple represents one cell in the result matrix. + index_name : str + Name for the index column in the golden file. + """ + generating = self.is_generating_golden() + + golden_csv = f"{golden_file}.csv" + golden_md = f"{golden_file}.md" + + if not generating: + golden = self.load_golden_csv(golden_csv) + errs = [] + for row_key, col_name, value, err in results: + if err is not None: + errs.append(err) + else: + expected = golden.loc[row_key, col_name] + if expected != value: + errs.append(f"{row_key} => {col_name}: expects {expected} but got {value}") + self.assertTrue(len(errs) == 0, "\n" + "\n".join(errs) + "\n") + else: + index = pd.Index(row_keys, name=index_name) + new_golden = pd.DataFrame("?", index=index, columns=column_names) + + for row_key, col_name, value, _ in results: + new_golden.loc[row_key, col_name] = value + + self.save_golden(new_golden, golden_csv, golden_md) + + @property + def column_names(self) -> list[str]: + """Column names for the golden file. Override in subclass.""" + raise NotImplementedError("Subclass must define column_names property") + + def run_single_test(self, test_item: Any) -> tuple[str, list[tuple[str, str]]]: + """ + Run a single test item and return results. + + Override in subclass to implement test logic. + + Returns + ------- + tuple[str, list[tuple[str, str]]] + (row_key, [(column_name, cell_value), ...]) + """ + raise NotImplementedError("Subclass must implement run_single_test method") + + @property + def test_cases(self) -> Iterable[Any]: + """Test cases to iterate over. Override in subclass.""" + raise NotImplementedError("Subclass must define test_cases property") + + def run_tests(self, golden_name: str) -> None: + """ + Run golden file tests using class properties. + + Uses self.test_cases, self.column_names, and self.run_single_test. + """ + self._run_golden_tests( + golden_name=golden_name, + test_items=self.test_cases, + run_test=self.run_single_test, + column_names=self.column_names, + parallel=True, + ) + + def _run_golden_tests( Review Comment: yes. here the framework is pretty general. the test classes who uses this mixin can define properties `test_cases`, `column_names`, and also define the method `run_single_test` to handle how to execute or check error. The framework logic here is simple: 1. run the provided `test_cases` by calling the provided `run_single_test`. 2. execute in parallel, if permitted. 3. collect results, serialize to string 4. compare or generate golden file. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
