acryl-datahub 1.2.0.6__py3-none-any.whl → 1.2.0.7rc1__py3-none-any.whl

datahub/ingestion/source/excel/profiling.py

@@ -0,0 +1,308 @@
+import logging
+import time
+from dataclasses import dataclass, field
+from typing import Any, Dict, Iterable, List, Optional, Union
+
+import numpy as np
+import pandas as pd
+
+from datahub.emitter.mcp import MetadataChangeProposalWrapper
+from datahub.ingestion.api.workunit import MetadataWorkUnit
+from datahub.ingestion.source.excel.config import ExcelSourceConfig
+from datahub.ingestion.source.excel.report import ExcelSourceReport
+from datahub.ingestion.source.excel.util import gen_dataset_name
+from datahub.metadata.schema_classes import (
+    DatasetFieldProfileClass,
+    DatasetProfileClass,
+    QuantileClass,
+)
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass
+class ColumnMetric:
+    col_type: Union[str, None] = None
+    values: List[Any] = field(default_factory=list)
+    null_count: int = 0
+    total_count: int = 0
+    distinct_count: Optional[int] = None
+    min: Optional[Any] = None
+    max: Optional[Any] = None
+    mean: Optional[float] = None
+    stdev: Optional[float] = None
+    median: Optional[float] = None
+    quantiles: Optional[List[float]] = None
+    sample_values: Optional[Any] = None
+
+
+@dataclass
+class ProfileData:
+    row_count: Optional[int] = 0
+    column_count: Optional[int] = 0
+    column_metrics: Dict[str, ColumnMetric] = field(default_factory=dict)
+
+
+class ExcelProfiler:
+    config: ExcelSourceConfig
+    report: ExcelSourceReport
+    df: pd.DataFrame
+    filename: str
+    sheet_name: str
+    dataset_urn: str
+    path: str
+
+    def __init__(
+        self,
+        config: ExcelSourceConfig,
+        report: ExcelSourceReport,
+        df: pd.DataFrame,
+        filename: str,
+        sheet_name: str,
+        dataset_urn: str,
+        path: str,
+    ) -> None:
+        self.config = config
+        self.report = report
+        self.df = df
+        self.filename = filename
+        self.sheet_name = sheet_name
+        self.dataset_urn = dataset_urn
+        self.path = path
+        self.sheet_path = f"[{self.filename}]{self.sheet_name}"
+
+        if self.config.profiling.use_sampling:
+            self.sample_size = self.config.profiling.sample_size
+        else:
+            self.sample_size = 0
+
+        self.field_sample_count = self.config.profiling.field_sample_values_limit
+
+        if self.config.profiling.max_number_of_fields_to_profile:
+            self.sample_fields = self.config.profiling.max_number_of_fields_to_profile
+        else:
+            self.sample_fields = 0
+
+    def get_workunits(self) -> Iterable[MetadataWorkUnit]:
+        logger.info(f"Profiling worksheet {self.sheet_path}")
+
+        try:
+            yield from self.generate_profile()
+        except Exception as exc:
+            self.report.profiling_skipped_other[self.filename] += 1
+            self.report.failure(
+                message="Failed to profile Excel worksheet",
+                context=f"Worksheet={self.sheet_path}",
+                exc=exc,
+            )
+
+    def generate_profile(self) -> Iterable[MetadataWorkUnit]:
+        if (
+            not self.config.profile_pattern.allowed(
+                gen_dataset_name(
+                    self.path,
+                    self.sheet_name,
+                    self.config.convert_urns_to_lowercase,
+                )
+            )
+            and self.config.profiling.report_dropped_profiles
+            and self.config.profiling.report_dropped_profiles
+        ):
+            self.report.profiling_skipped_table_profile_pattern[self.filename] += 1
+            logger.info(f"Profiling not allowed for worksheet {self.sheet_path}")
+            return
+
+        try:
+            profile_data = self.profile_workbook()
+        except Exception as exc:
+            self.report.warning(
+                message="Failed to profile Excel worksheet",
+                context=f"Worksheet={self.sheet_path}",
+                exc=exc,
+            )
+            return
+
+        profile_aspect = self.populate_profile_aspect(profile_data)
+
+        if profile_aspect:
+            self.report.report_entity_profiled()
+            mcp = MetadataChangeProposalWrapper(
+                entityUrn=self.dataset_urn, aspect=profile_aspect
+            )
+            yield mcp.as_workunit()
+
+    def populate_profile_aspect(self, profile_data: ProfileData) -> DatasetProfileClass:
+        field_profiles = [
+            self._create_field_profile(column_name, column_metrics)
+            for column_name, column_metrics in profile_data.column_metrics.items()
+        ]
+        return DatasetProfileClass(
+            timestampMillis=round(time.time() * 1000),
+            rowCount=profile_data.row_count,
+            columnCount=profile_data.column_count,
+            fieldProfiles=field_profiles,
+        )
+
+    @staticmethod
+    def _create_field_profile(
+        field_name: str, field_stats: ColumnMetric
+    ) -> DatasetFieldProfileClass:
+        quantiles = field_stats.quantiles
+        return DatasetFieldProfileClass(
+            fieldPath=field_name,
+            uniqueCount=field_stats.distinct_count,
+            nullCount=field_stats.null_count,
+            min=str(field_stats.min) if field_stats.min else None,
+            max=str(field_stats.max) if field_stats.max else None,
+            mean=str(field_stats.mean) if field_stats.mean else None,
+            median=str(field_stats.median) if field_stats.median else None,
+            stdev=str(field_stats.stdev) if field_stats.stdev else None,
+            quantiles=[
+                QuantileClass(quantile=str(0.25), value=str(quantiles[0])),
+                QuantileClass(quantile=str(0.75), value=str(quantiles[1])),
+            ]
+            if quantiles
+            else None,
+            sampleValues=field_stats.sample_values
+            if field_stats.sample_values
+            else None,
+        )
+
+    def profile_workbook(self) -> ProfileData:
+        profile_data = ProfileData()
+
+        if not self.config.profiling.profile_table_level_only:
+            return self.collect_column_data(profile_data)
+        else:
+            return self.collect_dataset_data(profile_data)
+
+    def collect_dataset_data(self, profile_data: ProfileData) -> ProfileData:
+        profile_data.row_count = self.df.shape[0]
+        profile_data.column_count = self.df.shape[1]
+
+        return profile_data
+
+    def collect_column_data(self, profile_data: ProfileData) -> ProfileData:
+        dropped_fields = set()
+        dataset_name = gen_dataset_name(
+            self.path, self.sheet_name, self.config.convert_urns_to_lowercase
+        )
+
+        logger.info(f"Attempting to profile dataset {dataset_name}")
+
+        # Get data types for each column
+        data_types = self.df.dtypes.to_dict()
+
+        # Convert numpy types to string representation for better readability
+        data_types = {col: str(dtype) for col, dtype in data_types.items()}
+
+        for n, (f_name, f_type) in enumerate(data_types.items()):
+            if 0 < self.sample_fields <= n:
+                dropped_fields.add(f_name)
+                continue
+            values = self.df[f_name].tolist()
+            profile_data.column_metrics[f_name] = ColumnMetric()
+            profile_data.column_metrics[f_name].values.extend(values)
+            profile_data.column_metrics[f_name].col_type = f_type
+
+        if len(dropped_fields) > 0:
+            if self.config.profiling.report_dropped_profiles:
+                self.report.report_dropped(
+                    f"The max_number_of_fields_to_profile={self.sample_fields} reached. "
+                    f"Dropped fields for {dataset_name} ({', '.join(sorted(dropped_fields))})"
+                )
+
+        profile_data.row_count = self.df.shape[0]
+        profile_data.column_count = self.df.shape[1]
+
+        return self.add_field_statistics(profile_data)
+
+    def add_field_statistics(self, profile_data: ProfileData) -> ProfileData:
+        for field_name, column_metrics in profile_data.column_metrics.items():
+            if column_metrics.values:
+                try:
+                    self.compute_field_statistics(column_metrics)
+                except Exception as exc:
+                    self.report.warning(
+                        message="Profiling Failed For Column Statistics",
+                        context=field_name,
+                        exc=exc,
+                    )
+                    raise exc
+
+        return profile_data
+
+    def compute_field_statistics(self, column_metrics: ColumnMetric) -> None:
+        values = column_metrics.values
+        if not values:
+            return
+
+        logger.debug(
+            f"Computing statistics for column of type {column_metrics.col_type}"
+        )
+
+        column_metrics.total_count = len(values)
+
+        # ByDefault Null count is added
+        if not self.config.profiling.include_field_null_count:
+            column_metrics.null_count = 0
+
+        if self.config.profiling.include_field_distinct_count:
+            column_metrics.distinct_count = len(set(values))
+
+        if values and self._is_numeric_type(column_metrics.col_type):
+            if self.config.profiling.include_field_min_value:
+                column_metrics.min = min(values)
+            if self.config.profiling.include_field_max_value:
+                column_metrics.max = max(values)
+            if self.config.profiling.include_field_mean_value:
+                column_metrics.mean = round(float(np.mean(values)), 2)
+            if self.config.profiling.include_field_stddev_value:
+                column_metrics.stdev = round(float(np.std(values)), 2)
+            if self.config.profiling.include_field_median_value:
+                column_metrics.median = round(float(np.median(values)), 2)
+            if self.config.profiling.include_field_quantiles:
+                column_metrics.quantiles = [
+                    float(np.percentile(values, 25)),
+                    float(np.percentile(values, 75)),
+                ]
+
+        if values and self.config.profiling.include_field_sample_values:
+            column_metrics.sample_values = [
+                str(v) for v in values[: self.field_sample_count]
+            ]
+
+    @staticmethod
+    def _is_numeric_type(data_type: Union[str, None]) -> bool:
+        if not data_type:
+            return False
+        else:
+            return data_type.lower() in [
+                "int8",
+                "int16",
+                "int32",
+                "int64",
+                "uint8",
+                "uint16",
+                "uint32",
+                "uint64",
+                "intp",
+                "uintp",
+                "int8",
+                "int16",
+                "int32",
+                "int64",
+                "uint8",
+                "uint16",
+                "uint32",
+                "uint64",
+                "float16",
+                "float32",
+                "float64",
+                "float128",
+                "float32",
+                "float64",
+                "complex64",
+                "complex128",
+                "complex256",
+            ]

datahub/ingestion/source/excel/report.py

@@ -0,0 +1,49 @@
+from dataclasses import dataclass, field
+from typing import List
+
+from datahub.ingestion.source.state.stale_entity_removal_handler import (
+    StaleEntityRemovalSourceReport,
+)
+from datahub.utilities.perf_timer import PerfTimer
+from datahub.utilities.stats_collections import TopKDict, int_top_k_dict
+
+
+@dataclass
+class ExcelSourceReport(StaleEntityRemovalSourceReport):
+    files_scanned = 0
+    files_processed = 0
+    worksheets_scanned = 0
+    worksheets_processed = 0
+    datasets_profiled = 0
+    local_file_get_timer: PerfTimer = field(default_factory=PerfTimer)
+    s3_file_get_timer: PerfTimer = field(default_factory=PerfTimer)
+    abs_file_get_timer: PerfTimer = field(default_factory=PerfTimer)
+    filtered: List[str] = field(default_factory=list)
+    profiling_skipped_other: TopKDict[str, int] = field(default_factory=int_top_k_dict)
+    profiling_skipped_table_profile_pattern: TopKDict[str, int] = field(
+        default_factory=int_top_k_dict
+    )
+
+    def report_dropped(self, name: str) -> None:
+        self.filtered.append(name)
+
+    def report_entity_profiled(self) -> None:
+        self.datasets_profiled += 1
+
+    def report_file_scanned(self) -> None:
+        self.files_scanned += 1
+
+    def report_file_processed(self) -> None:
+        self.files_processed += 1
+
+    def report_worksheet_scanned(self) -> None:
+        self.worksheets_scanned += 1
+
+    def report_worksheet_processed(self) -> None:
+        self.worksheets_processed += 1
+
+    def report_file_dropped(self, file: str) -> None:
+        self.filtered.append(file)
+
+    def report_worksheet_dropped(self, worksheet: str) -> None:
+        self.filtered.append(worksheet)