Fokko commented on code in PR #7831: URL: https://github.com/apache/iceberg/pull/7831#discussion_r1259665559
########## python/pyiceberg/utils/file_stats.py: ########## @@ -0,0 +1,333 @@ +# 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. + +import struct +from typing import ( + Any, + Dict, + List, + Union, +) + +import pyarrow.lib +import pyarrow.parquet as pq + +from pyiceberg.manifest import DataFile, FileFormat +from pyiceberg.schema import Schema, SchemaVisitor, visit +from pyiceberg.types import ( + IcebergType, + ListType, + MapType, + NestedField, + PrimitiveType, + StructType, +) + +BOUND_TRUNCATED_LENGHT = 16 + +# Serialization rules: https://iceberg.apache.org/spec/#binary-single-value-serialization +# +# Type Binary serialization +# boolean 0x00 for false, non-zero byte for true +# int Stored as 4-byte little-endian +# long Stored as 8-byte little-endian +# float Stored as 4-byte little-endian +# double Stored as 8-byte little-endian +# date Stores days from the 1970-01-01 in an 4-byte little-endian int +# time Stores microseconds from midnight in an 8-byte little-endian long +# timestamp without zone Stores microseconds from 1970-01-01 00:00:00.000000 in an 8-byte little-endian long +# timestamp with zone Stores microseconds from 1970-01-01 00:00:00.000000 UTC in an 8-byte little-endian long +# string UTF-8 bytes (without length) +# uuid 16-byte big-endian value, see example in Appendix B +# fixed(L) Binary value +# binary Binary value (without length) +# + + +def bool_to_avro(value: bool) -> bytes: + return struct.pack("?", value) + + +def int32_to_avro(value: int) -> bytes: + return struct.pack("<i", value) + + +def int64_to_avro(value: int) -> bytes: + return struct.pack("<q", value) + + +def float_to_avro(value: float) -> bytes: + return struct.pack("<f", value) + + +def double_to_avro(value: float) -> bytes: + return struct.pack("<d", value) + + +def bytes_to_avro(value: Union[bytes, str]) -> bytes: + if type(value) == str: + return value.encode() + else: + assert isinstance(value, bytes) # appeases mypy + return value + + +class StatsAggregator: + def __init__(self, type_string: str): + self.current_min: Any = None + self.current_max: Any = None + self.serialize: Any = None + + if type_string == "BOOLEAN": + self.serialize = bool_to_avro + elif type_string == "INT32": + self.serialize = int32_to_avro + elif type_string == "INT64": + self.serialize = int64_to_avro + elif type_string == "INT96": + raise NotImplementedError("Statistics not implemented for INT96 physical type") + elif type_string == "FLOAT": + self.serialize = float_to_avro + elif type_string == "DOUBLE": + self.serialize = double_to_avro + elif type_string == "BYTE_ARRAY": + self.serialize = bytes_to_avro + elif type_string == "FIXED_LEN_BYTE_ARRAY": + self.serialize = bytes_to_avro + else: + raise AssertionError(f"Unknown physical type {type_string}") + + def add_min(self, val: bytes) -> None: + if not self.current_min: + self.current_min = val + elif val < self.current_min: + self.current_min = val + + def add_max(self, val: bytes) -> None: + if not self.current_max: + self.current_max = val + elif self.current_max < val: + self.current_max = val + + def get_min(self) -> bytes: + return self.serialize(self.current_min)[:BOUND_TRUNCATED_LENGHT] + + def get_max(self) -> bytes: + return self.serialize(self.current_max)[:BOUND_TRUNCATED_LENGHT] + + +def fill_parquet_file_metadata( + df: DataFile, metadata: pq.FileMetaData, col_path_2_iceberg_id: Dict[str, int], file_size: int +) -> None: + """ + Computes and fills the following fields of the DataFile object. + + - file_format + - record_count + - file_size_in_bytes + - column_sizes + - value_counts + - null_value_counts + - nan_value_counts + - lower_bounds + - upper_bounds + - split_offsets + + Args: + df (DataFile): A DataFile object representing the Parquet file for which metadata is to be filled. + metadata (pyarrow.parquet.FileMetaData): A pyarrow metadata object. + col_path_2_iceberg_id: A mapping of column paths as in the `path_in_schema` attribute of the colum Review Comment: Since Arrow is column-oriented, I'm sure that it will follow the order of the write schema:  The reason I try to avoid using too many internal details from PyArrow is that we support PyArrow `[9.0.0, 12.0.1]` currently. There is no guarantee that all these internals stay the same, therefore I think we should do as much as possible within our own control (also regarding the internal Parquet types, or how the column names are structured). We already have the write schema, so we can easily filter out the primitive types: ```python class PyArrowStatisticsCollector(PreOrderSchemaVisitor[List[StatisticsCollector]]): _field_id = 0 _schema: Schema _properties: Properties def __init__(self, schema: Schema, properties: Dict[str, str]): self._schema = schema self._properties = properties def schema(self, schema: Schema, struct_result: Callable[[], List[StatisticsCollector]]) -> List[StatisticsCollector]: return struct_result() def struct(self, struct: StructType, field_results: List[Callable[[], List[StatisticsCollector]]]) -> List[StatisticsCollector]: return list(chain(*[result() for result in field_results])) def field(self, field: NestedField, field_result: Callable[[], List[StatisticsCollector]]) -> List[StatisticsCollector]: self._field_id = field.field_id result = field_result() return result def list(self, list_type: ListType, element_result: Callable[[], List[StatisticsCollector]]) -> List[StatisticsCollector]: self._field_id = list_type.element_id return element_result() def map(self, map_type: MapType, key_result: Callable[[], List[StatisticsCollector]], value_result: Callable[[], List[StatisticsCollector]]) -> List[StatisticsCollector]: self._field_id = map_type.key_id k = key_result() self._field_id = map_type.value_id v = value_result() return k + v def primitive(self, primitive: PrimitiveType) -> List[StatisticsCollector]: return [StatisticsCollector( field_id=self._field_id, iceberg_type=primitive, mode=MetricsMode.TRUNC # schema # self._properties.get(f"write.metadata.metrics.column.{schema.find_column_name(self._field_id)}") )] ``` This way we get a nice list of columns that we need to collect statistics for. We have: ```python @dataclass(frozen=True) class StatisticsCollector: field_id: int iceberg_type: PrimitiveType mode: MetricsMode ``` Where we can use the `field_id` to properly populate the maps, and the `iceberg_type` to feed into `to_bytes` to do the conversion (so we don't have to have yet another one for PyArrow). I did a quick test, and it seems to work: ```python def test_complex_schema(table_schema_nested: Schema): tbl = pa.Table.from_pydict({ "foo": ["a", "b"], "bar": [1, 2], "baz": [False, True], "qux": [["a", "b"], ["c", "d"]], "quux": [[("a", (("aa", 1), ("ab", 2)))], [("b", (("ba", 3), ("bb", 4)))]], "location": [[(52.377956, 4.897070), (4.897070, -122.431297)], [(43.618881, -116.215019), (41.881832, -87.623177)]], "person": [("Fokko", 33), ("Max", 42)] # Possible data quality issue }, schema=schema_to_pyarrow(table_schema_nested) ) stats_columns = pre_order_visit(table_schema_nested, PyArrowStatisticsCollector(table_schema_nested, {})) visited_paths = [] def file_visitor(written_file: Any) -> None: visited_paths.append(written_file) with TemporaryDirectory() as tmpdir: pq.write_to_dataset(tbl, tmpdir, file_visitor=file_visitor) assert visited_paths[0].metadata.num_columns == len(stats_columns) ``` -- 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]
