liurenjie1024 commented on code in PR #1328: URL: https://github.com/apache/iceberg-rust/pull/1328#discussion_r2094987303
########## bindings/python/src/manifest.rs: ########## @@ -0,0 +1,420 @@ +// 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. + +use iceberg::spec::{DataContentType, DataFile, DataFileFormat, FieldSummary, FormatVersion, Literal, Manifest, ManifestContentType, ManifestEntry, ManifestFile, ManifestList, ManifestStatus, PrimitiveLiteral, StructType, Type}; +use pyo3::prelude::*; +use std::collections::HashMap; +use std::sync::Arc; +use pyo3::types::PyAny; +use iceberg::{Error, ErrorKind}; + +#[pyclass] +pub struct PyLiteral { + inner: Literal, +} + + +#[pyclass] +pub struct PyPrimitiveLiteral { + inner: PrimitiveLiteral, +} + + +#[pyclass] +pub struct PyDataFile { + inner: DataFile, +} + +#[pymethods] +impl PyDataFile { + + #[getter] + fn content(&self) -> i32 { + // TODO: Maps the enum back to the int + // I have a feeling this can be done more intelligently + match self.inner.content_type() { + DataContentType::Data => 0, + DataContentType::PositionDeletes => 1, + DataContentType::EqualityDeletes => 2, + } + } + + #[getter] + fn file_path(&self) -> &str { + self.inner.file_path() + } + + #[getter] + fn file_format(&self) -> &str { + // TODO: Maps the enum back to the int + // I have a feeling this can be done more intelligently + match self.inner.file_format() { + DataFileFormat::Avro => "avro", + DataFileFormat::Orc => "orc", + DataFileFormat::Parquet => "parquet", + DataFileFormat::Puffin => "puffin", + } + } + + #[getter] + fn partition(&self) -> Vec<Option<PyLiteral>> { + self.inner.partition().fields().iter().map(|p| match p { + Some(lit) => Some(PyLiteral { inner: lit.clone() }), + _ => None + } ).collect() + } + + #[getter] + fn record_count(&self) -> u64 { + self.inner.record_count() + } + + #[getter] + fn file_size_in_bytes(&self) -> u64 { + self.inner.file_size_in_bytes() + } + + #[getter] + fn column_sizes(&self) -> &HashMap<i32, u64> { + self.inner.column_sizes() + } + + #[getter] + fn value_counts(&self) -> &HashMap<i32, u64> { + self.inner.value_counts() + } + + #[getter] + fn null_value_counts(&self) -> &HashMap<i32, u64> { + self.inner.null_value_counts() + } + + #[getter] + fn nan_value_counts(&self) -> &HashMap<i32, u64> { + self.inner.nan_value_counts() + } + + #[getter] + fn upper_bounds(&self) -> HashMap<i32, Vec<u8>> { + self.inner.upper_bounds().into_iter().map(|(k, v)| (k.clone(), v.to_bytes().unwrap().to_vec())).collect() + } + + #[getter] + fn lower_bounds(&self) -> HashMap<i32, Vec<u8>> { + self.inner.lower_bounds().into_iter().map(|(k, v)| (k.clone(), v.to_bytes().unwrap().to_vec())).collect() + } + + #[getter] + fn key_metadata(&self) -> Option<&[u8]> { + self.inner.key_metadata() + } + + #[getter] + fn split_offsets(&self) -> &[i64] { + self.inner.split_offsets() + } + + #[getter] + fn equality_ids(&self) -> &[i32] { + self.inner.equality_ids() + } + + #[getter] + fn sort_order_id(&self) -> Option<i32> { + self.inner.sort_order_id() + } + +} + +#[pyclass] +pub struct PyManifest { + inner: Manifest, +} + + +#[pymethods] +impl PyManifest { + fn entries(&self) -> Vec<PyManifestEntry> { + // TODO: Most of the time, we're only interested in 'alive' entries, + // that are the ones that are either ADDED or EXISTING + // so we can add a boolean to skip the DELETED entries right away before + // moving it into the Python world + self.inner.entries().iter().map(|entry| PyManifestEntry { inner: entry.clone() }).collect() + } +} + + +#[pyclass] +pub struct PyFieldSummary { + inner: FieldSummary, +} + + +#[pymethods] +impl crate::manifest::PyFieldSummary { + + #[getter] + fn contains_null(&self) -> bool { + self.inner.contains_null + } + + #[getter] + fn contains_nan(&self) -> Option<bool> { + self.inner.contains_nan + } + + #[getter] + fn lower_bound(&self) -> Option<PyPrimitiveLiteral> { + self.inner.lower_bound.clone().map(|v| PyPrimitiveLiteral{ inner: v.literal().clone() }) + } + + #[getter] + fn upper_bound(&self) -> Option<PyPrimitiveLiteral> { + self.inner.upper_bound.clone().map(|v| PyPrimitiveLiteral{ inner: v.literal().clone() }) + } + + + +} + +#[pyclass] +pub struct PyManifestFile { + inner: ManifestFile, +} + + +#[pymethods] +impl crate::manifest::PyManifestFile { + #[getter] + fn manifest_path(&self) -> &str { + self.inner.manifest_path.as_str() + } + #[getter] + fn manifest_length(&self) -> i64 { + self.inner.manifest_length + } + #[getter] + fn partition_spec_id(&self) -> i32 { + self.inner.partition_spec_id + } + + #[getter] + fn content(&self) -> i32 { + // TODO: Maps the enum back to the int + // I have a feeling this can be done more intelligently + match self.inner.content { + ManifestContentType::Data => 0, + ManifestContentType::Deletes => 1, + } + } + + #[getter] + fn sequence_number(&self) -> i64 { + self.inner.sequence_number + } + + + #[getter] + fn min_sequence_number(&self) -> i64 { + self.inner.min_sequence_number + } + + #[getter] + fn added_snapshot_id(&self) -> i64 { + self.inner.added_snapshot_id + } + + + #[getter] + fn added_files_count(&self) -> Option<u32> { + self.inner.added_files_count + } + + #[getter] + fn existing_files_count(&self) -> Option<u32> { + self.inner.existing_files_count + } + + #[getter] + fn deleted_files_count(&self) -> Option<u32> { + self.inner.deleted_files_count + } + + #[getter] + fn added_rows_count(&self) -> Option<u64> { + self.inner.added_rows_count + } + + #[getter] + fn existing_rows_count(&self) -> Option<u64> { + self.inner.existing_rows_count + } + + #[getter] + fn deleted_rows_count(&self) -> Option<u64> { + self.inner.deleted_rows_count + } + + #[getter] + fn partitions(&self) -> Vec<PyFieldSummary> { + self.inner.partitions.iter().map(|s| PyFieldSummary { + inner: s.clone() + }).collect() + } + + #[getter] + fn key_metadata(&self) -> Vec<u8> { + self.inner.key_metadata.clone() + } + +} + +#[pyclass] +pub struct PyManifestEntry { + inner: Arc<ManifestEntry>, +} + +#[pymethods] +impl PyManifestEntry { + + #[getter] + fn status(&self) -> i32 { + // TODO: Maps the enum back to the int + // I have a feeling this can be done more intelligently + match self.inner.status { + ManifestStatus::Existing => 0, + ManifestStatus::Added => 1, + ManifestStatus::Deleted => 2, + } + } + + #[getter] + fn snapshot_id(&self) -> Option<i64> { + self.inner.snapshot_id + } + + #[getter] + fn sequence_number(&self) -> Option<i64> { + self.inner.sequence_number + } + + #[getter] + fn file_sequence_number(&self) -> Option<i64> { + self.inner.file_sequence_number + } + + #[getter] + fn data_file(&self) -> PyDataFile { + PyDataFile { + inner: self.inner.data_file.clone() + } + } +} + + +#[pyfunction] +pub fn read_manifest_entries(bs: &[u8]) -> PyManifest { + // TODO: Some error handling + PyManifest { + inner: Manifest::parse_avro(bs).unwrap() + } +} + +#[pyclass] +// #[derive(Clone)] +pub struct PartitionSpecProviderCallbackHolder { + callback: Py<PyAny>, +} + +#[pymethods] +impl PartitionSpecProviderCallbackHolder { + #[new] + fn new(callback: Py<PyAny>) -> Self { + Self { callback } + } + + fn trigger_from_python(&self, id: i32) -> PyResult<String> { + Self::do_the_callback(self, id) + } +} + +impl PartitionSpecProviderCallbackHolder { + /// Simulate calling the Python callback from "pure Rust" + pub fn do_the_callback(&self, id: i32) -> PyResult<String> { + Python::with_gil(|py| { + let result = self.callback.call1(py, (id,))?; // Call the Python function + let string = result.extract::<String>(py)?; // Try converting the result to a Rust String + Ok(string) + }) + } +} + + +#[pyclass] +pub struct PyManifestList { + inner: ManifestList, +} + + +#[pymethods] +impl crate::manifest::PyManifestList { + fn entries(&self) -> Vec<PyManifestFile> { + self.inner.entries().iter().map(|file| PyManifestFile { inner: file.clone() }).collect() + } +} + + +#[pyfunction] +pub fn read_manifest_list(bs: &[u8], cb: &PartitionSpecProviderCallbackHolder) -> PyManifestList { Review Comment: I think it has some extra runtime cost, also it's not general purpose approach. -- 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: issues-unsubscr...@iceberg.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@iceberg.apache.org For additional commands, e-mail: issues-h...@iceberg.apache.org
