@loaders.gl/parquet 4.4.5 → 5.0.0-alpha.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/LICENSE +14 -3
- package/dist/bundled.d.ts +9 -0
- package/dist/bundled.d.ts.map +1 -0
- package/dist/bundled.js +10 -0
- package/dist/bundled.js.map +1 -0
- package/dist/geoparquet-loader-with-parser.d.ts +33 -0
- package/dist/geoparquet-loader-with-parser.d.ts.map +1 -0
- package/dist/geoparquet-loader-with-parser.js +52 -0
- package/dist/geoparquet-loader-with-parser.js.map +1 -0
- package/dist/geoparquet-loader.d.ts +59 -0
- package/dist/geoparquet-loader.d.ts.map +1 -0
- package/dist/geoparquet-loader.js +30 -0
- package/dist/geoparquet-loader.js.map +1 -0
- package/dist/index.cjs +25873 -4776
- package/dist/index.cjs.map +3 -4
- package/dist/index.d.ts +10 -7
- package/dist/index.d.ts.map +1 -1
- package/dist/index.js +7 -11
- package/dist/index.js.map +1 -1
- package/dist/lib/arrow/convert-columns-to-row-group.js +3 -0
- package/dist/lib/arrow/convert-columns-to-row-group.js.map +1 -1
- package/dist/lib/arrow/convert-schema-from-parquet.d.ts.map +1 -1
- package/dist/lib/arrow/convert-schema-from-parquet.js +86 -26
- package/dist/lib/arrow/convert-schema-from-parquet.js.map +1 -1
- package/dist/lib/arrow/convert-schema-to-parquet.d.ts.map +1 -1
- package/dist/lib/arrow/convert-schema-to-parquet.js +16 -7
- package/dist/lib/arrow/convert-schema-to-parquet.js.map +1 -1
- package/dist/lib/arrow/create-nested-arrow-vector.d.ts +8 -0
- package/dist/lib/arrow/create-nested-arrow-vector.d.ts.map +1 -0
- package/dist/lib/arrow/create-nested-arrow-vector.js +396 -0
- package/dist/lib/arrow/create-nested-arrow-vector.js.map +1 -0
- package/dist/lib/constants.d.ts +0 -1
- package/dist/lib/constants.d.ts.map +1 -1
- package/dist/lib/constants.js +1 -2
- package/dist/lib/constants.js.map +1 -1
- package/dist/lib/encoders/encode-arrow-to-parquet.d.ts +2 -2
- package/dist/lib/encoders/encode-arrow-to-parquet.d.ts.map +1 -1
- package/dist/lib/encoders/encode-arrow-to-parquet.js +3 -10
- package/dist/lib/encoders/encode-arrow-to-parquet.js.map +1 -1
- package/dist/lib/encoders/encode-table-to-parquet-js.d.ts +5 -0
- package/dist/lib/encoders/encode-table-to-parquet-js.d.ts.map +1 -0
- package/dist/lib/encoders/encode-table-to-parquet-js.js +199 -0
- package/dist/lib/encoders/encode-table-to-parquet-js.js.map +1 -0
- package/dist/lib/geo/geospatial-metadata.d.ts +67 -0
- package/dist/lib/geo/geospatial-metadata.d.ts.map +1 -0
- package/dist/lib/geo/geospatial-metadata.js +338 -0
- package/dist/lib/geo/geospatial-metadata.js.map +1 -0
- package/dist/lib/parquet-page-index.d.ts +62 -0
- package/dist/lib/parquet-page-index.d.ts.map +1 -0
- package/dist/lib/parquet-page-index.js +377 -0
- package/dist/lib/parquet-page-index.js.map +1 -0
- package/dist/lib/parquet-predicate.d.ts +17 -0
- package/dist/lib/parquet-predicate.d.ts.map +1 -0
- package/dist/lib/parquet-predicate.js +300 -0
- package/dist/lib/parquet-predicate.js.map +1 -0
- package/dist/lib/parquet-source-worker-client.d.ts +6 -0
- package/dist/lib/parquet-source-worker-client.d.ts.map +1 -0
- package/dist/lib/parquet-source-worker-client.js +22 -0
- package/dist/lib/parquet-source-worker-client.js.map +1 -0
- package/dist/lib/parquet-source-worker-decoder.d.ts +4 -0
- package/dist/lib/parquet-source-worker-decoder.d.ts.map +1 -0
- package/dist/lib/parquet-source-worker-decoder.js +115 -0
- package/dist/lib/parquet-source-worker-decoder.js.map +1 -0
- package/dist/lib/parquet-source-worker-file.d.ts +24 -0
- package/dist/lib/parquet-source-worker-file.d.ts.map +1 -0
- package/dist/lib/parquet-source-worker-file.js +40 -0
- package/dist/lib/parquet-source-worker-file.js.map +1 -0
- package/dist/lib/parquet-source-worker-types.d.ts +97 -0
- package/dist/lib/parquet-source-worker-types.d.ts.map +1 -0
- package/dist/lib/parquet-source-worker-types.js +5 -0
- package/dist/lib/parquet-source-worker-types.js.map +1 -0
- package/dist/lib/parquet-worker-transport.d.ts +8 -0
- package/dist/lib/parquet-worker-transport.d.ts.map +1 -0
- package/dist/lib/parquet-worker-transport.js +40 -0
- package/dist/lib/parquet-worker-transport.js.map +1 -0
- package/dist/lib/parsers/convert-parquet-tables.d.ts +16 -0
- package/dist/lib/parsers/convert-parquet-tables.d.ts.map +1 -0
- package/dist/lib/parsers/convert-parquet-tables.js +30 -0
- package/dist/lib/parsers/convert-parquet-tables.js.map +1 -0
- package/dist/lib/parsers/get-parquet-schema.d.ts.map +1 -1
- package/dist/lib/parsers/get-parquet-schema.js +3 -1
- package/dist/lib/parsers/get-parquet-schema.js.map +1 -1
- package/dist/lib/parsers/parse-geoparquet-to-geojson.d.ts +4 -4
- package/dist/lib/parsers/parse-geoparquet-to-geojson.d.ts.map +1 -1
- package/dist/lib/parsers/parse-geoparquet-to-geojson.js +19 -6
- package/dist/lib/parsers/parse-geoparquet-to-geojson.js.map +1 -1
- package/dist/lib/parsers/parse-parquet-tables.d.ts +37 -0
- package/dist/lib/parsers/parse-parquet-tables.d.ts.map +1 -0
- package/dist/lib/parsers/parse-parquet-tables.js +50 -0
- package/dist/lib/parsers/parse-parquet-tables.js.map +1 -0
- package/dist/lib/parsers/parse-parquet-to-arrow-js.d.ts +19 -0
- package/dist/lib/parsers/parse-parquet-to-arrow-js.d.ts.map +1 -0
- package/dist/lib/parsers/parse-parquet-to-arrow-js.js +608 -0
- package/dist/lib/parsers/parse-parquet-to-arrow-js.js.map +1 -0
- package/dist/lib/parsers/parse-parquet-to-arrow.d.ts +7 -7
- package/dist/lib/parsers/parse-parquet-to-arrow.d.ts.map +1 -1
- package/dist/lib/parsers/parse-parquet-to-arrow.js +45 -32
- package/dist/lib/parsers/parse-parquet-to-arrow.js.map +1 -1
- package/dist/lib/parsers/parse-parquet-to-columns.d.ts +3 -3
- package/dist/lib/parsers/parse-parquet-to-columns.d.ts.map +1 -1
- package/dist/lib/parsers/parse-parquet-to-columns.js +1 -4
- package/dist/lib/parsers/parse-parquet-to-columns.js.map +1 -1
- package/dist/lib/parsers/parse-parquet-to-json.d.ts +3 -3
- package/dist/lib/parsers/parse-parquet-to-json.d.ts.map +1 -1
- package/dist/lib/parsers/parse-parquet-to-json.js +77 -26
- package/dist/lib/parsers/parse-parquet-to-json.js.map +1 -1
- package/dist/lib/sources/parquet-range-file.d.ts +80 -0
- package/dist/lib/sources/parquet-range-file.d.ts.map +1 -0
- package/dist/lib/sources/parquet-range-file.js +261 -0
- package/dist/lib/sources/parquet-range-file.js.map +1 -0
- package/dist/lib/sources/single-batch-queue.d.ts +24 -0
- package/dist/lib/sources/single-batch-queue.d.ts.map +1 -0
- package/dist/lib/sources/single-batch-queue.js +91 -0
- package/dist/lib/sources/single-batch-queue.js.map +1 -0
- package/dist/lib/utils/load-wasm-node.d.ts +3 -0
- package/dist/lib/utils/load-wasm-node.d.ts.map +1 -0
- package/dist/lib/utils/load-wasm-node.js +23 -0
- package/dist/lib/utils/load-wasm-node.js.map +1 -0
- package/dist/lib/utils/load-wasm.d.ts +2 -2
- package/dist/lib/utils/load-wasm.d.ts.map +1 -1
- package/dist/lib/utils/load-wasm.js +28 -12
- package/dist/lib/utils/load-wasm.js.map +1 -1
- package/dist/lib/utils/make-stream-iterator.d.ts +2 -0
- package/dist/lib/utils/make-stream-iterator.d.ts.map +1 -1
- package/dist/lib/utils/make-stream-iterator.js +123 -14
- package/dist/lib/utils/make-stream-iterator.js.map +1 -1
- package/dist/lib/utils/normalize-parquet-options.d.ts +5 -0
- package/dist/lib/utils/normalize-parquet-options.d.ts.map +1 -0
- package/dist/lib/utils/normalize-parquet-options.js +11 -0
- package/dist/lib/utils/normalize-parquet-options.js.map +1 -0
- package/dist/parquet-dataset-source.cjs +26413 -0
- package/dist/parquet-dataset-source.cjs.map +6 -0
- package/dist/parquet-dataset-source.d.ts +54 -0
- package/dist/parquet-dataset-source.d.ts.map +1 -0
- package/dist/parquet-dataset-source.js +380 -0
- package/dist/parquet-dataset-source.js.map +1 -0
- package/dist/parquet-format.d.ts +2 -0
- package/dist/parquet-format.d.ts.map +1 -1
- package/dist/parquet-format.js +2 -0
- package/dist/parquet-format.js.map +1 -1
- package/dist/parquet-js-loader-types.d.ts +57 -0
- package/dist/parquet-js-loader-types.d.ts.map +1 -0
- package/dist/parquet-js-loader-types.js +33 -0
- package/dist/parquet-js-loader-types.js.map +1 -0
- package/dist/parquet-js-loader.d.ts +32 -0
- package/dist/parquet-js-loader.d.ts.map +1 -0
- package/dist/parquet-js-loader.js +58 -0
- package/dist/parquet-js-loader.js.map +1 -0
- package/dist/parquet-js-writer.d.ts +44 -0
- package/dist/parquet-js-writer.d.ts.map +1 -0
- package/dist/parquet-js-writer.js +28 -0
- package/dist/parquet-js-writer.js.map +1 -0
- package/dist/parquet-loader-base.d.ts +25 -0
- package/dist/parquet-loader-base.d.ts.map +1 -0
- package/dist/parquet-loader-base.js +23 -0
- package/dist/parquet-loader-base.js.map +1 -0
- package/dist/parquet-loader-options.d.ts +44 -0
- package/dist/parquet-loader-options.d.ts.map +1 -0
- package/dist/parquet-loader-options.js +10 -0
- package/dist/parquet-loader-options.js.map +1 -0
- package/dist/parquet-loader-types.d.ts +61 -0
- package/dist/parquet-loader-types.d.ts.map +1 -0
- package/dist/parquet-loader-types.js +17 -0
- package/dist/parquet-loader-types.js.map +1 -0
- package/dist/parquet-loader.d.ts +42 -0
- package/dist/parquet-loader.d.ts.map +1 -0
- package/dist/parquet-loader.js +68 -0
- package/dist/parquet-loader.js.map +1 -0
- package/dist/parquet-source-capabilities.d.ts +36 -0
- package/dist/parquet-source-capabilities.d.ts.map +1 -0
- package/dist/parquet-source-capabilities.js +22 -0
- package/dist/parquet-source-capabilities.js.map +1 -0
- package/dist/parquet-source-loader-types.d.ts +72 -0
- package/dist/parquet-source-loader-types.d.ts.map +1 -0
- package/dist/parquet-source-loader-types.js +74 -0
- package/dist/parquet-source-loader-types.js.map +1 -0
- package/dist/parquet-source-loader.cjs +25975 -0
- package/dist/parquet-source-loader.cjs.map +6 -0
- package/dist/parquet-source-loader.d.ts +120 -0
- package/dist/parquet-source-loader.d.ts.map +1 -0
- package/dist/parquet-source-loader.js +812 -0
- package/dist/parquet-source-loader.js.map +1 -0
- package/dist/parquet-source-types.d.ts +384 -0
- package/dist/parquet-source-types.d.ts.map +1 -0
- package/dist/parquet-source-types.js +5 -0
- package/dist/parquet-source-types.js.map +1 -0
- package/dist/parquet-source-worker-url.d.ts +3 -0
- package/dist/parquet-source-worker-url.d.ts.map +1 -0
- package/dist/parquet-source-worker-url.js +22 -0
- package/dist/parquet-source-worker-url.js.map +1 -0
- package/dist/parquet-source-worker.js +38491 -0
- package/dist/parquet-source-worker.js.map +7 -0
- package/dist/parquet-worker-node.cjs +54779 -0
- package/dist/parquet-worker-node.cjs.map +7 -0
- package/dist/parquet-worker-url.d.ts +3 -0
- package/dist/parquet-worker-url.d.ts.map +1 -0
- package/dist/parquet-worker-url.js +15 -0
- package/dist/parquet-worker-url.js.map +1 -0
- package/dist/parquet-worker.js +19920 -0
- package/dist/parquet-worker.js.map +7 -0
- package/dist/parquet-writer.d.ts +10 -0
- package/dist/parquet-writer.d.ts.map +1 -0
- package/dist/parquet-writer.js +27 -0
- package/dist/parquet-writer.js.map +1 -0
- package/dist/parquet_wasm_bg.wasm +0 -0
- package/dist/parquetjs/codecs/byte-stream-split.d.ts +9 -0
- package/dist/parquetjs/codecs/byte-stream-split.d.ts.map +1 -0
- package/dist/parquetjs/codecs/byte-stream-split.js +159 -0
- package/dist/parquetjs/codecs/byte-stream-split.js.map +1 -0
- package/dist/parquetjs/codecs/declare.d.ts +20 -3
- package/dist/parquetjs/codecs/declare.d.ts.map +1 -1
- package/dist/parquetjs/codecs/declare.js +7 -1
- package/dist/parquetjs/codecs/declare.js.map +1 -1
- package/dist/parquetjs/codecs/delta.d.ts +17 -0
- package/dist/parquetjs/codecs/delta.d.ts.map +1 -0
- package/dist/parquetjs/codecs/delta.js +439 -0
- package/dist/parquetjs/codecs/delta.js.map +1 -0
- package/dist/parquetjs/codecs/dictionary.d.ts +5 -2
- package/dist/parquetjs/codecs/dictionary.d.ts.map +1 -1
- package/dist/parquetjs/codecs/dictionary.js +16 -4
- package/dist/parquetjs/codecs/dictionary.js.map +1 -1
- package/dist/parquetjs/codecs/index.d.ts.map +1 -1
- package/dist/parquetjs/codecs/index.js +18 -0
- package/dist/parquetjs/codecs/index.js.map +1 -1
- package/dist/parquetjs/codecs/plain.d.ts +3 -3
- package/dist/parquetjs/codecs/plain.d.ts.map +1 -1
- package/dist/parquetjs/codecs/plain.js +88 -76
- package/dist/parquetjs/codecs/plain.js.map +1 -1
- package/dist/parquetjs/codecs/rle.d.ts +3 -3
- package/dist/parquetjs/codecs/rle.d.ts.map +1 -1
- package/dist/parquetjs/codecs/rle.js +196 -68
- package/dist/parquetjs/codecs/rle.js.map +1 -1
- package/dist/parquetjs/compression.d.ts +7 -8
- package/dist/parquetjs/compression.d.ts.map +1 -1
- package/dist/parquetjs/compression.js +95 -141
- package/dist/parquetjs/compression.js.map +1 -1
- package/dist/parquetjs/encoder/dictionary-planner.d.ts +17 -0
- package/dist/parquetjs/encoder/dictionary-planner.d.ts.map +1 -0
- package/dist/parquetjs/encoder/dictionary-planner.js +64 -0
- package/dist/parquetjs/encoder/dictionary-planner.js.map +1 -0
- package/dist/parquetjs/encoder/page-planner.d.ts +11 -0
- package/dist/parquetjs/encoder/page-planner.d.ts.map +1 -0
- package/dist/parquetjs/encoder/page-planner.js +60 -0
- package/dist/parquetjs/encoder/page-planner.js.map +1 -0
- package/dist/parquetjs/encoder/parquet-encoder.d.ts +11 -4
- package/dist/parquetjs/encoder/parquet-encoder.d.ts.map +1 -1
- package/dist/parquetjs/encoder/parquet-encoder.js +259 -52
- package/dist/parquetjs/encoder/parquet-encoder.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/BsonType.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/BsonType.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/BsonType.js +1 -1
- package/dist/parquetjs/parquet-thrift/BsonType.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/ColumnChunk.d.ts +2 -2
- package/dist/parquetjs/parquet-thrift/ColumnChunk.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/ColumnChunk.js +2 -2
- package/dist/parquetjs/parquet-thrift/ColumnChunk.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/ColumnIndex.d.ts +6 -6
- package/dist/parquetjs/parquet-thrift/ColumnIndex.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/ColumnIndex.js +3 -3
- package/dist/parquetjs/parquet-thrift/ColumnIndex.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/ColumnMetaData.d.ts +2 -2
- package/dist/parquetjs/parquet-thrift/ColumnMetaData.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/ColumnMetaData.js +2 -2
- package/dist/parquetjs/parquet-thrift/ColumnMetaData.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/ColumnOrder.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/ColumnOrder.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/ColumnOrder.js +1 -1
- package/dist/parquetjs/parquet-thrift/ColumnOrder.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/DataPageHeader.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/DataPageHeader.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/DataPageHeader.js +1 -1
- package/dist/parquetjs/parquet-thrift/DataPageHeader.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/DataPageHeaderV2.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/DataPageHeaderV2.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/DataPageHeaderV2.js +1 -1
- package/dist/parquetjs/parquet-thrift/DataPageHeaderV2.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/DateType.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/DateType.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/DateType.js +1 -1
- package/dist/parquetjs/parquet-thrift/DateType.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/DecimalType.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/DecimalType.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/DecimalType.js +1 -1
- package/dist/parquetjs/parquet-thrift/DecimalType.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/DictionaryPageHeader.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/DictionaryPageHeader.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/DictionaryPageHeader.js +1 -1
- package/dist/parquetjs/parquet-thrift/DictionaryPageHeader.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/EdgeInterpolationAlgorithm.d.ts +8 -0
- package/dist/parquetjs/parquet-thrift/EdgeInterpolationAlgorithm.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/EdgeInterpolationAlgorithm.js +14 -0
- package/dist/parquetjs/parquet-thrift/EdgeInterpolationAlgorithm.js.map +1 -0
- package/dist/parquetjs/parquet-thrift/Encoding.d.ts +2 -1
- package/dist/parquetjs/parquet-thrift/Encoding.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/Encoding.js +1 -0
- package/dist/parquetjs/parquet-thrift/Encoding.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/EnumType.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/EnumType.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/EnumType.js +1 -1
- package/dist/parquetjs/parquet-thrift/EnumType.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/FileMetaData.d.ts +2 -2
- package/dist/parquetjs/parquet-thrift/FileMetaData.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/FileMetaData.js +2 -2
- package/dist/parquetjs/parquet-thrift/FileMetaData.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/Float16Type.d.ts +9 -0
- package/dist/parquetjs/parquet-thrift/Float16Type.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/Float16Type.js +27 -0
- package/dist/parquetjs/parquet-thrift/Float16Type.js.map +1 -0
- package/dist/parquetjs/parquet-thrift/GeographyType.d.ts +14 -0
- package/dist/parquetjs/parquet-thrift/GeographyType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/GeographyType.js +51 -0
- package/dist/parquetjs/parquet-thrift/GeographyType.js.map +1 -0
- package/dist/parquetjs/parquet-thrift/GeometryType.d.ts +11 -0
- package/dist/parquetjs/parquet-thrift/GeometryType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/GeometryType.js +41 -0
- package/dist/parquetjs/parquet-thrift/GeometryType.js.map +1 -0
- package/dist/parquetjs/parquet-thrift/IndexPageHeader.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/IndexPageHeader.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/IndexPageHeader.js +1 -1
- package/dist/parquetjs/parquet-thrift/IndexPageHeader.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/IntType.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/IntType.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/IntType.js +1 -1
- package/dist/parquetjs/parquet-thrift/IntType.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/JsonType.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/JsonType.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/JsonType.js +1 -1
- package/dist/parquetjs/parquet-thrift/JsonType.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/KeyValue.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/KeyValue.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/KeyValue.js +1 -1
- package/dist/parquetjs/parquet-thrift/KeyValue.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/ListType.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/ListType.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/ListType.js +1 -1
- package/dist/parquetjs/parquet-thrift/ListType.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/LogicalType.d.ts +17 -1
- package/dist/parquetjs/parquet-thrift/LogicalType.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/LogicalType.js +97 -1
- package/dist/parquetjs/parquet-thrift/LogicalType.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/MapType.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/MapType.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/MapType.js +1 -1
- package/dist/parquetjs/parquet-thrift/MapType.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/MicroSeconds.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/MicroSeconds.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/MicroSeconds.js +1 -1
- package/dist/parquetjs/parquet-thrift/MicroSeconds.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/MilliSeconds.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/MilliSeconds.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/MilliSeconds.js +1 -1
- package/dist/parquetjs/parquet-thrift/MilliSeconds.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/NanoSeconds.d.ts +9 -0
- package/dist/parquetjs/parquet-thrift/NanoSeconds.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/NanoSeconds.js +27 -0
- package/dist/parquetjs/parquet-thrift/NanoSeconds.js.map +1 -0
- package/dist/parquetjs/parquet-thrift/NullType.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/NullType.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/NullType.js +1 -1
- package/dist/parquetjs/parquet-thrift/NullType.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/OffsetIndex.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/OffsetIndex.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/OffsetIndex.js +1 -1
- package/dist/parquetjs/parquet-thrift/OffsetIndex.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/PageEncodingStats.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/PageEncodingStats.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/PageEncodingStats.js +1 -1
- package/dist/parquetjs/parquet-thrift/PageEncodingStats.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/PageHeader.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/PageHeader.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/PageHeader.js +1 -1
- package/dist/parquetjs/parquet-thrift/PageHeader.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/PageLocation.d.ts +2 -2
- package/dist/parquetjs/parquet-thrift/PageLocation.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/PageLocation.js +2 -2
- package/dist/parquetjs/parquet-thrift/PageLocation.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/RowGroup.d.ts +2 -2
- package/dist/parquetjs/parquet-thrift/RowGroup.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/RowGroup.js +2 -2
- package/dist/parquetjs/parquet-thrift/RowGroup.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/SchemaElement.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/SchemaElement.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/SchemaElement.js +1 -1
- package/dist/parquetjs/parquet-thrift/SchemaElement.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/SortingColumn.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/SortingColumn.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/SortingColumn.js +1 -1
- package/dist/parquetjs/parquet-thrift/SortingColumn.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/Statistics.d.ts +14 -10
- package/dist/parquetjs/parquet-thrift/Statistics.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/Statistics.js +38 -2
- package/dist/parquetjs/parquet-thrift/Statistics.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/StringType.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/StringType.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/StringType.js +1 -1
- package/dist/parquetjs/parquet-thrift/StringType.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/TimeType.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/TimeType.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/TimeType.js +1 -1
- package/dist/parquetjs/parquet-thrift/TimeType.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/TimeUnit.d.ts +5 -1
- package/dist/parquetjs/parquet-thrift/TimeUnit.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/TimeUnit.js +25 -1
- package/dist/parquetjs/parquet-thrift/TimeUnit.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/TimestampType.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/TimestampType.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/TimestampType.js +1 -1
- package/dist/parquetjs/parquet-thrift/TimestampType.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/TypeDefinedOrder.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/TypeDefinedOrder.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/TypeDefinedOrder.js +1 -1
- package/dist/parquetjs/parquet-thrift/TypeDefinedOrder.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/UUIDType.d.ts +1 -1
- package/dist/parquetjs/parquet-thrift/UUIDType.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/UUIDType.js +1 -1
- package/dist/parquetjs/parquet-thrift/UUIDType.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/VariantType.d.ts +11 -0
- package/dist/parquetjs/parquet-thrift/VariantType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/VariantType.js +41 -0
- package/dist/parquetjs/parquet-thrift/VariantType.js.map +1 -0
- package/dist/parquetjs/parquet-thrift/index.d.ts +6 -2
- package/dist/parquetjs/parquet-thrift/index.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/index.js +6 -2
- package/dist/parquetjs/parquet-thrift/index.js.map +1 -1
- package/dist/parquetjs/parser/decoders.d.ts +14 -3
- package/dist/parquetjs/parser/decoders.d.ts.map +1 -1
- package/dist/parquetjs/parser/decoders.js +381 -117
- package/dist/parquetjs/parser/decoders.js.map +1 -1
- package/dist/parquetjs/parser/parquet-reader.d.ts +35 -10
- package/dist/parquetjs/parser/parquet-reader.d.ts.map +1 -1
- package/dist/parquetjs/parser/parquet-reader.js +179 -59
- package/dist/parquetjs/parser/parquet-reader.js.map +1 -1
- package/dist/parquetjs/schema/declare.d.ts +62 -8
- package/dist/parquetjs/schema/declare.d.ts.map +1 -1
- package/dist/parquetjs/schema/declare.js.map +1 -1
- package/dist/parquetjs/schema/schema.d.ts +3 -0
- package/dist/parquetjs/schema/schema.d.ts.map +1 -1
- package/dist/parquetjs/schema/schema.js +21 -3
- package/dist/parquetjs/schema/schema.js.map +1 -1
- package/dist/parquetjs/schema/shred.d.ts +2 -0
- package/dist/parquetjs/schema/shred.d.ts.map +1 -1
- package/dist/parquetjs/schema/shred.js +71 -23
- package/dist/parquetjs/schema/shred.js.map +1 -1
- package/dist/parquetjs/schema/types.d.ts.map +1 -1
- package/dist/parquetjs/schema/types.js +272 -51
- package/dist/parquetjs/schema/types.js.map +1 -1
- package/dist/parquetjs/utils/binary-utils.d.ts +41 -0
- package/dist/parquetjs/utils/binary-utils.d.ts.map +1 -0
- package/dist/parquetjs/utils/binary-utils.js +106 -0
- package/dist/parquetjs/utils/binary-utils.js.map +1 -0
- package/dist/parquetjs/utils/file-utils.d.ts +1 -1
- package/dist/parquetjs/utils/file-utils.d.ts.map +1 -1
- package/dist/parquetjs/utils/file-utils.js +3 -3
- package/dist/parquetjs/utils/file-utils.js.map +1 -1
- package/dist/parquetjs/utils/read-utils.d.ts +14 -6
- package/dist/parquetjs/utils/read-utils.d.ts.map +1 -1
- package/dist/parquetjs/utils/read-utils.js +46 -39
- package/dist/parquetjs/utils/read-utils.js.map +1 -1
- package/dist/parquetjs/utils/thrift-runtime.d.ts +106 -0
- package/dist/parquetjs/utils/thrift-runtime.d.ts.map +1 -0
- package/dist/parquetjs/utils/thrift-runtime.js +50 -0
- package/dist/parquetjs/utils/thrift-runtime.js.map +1 -0
- package/dist/parquetjs/utils/uint8-array-compact-protocol-writer.d.ts +78 -0
- package/dist/parquetjs/utils/uint8-array-compact-protocol-writer.d.ts.map +1 -0
- package/dist/parquetjs/utils/uint8-array-compact-protocol-writer.js +213 -0
- package/dist/parquetjs/utils/uint8-array-compact-protocol-writer.js.map +1 -0
- package/dist/parquetjs/utils/uint8-array-compact-protocol.d.ts +80 -0
- package/dist/parquetjs/utils/uint8-array-compact-protocol.d.ts.map +1 -0
- package/dist/parquetjs/utils/uint8-array-compact-protocol.js +276 -0
- package/dist/parquetjs/utils/uint8-array-compact-protocol.js.map +1 -0
- package/dist/parquetjs/utils/uint8-array-transport.d.ts +25 -0
- package/dist/parquetjs/utils/uint8-array-transport.d.ts.map +1 -0
- package/dist/parquetjs/utils/uint8-array-transport.js +62 -0
- package/dist/parquetjs/utils/uint8-array-transport.js.map +1 -0
- package/dist/polyfills/process.d.ts +12 -0
- package/dist/polyfills/process.d.ts.map +1 -0
- package/dist/polyfills/process.js +17 -0
- package/dist/polyfills/process.js.map +1 -0
- package/dist/unbundled.d.ts +9 -0
- package/dist/unbundled.d.ts.map +1 -0
- package/dist/unbundled.js +10 -0
- package/dist/unbundled.js.map +1 -0
- package/dist/wasm.d.ts +3 -0
- package/dist/wasm.d.ts.map +1 -0
- package/dist/wasm.js +6 -0
- package/dist/wasm.js.map +1 -0
- package/dist/workers/parquet-source-worker.d.ts +2 -0
- package/dist/workers/parquet-source-worker.d.ts.map +1 -0
- package/dist/workers/parquet-source-worker.js +17 -0
- package/dist/workers/parquet-source-worker.js.map +1 -0
- package/dist/workers/parquet-worker-node.d.ts +2 -0
- package/dist/workers/parquet-worker-node.d.ts.map +1 -0
- package/dist/workers/parquet-worker-node.js +5 -0
- package/dist/workers/parquet-worker-node.js.map +1 -0
- package/dist/workers/parquet-worker.d.ts +2 -0
- package/dist/workers/parquet-worker.d.ts.map +1 -0
- package/dist/workers/parquet-worker.js +27 -0
- package/dist/workers/parquet-worker.js.map +1 -0
- package/package.json +63 -33
- package/src/bundled.ts +33 -0
- package/src/geoparquet-loader-with-parser.ts +95 -0
- package/src/geoparquet-loader.ts +54 -0
- package/src/index.ts +49 -23
- package/src/lib/arrow/convert-columns-to-row-group.ts +3 -0
- package/src/lib/arrow/convert-schema-from-parquet.ts +97 -30
- package/src/lib/arrow/convert-schema-to-parquet.ts +16 -7
- package/src/lib/arrow/create-nested-arrow-vector.ts +589 -0
- package/src/lib/constants.ts +0 -2
- package/src/lib/encoders/encode-arrow-to-parquet.ts +10 -16
- package/src/lib/encoders/encode-table-to-parquet-js.ts +261 -0
- package/src/lib/geo/geospatial-metadata.ts +457 -0
- package/src/lib/parquet-page-index.ts +581 -0
- package/src/lib/parquet-predicate.ts +374 -0
- package/src/lib/parquet-source-worker-client.ts +39 -0
- package/src/lib/parquet-source-worker-decoder.ts +156 -0
- package/src/lib/parquet-source-worker-file.ts +56 -0
- package/src/lib/parquet-source-worker-types.ts +107 -0
- package/src/lib/parquet-worker-transport.ts +57 -0
- package/src/lib/parsers/convert-parquet-tables.ts +39 -0
- package/src/lib/parsers/get-parquet-schema.ts +3 -1
- package/src/lib/parsers/parse-geoparquet-to-geojson.ts +27 -14
- package/src/lib/parsers/parse-parquet-tables.ts +82 -0
- package/src/lib/parsers/parse-parquet-to-arrow-js.ts +829 -0
- package/src/lib/parsers/parse-parquet-to-arrow.ts +63 -36
- package/src/lib/parsers/parse-parquet-to-columns.ts +6 -7
- package/src/lib/parsers/parse-parquet-to-json.ts +101 -33
- package/src/lib/sources/parquet-range-file.ts +359 -0
- package/src/lib/sources/single-batch-queue.ts +95 -0
- package/src/lib/utils/load-wasm-node.ts +25 -0
- package/src/lib/utils/load-wasm.ts +36 -13
- package/src/lib/utils/make-stream-iterator.ts +126 -14
- package/src/lib/utils/normalize-parquet-options.ts +14 -0
- package/src/parquet-dataset-source.ts +488 -0
- package/src/parquet-format.ts +2 -0
- package/src/parquet-js-loader-types.ts +52 -0
- package/src/parquet-js-loader.ts +83 -0
- package/src/parquet-js-writer.ts +68 -0
- package/src/parquet-loader-base.ts +34 -0
- package/src/parquet-loader-options.ts +51 -0
- package/src/parquet-loader-types.ts +35 -0
- package/src/parquet-loader.ts +115 -0
- package/src/parquet-source-capabilities.ts +56 -0
- package/src/parquet-source-loader-types.ts +87 -0
- package/src/parquet-source-loader.ts +1265 -0
- package/src/parquet-source-types.ts +458 -0
- package/src/parquet-source-worker-url.ts +23 -0
- package/src/parquet-worker-url.ts +18 -0
- package/src/parquet-writer.ts +42 -0
- package/src/parquetjs/LICENSE +23 -1
- package/src/parquetjs/codecs/byte-stream-split.ts +221 -0
- package/src/parquetjs/codecs/declare.ts +35 -3
- package/src/parquetjs/codecs/delta.ts +587 -0
- package/src/parquetjs/codecs/dictionary.ts +22 -4
- package/src/parquetjs/codecs/index.ts +18 -0
- package/src/parquetjs/codecs/plain.ts +149 -87
- package/src/parquetjs/codecs/rle.ts +268 -82
- package/src/parquetjs/compression.ts +110 -155
- package/src/parquetjs/encoder/dictionary-planner.ts +99 -0
- package/src/parquetjs/encoder/page-planner.ts +83 -0
- package/src/parquetjs/encoder/parquet-encoder.ts +378 -77
- package/src/parquetjs/parquet-thrift/BsonType.ts +1 -1
- package/src/parquetjs/parquet-thrift/ColumnChunk.ts +2 -2
- package/src/parquetjs/parquet-thrift/ColumnIndex.ts +13 -13
- package/src/parquetjs/parquet-thrift/ColumnMetaData.ts +2 -2
- package/src/parquetjs/parquet-thrift/ColumnOrder.ts +1 -1
- package/src/parquetjs/parquet-thrift/DataPageHeader.ts +1 -1
- package/src/parquetjs/parquet-thrift/DataPageHeaderV2.ts +1 -1
- package/src/parquetjs/parquet-thrift/DateType.ts +1 -1
- package/src/parquetjs/parquet-thrift/DecimalType.ts +1 -1
- package/src/parquetjs/parquet-thrift/DictionaryPageHeader.ts +1 -1
- package/src/parquetjs/parquet-thrift/EdgeInterpolationAlgorithm.ts +13 -0
- package/src/parquetjs/parquet-thrift/Encoding.ts +2 -1
- package/src/parquetjs/parquet-thrift/EnumType.ts +1 -1
- package/src/parquetjs/parquet-thrift/FileMetaData.ts +2 -2
- package/src/parquetjs/parquet-thrift/Float16Type.ts +28 -0
- package/src/parquetjs/parquet-thrift/GeographyType.ts +54 -0
- package/src/parquetjs/parquet-thrift/GeometryType.ts +43 -0
- package/src/parquetjs/parquet-thrift/IndexPageHeader.ts +1 -1
- package/src/parquetjs/parquet-thrift/IntType.ts +1 -1
- package/src/parquetjs/parquet-thrift/JsonType.ts +1 -1
- package/src/parquetjs/parquet-thrift/KeyValue.ts +1 -1
- package/src/parquetjs/parquet-thrift/ListType.ts +1 -1
- package/src/parquetjs/parquet-thrift/LogicalType.ts +97 -1
- package/src/parquetjs/parquet-thrift/MapType.ts +1 -1
- package/src/parquetjs/parquet-thrift/MicroSeconds.ts +1 -1
- package/src/parquetjs/parquet-thrift/MilliSeconds.ts +1 -1
- package/src/parquetjs/parquet-thrift/NanoSeconds.ts +28 -0
- package/src/parquetjs/parquet-thrift/NullType.ts +1 -1
- package/src/parquetjs/parquet-thrift/OffsetIndex.ts +1 -1
- package/src/parquetjs/parquet-thrift/PageEncodingStats.ts +1 -1
- package/src/parquetjs/parquet-thrift/PageHeader.ts +1 -1
- package/src/parquetjs/parquet-thrift/PageLocation.ts +2 -2
- package/src/parquetjs/parquet-thrift/RowGroup.ts +2 -2
- package/src/parquetjs/parquet-thrift/SchemaElement.ts +1 -1
- package/src/parquetjs/parquet-thrift/SortingColumn.ts +1 -1
- package/src/parquetjs/parquet-thrift/Statistics.ts +50 -14
- package/src/parquetjs/parquet-thrift/StringType.ts +1 -1
- package/src/parquetjs/parquet-thrift/TimeType.ts +1 -1
- package/src/parquetjs/parquet-thrift/TimeUnit.ts +25 -1
- package/src/parquetjs/parquet-thrift/TimestampType.ts +1 -1
- package/src/parquetjs/parquet-thrift/TypeDefinedOrder.ts +1 -1
- package/src/parquetjs/parquet-thrift/UUIDType.ts +1 -1
- package/src/parquetjs/parquet-thrift/VariantType.ts +43 -0
- package/src/parquetjs/parquet-thrift/index.ts +6 -4
- package/src/parquetjs/parser/decoders.ts +487 -133
- package/src/parquetjs/parser/parquet-reader.ts +297 -71
- package/src/parquetjs/schema/declare.ts +99 -7
- package/src/parquetjs/schema/schema.ts +23 -3
- package/src/parquetjs/schema/shred.ts +112 -35
- package/src/parquetjs/schema/types.ts +294 -61
- package/src/parquetjs/utils/binary-utils.ts +132 -0
- package/src/parquetjs/utils/file-utils.ts +4 -4
- package/src/parquetjs/utils/read-utils.ts +56 -55
- package/src/parquetjs/utils/thrift-runtime.ts +109 -0
- package/src/parquetjs/utils/uint8-array-compact-protocol-writer.ts +254 -0
- package/src/parquetjs/utils/uint8-array-compact-protocol.ts +322 -0
- package/src/parquetjs/utils/uint8-array-transport.ts +79 -0
- package/src/polyfills/process.ts +17 -0
- package/src/unbundled.ts +33 -0
- package/src/wasm.ts +6 -0
- package/src/workers/parquet-source-worker.ts +21 -0
- package/src/workers/parquet-worker-node.ts +5 -0
- package/src/workers/parquet-worker.ts +31 -0
- package/dist/parquet-arrow-loader.d.ts +0 -43
- package/dist/parquet-arrow-loader.d.ts.map +0 -1
- package/dist/parquet-arrow-loader.js +0 -49
- package/dist/parquet-arrow-loader.js.map +0 -1
- package/dist/parquet-arrow-writer.d.ts +0 -9
- package/dist/parquet-arrow-writer.d.ts.map +0 -1
- package/dist/parquet-arrow-writer.js +0 -23
- package/dist/parquet-arrow-writer.js.map +0 -1
- package/dist/parquet-json-loader.d.ts +0 -148
- package/dist/parquet-json-loader.d.ts.map +0 -1
- package/dist/parquet-json-loader.js +0 -106
- package/dist/parquet-json-loader.js.map +0 -1
- package/dist/parquet-json-writer.d.ts +0 -19
- package/dist/parquet-json-writer.d.ts.map +0 -1
- package/dist/parquet-json-writer.js +0 -20
- package/dist/parquet-json-writer.js.map +0 -1
- package/dist/polyfills/buffer/buffer-polyfill.browser.d.ts +0 -3
- package/dist/polyfills/buffer/buffer-polyfill.browser.d.ts.map +0 -1
- package/dist/polyfills/buffer/buffer-polyfill.browser.js +0 -12
- package/dist/polyfills/buffer/buffer-polyfill.browser.js.map +0 -1
- package/dist/polyfills/buffer/buffer-polyfill.node.d.ts +0 -3
- package/dist/polyfills/buffer/buffer-polyfill.node.d.ts.map +0 -1
- package/dist/polyfills/buffer/buffer-polyfill.node.js +0 -17
- package/dist/polyfills/buffer/buffer-polyfill.node.js.map +0 -1
- package/dist/polyfills/buffer/buffer.d.ts +0 -222
- package/dist/polyfills/buffer/buffer.d.ts.map +0 -1
- package/dist/polyfills/buffer/buffer.js +0 -1868
- package/dist/polyfills/buffer/buffer.js.map +0 -1
- package/dist/polyfills/buffer/index.d.ts +0 -4
- package/dist/polyfills/buffer/index.d.ts.map +0 -1
- package/dist/polyfills/buffer/index.js +0 -9
- package/dist/polyfills/buffer/index.js.map +0 -1
- package/dist/polyfills/buffer/install-buffer-polyfill.d.ts +0 -2
- package/dist/polyfills/buffer/install-buffer-polyfill.d.ts.map +0 -1
- package/dist/polyfills/buffer/install-buffer-polyfill.js +0 -10
- package/dist/polyfills/buffer/install-buffer-polyfill.js.map +0 -1
- package/dist/workers/parquet-json-worker.d.ts +0 -2
- package/dist/workers/parquet-json-worker.d.ts.map +0 -1
- package/dist/workers/parquet-json-worker.js +0 -7
- package/dist/workers/parquet-json-worker.js.map +0 -1
- package/src/parquet-arrow-loader.ts +0 -74
- package/src/parquet-arrow-writer.ts +0 -33
- package/src/parquet-json-loader.ts +0 -160
- package/src/parquet-json-writer.ts +0 -27
- package/src/parquetjs/modules.d.ts +0 -25
- package/src/polyfills/buffer/buffer-polyfill.browser.ts +0 -14
- package/src/polyfills/buffer/buffer-polyfill.node.ts +0 -19
- package/src/polyfills/buffer/buffer.ts +0 -2205
- package/src/polyfills/buffer/index.ts +0 -11
- package/src/polyfills/buffer/install-buffer-polyfill.ts +0 -12
- package/src/workers/parquet-json-worker.ts +0 -8
|
@@ -0,0 +1,62 @@
|
|
|
1
|
+
import type { ReadableFile } from '@loaders.gl/loader-utils';
|
|
2
|
+
import type { ParquetPredicate } from "../parquet-source-types.js";
|
|
3
|
+
import { type ColumnChunk, type RowGroup } from "../parquetjs/parquet-thrift/index.js";
|
|
4
|
+
import type { ParquetField } from "../parquetjs/schema/declare.js";
|
|
5
|
+
import type { ParquetSchema } from "../parquetjs/schema/schema.js";
|
|
6
|
+
/** Half-open logical row range relative to one Parquet row group. */
|
|
7
|
+
export type ParquetRowRange = {
|
|
8
|
+
/** First included row. */
|
|
9
|
+
start: number;
|
|
10
|
+
/** First excluded row. */
|
|
11
|
+
end: number;
|
|
12
|
+
};
|
|
13
|
+
/** One data-page location decoded from a Parquet offset index. */
|
|
14
|
+
export type ParquetDataPageLocation = {
|
|
15
|
+
/** Absolute page-header offset in the source object. */
|
|
16
|
+
offset: number;
|
|
17
|
+
/** Page header plus compressed body byte length. */
|
|
18
|
+
compressedByteLength: number;
|
|
19
|
+
/** First logical row represented by the page. */
|
|
20
|
+
firstRowIndex: number;
|
|
21
|
+
/** First logical row represented by the next page. */
|
|
22
|
+
endRowIndex: number;
|
|
23
|
+
};
|
|
24
|
+
/** Page locations keyed by the physical Parquet column path. */
|
|
25
|
+
export type ParquetPageLocations = Record<string, ParquetDataPageLocation[]>;
|
|
26
|
+
/** Selective page plan for one row group. */
|
|
27
|
+
export type ParquetPagePruningPlan = {
|
|
28
|
+
/** Candidate row ranges after conservative page-statistics pruning. */
|
|
29
|
+
rowRanges: ParquetRowRange[];
|
|
30
|
+
/** Offset-index page locations for every decoded column. */
|
|
31
|
+
pageLocations: ParquetPageLocations;
|
|
32
|
+
/** Number of column and offset index blobs decoded. */
|
|
33
|
+
indexCount: number;
|
|
34
|
+
/** Data pages in all decoded column chunks. */
|
|
35
|
+
totalPageCount: number;
|
|
36
|
+
/** Data pages required by the candidate ranges. */
|
|
37
|
+
selectedPageCount: number;
|
|
38
|
+
/** Logical rows eliminated before data-page reads. */
|
|
39
|
+
prunedRowCount: number;
|
|
40
|
+
};
|
|
41
|
+
/**
|
|
42
|
+
* Builds a conservative selective-page plan from Parquet column and offset indexes.
|
|
43
|
+
*
|
|
44
|
+
* Returns `undefined` when indexes or the current flat-column decoder cannot safely avoid full
|
|
45
|
+
* column-chunk reads. An empty `rowRanges` array means the indexes prove the predicate cannot match.
|
|
46
|
+
*/
|
|
47
|
+
export declare function createParquetPagePruningPlan(file: ReadableFile, rowGroup: RowGroup, schema: ParquetSchema, selectedColumnPaths: readonly string[][], predicate: ParquetPredicate, signal?: AbortSignal): Promise<ParquetPagePruningPlan | undefined>;
|
|
48
|
+
/** Returns data-page byte ranges needed to decode a selective row-group plan. */
|
|
49
|
+
export declare function getParquetPageReadRanges(rowGroup: RowGroup, selectedColumnPaths: readonly string[][], plan: ParquetPagePruningPlan): Array<{
|
|
50
|
+
offset: number;
|
|
51
|
+
length: number;
|
|
52
|
+
}>;
|
|
53
|
+
/** Decodes one compact-protocol Parquet offset index. */
|
|
54
|
+
export declare function decodeOffsetIndex(bytes: Uint8Array, rowCount: number): ParquetDataPageLocation[];
|
|
55
|
+
/** Decodes one physical page-index statistic and applies its Parquet logical annotation. */
|
|
56
|
+
export declare function decodeParquetPageStatisticsValue(bytes: Uint8Array, field: ParquetField): unknown;
|
|
57
|
+
/** Validates one optional footer index byte range against the containing file. */
|
|
58
|
+
export declare function getParquetIndexRange(offsetValue: ColumnChunk['offset_index_offset'], lengthValue: number | undefined, fileSize: number): {
|
|
59
|
+
offset: number;
|
|
60
|
+
length: number;
|
|
61
|
+
} | undefined;
|
|
62
|
+
//# sourceMappingURL=parquet-page-index.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"parquet-page-index.d.ts","sourceRoot":"","sources":["../../src/lib/parquet-page-index.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,EAAC,YAAY,EAAC,MAAM,0BAA0B,CAAC;AAE3D,OAAO,KAAK,EAIV,gBAAgB,EACjB,mCAAgC;AACjC,OAAO,EAIL,KAAK,WAAW,EAChB,KAAK,QAAQ,EACd,6CAA0C;AAC3C,OAAO,KAAK,EAAC,YAAY,EAAC,uCAAoC;AAC9D,OAAO,KAAK,EAAC,aAAa,EAAC,sCAAmC;AAgB9D,qEAAqE;AACrE,MAAM,MAAM,eAAe,GAAG;IAC5B,0BAA0B;IAC1B,KAAK,EAAE,MAAM,CAAC;IACd,0BAA0B;IAC1B,GAAG,EAAE,MAAM,CAAC;CACb,CAAC;AAEF,kEAAkE;AAClE,MAAM,MAAM,uBAAuB,GAAG;IACpC,wDAAwD;IACxD,MAAM,EAAE,MAAM,CAAC;IACf,oDAAoD;IACpD,oBAAoB,EAAE,MAAM,CAAC;IAC7B,iDAAiD;IACjD,aAAa,EAAE,MAAM,CAAC;IACtB,sDAAsD;IACtD,WAAW,EAAE,MAAM,CAAC;CACrB,CAAC;AAEF,gEAAgE;AAChE,MAAM,MAAM,oBAAoB,GAAG,MAAM,CAAC,MAAM,EAAE,uBAAuB,EAAE,CAAC,CAAC;AAE7E,6CAA6C;AAC7C,MAAM,MAAM,sBAAsB,GAAG;IACnC,uEAAuE;IACvE,SAAS,EAAE,eAAe,EAAE,CAAC;IAC7B,4DAA4D;IAC5D,aAAa,EAAE,oBAAoB,CAAC;IACpC,uDAAuD;IACvD,UAAU,EAAE,MAAM,CAAC;IACnB,+CAA+C;IAC/C,cAAc,EAAE,MAAM,CAAC;IACvB,mDAAmD;IACnD,iBAAiB,EAAE,MAAM,CAAC;IAC1B,sDAAsD;IACtD,cAAc,EAAE,MAAM,CAAC;CACxB,CAAC;AAeF;;;;;GAKG;AACH,wBAAsB,4BAA4B,CAChD,IAAI,EAAE,YAAY,EAClB,QAAQ,EAAE,QAAQ,EAClB,MAAM,EAAE,aAAa,EACrB,mBAAmB,EAAE,SAAS,MAAM,EAAE,EAAE,EACxC,SAAS,EAAE,gBAAgB,EAC3B,MAAM,CAAC,EAAE,WAAW,GACnB,OAAO,CAAC,sBAAsB,GAAG,SAAS,CAAC,CAqG7C;AAED,iFAAiF;AACjF,wBAAgB,wBAAwB,CACtC,QAAQ,EAAE,QAAQ,EAClB,mBAAmB,EAAE,SAAS,MAAM,EAAE,EAAE,EACxC,IAAI,EAAE,sBAAsB,GAC3B,KAAK,CAAC;IAAC,MAAM,EAAE,MAAM,CAAC;IAAC,MAAM,EAAE,MAAM,CAAA;CAAC,CAAC,CAkCzC;AAED,yDAAyD;AACzD,wBAAgB,iBAAiB,CAAC,KAAK,EAAE,UAAU,EAAE,QAAQ,EAAE,MAAM,GAAG,uBAAuB,EAAE,CA8BhG;AAwFD,4FAA4F;AAC5F,wBAAgB,gCAAgC,CAAC,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,YAAY,GAAG,OAAO,CA6BhG;AA6CD,kFAAkF;AAClF,wBAAgB,oBAAoB,CAClC,WAAW,EAAE,WAAW,CAAC,qBAAqB,CAAC,EAC/C,WAAW,EAAE,MAAM,GAAG,SAAS,EAC/B,QAAQ,EAAE,MAAM,GACf;IAAC,MAAM,EAAE,MAAM,CAAC;IAAC,MAAM,EAAE,MAAM,CAAA;CAAC,GAAG,SAAS,CAiB9C"}
|
|
@@ -0,0 +1,377 @@
|
|
|
1
|
+
// loaders.gl
|
|
2
|
+
// SPDX-License-Identifier: MIT
|
|
3
|
+
// Copyright (c) vis.gl contributors
|
|
4
|
+
import { ColumnIndex, Encoding, OffsetIndex } from "../parquetjs/parquet-thrift/index.js";
|
|
5
|
+
import * as Types from "../parquetjs/schema/types.js";
|
|
6
|
+
import { copyUint8Array, readDoubleLE, readFloatLE, readInt32LE, readInt64LE, readUInt32LE, readUInt64LE, toUint8Array } from "../parquetjs/utils/binary-utils.js";
|
|
7
|
+
import { Uint8ArrayCompactProtocol } from "../parquetjs/utils/uint8-array-compact-protocol.js";
|
|
8
|
+
import { Uint8ArrayTransport } from "../parquetjs/utils/uint8-array-transport.js";
|
|
9
|
+
import { canParquetStatisticsMatch, getParquetPredicateColumns } from "./parquet-predicate.js";
|
|
10
|
+
/**
|
|
11
|
+
* Builds a conservative selective-page plan from Parquet column and offset indexes.
|
|
12
|
+
*
|
|
13
|
+
* Returns `undefined` when indexes or the current flat-column decoder cannot safely avoid full
|
|
14
|
+
* column-chunk reads. An empty `rowRanges` array means the indexes prove the predicate cannot match.
|
|
15
|
+
*/
|
|
16
|
+
export async function createParquetPagePruningPlan(file, rowGroup, schema, selectedColumnPaths, predicate, signal) {
|
|
17
|
+
const rowCount = Number(rowGroup.num_rows);
|
|
18
|
+
const selectedColumnChunks = getSelectedColumnChunks(rowGroup, selectedColumnPaths);
|
|
19
|
+
if (!isSafeFlatPageSelection(schema, selectedColumnChunks) || rowCount <= 0) {
|
|
20
|
+
return undefined;
|
|
21
|
+
}
|
|
22
|
+
const pageLocations = {};
|
|
23
|
+
const pageStatistics = {};
|
|
24
|
+
const predicateColumns = new Set(getParquetPredicateColumns(predicate));
|
|
25
|
+
let indexCount = 0;
|
|
26
|
+
await Promise.all(selectedColumnChunks.map(async (columnChunk) => {
|
|
27
|
+
const path = columnChunk.meta_data.path_in_schema;
|
|
28
|
+
const pathKey = path.join('.');
|
|
29
|
+
const offsetIndexRange = getParquetIndexRange(columnChunk.offset_index_offset, columnChunk.offset_index_length, file.size);
|
|
30
|
+
if (!offsetIndexRange) {
|
|
31
|
+
return;
|
|
32
|
+
}
|
|
33
|
+
const offsetIndexBytes = await file.read(offsetIndexRange.offset, offsetIndexRange.length, signal);
|
|
34
|
+
let pages;
|
|
35
|
+
try {
|
|
36
|
+
pages = decodeOffsetIndex(toUint8Array(offsetIndexBytes), rowCount);
|
|
37
|
+
}
|
|
38
|
+
catch {
|
|
39
|
+
return;
|
|
40
|
+
}
|
|
41
|
+
pageLocations[pathKey] = pages;
|
|
42
|
+
indexCount++;
|
|
43
|
+
if (path.length !== 1 || !predicateColumns.has(path[0])) {
|
|
44
|
+
return;
|
|
45
|
+
}
|
|
46
|
+
const columnIndexRange = getParquetIndexRange(columnChunk.column_index_offset, columnChunk.column_index_length, file.size);
|
|
47
|
+
if (!columnIndexRange) {
|
|
48
|
+
return;
|
|
49
|
+
}
|
|
50
|
+
const columnIndexBytes = await file.read(columnIndexRange.offset, columnIndexRange.length, signal);
|
|
51
|
+
const field = schema.findField(path);
|
|
52
|
+
try {
|
|
53
|
+
pageStatistics[path[0]] = {
|
|
54
|
+
pages,
|
|
55
|
+
statistics: decodeColumnIndex(toUint8Array(columnIndexBytes), pages, field)
|
|
56
|
+
};
|
|
57
|
+
}
|
|
58
|
+
catch {
|
|
59
|
+
return;
|
|
60
|
+
}
|
|
61
|
+
indexCount++;
|
|
62
|
+
}));
|
|
63
|
+
if (selectedColumnChunks.some(columnChunk => !pageLocations[columnChunk.meta_data.path_in_schema.join('.')])) {
|
|
64
|
+
return undefined;
|
|
65
|
+
}
|
|
66
|
+
const candidateRanges = getPredicateRowRanges(predicate, pageStatistics, rowCount);
|
|
67
|
+
if (candidateRanges === undefined) {
|
|
68
|
+
return undefined;
|
|
69
|
+
}
|
|
70
|
+
const rowRanges = expandAndMergeRowRanges(candidateRanges, pageLocations, rowCount);
|
|
71
|
+
const prunedRowCount = rowCount - getRowRangeCount(rowRanges);
|
|
72
|
+
if (prunedRowCount <= 0) {
|
|
73
|
+
return undefined;
|
|
74
|
+
}
|
|
75
|
+
const totalPageCount = Object.values(pageLocations).reduce((count, pages) => count + pages.length, 0);
|
|
76
|
+
const selectedPageCount = Object.values(pageLocations).reduce((count, pages) => count + countSelectedPages(pages, rowRanges), 0);
|
|
77
|
+
return {
|
|
78
|
+
rowRanges,
|
|
79
|
+
pageLocations,
|
|
80
|
+
indexCount,
|
|
81
|
+
totalPageCount,
|
|
82
|
+
selectedPageCount,
|
|
83
|
+
prunedRowCount
|
|
84
|
+
};
|
|
85
|
+
}
|
|
86
|
+
/** Returns data-page byte ranges needed to decode a selective row-group plan. */
|
|
87
|
+
export function getParquetPageReadRanges(rowGroup, selectedColumnPaths, plan) {
|
|
88
|
+
const ranges = [];
|
|
89
|
+
for (const columnChunk of getSelectedColumnChunks(rowGroup, selectedColumnPaths)) {
|
|
90
|
+
const columnMetadata = columnChunk.meta_data;
|
|
91
|
+
const pages = plan.pageLocations[columnMetadata.path_in_schema.join('.')];
|
|
92
|
+
const selectedPages = pages.filter(page => plan.rowRanges.some(range => page.endRowIndex > range.start && page.firstRowIndex < range.end));
|
|
93
|
+
if (!selectedPages.length) {
|
|
94
|
+
continue;
|
|
95
|
+
}
|
|
96
|
+
const dictionaryPageOffset = Number(columnMetadata.dictionary_page_offset);
|
|
97
|
+
if (Number.isSafeInteger(dictionaryPageOffset) && dictionaryPageOffset > 0) {
|
|
98
|
+
ranges.push({
|
|
99
|
+
offset: dictionaryPageOffset,
|
|
100
|
+
length: Math.max(0, pages[0].offset - dictionaryPageOffset)
|
|
101
|
+
});
|
|
102
|
+
}
|
|
103
|
+
for (const rowRange of plan.rowRanges) {
|
|
104
|
+
const overlappingPages = pages.filter(page => page.endRowIndex > rowRange.start && page.firstRowIndex < rowRange.end);
|
|
105
|
+
if (!overlappingPages.length) {
|
|
106
|
+
continue;
|
|
107
|
+
}
|
|
108
|
+
const firstPage = overlappingPages[0];
|
|
109
|
+
const lastPage = overlappingPages[overlappingPages.length - 1];
|
|
110
|
+
ranges.push({
|
|
111
|
+
offset: firstPage.offset,
|
|
112
|
+
length: lastPage.offset + lastPage.compressedByteLength - firstPage.offset
|
|
113
|
+
});
|
|
114
|
+
}
|
|
115
|
+
}
|
|
116
|
+
return mergeByteRanges(ranges.filter(range => range.length > 0));
|
|
117
|
+
}
|
|
118
|
+
/** Decodes one compact-protocol Parquet offset index. */
|
|
119
|
+
export function decodeOffsetIndex(bytes, rowCount) {
|
|
120
|
+
const protocol = new Uint8ArrayCompactProtocol(new Uint8ArrayTransport(bytes));
|
|
121
|
+
const offsetIndex = OffsetIndex.read(protocol);
|
|
122
|
+
if (offsetIndex.page_locations.length === 0 ||
|
|
123
|
+
Number(offsetIndex.page_locations[0].first_row_index) !== 0) {
|
|
124
|
+
throw new Error('Invalid Parquet offset index page locations');
|
|
125
|
+
}
|
|
126
|
+
return offsetIndex.page_locations.map((location, index, locations) => {
|
|
127
|
+
const firstRowIndex = Number(location.first_row_index);
|
|
128
|
+
const endRowIndex = index + 1 < locations.length ? Number(locations[index + 1].first_row_index) : rowCount;
|
|
129
|
+
const offset = Number(location.offset);
|
|
130
|
+
const compressedByteLength = location.compressed_page_size;
|
|
131
|
+
if (!Number.isSafeInteger(offset) ||
|
|
132
|
+
!Number.isSafeInteger(compressedByteLength) ||
|
|
133
|
+
!Number.isSafeInteger(firstRowIndex) ||
|
|
134
|
+
!Number.isSafeInteger(endRowIndex) ||
|
|
135
|
+
offset < 0 ||
|
|
136
|
+
compressedByteLength <= 0 ||
|
|
137
|
+
firstRowIndex < 0 ||
|
|
138
|
+
endRowIndex <= firstRowIndex ||
|
|
139
|
+
endRowIndex > rowCount) {
|
|
140
|
+
throw new Error('Invalid Parquet offset index page location');
|
|
141
|
+
}
|
|
142
|
+
return { offset, compressedByteLength, firstRowIndex, endRowIndex };
|
|
143
|
+
});
|
|
144
|
+
}
|
|
145
|
+
/** Returns whether one predicate leaf can match supplied page statistics. */
|
|
146
|
+
function getPredicateRowRanges(predicate, columns, rowCount) {
|
|
147
|
+
if (predicate.op === 'and' || predicate.op === 'or') {
|
|
148
|
+
if (predicate.op === 'and') {
|
|
149
|
+
let ranges;
|
|
150
|
+
for (const child of predicate.args) {
|
|
151
|
+
ranges = intersectRowRanges(ranges, getPredicateRowRanges(child, columns, rowCount));
|
|
152
|
+
}
|
|
153
|
+
return ranges;
|
|
154
|
+
}
|
|
155
|
+
let ranges = [];
|
|
156
|
+
for (const child of predicate.args) {
|
|
157
|
+
const childRanges = getPredicateRowRanges(child, columns, rowCount);
|
|
158
|
+
if (childRanges === undefined) {
|
|
159
|
+
return undefined;
|
|
160
|
+
}
|
|
161
|
+
ranges = unionRowRanges(ranges, childRanges);
|
|
162
|
+
}
|
|
163
|
+
return ranges;
|
|
164
|
+
}
|
|
165
|
+
if (predicate.op === 'not') {
|
|
166
|
+
return undefined;
|
|
167
|
+
}
|
|
168
|
+
const leafPredicate = predicate;
|
|
169
|
+
const column = columns[leafPredicate.args[0].property];
|
|
170
|
+
if (!column) {
|
|
171
|
+
return undefined;
|
|
172
|
+
}
|
|
173
|
+
const ranges = [];
|
|
174
|
+
for (let pageIndex = 0; pageIndex < column.pages.length; pageIndex++) {
|
|
175
|
+
const page = column.pages[pageIndex];
|
|
176
|
+
if (canParquetStatisticsMatch(leafPredicate, column.statistics[pageIndex], page.endRowIndex - page.firstRowIndex)) {
|
|
177
|
+
ranges.push({ start: page.firstRowIndex, end: page.endRowIndex });
|
|
178
|
+
}
|
|
179
|
+
}
|
|
180
|
+
return mergeRowRanges(ranges);
|
|
181
|
+
}
|
|
182
|
+
/** Decodes column-index values into the logical representation used by exact predicates. */
|
|
183
|
+
function decodeColumnIndex(bytes, pages, field) {
|
|
184
|
+
const protocol = new Uint8ArrayCompactProtocol(new Uint8ArrayTransport(bytes));
|
|
185
|
+
const columnIndex = ColumnIndex.read(protocol);
|
|
186
|
+
if (columnIndex.null_pages.length !== pages.length ||
|
|
187
|
+
columnIndex.min_values.length !== pages.length ||
|
|
188
|
+
columnIndex.max_values.length !== pages.length ||
|
|
189
|
+
(columnIndex.null_counts !== undefined && columnIndex.null_counts.length !== pages.length)) {
|
|
190
|
+
throw new Error('Parquet column and offset indexes contain different page counts');
|
|
191
|
+
}
|
|
192
|
+
return pages.map((_page, pageIndex) => ({
|
|
193
|
+
min: columnIndex.null_pages[pageIndex]
|
|
194
|
+
? undefined
|
|
195
|
+
: decodeParquetPageStatisticsValue(columnIndex.min_values[pageIndex], field),
|
|
196
|
+
max: columnIndex.null_pages[pageIndex]
|
|
197
|
+
? undefined
|
|
198
|
+
: decodeParquetPageStatisticsValue(columnIndex.max_values[pageIndex], field),
|
|
199
|
+
nullCount: columnIndex.null_counts?.[pageIndex] !== undefined
|
|
200
|
+
? Number(columnIndex.null_counts[pageIndex])
|
|
201
|
+
: columnIndex.null_pages[pageIndex]
|
|
202
|
+
? pages[pageIndex].endRowIndex - pages[pageIndex].firstRowIndex
|
|
203
|
+
: undefined,
|
|
204
|
+
minIsExact: true,
|
|
205
|
+
maxIsExact: true
|
|
206
|
+
}));
|
|
207
|
+
}
|
|
208
|
+
/** Decodes one physical page-index statistic and applies its Parquet logical annotation. */
|
|
209
|
+
export function decodeParquetPageStatisticsValue(bytes, field) {
|
|
210
|
+
let primitiveValue;
|
|
211
|
+
switch (field.primitiveType) {
|
|
212
|
+
case 'BOOLEAN':
|
|
213
|
+
primitiveValue = Boolean(bytes[0]);
|
|
214
|
+
break;
|
|
215
|
+
case 'INT32':
|
|
216
|
+
primitiveValue =
|
|
217
|
+
field.originalType === 'UINT_32' ? readUInt32LE(bytes, 0) : readInt32LE(bytes, 0);
|
|
218
|
+
break;
|
|
219
|
+
case 'INT64':
|
|
220
|
+
primitiveValue =
|
|
221
|
+
field.originalType === 'UINT_64' ? readUInt64LE(bytes, 0) : readInt64LE(bytes, 0);
|
|
222
|
+
break;
|
|
223
|
+
case 'FLOAT':
|
|
224
|
+
primitiveValue = readFloatLE(bytes, 0);
|
|
225
|
+
break;
|
|
226
|
+
case 'DOUBLE':
|
|
227
|
+
primitiveValue = readDoubleLE(bytes, 0);
|
|
228
|
+
break;
|
|
229
|
+
case 'INT96':
|
|
230
|
+
case 'BYTE_ARRAY':
|
|
231
|
+
case 'FIXED_LEN_BYTE_ARRAY':
|
|
232
|
+
primitiveValue = copyUint8Array(bytes);
|
|
233
|
+
break;
|
|
234
|
+
default:
|
|
235
|
+
return undefined;
|
|
236
|
+
}
|
|
237
|
+
return Types.fromPrimitive(field.originalType || field.primitiveType, primitiveValue, field);
|
|
238
|
+
}
|
|
239
|
+
/** Returns selected physical chunks, using an empty path list as the all-column sentinel. */
|
|
240
|
+
function getSelectedColumnChunks(rowGroup, selectedColumnPaths) {
|
|
241
|
+
return rowGroup.columns.filter(columnChunk => {
|
|
242
|
+
const path = columnChunk.meta_data?.path_in_schema;
|
|
243
|
+
return Boolean(path &&
|
|
244
|
+
(selectedColumnPaths.length === 0 ||
|
|
245
|
+
selectedColumnPaths.some(selectedPath => selectedPath[0] === path[0])));
|
|
246
|
+
});
|
|
247
|
+
}
|
|
248
|
+
/** Restricts selective page reads to independently materializable flat primitive columns. */
|
|
249
|
+
function isSafeFlatPageSelection(schema, columnChunks) {
|
|
250
|
+
return (columnChunks.length > 0 &&
|
|
251
|
+
columnChunks.every(columnChunk => {
|
|
252
|
+
const path = columnChunk.meta_data?.path_in_schema;
|
|
253
|
+
if (!path || path.length !== 1) {
|
|
254
|
+
return false;
|
|
255
|
+
}
|
|
256
|
+
const field = schema.findField(path);
|
|
257
|
+
const dictionaryEncoded = columnChunk.meta_data.encodings.some(encoding => encoding === Encoding.PLAIN_DICTIONARY || encoding === Encoding.RLE_DICTIONARY);
|
|
258
|
+
const dictionaryPageOffset = Number(columnChunk.meta_data.dictionary_page_offset);
|
|
259
|
+
return (field.path.length === 1 &&
|
|
260
|
+
field.repetitionType !== 'REPEATED' &&
|
|
261
|
+
field.rLevelMax === 0 &&
|
|
262
|
+
(!dictionaryEncoded ||
|
|
263
|
+
(Number.isSafeInteger(dictionaryPageOffset) && dictionaryPageOffset > 0)));
|
|
264
|
+
}));
|
|
265
|
+
}
|
|
266
|
+
/** Validates one optional footer index byte range against the containing file. */
|
|
267
|
+
export function getParquetIndexRange(offsetValue, lengthValue, fileSize) {
|
|
268
|
+
if (offsetValue === undefined || lengthValue === undefined) {
|
|
269
|
+
return undefined;
|
|
270
|
+
}
|
|
271
|
+
const offset = Number(offsetValue);
|
|
272
|
+
if (!Number.isSafeInteger(offset) ||
|
|
273
|
+
!Number.isSafeInteger(lengthValue) ||
|
|
274
|
+
!Number.isSafeInteger(fileSize) ||
|
|
275
|
+
offset < 0 ||
|
|
276
|
+
lengthValue <= 0 ||
|
|
277
|
+
fileSize < 0 ||
|
|
278
|
+
offset > fileSize - lengthValue) {
|
|
279
|
+
return undefined;
|
|
280
|
+
}
|
|
281
|
+
return { offset, length: lengthValue };
|
|
282
|
+
}
|
|
283
|
+
/** Expands ranges to selected-column page boundaries until all columns agree, then merges them. */
|
|
284
|
+
function expandAndMergeRowRanges(ranges, pageLocations, rowCount) {
|
|
285
|
+
const pagesByColumn = Object.values(pageLocations);
|
|
286
|
+
const expanded = ranges.map(range => {
|
|
287
|
+
let current = { ...range };
|
|
288
|
+
let changed = true;
|
|
289
|
+
while (changed) {
|
|
290
|
+
changed = false;
|
|
291
|
+
for (const pages of pagesByColumn) {
|
|
292
|
+
const overlapping = pages.filter(page => page.endRowIndex > current.start && page.firstRowIndex < current.end);
|
|
293
|
+
if (!overlapping.length) {
|
|
294
|
+
continue;
|
|
295
|
+
}
|
|
296
|
+
const start = Math.min(current.start, overlapping[0].firstRowIndex);
|
|
297
|
+
const end = Math.max(current.end, overlapping[overlapping.length - 1].endRowIndex);
|
|
298
|
+
if (start !== current.start || end !== current.end) {
|
|
299
|
+
current = { start, end };
|
|
300
|
+
changed = true;
|
|
301
|
+
}
|
|
302
|
+
}
|
|
303
|
+
}
|
|
304
|
+
return { start: Math.max(0, current.start), end: Math.min(rowCount, current.end) };
|
|
305
|
+
});
|
|
306
|
+
return mergeRowRanges(expanded);
|
|
307
|
+
}
|
|
308
|
+
/** Intersects sorted row ranges; `undefined` represents every row. */
|
|
309
|
+
function intersectRowRanges(left, right) {
|
|
310
|
+
if (left === undefined) {
|
|
311
|
+
return right;
|
|
312
|
+
}
|
|
313
|
+
if (right === undefined) {
|
|
314
|
+
return left;
|
|
315
|
+
}
|
|
316
|
+
const ranges = [];
|
|
317
|
+
let leftIndex = 0;
|
|
318
|
+
let rightIndex = 0;
|
|
319
|
+
while (leftIndex < left.length && rightIndex < right.length) {
|
|
320
|
+
const start = Math.max(left[leftIndex].start, right[rightIndex].start);
|
|
321
|
+
const end = Math.min(left[leftIndex].end, right[rightIndex].end);
|
|
322
|
+
if (start < end) {
|
|
323
|
+
ranges.push({ start, end });
|
|
324
|
+
}
|
|
325
|
+
if (left[leftIndex].end < right[rightIndex].end) {
|
|
326
|
+
leftIndex++;
|
|
327
|
+
}
|
|
328
|
+
else {
|
|
329
|
+
rightIndex++;
|
|
330
|
+
}
|
|
331
|
+
}
|
|
332
|
+
return ranges;
|
|
333
|
+
}
|
|
334
|
+
/** Unions two sorted row-range sets. */
|
|
335
|
+
function unionRowRanges(left, right) {
|
|
336
|
+
return mergeRowRanges([...left, ...right]);
|
|
337
|
+
}
|
|
338
|
+
/** Sorts and merges overlapping or touching logical row ranges. */
|
|
339
|
+
function mergeRowRanges(ranges) {
|
|
340
|
+
const sorted = [...ranges].sort((left, right) => left.start - right.start || left.end - right.end);
|
|
341
|
+
const merged = [];
|
|
342
|
+
for (const range of sorted) {
|
|
343
|
+
const previous = merged[merged.length - 1];
|
|
344
|
+
if (previous && range.start <= previous.end) {
|
|
345
|
+
previous.end = Math.max(previous.end, range.end);
|
|
346
|
+
}
|
|
347
|
+
else {
|
|
348
|
+
merged.push({ ...range });
|
|
349
|
+
}
|
|
350
|
+
}
|
|
351
|
+
return merged;
|
|
352
|
+
}
|
|
353
|
+
/** Counts logical rows represented by disjoint ranges. */
|
|
354
|
+
function getRowRangeCount(ranges) {
|
|
355
|
+
return ranges.reduce((count, range) => count + range.end - range.start, 0);
|
|
356
|
+
}
|
|
357
|
+
/** Counts unique pages overlapping at least one selected row range. */
|
|
358
|
+
function countSelectedPages(pages, ranges) {
|
|
359
|
+
return pages.filter(page => ranges.some(range => page.endRowIndex > range.start && page.firstRowIndex < range.end)).length;
|
|
360
|
+
}
|
|
361
|
+
/** Sorts and merges overlapping or touching byte ranges without overfetch gaps. */
|
|
362
|
+
function mergeByteRanges(ranges) {
|
|
363
|
+
const sorted = [...ranges].sort((left, right) => left.offset - right.offset);
|
|
364
|
+
const merged = [];
|
|
365
|
+
for (const range of sorted) {
|
|
366
|
+
const previous = merged[merged.length - 1];
|
|
367
|
+
const end = range.offset + range.length;
|
|
368
|
+
if (previous && range.offset <= previous.offset + previous.length) {
|
|
369
|
+
previous.length = Math.max(previous.offset + previous.length, end) - previous.offset;
|
|
370
|
+
}
|
|
371
|
+
else {
|
|
372
|
+
merged.push({ ...range });
|
|
373
|
+
}
|
|
374
|
+
}
|
|
375
|
+
return merged;
|
|
376
|
+
}
|
|
377
|
+
//# sourceMappingURL=parquet-page-index.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"parquet-page-index.js","sourceRoot":"","sources":["../../src/lib/parquet-page-index.ts"],"names":[],"mappings":"AAAA,aAAa;AACb,+BAA+B;AAC/B,oCAAoC;AAUpC,OAAO,EACL,WAAW,EACX,QAAQ,EACR,WAAW,EAGZ,6CAA0C;AAG3C,OAAO,KAAK,KAAK,qCAAkC;AACnD,OAAO,EACL,cAAc,EACd,YAAY,EACZ,WAAW,EACX,WAAW,EACX,WAAW,EACX,YAAY,EACZ,YAAY,EACZ,YAAY,EACb,2CAAwC;AACzC,OAAO,EAAC,yBAAyB,EAAC,2DAAwD;AAC1F,OAAO,EAAC,mBAAmB,EAAC,oDAAiD;AAC7E,OAAO,EAAC,yBAAyB,EAAE,0BAA0B,EAAC,+BAA4B;AAsD1F;;;;;GAKG;AACH,MAAM,CAAC,KAAK,UAAU,4BAA4B,CAChD,IAAkB,EAClB,QAAkB,EAClB,MAAqB,EACrB,mBAAwC,EACxC,SAA2B,EAC3B,MAAoB;IAEpB,MAAM,QAAQ,GAAG,MAAM,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC;IAC3C,MAAM,oBAAoB,GAAG,uBAAuB,CAAC,QAAQ,EAAE,mBAAmB,CAAC,CAAC;IACpF,IAAI,CAAC,uBAAuB,CAAC,MAAM,EAAE,oBAAoB,CAAC,IAAI,QAAQ,IAAI,CAAC,EAAE,CAAC;QAC5E,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,MAAM,aAAa,GAAyB,EAAE,CAAC;IAC/C,MAAM,cAAc,GAAgD,EAAE,CAAC;IACvE,MAAM,gBAAgB,GAAG,IAAI,GAAG,CAAC,0BAA0B,CAAC,SAAS,CAAC,CAAC,CAAC;IACxE,IAAI,UAAU,GAAG,CAAC,CAAC;IAEnB,MAAM,OAAO,CAAC,GAAG,CACf,oBAAoB,CAAC,GAAG,CAAC,KAAK,EAAC,WAAW,EAAC,EAAE;QAC3C,MAAM,IAAI,GAAG,WAAW,CAAC,SAAU,CAAC,cAAc,CAAC;QACnD,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QAC/B,MAAM,gBAAgB,GAAG,oBAAoB,CAC3C,WAAW,CAAC,mBAAmB,EAC/B,WAAW,CAAC,mBAAmB,EAC/B,IAAI,CAAC,IAAI,CACV,CAAC;QACF,IAAI,CAAC,gBAAgB,EAAE,CAAC;YACtB,OAAO;QACT,CAAC;QACD,MAAM,gBAAgB,GAAG,MAAM,IAAI,CAAC,IAAI,CACtC,gBAAgB,CAAC,MAAM,EACvB,gBAAgB,CAAC,MAAM,EACvB,MAAM,CACP,CAAC;QACF,IAAI,KAAgC,CAAC;QACrC,IAAI,CAAC;YACH,KAAK,GAAG,iBAAiB,CAAC,YAAY,CAAC,gBAAgB,CAAC,EAAE,QAAQ,CAAC,CAAC;QACtE,CAAC;QAAC,MAAM,CAAC;YACP,OAAO;QACT,CAAC;QACD,aAAa,CAAC,OAAO,CAAC,GAAG,KAAK,CAAC;QAC/B,UAAU,EAAE,CAAC;QAEb,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,IAAI,CAAC,gBAAgB,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;YACxD,OAAO;QACT,CAAC;QACD,MAAM,gBAAgB,GAAG,oBAAoB,CAC3C,WAAW,CAAC,mBAAmB,EAC/B,WAAW,CAAC,mBAAmB,EAC/B,IAAI,CAAC,IAAI,CACV,CAAC;QACF,IAAI,CAAC,gBAAgB,EAAE,CAAC;YACtB,OAAO;QACT,CAAC;QACD,MAAM,gBAAgB,GAAG,MAAM,IAAI,CAAC,IAAI,CACtC,gBAAgB,CAAC,MAAM,EACvB,gBAAgB,CAAC,MAAM,EACvB,MAAM,CACP,CAAC;QACF,MAAM,KAAK,GAAG,MAAM,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC;QACrC,IAAI,CAAC;YACH,cAAc,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,GAAG;gBACxB,KAAK;gBACL,UAAU,EAAE,iBAAiB,CAAC,YAAY,CAAC,gBAAgB,CAAC,EAAE,KAAK,EAAE,KAAK,CAAC;aAC5E,CAAC;QACJ,CAAC;QAAC,MAAM,CAAC;YACP,OAAO;QACT,CAAC;QACD,UAAU,EAAE,CAAC;IACf,CAAC,CAAC,CACH,CAAC;IAEF,IACE,oBAAoB,CAAC,IAAI,CACvB,WAAW,CAAC,EAAE,CAAC,CAAC,aAAa,CAAC,WAAW,CAAC,SAAU,CAAC,cAAc,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAC/E,EACD,CAAC;QACD,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,MAAM,eAAe,GAAG,qBAAqB,CAAC,SAAS,EAAE,cAAc,EAAE,QAAQ,CAAC,CAAC;IACnF,IAAI,eAAe,KAAK,SAAS,EAAE,CAAC;QAClC,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,MAAM,SAAS,GAAG,uBAAuB,CAAC,eAAe,EAAE,aAAa,EAAE,QAAQ,CAAC,CAAC;IACpF,MAAM,cAAc,GAAG,QAAQ,GAAG,gBAAgB,CAAC,SAAS,CAAC,CAAC;IAC9D,IAAI,cAAc,IAAI,CAAC,EAAE,CAAC;QACxB,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,MAAM,cAAc,GAAG,MAAM,CAAC,MAAM,CAAC,aAAa,CAAC,CAAC,MAAM,CACxD,CAAC,KAAK,EAAE,KAAK,EAAE,EAAE,CAAC,KAAK,GAAG,KAAK,CAAC,MAAM,EACtC,CAAC,CACF,CAAC;IACF,MAAM,iBAAiB,GAAG,MAAM,CAAC,MAAM,CAAC,aAAa,CAAC,CAAC,MAAM,CAC3D,CAAC,KAAK,EAAE,KAAK,EAAE,EAAE,CAAC,KAAK,GAAG,kBAAkB,CAAC,KAAK,EAAE,SAAS,CAAC,EAC9D,CAAC,CACF,CAAC;IACF,OAAO;QACL,SAAS;QACT,aAAa;QACb,UAAU;QACV,cAAc;QACd,iBAAiB;QACjB,cAAc;KACf,CAAC;AACJ,CAAC;AAED,iFAAiF;AACjF,MAAM,UAAU,wBAAwB,CACtC,QAAkB,EAClB,mBAAwC,EACxC,IAA4B;IAE5B,MAAM,MAAM,GAA4C,EAAE,CAAC;IAC3D,KAAK,MAAM,WAAW,IAAI,uBAAuB,CAAC,QAAQ,EAAE,mBAAmB,CAAC,EAAE,CAAC;QACjF,MAAM,cAAc,GAAG,WAAW,CAAC,SAAU,CAAC;QAC9C,MAAM,KAAK,GAAG,IAAI,CAAC,aAAa,CAAC,cAAc,CAAC,cAAc,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;QAC1E,MAAM,aAAa,GAAG,KAAK,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,CACxC,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC,IAAI,CAAC,WAAW,GAAG,KAAK,CAAC,KAAK,IAAI,IAAI,CAAC,aAAa,GAAG,KAAK,CAAC,GAAG,CAAC,CAC/F,CAAC;QACF,IAAI,CAAC,aAAa,CAAC,MAAM,EAAE,CAAC;YAC1B,SAAS;QACX,CAAC;QACD,MAAM,oBAAoB,GAAG,MAAM,CAAC,cAAc,CAAC,sBAAsB,CAAC,CAAC;QAC3E,IAAI,MAAM,CAAC,aAAa,CAAC,oBAAoB,CAAC,IAAI,oBAAoB,GAAG,CAAC,EAAE,CAAC;YAC3E,MAAM,CAAC,IAAI,CAAC;gBACV,MAAM,EAAE,oBAAoB;gBAC5B,MAAM,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM,GAAG,oBAAoB,CAAC;aAC5D,CAAC,CAAC;QACL,CAAC;QACD,KAAK,MAAM,QAAQ,IAAI,IAAI,CAAC,SAAS,EAAE,CAAC;YACtC,MAAM,gBAAgB,GAAG,KAAK,CAAC,MAAM,CACnC,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,WAAW,GAAG,QAAQ,CAAC,KAAK,IAAI,IAAI,CAAC,aAAa,GAAG,QAAQ,CAAC,GAAG,CAC/E,CAAC;YACF,IAAI,CAAC,gBAAgB,CAAC,MAAM,EAAE,CAAC;gBAC7B,SAAS;YACX,CAAC;YACD,MAAM,SAAS,GAAG,gBAAgB,CAAC,CAAC,CAAC,CAAC;YACtC,MAAM,QAAQ,GAAG,gBAAgB,CAAC,gBAAgB,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;YAC/D,MAAM,CAAC,IAAI,CAAC;gBACV,MAAM,EAAE,SAAS,CAAC,MAAM;gBACxB,MAAM,EAAE,QAAQ,CAAC,MAAM,GAAG,QAAQ,CAAC,oBAAoB,GAAG,SAAS,CAAC,MAAM;aAC3E,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IACD,OAAO,eAAe,CAAC,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC;AACnE,CAAC;AAED,yDAAyD;AACzD,MAAM,UAAU,iBAAiB,CAAC,KAAiB,EAAE,QAAgB;IACnE,MAAM,QAAQ,GAAG,IAAI,yBAAyB,CAAC,IAAI,mBAAmB,CAAC,KAAK,CAAC,CAAC,CAAC;IAC/E,MAAM,WAAW,GAAG,WAAW,CAAC,IAAI,CAAC,QAAe,CAAC,CAAC;IACtD,IACE,WAAW,CAAC,cAAc,CAAC,MAAM,KAAK,CAAC;QACvC,MAAM,CAAC,WAAW,CAAC,cAAc,CAAC,CAAC,CAAC,CAAC,eAAe,CAAC,KAAK,CAAC,EAC3D,CAAC;QACD,MAAM,IAAI,KAAK,CAAC,6CAA6C,CAAC,CAAC;IACjE,CAAC;IACD,OAAO,WAAW,CAAC,cAAc,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,KAAK,EAAE,SAAS,EAAE,EAAE;QACnE,MAAM,aAAa,GAAG,MAAM,CAAC,QAAQ,CAAC,eAAe,CAAC,CAAC;QACvD,MAAM,WAAW,GACf,KAAK,GAAG,CAAC,GAAG,SAAS,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC;QACzF,MAAM,MAAM,GAAG,MAAM,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;QACvC,MAAM,oBAAoB,GAAG,QAAQ,CAAC,oBAAoB,CAAC;QAC3D,IACE,CAAC,MAAM,CAAC,aAAa,CAAC,MAAM,CAAC;YAC7B,CAAC,MAAM,CAAC,aAAa,CAAC,oBAAoB,CAAC;YAC3C,CAAC,MAAM,CAAC,aAAa,CAAC,aAAa,CAAC;YACpC,CAAC,MAAM,CAAC,aAAa,CAAC,WAAW,CAAC;YAClC,MAAM,GAAG,CAAC;YACV,oBAAoB,IAAI,CAAC;YACzB,aAAa,GAAG,CAAC;YACjB,WAAW,IAAI,aAAa;YAC5B,WAAW,GAAG,QAAQ,EACtB,CAAC;YACD,MAAM,IAAI,KAAK,CAAC,4CAA4C,CAAC,CAAC;QAChE,CAAC;QACD,OAAO,EAAC,MAAM,EAAE,oBAAoB,EAAE,aAAa,EAAE,WAAW,EAAC,CAAC;IACpE,CAAC,CAAC,CAAC;AACL,CAAC;AAED,6EAA6E;AAC7E,SAAS,qBAAqB,CAC5B,SAA2B,EAC3B,OAAoD,EACpD,QAAgB;IAEhB,IAAI,SAAS,CAAC,EAAE,KAAK,KAAK,IAAI,SAAS,CAAC,EAAE,KAAK,IAAI,EAAE,CAAC;QACpD,IAAI,SAAS,CAAC,EAAE,KAAK,KAAK,EAAE,CAAC;YAC3B,IAAI,MAAqC,CAAC;YAC1C,KAAK,MAAM,KAAK,IAAI,SAAS,CAAC,IAAI,EAAE,CAAC;gBACnC,MAAM,GAAG,kBAAkB,CAAC,MAAM,EAAE,qBAAqB,CAAC,KAAK,EAAE,OAAO,EAAE,QAAQ,CAAC,CAAC,CAAC;YACvF,CAAC;YACD,OAAO,MAAM,CAAC;QAChB,CAAC;QACD,IAAI,MAAM,GAAsB,EAAE,CAAC;QACnC,KAAK,MAAM,KAAK,IAAI,SAAS,CAAC,IAAI,EAAE,CAAC;YACnC,MAAM,WAAW,GAAG,qBAAqB,CAAC,KAAK,EAAE,OAAO,EAAE,QAAQ,CAAC,CAAC;YACpE,IAAI,WAAW,KAAK,SAAS,EAAE,CAAC;gBAC9B,OAAO,SAAS,CAAC;YACnB,CAAC;YACD,MAAM,GAAG,cAAc,CAAC,MAAM,EAAE,WAAW,CAAC,CAAC;QAC/C,CAAC;QACD,OAAO,MAAM,CAAC;IAChB,CAAC;IACD,IAAI,SAAS,CAAC,EAAE,KAAK,KAAK,EAAE,CAAC;QAC3B,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,MAAM,aAAa,GAAG,SAGE,CAAC;IACzB,MAAM,MAAM,GAAG,OAAO,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC;IACvD,IAAI,CAAC,MAAM,EAAE,CAAC;QACZ,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,MAAM,MAAM,GAAsB,EAAE,CAAC;IACrC,KAAK,IAAI,SAAS,GAAG,CAAC,EAAE,SAAS,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,EAAE,SAAS,EAAE,EAAE,CAAC;QACrE,MAAM,IAAI,GAAG,MAAM,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC;QACrC,IACE,yBAAyB,CACvB,aAAa,EACb,MAAM,CAAC,UAAU,CAAC,SAAS,CAAC,EAC5B,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,aAAa,CACtC,EACD,CAAC;YACD,MAAM,CAAC,IAAI,CAAC,EAAC,KAAK,EAAE,IAAI,CAAC,aAAa,EAAE,GAAG,EAAE,IAAI,CAAC,WAAW,EAAC,CAAC,CAAC;QAClE,CAAC;IACH,CAAC;IACD,OAAO,cAAc,CAAC,MAAM,CAAC,CAAC;AAChC,CAAC;AAED,4FAA4F;AAC5F,SAAS,iBAAiB,CACxB,KAAiB,EACjB,KAAyC,EACzC,KAAmB;IAEnB,MAAM,QAAQ,GAAG,IAAI,yBAAyB,CAAC,IAAI,mBAAmB,CAAC,KAAK,CAAC,CAAC,CAAC;IAC/E,MAAM,WAAW,GAAG,WAAW,CAAC,IAAI,CAAC,QAAe,CAAC,CAAC;IACtD,IACE,WAAW,CAAC,UAAU,CAAC,MAAM,KAAK,KAAK,CAAC,MAAM;QAC9C,WAAW,CAAC,UAAU,CAAC,MAAM,KAAK,KAAK,CAAC,MAAM;QAC9C,WAAW,CAAC,UAAU,CAAC,MAAM,KAAK,KAAK,CAAC,MAAM;QAC9C,CAAC,WAAW,CAAC,WAAW,KAAK,SAAS,IAAI,WAAW,CAAC,WAAW,CAAC,MAAM,KAAK,KAAK,CAAC,MAAM,CAAC,EAC1F,CAAC;QACD,MAAM,IAAI,KAAK,CAAC,iEAAiE,CAAC,CAAC;IACrF,CAAC;IACD,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,SAAS,EAAE,EAAE,CAAC,CAAC;QACtC,GAAG,EAAE,WAAW,CAAC,UAAU,CAAC,SAAS,CAAC;YACpC,CAAC,CAAC,SAAS;YACX,CAAC,CAAC,gCAAgC,CAAC,WAAW,CAAC,UAAU,CAAC,SAAS,CAAC,EAAE,KAAK,CAAC;QAC9E,GAAG,EAAE,WAAW,CAAC,UAAU,CAAC,SAAS,CAAC;YACpC,CAAC,CAAC,SAAS;YACX,CAAC,CAAC,gCAAgC,CAAC,WAAW,CAAC,UAAU,CAAC,SAAS,CAAC,EAAE,KAAK,CAAC;QAC9E,SAAS,EACP,WAAW,CAAC,WAAW,EAAE,CAAC,SAAS,CAAC,KAAK,SAAS;YAChD,CAAC,CAAC,MAAM,CAAC,WAAW,CAAC,WAAW,CAAC,SAAS,CAAC,CAAC;YAC5C,CAAC,CAAC,WAAW,CAAC,UAAU,CAAC,SAAS,CAAC;gBACjC,CAAC,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC,WAAW,GAAG,KAAK,CAAC,SAAS,CAAC,CAAC,aAAa;gBAC/D,CAAC,CAAC,SAAS;QACjB,UAAU,EAAE,IAAI;QAChB,UAAU,EAAE,IAAI;KACjB,CAAC,CAAC,CAAC;AACN,CAAC;AAED,4FAA4F;AAC5F,MAAM,UAAU,gCAAgC,CAAC,KAAiB,EAAE,KAAmB;IACrF,IAAI,cAAuB,CAAC;IAC5B,QAAQ,KAAK,CAAC,aAAa,EAAE,CAAC;QAC5B,KAAK,SAAS;YACZ,cAAc,GAAG,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;YACnC,MAAM;QACR,KAAK,OAAO;YACV,cAAc;gBACZ,KAAK,CAAC,YAAY,KAAK,SAAS,CAAC,CAAC,CAAC,YAAY,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,WAAW,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC;YACpF,MAAM;QACR,KAAK,OAAO;YACV,cAAc;gBACZ,KAAK,CAAC,YAAY,KAAK,SAAS,CAAC,CAAC,CAAC,YAAY,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,WAAW,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC;YACpF,MAAM;QACR,KAAK,OAAO;YACV,cAAc,GAAG,WAAW,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC;YACvC,MAAM;QACR,KAAK,QAAQ;YACX,cAAc,GAAG,YAAY,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC;YACxC,MAAM;QACR,KAAK,OAAO,CAAC;QACb,KAAK,YAAY,CAAC;QAClB,KAAK,sBAAsB;YACzB,cAAc,GAAG,cAAc,CAAC,KAAK,CAAC,CAAC;YACvC,MAAM;QACR;YACE,OAAO,SAAS,CAAC;IACrB,CAAC;IACD,OAAO,KAAK,CAAC,aAAa,CAAC,KAAK,CAAC,YAAY,IAAI,KAAK,CAAC,aAAa,EAAE,cAAc,EAAE,KAAK,CAAC,CAAC;AAC/F,CAAC;AAED,6FAA6F;AAC7F,SAAS,uBAAuB,CAC9B,QAAkB,EAClB,mBAAwC;IAExC,OAAO,QAAQ,CAAC,OAAO,CAAC,MAAM,CAAC,WAAW,CAAC,EAAE;QAC3C,MAAM,IAAI,GAAG,WAAW,CAAC,SAAS,EAAE,cAAc,CAAC;QACnD,OAAO,OAAO,CACZ,IAAI;YACF,CAAC,mBAAmB,CAAC,MAAM,KAAK,CAAC;gBAC/B,mBAAmB,CAAC,IAAI,CAAC,YAAY,CAAC,EAAE,CAAC,YAAY,CAAC,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAC3E,CAAC;IACJ,CAAC,CAAC,CAAC;AACL,CAAC;AAED,6FAA6F;AAC7F,SAAS,uBAAuB,CAC9B,MAAqB,EACrB,YAAoC;IAEpC,OAAO,CACL,YAAY,CAAC,MAAM,GAAG,CAAC;QACvB,YAAY,CAAC,KAAK,CAAC,WAAW,CAAC,EAAE;YAC/B,MAAM,IAAI,GAAG,WAAW,CAAC,SAAS,EAAE,cAAc,CAAC;YACnD,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;gBAC/B,OAAO,KAAK,CAAC;YACf,CAAC;YACD,MAAM,KAAK,GAAG,MAAM,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC;YACrC,MAAM,iBAAiB,GAAG,WAAW,CAAC,SAAU,CAAC,SAAS,CAAC,IAAI,CAC7D,QAAQ,CAAC,EAAE,CAAC,QAAQ,KAAK,QAAQ,CAAC,gBAAgB,IAAI,QAAQ,KAAK,QAAQ,CAAC,cAAc,CAC3F,CAAC;YACF,MAAM,oBAAoB,GAAG,MAAM,CAAC,WAAW,CAAC,SAAU,CAAC,sBAAsB,CAAC,CAAC;YACnF,OAAO,CACL,KAAK,CAAC,IAAI,CAAC,MAAM,KAAK,CAAC;gBACvB,KAAK,CAAC,cAAc,KAAK,UAAU;gBACnC,KAAK,CAAC,SAAS,KAAK,CAAC;gBACrB,CAAC,CAAC,iBAAiB;oBACjB,CAAC,MAAM,CAAC,aAAa,CAAC,oBAAoB,CAAC,IAAI,oBAAoB,GAAG,CAAC,CAAC,CAAC,CAC5E,CAAC;QACJ,CAAC,CAAC,CACH,CAAC;AACJ,CAAC;AAED,kFAAkF;AAClF,MAAM,UAAU,oBAAoB,CAClC,WAA+C,EAC/C,WAA+B,EAC/B,QAAgB;IAEhB,IAAI,WAAW,KAAK,SAAS,IAAI,WAAW,KAAK,SAAS,EAAE,CAAC;QAC3D,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,MAAM,MAAM,GAAG,MAAM,CAAC,WAAW,CAAC,CAAC;IACnC,IACE,CAAC,MAAM,CAAC,aAAa,CAAC,MAAM,CAAC;QAC7B,CAAC,MAAM,CAAC,aAAa,CAAC,WAAW,CAAC;QAClC,CAAC,MAAM,CAAC,aAAa,CAAC,QAAQ,CAAC;QAC/B,MAAM,GAAG,CAAC;QACV,WAAW,IAAI,CAAC;QAChB,QAAQ,GAAG,CAAC;QACZ,MAAM,GAAG,QAAQ,GAAG,WAAW,EAC/B,CAAC;QACD,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,OAAO,EAAC,MAAM,EAAE,MAAM,EAAE,WAAW,EAAC,CAAC;AACvC,CAAC;AAED,mGAAmG;AACnG,SAAS,uBAAuB,CAC9B,MAAkC,EAClC,aAAmC,EACnC,QAAgB;IAEhB,MAAM,aAAa,GAAG,MAAM,CAAC,MAAM,CAAC,aAAa,CAAC,CAAC;IACnD,MAAM,QAAQ,GAAG,MAAM,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE;QAClC,IAAI,OAAO,GAAG,EAAC,GAAG,KAAK,EAAC,CAAC;QACzB,IAAI,OAAO,GAAG,IAAI,CAAC;QACnB,OAAO,OAAO,EAAE,CAAC;YACf,OAAO,GAAG,KAAK,CAAC;YAChB,KAAK,MAAM,KAAK,IAAI,aAAa,EAAE,CAAC;gBAClC,MAAM,WAAW,GAAG,KAAK,CAAC,MAAM,CAC9B,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,WAAW,GAAG,OAAO,CAAC,KAAK,IAAI,IAAI,CAAC,aAAa,GAAG,OAAO,CAAC,GAAG,CAC7E,CAAC;gBACF,IAAI,CAAC,WAAW,CAAC,MAAM,EAAE,CAAC;oBACxB,SAAS;gBACX,CAAC;gBACD,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,KAAK,EAAE,WAAW,CAAC,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC;gBACpE,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,GAAG,EAAE,WAAW,CAAC,WAAW,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC;gBACnF,IAAI,KAAK,KAAK,OAAO,CAAC,KAAK,IAAI,GAAG,KAAK,OAAO,CAAC,GAAG,EAAE,CAAC;oBACnD,OAAO,GAAG,EAAC,KAAK,EAAE,GAAG,EAAC,CAAC;oBACvB,OAAO,GAAG,IAAI,CAAC;gBACjB,CAAC;YACH,CAAC;QACH,CAAC;QACD,OAAO,EAAC,KAAK,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,OAAO,CAAC,KAAK,CAAC,EAAE,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,QAAQ,EAAE,OAAO,CAAC,GAAG,CAAC,EAAC,CAAC;IACnF,CAAC,CAAC,CAAC;IACH,OAAO,cAAc,CAAC,QAAQ,CAAC,CAAC;AAClC,CAAC;AAED,sEAAsE;AACtE,SAAS,kBAAkB,CACzB,IAAmC,EACnC,KAAoC;IAEpC,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;QACvB,OAAO,KAAK,CAAC;IACf,CAAC;IACD,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;QACxB,OAAO,IAAI,CAAC;IACd,CAAC;IACD,MAAM,MAAM,GAAsB,EAAE,CAAC;IACrC,IAAI,SAAS,GAAG,CAAC,CAAC;IAClB,IAAI,UAAU,GAAG,CAAC,CAAC;IACnB,OAAO,SAAS,GAAG,IAAI,CAAC,MAAM,IAAI,UAAU,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC;QAC5D,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,KAAK,EAAE,KAAK,CAAC,UAAU,CAAC,CAAC,KAAK,CAAC,CAAC;QACvE,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,GAAG,EAAE,KAAK,CAAC,UAAU,CAAC,CAAC,GAAG,CAAC,CAAC;QACjE,IAAI,KAAK,GAAG,GAAG,EAAE,CAAC;YAChB,MAAM,CAAC,IAAI,CAAC,EAAC,KAAK,EAAE,GAAG,EAAC,CAAC,CAAC;QAC5B,CAAC;QACD,IAAI,IAAI,CAAC,SAAS,CAAC,CAAC,GAAG,GAAG,KAAK,CAAC,UAAU,CAAC,CAAC,GAAG,EAAE,CAAC;YAChD,SAAS,EAAE,CAAC;QACd,CAAC;aAAM,CAAC;YACN,UAAU,EAAE,CAAC;QACf,CAAC;IACH,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,wCAAwC;AACxC,SAAS,cAAc,CACrB,IAAgC,EAChC,KAAiC;IAEjC,OAAO,cAAc,CAAC,CAAC,GAAG,IAAI,EAAE,GAAG,KAAK,CAAC,CAAC,CAAC;AAC7C,CAAC;AAED,mEAAmE;AACnE,SAAS,cAAc,CAAC,MAAkC;IACxD,MAAM,MAAM,GAAG,CAAC,GAAG,MAAM,CAAC,CAAC,IAAI,CAC7B,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC,KAAK,IAAI,IAAI,CAAC,GAAG,GAAG,KAAK,CAAC,GAAG,CAClE,CAAC;IACF,MAAM,MAAM,GAAsB,EAAE,CAAC;IACrC,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC3B,MAAM,QAAQ,GAAG,MAAM,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QAC3C,IAAI,QAAQ,IAAI,KAAK,CAAC,KAAK,IAAI,QAAQ,CAAC,GAAG,EAAE,CAAC;YAC5C,QAAQ,CAAC,GAAG,GAAG,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,GAAG,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC;QACnD,CAAC;aAAM,CAAC;YACN,MAAM,CAAC,IAAI,CAAC,EAAC,GAAG,KAAK,EAAC,CAAC,CAAC;QAC1B,CAAC;IACH,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,0DAA0D;AAC1D,SAAS,gBAAgB,CAAC,MAAkC;IAC1D,OAAO,MAAM,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,KAAK,EAAE,EAAE,CAAC,KAAK,GAAG,KAAK,CAAC,GAAG,GAAG,KAAK,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC;AAC7E,CAAC;AAED,uEAAuE;AACvE,SAAS,kBAAkB,CACzB,KAAyC,EACzC,MAAkC;IAElC,OAAO,KAAK,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,CACzB,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC,IAAI,CAAC,WAAW,GAAG,KAAK,CAAC,KAAK,IAAI,IAAI,CAAC,aAAa,GAAG,KAAK,CAAC,GAAG,CAAC,CACvF,CAAC,MAAM,CAAC;AACX,CAAC;AAED,mFAAmF;AACnF,SAAS,eAAe,CACtB,MAAmD;IAEnD,MAAM,MAAM,GAAG,CAAC,GAAG,MAAM,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC;IAC7E,MAAM,MAAM,GAA4C,EAAE,CAAC;IAC3D,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC3B,MAAM,QAAQ,GAAG,MAAM,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QAC3C,MAAM,GAAG,GAAG,KAAK,CAAC,MAAM,GAAG,KAAK,CAAC,MAAM,CAAC;QACxC,IAAI,QAAQ,IAAI,KAAK,CAAC,MAAM,IAAI,QAAQ,CAAC,MAAM,GAAG,QAAQ,CAAC,MAAM,EAAE,CAAC;YAClE,QAAQ,CAAC,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,MAAM,GAAG,QAAQ,CAAC,MAAM,EAAE,GAAG,CAAC,GAAG,QAAQ,CAAC,MAAM,CAAC;QACvF,CAAC;aAAM,CAAC;YACN,MAAM,CAAC,IAAI,CAAC,EAAC,GAAG,KAAK,EAAC,CAAC,CAAC;QAC1B,CAAC;IACH,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC"}
|
|
@@ -0,0 +1,17 @@
|
|
|
1
|
+
import type { ArrayType } from '@loaders.gl/schema';
|
|
2
|
+
import type { ParquetColumnChunkStatistics, ParquetComparisonPredicate, ParquetInPredicate, ParquetNullPredicate, ParquetPredicate, ParquetRowGroupMetadata } from "../parquet-source-types.js";
|
|
3
|
+
/** Returns the unique top-level columns referenced by a predicate. */
|
|
4
|
+
export declare function getParquetPredicateColumns(predicate: ParquetPredicate): string[];
|
|
5
|
+
/** Copies a predicate tree and mutable scalar values for read-scoped option snapshots. */
|
|
6
|
+
export declare function copyParquetPredicate(predicate: ParquetPredicate): ParquetPredicate;
|
|
7
|
+
/** Validates a predicate and every referenced top-level column. */
|
|
8
|
+
export declare function validateParquetPredicate(predicate: ParquetPredicate, availableColumns: ReadonlySet<string>): void;
|
|
9
|
+
/** Conservatively determines whether a row group can contain a matching row. */
|
|
10
|
+
export declare function canParquetRowGroupMatch(predicate: ParquetPredicate, rowGroup: ParquetRowGroupMetadata): boolean;
|
|
11
|
+
/** Conservatively determines whether one predicate leaf can match supplied statistics. */
|
|
12
|
+
export declare function canParquetStatisticsMatch(predicate: ParquetComparisonPredicate | ParquetInPredicate | ParquetNullPredicate, statistics: ParquetColumnChunkStatistics, rowCount: number): boolean;
|
|
13
|
+
/** Returns exact source row indexes matching a predicate. */
|
|
14
|
+
export declare function filterParquetRowIndices(predicate: ParquetPredicate | undefined, columns: Record<string, ArrayType>, rowCount: number): number[];
|
|
15
|
+
/** Gathers selected source rows without constructing row objects. */
|
|
16
|
+
export declare function gatherParquetColumns(columns: Record<string, ArrayType>, rowIndices: readonly number[], selectedColumns?: ReadonlySet<string>): Record<string, ArrayType>;
|
|
17
|
+
//# sourceMappingURL=parquet-predicate.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"parquet-predicate.d.ts","sourceRoot":"","sources":["../../src/lib/parquet-predicate.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,EAAC,SAAS,EAAC,MAAM,oBAAoB,CAAC;AAElD,OAAO,KAAK,EACV,4BAA4B,EAC5B,0BAA0B,EAC1B,kBAAkB,EAGlB,oBAAoB,EACpB,gBAAgB,EAEhB,uBAAuB,EACxB,mCAAgC;AAGjC,sEAAsE;AACtE,wBAAgB,0BAA0B,CAAC,SAAS,EAAE,gBAAgB,GAAG,MAAM,EAAE,CAShF;AAED,0FAA0F;AAC1F,wBAAgB,oBAAoB,CAAC,SAAS,EAAE,gBAAgB,GAAG,gBAAgB,CA2BlF;AAED,mEAAmE;AACnE,wBAAgB,wBAAwB,CACtC,SAAS,EAAE,gBAAgB,EAC3B,gBAAgB,EAAE,WAAW,CAAC,MAAM,CAAC,GACpC,IAAI,CAmBN;AAED,gFAAgF;AAChF,wBAAgB,uBAAuB,CACrC,SAAS,EAAE,gBAAgB,EAC3B,QAAQ,EAAE,uBAAuB,GAChC,OAAO,CAqBT;AAED,0FAA0F;AAC1F,wBAAgB,yBAAyB,CACvC,SAAS,EAAE,0BAA0B,GAAG,kBAAkB,GAAG,oBAAoB,EACjF,UAAU,EAAE,4BAA4B,EACxC,QAAQ,EAAE,MAAM,GACf,OAAO,CAaT;AAED,6DAA6D;AAC7D,wBAAgB,uBAAuB,CACrC,SAAS,EAAE,gBAAgB,GAAG,SAAS,EACvC,OAAO,EAAE,MAAM,CAAC,MAAM,EAAE,SAAS,CAAC,EAClC,QAAQ,EAAE,MAAM,GACf,MAAM,EAAE,CAWV;AAED,qEAAqE;AACrE,wBAAgB,oBAAoB,CAClC,OAAO,EAAE,MAAM,CAAC,MAAM,EAAE,SAAS,CAAC,EAClC,UAAU,EAAE,SAAS,MAAM,EAAE,EAC7B,eAAe,CAAC,EAAE,WAAW,CAAC,MAAM,CAAC,GACpC,MAAM,CAAC,MAAM,EAAE,SAAS,CAAC,CAS3B"}
|