@loaders.gl/parquet 4.5.3 → 5.0.0-alpha.10
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/delta-format.d.ts +9 -0
- package/dist/delta-format.d.ts.map +1 -0
- package/dist/delta-format.js +12 -0
- package/dist/delta-format.js.map +1 -0
- package/dist/delta-source-loader-types.d.ts +27 -0
- package/dist/delta-source-loader-types.d.ts.map +1 -0
- package/dist/delta-source-loader-types.js +28 -0
- package/dist/delta-source-loader-types.js.map +1 -0
- package/dist/delta-source.d.ts +48 -0
- package/dist/delta-source.d.ts.map +1 -0
- package/dist/delta-source.js +222 -0
- package/dist/delta-source.js.map +1 -0
- package/dist/delta-types.d.ts +38 -0
- package/dist/delta-types.d.ts.map +1 -0
- package/dist/delta-types.js +5 -0
- package/dist/delta-types.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/iceberg-rest-catalog.d.ts +49 -0
- package/dist/iceberg-rest-catalog.d.ts.map +1 -0
- package/dist/iceberg-rest-catalog.js +78 -0
- package/dist/iceberg-rest-catalog.js.map +1 -0
- package/dist/iceberg-table-source.d.ts +63 -0
- package/dist/iceberg-table-source.d.ts.map +1 -0
- package/dist/iceberg-table-source.js +797 -0
- package/dist/iceberg-table-source.js.map +1 -0
- package/dist/iceberg-types.d.ts +150 -0
- package/dist/iceberg-types.d.ts.map +1 -0
- package/dist/iceberg-types.js +5 -0
- package/dist/iceberg-types.js.map +1 -0
- package/dist/index.cjs +41568 -6781
- package/dist/index.cjs.map +3 -4
- package/dist/index.d.ts +15 -9
- package/dist/index.d.ts.map +1 -1
- package/dist/index.js +11 -12
- 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 +3 -1
- package/dist/lib/arrow/convert-schema-from-parquet.d.ts.map +1 -1
- package/dist/lib/arrow/convert-schema-from-parquet.js +164 -29
- 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 +485 -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 +394 -0
- package/dist/lib/encoders/encode-table-to-parquet-js.js.map +1 -0
- package/dist/lib/geo/geoparquet-covering.d.ts +19 -0
- package/dist/lib/geo/geoparquet-covering.d.ts.map +1 -0
- package/dist/lib/geo/geoparquet-covering.js +137 -0
- package/dist/lib/geo/geoparquet-covering.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 +412 -0
- package/dist/lib/geo/geospatial-metadata.js.map +1 -0
- package/dist/lib/parquet-bloom-filter-planner.d.ts +16 -0
- package/dist/lib/parquet-bloom-filter-planner.d.ts.map +1 -0
- package/dist/lib/parquet-bloom-filter-planner.js +57 -0
- package/dist/lib/parquet-bloom-filter-planner.js.map +1 -0
- package/dist/lib/parquet-bloom-filter.d.ts +30 -0
- package/dist/lib/parquet-bloom-filter.d.ts.map +1 -0
- package/dist/lib/parquet-bloom-filter.js +289 -0
- package/dist/lib/parquet-bloom-filter.js.map +1 -0
- package/dist/lib/parquet-encryption.d.ts +80 -0
- package/dist/lib/parquet-encryption.d.ts.map +1 -0
- package/dist/lib/parquet-encryption.js +190 -0
- package/dist/lib/parquet-encryption.js.map +1 -0
- package/dist/lib/parquet-page-index.d.ts +91 -0
- package/dist/lib/parquet-page-index.d.ts.map +1 -0
- package/dist/lib/parquet-page-index.js +463 -0
- package/dist/lib/parquet-page-index.js.map +1 -0
- package/dist/lib/parquet-predicate.d.ts +21 -0
- package/dist/lib/parquet-predicate.d.ts.map +1 -0
- package/dist/lib/parquet-predicate.js +328 -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 +24 -0
- package/dist/lib/parquet-source-worker-client.js.map +1 -0
- package/dist/lib/parquet-source-worker-decoder.d.ts +7 -0
- package/dist/lib/parquet-source-worker-decoder.d.ts.map +1 -0
- package/dist/lib/parquet-source-worker-decoder.js +181 -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 +121 -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 +6 -2
- 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 +921 -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 +4 -5
- 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 +95 -28
- package/dist/lib/parsers/parse-parquet-to-json.js.map +1 -1
- package/dist/lib/slice-parquet-batch.d.ts +9 -0
- package/dist/lib/slice-parquet-batch.d.ts.map +1 -0
- package/dist/lib/slice-parquet-batch.js +28 -0
- package/dist/lib/slice-parquet-batch.js.map +1 -0
- 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 +267 -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 +25 -0
- package/dist/lib/utils/normalize-parquet-options.js.map +1 -0
- package/dist/parquet-dataset-source.cjs +38974 -0
- package/dist/parquet-dataset-source.cjs.map +6 -0
- package/dist/parquet-dataset-source.d.ts +81 -0
- package/dist/parquet-dataset-source.d.ts.map +1 -0
- package/dist/parquet-dataset-source.js +602 -0
- package/dist/parquet-dataset-source.js.map +1 -0
- package/dist/parquet-encoded-page-types.d.ts +108 -0
- package/dist/parquet-encoded-page-types.d.ts.map +1 -0
- package/dist/parquet-encoded-page-types.js +5 -0
- package/dist/parquet-encoded-page-types.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 +59 -0
- package/dist/parquet-js-loader-types.d.ts.map +1 -0
- package/dist/parquet-js-loader-types.js +34 -0
- package/dist/parquet-js-loader-types.js.map +1 -0
- package/dist/parquet-js-loader.d.ts +18 -22
- package/dist/parquet-js-loader.d.ts.map +1 -1
- package/dist/parquet-js-loader.js +53 -14
- package/dist/parquet-js-loader.js.map +1 -1
- package/dist/parquet-js-writer.d.ts +65 -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 +55 -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 +43 -0
- package/dist/parquet-source-capabilities.d.ts.map +1 -0
- package/dist/parquet-source-capabilities.js +32 -0
- package/dist/parquet-source-capabilities.js.map +1 -0
- package/dist/parquet-source-loader-types.d.ts +82 -0
- package/dist/parquet-source-loader-types.d.ts.map +1 -0
- package/dist/parquet-source-loader-types.js +84 -0
- package/dist/parquet-source-loader-types.js.map +1 -0
- package/dist/parquet-source-loader.cjs +38261 -0
- package/dist/parquet-source-loader.cjs.map +6 -0
- package/dist/parquet-source-loader.d.ts +167 -0
- package/dist/parquet-source-loader.d.ts.map +1 -0
- package/dist/parquet-source-loader.js +1438 -0
- package/dist/parquet-source-loader.js.map +1 -0
- package/dist/parquet-source-types.d.ts +534 -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-factory.d.ts +4 -0
- package/dist/parquet-source-worker-factory.d.ts.map +1 -0
- package/dist/parquet-source-worker-factory.js +6 -0
- package/dist/parquet-source-worker-factory.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 +38871 -0
- package/dist/parquet-source-worker.js.map +7 -0
- package/dist/parquet-worker-node.cjs +78507 -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 +19 -0
- package/dist/parquet-worker-url.js.map +1 -0
- package/dist/parquet-worker.js +20094 -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 +35 -3
- package/dist/parquetjs/codecs/declare.d.ts.map +1 -1
- package/dist/parquetjs/codecs/declare.js +17 -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 +788 -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 +22 -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 +231 -78
- package/dist/parquetjs/codecs/plain.js.map +1 -1
- package/dist/parquetjs/codecs/rle.d.ts +7 -3
- package/dist/parquetjs/codecs/rle.d.ts.map +1 -1
- package/dist/parquetjs/codecs/rle.js +613 -69
- package/dist/parquetjs/codecs/rle.js.map +1 -1
- package/dist/parquetjs/compression.d.ts +16 -8
- package/dist/parquetjs/compression.d.ts.map +1 -1
- package/dist/parquetjs/compression.js +114 -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 +67 -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 +54 -5
- package/dist/parquetjs/encoder/parquet-encoder.d.ts.map +1 -1
- package/dist/parquetjs/encoder/parquet-encoder.js +930 -66
- package/dist/parquetjs/encoder/parquet-encoder.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/BoundingBox.d.ts +29 -0
- package/dist/parquetjs/parquet-thrift/BoundingBox.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/BoundingBox.js +85 -0
- package/dist/parquetjs/parquet-thrift/BoundingBox.js.map +1 -0
- 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 +7 -2
- package/dist/parquetjs/parquet-thrift/ColumnChunk.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/ColumnChunk.js +37 -2
- package/dist/parquetjs/parquet-thrift/ColumnChunk.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/ColumnCryptoMetaData.d.ts +35 -0
- package/dist/parquetjs/parquet-thrift/ColumnCryptoMetaData.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/ColumnCryptoMetaData.js +135 -0
- package/dist/parquetjs/parquet-thrift/ColumnCryptoMetaData.js.map +1 -0
- 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 +12 -2
- package/dist/parquetjs/parquet-thrift/ColumnMetaData.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/ColumnMetaData.js +75 -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/EncryptionAlgorithm.d.ts +22 -0
- package/dist/parquetjs/parquet-thrift/EncryptionAlgorithm.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/EncryptionAlgorithm.js +111 -0
- package/dist/parquetjs/parquet-thrift/EncryptionAlgorithm.js.map +1 -0
- 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/FileCryptoMetaData.d.ts +16 -0
- package/dist/parquetjs/parquet-thrift/FileCryptoMetaData.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/FileCryptoMetaData.js +57 -0
- package/dist/parquetjs/parquet-thrift/FileCryptoMetaData.js.map +1 -0
- package/dist/parquetjs/parquet-thrift/FileMetaData.d.ts +7 -2
- package/dist/parquetjs/parquet-thrift/FileMetaData.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/FileMetaData.js +37 -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/GeospatialStatistics.d.ts +18 -0
- package/dist/parquetjs/parquet-thrift/GeospatialStatistics.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/GeospatialStatistics.js +62 -0
- package/dist/parquetjs/parquet-thrift/GeospatialStatistics.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/SizeStatistics.d.ts +26 -0
- package/dist/parquetjs/parquet-thrift/SizeStatistics.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/SizeStatistics.js +85 -0
- package/dist/parquetjs/parquet-thrift/SizeStatistics.js.map +1 -0
- 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 +12 -2
- package/dist/parquetjs/parquet-thrift/index.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/index.js +12 -2
- package/dist/parquetjs/parquet-thrift/index.js.map +1 -1
- package/dist/parquetjs/parser/decoders.d.ts +23 -3
- package/dist/parquetjs/parser/decoders.d.ts.map +1 -1
- package/dist/parquetjs/parser/decoders.js +687 -119
- package/dist/parquetjs/parser/decoders.js.map +1 -1
- package/dist/parquetjs/parser/encoded-pages.d.ts +24 -0
- package/dist/parquetjs/parser/encoded-pages.d.ts.map +1 -0
- package/dist/parquetjs/parser/encoded-pages.js +206 -0
- package/dist/parquetjs/parser/encoded-pages.js.map +1 -0
- package/dist/parquetjs/parser/parquet-reader.d.ts +111 -10
- package/dist/parquetjs/parser/parquet-reader.d.ts.map +1 -1
- package/dist/parquetjs/parser/parquet-reader.js +836 -69
- package/dist/parquetjs/parser/parquet-reader.js.map +1 -1
- package/dist/parquetjs/schema/declare.d.ts +83 -10
- 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 +31 -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 +285 -29
- 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 +290 -71
- package/dist/parquetjs/schema/types.js.map +1 -1
- package/dist/parquetjs/schema/variant.d.ts +3 -0
- package/dist/parquetjs/schema/variant.d.ts.map +1 -0
- package/dist/parquetjs/schema/variant.js +248 -0
- package/dist/parquetjs/schema/variant.js.map +1 -0
- 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/crc32.d.ts +3 -0
- package/dist/parquetjs/utils/crc32.d.ts.map +1 -0
- package/dist/parquetjs/utils/crc32.js +15 -0
- package/dist/parquetjs/utils/crc32.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 +25 -7
- package/dist/parquetjs/utils/read-utils.d.ts.map +1 -1
- package/dist/parquetjs/utils/read-utils.js +66 -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 +10 -0
- package/dist/unbundled.d.ts.map +1 -0
- package/dist/unbundled.js +11 -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 +84 -40
- package/src/bundled.ts +43 -0
- package/src/delta-format.ts +12 -0
- package/src/delta-source-loader-types.ts +40 -0
- package/src/delta-source.ts +305 -0
- package/src/delta-types.ts +40 -0
- package/src/geoparquet-loader-with-parser.ts +95 -0
- package/src/geoparquet-loader.ts +54 -0
- package/src/iceberg-rest-catalog.ts +122 -0
- package/src/iceberg-table-source.ts +1037 -0
- package/src/iceberg-types.ts +165 -0
- package/src/index.ts +105 -26
- package/src/lib/arrow/convert-columns-to-row-group.ts +3 -0
- package/src/lib/arrow/convert-schema-from-parquet.ts +193 -32
- package/src/lib/arrow/convert-schema-to-parquet.ts +16 -7
- package/src/lib/arrow/create-nested-arrow-vector.ts +705 -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 +526 -0
- package/src/lib/geo/geoparquet-covering.ts +201 -0
- package/src/lib/geo/geospatial-metadata.ts +547 -0
- package/src/lib/parquet-bloom-filter-planner.ts +86 -0
- package/src/lib/parquet-bloom-filter.ts +345 -0
- package/src/lib/parquet-encryption.ts +344 -0
- package/src/lib/parquet-page-index.ts +727 -0
- package/src/lib/parquet-predicate.ts +410 -0
- package/src/lib/parquet-source-worker-client.ts +41 -0
- package/src/lib/parquet-source-worker-decoder.ts +252 -0
- package/src/lib/parquet-source-worker-file.ts +56 -0
- package/src/lib/parquet-source-worker-types.ts +132 -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 +6 -2
- 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 +1213 -0
- package/src/lib/parsers/parse-parquet-to-arrow.ts +62 -36
- package/src/lib/parsers/parse-parquet-to-columns.ts +9 -8
- package/src/lib/parsers/parse-parquet-to-json.ts +119 -35
- package/src/lib/slice-parquet-batch.ts +32 -0
- package/src/lib/sources/parquet-range-file.ts +365 -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 +30 -0
- package/src/parquet-dataset-source.ts +800 -0
- package/src/parquet-encoded-page-types.ts +122 -0
- package/src/parquet-format.ts +2 -0
- package/src/parquet-js-loader-types.ts +53 -0
- package/src/parquet-js-loader.ts +76 -19
- package/src/parquet-js-writer.ts +98 -0
- package/src/parquet-loader-base.ts +34 -0
- package/src/parquet-loader-options.ts +62 -0
- package/src/parquet-loader-types.ts +35 -0
- package/src/parquet-loader.ts +115 -0
- package/src/parquet-source-capabilities.ts +73 -0
- package/src/parquet-source-loader-types.ts +97 -0
- package/src/parquet-source-loader.ts +2254 -0
- package/src/parquet-source-types.ts +627 -0
- package/src/parquet-source-worker-factory.ts +8 -0
- package/src/parquet-source-worker-url.ts +23 -0
- package/src/parquet-worker-url.ts +23 -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 +62 -3
- package/src/parquetjs/codecs/delta.ts +1047 -0
- package/src/parquetjs/codecs/dictionary.ts +22 -4
- package/src/parquetjs/codecs/index.ts +22 -0
- package/src/parquetjs/codecs/plain.ts +317 -89
- package/src/parquetjs/codecs/rle.ts +852 -82
- package/src/parquetjs/compression.ts +137 -156
- package/src/parquetjs/encoder/dictionary-planner.ts +104 -0
- package/src/parquetjs/encoder/page-planner.ts +83 -0
- package/src/parquetjs/encoder/parquet-encoder.ts +1374 -95
- package/src/parquetjs/parquet-thrift/BoundingBox.ts +108 -0
- package/src/parquetjs/parquet-thrift/BsonType.ts +1 -1
- package/src/parquetjs/parquet-thrift/ColumnChunk.ts +37 -2
- package/src/parquetjs/parquet-thrift/ColumnCryptoMetaData.ts +139 -0
- package/src/parquetjs/parquet-thrift/ColumnIndex.ts +13 -13
- package/src/parquetjs/parquet-thrift/ColumnMetaData.ts +75 -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/EncryptionAlgorithm.ts +127 -0
- package/src/parquetjs/parquet-thrift/EnumType.ts +1 -1
- package/src/parquetjs/parquet-thrift/FileCryptoMetaData.ts +60 -0
- package/src/parquetjs/parquet-thrift/FileMetaData.ts +37 -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/GeospatialStatistics.ts +68 -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/SizeStatistics.ts +98 -0
- 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 +12 -4
- package/src/parquetjs/parser/decoders.ts +922 -143
- package/src/parquetjs/parser/encoded-pages.ts +278 -0
- package/src/parquetjs/parser/parquet-reader.ts +1334 -83
- package/src/parquetjs/schema/declare.ts +121 -9
- package/src/parquetjs/schema/schema.ts +33 -3
- package/src/parquetjs/schema/shred.ts +345 -41
- package/src/parquetjs/schema/types.ts +308 -84
- package/src/parquetjs/schema/variant.ts +313 -0
- package/src/parquetjs/utils/binary-utils.ts +132 -0
- package/src/parquetjs/utils/crc32.ts +15 -0
- package/src/parquetjs/utils/file-utils.ts +4 -4
- package/src/parquetjs/utils/read-utils.ts +77 -51
- 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 +47 -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 +1,547 @@
|
|
|
1
|
+
// loaders.gl
|
|
2
|
+
// SPDX-License-Identifier: MIT
|
|
3
|
+
// Copyright (c) vis.gl contributors
|
|
4
|
+
|
|
5
|
+
import * as arrow from 'apache-arrow';
|
|
6
|
+
|
|
7
|
+
import {
|
|
8
|
+
getGeoMetadata,
|
|
9
|
+
getMetadataValue,
|
|
10
|
+
setGeoMetadata,
|
|
11
|
+
setMetadataValue,
|
|
12
|
+
type GeoArrowEncoding,
|
|
13
|
+
type GeoArrowMetadata,
|
|
14
|
+
type GeoColumnMetadata,
|
|
15
|
+
type GeoMetadata,
|
|
16
|
+
type GeoParquetGeometryType,
|
|
17
|
+
type Field,
|
|
18
|
+
type Schema
|
|
19
|
+
} from '@loaders.gl/schema';
|
|
20
|
+
import {
|
|
21
|
+
GeoArrowMetadataSchema,
|
|
22
|
+
GeoParquetMetadataSchema
|
|
23
|
+
} from '@loaders.gl/gis/geospatial-metadata-zod-schema';
|
|
24
|
+
import {convertArrowToSchema} from '@loaders.gl/schema-utils';
|
|
25
|
+
|
|
26
|
+
const GEOARROW_EXTENSION_NAME_KEY = 'ARROW:extension:name';
|
|
27
|
+
const GEOARROW_EXTENSION_METADATA_KEY = 'ARROW:extension:metadata';
|
|
28
|
+
const GEOPARQUET_VERSION = '1.1.0';
|
|
29
|
+
|
|
30
|
+
const GEOARROW_ENCODINGS = [
|
|
31
|
+
'geoarrow.geometry',
|
|
32
|
+
'geoarrow.geometrycollection',
|
|
33
|
+
'geoarrow.multipolygon',
|
|
34
|
+
'geoarrow.polygon',
|
|
35
|
+
'geoarrow.multilinestring',
|
|
36
|
+
'geoarrow.linestring',
|
|
37
|
+
'geoarrow.multipoint',
|
|
38
|
+
'geoarrow.point',
|
|
39
|
+
'geoarrow.box',
|
|
40
|
+
'geoarrow.wkb',
|
|
41
|
+
'geoarrow.wkt'
|
|
42
|
+
] as const satisfies GeoArrowEncoding[];
|
|
43
|
+
|
|
44
|
+
const GEOPARQUET_TO_GEOARROW_ENCODINGS = {
|
|
45
|
+
wkb: 'geoarrow.wkb',
|
|
46
|
+
geometry: 'geoarrow.wkb',
|
|
47
|
+
geography: 'geoarrow.wkb',
|
|
48
|
+
point: 'geoarrow.point',
|
|
49
|
+
linestring: 'geoarrow.linestring',
|
|
50
|
+
polygon: 'geoarrow.polygon',
|
|
51
|
+
multipoint: 'geoarrow.multipoint',
|
|
52
|
+
multilinestring: 'geoarrow.multilinestring',
|
|
53
|
+
multipolygon: 'geoarrow.multipolygon'
|
|
54
|
+
} as const satisfies Record<string, GeoArrowEncoding>;
|
|
55
|
+
|
|
56
|
+
const GEOARROW_TO_GEOPARQUET_ENCODINGS = {
|
|
57
|
+
'geoarrow.wkb': 'wkb',
|
|
58
|
+
'geoarrow.point': 'point',
|
|
59
|
+
'geoarrow.linestring': 'linestring',
|
|
60
|
+
'geoarrow.polygon': 'polygon',
|
|
61
|
+
'geoarrow.multipoint': 'multipoint',
|
|
62
|
+
'geoarrow.multilinestring': 'multilinestring',
|
|
63
|
+
'geoarrow.multipolygon': 'multipolygon'
|
|
64
|
+
} as const;
|
|
65
|
+
|
|
66
|
+
const GEOARROW_TO_GEOMETRY_TYPE = {
|
|
67
|
+
'geoarrow.point': 'Point',
|
|
68
|
+
'geoarrow.linestring': 'LineString',
|
|
69
|
+
'geoarrow.polygon': 'Polygon',
|
|
70
|
+
'geoarrow.multipoint': 'MultiPoint',
|
|
71
|
+
'geoarrow.multilinestring': 'MultiLineString',
|
|
72
|
+
'geoarrow.multipolygon': 'MultiPolygon'
|
|
73
|
+
} as const satisfies Record<string, GeoParquetGeometryType>;
|
|
74
|
+
|
|
75
|
+
/** Returns GeoArrow geometry field metadata keyed by column name. */
|
|
76
|
+
function getGeometryColumnsFromSchema(schema: Schema): Record<string, GeoArrowMetadata> {
|
|
77
|
+
const geometryColumns: Record<string, GeoArrowMetadata> = {};
|
|
78
|
+
for (const field of schema.fields || []) {
|
|
79
|
+
const metadata = getGeometryMetadataForField(field.metadata || {});
|
|
80
|
+
if (metadata) {
|
|
81
|
+
geometryColumns[field.name] = metadata;
|
|
82
|
+
}
|
|
83
|
+
}
|
|
84
|
+
return geometryColumns;
|
|
85
|
+
}
|
|
86
|
+
|
|
87
|
+
/** Extracts GeoArrow metadata from one schema field metadata object. */
|
|
88
|
+
function getGeometryMetadataForField(
|
|
89
|
+
fieldMetadata: Record<string, string>
|
|
90
|
+
): GeoArrowMetadata | null {
|
|
91
|
+
let metadata: GeoArrowMetadata | null = null;
|
|
92
|
+
let geoEncoding = getMetadataValue(fieldMetadata, GEOARROW_EXTENSION_NAME_KEY);
|
|
93
|
+
if (geoEncoding) {
|
|
94
|
+
geoEncoding = geoEncoding.toLowerCase();
|
|
95
|
+
if (geoEncoding === 'wkb') {
|
|
96
|
+
geoEncoding = 'geoarrow.wkb';
|
|
97
|
+
}
|
|
98
|
+
if (geoEncoding === 'wkt') {
|
|
99
|
+
geoEncoding = 'geoarrow.wkt';
|
|
100
|
+
}
|
|
101
|
+
if (GEOARROW_ENCODINGS.includes(geoEncoding as GeoArrowEncoding)) {
|
|
102
|
+
metadata ||= {};
|
|
103
|
+
metadata.encoding = geoEncoding as GeoArrowEncoding;
|
|
104
|
+
}
|
|
105
|
+
}
|
|
106
|
+
|
|
107
|
+
const columnMetadata = getMetadataValue(fieldMetadata, GEOARROW_EXTENSION_METADATA_KEY);
|
|
108
|
+
if (columnMetadata) {
|
|
109
|
+
try {
|
|
110
|
+
const parsedMetadata = parseGeoArrowMetadata(JSON.parse(columnMetadata));
|
|
111
|
+
metadata = {
|
|
112
|
+
...(metadata || {}),
|
|
113
|
+
...parsedMetadata
|
|
114
|
+
};
|
|
115
|
+
} catch {
|
|
116
|
+
return metadata;
|
|
117
|
+
}
|
|
118
|
+
}
|
|
119
|
+
|
|
120
|
+
return metadata;
|
|
121
|
+
}
|
|
122
|
+
|
|
123
|
+
/** Validates GeoArrow metadata, treating a mislabeled string CRS as opaque metadata. */
|
|
124
|
+
function parseGeoArrowMetadata(value: unknown): GeoArrowMetadata {
|
|
125
|
+
const result = GeoArrowMetadataSchema.safeParse(value);
|
|
126
|
+
if (result.success) {
|
|
127
|
+
return result.data;
|
|
128
|
+
}
|
|
129
|
+
if (value && typeof value === 'object' && typeof (value as {crs?: unknown}).crs === 'string') {
|
|
130
|
+
const {crs_type: _untrustedCRSType, ...opaqueMetadata} = value as Record<string, unknown>;
|
|
131
|
+
const opaqueResult = GeoArrowMetadataSchema.safeParse(opaqueMetadata);
|
|
132
|
+
if (opaqueResult.success) {
|
|
133
|
+
return opaqueResult.data;
|
|
134
|
+
}
|
|
135
|
+
}
|
|
136
|
+
throw result.error;
|
|
137
|
+
}
|
|
138
|
+
|
|
139
|
+
/**
|
|
140
|
+
* Applies GeoParquet schema metadata to matching geometry fields as GeoArrow field metadata.
|
|
141
|
+
*
|
|
142
|
+
* @param schema - Schema to annotate.
|
|
143
|
+
* @returns The same schema reference.
|
|
144
|
+
*/
|
|
145
|
+
export function applyGeoParquetToFieldMetadata(schema: Schema): Schema {
|
|
146
|
+
const geoMetadata = getGeoMetadata(schema.metadata);
|
|
147
|
+
if (!geoMetadata) {
|
|
148
|
+
return schema;
|
|
149
|
+
}
|
|
150
|
+
|
|
151
|
+
for (const field of schema.fields) {
|
|
152
|
+
const columnMetadata = geoMetadata.columns?.[field.name];
|
|
153
|
+
if (!columnMetadata) {
|
|
154
|
+
continue;
|
|
155
|
+
}
|
|
156
|
+
|
|
157
|
+
const fieldMetadata = getGeoArrowMetadataFromGeoParquetField(field, columnMetadata);
|
|
158
|
+
if (!fieldMetadata) {
|
|
159
|
+
continue;
|
|
160
|
+
}
|
|
161
|
+
|
|
162
|
+
field.metadata ||= {};
|
|
163
|
+
setMetadataValue(field.metadata, GEOARROW_EXTENSION_NAME_KEY, fieldMetadata.extensionName);
|
|
164
|
+
|
|
165
|
+
if (fieldMetadata.extensionMetadata) {
|
|
166
|
+
setMetadataValue(
|
|
167
|
+
field.metadata,
|
|
168
|
+
GEOARROW_EXTENSION_METADATA_KEY,
|
|
169
|
+
JSON.stringify(fieldMetadata.extensionMetadata)
|
|
170
|
+
);
|
|
171
|
+
}
|
|
172
|
+
}
|
|
173
|
+
|
|
174
|
+
return schema;
|
|
175
|
+
}
|
|
176
|
+
|
|
177
|
+
/**
|
|
178
|
+
* Creates an Arrow JS table whose schema metadata matches a loaders.gl schema.
|
|
179
|
+
*
|
|
180
|
+
* @param table - Arrow JS table to rewrap.
|
|
181
|
+
* @param schema - loaders.gl schema containing desired field/schema metadata.
|
|
182
|
+
* @returns Table using the updated Arrow JS schema.
|
|
183
|
+
*/
|
|
184
|
+
export function applyLoadersSchemaToArrowTable(table: arrow.Table, schema: Schema): arrow.Table {
|
|
185
|
+
const arrowSchema = createArrowJsSchema(table.schema, schema);
|
|
186
|
+
return new arrow.Table(arrowSchema, table.batches);
|
|
187
|
+
}
|
|
188
|
+
|
|
189
|
+
/**
|
|
190
|
+
* Normalizes an Arrow wrapper so GeoParquet metadata is preserved and GeoArrow field metadata is attached.
|
|
191
|
+
*
|
|
192
|
+
* @param arrowTable - loaders.gl Arrow table wrapper.
|
|
193
|
+
* @param schemaMetadata - Optional schema metadata override from the Parquet footer.
|
|
194
|
+
* @returns Arrow table wrapper with synchronized schema metadata.
|
|
195
|
+
*/
|
|
196
|
+
export function normalizeArrowTableGeoMetadata(
|
|
197
|
+
arrowTable: {shape: 'arrow-table'; data: arrow.Table; schema?: Schema},
|
|
198
|
+
schemaMetadata?: Map<string, string> | Record<string, string>
|
|
199
|
+
): {shape: 'arrow-table'; data: arrow.Table; schema: Schema} {
|
|
200
|
+
const schema = convertArrowToSchema(arrowTable.data.schema);
|
|
201
|
+
mergeSchemaMetadata(schema, schemaMetadata);
|
|
202
|
+
applyGeoParquetToFieldMetadata(schema);
|
|
203
|
+
|
|
204
|
+
return {
|
|
205
|
+
shape: 'arrow-table',
|
|
206
|
+
data: applyLoadersSchemaToArrowTable(arrowTable.data, schema),
|
|
207
|
+
schema
|
|
208
|
+
};
|
|
209
|
+
}
|
|
210
|
+
|
|
211
|
+
/**
|
|
212
|
+
* Synthesizes GeoParquet schema metadata from GeoArrow field metadata when needed.
|
|
213
|
+
*
|
|
214
|
+
* @param schema - Schema to inspect and update.
|
|
215
|
+
* @returns The same schema reference.
|
|
216
|
+
*/
|
|
217
|
+
export function ensureGeoParquetMetadata(schema: Schema): Schema {
|
|
218
|
+
schema.metadata ||= {};
|
|
219
|
+
const rawExistingGeoMetadata = getGeoMetadata(schema.metadata);
|
|
220
|
+
if (isValidGeoParquetMetadata(rawExistingGeoMetadata)) {
|
|
221
|
+
return schema;
|
|
222
|
+
}
|
|
223
|
+
const existingGeoMetadata: GeoMetadata | undefined = rawExistingGeoMetadata || undefined;
|
|
224
|
+
|
|
225
|
+
const geometryColumns = getGeometryColumnsFromSchema(schema);
|
|
226
|
+
const geometryColumnNames = Object.keys(geometryColumns);
|
|
227
|
+
if (geometryColumnNames.length === 0) {
|
|
228
|
+
return schema;
|
|
229
|
+
}
|
|
230
|
+
|
|
231
|
+
const geoMetadata: GeoMetadata = {
|
|
232
|
+
version: GEOPARQUET_VERSION,
|
|
233
|
+
primary_column: geometryColumnNames[0],
|
|
234
|
+
columns: {}
|
|
235
|
+
};
|
|
236
|
+
|
|
237
|
+
for (const geometryColumnName of geometryColumnNames) {
|
|
238
|
+
const geometryMetadata = geometryColumns[geometryColumnName];
|
|
239
|
+
const synthesizedColumnMetadata = synthesizeGeoParquetColumnMetadata(geometryMetadata);
|
|
240
|
+
if (!synthesizedColumnMetadata) {
|
|
241
|
+
continue;
|
|
242
|
+
}
|
|
243
|
+
|
|
244
|
+
const existingColumnMetadata = existingGeoMetadata?.columns?.[geometryColumnName];
|
|
245
|
+
geoMetadata.columns[geometryColumnName] = {
|
|
246
|
+
...pickValidOptionalGeoParquetColumnMetadata(existingColumnMetadata),
|
|
247
|
+
...synthesizedColumnMetadata
|
|
248
|
+
};
|
|
249
|
+
}
|
|
250
|
+
|
|
251
|
+
if (Object.keys(geoMetadata.columns).length === 0) {
|
|
252
|
+
return schema;
|
|
253
|
+
}
|
|
254
|
+
|
|
255
|
+
setGeoMetadata(schema.metadata, geoMetadata);
|
|
256
|
+
return schema;
|
|
257
|
+
}
|
|
258
|
+
|
|
259
|
+
/**
|
|
260
|
+
* Ensures an Arrow table wrapper carries valid GeoParquet schema metadata when GeoArrow fields are present.
|
|
261
|
+
*
|
|
262
|
+
* @param arrowTable - Arrow table wrapper to inspect and update.
|
|
263
|
+
* @returns Updated Arrow table wrapper.
|
|
264
|
+
*/
|
|
265
|
+
export function ensureGeoParquetMetadataOnArrowTable(arrowTable: {
|
|
266
|
+
shape: 'arrow-table';
|
|
267
|
+
data: arrow.Table;
|
|
268
|
+
schema?: Schema;
|
|
269
|
+
}): {shape: 'arrow-table'; data: arrow.Table; schema: Schema} {
|
|
270
|
+
const schema = arrowTable.schema || convertArrowToSchema(arrowTable.data.schema);
|
|
271
|
+
ensureGeoParquetMetadata(schema);
|
|
272
|
+
|
|
273
|
+
return {
|
|
274
|
+
shape: 'arrow-table',
|
|
275
|
+
data: applyLoadersSchemaToArrowTable(arrowTable.data, schema),
|
|
276
|
+
schema
|
|
277
|
+
};
|
|
278
|
+
}
|
|
279
|
+
|
|
280
|
+
/**
|
|
281
|
+
* Returns Parquet file key/value metadata as a plain map.
|
|
282
|
+
*
|
|
283
|
+
* @param parquetMetadata - parquet-wasm metadata object.
|
|
284
|
+
* @returns Metadata map.
|
|
285
|
+
*/
|
|
286
|
+
export function getParquetFileMetadataMap(parquetMetadata: {
|
|
287
|
+
fileMetadata(): {keyValueMetadata(): Map<string, string>};
|
|
288
|
+
}): Map<string, string> {
|
|
289
|
+
return parquetMetadata.fileMetadata().keyValueMetadata();
|
|
290
|
+
}
|
|
291
|
+
|
|
292
|
+
function mergeSchemaMetadata(
|
|
293
|
+
schema: Schema,
|
|
294
|
+
schemaMetadata?: Map<string, string> | Record<string, string>
|
|
295
|
+
): void {
|
|
296
|
+
if (!schemaMetadata) {
|
|
297
|
+
return;
|
|
298
|
+
}
|
|
299
|
+
|
|
300
|
+
schema.metadata ||= {};
|
|
301
|
+
for (const [metadataKey, metadataValue] of getMetadataEntries(schemaMetadata)) {
|
|
302
|
+
setMetadataValue(schema.metadata, metadataKey, metadataValue);
|
|
303
|
+
}
|
|
304
|
+
}
|
|
305
|
+
|
|
306
|
+
function createArrowJsSchema(baseSchema: arrow.Schema, schema: Schema): arrow.Schema {
|
|
307
|
+
const fields = baseSchema.fields.map(baseField => {
|
|
308
|
+
const nextField = schema.fields.find(field => field.name === baseField.name);
|
|
309
|
+
if (!nextField?.metadata) {
|
|
310
|
+
return baseField;
|
|
311
|
+
}
|
|
312
|
+
|
|
313
|
+
return baseField.clone({
|
|
314
|
+
metadata: createMetadataMap(nextField.metadata)
|
|
315
|
+
});
|
|
316
|
+
});
|
|
317
|
+
|
|
318
|
+
return new arrow.Schema(
|
|
319
|
+
fields,
|
|
320
|
+
createMetadataMap(schema.metadata),
|
|
321
|
+
baseSchema.dictionaries,
|
|
322
|
+
baseSchema.metadataVersion
|
|
323
|
+
);
|
|
324
|
+
}
|
|
325
|
+
|
|
326
|
+
function getGeoArrowMetadataFromGeoParquetField(
|
|
327
|
+
field: Field,
|
|
328
|
+
columnMetadata: GeoColumnMetadata
|
|
329
|
+
): {extensionName: GeoArrowEncoding; extensionMetadata?: Record<string, unknown>} | null {
|
|
330
|
+
const normalizedEncoding = String(columnMetadata.encoding || '').toLowerCase();
|
|
331
|
+
const extensionName =
|
|
332
|
+
GEOPARQUET_TO_GEOARROW_ENCODINGS[
|
|
333
|
+
normalizedEncoding as keyof typeof GEOPARQUET_TO_GEOARROW_ENCODINGS
|
|
334
|
+
];
|
|
335
|
+
|
|
336
|
+
if (!extensionName) {
|
|
337
|
+
return null;
|
|
338
|
+
}
|
|
339
|
+
|
|
340
|
+
const isBinaryGeospatialEncoding =
|
|
341
|
+
normalizedEncoding === 'wkb' ||
|
|
342
|
+
normalizedEncoding === 'geometry' ||
|
|
343
|
+
normalizedEncoding === 'geography';
|
|
344
|
+
if (isBinaryGeospatialEncoding && !isBinaryFieldType(field.type)) {
|
|
345
|
+
return null;
|
|
346
|
+
}
|
|
347
|
+
|
|
348
|
+
if (
|
|
349
|
+
!isBinaryGeospatialEncoding &&
|
|
350
|
+
(isBinaryFieldType(field.type) || isStringFieldType(field.type))
|
|
351
|
+
) {
|
|
352
|
+
return null;
|
|
353
|
+
}
|
|
354
|
+
|
|
355
|
+
const extensionMetadata: Record<string, unknown> = {};
|
|
356
|
+
if (columnMetadata.crs === undefined) {
|
|
357
|
+
extensionMetadata.crs = 'OGC:CRS84';
|
|
358
|
+
extensionMetadata.crs_type = 'authority_code';
|
|
359
|
+
} else if (columnMetadata.crs !== null) {
|
|
360
|
+
extensionMetadata.crs = columnMetadata.crs;
|
|
361
|
+
extensionMetadata.crs_type = 'projjson';
|
|
362
|
+
}
|
|
363
|
+
if (columnMetadata.edges && columnMetadata.edges !== 'planar') {
|
|
364
|
+
extensionMetadata.edges = columnMetadata.edges;
|
|
365
|
+
}
|
|
366
|
+
if (normalizedEncoding === 'geography' && !extensionMetadata.edges) {
|
|
367
|
+
extensionMetadata.edges = 'spherical';
|
|
368
|
+
}
|
|
369
|
+
for (const [metadataKey, metadataValue] of Object.entries(columnMetadata)) {
|
|
370
|
+
if (
|
|
371
|
+
metadataValue !== undefined &&
|
|
372
|
+
metadataValue !== null &&
|
|
373
|
+
!['encoding', 'geometry_types', 'crs', 'edges'].includes(metadataKey)
|
|
374
|
+
) {
|
|
375
|
+
extensionMetadata[metadataKey] = metadataValue;
|
|
376
|
+
}
|
|
377
|
+
}
|
|
378
|
+
|
|
379
|
+
return Object.keys(extensionMetadata).length > 0
|
|
380
|
+
? {extensionName, extensionMetadata}
|
|
381
|
+
: {extensionName};
|
|
382
|
+
}
|
|
383
|
+
|
|
384
|
+
function isValidGeoParquetMetadata(geoMetadata: GeoMetadata | null): boolean {
|
|
385
|
+
return GeoParquetMetadataSchema.safeParse(geoMetadata).success;
|
|
386
|
+
}
|
|
387
|
+
|
|
388
|
+
function synthesizeGeoParquetColumnMetadata(
|
|
389
|
+
geometryMetadata: GeoArrowMetadata
|
|
390
|
+
): GeoColumnMetadata | null {
|
|
391
|
+
const encoding = geometryMetadata.encoding
|
|
392
|
+
? GEOARROW_TO_GEOPARQUET_ENCODINGS[
|
|
393
|
+
geometryMetadata.encoding as keyof typeof GEOARROW_TO_GEOPARQUET_ENCODINGS
|
|
394
|
+
]
|
|
395
|
+
: null;
|
|
396
|
+
|
|
397
|
+
if (!encoding) {
|
|
398
|
+
return null;
|
|
399
|
+
}
|
|
400
|
+
|
|
401
|
+
const geometryTypes = Array.isArray(geometryMetadata.geometry_types)
|
|
402
|
+
? geometryMetadata.geometry_types
|
|
403
|
+
: inferGeometryTypesFromGeoArrowEncoding(geometryMetadata.encoding);
|
|
404
|
+
|
|
405
|
+
const columnMetadata: GeoColumnMetadata = {
|
|
406
|
+
encoding,
|
|
407
|
+
geometry_types: geometryTypes
|
|
408
|
+
};
|
|
409
|
+
|
|
410
|
+
const preservedMetadata = preserveUnknownGeoParquetColumnMetadata(geometryMetadata);
|
|
411
|
+
|
|
412
|
+
if (geometryMetadata.crs === undefined) {
|
|
413
|
+
columnMetadata.crs = null;
|
|
414
|
+
} else if (typeof geometryMetadata.crs === 'object') {
|
|
415
|
+
columnMetadata.crs = geometryMetadata.crs;
|
|
416
|
+
} else if (
|
|
417
|
+
typeof geometryMetadata.crs !== 'string' ||
|
|
418
|
+
!isDefaultCRS84Identifier(geometryMetadata.crs)
|
|
419
|
+
) {
|
|
420
|
+
return null;
|
|
421
|
+
}
|
|
422
|
+
if (geometryMetadata.edges) {
|
|
423
|
+
columnMetadata.edges = geometryMetadata.edges;
|
|
424
|
+
}
|
|
425
|
+
|
|
426
|
+
return {...preservedMetadata, ...columnMetadata};
|
|
427
|
+
}
|
|
428
|
+
|
|
429
|
+
/** Returns whether a compact GeoArrow CRS identifier has GeoParquet's default CRS semantics. */
|
|
430
|
+
function isDefaultCRS84Identifier(crs: string): boolean {
|
|
431
|
+
const normalizedCRS = crs.toUpperCase();
|
|
432
|
+
return normalizedCRS === 'OGC:CRS84' || normalizedCRS === 'EPSG:4326';
|
|
433
|
+
}
|
|
434
|
+
|
|
435
|
+
function inferGeometryTypesFromGeoArrowEncoding(
|
|
436
|
+
encoding?: GeoArrowEncoding
|
|
437
|
+
): GeoParquetGeometryType[] {
|
|
438
|
+
const geometryType = encoding
|
|
439
|
+
? GEOARROW_TO_GEOMETRY_TYPE[encoding as keyof typeof GEOARROW_TO_GEOMETRY_TYPE]
|
|
440
|
+
: null;
|
|
441
|
+
return geometryType ? [geometryType] : [];
|
|
442
|
+
}
|
|
443
|
+
|
|
444
|
+
function pickValidOptionalGeoParquetColumnMetadata(
|
|
445
|
+
columnMetadata: GeoColumnMetadata | undefined
|
|
446
|
+
): Partial<GeoColumnMetadata> {
|
|
447
|
+
if (!columnMetadata) {
|
|
448
|
+
return {};
|
|
449
|
+
}
|
|
450
|
+
|
|
451
|
+
const nextColumnMetadata = preserveUnknownGeoParquetColumnMetadata(columnMetadata);
|
|
452
|
+
|
|
453
|
+
if (columnMetadata.orientation === 'counterclockwise') {
|
|
454
|
+
nextColumnMetadata.orientation = columnMetadata.orientation;
|
|
455
|
+
}
|
|
456
|
+
|
|
457
|
+
if (isValidBBox(columnMetadata.bbox)) {
|
|
458
|
+
nextColumnMetadata.bbox = columnMetadata.bbox;
|
|
459
|
+
}
|
|
460
|
+
|
|
461
|
+
if (typeof columnMetadata.epoch === 'number' && Number.isFinite(columnMetadata.epoch)) {
|
|
462
|
+
nextColumnMetadata.epoch = columnMetadata.epoch;
|
|
463
|
+
}
|
|
464
|
+
|
|
465
|
+
if (columnMetadata.covering && typeof columnMetadata.covering === 'object') {
|
|
466
|
+
nextColumnMetadata.covering = columnMetadata.covering;
|
|
467
|
+
}
|
|
468
|
+
|
|
469
|
+
return nextColumnMetadata;
|
|
470
|
+
}
|
|
471
|
+
|
|
472
|
+
/** Preserves vendor and future GeoParquet column keys without trusting core fields. */
|
|
473
|
+
function preserveUnknownGeoParquetColumnMetadata(
|
|
474
|
+
columnMetadata: Record<string, unknown> | undefined
|
|
475
|
+
): Record<string, unknown> {
|
|
476
|
+
if (!columnMetadata) {
|
|
477
|
+
return {};
|
|
478
|
+
}
|
|
479
|
+
|
|
480
|
+
const preservedMetadata: Record<string, unknown> = {};
|
|
481
|
+
for (const [metadataKey, metadataValue] of Object.entries(columnMetadata)) {
|
|
482
|
+
if (
|
|
483
|
+
metadataValue !== undefined &&
|
|
484
|
+
![
|
|
485
|
+
'encoding',
|
|
486
|
+
'geometry_types',
|
|
487
|
+
'crs',
|
|
488
|
+
'edges',
|
|
489
|
+
'orientation',
|
|
490
|
+
'bbox',
|
|
491
|
+
'epoch',
|
|
492
|
+
'covering'
|
|
493
|
+
].includes(metadataKey)
|
|
494
|
+
) {
|
|
495
|
+
preservedMetadata[metadataKey] = metadataValue;
|
|
496
|
+
}
|
|
497
|
+
}
|
|
498
|
+
return preservedMetadata;
|
|
499
|
+
}
|
|
500
|
+
|
|
501
|
+
/** Returns whether an Arrow field is a binary value capable of carrying WKB bytes. */
|
|
502
|
+
function isBinaryFieldType(fieldType: Field['type']): boolean {
|
|
503
|
+
return typeof fieldType === 'string' && (fieldType === 'binary' || fieldType === 'binary-view');
|
|
504
|
+
}
|
|
505
|
+
|
|
506
|
+
/** Returns whether an Arrow field is a UTF-8 value capable of carrying WKT text. */
|
|
507
|
+
function isStringFieldType(fieldType: Field['type']): boolean {
|
|
508
|
+
return typeof fieldType === 'string' && (fieldType === 'utf8' || fieldType === 'utf8-view');
|
|
509
|
+
}
|
|
510
|
+
|
|
511
|
+
function isValidBBox(
|
|
512
|
+
bbox: GeoColumnMetadata['bbox']
|
|
513
|
+
): bbox is NonNullable<GeoColumnMetadata['bbox']> {
|
|
514
|
+
if (!Array.isArray(bbox) || ![4, 6, 8].includes(bbox.length)) {
|
|
515
|
+
return false;
|
|
516
|
+
}
|
|
517
|
+
|
|
518
|
+
return bbox.every(value => typeof value === 'number' && Number.isFinite(value));
|
|
519
|
+
}
|
|
520
|
+
|
|
521
|
+
function getMetadataEntries(
|
|
522
|
+
metadata: Map<string, string> | Record<string, string>
|
|
523
|
+
): [string, string][] {
|
|
524
|
+
if (metadata instanceof Map) {
|
|
525
|
+
return [...metadata.entries()];
|
|
526
|
+
}
|
|
527
|
+
return Object.entries(metadata);
|
|
528
|
+
}
|
|
529
|
+
|
|
530
|
+
function createMetadataMap(
|
|
531
|
+
metadata: Map<string, string> | Record<string, string> | undefined
|
|
532
|
+
): Map<string, string> {
|
|
533
|
+
if (!metadata) {
|
|
534
|
+
return new Map();
|
|
535
|
+
}
|
|
536
|
+
|
|
537
|
+
if (metadata instanceof Map) {
|
|
538
|
+
return new Map(metadata);
|
|
539
|
+
}
|
|
540
|
+
|
|
541
|
+
return new Map(
|
|
542
|
+
Object.entries(metadata).map(([metadataKey, metadataValue]) => [
|
|
543
|
+
metadataKey,
|
|
544
|
+
String(metadataValue)
|
|
545
|
+
])
|
|
546
|
+
);
|
|
547
|
+
}
|
|
@@ -0,0 +1,86 @@
|
|
|
1
|
+
// loaders.gl
|
|
2
|
+
// SPDX-License-Identifier: MIT
|
|
3
|
+
// Copyright (c) vis.gl contributors
|
|
4
|
+
|
|
5
|
+
import type {
|
|
6
|
+
ParquetColumnChunkMetadata,
|
|
7
|
+
ParquetPredicate,
|
|
8
|
+
ParquetPredicateValue,
|
|
9
|
+
ParquetRowGroupMetadata
|
|
10
|
+
} from '../parquet-source-types';
|
|
11
|
+
|
|
12
|
+
/** One conservative Bloom-filter probe that can be applied before decoding a row group. */
|
|
13
|
+
export type ParquetBloomFilterProbe = {
|
|
14
|
+
/** Column chunk containing the split-block Bloom filter. */
|
|
15
|
+
readonly column: ParquetColumnChunkMetadata;
|
|
16
|
+
/** Plain-encoded values to hash and probe. */
|
|
17
|
+
readonly values: readonly ParquetPredicateValue[];
|
|
18
|
+
};
|
|
19
|
+
|
|
20
|
+
/**
|
|
21
|
+
* Finds equality and membership predicates backed by row-group Bloom-filter metadata.
|
|
22
|
+
*
|
|
23
|
+
* Only leaves and conjunctions are returned. Disjunctions and negations are intentionally
|
|
24
|
+
* excluded because probing an individual branch could otherwise create a false negative.
|
|
25
|
+
*/
|
|
26
|
+
export function getParquetBloomFilterProbes(
|
|
27
|
+
predicate: ParquetPredicate,
|
|
28
|
+
rowGroup: ParquetRowGroupMetadata
|
|
29
|
+
): ParquetBloomFilterProbe[] {
|
|
30
|
+
const probes: ParquetBloomFilterProbe[] = [];
|
|
31
|
+
collectParquetBloomFilterProbes(predicate, rowGroup, probes);
|
|
32
|
+
const uniqueProbes = new Map<string, ParquetBloomFilterProbe>();
|
|
33
|
+
for (const probe of probes) {
|
|
34
|
+
const key = probe.column.path.join('\0');
|
|
35
|
+
const previousProbe = uniqueProbes.get(key);
|
|
36
|
+
if (previousProbe) {
|
|
37
|
+
uniqueProbes.set(key, {
|
|
38
|
+
column: probe.column,
|
|
39
|
+
values: [...previousProbe.values, ...probe.values]
|
|
40
|
+
});
|
|
41
|
+
} else {
|
|
42
|
+
uniqueProbes.set(key, probe);
|
|
43
|
+
}
|
|
44
|
+
}
|
|
45
|
+
return [...uniqueProbes.values()];
|
|
46
|
+
}
|
|
47
|
+
|
|
48
|
+
function collectParquetBloomFilterProbes(
|
|
49
|
+
predicate: ParquetPredicate,
|
|
50
|
+
rowGroup: ParquetRowGroupMetadata,
|
|
51
|
+
probes: ParquetBloomFilterProbe[]
|
|
52
|
+
): void {
|
|
53
|
+
if (predicate.op === 'and') {
|
|
54
|
+
for (const child of predicate.args) {
|
|
55
|
+
collectParquetBloomFilterProbes(child, rowGroup, probes);
|
|
56
|
+
}
|
|
57
|
+
return;
|
|
58
|
+
}
|
|
59
|
+
if (predicate.op !== '=' && predicate.op !== 'in') {
|
|
60
|
+
return;
|
|
61
|
+
}
|
|
62
|
+
const path =
|
|
63
|
+
typeof predicate.args[0].property === 'string'
|
|
64
|
+
? [predicate.args[0].property]
|
|
65
|
+
: [...predicate.args[0].property];
|
|
66
|
+
const column = rowGroup.columns.find(
|
|
67
|
+
candidate =>
|
|
68
|
+
candidate.path.length === path.length &&
|
|
69
|
+
candidate.path.every((part, index) => part === path[index])
|
|
70
|
+
);
|
|
71
|
+
if (
|
|
72
|
+
!column ||
|
|
73
|
+
column.bloomFilterOffset === undefined ||
|
|
74
|
+
column.bloomFilterByteLength === undefined ||
|
|
75
|
+
column.bloomFilterByteLength <= 0
|
|
76
|
+
) {
|
|
77
|
+
return;
|
|
78
|
+
}
|
|
79
|
+
probes.push({
|
|
80
|
+
column,
|
|
81
|
+
values:
|
|
82
|
+
predicate.op === '='
|
|
83
|
+
? [predicate.args[1]]
|
|
84
|
+
: [...(predicate.args[1] as readonly ParquetPredicateValue[])]
|
|
85
|
+
});
|
|
86
|
+
}
|