@loaders.gl/parquet 4.0.0-alpha.5 → 4.0.0-alpha.6
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/dist/bundle.js +2 -2
- package/dist/constants.js +18 -6
- package/dist/dist.min.js +27 -25
- package/dist/dist.min.js.map +3 -3
- package/dist/es5/bundle.js +6 -0
- package/dist/es5/bundle.js.map +1 -0
- package/dist/es5/constants.js +17 -0
- package/dist/es5/constants.js.map +1 -0
- package/dist/es5/index.js +128 -0
- package/dist/es5/index.js.map +1 -0
- package/dist/es5/lib/arrow/convert-columns-to-row-group.js +2 -0
- package/dist/es5/lib/arrow/convert-columns-to-row-group.js.map +1 -0
- package/dist/es5/lib/arrow/convert-row-group-to-columns.js +19 -0
- package/dist/es5/lib/arrow/convert-row-group-to-columns.js.map +1 -0
- package/dist/es5/lib/arrow/convert-schema-from-parquet.js +114 -0
- package/dist/es5/lib/arrow/convert-schema-from-parquet.js.map +1 -0
- package/dist/es5/lib/arrow/convert-schema-to-parquet.js +47 -0
- package/dist/es5/lib/arrow/convert-schema-to-parquet.js.map +1 -0
- package/dist/es5/lib/geo/decode-geo-metadata.js +81 -0
- package/dist/es5/lib/geo/decode-geo-metadata.js.map +1 -0
- package/dist/es5/lib/geo/geoparquet-schema.js +83 -0
- package/dist/es5/lib/geo/geoparquet-schema.js.map +1 -0
- package/dist/es5/lib/parsers/parse-parquet-to-columns.js +177 -0
- package/dist/es5/lib/parsers/parse-parquet-to-columns.js.map +1 -0
- package/dist/es5/lib/parsers/parse-parquet-to-rows.js +172 -0
- package/dist/es5/lib/parsers/parse-parquet-to-rows.js.map +1 -0
- package/dist/es5/lib/wasm/encode-parquet-wasm.js +43 -0
- package/dist/es5/lib/wasm/encode-parquet-wasm.js.map +1 -0
- package/dist/es5/lib/wasm/load-wasm/index.js +13 -0
- package/dist/es5/lib/wasm/load-wasm/index.js.map +1 -0
- package/dist/es5/lib/wasm/load-wasm/load-wasm-browser.js +42 -0
- package/dist/es5/lib/wasm/load-wasm/load-wasm-browser.js.map +1 -0
- package/dist/es5/lib/wasm/load-wasm/load-wasm-node.js +31 -0
- package/dist/es5/lib/wasm/load-wasm/load-wasm-node.js.map +1 -0
- package/dist/es5/lib/wasm/parse-parquet-wasm.js +60 -0
- package/dist/es5/lib/wasm/parse-parquet-wasm.js.map +1 -0
- package/dist/es5/lib/wip/convert-schema-deep.java.disabled +910 -0
- package/dist/es5/lib/wip/convert-schema-deep.rs.disabled +976 -0
- package/dist/es5/parquet-loader.js +44 -0
- package/dist/es5/parquet-loader.js.map +1 -0
- package/dist/es5/parquet-wasm-loader.js +30 -0
- package/dist/es5/parquet-wasm-loader.js.map +1 -0
- package/dist/es5/parquet-wasm-writer.js +26 -0
- package/dist/es5/parquet-wasm-writer.js.map +1 -0
- package/dist/es5/parquet-writer.js +24 -0
- package/dist/es5/parquet-writer.js.map +1 -0
- package/dist/es5/parquetjs/codecs/declare.js +2 -0
- package/dist/es5/parquetjs/codecs/declare.js.map +1 -0
- package/dist/es5/parquetjs/codecs/dictionary.js +23 -0
- package/dist/es5/parquetjs/codecs/dictionary.js.map +1 -0
- package/dist/es5/parquetjs/codecs/index.js +47 -0
- package/dist/es5/parquetjs/codecs/index.js.map +1 -0
- package/dist/es5/parquetjs/codecs/plain.js +208 -0
- package/dist/es5/parquetjs/codecs/plain.js.map +1 -0
- package/dist/es5/parquetjs/codecs/rle.js +132 -0
- package/dist/es5/parquetjs/codecs/rle.js.map +1 -0
- package/dist/es5/parquetjs/compression.js +137 -0
- package/dist/es5/parquetjs/compression.js.map +1 -0
- package/dist/es5/parquetjs/encoder/parquet-encoder.js +625 -0
- package/dist/es5/parquetjs/encoder/parquet-encoder.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/BoundaryOrder.js +14 -0
- package/dist/es5/parquetjs/parquet-thrift/BoundaryOrder.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/BsonType.js +52 -0
- package/dist/es5/parquetjs/parquet-thrift/BsonType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/ColumnChunk.js +193 -0
- package/dist/es5/parquetjs/parquet-thrift/ColumnChunk.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/ColumnIndex.js +198 -0
- package/dist/es5/parquetjs/parquet-thrift/ColumnIndex.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/ColumnMetaData.js +367 -0
- package/dist/es5/parquetjs/parquet-thrift/ColumnMetaData.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/ColumnOrder.js +99 -0
- package/dist/es5/parquetjs/parquet-thrift/ColumnOrder.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/CompressionCodec.js +19 -0
- package/dist/es5/parquetjs/parquet-thrift/CompressionCodec.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/ConvertedType.js +33 -0
- package/dist/es5/parquetjs/parquet-thrift/ConvertedType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/DataPageHeader.js +152 -0
- package/dist/es5/parquetjs/parquet-thrift/DataPageHeader.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/DataPageHeaderV2.js +207 -0
- package/dist/es5/parquetjs/parquet-thrift/DataPageHeaderV2.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/DateType.js +52 -0
- package/dist/es5/parquetjs/parquet-thrift/DateType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/DecimalType.js +96 -0
- package/dist/es5/parquetjs/parquet-thrift/DecimalType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/DictionaryPageHeader.js +113 -0
- package/dist/es5/parquetjs/parquet-thrift/DictionaryPageHeader.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/Encoding.js +19 -0
- package/dist/es5/parquetjs/parquet-thrift/Encoding.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/EnumType.js +52 -0
- package/dist/es5/parquetjs/parquet-thrift/EnumType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/FieldRepetitionType.js +14 -0
- package/dist/es5/parquetjs/parquet-thrift/FieldRepetitionType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/FileMetaData.js +239 -0
- package/dist/es5/parquetjs/parquet-thrift/FileMetaData.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/IndexPageHeader.js +52 -0
- package/dist/es5/parquetjs/parquet-thrift/IndexPageHeader.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/IntType.js +96 -0
- package/dist/es5/parquetjs/parquet-thrift/IntType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/JsonType.js +52 -0
- package/dist/es5/parquetjs/parquet-thrift/JsonType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/KeyValue.js +94 -0
- package/dist/es5/parquetjs/parquet-thrift/KeyValue.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/ListType.js +52 -0
- package/dist/es5/parquetjs/parquet-thrift/ListType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/LogicalType.js +423 -0
- package/dist/es5/parquetjs/parquet-thrift/LogicalType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/MapType.js +52 -0
- package/dist/es5/parquetjs/parquet-thrift/MapType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/MicroSeconds.js +52 -0
- package/dist/es5/parquetjs/parquet-thrift/MicroSeconds.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/MilliSeconds.js +52 -0
- package/dist/es5/parquetjs/parquet-thrift/MilliSeconds.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/NullType.js +52 -0
- package/dist/es5/parquetjs/parquet-thrift/NullType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/OffsetIndex.js +89 -0
- package/dist/es5/parquetjs/parquet-thrift/OffsetIndex.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/PageEncodingStats.js +115 -0
- package/dist/es5/parquetjs/parquet-thrift/PageEncodingStats.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/PageHeader.js +204 -0
- package/dist/es5/parquetjs/parquet-thrift/PageHeader.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/PageLocation.js +124 -0
- package/dist/es5/parquetjs/parquet-thrift/PageLocation.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/PageType.js +15 -0
- package/dist/es5/parquetjs/parquet-thrift/PageType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/RowGroup.js +165 -0
- package/dist/es5/parquetjs/parquet-thrift/RowGroup.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/SchemaElement.js +231 -0
- package/dist/es5/parquetjs/parquet-thrift/SchemaElement.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/SortingColumn.js +115 -0
- package/dist/es5/parquetjs/parquet-thrift/SortingColumn.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/Statistics.js +165 -0
- package/dist/es5/parquetjs/parquet-thrift/Statistics.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/StringType.js +52 -0
- package/dist/es5/parquetjs/parquet-thrift/StringType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/TimeType.js +97 -0
- package/dist/es5/parquetjs/parquet-thrift/TimeType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/TimeUnit.js +126 -0
- package/dist/es5/parquetjs/parquet-thrift/TimeUnit.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/TimestampType.js +97 -0
- package/dist/es5/parquetjs/parquet-thrift/TimestampType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/Type.js +19 -0
- package/dist/es5/parquetjs/parquet-thrift/Type.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/TypeDefinedOrder.js +52 -0
- package/dist/es5/parquetjs/parquet-thrift/TypeDefinedOrder.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/UUIDType.js +52 -0
- package/dist/es5/parquetjs/parquet-thrift/UUIDType.js.map +1 -0
- package/dist/es5/parquetjs/parquet-thrift/index.js +479 -0
- package/dist/es5/parquetjs/parquet-thrift/index.js.map +1 -0
- package/dist/es5/parquetjs/parser/decoders.js +393 -0
- package/dist/es5/parquetjs/parser/decoders.js.map +1 -0
- package/dist/es5/parquetjs/parser/parquet-reader.js +610 -0
- package/dist/es5/parquetjs/parser/parquet-reader.js.map +1 -0
- package/dist/es5/parquetjs/schema/declare.js +21 -0
- package/dist/es5/parquetjs/schema/declare.js.map +1 -0
- package/dist/es5/parquetjs/schema/schema.js +165 -0
- package/dist/es5/parquetjs/schema/schema.js.map +1 -0
- package/dist/es5/parquetjs/schema/shred.js +282 -0
- package/dist/es5/parquetjs/schema/shred.js.map +1 -0
- package/dist/es5/parquetjs/schema/types.js +406 -0
- package/dist/es5/parquetjs/schema/types.js.map +1 -0
- package/dist/es5/parquetjs/utils/file-utils.js +47 -0
- package/dist/es5/parquetjs/utils/file-utils.js.map +1 -0
- package/dist/es5/parquetjs/utils/read-utils.js +120 -0
- package/dist/es5/parquetjs/utils/read-utils.js.map +1 -0
- package/dist/es5/workers/parquet-worker.js +6 -0
- package/dist/es5/workers/parquet-worker.js.map +1 -0
- package/dist/esm/bundle.js +4 -0
- package/dist/esm/bundle.js.map +1 -0
- package/dist/esm/constants.js +6 -0
- package/dist/esm/constants.js.map +1 -0
- package/dist/esm/index.js +31 -0
- package/dist/esm/index.js.map +1 -0
- package/dist/esm/lib/arrow/convert-columns-to-row-group.js +2 -0
- package/dist/esm/lib/arrow/convert-columns-to-row-group.js.map +1 -0
- package/dist/esm/lib/arrow/convert-row-group-to-columns.js +8 -0
- package/dist/esm/lib/arrow/convert-row-group-to-columns.js.map +1 -0
- package/dist/esm/lib/arrow/convert-schema-from-parquet.js +95 -0
- package/dist/esm/lib/arrow/convert-schema-from-parquet.js.map +1 -0
- package/dist/esm/lib/arrow/convert-schema-to-parquet.js +39 -0
- package/dist/esm/lib/arrow/convert-schema-to-parquet.js.map +1 -0
- package/dist/esm/lib/geo/decode-geo-metadata.js +62 -0
- package/dist/esm/lib/geo/decode-geo-metadata.js.map +1 -0
- package/dist/esm/lib/geo/geoparquet-schema.js +76 -0
- package/dist/esm/lib/geo/geoparquet-schema.js.map +1 -0
- package/dist/esm/lib/parsers/parse-parquet-to-columns.js +39 -0
- package/dist/esm/lib/parsers/parse-parquet-to-columns.js.map +1 -0
- package/dist/esm/lib/parsers/parse-parquet-to-rows.js +29 -0
- package/dist/esm/lib/parsers/parse-parquet-to-rows.js.map +1 -0
- package/dist/esm/lib/wasm/encode-parquet-wasm.js +15 -0
- package/dist/esm/lib/wasm/encode-parquet-wasm.js.map +1 -0
- package/dist/esm/lib/wasm/load-wasm/index.js +2 -0
- package/dist/esm/lib/wasm/load-wasm/index.js.map +1 -0
- package/dist/esm/lib/wasm/load-wasm/load-wasm-browser.js +11 -0
- package/dist/esm/lib/wasm/load-wasm/load-wasm-browser.js.map +1 -0
- package/dist/esm/lib/wasm/load-wasm/load-wasm-node.js +5 -0
- package/dist/esm/lib/wasm/load-wasm/load-wasm-node.js.map +1 -0
- package/dist/esm/lib/wasm/parse-parquet-wasm.js +21 -0
- package/dist/esm/lib/wasm/parse-parquet-wasm.js.map +1 -0
- package/dist/esm/lib/wip/convert-schema-deep.java.disabled +910 -0
- package/dist/esm/lib/wip/convert-schema-deep.rs.disabled +976 -0
- package/dist/esm/parquet-loader.js +36 -0
- package/dist/esm/parquet-loader.js.map +1 -0
- package/dist/esm/parquet-wasm-loader.js +22 -0
- package/dist/esm/parquet-wasm-loader.js.map +1 -0
- package/dist/esm/parquet-wasm-writer.js +19 -0
- package/dist/esm/parquet-wasm-writer.js.map +1 -0
- package/dist/esm/parquet-writer.js +17 -0
- package/dist/esm/parquet-writer.js.map +1 -0
- package/dist/esm/parquetjs/LICENSE +20 -0
- package/dist/esm/parquetjs/codecs/declare.js +2 -0
- package/dist/esm/parquetjs/codecs/declare.js.map +1 -0
- package/dist/esm/parquetjs/codecs/dictionary.js +13 -0
- package/dist/esm/parquetjs/codecs/dictionary.js.map +1 -0
- package/dist/esm/parquetjs/codecs/index.js +23 -0
- package/dist/esm/parquetjs/codecs/index.js.map +1 -0
- package/dist/esm/parquetjs/codecs/plain.js +200 -0
- package/dist/esm/parquetjs/codecs/plain.js.map +1 -0
- package/dist/esm/parquetjs/codecs/rle.js +119 -0
- package/dist/esm/parquetjs/codecs/rle.js.map +1 -0
- package/dist/esm/parquetjs/compression.js +61 -0
- package/dist/esm/parquetjs/compression.js.map +1 -0
- package/dist/{parquetjs/encoder/writer.js → esm/parquetjs/encoder/parquet-encoder.js} +8 -106
- package/dist/esm/parquetjs/encoder/parquet-encoder.js.map +1 -0
- package/dist/esm/parquetjs/modules.d.ts +21 -0
- package/dist/esm/parquetjs/parquet-thrift/BoundaryOrder.js +7 -0
- package/dist/esm/parquetjs/parquet-thrift/BoundaryOrder.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/BsonType.js +31 -0
- package/dist/esm/parquetjs/parquet-thrift/BsonType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/ColumnChunk.js +173 -0
- package/dist/esm/parquetjs/parquet-thrift/ColumnChunk.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/ColumnIndex.js +176 -0
- package/dist/esm/parquetjs/parquet-thrift/ColumnIndex.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/ColumnMetaData.js +347 -0
- package/dist/esm/parquetjs/parquet-thrift/ColumnMetaData.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/ColumnOrder.js +77 -0
- package/dist/esm/parquetjs/parquet-thrift/ColumnOrder.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/CompressionCodec.js +12 -0
- package/dist/esm/parquetjs/parquet-thrift/CompressionCodec.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/ConvertedType.js +26 -0
- package/dist/esm/parquetjs/parquet-thrift/ConvertedType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/DataPageHeader.js +132 -0
- package/dist/esm/parquetjs/parquet-thrift/DataPageHeader.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/DataPageHeaderV2.js +187 -0
- package/dist/esm/parquetjs/parquet-thrift/DataPageHeaderV2.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/DateType.js +31 -0
- package/dist/esm/parquetjs/parquet-thrift/DateType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/DecimalType.js +76 -0
- package/dist/esm/parquetjs/parquet-thrift/DecimalType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/DictionaryPageHeader.js +93 -0
- package/dist/esm/parquetjs/parquet-thrift/DictionaryPageHeader.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/Encoding.js +12 -0
- package/dist/esm/parquetjs/parquet-thrift/Encoding.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/EnumType.js +31 -0
- package/dist/esm/parquetjs/parquet-thrift/EnumType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/FieldRepetitionType.js +7 -0
- package/dist/esm/parquetjs/parquet-thrift/FieldRepetitionType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/FileMetaData.js +219 -0
- package/dist/esm/parquetjs/parquet-thrift/FileMetaData.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/IndexPageHeader.js +31 -0
- package/dist/esm/parquetjs/parquet-thrift/IndexPageHeader.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/IntType.js +76 -0
- package/dist/esm/parquetjs/parquet-thrift/IntType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/JsonType.js +31 -0
- package/dist/esm/parquetjs/parquet-thrift/JsonType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/KeyValue.js +74 -0
- package/dist/esm/parquetjs/parquet-thrift/KeyValue.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/ListType.js +31 -0
- package/dist/esm/parquetjs/parquet-thrift/ListType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/LogicalType.js +377 -0
- package/dist/esm/parquetjs/parquet-thrift/LogicalType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/MapType.js +31 -0
- package/dist/esm/parquetjs/parquet-thrift/MapType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/MicroSeconds.js +31 -0
- package/dist/esm/parquetjs/parquet-thrift/MicroSeconds.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/MilliSeconds.js +31 -0
- package/dist/esm/parquetjs/parquet-thrift/MilliSeconds.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/NullType.js +31 -0
- package/dist/esm/parquetjs/parquet-thrift/NullType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/OffsetIndex.js +69 -0
- package/dist/esm/parquetjs/parquet-thrift/OffsetIndex.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/PageEncodingStats.js +95 -0
- package/dist/esm/parquetjs/parquet-thrift/PageEncodingStats.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/PageHeader.js +184 -0
- package/dist/esm/parquetjs/parquet-thrift/PageHeader.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/PageLocation.js +104 -0
- package/dist/esm/parquetjs/parquet-thrift/PageLocation.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/PageType.js +8 -0
- package/dist/esm/parquetjs/parquet-thrift/PageType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/RowGroup.js +145 -0
- package/dist/esm/parquetjs/parquet-thrift/RowGroup.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/SchemaElement.js +211 -0
- package/dist/esm/parquetjs/parquet-thrift/SchemaElement.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/SortingColumn.js +95 -0
- package/dist/esm/parquetjs/parquet-thrift/SortingColumn.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/Statistics.js +145 -0
- package/dist/esm/parquetjs/parquet-thrift/Statistics.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/StringType.js +31 -0
- package/dist/esm/parquetjs/parquet-thrift/StringType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/TimeType.js +77 -0
- package/dist/esm/parquetjs/parquet-thrift/TimeType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/TimeUnit.js +102 -0
- package/dist/esm/parquetjs/parquet-thrift/TimeUnit.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/TimestampType.js +77 -0
- package/dist/esm/parquetjs/parquet-thrift/TimestampType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/Type.js +12 -0
- package/dist/esm/parquetjs/parquet-thrift/Type.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/TypeDefinedOrder.js +31 -0
- package/dist/esm/parquetjs/parquet-thrift/TypeDefinedOrder.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/UUIDType.js +31 -0
- package/dist/esm/parquetjs/parquet-thrift/UUIDType.js.map +1 -0
- package/dist/esm/parquetjs/parquet-thrift/index.js +44 -0
- package/dist/esm/parquetjs/parquet-thrift/index.js.map +1 -0
- package/dist/esm/parquetjs/parser/decoders.js +253 -0
- package/dist/esm/parquetjs/parser/decoders.js.map +1 -0
- package/dist/{parquetjs/parser/parquet-envelope-reader.js → esm/parquetjs/parser/parquet-reader.js} +95 -74
- package/dist/esm/parquetjs/parser/parquet-reader.js.map +1 -0
- package/dist/esm/parquetjs/schema/declare.js +12 -0
- package/dist/esm/parquetjs/schema/declare.js.map +1 -0
- package/dist/esm/parquetjs/schema/schema.js +140 -0
- package/dist/esm/parquetjs/schema/schema.js.map +1 -0
- package/dist/esm/parquetjs/schema/shred.js +228 -0
- package/dist/esm/parquetjs/schema/shred.js.map +1 -0
- package/dist/esm/parquetjs/schema/types.js +397 -0
- package/dist/esm/parquetjs/schema/types.js.map +1 -0
- package/dist/esm/parquetjs/utils/file-utils.js +34 -0
- package/dist/esm/parquetjs/utils/file-utils.js.map +1 -0
- package/dist/esm/parquetjs/utils/read-utils.js +90 -0
- package/dist/esm/parquetjs/utils/read-utils.js.map +1 -0
- package/dist/esm/workers/parquet-worker.js +4 -0
- package/dist/esm/workers/parquet-worker.js.map +1 -0
- package/dist/index.d.ts +16 -20
- package/dist/index.d.ts.map +1 -1
- package/dist/index.js +58 -15
- package/dist/lib/arrow/convert-columns-to-row-group.d.ts +1 -0
- package/dist/lib/arrow/convert-columns-to-row-group.d.ts.map +1 -0
- package/dist/lib/arrow/convert-columns-to-row-group.js +1 -0
- package/dist/lib/arrow/convert-row-group-to-columns.d.ts +4 -0
- package/dist/lib/arrow/convert-row-group-to-columns.d.ts.map +1 -0
- package/dist/lib/arrow/convert-row-group-to-columns.js +12 -0
- package/dist/lib/arrow/convert-schema-from-parquet.d.ts +9 -0
- package/dist/lib/arrow/convert-schema-from-parquet.d.ts.map +1 -0
- package/dist/lib/arrow/convert-schema-from-parquet.js +86 -0
- package/dist/lib/arrow/convert-schema-to-parquet.d.ts +7 -0
- package/dist/lib/arrow/convert-schema-to-parquet.d.ts.map +1 -0
- package/dist/lib/arrow/convert-schema-to-parquet.js +71 -0
- package/dist/lib/geo/decode-geo-metadata.d.ts +31 -0
- package/dist/lib/geo/decode-geo-metadata.d.ts.map +1 -0
- package/dist/lib/geo/decode-geo-metadata.js +77 -0
- package/dist/lib/geo/geoparquet-schema.d.ts +80 -0
- package/dist/lib/geo/geoparquet-schema.d.ts.map +1 -0
- package/dist/lib/geo/geoparquet-schema.js +69 -0
- package/dist/lib/parsers/parse-parquet-to-columns.d.ts +5 -0
- package/dist/lib/parsers/parse-parquet-to-columns.d.ts.map +1 -0
- package/dist/lib/parsers/parse-parquet-to-columns.js +46 -0
- package/dist/lib/parsers/parse-parquet-to-rows.d.ts +5 -0
- package/dist/lib/parsers/parse-parquet-to-rows.d.ts.map +1 -0
- package/dist/lib/parsers/parse-parquet-to-rows.js +37 -0
- package/dist/lib/wasm/encode-parquet-wasm.d.ts +21 -0
- package/dist/lib/wasm/encode-parquet-wasm.d.ts.map +1 -0
- package/dist/lib/wasm/encode-parquet-wasm.js +30 -0
- package/dist/lib/wasm/load-wasm/index.d.ts +2 -0
- package/dist/lib/wasm/load-wasm/index.d.ts.map +1 -0
- package/dist/lib/wasm/load-wasm/index.js +5 -0
- package/dist/lib/wasm/load-wasm/load-wasm-browser.d.ts +3 -0
- package/dist/lib/wasm/load-wasm/load-wasm-browser.d.ts.map +1 -0
- package/dist/lib/wasm/load-wasm/load-wasm-browser.js +38 -0
- package/dist/lib/wasm/load-wasm/load-wasm-node.d.ts +3 -0
- package/dist/lib/wasm/load-wasm/load-wasm-node.d.ts.map +1 -0
- package/dist/lib/wasm/load-wasm/load-wasm-node.js +31 -0
- package/dist/lib/wasm/parse-parquet-wasm.d.ts +10 -0
- package/dist/lib/wasm/parse-parquet-wasm.d.ts.map +1 -0
- package/dist/lib/wasm/parse-parquet-wasm.js +27 -0
- package/dist/parquet-loader.d.ts +6 -15
- package/dist/parquet-loader.d.ts.map +1 -1
- package/dist/parquet-loader.js +38 -19
- package/dist/parquet-wasm-loader.d.ts +23 -0
- package/dist/parquet-wasm-loader.d.ts.map +1 -0
- package/dist/parquet-wasm-loader.js +27 -0
- package/dist/parquet-wasm-writer.d.ts +3 -0
- package/dist/parquet-wasm-writer.d.ts.map +1 -0
- package/dist/parquet-wasm-writer.js +23 -0
- package/dist/parquet-worker.js +27 -25
- package/dist/parquet-worker.js.map +3 -3
- package/dist/parquet-writer.d.ts +3 -2
- package/dist/parquet-writer.d.ts.map +1 -1
- package/dist/parquet-writer.js +18 -14
- package/dist/parquetjs/codecs/declare.js +2 -2
- package/dist/parquetjs/codecs/dictionary.js +12 -10
- package/dist/parquetjs/codecs/index.js +54 -22
- package/dist/parquetjs/codecs/plain.js +173 -232
- package/dist/parquetjs/codecs/rle.js +134 -140
- package/dist/parquetjs/compression.d.ts +3 -0
- package/dist/parquetjs/compression.d.ts.map +1 -1
- package/dist/parquetjs/compression.js +169 -48
- package/dist/parquetjs/encoder/{writer.d.ts → parquet-encoder.d.ts} +15 -23
- package/dist/parquetjs/encoder/parquet-encoder.d.ts.map +1 -0
- package/dist/parquetjs/encoder/parquet-encoder.js +484 -0
- package/dist/parquetjs/parquet-thrift/BoundaryOrder.js +14 -7
- package/dist/parquetjs/parquet-thrift/BsonType.js +60 -37
- package/dist/parquetjs/parquet-thrift/ColumnChunk.js +209 -215
- package/dist/parquetjs/parquet-thrift/ColumnIndex.js +210 -211
- package/dist/parquetjs/parquet-thrift/ColumnMetaData.js +394 -421
- package/dist/parquetjs/parquet-thrift/ColumnOrder.js +102 -89
- package/dist/parquetjs/parquet-thrift/CompressionCodec.js +19 -12
- package/dist/parquetjs/parquet-thrift/ConvertedType.js +33 -26
- package/dist/parquetjs/parquet-thrift/DataPageHeader.js +165 -161
- package/dist/parquetjs/parquet-thrift/DataPageHeaderV2.js +228 -234
- package/dist/parquetjs/parquet-thrift/DateType.js +60 -37
- package/dist/parquetjs/parquet-thrift/DecimalType.js +104 -90
- package/dist/parquetjs/parquet-thrift/DictionaryPageHeader.js +121 -112
- package/dist/parquetjs/parquet-thrift/Encoding.js +19 -12
- package/dist/parquetjs/parquet-thrift/EnumType.js +60 -37
- package/dist/parquetjs/parquet-thrift/FieldRepetitionType.js +14 -7
- package/dist/parquetjs/parquet-thrift/FileMetaData.js +253 -263
- package/dist/parquetjs/parquet-thrift/IndexPageHeader.js +60 -37
- package/dist/parquetjs/parquet-thrift/IntType.js +104 -90
- package/dist/parquetjs/parquet-thrift/JsonType.js +60 -37
- package/dist/parquetjs/parquet-thrift/KeyValue.js +101 -88
- package/dist/parquetjs/parquet-thrift/ListType.js +60 -37
- package/dist/parquetjs/parquet-thrift/LogicalType.js +366 -449
- package/dist/parquetjs/parquet-thrift/MapType.js +60 -37
- package/dist/parquetjs/parquet-thrift/MicroSeconds.js +60 -37
- package/dist/parquetjs/parquet-thrift/MilliSeconds.js +60 -37
- package/dist/parquetjs/parquet-thrift/NullType.js +60 -37
- package/dist/parquetjs/parquet-thrift/OffsetIndex.js +96 -80
- package/dist/parquetjs/parquet-thrift/PageEncodingStats.js +126 -114
- package/dist/parquetjs/parquet-thrift/PageHeader.js +218 -231
- package/dist/parquetjs/parquet-thrift/PageLocation.js +140 -123
- package/dist/parquetjs/parquet-thrift/PageType.js +15 -8
- package/dist/parquetjs/parquet-thrift/RowGroup.js +179 -171
- package/dist/parquetjs/parquet-thrift/SchemaElement.js +241 -268
- package/dist/parquetjs/parquet-thrift/SortingColumn.js +126 -114
- package/dist/parquetjs/parquet-thrift/Statistics.js +175 -178
- package/dist/parquetjs/parquet-thrift/StringType.js +60 -37
- package/dist/parquetjs/parquet-thrift/TimeType.js +105 -91
- package/dist/parquetjs/parquet-thrift/TimeUnit.js +124 -119
- package/dist/parquetjs/parquet-thrift/TimestampType.js +105 -91
- package/dist/parquetjs/parquet-thrift/Type.js +19 -12
- package/dist/parquetjs/parquet-thrift/TypeDefinedOrder.js +60 -37
- package/dist/parquetjs/parquet-thrift/UUIDType.js +60 -37
- package/dist/parquetjs/parquet-thrift/index.js +65 -44
- package/dist/parquetjs/parser/decoders.d.ts +2 -2
- package/dist/parquetjs/parser/decoders.d.ts.map +1 -1
- package/dist/parquetjs/parser/decoders.js +301 -283
- package/dist/parquetjs/parser/parquet-reader.d.ts +47 -57
- package/dist/parquetjs/parser/parquet-reader.d.ts.map +1 -1
- package/dist/parquetjs/parser/parquet-reader.js +193 -113
- package/dist/parquetjs/schema/declare.d.ts +26 -18
- package/dist/parquetjs/schema/declare.d.ts.map +1 -1
- package/dist/parquetjs/schema/declare.js +11 -12
- package/dist/parquetjs/schema/schema.d.ts +4 -4
- package/dist/parquetjs/schema/schema.d.ts.map +1 -1
- package/dist/parquetjs/schema/schema.js +148 -162
- package/dist/parquetjs/schema/shred.d.ts +33 -12
- package/dist/parquetjs/schema/shred.d.ts.map +1 -1
- package/dist/parquetjs/schema/shred.js +340 -147
- package/dist/parquetjs/schema/types.d.ts +2 -2
- package/dist/parquetjs/schema/types.d.ts.map +1 -1
- package/dist/parquetjs/schema/types.js +355 -415
- package/dist/parquetjs/utils/file-utils.d.ts +5 -4
- package/dist/parquetjs/utils/file-utils.d.ts.map +1 -1
- package/dist/parquetjs/utils/file-utils.js +37 -28
- package/dist/parquetjs/utils/read-utils.js +99 -95
- package/dist/workers/parquet-worker.js +5 -4
- package/package.json +17 -12
- package/src/index.ts +58 -7
- package/src/lib/arrow/convert-columns-to-row-group.ts +0 -0
- package/src/lib/arrow/convert-row-group-to-columns.ts +15 -0
- package/src/lib/arrow/convert-schema-from-parquet.ts +104 -0
- package/src/lib/arrow/convert-schema-to-parquet.ts +90 -0
- package/src/lib/geo/decode-geo-metadata.ts +108 -0
- package/src/lib/geo/geoparquet-schema.ts +69 -0
- package/src/lib/parsers/parse-parquet-to-columns.ts +60 -0
- package/src/lib/parsers/parse-parquet-to-rows.ts +45 -0
- package/src/lib/wasm/encode-parquet-wasm.ts +40 -0
- package/src/lib/wasm/load-wasm/index.ts +1 -0
- package/src/lib/wasm/load-wasm/load-wasm-browser.ts +15 -0
- package/src/lib/wasm/load-wasm/load-wasm-node.ts +5 -0
- package/src/lib/wasm/parse-parquet-wasm.ts +42 -0
- package/src/lib/wip/convert-schema-deep.java.disabled +910 -0
- package/src/lib/wip/convert-schema-deep.rs.disabled +976 -0
- package/src/parquet-loader.ts +30 -3
- package/src/parquet-wasm-loader.ts +36 -0
- package/src/parquet-wasm-writer.ts +24 -0
- package/src/parquet-writer.ts +4 -1
- package/src/parquetjs/compression.ts +24 -7
- package/src/parquetjs/encoder/{writer.ts → parquet-encoder.ts} +33 -38
- package/src/parquetjs/parser/decoders.ts +3 -3
- package/src/parquetjs/parser/parquet-reader.ts +239 -122
- package/src/parquetjs/schema/declare.ts +22 -13
- package/src/parquetjs/schema/schema.ts +8 -8
- package/src/parquetjs/schema/shred.ts +239 -71
- package/src/parquetjs/schema/types.ts +25 -30
- package/src/parquetjs/utils/file-utils.ts +3 -4
- package/dist/bundle.js.map +0 -1
- package/dist/constants.js.map +0 -1
- package/dist/index.js.map +0 -1
- package/dist/lib/convert-schema.d.ts +0 -8
- package/dist/lib/convert-schema.d.ts.map +0 -1
- package/dist/lib/convert-schema.js +0 -71
- package/dist/lib/convert-schema.js.map +0 -1
- package/dist/lib/parse-parquet.d.ts +0 -4
- package/dist/lib/parse-parquet.d.ts.map +0 -1
- package/dist/lib/parse-parquet.js +0 -28
- package/dist/lib/parse-parquet.js.map +0 -1
- package/dist/lib/read-array-buffer.d.ts +0 -19
- package/dist/lib/read-array-buffer.d.ts.map +0 -1
- package/dist/lib/read-array-buffer.js +0 -9
- package/dist/lib/read-array-buffer.js.map +0 -1
- package/dist/parquet-loader.js.map +0 -1
- package/dist/parquet-writer.js.map +0 -1
- package/dist/parquetjs/codecs/declare.js.map +0 -1
- package/dist/parquetjs/codecs/dictionary.js.map +0 -1
- package/dist/parquetjs/codecs/index.js.map +0 -1
- package/dist/parquetjs/codecs/plain.js.map +0 -1
- package/dist/parquetjs/codecs/rle.js.map +0 -1
- package/dist/parquetjs/compression.js.map +0 -1
- package/dist/parquetjs/encoder/writer.d.ts.map +0 -1
- package/dist/parquetjs/encoder/writer.js.map +0 -1
- package/dist/parquetjs/file.d.ts +0 -10
- package/dist/parquetjs/file.d.ts.map +0 -1
- package/dist/parquetjs/file.js +0 -80
- package/dist/parquetjs/file.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/BoundaryOrder.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/BsonType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/ColumnChunk.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/ColumnIndex.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/ColumnMetaData.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/ColumnOrder.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/CompressionCodec.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/ConvertedType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/DataPageHeader.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/DataPageHeaderV2.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/DateType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/DecimalType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/DictionaryPageHeader.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/Encoding.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/EnumType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/FieldRepetitionType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/FileMetaData.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/IndexPageHeader.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/IntType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/JsonType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/KeyValue.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/ListType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/LogicalType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/MapType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/MicroSeconds.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/MilliSeconds.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/NullType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/OffsetIndex.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/PageEncodingStats.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/PageHeader.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/PageLocation.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/PageType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/RowGroup.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/SchemaElement.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/SortingColumn.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/Statistics.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/StringType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/TimeType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/TimeUnit.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/TimestampType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/Type.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/TypeDefinedOrder.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/UUIDType.js.map +0 -1
- package/dist/parquetjs/parquet-thrift/index.js.map +0 -1
- package/dist/parquetjs/parser/decoders.js.map +0 -1
- package/dist/parquetjs/parser/parquet-cursor.d.ts +0 -36
- package/dist/parquetjs/parser/parquet-cursor.d.ts.map +0 -1
- package/dist/parquetjs/parser/parquet-cursor.js +0 -90
- package/dist/parquetjs/parser/parquet-cursor.js.map +0 -1
- package/dist/parquetjs/parser/parquet-envelope-reader.d.ts +0 -40
- package/dist/parquetjs/parser/parquet-envelope-reader.d.ts.map +0 -1
- package/dist/parquetjs/parser/parquet-envelope-reader.js.map +0 -1
- package/dist/parquetjs/parser/parquet-reader.js.map +0 -1
- package/dist/parquetjs/schema/declare.js.map +0 -1
- package/dist/parquetjs/schema/schema.js.map +0 -1
- package/dist/parquetjs/schema/shred.js.map +0 -1
- package/dist/parquetjs/schema/types.js.map +0 -1
- package/dist/parquetjs/utils/buffer-utils.d.ts +0 -10
- package/dist/parquetjs/utils/buffer-utils.d.ts.map +0 -1
- package/dist/parquetjs/utils/buffer-utils.js +0 -12
- package/dist/parquetjs/utils/buffer-utils.js.map +0 -1
- package/dist/parquetjs/utils/file-utils.js.map +0 -1
- package/dist/parquetjs/utils/read-utils.js.map +0 -1
- package/dist/workers/parquet-worker.js.map +0 -1
- package/src/lib/convert-schema.ts +0 -95
- package/src/lib/parse-parquet.ts +0 -27
- package/src/lib/read-array-buffer.ts +0 -31
- package/src/parquetjs/file.ts +0 -90
- package/src/parquetjs/parser/parquet-cursor.ts +0 -94
- package/src/parquetjs/parser/parquet-envelope-reader.ts +0 -199
- package/src/parquetjs/utils/buffer-utils.ts +0 -18
- /package/dist/{parquetjs → es5/parquetjs}/LICENSE +0 -0
- /package/dist/{parquetjs → es5/parquetjs}/modules.d.ts +0 -0
|
@@ -1,6 +1,7 @@
|
|
|
1
1
|
/// <reference types="node" />
|
|
2
|
-
|
|
3
|
-
|
|
2
|
+
/// <reference types="node" />
|
|
3
|
+
/// <reference types="node" />
|
|
4
|
+
import { fs, stream } from '@loaders.gl/loader-utils';
|
|
4
5
|
export declare function load(name: string): any;
|
|
5
6
|
export interface WriteStreamOptions {
|
|
6
7
|
flags?: string;
|
|
@@ -10,7 +11,7 @@ export interface WriteStreamOptions {
|
|
|
10
11
|
autoClose?: boolean;
|
|
11
12
|
start?: number;
|
|
12
13
|
}
|
|
13
|
-
export declare function oswrite(os: Writable, buf: Buffer): Promise<void>;
|
|
14
|
-
export declare function osclose(os: Writable): Promise<void>;
|
|
14
|
+
export declare function oswrite(os: stream.Writable, buf: Buffer): Promise<void>;
|
|
15
|
+
export declare function osclose(os: stream.Writable): Promise<void>;
|
|
15
16
|
export declare function osopen(path: string, opts?: WriteStreamOptions): Promise<fs.WriteStream>;
|
|
16
17
|
//# sourceMappingURL=file-utils.d.ts.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"file-utils.d.ts","sourceRoot":"","sources":["../../../src/parquetjs/utils/file-utils.ts"],"names":[],"mappings":"
|
|
1
|
+
{"version":3,"file":"file-utils.d.ts","sourceRoot":"","sources":["../../../src/parquetjs/utils/file-utils.ts"],"names":[],"mappings":";;;AACA,OAAO,EAAC,EAAE,EAAE,MAAM,EAAC,MAAM,0BAA0B,CAAC;AAEpD,wBAAgB,IAAI,CAAC,IAAI,EAAE,MAAM,GAAG,GAAG,CAEtC;AACD,MAAM,WAAW,kBAAkB;IACjC,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,EAAE,CAAC,EAAE,MAAM,CAAC;IACZ,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,SAAS,CAAC,EAAE,OAAO,CAAC;IACpB,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAED,wBAAgB,OAAO,CAAC,EAAE,EAAE,MAAM,CAAC,QAAQ,EAAE,GAAG,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC,CAUvE;AAED,wBAAgB,OAAO,CAAC,EAAE,EAAE,MAAM,CAAC,QAAQ,GAAG,OAAO,CAAC,IAAI,CAAC,CAU1D;AAED,wBAAgB,MAAM,CAAC,IAAI,EAAE,MAAM,EAAE,IAAI,CAAC,EAAE,kBAAkB,GAAG,OAAO,CAAC,EAAE,CAAC,WAAW,CAAC,CAMvF"}
|
|
@@ -1,34 +1,43 @@
|
|
|
1
|
-
|
|
2
|
-
|
|
3
|
-
|
|
1
|
+
"use strict";
|
|
2
|
+
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
+
exports.osopen = exports.osclose = exports.oswrite = exports.load = void 0;
|
|
4
|
+
// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)
|
|
5
|
+
const loader_utils_1 = require("@loaders.gl/loader-utils");
|
|
6
|
+
function load(name) {
|
|
7
|
+
return (module || global).require(name);
|
|
4
8
|
}
|
|
5
|
-
|
|
6
|
-
|
|
7
|
-
|
|
8
|
-
|
|
9
|
-
|
|
10
|
-
|
|
11
|
-
|
|
12
|
-
|
|
9
|
+
exports.load = load;
|
|
10
|
+
function oswrite(os, buf) {
|
|
11
|
+
return new Promise((resolve, reject) => {
|
|
12
|
+
os.write(buf, (err) => {
|
|
13
|
+
if (err) {
|
|
14
|
+
reject(err);
|
|
15
|
+
}
|
|
16
|
+
else {
|
|
17
|
+
resolve();
|
|
18
|
+
}
|
|
19
|
+
});
|
|
13
20
|
});
|
|
14
|
-
});
|
|
15
21
|
}
|
|
16
|
-
|
|
17
|
-
|
|
18
|
-
|
|
19
|
-
|
|
20
|
-
|
|
21
|
-
|
|
22
|
-
|
|
23
|
-
|
|
22
|
+
exports.oswrite = oswrite;
|
|
23
|
+
function osclose(os) {
|
|
24
|
+
return new Promise((resolve, reject) => {
|
|
25
|
+
os.close((err) => {
|
|
26
|
+
if (err) {
|
|
27
|
+
reject(err);
|
|
28
|
+
}
|
|
29
|
+
else {
|
|
30
|
+
resolve();
|
|
31
|
+
}
|
|
32
|
+
});
|
|
24
33
|
});
|
|
25
|
-
});
|
|
26
34
|
}
|
|
27
|
-
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
32
|
-
|
|
35
|
+
exports.osclose = osclose;
|
|
36
|
+
function osopen(path, opts) {
|
|
37
|
+
return new Promise((resolve, reject) => {
|
|
38
|
+
const outputStream = loader_utils_1.fs.createWriteStream(path, opts);
|
|
39
|
+
outputStream.once('open', (fd) => resolve(outputStream));
|
|
40
|
+
outputStream.once('error', (err) => reject(err));
|
|
41
|
+
});
|
|
33
42
|
}
|
|
34
|
-
|
|
43
|
+
exports.osopen = osopen;
|
|
@@ -1,105 +1,109 @@
|
|
|
1
|
-
|
|
2
|
-
|
|
3
|
-
|
|
4
|
-
|
|
5
|
-
|
|
6
|
-
|
|
7
|
-
|
|
8
|
-
|
|
9
|
-
|
|
10
|
-
|
|
11
|
-
|
|
12
|
-
}
|
|
13
|
-
|
|
14
|
-
export function serializeThrift(obj) {
|
|
15
|
-
const output = [];
|
|
16
|
-
const transport = new TBufferedTransport(undefined, buf => {
|
|
17
|
-
output.push(buf);
|
|
18
|
-
});
|
|
19
|
-
const protocol = new TCompactProtocol(transport);
|
|
20
|
-
obj.write(protocol);
|
|
21
|
-
transport.flush();
|
|
22
|
-
return Buffer.concat(output);
|
|
1
|
+
"use strict";
|
|
2
|
+
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
+
exports.fieldIndexOf = exports.getBitWidth = exports.decodePageHeader = exports.decodeFileMetadata = exports.getThriftEnum = exports.decodeThrift = exports.serializeThrift = void 0;
|
|
4
|
+
const thrift_1 = require("thrift");
|
|
5
|
+
const parquet_thrift_1 = require("../parquet-thrift");
|
|
6
|
+
class UFramedTransport extends thrift_1.TFramedTransport {
|
|
7
|
+
constructor() {
|
|
8
|
+
super(...arguments);
|
|
9
|
+
this.readPos = 0;
|
|
10
|
+
}
|
|
23
11
|
}
|
|
24
|
-
|
|
25
|
-
|
|
26
|
-
|
|
27
|
-
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
32
|
-
|
|
33
|
-
|
|
12
|
+
/**
|
|
13
|
+
* Helper function that serializes a thrift object into a buffer
|
|
14
|
+
*/
|
|
15
|
+
function serializeThrift(obj) {
|
|
16
|
+
const output = [];
|
|
17
|
+
const transport = new thrift_1.TBufferedTransport(undefined, (buf) => {
|
|
18
|
+
output.push(buf);
|
|
19
|
+
});
|
|
20
|
+
const protocol = new thrift_1.TCompactProtocol(transport);
|
|
21
|
+
obj.write(protocol);
|
|
22
|
+
transport.flush();
|
|
23
|
+
return Buffer.concat(output);
|
|
34
24
|
}
|
|
35
|
-
|
|
36
|
-
|
|
37
|
-
if (
|
|
38
|
-
|
|
25
|
+
exports.serializeThrift = serializeThrift;
|
|
26
|
+
function decodeThrift(obj, buf, offset) {
|
|
27
|
+
if (!offset) {
|
|
28
|
+
// tslint:disable-next-line:no-parameter-reassignment
|
|
29
|
+
offset = 0;
|
|
39
30
|
}
|
|
40
|
-
|
|
41
|
-
|
|
42
|
-
|
|
31
|
+
const transport = new UFramedTransport(buf);
|
|
32
|
+
transport.readPos = offset;
|
|
33
|
+
const protocol = new thrift_1.TCompactProtocol(transport);
|
|
34
|
+
obj.read(protocol);
|
|
35
|
+
return transport.readPos - offset;
|
|
43
36
|
}
|
|
44
|
-
|
|
45
|
-
|
|
46
|
-
|
|
47
|
-
|
|
48
|
-
|
|
49
|
-
|
|
50
|
-
|
|
51
|
-
|
|
52
|
-
|
|
53
|
-
|
|
54
|
-
|
|
55
|
-
metadata
|
|
56
|
-
};
|
|
37
|
+
exports.decodeThrift = decodeThrift;
|
|
38
|
+
/**
|
|
39
|
+
* FIXME not ideal that this is linear
|
|
40
|
+
*/
|
|
41
|
+
function getThriftEnum(klass, value) {
|
|
42
|
+
for (const k in klass) {
|
|
43
|
+
if (klass[k] === value) {
|
|
44
|
+
return k;
|
|
45
|
+
}
|
|
46
|
+
}
|
|
47
|
+
throw new Error('Invalid ENUM value');
|
|
57
48
|
}
|
|
58
|
-
|
|
59
|
-
|
|
60
|
-
offset
|
|
61
|
-
|
|
62
|
-
|
|
63
|
-
|
|
64
|
-
|
|
65
|
-
|
|
66
|
-
|
|
67
|
-
|
|
68
|
-
length: transport.readPos - offset,
|
|
69
|
-
pageHeader
|
|
70
|
-
};
|
|
49
|
+
exports.getThriftEnum = getThriftEnum;
|
|
50
|
+
function decodeFileMetadata(buf, offset) {
|
|
51
|
+
if (!offset) {
|
|
52
|
+
// tslint:disable-next-line:no-parameter-reassignment
|
|
53
|
+
offset = 0;
|
|
54
|
+
}
|
|
55
|
+
const transport = new UFramedTransport(buf);
|
|
56
|
+
transport.readPos = offset;
|
|
57
|
+
const protocol = new thrift_1.TCompactProtocol(transport);
|
|
58
|
+
const metadata = parquet_thrift_1.FileMetaData.read(protocol);
|
|
59
|
+
return { length: transport.readPos - offset, metadata };
|
|
71
60
|
}
|
|
72
|
-
|
|
73
|
-
|
|
74
|
-
|
|
75
|
-
|
|
76
|
-
|
|
77
|
-
|
|
61
|
+
exports.decodeFileMetadata = decodeFileMetadata;
|
|
62
|
+
function decodePageHeader(buf, offset) {
|
|
63
|
+
if (!offset) {
|
|
64
|
+
// tslint:disable-next-line:no-parameter-reassignment
|
|
65
|
+
offset = 0;
|
|
66
|
+
}
|
|
67
|
+
const transport = new UFramedTransport(buf);
|
|
68
|
+
transport.readPos = offset;
|
|
69
|
+
const protocol = new thrift_1.TCompactProtocol(transport);
|
|
70
|
+
const pageHeader = parquet_thrift_1.PageHeader.read(protocol);
|
|
71
|
+
return { length: transport.readPos - offset, pageHeader };
|
|
78
72
|
}
|
|
79
|
-
|
|
80
|
-
|
|
81
|
-
|
|
82
|
-
|
|
73
|
+
exports.decodePageHeader = decodePageHeader;
|
|
74
|
+
/**
|
|
75
|
+
* Get the number of bits required to store a given value
|
|
76
|
+
*/
|
|
77
|
+
function getBitWidth(val) {
|
|
78
|
+
if (val === 0) {
|
|
79
|
+
return 0;
|
|
80
|
+
// tslint:disable-next-line:no-else-after-return
|
|
83
81
|
}
|
|
84
|
-
|
|
85
|
-
|
|
86
|
-
|
|
87
|
-
|
|
88
|
-
|
|
89
|
-
|
|
90
|
-
|
|
91
|
-
|
|
92
|
-
|
|
93
|
-
|
|
94
|
-
|
|
95
|
-
|
|
96
|
-
|
|
97
|
-
|
|
82
|
+
return Math.ceil(Math.log2(val + 1));
|
|
83
|
+
}
|
|
84
|
+
exports.getBitWidth = getBitWidth;
|
|
85
|
+
// Supports MQTT path wildcards
|
|
86
|
+
// + all immediate children
|
|
87
|
+
// # all descendents
|
|
88
|
+
function fieldIndexOf(arr, elem) {
|
|
89
|
+
for (let j = 0; j < arr.length; j++) {
|
|
90
|
+
if (arr[j].length > elem.length) {
|
|
91
|
+
continue; // eslint-disable-line no-continue
|
|
92
|
+
}
|
|
93
|
+
let m = true;
|
|
94
|
+
for (let i = 0; i < elem.length; i++) {
|
|
95
|
+
if (arr[j][i] === elem[i] || arr[j][i] === '+' || arr[j][i] === '#') {
|
|
96
|
+
continue; // eslint-disable-line no-continue
|
|
97
|
+
}
|
|
98
|
+
if (i >= arr[j].length && arr[j][arr[j].length - 1] === '#') {
|
|
99
|
+
continue; // eslint-disable-line no-continue
|
|
100
|
+
}
|
|
101
|
+
m = false;
|
|
102
|
+
break;
|
|
103
|
+
}
|
|
104
|
+
if (m)
|
|
105
|
+
return j;
|
|
98
106
|
}
|
|
99
|
-
|
|
100
|
-
if (m) return j;
|
|
101
|
-
}
|
|
102
|
-
|
|
103
|
-
return -1;
|
|
107
|
+
return -1;
|
|
104
108
|
}
|
|
105
|
-
|
|
109
|
+
exports.fieldIndexOf = fieldIndexOf;
|
|
@@ -1,4 +1,5 @@
|
|
|
1
|
-
|
|
2
|
-
|
|
3
|
-
|
|
4
|
-
|
|
1
|
+
"use strict";
|
|
2
|
+
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
+
const loader_utils_1 = require("@loaders.gl/loader-utils");
|
|
4
|
+
const index_1 = require("../index");
|
|
5
|
+
(0, loader_utils_1.createLoaderWorker)(index_1.ParquetLoader);
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@loaders.gl/parquet",
|
|
3
|
-
"version": "4.0.0-alpha.
|
|
3
|
+
"version": "4.0.0-alpha.6",
|
|
4
4
|
"description": "Framework-independent loader for Apache Parquet files",
|
|
5
5
|
"license": "MIT",
|
|
6
6
|
"publishConfig": {
|
|
@@ -19,8 +19,8 @@
|
|
|
19
19
|
"Apache Parquet"
|
|
20
20
|
],
|
|
21
21
|
"types": "dist/index.d.ts",
|
|
22
|
-
"main": "dist/index.js",
|
|
23
|
-
"module": "dist/index.js",
|
|
22
|
+
"main": "dist/es5/index.js",
|
|
23
|
+
"module": "dist/esm/index.js",
|
|
24
24
|
"sideEffects": false,
|
|
25
25
|
"files": [
|
|
26
26
|
"src",
|
|
@@ -36,31 +36,36 @@
|
|
|
36
36
|
"child_process": false,
|
|
37
37
|
"net": false,
|
|
38
38
|
"tls": false,
|
|
39
|
-
"
|
|
39
|
+
"stream": false,
|
|
40
|
+
"fs": false,
|
|
41
|
+
"./src/lib/wasm/load-wasm/load-wasm-node.ts": "./src/lib/wasm/load-wasm/load-wasm-browser.ts"
|
|
40
42
|
},
|
|
41
43
|
"dependencies": {
|
|
42
|
-
"@loaders.gl/
|
|
43
|
-
"@loaders.gl/
|
|
44
|
-
"@loaders.gl/
|
|
44
|
+
"@loaders.gl/bson": "4.0.0-alpha.6",
|
|
45
|
+
"@loaders.gl/compression": "4.0.0-alpha.6",
|
|
46
|
+
"@loaders.gl/loader-utils": "4.0.0-alpha.6",
|
|
47
|
+
"@loaders.gl/schema": "4.0.0-alpha.6",
|
|
45
48
|
"async-mutex": "^0.2.2",
|
|
46
49
|
"brotli": "^1.3.2",
|
|
47
|
-
"bson": "^1.0.4",
|
|
48
50
|
"int53": "^0.2.4",
|
|
49
51
|
"lz4js": "^0.2.0",
|
|
50
|
-
"lzo": "^0.4.11",
|
|
51
52
|
"node-int64": "^0.4.0",
|
|
52
53
|
"object-stream": "0.0.1",
|
|
54
|
+
"parquet-wasm": "^0.3.1",
|
|
53
55
|
"snappyjs": "^0.6.0",
|
|
54
56
|
"thrift": "^0.14.2",
|
|
55
57
|
"varint": "^5.0.0",
|
|
56
58
|
"zstd-codec": "^0.1"
|
|
57
59
|
},
|
|
60
|
+
"peerDependencies": {
|
|
61
|
+
"apache-arrow": "^9.0.0"
|
|
62
|
+
},
|
|
58
63
|
"devDependencies": {
|
|
59
|
-
"@types/bson": "^4.0.0",
|
|
60
64
|
"@types/node": "^10.14.15",
|
|
61
65
|
"@types/node-int64": "^0.4.29",
|
|
62
66
|
"@types/thrift": "^0.10.8",
|
|
63
|
-
"@types/varint": "^5.0.0"
|
|
67
|
+
"@types/varint": "^5.0.0",
|
|
68
|
+
"apache-arrow": "^9.0.0"
|
|
64
69
|
},
|
|
65
|
-
"gitHead": "
|
|
70
|
+
"gitHead": "acc1985050dfaa0f1f0c066f8da5bce7454a046c"
|
|
66
71
|
}
|
package/src/index.ts
CHANGED
|
@@ -1,22 +1,64 @@
|
|
|
1
|
+
// loaders.gl, MIT license
|
|
2
|
+
|
|
1
3
|
import type {LoaderWithParser} from '@loaders.gl/loader-utils';
|
|
4
|
+
import type {
|
|
5
|
+
ObjectRowTable,
|
|
6
|
+
ObjectRowTableBatch,
|
|
7
|
+
ColumnarTable,
|
|
8
|
+
ColumnarTableBatch
|
|
9
|
+
} from '@loaders.gl/schema';
|
|
10
|
+
import type {Table as ArrowTable} from 'apache-arrow';
|
|
2
11
|
|
|
3
12
|
// ParquetLoader
|
|
4
13
|
|
|
5
|
-
import {
|
|
6
|
-
|
|
14
|
+
import {
|
|
15
|
+
ParquetLoader as ParquetWorkerLoader,
|
|
16
|
+
ParquetLoader as ParquetColumnarWorkerLoader,
|
|
17
|
+
ParquetLoaderOptions
|
|
18
|
+
} from './parquet-loader';
|
|
19
|
+
import {parseParquet, parseParquetFileInBatches} from './lib/parsers/parse-parquet-to-rows';
|
|
20
|
+
import {
|
|
21
|
+
parseParquetInColumns,
|
|
22
|
+
parseParquetFileInColumnarBatches
|
|
23
|
+
} from './lib/parsers/parse-parquet-to-columns';
|
|
24
|
+
|
|
25
|
+
import {parseParquetWasm, ParquetWasmLoaderOptions} from './lib/wasm/parse-parquet-wasm';
|
|
26
|
+
import {ParquetWasmLoader as ParquetWasmWorkerLoader} from './parquet-wasm-loader';
|
|
7
27
|
|
|
8
|
-
export {ParquetWorkerLoader};
|
|
28
|
+
export {ParquetWorkerLoader, ParquetWasmWorkerLoader};
|
|
9
29
|
|
|
10
30
|
/** ParquetJS table loader */
|
|
11
|
-
export const ParquetLoader
|
|
31
|
+
export const ParquetLoader: LoaderWithParser<
|
|
32
|
+
ObjectRowTable,
|
|
33
|
+
ObjectRowTableBatch,
|
|
34
|
+
ParquetLoaderOptions
|
|
35
|
+
> = {
|
|
12
36
|
...ParquetWorkerLoader,
|
|
13
37
|
parse: parseParquet,
|
|
14
38
|
parseFileInBatches: parseParquetFileInBatches
|
|
15
39
|
};
|
|
16
40
|
|
|
41
|
+
/** ParquetJS table loader */
|
|
42
|
+
// @ts-expect-error
|
|
43
|
+
export const ParquetColumnarLoader: LoaderWithParser<
|
|
44
|
+
ColumnarTable,
|
|
45
|
+
ColumnarTableBatch,
|
|
46
|
+
ParquetLoaderOptions
|
|
47
|
+
> = {
|
|
48
|
+
...ParquetColumnarWorkerLoader,
|
|
49
|
+
parse: parseParquetInColumns,
|
|
50
|
+
parseFileInBatches: parseParquetFileInColumnarBatches
|
|
51
|
+
};
|
|
52
|
+
|
|
53
|
+
export const ParquetWasmLoader: LoaderWithParser<ArrowTable, never, ParquetWasmLoaderOptions> = {
|
|
54
|
+
...ParquetWasmWorkerLoader,
|
|
55
|
+
parse: parseParquetWasm
|
|
56
|
+
};
|
|
57
|
+
|
|
17
58
|
// ParquetWriter
|
|
18
59
|
|
|
19
60
|
export {ParquetWriter as _ParquetWriter} from './parquet-writer';
|
|
61
|
+
export {ParquetWasmWriter} from './parquet-wasm-writer';
|
|
20
62
|
|
|
21
63
|
// EXPERIMENTAL - expose the internal parquetjs API
|
|
22
64
|
|
|
@@ -24,9 +66,18 @@ export {preloadCompressions} from './parquetjs/compression';
|
|
|
24
66
|
|
|
25
67
|
export {ParquetSchema} from './parquetjs/schema/schema';
|
|
26
68
|
export {ParquetReader} from './parquetjs/parser/parquet-reader';
|
|
27
|
-
export {
|
|
28
|
-
|
|
29
|
-
export {
|
|
69
|
+
export {ParquetEncoder} from './parquetjs/encoder/parquet-encoder';
|
|
70
|
+
|
|
71
|
+
export {
|
|
72
|
+
convertParquetSchema,
|
|
73
|
+
convertParquetSchema as convertParquetToArrowSchema
|
|
74
|
+
} from './lib/arrow/convert-schema-from-parquet';
|
|
30
75
|
|
|
31
76
|
// TESTS
|
|
32
77
|
export const _typecheckParquetLoader: LoaderWithParser = ParquetLoader;
|
|
78
|
+
|
|
79
|
+
// Geo Metadata
|
|
80
|
+
export {default as geoJSONSchema} from './lib/geo/geoparquet-schema';
|
|
81
|
+
|
|
82
|
+
export type {GeoMetadata} from './lib/geo/decode-geo-metadata';
|
|
83
|
+
export {getGeoMetadata, setGeoMetadata, unpackGeoMetadata} from './lib/geo/decode-geo-metadata';
|
|
File without changes
|
|
@@ -0,0 +1,15 @@
|
|
|
1
|
+
// loaders.gl, MIT license
|
|
2
|
+
|
|
3
|
+
import {Schema} from '@loaders.gl/schema';
|
|
4
|
+
import {ParquetRowGroup} from '@loaders.gl/parquet/parquetjs/schema/declare';
|
|
5
|
+
|
|
6
|
+
export function convertParquetRowGroupToColumns(
|
|
7
|
+
schema: Schema,
|
|
8
|
+
rowGroup: ParquetRowGroup
|
|
9
|
+
): Record<string, any[]> {
|
|
10
|
+
const columns: Record<string, any[]> = {};
|
|
11
|
+
for (const [columnName, data] of Object.entries(rowGroup.columnData)) {
|
|
12
|
+
columns[columnName] = columns[columnName] || data.values;
|
|
13
|
+
}
|
|
14
|
+
return columns;
|
|
15
|
+
}
|
|
@@ -0,0 +1,104 @@
|
|
|
1
|
+
// loaders.gl, MIT license
|
|
2
|
+
|
|
3
|
+
import {Schema, Field, DataType} from '@loaders.gl/schema';
|
|
4
|
+
|
|
5
|
+
import type {ParquetSchema} from '../../parquetjs/schema/schema';
|
|
6
|
+
import type {FieldDefinition, ParquetField, ParquetType} from '../../parquetjs/schema/declare';
|
|
7
|
+
import {FileMetaData} from '../../parquetjs/parquet-thrift';
|
|
8
|
+
|
|
9
|
+
export const PARQUET_TYPE_MAPPING: {[type in ParquetType]: DataType} = {
|
|
10
|
+
BOOLEAN: 'bool',
|
|
11
|
+
INT32: 'int32',
|
|
12
|
+
INT64: 'float64',
|
|
13
|
+
INT96: 'float64',
|
|
14
|
+
FLOAT: 'float32',
|
|
15
|
+
DOUBLE: 'float64',
|
|
16
|
+
BYTE_ARRAY: 'binary',
|
|
17
|
+
FIXED_LEN_BYTE_ARRAY: 'binary',
|
|
18
|
+
UTF8: 'utf8',
|
|
19
|
+
DATE: 'int32',
|
|
20
|
+
TIME_MILLIS: 'int64',
|
|
21
|
+
TIME_MICROS: 'int64',
|
|
22
|
+
TIMESTAMP_MILLIS: 'int64',
|
|
23
|
+
TIMESTAMP_MICROS: 'int64',
|
|
24
|
+
UINT_8: 'int32',
|
|
25
|
+
UINT_16: 'uint16',
|
|
26
|
+
UINT_32: 'uint32',
|
|
27
|
+
UINT_64: 'uint64',
|
|
28
|
+
INT_8: 'int8',
|
|
29
|
+
INT_16: 'int16',
|
|
30
|
+
INT_32: 'int32',
|
|
31
|
+
INT_64: 'int64',
|
|
32
|
+
JSON: 'binary',
|
|
33
|
+
BSON: 'binary',
|
|
34
|
+
// TODO check interal type
|
|
35
|
+
INTERVAL: 'binary',
|
|
36
|
+
DECIMAL_INT32: 'float32',
|
|
37
|
+
DECIMAL_INT64: 'float64',
|
|
38
|
+
DECIMAL_BYTE_ARRAY: 'float64',
|
|
39
|
+
DECIMAL_FIXED_LEN_BYTE_ARRAY: 'float64'
|
|
40
|
+
};
|
|
41
|
+
|
|
42
|
+
export function convertParquetSchema(
|
|
43
|
+
parquetSchema: ParquetSchema,
|
|
44
|
+
parquetMetadata: FileMetaData | null
|
|
45
|
+
): Schema {
|
|
46
|
+
const fields = getFields(parquetSchema.schema);
|
|
47
|
+
const metadata = parquetMetadata && getSchemaMetadata(parquetMetadata);
|
|
48
|
+
|
|
49
|
+
const schema: Schema = {
|
|
50
|
+
fields,
|
|
51
|
+
metadata: metadata || {}
|
|
52
|
+
};
|
|
53
|
+
|
|
54
|
+
return schema;
|
|
55
|
+
}
|
|
56
|
+
|
|
57
|
+
function getFields(schema: FieldDefinition): Field[] {
|
|
58
|
+
const fields: Field[] = [];
|
|
59
|
+
|
|
60
|
+
for (const name in schema) {
|
|
61
|
+
const field = schema[name];
|
|
62
|
+
|
|
63
|
+
if (field.fields) {
|
|
64
|
+
const children = getFields(field.fields);
|
|
65
|
+
fields.push({name, type: {type: 'struct', children}, nullable: field.optional});
|
|
66
|
+
} else {
|
|
67
|
+
const type = PARQUET_TYPE_MAPPING[field.type];
|
|
68
|
+
const metadata = getFieldMetadata(field);
|
|
69
|
+
const arrowField = {name, type, nullable: field.optional, metadata};
|
|
70
|
+
fields.push(arrowField);
|
|
71
|
+
}
|
|
72
|
+
}
|
|
73
|
+
|
|
74
|
+
return fields;
|
|
75
|
+
}
|
|
76
|
+
|
|
77
|
+
function getFieldMetadata(field: ParquetField): Record<string, string> | undefined {
|
|
78
|
+
let metadata: Record<string, string> | undefined;
|
|
79
|
+
|
|
80
|
+
for (const key in field) {
|
|
81
|
+
if (key !== 'name') {
|
|
82
|
+
let value = field[key] || '';
|
|
83
|
+
value = typeof field[key] !== 'string' ? JSON.stringify(field[key]) : field[key];
|
|
84
|
+
metadata = metadata || {};
|
|
85
|
+
metadata[key] = value;
|
|
86
|
+
}
|
|
87
|
+
}
|
|
88
|
+
|
|
89
|
+
return metadata;
|
|
90
|
+
}
|
|
91
|
+
|
|
92
|
+
function getSchemaMetadata(parquetMetadata: FileMetaData): Record<string, string> | undefined {
|
|
93
|
+
let metadata: Record<string, string> | undefined;
|
|
94
|
+
|
|
95
|
+
const keyValueList = parquetMetadata.key_value_metadata || [];
|
|
96
|
+
for (const {key, value} of keyValueList) {
|
|
97
|
+
if (typeof value === 'string') {
|
|
98
|
+
metadata = metadata || {};
|
|
99
|
+
metadata[key] = value;
|
|
100
|
+
}
|
|
101
|
+
}
|
|
102
|
+
|
|
103
|
+
return metadata;
|
|
104
|
+
}
|