@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
|
@@ -0,0 +1,610 @@
|
|
|
1
|
+
"use strict";
|
|
2
|
+
|
|
3
|
+
var _interopRequireDefault = require("@babel/runtime/helpers/interopRequireDefault");
|
|
4
|
+
Object.defineProperty(exports, "__esModule", {
|
|
5
|
+
value: true
|
|
6
|
+
});
|
|
7
|
+
exports.ParquetReader = void 0;
|
|
8
|
+
var _regenerator = _interopRequireDefault(require("@babel/runtime/regenerator"));
|
|
9
|
+
var _asyncToGenerator2 = _interopRequireDefault(require("@babel/runtime/helpers/asyncToGenerator"));
|
|
10
|
+
var _classCallCheck2 = _interopRequireDefault(require("@babel/runtime/helpers/classCallCheck"));
|
|
11
|
+
var _createClass2 = _interopRequireDefault(require("@babel/runtime/helpers/createClass"));
|
|
12
|
+
var _defineProperty2 = _interopRequireDefault(require("@babel/runtime/helpers/defineProperty"));
|
|
13
|
+
var _awaitAsyncGenerator2 = _interopRequireDefault(require("@babel/runtime/helpers/awaitAsyncGenerator"));
|
|
14
|
+
var _wrapAsyncGenerator2 = _interopRequireDefault(require("@babel/runtime/helpers/wrapAsyncGenerator"));
|
|
15
|
+
var _schema = require("../schema/schema");
|
|
16
|
+
var _decoders = require("./decoders");
|
|
17
|
+
var _shred = require("../schema/shred");
|
|
18
|
+
var _constants = require("../../constants");
|
|
19
|
+
var _parquetThrift = require("../parquet-thrift");
|
|
20
|
+
var _readUtils = require("../utils/read-utils");
|
|
21
|
+
function _createForOfIteratorHelper(o, allowArrayLike) { var it = typeof Symbol !== "undefined" && o[Symbol.iterator] || o["@@iterator"]; if (!it) { if (Array.isArray(o) || (it = _unsupportedIterableToArray(o)) || allowArrayLike && o && typeof o.length === "number") { if (it) o = it; var i = 0; var F = function F() {}; return { s: F, n: function n() { if (i >= o.length) return { done: true }; return { done: false, value: o[i++] }; }, e: function e(_e) { throw _e; }, f: F }; } throw new TypeError("Invalid attempt to iterate non-iterable instance.\nIn order to be iterable, non-array objects must have a [Symbol.iterator]() method."); } var normalCompletion = true, didErr = false, err; return { s: function s() { it = it.call(o); }, n: function n() { var step = it.next(); normalCompletion = step.done; return step; }, e: function e(_e2) { didErr = true; err = _e2; }, f: function f() { try { if (!normalCompletion && it.return != null) it.return(); } finally { if (didErr) throw err; } } }; }
|
|
22
|
+
function _unsupportedIterableToArray(o, minLen) { if (!o) return; if (typeof o === "string") return _arrayLikeToArray(o, minLen); var n = Object.prototype.toString.call(o).slice(8, -1); if (n === "Object" && o.constructor) n = o.constructor.name; if (n === "Map" || n === "Set") return Array.from(o); if (n === "Arguments" || /^(?:Ui|I)nt(?:8|16|32)(?:Clamped)?Array$/.test(n)) return _arrayLikeToArray(o, minLen); }
|
|
23
|
+
function _arrayLikeToArray(arr, len) { if (len == null || len > arr.length) len = arr.length; for (var i = 0, arr2 = new Array(len); i < len; i++) arr2[i] = arr[i]; return arr2; }
|
|
24
|
+
function ownKeys(object, enumerableOnly) { var keys = Object.keys(object); if (Object.getOwnPropertySymbols) { var symbols = Object.getOwnPropertySymbols(object); enumerableOnly && (symbols = symbols.filter(function (sym) { return Object.getOwnPropertyDescriptor(object, sym).enumerable; })), keys.push.apply(keys, symbols); } return keys; }
|
|
25
|
+
function _objectSpread(target) { for (var i = 1; i < arguments.length; i++) { var source = null != arguments[i] ? arguments[i] : {}; i % 2 ? ownKeys(Object(source), !0).forEach(function (key) { (0, _defineProperty2.default)(target, key, source[key]); }) : Object.getOwnPropertyDescriptors ? Object.defineProperties(target, Object.getOwnPropertyDescriptors(source)) : ownKeys(Object(source)).forEach(function (key) { Object.defineProperty(target, key, Object.getOwnPropertyDescriptor(source, key)); }); } return target; }
|
|
26
|
+
function _asyncIterator(iterable) { var method, async, sync, retry = 2; for ("undefined" != typeof Symbol && (async = Symbol.asyncIterator, sync = Symbol.iterator); retry--;) { if (async && null != (method = iterable[async])) return method.call(iterable); if (sync && null != (method = iterable[sync])) return new AsyncFromSyncIterator(method.call(iterable)); async = "@@asyncIterator", sync = "@@iterator"; } throw new TypeError("Object is not async iterable"); }
|
|
27
|
+
function AsyncFromSyncIterator(s) { function AsyncFromSyncIteratorContinuation(r) { if (Object(r) !== r) return Promise.reject(new TypeError(r + " is not an object.")); var done = r.done; return Promise.resolve(r.value).then(function (value) { return { value: value, done: done }; }); } return AsyncFromSyncIterator = function AsyncFromSyncIterator(s) { this.s = s, this.n = s.next; }, AsyncFromSyncIterator.prototype = { s: null, n: null, next: function next() { return AsyncFromSyncIteratorContinuation(this.n.apply(this.s, arguments)); }, return: function _return(value) { var ret = this.s.return; return void 0 === ret ? Promise.resolve({ value: value, done: !0 }) : AsyncFromSyncIteratorContinuation(ret.apply(this.s, arguments)); }, throw: function _throw(value) { var thr = this.s.return; return void 0 === thr ? Promise.reject(value) : AsyncFromSyncIteratorContinuation(thr.apply(this.s, arguments)); } }, new AsyncFromSyncIterator(s); }
|
|
28
|
+
var DEFAULT_PROPS = {
|
|
29
|
+
defaultDictionarySize: 1e6
|
|
30
|
+
};
|
|
31
|
+
var ParquetReader = function () {
|
|
32
|
+
function ParquetReader(file, props) {
|
|
33
|
+
(0, _classCallCheck2.default)(this, ParquetReader);
|
|
34
|
+
(0, _defineProperty2.default)(this, "props", void 0);
|
|
35
|
+
(0, _defineProperty2.default)(this, "file", void 0);
|
|
36
|
+
(0, _defineProperty2.default)(this, "metadata", null);
|
|
37
|
+
this.file = file;
|
|
38
|
+
this.props = _objectSpread(_objectSpread({}, DEFAULT_PROPS), props);
|
|
39
|
+
}
|
|
40
|
+
(0, _createClass2.default)(ParquetReader, [{
|
|
41
|
+
key: "close",
|
|
42
|
+
value: function close() {
|
|
43
|
+
this.file.close();
|
|
44
|
+
}
|
|
45
|
+
}, {
|
|
46
|
+
key: "rowIterator",
|
|
47
|
+
value: function rowIterator(props) {
|
|
48
|
+
var _this = this;
|
|
49
|
+
return (0, _wrapAsyncGenerator2.default)(_regenerator.default.mark(function _callee() {
|
|
50
|
+
var _iteratorAbruptCompletion, _didIteratorError, _iteratorError, _iterator, _step, rows, _iterator3, _step3, row;
|
|
51
|
+
return _regenerator.default.wrap(function _callee$(_context) {
|
|
52
|
+
while (1) switch (_context.prev = _context.next) {
|
|
53
|
+
case 0:
|
|
54
|
+
_iteratorAbruptCompletion = false;
|
|
55
|
+
_didIteratorError = false;
|
|
56
|
+
_context.prev = 2;
|
|
57
|
+
_iterator = _asyncIterator(_this.rowBatchIterator(props));
|
|
58
|
+
case 4:
|
|
59
|
+
_context.next = 6;
|
|
60
|
+
return (0, _awaitAsyncGenerator2.default)(_iterator.next());
|
|
61
|
+
case 6:
|
|
62
|
+
if (!(_iteratorAbruptCompletion = !(_step = _context.sent).done)) {
|
|
63
|
+
_context.next = 28;
|
|
64
|
+
break;
|
|
65
|
+
}
|
|
66
|
+
rows = _step.value;
|
|
67
|
+
_iterator3 = _createForOfIteratorHelper(rows);
|
|
68
|
+
_context.prev = 9;
|
|
69
|
+
_iterator3.s();
|
|
70
|
+
case 11:
|
|
71
|
+
if ((_step3 = _iterator3.n()).done) {
|
|
72
|
+
_context.next = 17;
|
|
73
|
+
break;
|
|
74
|
+
}
|
|
75
|
+
row = _step3.value;
|
|
76
|
+
_context.next = 15;
|
|
77
|
+
return row;
|
|
78
|
+
case 15:
|
|
79
|
+
_context.next = 11;
|
|
80
|
+
break;
|
|
81
|
+
case 17:
|
|
82
|
+
_context.next = 22;
|
|
83
|
+
break;
|
|
84
|
+
case 19:
|
|
85
|
+
_context.prev = 19;
|
|
86
|
+
_context.t0 = _context["catch"](9);
|
|
87
|
+
_iterator3.e(_context.t0);
|
|
88
|
+
case 22:
|
|
89
|
+
_context.prev = 22;
|
|
90
|
+
_iterator3.f();
|
|
91
|
+
return _context.finish(22);
|
|
92
|
+
case 25:
|
|
93
|
+
_iteratorAbruptCompletion = false;
|
|
94
|
+
_context.next = 4;
|
|
95
|
+
break;
|
|
96
|
+
case 28:
|
|
97
|
+
_context.next = 34;
|
|
98
|
+
break;
|
|
99
|
+
case 30:
|
|
100
|
+
_context.prev = 30;
|
|
101
|
+
_context.t1 = _context["catch"](2);
|
|
102
|
+
_didIteratorError = true;
|
|
103
|
+
_iteratorError = _context.t1;
|
|
104
|
+
case 34:
|
|
105
|
+
_context.prev = 34;
|
|
106
|
+
_context.prev = 35;
|
|
107
|
+
if (!(_iteratorAbruptCompletion && _iterator.return != null)) {
|
|
108
|
+
_context.next = 39;
|
|
109
|
+
break;
|
|
110
|
+
}
|
|
111
|
+
_context.next = 39;
|
|
112
|
+
return (0, _awaitAsyncGenerator2.default)(_iterator.return());
|
|
113
|
+
case 39:
|
|
114
|
+
_context.prev = 39;
|
|
115
|
+
if (!_didIteratorError) {
|
|
116
|
+
_context.next = 42;
|
|
117
|
+
break;
|
|
118
|
+
}
|
|
119
|
+
throw _iteratorError;
|
|
120
|
+
case 42:
|
|
121
|
+
return _context.finish(39);
|
|
122
|
+
case 43:
|
|
123
|
+
return _context.finish(34);
|
|
124
|
+
case 44:
|
|
125
|
+
case "end":
|
|
126
|
+
return _context.stop();
|
|
127
|
+
}
|
|
128
|
+
}, _callee, null, [[2, 30, 34, 44], [9, 19, 22, 25], [35,, 39, 43]]);
|
|
129
|
+
}))();
|
|
130
|
+
}
|
|
131
|
+
}, {
|
|
132
|
+
key: "rowBatchIterator",
|
|
133
|
+
value: function rowBatchIterator(props) {
|
|
134
|
+
var _this2 = this;
|
|
135
|
+
return (0, _wrapAsyncGenerator2.default)(_regenerator.default.mark(function _callee2() {
|
|
136
|
+
var schema, _iteratorAbruptCompletion2, _didIteratorError2, _iteratorError2, _iterator2, _step2, rowGroup;
|
|
137
|
+
return _regenerator.default.wrap(function _callee2$(_context2) {
|
|
138
|
+
while (1) switch (_context2.prev = _context2.next) {
|
|
139
|
+
case 0:
|
|
140
|
+
_context2.next = 2;
|
|
141
|
+
return (0, _awaitAsyncGenerator2.default)(_this2.getSchema());
|
|
142
|
+
case 2:
|
|
143
|
+
schema = _context2.sent;
|
|
144
|
+
_iteratorAbruptCompletion2 = false;
|
|
145
|
+
_didIteratorError2 = false;
|
|
146
|
+
_context2.prev = 5;
|
|
147
|
+
_iterator2 = _asyncIterator(_this2.rowGroupIterator(props));
|
|
148
|
+
case 7:
|
|
149
|
+
_context2.next = 9;
|
|
150
|
+
return (0, _awaitAsyncGenerator2.default)(_iterator2.next());
|
|
151
|
+
case 9:
|
|
152
|
+
if (!(_iteratorAbruptCompletion2 = !(_step2 = _context2.sent).done)) {
|
|
153
|
+
_context2.next = 16;
|
|
154
|
+
break;
|
|
155
|
+
}
|
|
156
|
+
rowGroup = _step2.value;
|
|
157
|
+
_context2.next = 13;
|
|
158
|
+
return (0, _shred.materializeRows)(schema, rowGroup);
|
|
159
|
+
case 13:
|
|
160
|
+
_iteratorAbruptCompletion2 = false;
|
|
161
|
+
_context2.next = 7;
|
|
162
|
+
break;
|
|
163
|
+
case 16:
|
|
164
|
+
_context2.next = 22;
|
|
165
|
+
break;
|
|
166
|
+
case 18:
|
|
167
|
+
_context2.prev = 18;
|
|
168
|
+
_context2.t0 = _context2["catch"](5);
|
|
169
|
+
_didIteratorError2 = true;
|
|
170
|
+
_iteratorError2 = _context2.t0;
|
|
171
|
+
case 22:
|
|
172
|
+
_context2.prev = 22;
|
|
173
|
+
_context2.prev = 23;
|
|
174
|
+
if (!(_iteratorAbruptCompletion2 && _iterator2.return != null)) {
|
|
175
|
+
_context2.next = 27;
|
|
176
|
+
break;
|
|
177
|
+
}
|
|
178
|
+
_context2.next = 27;
|
|
179
|
+
return (0, _awaitAsyncGenerator2.default)(_iterator2.return());
|
|
180
|
+
case 27:
|
|
181
|
+
_context2.prev = 27;
|
|
182
|
+
if (!_didIteratorError2) {
|
|
183
|
+
_context2.next = 30;
|
|
184
|
+
break;
|
|
185
|
+
}
|
|
186
|
+
throw _iteratorError2;
|
|
187
|
+
case 30:
|
|
188
|
+
return _context2.finish(27);
|
|
189
|
+
case 31:
|
|
190
|
+
return _context2.finish(22);
|
|
191
|
+
case 32:
|
|
192
|
+
case "end":
|
|
193
|
+
return _context2.stop();
|
|
194
|
+
}
|
|
195
|
+
}, _callee2, null, [[5, 18, 22, 32], [23,, 27, 31]]);
|
|
196
|
+
}))();
|
|
197
|
+
}
|
|
198
|
+
}, {
|
|
199
|
+
key: "rowGroupIterator",
|
|
200
|
+
value: function rowGroupIterator(props) {
|
|
201
|
+
var _this3 = this;
|
|
202
|
+
return (0, _wrapAsyncGenerator2.default)(_regenerator.default.mark(function _callee3() {
|
|
203
|
+
var columnList, metadata, schema, rowGroupCount, rowGroupIndex, rowGroup;
|
|
204
|
+
return _regenerator.default.wrap(function _callee3$(_context3) {
|
|
205
|
+
while (1) switch (_context3.prev = _context3.next) {
|
|
206
|
+
case 0:
|
|
207
|
+
columnList = ((props === null || props === void 0 ? void 0 : props.columnList) || []).map(function (x) {
|
|
208
|
+
return Array.isArray(x) ? x : [x];
|
|
209
|
+
});
|
|
210
|
+
_context3.next = 3;
|
|
211
|
+
return (0, _awaitAsyncGenerator2.default)(_this3.getFileMetadata());
|
|
212
|
+
case 3:
|
|
213
|
+
metadata = _context3.sent;
|
|
214
|
+
_context3.next = 6;
|
|
215
|
+
return (0, _awaitAsyncGenerator2.default)(_this3.getSchema());
|
|
216
|
+
case 6:
|
|
217
|
+
schema = _context3.sent;
|
|
218
|
+
rowGroupCount = (metadata === null || metadata === void 0 ? void 0 : metadata.row_groups.length) || 0;
|
|
219
|
+
rowGroupIndex = 0;
|
|
220
|
+
case 9:
|
|
221
|
+
if (!(rowGroupIndex < rowGroupCount)) {
|
|
222
|
+
_context3.next = 18;
|
|
223
|
+
break;
|
|
224
|
+
}
|
|
225
|
+
_context3.next = 12;
|
|
226
|
+
return (0, _awaitAsyncGenerator2.default)(_this3.readRowGroup(schema, metadata.row_groups[rowGroupIndex], columnList));
|
|
227
|
+
case 12:
|
|
228
|
+
rowGroup = _context3.sent;
|
|
229
|
+
_context3.next = 15;
|
|
230
|
+
return rowGroup;
|
|
231
|
+
case 15:
|
|
232
|
+
rowGroupIndex++;
|
|
233
|
+
_context3.next = 9;
|
|
234
|
+
break;
|
|
235
|
+
case 18:
|
|
236
|
+
case "end":
|
|
237
|
+
return _context3.stop();
|
|
238
|
+
}
|
|
239
|
+
}, _callee3);
|
|
240
|
+
}))();
|
|
241
|
+
}
|
|
242
|
+
}, {
|
|
243
|
+
key: "getRowCount",
|
|
244
|
+
value: function () {
|
|
245
|
+
var _getRowCount = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee4() {
|
|
246
|
+
var metadata;
|
|
247
|
+
return _regenerator.default.wrap(function _callee4$(_context4) {
|
|
248
|
+
while (1) switch (_context4.prev = _context4.next) {
|
|
249
|
+
case 0:
|
|
250
|
+
_context4.next = 2;
|
|
251
|
+
return this.getFileMetadata();
|
|
252
|
+
case 2:
|
|
253
|
+
metadata = _context4.sent;
|
|
254
|
+
return _context4.abrupt("return", Number(metadata.num_rows));
|
|
255
|
+
case 4:
|
|
256
|
+
case "end":
|
|
257
|
+
return _context4.stop();
|
|
258
|
+
}
|
|
259
|
+
}, _callee4, this);
|
|
260
|
+
}));
|
|
261
|
+
function getRowCount() {
|
|
262
|
+
return _getRowCount.apply(this, arguments);
|
|
263
|
+
}
|
|
264
|
+
return getRowCount;
|
|
265
|
+
}()
|
|
266
|
+
}, {
|
|
267
|
+
key: "getSchema",
|
|
268
|
+
value: function () {
|
|
269
|
+
var _getSchema = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee5() {
|
|
270
|
+
var metadata, root, _decodeSchema, schemaDefinition, schema;
|
|
271
|
+
return _regenerator.default.wrap(function _callee5$(_context5) {
|
|
272
|
+
while (1) switch (_context5.prev = _context5.next) {
|
|
273
|
+
case 0:
|
|
274
|
+
_context5.next = 2;
|
|
275
|
+
return this.getFileMetadata();
|
|
276
|
+
case 2:
|
|
277
|
+
metadata = _context5.sent;
|
|
278
|
+
root = metadata.schema[0];
|
|
279
|
+
_decodeSchema = (0, _decoders.decodeSchema)(metadata.schema, 1, root.num_children), schemaDefinition = _decodeSchema.schema;
|
|
280
|
+
schema = new _schema.ParquetSchema(schemaDefinition);
|
|
281
|
+
return _context5.abrupt("return", schema);
|
|
282
|
+
case 7:
|
|
283
|
+
case "end":
|
|
284
|
+
return _context5.stop();
|
|
285
|
+
}
|
|
286
|
+
}, _callee5, this);
|
|
287
|
+
}));
|
|
288
|
+
function getSchema() {
|
|
289
|
+
return _getSchema.apply(this, arguments);
|
|
290
|
+
}
|
|
291
|
+
return getSchema;
|
|
292
|
+
}()
|
|
293
|
+
}, {
|
|
294
|
+
key: "getSchemaMetadata",
|
|
295
|
+
value: function () {
|
|
296
|
+
var _getSchemaMetadata = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee6() {
|
|
297
|
+
var metadata, md, _iterator4, _step4, kv;
|
|
298
|
+
return _regenerator.default.wrap(function _callee6$(_context6) {
|
|
299
|
+
while (1) switch (_context6.prev = _context6.next) {
|
|
300
|
+
case 0:
|
|
301
|
+
_context6.next = 2;
|
|
302
|
+
return this.getFileMetadata();
|
|
303
|
+
case 2:
|
|
304
|
+
metadata = _context6.sent;
|
|
305
|
+
md = {};
|
|
306
|
+
_iterator4 = _createForOfIteratorHelper(metadata.key_value_metadata);
|
|
307
|
+
try {
|
|
308
|
+
for (_iterator4.s(); !(_step4 = _iterator4.n()).done;) {
|
|
309
|
+
kv = _step4.value;
|
|
310
|
+
md[kv.key] = kv.value;
|
|
311
|
+
}
|
|
312
|
+
} catch (err) {
|
|
313
|
+
_iterator4.e(err);
|
|
314
|
+
} finally {
|
|
315
|
+
_iterator4.f();
|
|
316
|
+
}
|
|
317
|
+
return _context6.abrupt("return", md);
|
|
318
|
+
case 7:
|
|
319
|
+
case "end":
|
|
320
|
+
return _context6.stop();
|
|
321
|
+
}
|
|
322
|
+
}, _callee6, this);
|
|
323
|
+
}));
|
|
324
|
+
function getSchemaMetadata() {
|
|
325
|
+
return _getSchemaMetadata.apply(this, arguments);
|
|
326
|
+
}
|
|
327
|
+
return getSchemaMetadata;
|
|
328
|
+
}()
|
|
329
|
+
}, {
|
|
330
|
+
key: "getFileMetadata",
|
|
331
|
+
value: function () {
|
|
332
|
+
var _getFileMetadata = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee7() {
|
|
333
|
+
return _regenerator.default.wrap(function _callee7$(_context7) {
|
|
334
|
+
while (1) switch (_context7.prev = _context7.next) {
|
|
335
|
+
case 0:
|
|
336
|
+
if (this.metadata) {
|
|
337
|
+
_context7.next = 4;
|
|
338
|
+
break;
|
|
339
|
+
}
|
|
340
|
+
_context7.next = 3;
|
|
341
|
+
return this.readHeader();
|
|
342
|
+
case 3:
|
|
343
|
+
this.metadata = this.readFooter();
|
|
344
|
+
case 4:
|
|
345
|
+
return _context7.abrupt("return", this.metadata);
|
|
346
|
+
case 5:
|
|
347
|
+
case "end":
|
|
348
|
+
return _context7.stop();
|
|
349
|
+
}
|
|
350
|
+
}, _callee7, this);
|
|
351
|
+
}));
|
|
352
|
+
function getFileMetadata() {
|
|
353
|
+
return _getFileMetadata.apply(this, arguments);
|
|
354
|
+
}
|
|
355
|
+
return getFileMetadata;
|
|
356
|
+
}()
|
|
357
|
+
}, {
|
|
358
|
+
key: "readHeader",
|
|
359
|
+
value: function () {
|
|
360
|
+
var _readHeader = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee8() {
|
|
361
|
+
var buffer, magic;
|
|
362
|
+
return _regenerator.default.wrap(function _callee8$(_context8) {
|
|
363
|
+
while (1) switch (_context8.prev = _context8.next) {
|
|
364
|
+
case 0:
|
|
365
|
+
_context8.next = 2;
|
|
366
|
+
return this.file.read(0, _constants.PARQUET_MAGIC.length);
|
|
367
|
+
case 2:
|
|
368
|
+
buffer = _context8.sent;
|
|
369
|
+
magic = buffer.toString();
|
|
370
|
+
_context8.t0 = magic;
|
|
371
|
+
_context8.next = _context8.t0 === _constants.PARQUET_MAGIC ? 7 : _context8.t0 === _constants.PARQUET_MAGIC_ENCRYPTED ? 8 : 9;
|
|
372
|
+
break;
|
|
373
|
+
case 7:
|
|
374
|
+
return _context8.abrupt("break", 10);
|
|
375
|
+
case 8:
|
|
376
|
+
throw new Error('Encrypted parquet file not supported');
|
|
377
|
+
case 9:
|
|
378
|
+
throw new Error("Invalid parquet file (magic=".concat(magic, ")"));
|
|
379
|
+
case 10:
|
|
380
|
+
case "end":
|
|
381
|
+
return _context8.stop();
|
|
382
|
+
}
|
|
383
|
+
}, _callee8, this);
|
|
384
|
+
}));
|
|
385
|
+
function readHeader() {
|
|
386
|
+
return _readHeader.apply(this, arguments);
|
|
387
|
+
}
|
|
388
|
+
return readHeader;
|
|
389
|
+
}()
|
|
390
|
+
}, {
|
|
391
|
+
key: "readFooter",
|
|
392
|
+
value: function () {
|
|
393
|
+
var _readFooter = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee9() {
|
|
394
|
+
var trailerLen, trailerBuf, magic, metadataSize, metadataOffset, metadataBuf, _decodeFileMetadata, metadata;
|
|
395
|
+
return _regenerator.default.wrap(function _callee9$(_context9) {
|
|
396
|
+
while (1) switch (_context9.prev = _context9.next) {
|
|
397
|
+
case 0:
|
|
398
|
+
trailerLen = _constants.PARQUET_MAGIC.length + 4;
|
|
399
|
+
_context9.next = 3;
|
|
400
|
+
return this.file.read(this.file.size - trailerLen, trailerLen);
|
|
401
|
+
case 3:
|
|
402
|
+
trailerBuf = _context9.sent;
|
|
403
|
+
magic = trailerBuf.slice(4).toString();
|
|
404
|
+
if (!(magic !== _constants.PARQUET_MAGIC)) {
|
|
405
|
+
_context9.next = 7;
|
|
406
|
+
break;
|
|
407
|
+
}
|
|
408
|
+
throw new Error("Not a valid parquet file (magic=\"".concat(magic, ")"));
|
|
409
|
+
case 7:
|
|
410
|
+
metadataSize = trailerBuf.readUInt32LE(0);
|
|
411
|
+
metadataOffset = this.file.size - metadataSize - trailerLen;
|
|
412
|
+
if (!(metadataOffset < _constants.PARQUET_MAGIC.length)) {
|
|
413
|
+
_context9.next = 11;
|
|
414
|
+
break;
|
|
415
|
+
}
|
|
416
|
+
throw new Error("Invalid metadata size ".concat(metadataOffset));
|
|
417
|
+
case 11:
|
|
418
|
+
_context9.next = 13;
|
|
419
|
+
return this.file.read(metadataOffset, metadataSize);
|
|
420
|
+
case 13:
|
|
421
|
+
metadataBuf = _context9.sent;
|
|
422
|
+
_decodeFileMetadata = (0, _readUtils.decodeFileMetadata)(metadataBuf), metadata = _decodeFileMetadata.metadata;
|
|
423
|
+
return _context9.abrupt("return", metadata);
|
|
424
|
+
case 16:
|
|
425
|
+
case "end":
|
|
426
|
+
return _context9.stop();
|
|
427
|
+
}
|
|
428
|
+
}, _callee9, this);
|
|
429
|
+
}));
|
|
430
|
+
function readFooter() {
|
|
431
|
+
return _readFooter.apply(this, arguments);
|
|
432
|
+
}
|
|
433
|
+
return readFooter;
|
|
434
|
+
}()
|
|
435
|
+
}, {
|
|
436
|
+
key: "readRowGroup",
|
|
437
|
+
value: function () {
|
|
438
|
+
var _readRowGroup = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee10(schema, rowGroup, columnList) {
|
|
439
|
+
var buffer, _iterator5, _step5, colChunk, colMetadata, colKey;
|
|
440
|
+
return _regenerator.default.wrap(function _callee10$(_context10) {
|
|
441
|
+
while (1) switch (_context10.prev = _context10.next) {
|
|
442
|
+
case 0:
|
|
443
|
+
buffer = {
|
|
444
|
+
rowCount: Number(rowGroup.num_rows),
|
|
445
|
+
columnData: {}
|
|
446
|
+
};
|
|
447
|
+
_iterator5 = _createForOfIteratorHelper(rowGroup.columns);
|
|
448
|
+
_context10.prev = 2;
|
|
449
|
+
_iterator5.s();
|
|
450
|
+
case 4:
|
|
451
|
+
if ((_step5 = _iterator5.n()).done) {
|
|
452
|
+
_context10.next = 15;
|
|
453
|
+
break;
|
|
454
|
+
}
|
|
455
|
+
colChunk = _step5.value;
|
|
456
|
+
colMetadata = colChunk.meta_data;
|
|
457
|
+
colKey = colMetadata === null || colMetadata === void 0 ? void 0 : colMetadata.path_in_schema;
|
|
458
|
+
if (!(columnList.length > 0 && (0, _readUtils.fieldIndexOf)(columnList, colKey) < 0)) {
|
|
459
|
+
_context10.next = 10;
|
|
460
|
+
break;
|
|
461
|
+
}
|
|
462
|
+
return _context10.abrupt("continue", 13);
|
|
463
|
+
case 10:
|
|
464
|
+
_context10.next = 12;
|
|
465
|
+
return this.readColumnChunk(schema, colChunk);
|
|
466
|
+
case 12:
|
|
467
|
+
buffer.columnData[colKey.join()] = _context10.sent;
|
|
468
|
+
case 13:
|
|
469
|
+
_context10.next = 4;
|
|
470
|
+
break;
|
|
471
|
+
case 15:
|
|
472
|
+
_context10.next = 20;
|
|
473
|
+
break;
|
|
474
|
+
case 17:
|
|
475
|
+
_context10.prev = 17;
|
|
476
|
+
_context10.t0 = _context10["catch"](2);
|
|
477
|
+
_iterator5.e(_context10.t0);
|
|
478
|
+
case 20:
|
|
479
|
+
_context10.prev = 20;
|
|
480
|
+
_iterator5.f();
|
|
481
|
+
return _context10.finish(20);
|
|
482
|
+
case 23:
|
|
483
|
+
return _context10.abrupt("return", buffer);
|
|
484
|
+
case 24:
|
|
485
|
+
case "end":
|
|
486
|
+
return _context10.stop();
|
|
487
|
+
}
|
|
488
|
+
}, _callee10, this, [[2, 17, 20, 23]]);
|
|
489
|
+
}));
|
|
490
|
+
function readRowGroup(_x, _x2, _x3) {
|
|
491
|
+
return _readRowGroup.apply(this, arguments);
|
|
492
|
+
}
|
|
493
|
+
return readRowGroup;
|
|
494
|
+
}()
|
|
495
|
+
}, {
|
|
496
|
+
key: "readColumnChunk",
|
|
497
|
+
value: function () {
|
|
498
|
+
var _readColumnChunk = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee11(schema, colChunk) {
|
|
499
|
+
var _colChunk$meta_data, _colChunk$meta_data2, _colChunk$meta_data3, _colChunk$meta_data4, _colChunk$meta_data5, _colChunk$meta_data7, _colChunk$meta_data8, _options$dictionary;
|
|
500
|
+
var field, type, compression, pagesOffset, pagesSize, _colChunk$meta_data6, options, dictionary, dictionaryPageOffset, dictionaryOffset, pagesBuf;
|
|
501
|
+
return _regenerator.default.wrap(function _callee11$(_context11) {
|
|
502
|
+
while (1) switch (_context11.prev = _context11.next) {
|
|
503
|
+
case 0:
|
|
504
|
+
if (!(colChunk.file_path !== undefined && colChunk.file_path !== null)) {
|
|
505
|
+
_context11.next = 2;
|
|
506
|
+
break;
|
|
507
|
+
}
|
|
508
|
+
throw new Error('external references are not supported');
|
|
509
|
+
case 2:
|
|
510
|
+
field = schema.findField((_colChunk$meta_data = colChunk.meta_data) === null || _colChunk$meta_data === void 0 ? void 0 : _colChunk$meta_data.path_in_schema);
|
|
511
|
+
type = (0, _readUtils.getThriftEnum)(_parquetThrift.Type, (_colChunk$meta_data2 = colChunk.meta_data) === null || _colChunk$meta_data2 === void 0 ? void 0 : _colChunk$meta_data2.type);
|
|
512
|
+
if (!(type !== field.primitiveType)) {
|
|
513
|
+
_context11.next = 6;
|
|
514
|
+
break;
|
|
515
|
+
}
|
|
516
|
+
throw new Error("chunk type not matching schema: ".concat(type));
|
|
517
|
+
case 6:
|
|
518
|
+
compression = (0, _readUtils.getThriftEnum)(_parquetThrift.CompressionCodec, (_colChunk$meta_data3 = colChunk.meta_data) === null || _colChunk$meta_data3 === void 0 ? void 0 : _colChunk$meta_data3.codec);
|
|
519
|
+
pagesOffset = Number((_colChunk$meta_data4 = colChunk.meta_data) === null || _colChunk$meta_data4 === void 0 ? void 0 : _colChunk$meta_data4.data_page_offset);
|
|
520
|
+
pagesSize = Number((_colChunk$meta_data5 = colChunk.meta_data) === null || _colChunk$meta_data5 === void 0 ? void 0 : _colChunk$meta_data5.total_compressed_size);
|
|
521
|
+
if (!colChunk.file_path) {
|
|
522
|
+
pagesSize = Math.min(this.file.size - pagesOffset, Number((_colChunk$meta_data6 = colChunk.meta_data) === null || _colChunk$meta_data6 === void 0 ? void 0 : _colChunk$meta_data6.total_compressed_size));
|
|
523
|
+
}
|
|
524
|
+
options = {
|
|
525
|
+
type: type,
|
|
526
|
+
rLevelMax: field.rLevelMax,
|
|
527
|
+
dLevelMax: field.dLevelMax,
|
|
528
|
+
compression: compression,
|
|
529
|
+
column: field,
|
|
530
|
+
numValues: (_colChunk$meta_data7 = colChunk.meta_data) === null || _colChunk$meta_data7 === void 0 ? void 0 : _colChunk$meta_data7.num_values,
|
|
531
|
+
dictionary: []
|
|
532
|
+
};
|
|
533
|
+
dictionaryPageOffset = colChunk === null || colChunk === void 0 ? void 0 : (_colChunk$meta_data8 = colChunk.meta_data) === null || _colChunk$meta_data8 === void 0 ? void 0 : _colChunk$meta_data8.dictionary_page_offset;
|
|
534
|
+
if (!dictionaryPageOffset) {
|
|
535
|
+
_context11.next = 17;
|
|
536
|
+
break;
|
|
537
|
+
}
|
|
538
|
+
dictionaryOffset = Number(dictionaryPageOffset);
|
|
539
|
+
_context11.next = 16;
|
|
540
|
+
return this.getDictionary(dictionaryOffset, options, pagesOffset);
|
|
541
|
+
case 16:
|
|
542
|
+
dictionary = _context11.sent;
|
|
543
|
+
case 17:
|
|
544
|
+
dictionary = (_options$dictionary = options.dictionary) !== null && _options$dictionary !== void 0 && _options$dictionary.length ? options.dictionary : dictionary;
|
|
545
|
+
_context11.next = 20;
|
|
546
|
+
return this.file.read(pagesOffset, pagesSize);
|
|
547
|
+
case 20:
|
|
548
|
+
pagesBuf = _context11.sent;
|
|
549
|
+
_context11.next = 23;
|
|
550
|
+
return (0, _decoders.decodeDataPages)(pagesBuf, _objectSpread(_objectSpread({}, options), {}, {
|
|
551
|
+
dictionary: dictionary
|
|
552
|
+
}));
|
|
553
|
+
case 23:
|
|
554
|
+
return _context11.abrupt("return", _context11.sent);
|
|
555
|
+
case 24:
|
|
556
|
+
case "end":
|
|
557
|
+
return _context11.stop();
|
|
558
|
+
}
|
|
559
|
+
}, _callee11, this);
|
|
560
|
+
}));
|
|
561
|
+
function readColumnChunk(_x4, _x5) {
|
|
562
|
+
return _readColumnChunk.apply(this, arguments);
|
|
563
|
+
}
|
|
564
|
+
return readColumnChunk;
|
|
565
|
+
}()
|
|
566
|
+
}, {
|
|
567
|
+
key: "getDictionary",
|
|
568
|
+
value: function () {
|
|
569
|
+
var _getDictionary = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee12(dictionaryPageOffset, options, pagesOffset) {
|
|
570
|
+
var dictionarySize, pagesBuf, cursor, decodedPage;
|
|
571
|
+
return _regenerator.default.wrap(function _callee12$(_context12) {
|
|
572
|
+
while (1) switch (_context12.prev = _context12.next) {
|
|
573
|
+
case 0:
|
|
574
|
+
if (!(dictionaryPageOffset === 0)) {
|
|
575
|
+
_context12.next = 2;
|
|
576
|
+
break;
|
|
577
|
+
}
|
|
578
|
+
return _context12.abrupt("return", []);
|
|
579
|
+
case 2:
|
|
580
|
+
dictionarySize = Math.min(this.file.size - dictionaryPageOffset, this.props.defaultDictionarySize);
|
|
581
|
+
_context12.next = 5;
|
|
582
|
+
return this.file.read(dictionaryPageOffset, dictionarySize);
|
|
583
|
+
case 5:
|
|
584
|
+
pagesBuf = _context12.sent;
|
|
585
|
+
cursor = {
|
|
586
|
+
buffer: pagesBuf,
|
|
587
|
+
offset: 0,
|
|
588
|
+
size: pagesBuf.length
|
|
589
|
+
};
|
|
590
|
+
_context12.next = 9;
|
|
591
|
+
return (0, _decoders.decodePage)(cursor, options);
|
|
592
|
+
case 9:
|
|
593
|
+
decodedPage = _context12.sent;
|
|
594
|
+
return _context12.abrupt("return", decodedPage.dictionary);
|
|
595
|
+
case 11:
|
|
596
|
+
case "end":
|
|
597
|
+
return _context12.stop();
|
|
598
|
+
}
|
|
599
|
+
}, _callee12, this);
|
|
600
|
+
}));
|
|
601
|
+
function getDictionary(_x6, _x7, _x8) {
|
|
602
|
+
return _getDictionary.apply(this, arguments);
|
|
603
|
+
}
|
|
604
|
+
return getDictionary;
|
|
605
|
+
}()
|
|
606
|
+
}]);
|
|
607
|
+
return ParquetReader;
|
|
608
|
+
}();
|
|
609
|
+
exports.ParquetReader = ParquetReader;
|
|
610
|
+
//# sourceMappingURL=parquet-reader.js.map
|