@loaders.gl/parquet 3.1.0-alpha.4 → 3.1.0-beta.3
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.d.ts +2 -0
- package/dist/bundle.d.ts.map +1 -0
- package/dist/bundle.js +5 -0
- package/dist/constants.d.ts +15 -0
- package/dist/constants.d.ts.map +1 -0
- package/dist/constants.js +18 -0
- package/dist/dist.min.js +27 -13
- package/dist/dist.min.js.map +7 -1
- package/dist/es5/bundle.js +1 -1
- package/dist/es5/bundle.js.map +1 -1
- package/dist/es5/constants.js +5 -5
- package/dist/es5/constants.js.map +1 -1
- package/dist/es5/index.js +16 -45
- package/dist/es5/index.js.map +1 -1
- package/dist/es5/lib/convert-schema.js +13 -13
- package/dist/es5/lib/convert-schema.js.map +1 -1
- package/dist/es5/lib/parse-parquet.js +19 -154
- package/dist/es5/lib/parse-parquet.js.map +1 -1
- package/dist/es5/lib/read-array-buffer.js +6 -43
- package/dist/es5/lib/read-array-buffer.js.map +1 -1
- package/dist/es5/parquet-loader.js +4 -4
- package/dist/es5/parquet-loader.js.map +1 -1
- package/dist/es5/parquet-writer.js +4 -4
- package/dist/es5/parquet-writer.js.map +1 -1
- package/dist/es5/parquetjs/codecs/dictionary.js +2 -10
- package/dist/es5/parquetjs/codecs/dictionary.js.map +1 -1
- package/dist/es5/parquetjs/codecs/index.js +4 -6
- package/dist/es5/parquetjs/codecs/index.js.map +1 -1
- package/dist/es5/parquetjs/codecs/plain.js +41 -43
- package/dist/es5/parquetjs/codecs/plain.js.map +1 -1
- package/dist/es5/parquetjs/codecs/rle.js +25 -35
- package/dist/es5/parquetjs/codecs/rle.js.map +1 -1
- package/dist/es5/parquetjs/compression.js +28 -122
- package/dist/es5/parquetjs/compression.js.map +1 -1
- package/dist/es5/parquetjs/encoder/writer.js +301 -737
- package/dist/es5/parquetjs/encoder/writer.js.map +1 -1
- package/dist/es5/parquetjs/file.js +15 -15
- package/dist/es5/parquetjs/file.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/BoundaryOrder.js +1 -1
- package/dist/es5/parquetjs/parquet-thrift/BsonType.js +31 -45
- package/dist/es5/parquetjs/parquet-thrift/BsonType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/ColumnChunk.js +141 -152
- package/dist/es5/parquetjs/parquet-thrift/ColumnChunk.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/ColumnIndex.js +147 -160
- package/dist/es5/parquetjs/parquet-thrift/ColumnIndex.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/ColumnMetaData.js +248 -259
- package/dist/es5/parquetjs/parquet-thrift/ColumnMetaData.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/ColumnOrder.js +67 -79
- package/dist/es5/parquetjs/parquet-thrift/ColumnOrder.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/CompressionCodec.js +1 -1
- package/dist/es5/parquetjs/parquet-thrift/ConvertedType.js +1 -1
- package/dist/es5/parquetjs/parquet-thrift/DataPageHeader.js +113 -124
- package/dist/es5/parquetjs/parquet-thrift/DataPageHeader.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/DataPageHeaderV2.js +158 -169
- package/dist/es5/parquetjs/parquet-thrift/DataPageHeaderV2.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/DateType.js +31 -45
- package/dist/es5/parquetjs/parquet-thrift/DateType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/DecimalType.js +68 -79
- package/dist/es5/parquetjs/parquet-thrift/DecimalType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/DictionaryPageHeader.js +83 -94
- package/dist/es5/parquetjs/parquet-thrift/DictionaryPageHeader.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/Encoding.js +1 -1
- package/dist/es5/parquetjs/parquet-thrift/EnumType.js +31 -45
- package/dist/es5/parquetjs/parquet-thrift/EnumType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/FieldRepetitionType.js +1 -1
- package/dist/es5/parquetjs/parquet-thrift/FileMetaData.js +170 -182
- package/dist/es5/parquetjs/parquet-thrift/FileMetaData.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/IndexPageHeader.js +31 -45
- package/dist/es5/parquetjs/parquet-thrift/IndexPageHeader.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/IntType.js +68 -79
- package/dist/es5/parquetjs/parquet-thrift/IntType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/JsonType.js +31 -45
- package/dist/es5/parquetjs/parquet-thrift/JsonType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/KeyValue.js +68 -79
- package/dist/es5/parquetjs/parquet-thrift/KeyValue.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/ListType.js +31 -45
- package/dist/es5/parquetjs/parquet-thrift/ListType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/LogicalType.js +319 -343
- package/dist/es5/parquetjs/parquet-thrift/LogicalType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/MapType.js +31 -45
- package/dist/es5/parquetjs/parquet-thrift/MapType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/MicroSeconds.js +31 -45
- package/dist/es5/parquetjs/parquet-thrift/MicroSeconds.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/MilliSeconds.js +31 -45
- package/dist/es5/parquetjs/parquet-thrift/MilliSeconds.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/NullType.js +31 -45
- package/dist/es5/parquetjs/parquet-thrift/NullType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/OffsetIndex.js +64 -75
- package/dist/es5/parquetjs/parquet-thrift/OffsetIndex.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/PageEncodingStats.js +83 -94
- package/dist/es5/parquetjs/parquet-thrift/PageEncodingStats.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/PageHeader.js +158 -169
- package/dist/es5/parquetjs/parquet-thrift/PageHeader.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/PageLocation.js +83 -94
- package/dist/es5/parquetjs/parquet-thrift/PageLocation.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/PageType.js +1 -1
- package/dist/es5/parquetjs/parquet-thrift/RowGroup.js +113 -124
- package/dist/es5/parquetjs/parquet-thrift/RowGroup.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/SchemaElement.js +188 -199
- package/dist/es5/parquetjs/parquet-thrift/SchemaElement.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/SortingColumn.js +83 -94
- package/dist/es5/parquetjs/parquet-thrift/SortingColumn.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/Statistics.js +124 -135
- package/dist/es5/parquetjs/parquet-thrift/Statistics.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/StringType.js +31 -45
- package/dist/es5/parquetjs/parquet-thrift/StringType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/TimeType.js +68 -79
- package/dist/es5/parquetjs/parquet-thrift/TimeType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/TimeUnit.js +88 -101
- package/dist/es5/parquetjs/parquet-thrift/TimeUnit.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/TimestampType.js +68 -79
- package/dist/es5/parquetjs/parquet-thrift/TimestampType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/Type.js +1 -1
- package/dist/es5/parquetjs/parquet-thrift/TypeDefinedOrder.js +31 -45
- package/dist/es5/parquetjs/parquet-thrift/TypeDefinedOrder.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/UUIDType.js +31 -45
- package/dist/es5/parquetjs/parquet-thrift/UUIDType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/index.js +43 -43
- package/dist/es5/parquetjs/parser/decoders.js +218 -397
- package/dist/es5/parquetjs/parser/decoders.js.map +1 -1
- package/dist/es5/parquetjs/parser/parquet-cursor.js +62 -180
- package/dist/es5/parquetjs/parser/parquet-cursor.js.map +1 -1
- package/dist/es5/parquetjs/parser/parquet-envelope-reader.js +124 -408
- package/dist/es5/parquetjs/parser/parquet-envelope-reader.js.map +1 -1
- package/dist/es5/parquetjs/parser/parquet-reader.js +91 -369
- package/dist/es5/parquetjs/parser/parquet-reader.js.map +1 -1
- package/dist/es5/parquetjs/schema/declare.js +9 -11
- package/dist/es5/parquetjs/schema/declare.js.map +1 -1
- package/dist/es5/parquetjs/schema/schema.js +73 -87
- package/dist/es5/parquetjs/schema/schema.js.map +1 -1
- package/dist/es5/parquetjs/schema/shred.js +56 -96
- package/dist/es5/parquetjs/schema/shred.js.map +1 -1
- package/dist/es5/parquetjs/schema/types.js +39 -40
- package/dist/es5/parquetjs/schema/types.js.map +1 -1
- package/dist/es5/parquetjs/utils/buffer-utils.js +1 -1
- package/dist/es5/parquetjs/utils/buffer-utils.js.map +1 -1
- package/dist/es5/parquetjs/utils/file-utils.js +8 -65
- package/dist/es5/parquetjs/utils/file-utils.js.map +1 -1
- package/dist/es5/parquetjs/utils/read-utils.js +22 -50
- package/dist/es5/parquetjs/utils/read-utils.js.map +1 -1
- package/dist/esm/index.js +2 -3
- package/dist/esm/index.js.map +1 -1
- package/dist/esm/parquet-loader.js +1 -1
- package/dist/esm/parquet-loader.js.map +1 -1
- package/dist/esm/parquet-writer.js +1 -1
- package/dist/esm/parquet-writer.js.map +1 -1
- package/dist/esm/parquetjs/codecs/plain.js +3 -3
- package/dist/esm/parquetjs/codecs/plain.js.map +1 -1
- package/dist/esm/parquetjs/codecs/rle.js +1 -1
- package/dist/esm/parquetjs/codecs/rle.js.map +1 -1
- package/dist/esm/parquetjs/compression.js +4 -13
- package/dist/esm/parquetjs/compression.js.map +1 -1
- package/dist/esm/parquetjs/encoder/writer.js +1 -1
- package/dist/esm/parquetjs/encoder/writer.js.map +1 -1
- package/dist/esm/parquetjs/parser/decoders.js +4 -4
- package/dist/esm/parquetjs/parser/decoders.js.map +1 -1
- package/dist/esm/parquetjs/parser/parquet-envelope-reader.js +4 -13
- package/dist/esm/parquetjs/parser/parquet-envelope-reader.js.map +1 -1
- package/dist/esm/parquetjs/parser/parquet-reader.js +0 -13
- package/dist/esm/parquetjs/parser/parquet-reader.js.map +1 -1
- package/dist/esm/parquetjs/schema/schema.js +3 -3
- package/dist/esm/parquetjs/schema/schema.js.map +1 -1
- package/dist/esm/parquetjs/schema/shred.js +2 -2
- package/dist/esm/parquetjs/schema/shred.js.map +1 -1
- package/dist/esm/parquetjs/schema/types.js +20 -20
- package/dist/esm/parquetjs/schema/types.js.map +1 -1
- package/dist/esm/parquetjs/utils/file-utils.js +0 -45
- package/dist/esm/parquetjs/utils/file-utils.js.map +1 -1
- package/dist/index.d.ts +28 -0
- package/dist/index.d.ts.map +1 -0
- package/dist/index.js +30 -0
- package/dist/lib/convert-schema.d.ts +8 -0
- package/dist/lib/convert-schema.d.ts.map +1 -0
- package/dist/lib/convert-schema.js +70 -0
- package/dist/lib/parse-parquet.d.ts +4 -0
- package/dist/lib/parse-parquet.d.ts.map +1 -0
- package/dist/lib/parse-parquet.js +28 -0
- package/dist/lib/read-array-buffer.d.ts +19 -0
- package/dist/lib/read-array-buffer.d.ts.map +1 -0
- package/dist/lib/read-array-buffer.js +29 -0
- package/dist/parquet-loader.d.ts +23 -0
- package/dist/parquet-loader.d.ts.map +1 -0
- package/dist/parquet-loader.js +27 -0
- package/dist/parquet-worker.js +27 -13
- package/dist/parquet-worker.js.map +7 -1
- package/dist/parquet-writer.d.ts +4 -0
- package/dist/parquet-writer.d.ts.map +1 -0
- package/dist/parquet-writer.js +21 -0
- package/dist/parquetjs/codecs/declare.d.ts +17 -0
- package/dist/parquetjs/codecs/declare.d.ts.map +1 -0
- package/dist/parquetjs/codecs/declare.js +2 -0
- package/dist/parquetjs/codecs/dictionary.d.ts +3 -0
- package/dist/parquetjs/codecs/dictionary.d.ts.map +1 -0
- package/dist/parquetjs/codecs/dictionary.js +14 -0
- package/dist/parquetjs/codecs/index.d.ts +5 -0
- package/dist/parquetjs/codecs/index.d.ts.map +1 -0
- package/dist/parquetjs/codecs/index.js +51 -0
- package/dist/parquetjs/codecs/plain.d.ts +6 -0
- package/dist/parquetjs/codecs/plain.d.ts.map +1 -0
- package/dist/parquetjs/codecs/plain.js +211 -0
- package/dist/parquetjs/codecs/rle.d.ts +6 -0
- package/dist/parquetjs/codecs/rle.d.ts.map +1 -0
- package/dist/parquetjs/codecs/rle.js +145 -0
- package/dist/parquetjs/compression.d.ts +23 -0
- package/dist/parquetjs/compression.d.ts.map +1 -0
- package/dist/parquetjs/compression.js +168 -0
- package/dist/parquetjs/encoder/writer.d.ts +123 -0
- package/dist/parquetjs/encoder/writer.d.ts.map +1 -0
- package/dist/parquetjs/encoder/writer.js +478 -0
- package/dist/parquetjs/file.d.ts +10 -0
- package/dist/parquetjs/file.d.ts.map +1 -0
- package/dist/parquetjs/file.js +99 -0
- package/dist/parquetjs/parquet-thrift/BoundaryOrder.d.ts +6 -0
- package/dist/parquetjs/parquet-thrift/BoundaryOrder.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/BoundaryOrder.js +15 -0
- package/dist/parquetjs/parquet-thrift/BsonType.d.ts +9 -0
- package/dist/parquetjs/parquet-thrift/BsonType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/BsonType.js +58 -0
- package/dist/parquetjs/parquet-thrift/ColumnChunk.d.ts +25 -0
- package/dist/parquetjs/parquet-thrift/ColumnChunk.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/ColumnChunk.js +207 -0
- package/dist/parquetjs/parquet-thrift/ColumnIndex.d.ts +22 -0
- package/dist/parquetjs/parquet-thrift/ColumnIndex.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/ColumnIndex.js +213 -0
- package/dist/parquetjs/parquet-thrift/ColumnMetaData.d.ts +42 -0
- package/dist/parquetjs/parquet-thrift/ColumnMetaData.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/ColumnMetaData.js +398 -0
- package/dist/parquetjs/parquet-thrift/ColumnOrder.d.ts +13 -0
- package/dist/parquetjs/parquet-thrift/ColumnOrder.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/ColumnOrder.js +104 -0
- package/dist/parquetjs/parquet-thrift/CompressionCodec.d.ts +11 -0
- package/dist/parquetjs/parquet-thrift/CompressionCodec.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/CompressionCodec.js +20 -0
- package/dist/parquetjs/parquet-thrift/ConvertedType.d.ts +25 -0
- package/dist/parquetjs/parquet-thrift/ConvertedType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/ConvertedType.js +34 -0
- package/dist/parquetjs/parquet-thrift/DataPageHeader.d.ts +21 -0
- package/dist/parquetjs/parquet-thrift/DataPageHeader.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/DataPageHeader.js +166 -0
- package/dist/parquetjs/parquet-thrift/DataPageHeaderV2.d.ts +27 -0
- package/dist/parquetjs/parquet-thrift/DataPageHeaderV2.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/DataPageHeaderV2.js +226 -0
- package/dist/parquetjs/parquet-thrift/DateType.d.ts +9 -0
- package/dist/parquetjs/parquet-thrift/DateType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/DateType.js +58 -0
- package/dist/parquetjs/parquet-thrift/DecimalType.d.ts +13 -0
- package/dist/parquetjs/parquet-thrift/DecimalType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/DecimalType.js +105 -0
- package/dist/parquetjs/parquet-thrift/DictionaryPageHeader.d.ts +16 -0
- package/dist/parquetjs/parquet-thrift/DictionaryPageHeader.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/DictionaryPageHeader.js +122 -0
- package/dist/parquetjs/parquet-thrift/Encoding.d.ts +11 -0
- package/dist/parquetjs/parquet-thrift/Encoding.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/Encoding.js +20 -0
- package/dist/parquetjs/parquet-thrift/EnumType.d.ts +9 -0
- package/dist/parquetjs/parquet-thrift/EnumType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/EnumType.js +58 -0
- package/dist/parquetjs/parquet-thrift/FieldRepetitionType.d.ts +6 -0
- package/dist/parquetjs/parquet-thrift/FieldRepetitionType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/FieldRepetitionType.js +15 -0
- package/dist/parquetjs/parquet-thrift/FileMetaData.d.ts +28 -0
- package/dist/parquetjs/parquet-thrift/FileMetaData.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/FileMetaData.js +256 -0
- package/dist/parquetjs/parquet-thrift/IndexPageHeader.d.ts +9 -0
- package/dist/parquetjs/parquet-thrift/IndexPageHeader.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/IndexPageHeader.js +58 -0
- package/dist/parquetjs/parquet-thrift/IntType.d.ts +13 -0
- package/dist/parquetjs/parquet-thrift/IntType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/IntType.js +105 -0
- package/dist/parquetjs/parquet-thrift/JsonType.d.ts +9 -0
- package/dist/parquetjs/parquet-thrift/JsonType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/JsonType.js +58 -0
- package/dist/parquetjs/parquet-thrift/KeyValue.d.ts +13 -0
- package/dist/parquetjs/parquet-thrift/KeyValue.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/KeyValue.js +102 -0
- package/dist/parquetjs/parquet-thrift/ListType.d.ts +9 -0
- package/dist/parquetjs/parquet-thrift/ListType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/ListType.js +58 -0
- package/dist/parquetjs/parquet-thrift/LogicalType.d.ts +61 -0
- package/dist/parquetjs/parquet-thrift/LogicalType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/LogicalType.js +380 -0
- package/dist/parquetjs/parquet-thrift/MapType.d.ts +9 -0
- package/dist/parquetjs/parquet-thrift/MapType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/MapType.js +58 -0
- package/dist/parquetjs/parquet-thrift/MicroSeconds.d.ts +9 -0
- package/dist/parquetjs/parquet-thrift/MicroSeconds.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/MicroSeconds.js +58 -0
- package/dist/parquetjs/parquet-thrift/MilliSeconds.d.ts +9 -0
- package/dist/parquetjs/parquet-thrift/MilliSeconds.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/MilliSeconds.js +58 -0
- package/dist/parquetjs/parquet-thrift/NullType.d.ts +9 -0
- package/dist/parquetjs/parquet-thrift/NullType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/NullType.js +58 -0
- package/dist/parquetjs/parquet-thrift/OffsetIndex.d.ts +12 -0
- package/dist/parquetjs/parquet-thrift/OffsetIndex.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/OffsetIndex.js +97 -0
- package/dist/parquetjs/parquet-thrift/PageEncodingStats.d.ts +17 -0
- package/dist/parquetjs/parquet-thrift/PageEncodingStats.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/PageEncodingStats.js +127 -0
- package/dist/parquetjs/parquet-thrift/PageHeader.d.ts +30 -0
- package/dist/parquetjs/parquet-thrift/PageHeader.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/PageHeader.js +216 -0
- package/dist/parquetjs/parquet-thrift/PageLocation.d.ts +16 -0
- package/dist/parquetjs/parquet-thrift/PageLocation.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/PageLocation.js +141 -0
- package/dist/parquetjs/parquet-thrift/PageType.d.ts +7 -0
- package/dist/parquetjs/parquet-thrift/PageType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/PageType.js +16 -0
- package/dist/parquetjs/parquet-thrift/RowGroup.d.ts +20 -0
- package/dist/parquetjs/parquet-thrift/RowGroup.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/RowGroup.js +182 -0
- package/dist/parquetjs/parquet-thrift/SchemaElement.d.ts +33 -0
- package/dist/parquetjs/parquet-thrift/SchemaElement.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/SchemaElement.js +239 -0
- package/dist/parquetjs/parquet-thrift/SortingColumn.d.ts +15 -0
- package/dist/parquetjs/parquet-thrift/SortingColumn.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/SortingColumn.js +127 -0
- package/dist/parquetjs/parquet-thrift/Statistics.d.ts +23 -0
- package/dist/parquetjs/parquet-thrift/Statistics.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/Statistics.js +176 -0
- package/dist/parquetjs/parquet-thrift/StringType.d.ts +9 -0
- package/dist/parquetjs/parquet-thrift/StringType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/StringType.js +58 -0
- package/dist/parquetjs/parquet-thrift/TimeType.d.ts +14 -0
- package/dist/parquetjs/parquet-thrift/TimeType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/TimeType.js +106 -0
- package/dist/parquetjs/parquet-thrift/TimeUnit.d.ts +17 -0
- package/dist/parquetjs/parquet-thrift/TimeUnit.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/TimeUnit.js +127 -0
- package/dist/parquetjs/parquet-thrift/TimestampType.d.ts +14 -0
- package/dist/parquetjs/parquet-thrift/TimestampType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/TimestampType.js +106 -0
- package/dist/parquetjs/parquet-thrift/Type.d.ts +11 -0
- package/dist/parquetjs/parquet-thrift/Type.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/Type.js +20 -0
- package/dist/parquetjs/parquet-thrift/TypeDefinedOrder.d.ts +9 -0
- package/dist/parquetjs/parquet-thrift/TypeDefinedOrder.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/TypeDefinedOrder.js +58 -0
- package/dist/parquetjs/parquet-thrift/UUIDType.d.ts +9 -0
- package/dist/parquetjs/parquet-thrift/UUIDType.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/UUIDType.js +58 -0
- package/dist/parquetjs/parquet-thrift/index.d.ts +44 -0
- package/dist/parquetjs/parquet-thrift/index.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/index.js +61 -0
- package/dist/parquetjs/parser/decoders.d.ts +34 -0
- package/dist/parquetjs/parser/decoders.d.ts.map +1 -0
- package/dist/parquetjs/parser/decoders.js +318 -0
- package/dist/parquetjs/parser/parquet-cursor.d.ts +36 -0
- package/dist/parquetjs/parser/parquet-cursor.d.ts.map +1 -0
- package/dist/parquetjs/parser/parquet-cursor.js +74 -0
- package/dist/parquetjs/parser/parquet-envelope-reader.d.ts +40 -0
- package/dist/parquetjs/parser/parquet-envelope-reader.d.ts.map +1 -0
- package/dist/parquetjs/parser/parquet-envelope-reader.js +136 -0
- package/dist/parquetjs/parser/parquet-reader.d.ts +68 -0
- package/dist/parquetjs/parser/parquet-reader.d.ts.map +1 -0
- package/dist/parquetjs/parser/parquet-reader.js +134 -0
- package/dist/parquetjs/schema/declare.d.ts +80 -0
- package/dist/parquetjs/schema/declare.d.ts.map +1 -0
- package/dist/parquetjs/schema/declare.js +10 -0
- package/dist/parquetjs/schema/schema.d.ts +26 -0
- package/dist/parquetjs/schema/schema.d.ts.map +1 -0
- package/dist/parquetjs/schema/schema.js +162 -0
- package/dist/parquetjs/schema/shred.d.ts +48 -0
- package/dist/parquetjs/schema/shred.d.ts.map +1 -0
- package/dist/parquetjs/schema/shred.js +225 -0
- package/dist/parquetjs/schema/types.d.ts +20 -0
- package/dist/parquetjs/schema/types.d.ts.map +1 -0
- package/dist/parquetjs/schema/types.js +418 -0
- package/dist/parquetjs/utils/buffer-utils.d.ts +10 -0
- package/dist/parquetjs/utils/buffer-utils.d.ts.map +1 -0
- package/dist/parquetjs/utils/buffer-utils.js +22 -0
- package/dist/parquetjs/utils/file-utils.d.ts +16 -0
- package/dist/parquetjs/utils/file-utils.d.ts.map +1 -0
- package/dist/parquetjs/utils/file-utils.js +46 -0
- package/dist/parquetjs/utils/read-utils.d.ts +25 -0
- package/dist/parquetjs/utils/read-utils.d.ts.map +1 -0
- package/dist/parquetjs/utils/read-utils.js +109 -0
- package/dist/workers/parquet-worker.d.ts +2 -0
- package/dist/workers/parquet-worker.d.ts.map +1 -0
- package/dist/workers/parquet-worker.js +5 -0
- package/package.json +8 -8
- package/src/index.ts +3 -3
- package/src/parquetjs/compression.ts +10 -10
- package/src/parquetjs/parser/decoders.ts +1 -1
- package/src/parquetjs/parser/parquet-envelope-reader.ts +0 -11
- package/src/parquetjs/parser/parquet-reader.ts +0 -16
- package/src/parquetjs/utils/file-utils.ts +0 -49
|
@@ -2,29 +2,11 @@
|
|
|
2
2
|
|
|
3
3
|
var _interopRequireDefault = require("@babel/runtime/helpers/interopRequireDefault");
|
|
4
4
|
|
|
5
|
-
var _typeof = require("@babel/runtime/helpers/typeof");
|
|
6
|
-
|
|
7
5
|
Object.defineProperty(exports, "__esModule", {
|
|
8
6
|
value: true
|
|
9
7
|
});
|
|
10
8
|
exports.ParquetTransformer = exports.ParquetEnvelopeWriter = exports.ParquetWriter = void 0;
|
|
11
9
|
|
|
12
|
-
var _regenerator = _interopRequireDefault(require("@babel/runtime/regenerator"));
|
|
13
|
-
|
|
14
|
-
var _assertThisInitialized2 = _interopRequireDefault(require("@babel/runtime/helpers/assertThisInitialized"));
|
|
15
|
-
|
|
16
|
-
var _inherits2 = _interopRequireDefault(require("@babel/runtime/helpers/inherits"));
|
|
17
|
-
|
|
18
|
-
var _possibleConstructorReturn2 = _interopRequireDefault(require("@babel/runtime/helpers/possibleConstructorReturn"));
|
|
19
|
-
|
|
20
|
-
var _getPrototypeOf2 = _interopRequireDefault(require("@babel/runtime/helpers/getPrototypeOf"));
|
|
21
|
-
|
|
22
|
-
var _asyncToGenerator2 = _interopRequireDefault(require("@babel/runtime/helpers/asyncToGenerator"));
|
|
23
|
-
|
|
24
|
-
var _classCallCheck2 = _interopRequireDefault(require("@babel/runtime/helpers/classCallCheck"));
|
|
25
|
-
|
|
26
|
-
var _createClass2 = _interopRequireDefault(require("@babel/runtime/helpers/createClass"));
|
|
27
|
-
|
|
28
10
|
var _defineProperty2 = _interopRequireDefault(require("@babel/runtime/helpers/defineProperty"));
|
|
29
11
|
|
|
30
12
|
var _stream = require("stream");
|
|
@@ -43,30 +25,33 @@ var _readUtils = require("../utils/read-utils");
|
|
|
43
25
|
|
|
44
26
|
var _nodeInt = _interopRequireDefault(require("node-int64"));
|
|
45
27
|
|
|
46
|
-
function _getRequireWildcardCache(nodeInterop) { if (typeof WeakMap !== "function") return null; var cacheBabelInterop = new WeakMap(); var cacheNodeInterop = new WeakMap(); return (_getRequireWildcardCache = function
|
|
47
|
-
|
|
48
|
-
function _interopRequireWildcard(obj, nodeInterop) { if (!nodeInterop && obj && obj.__esModule) { return obj; } if (obj === null || _typeof(obj) !== "object" && typeof obj !== "function") { return { default: obj }; } var cache = _getRequireWildcardCache(nodeInterop); if (cache && cache.has(obj)) { return cache.get(obj); } var newObj = {}; var hasPropertyDescriptor = Object.defineProperty && Object.getOwnPropertyDescriptor; for (var key in obj) { if (key !== "default" && Object.prototype.hasOwnProperty.call(obj, key)) { var desc = hasPropertyDescriptor ? Object.getOwnPropertyDescriptor(obj, key) : null; if (desc && (desc.get || desc.set)) { Object.defineProperty(newObj, key, desc); } else { newObj[key] = obj[key]; } } } newObj.default = obj; if (cache) { cache.set(obj, newObj); } return newObj; }
|
|
28
|
+
function _getRequireWildcardCache(nodeInterop) { if (typeof WeakMap !== "function") return null; var cacheBabelInterop = new WeakMap(); var cacheNodeInterop = new WeakMap(); return (_getRequireWildcardCache = function (nodeInterop) { return nodeInterop ? cacheNodeInterop : cacheBabelInterop; })(nodeInterop); }
|
|
49
29
|
|
|
50
|
-
function
|
|
30
|
+
function _interopRequireWildcard(obj, nodeInterop) { if (!nodeInterop && obj && obj.__esModule) { return obj; } if (obj === null || typeof obj !== "object" && typeof obj !== "function") { return { default: obj }; } var cache = _getRequireWildcardCache(nodeInterop); if (cache && cache.has(obj)) { return cache.get(obj); } var newObj = {}; var hasPropertyDescriptor = Object.defineProperty && Object.getOwnPropertyDescriptor; for (var key in obj) { if (key !== "default" && Object.prototype.hasOwnProperty.call(obj, key)) { var desc = hasPropertyDescriptor ? Object.getOwnPropertyDescriptor(obj, key) : null; if (desc && (desc.get || desc.set)) { Object.defineProperty(newObj, key, desc); } else { newObj[key] = obj[key]; } } } newObj.default = obj; if (cache) { cache.set(obj, newObj); } return newObj; }
|
|
51
31
|
|
|
52
|
-
|
|
32
|
+
const PARQUET_MAGIC = 'PAR1';
|
|
33
|
+
const PARQUET_VERSION = 1;
|
|
34
|
+
const PARQUET_DEFAULT_PAGE_SIZE = 8192;
|
|
35
|
+
const PARQUET_DEFAULT_ROW_GROUP_SIZE = 4096;
|
|
36
|
+
const PARQUET_RDLVL_TYPE = 'INT32';
|
|
37
|
+
const PARQUET_RDLVL_ENCODING = 'RLE';
|
|
53
38
|
|
|
54
|
-
|
|
55
|
-
|
|
56
|
-
|
|
39
|
+
class ParquetWriter {
|
|
40
|
+
static async openFile(schema, path, opts) {
|
|
41
|
+
const outputStream = await (0, _fileUtils.osopen)(path, opts);
|
|
42
|
+
return ParquetWriter.openStream(schema, outputStream, opts);
|
|
43
|
+
}
|
|
57
44
|
|
|
58
|
-
|
|
45
|
+
static async openStream(schema, outputStream, opts) {
|
|
46
|
+
if (!opts) {
|
|
47
|
+
opts = {};
|
|
48
|
+
}
|
|
59
49
|
|
|
60
|
-
|
|
61
|
-
|
|
62
|
-
|
|
63
|
-
var PARQUET_DEFAULT_ROW_GROUP_SIZE = 4096;
|
|
64
|
-
var PARQUET_RDLVL_TYPE = 'INT32';
|
|
65
|
-
var PARQUET_RDLVL_ENCODING = 'RLE';
|
|
50
|
+
const envelopeWriter = await ParquetEnvelopeWriter.openStream(schema, outputStream, opts);
|
|
51
|
+
return new ParquetWriter(schema, envelopeWriter, opts);
|
|
52
|
+
}
|
|
66
53
|
|
|
67
|
-
|
|
68
|
-
function ParquetWriter(schema, envelopeWriter, opts) {
|
|
69
|
-
(0, _classCallCheck2.default)(this, ParquetWriter);
|
|
54
|
+
constructor(schema, envelopeWriter, opts) {
|
|
70
55
|
(0, _defineProperty2.default)(this, "schema", void 0);
|
|
71
56
|
(0, _defineProperty2.default)(this, "envelopeWriter", void 0);
|
|
72
57
|
(0, _defineProperty2.default)(this, "rowBuffer", void 0);
|
|
@@ -82,217 +67,72 @@ var ParquetWriter = function () {
|
|
|
82
67
|
this.writeHeader();
|
|
83
68
|
}
|
|
84
69
|
|
|
85
|
-
(
|
|
86
|
-
|
|
87
|
-
|
|
88
|
-
|
|
89
|
-
|
|
90
|
-
|
|
91
|
-
switch (_context.prev = _context.next) {
|
|
92
|
-
case 0:
|
|
93
|
-
_context.prev = 0;
|
|
94
|
-
_context.next = 3;
|
|
95
|
-
return this.envelopeWriter.writeHeader();
|
|
96
|
-
|
|
97
|
-
case 3:
|
|
98
|
-
_context.next = 10;
|
|
99
|
-
break;
|
|
100
|
-
|
|
101
|
-
case 5:
|
|
102
|
-
_context.prev = 5;
|
|
103
|
-
_context.t0 = _context["catch"](0);
|
|
104
|
-
_context.next = 9;
|
|
105
|
-
return this.envelopeWriter.close();
|
|
106
|
-
|
|
107
|
-
case 9:
|
|
108
|
-
throw _context.t0;
|
|
109
|
-
|
|
110
|
-
case 10:
|
|
111
|
-
case "end":
|
|
112
|
-
return _context.stop();
|
|
113
|
-
}
|
|
114
|
-
}
|
|
115
|
-
}, _callee, this, [[0, 5]]);
|
|
116
|
-
}));
|
|
117
|
-
|
|
118
|
-
function writeHeader() {
|
|
119
|
-
return _writeHeader.apply(this, arguments);
|
|
120
|
-
}
|
|
121
|
-
|
|
122
|
-
return writeHeader;
|
|
123
|
-
}()
|
|
124
|
-
}, {
|
|
125
|
-
key: "appendRow",
|
|
126
|
-
value: function () {
|
|
127
|
-
var _appendRow = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee2(row) {
|
|
128
|
-
return _regenerator.default.wrap(function _callee2$(_context2) {
|
|
129
|
-
while (1) {
|
|
130
|
-
switch (_context2.prev = _context2.next) {
|
|
131
|
-
case 0:
|
|
132
|
-
if (!this.closed) {
|
|
133
|
-
_context2.next = 2;
|
|
134
|
-
break;
|
|
135
|
-
}
|
|
136
|
-
|
|
137
|
-
throw new Error('writer was closed');
|
|
138
|
-
|
|
139
|
-
case 2:
|
|
140
|
-
Shred.shredRecord(this.schema, row, this.rowBuffer);
|
|
141
|
-
|
|
142
|
-
if (this.rowBuffer.rowCount >= this.rowGroupSize) {
|
|
143
|
-
this.rowBuffer = {};
|
|
144
|
-
}
|
|
145
|
-
|
|
146
|
-
case 4:
|
|
147
|
-
case "end":
|
|
148
|
-
return _context2.stop();
|
|
149
|
-
}
|
|
150
|
-
}
|
|
151
|
-
}, _callee2, this);
|
|
152
|
-
}));
|
|
153
|
-
|
|
154
|
-
function appendRow(_x) {
|
|
155
|
-
return _appendRow.apply(this, arguments);
|
|
156
|
-
}
|
|
157
|
-
|
|
158
|
-
return appendRow;
|
|
159
|
-
}()
|
|
160
|
-
}, {
|
|
161
|
-
key: "close",
|
|
162
|
-
value: function () {
|
|
163
|
-
var _close = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee3(callback) {
|
|
164
|
-
return _regenerator.default.wrap(function _callee3$(_context3) {
|
|
165
|
-
while (1) {
|
|
166
|
-
switch (_context3.prev = _context3.next) {
|
|
167
|
-
case 0:
|
|
168
|
-
if (!this.closed) {
|
|
169
|
-
_context3.next = 2;
|
|
170
|
-
break;
|
|
171
|
-
}
|
|
172
|
-
|
|
173
|
-
throw new Error('writer was closed');
|
|
174
|
-
|
|
175
|
-
case 2:
|
|
176
|
-
this.closed = true;
|
|
177
|
-
|
|
178
|
-
if (this.rowBuffer.rowCount > 0 || this.rowBuffer.rowCount >= this.rowGroupSize) {
|
|
179
|
-
this.rowBuffer = {};
|
|
180
|
-
}
|
|
181
|
-
|
|
182
|
-
_context3.next = 6;
|
|
183
|
-
return this.envelopeWriter.writeFooter(this.userMetadata);
|
|
184
|
-
|
|
185
|
-
case 6:
|
|
186
|
-
_context3.next = 8;
|
|
187
|
-
return this.envelopeWriter.close();
|
|
188
|
-
|
|
189
|
-
case 8:
|
|
190
|
-
if (callback) {
|
|
191
|
-
callback();
|
|
192
|
-
}
|
|
193
|
-
|
|
194
|
-
case 9:
|
|
195
|
-
case "end":
|
|
196
|
-
return _context3.stop();
|
|
197
|
-
}
|
|
198
|
-
}
|
|
199
|
-
}, _callee3, this);
|
|
200
|
-
}));
|
|
201
|
-
|
|
202
|
-
function close(_x2) {
|
|
203
|
-
return _close.apply(this, arguments);
|
|
204
|
-
}
|
|
205
|
-
|
|
206
|
-
return close;
|
|
207
|
-
}()
|
|
208
|
-
}, {
|
|
209
|
-
key: "setMetadata",
|
|
210
|
-
value: function setMetadata(key, value) {
|
|
211
|
-
this.userMetadata[String(key)] = String(value);
|
|
70
|
+
async writeHeader() {
|
|
71
|
+
try {
|
|
72
|
+
await this.envelopeWriter.writeHeader();
|
|
73
|
+
} catch (err) {
|
|
74
|
+
await this.envelopeWriter.close();
|
|
75
|
+
throw err;
|
|
212
76
|
}
|
|
213
|
-
}
|
|
214
|
-
|
|
215
|
-
|
|
216
|
-
|
|
77
|
+
}
|
|
78
|
+
|
|
79
|
+
async appendRow(row) {
|
|
80
|
+
if (this.closed) {
|
|
81
|
+
throw new Error('writer was closed');
|
|
217
82
|
}
|
|
218
|
-
|
|
219
|
-
|
|
220
|
-
|
|
221
|
-
|
|
83
|
+
|
|
84
|
+
Shred.shredRecord(this.schema, row, this.rowBuffer);
|
|
85
|
+
|
|
86
|
+
if (this.rowBuffer.rowCount >= this.rowGroupSize) {
|
|
87
|
+
this.rowBuffer = {};
|
|
222
88
|
}
|
|
223
|
-
}
|
|
224
|
-
|
|
225
|
-
|
|
226
|
-
|
|
227
|
-
|
|
228
|
-
|
|
229
|
-
|
|
230
|
-
|
|
231
|
-
|
|
232
|
-
|
|
233
|
-
|
|
234
|
-
|
|
235
|
-
|
|
236
|
-
|
|
237
|
-
|
|
238
|
-
|
|
239
|
-
|
|
240
|
-
|
|
241
|
-
|
|
242
|
-
|
|
243
|
-
|
|
244
|
-
|
|
245
|
-
|
|
246
|
-
|
|
247
|
-
|
|
248
|
-
|
|
249
|
-
|
|
250
|
-
|
|
251
|
-
|
|
252
|
-
|
|
253
|
-
|
|
254
|
-
|
|
255
|
-
|
|
256
|
-
|
|
257
|
-
var envelopeWriter;
|
|
258
|
-
return _regenerator.default.wrap(function _callee5$(_context5) {
|
|
259
|
-
while (1) {
|
|
260
|
-
switch (_context5.prev = _context5.next) {
|
|
261
|
-
case 0:
|
|
262
|
-
if (!opts) {
|
|
263
|
-
opts = {};
|
|
264
|
-
}
|
|
265
|
-
|
|
266
|
-
_context5.next = 3;
|
|
267
|
-
return ParquetEnvelopeWriter.openStream(schema, outputStream, opts);
|
|
268
|
-
|
|
269
|
-
case 3:
|
|
270
|
-
envelopeWriter = _context5.sent;
|
|
271
|
-
return _context5.abrupt("return", new ParquetWriter(schema, envelopeWriter, opts));
|
|
272
|
-
|
|
273
|
-
case 5:
|
|
274
|
-
case "end":
|
|
275
|
-
return _context5.stop();
|
|
276
|
-
}
|
|
277
|
-
}
|
|
278
|
-
}, _callee5);
|
|
279
|
-
}));
|
|
280
|
-
|
|
281
|
-
function openStream(_x6, _x7, _x8) {
|
|
282
|
-
return _openStream.apply(this, arguments);
|
|
283
|
-
}
|
|
284
|
-
|
|
285
|
-
return openStream;
|
|
286
|
-
}()
|
|
287
|
-
}]);
|
|
288
|
-
return ParquetWriter;
|
|
289
|
-
}();
|
|
89
|
+
}
|
|
90
|
+
|
|
91
|
+
async close(callback) {
|
|
92
|
+
if (this.closed) {
|
|
93
|
+
throw new Error('writer was closed');
|
|
94
|
+
}
|
|
95
|
+
|
|
96
|
+
this.closed = true;
|
|
97
|
+
|
|
98
|
+
if (this.rowBuffer.rowCount > 0 || this.rowBuffer.rowCount >= this.rowGroupSize) {
|
|
99
|
+
this.rowBuffer = {};
|
|
100
|
+
}
|
|
101
|
+
|
|
102
|
+
await this.envelopeWriter.writeFooter(this.userMetadata);
|
|
103
|
+
await this.envelopeWriter.close();
|
|
104
|
+
|
|
105
|
+
if (callback) {
|
|
106
|
+
callback();
|
|
107
|
+
}
|
|
108
|
+
}
|
|
109
|
+
|
|
110
|
+
setMetadata(key, value) {
|
|
111
|
+
this.userMetadata[String(key)] = String(value);
|
|
112
|
+
}
|
|
113
|
+
|
|
114
|
+
setRowGroupSize(cnt) {
|
|
115
|
+
this.rowGroupSize = cnt;
|
|
116
|
+
}
|
|
117
|
+
|
|
118
|
+
setPageSize(cnt) {
|
|
119
|
+
this.envelopeWriter.setPageSize(cnt);
|
|
120
|
+
}
|
|
121
|
+
|
|
122
|
+
}
|
|
290
123
|
|
|
291
124
|
exports.ParquetWriter = ParquetWriter;
|
|
292
125
|
|
|
293
|
-
|
|
294
|
-
|
|
295
|
-
|
|
126
|
+
class ParquetEnvelopeWriter {
|
|
127
|
+
static async openStream(schema, outputStream, opts) {
|
|
128
|
+
const writeFn = _fileUtils.oswrite.bind(undefined, outputStream);
|
|
129
|
+
|
|
130
|
+
const closeFn = _fileUtils.osclose.bind(undefined, outputStream);
|
|
131
|
+
|
|
132
|
+
return new ParquetEnvelopeWriter(schema, writeFn, closeFn, 0, opts);
|
|
133
|
+
}
|
|
134
|
+
|
|
135
|
+
constructor(schema, writeFn, closeFn, fileOffset, opts) {
|
|
296
136
|
(0, _defineProperty2.default)(this, "schema", void 0);
|
|
297
137
|
(0, _defineProperty2.default)(this, "write", void 0);
|
|
298
138
|
(0, _defineProperty2.default)(this, "close", void 0);
|
|
@@ -311,194 +151,72 @@ var ParquetEnvelopeWriter = function () {
|
|
|
311
151
|
this.useDataPageV2 = 'useDataPageV2' in opts ? Boolean(opts.useDataPageV2) : false;
|
|
312
152
|
}
|
|
313
153
|
|
|
314
|
-
(
|
|
315
|
-
|
|
316
|
-
|
|
317
|
-
|
|
318
|
-
|
|
319
|
-
|
|
320
|
-
|
|
321
|
-
|
|
322
|
-
|
|
323
|
-
|
|
324
|
-
|
|
325
|
-
|
|
326
|
-
|
|
327
|
-
|
|
328
|
-
|
|
329
|
-
|
|
330
|
-
|
|
331
|
-
|
|
332
|
-
|
|
333
|
-
|
|
334
|
-
|
|
335
|
-
|
|
336
|
-
|
|
337
|
-
pageSize: this.pageSize,
|
|
338
|
-
useDataPageV2: this.useDataPageV2
|
|
339
|
-
});
|
|
340
|
-
|
|
341
|
-
case 2:
|
|
342
|
-
rgroup = _context6.sent;
|
|
343
|
-
this.rowCount += records.rowCount;
|
|
344
|
-
this.rowGroups.push(rgroup.metadata);
|
|
345
|
-
_context6.next = 7;
|
|
346
|
-
return this.writeSection(rgroup.body);
|
|
347
|
-
|
|
348
|
-
case 7:
|
|
349
|
-
return _context6.abrupt("return", _context6.sent);
|
|
350
|
-
|
|
351
|
-
case 8:
|
|
352
|
-
case "end":
|
|
353
|
-
return _context6.stop();
|
|
354
|
-
}
|
|
355
|
-
}
|
|
356
|
-
}, _callee6, this);
|
|
357
|
-
}));
|
|
358
|
-
|
|
359
|
-
function writeRowGroup(_x9) {
|
|
360
|
-
return _writeRowGroup.apply(this, arguments);
|
|
361
|
-
}
|
|
362
|
-
|
|
363
|
-
return writeRowGroup;
|
|
364
|
-
}()
|
|
365
|
-
}, {
|
|
366
|
-
key: "writeFooter",
|
|
367
|
-
value: function writeFooter(userMetadata) {
|
|
368
|
-
if (!userMetadata) {
|
|
369
|
-
userMetadata = {};
|
|
370
|
-
}
|
|
371
|
-
|
|
372
|
-
return this.writeSection(encodeFooter(this.schema, this.rowCount, this.rowGroups, userMetadata));
|
|
373
|
-
}
|
|
374
|
-
}, {
|
|
375
|
-
key: "setPageSize",
|
|
376
|
-
value: function setPageSize(cnt) {
|
|
377
|
-
this.pageSize = cnt;
|
|
154
|
+
writeSection(buf) {
|
|
155
|
+
this.offset += buf.length;
|
|
156
|
+
return this.write(buf);
|
|
157
|
+
}
|
|
158
|
+
|
|
159
|
+
writeHeader() {
|
|
160
|
+
return this.writeSection(Buffer.from(PARQUET_MAGIC));
|
|
161
|
+
}
|
|
162
|
+
|
|
163
|
+
async writeRowGroup(records) {
|
|
164
|
+
const rgroup = await encodeRowGroup(this.schema, records, {
|
|
165
|
+
baseOffset: this.offset,
|
|
166
|
+
pageSize: this.pageSize,
|
|
167
|
+
useDataPageV2: this.useDataPageV2
|
|
168
|
+
});
|
|
169
|
+
this.rowCount += records.rowCount;
|
|
170
|
+
this.rowGroups.push(rgroup.metadata);
|
|
171
|
+
return await this.writeSection(rgroup.body);
|
|
172
|
+
}
|
|
173
|
+
|
|
174
|
+
writeFooter(userMetadata) {
|
|
175
|
+
if (!userMetadata) {
|
|
176
|
+
userMetadata = {};
|
|
378
177
|
}
|
|
379
|
-
}], [{
|
|
380
|
-
key: "openStream",
|
|
381
|
-
value: function () {
|
|
382
|
-
var _openStream2 = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee7(schema, outputStream, opts) {
|
|
383
|
-
var writeFn, closeFn;
|
|
384
|
-
return _regenerator.default.wrap(function _callee7$(_context7) {
|
|
385
|
-
while (1) {
|
|
386
|
-
switch (_context7.prev = _context7.next) {
|
|
387
|
-
case 0:
|
|
388
|
-
writeFn = _fileUtils.oswrite.bind(undefined, outputStream);
|
|
389
|
-
closeFn = _fileUtils.osclose.bind(undefined, outputStream);
|
|
390
|
-
return _context7.abrupt("return", new ParquetEnvelopeWriter(schema, writeFn, closeFn, 0, opts));
|
|
391
|
-
|
|
392
|
-
case 3:
|
|
393
|
-
case "end":
|
|
394
|
-
return _context7.stop();
|
|
395
|
-
}
|
|
396
|
-
}
|
|
397
|
-
}, _callee7);
|
|
398
|
-
}));
|
|
399
|
-
|
|
400
|
-
function openStream(_x10, _x11, _x12) {
|
|
401
|
-
return _openStream2.apply(this, arguments);
|
|
402
|
-
}
|
|
403
|
-
|
|
404
|
-
return openStream;
|
|
405
|
-
}()
|
|
406
|
-
}]);
|
|
407
|
-
return ParquetEnvelopeWriter;
|
|
408
|
-
}();
|
|
409
178
|
|
|
410
|
-
|
|
179
|
+
return this.writeSection(encodeFooter(this.schema, this.rowCount, this.rowGroups, userMetadata));
|
|
180
|
+
}
|
|
411
181
|
|
|
412
|
-
|
|
413
|
-
|
|
182
|
+
setPageSize(cnt) {
|
|
183
|
+
this.pageSize = cnt;
|
|
184
|
+
}
|
|
414
185
|
|
|
415
|
-
|
|
186
|
+
}
|
|
416
187
|
|
|
417
|
-
|
|
418
|
-
var _this;
|
|
188
|
+
exports.ParquetEnvelopeWriter = ParquetEnvelopeWriter;
|
|
419
189
|
|
|
420
|
-
|
|
421
|
-
|
|
422
|
-
|
|
190
|
+
class ParquetTransformer extends _stream.Transform {
|
|
191
|
+
constructor(schema, opts = {}) {
|
|
192
|
+
super({
|
|
423
193
|
objectMode: true
|
|
424
194
|
});
|
|
425
|
-
(0, _defineProperty2.default)(
|
|
426
|
-
|
|
427
|
-
var writeProxy = function (t) {
|
|
428
|
-
return function () {
|
|
429
|
-
var _ref = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee8(b) {
|
|
430
|
-
return _regenerator.default.wrap(function _callee8$(_context8) {
|
|
431
|
-
while (1) {
|
|
432
|
-
switch (_context8.prev = _context8.next) {
|
|
433
|
-
case 0:
|
|
434
|
-
t.push(b);
|
|
435
|
-
|
|
436
|
-
case 1:
|
|
437
|
-
case "end":
|
|
438
|
-
return _context8.stop();
|
|
439
|
-
}
|
|
440
|
-
}
|
|
441
|
-
}, _callee8);
|
|
442
|
-
}));
|
|
443
|
-
|
|
444
|
-
return function (_x13) {
|
|
445
|
-
return _ref.apply(this, arguments);
|
|
446
|
-
};
|
|
447
|
-
}();
|
|
448
|
-
}((0, _assertThisInitialized2.default)(_this));
|
|
449
|
-
|
|
450
|
-
_this.writer = new ParquetWriter(schema, new ParquetEnvelopeWriter(schema, writeProxy, (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee9() {
|
|
451
|
-
return _regenerator.default.wrap(function _callee9$(_context9) {
|
|
452
|
-
while (1) {
|
|
453
|
-
switch (_context9.prev = _context9.next) {
|
|
454
|
-
case 0:
|
|
455
|
-
case "end":
|
|
456
|
-
return _context9.stop();
|
|
457
|
-
}
|
|
458
|
-
}
|
|
459
|
-
}, _callee9);
|
|
460
|
-
})), 0, opts), opts);
|
|
461
|
-
return _this;
|
|
462
|
-
}
|
|
195
|
+
(0, _defineProperty2.default)(this, "writer", void 0);
|
|
463
196
|
|
|
464
|
-
|
|
465
|
-
|
|
466
|
-
|
|
467
|
-
|
|
468
|
-
|
|
469
|
-
}
|
|
197
|
+
const writeProxy = function (t) {
|
|
198
|
+
return async function (b) {
|
|
199
|
+
t.push(b);
|
|
200
|
+
};
|
|
201
|
+
}(this);
|
|
470
202
|
|
|
471
|
-
|
|
472
|
-
|
|
203
|
+
this.writer = new ParquetWriter(schema, new ParquetEnvelopeWriter(schema, writeProxy, async () => {}, 0, opts), opts);
|
|
204
|
+
}
|
|
205
|
+
|
|
206
|
+
_transform(row, encoding, callback) {
|
|
207
|
+
if (row) {
|
|
208
|
+
return this.writer.appendRow(row).then(callback);
|
|
473
209
|
}
|
|
474
|
-
|
|
475
|
-
|
|
476
|
-
|
|
477
|
-
|
|
478
|
-
|
|
479
|
-
|
|
480
|
-
|
|
481
|
-
|
|
482
|
-
|
|
483
|
-
|
|
484
|
-
|
|
485
|
-
case 2:
|
|
486
|
-
case "end":
|
|
487
|
-
return _context10.stop();
|
|
488
|
-
}
|
|
489
|
-
}
|
|
490
|
-
}, _callee10, this);
|
|
491
|
-
}));
|
|
492
|
-
|
|
493
|
-
function _flush(_x14) {
|
|
494
|
-
return _flush2.apply(this, arguments);
|
|
495
|
-
}
|
|
496
|
-
|
|
497
|
-
return _flush;
|
|
498
|
-
}()
|
|
499
|
-
}]);
|
|
500
|
-
return ParquetTransformer;
|
|
501
|
-
}(_stream.Transform);
|
|
210
|
+
|
|
211
|
+
callback();
|
|
212
|
+
return Promise.resolve();
|
|
213
|
+
}
|
|
214
|
+
|
|
215
|
+
async _flush(callback) {
|
|
216
|
+
await this.writer.close(callback);
|
|
217
|
+
}
|
|
218
|
+
|
|
219
|
+
}
|
|
502
220
|
|
|
503
221
|
exports.ParquetTransformer = ParquetTransformer;
|
|
504
222
|
|
|
@@ -510,304 +228,160 @@ function encodeValues(type, encoding, values, opts) {
|
|
|
510
228
|
return _codecs.PARQUET_CODECS[encoding].encodeValues(type, values, opts);
|
|
511
229
|
}
|
|
512
230
|
|
|
513
|
-
function encodeDataPage(
|
|
514
|
-
|
|
515
|
-
}
|
|
231
|
+
async function encodeDataPage(column, data) {
|
|
232
|
+
let rLevelsBuf = Buffer.alloc(0);
|
|
516
233
|
|
|
517
|
-
|
|
518
|
-
|
|
519
|
-
|
|
520
|
-
|
|
521
|
-
|
|
522
|
-
switch (_context11.prev = _context11.next) {
|
|
523
|
-
case 0:
|
|
524
|
-
rLevelsBuf = Buffer.alloc(0);
|
|
525
|
-
|
|
526
|
-
if (column.rLevelMax > 0) {
|
|
527
|
-
rLevelsBuf = encodeValues(PARQUET_RDLVL_TYPE, PARQUET_RDLVL_ENCODING, data.rlevels, {
|
|
528
|
-
bitWidth: (0, _readUtils.getBitWidth)(column.rLevelMax)
|
|
529
|
-
});
|
|
530
|
-
}
|
|
531
|
-
|
|
532
|
-
dLevelsBuf = Buffer.alloc(0);
|
|
533
|
-
|
|
534
|
-
if (column.dLevelMax > 0) {
|
|
535
|
-
dLevelsBuf = encodeValues(PARQUET_RDLVL_TYPE, PARQUET_RDLVL_ENCODING, data.dlevels, {
|
|
536
|
-
bitWidth: (0, _readUtils.getBitWidth)(column.dLevelMax)
|
|
537
|
-
});
|
|
538
|
-
}
|
|
539
|
-
|
|
540
|
-
valuesBuf = encodeValues(column.primitiveType, column.encoding, data.values, {
|
|
541
|
-
typeLength: column.typeLength,
|
|
542
|
-
bitWidth: column.typeLength
|
|
543
|
-
});
|
|
544
|
-
dataBuf = Buffer.concat([rLevelsBuf, dLevelsBuf, valuesBuf]);
|
|
545
|
-
_context11.next = 8;
|
|
546
|
-
return Compression.deflate(column.compression, dataBuf);
|
|
547
|
-
|
|
548
|
-
case 8:
|
|
549
|
-
compressedBuf = _context11.sent;
|
|
550
|
-
header = new _parquetThrift.PageHeader({
|
|
551
|
-
type: _parquetThrift.PageType.DATA_PAGE,
|
|
552
|
-
data_page_header: new _parquetThrift.DataPageHeader({
|
|
553
|
-
num_values: data.count,
|
|
554
|
-
encoding: _parquetThrift.Encoding[column.encoding],
|
|
555
|
-
definition_level_encoding: _parquetThrift.Encoding[PARQUET_RDLVL_ENCODING],
|
|
556
|
-
repetition_level_encoding: _parquetThrift.Encoding[PARQUET_RDLVL_ENCODING]
|
|
557
|
-
}),
|
|
558
|
-
uncompressed_page_size: dataBuf.length,
|
|
559
|
-
compressed_page_size: compressedBuf.length
|
|
560
|
-
});
|
|
561
|
-
headerBuf = (0, _readUtils.serializeThrift)(header);
|
|
562
|
-
page = Buffer.concat([headerBuf, compressedBuf]);
|
|
563
|
-
return _context11.abrupt("return", {
|
|
564
|
-
header: header,
|
|
565
|
-
headerSize: headerBuf.length,
|
|
566
|
-
page: page
|
|
567
|
-
});
|
|
568
|
-
|
|
569
|
-
case 13:
|
|
570
|
-
case "end":
|
|
571
|
-
return _context11.stop();
|
|
572
|
-
}
|
|
573
|
-
}
|
|
574
|
-
}, _callee11);
|
|
575
|
-
}));
|
|
576
|
-
return _encodeDataPage.apply(this, arguments);
|
|
577
|
-
}
|
|
234
|
+
if (column.rLevelMax > 0) {
|
|
235
|
+
rLevelsBuf = encodeValues(PARQUET_RDLVL_TYPE, PARQUET_RDLVL_ENCODING, data.rlevels, {
|
|
236
|
+
bitWidth: (0, _readUtils.getBitWidth)(column.rLevelMax)
|
|
237
|
+
});
|
|
238
|
+
}
|
|
578
239
|
|
|
579
|
-
|
|
580
|
-
return _encodeDataPageV.apply(this, arguments);
|
|
581
|
-
}
|
|
240
|
+
let dLevelsBuf = Buffer.alloc(0);
|
|
582
241
|
|
|
583
|
-
|
|
584
|
-
|
|
585
|
-
|
|
586
|
-
|
|
587
|
-
|
|
588
|
-
switch (_context12.prev = _context12.next) {
|
|
589
|
-
case 0:
|
|
590
|
-
valuesBuf = encodeValues(column.primitiveType, column.encoding, data.values, {
|
|
591
|
-
typeLength: column.typeLength,
|
|
592
|
-
bitWidth: column.typeLength
|
|
593
|
-
});
|
|
594
|
-
_context12.next = 3;
|
|
595
|
-
return Compression.deflate(column.compression, valuesBuf);
|
|
596
|
-
|
|
597
|
-
case 3:
|
|
598
|
-
compressedBuf = _context12.sent;
|
|
599
|
-
rLevelsBuf = Buffer.alloc(0);
|
|
600
|
-
|
|
601
|
-
if (column.rLevelMax > 0) {
|
|
602
|
-
rLevelsBuf = encodeValues(PARQUET_RDLVL_TYPE, PARQUET_RDLVL_ENCODING, data.rlevels, {
|
|
603
|
-
bitWidth: (0, _readUtils.getBitWidth)(column.rLevelMax),
|
|
604
|
-
disableEnvelope: true
|
|
605
|
-
});
|
|
606
|
-
}
|
|
607
|
-
|
|
608
|
-
dLevelsBuf = Buffer.alloc(0);
|
|
609
|
-
|
|
610
|
-
if (column.dLevelMax > 0) {
|
|
611
|
-
dLevelsBuf = encodeValues(PARQUET_RDLVL_TYPE, PARQUET_RDLVL_ENCODING, data.dlevels, {
|
|
612
|
-
bitWidth: (0, _readUtils.getBitWidth)(column.dLevelMax),
|
|
613
|
-
disableEnvelope: true
|
|
614
|
-
});
|
|
615
|
-
}
|
|
616
|
-
|
|
617
|
-
header = new _parquetThrift.PageHeader({
|
|
618
|
-
type: _parquetThrift.PageType.DATA_PAGE_V2,
|
|
619
|
-
data_page_header_v2: new _parquetThrift.DataPageHeaderV2({
|
|
620
|
-
num_values: data.count,
|
|
621
|
-
num_nulls: data.count - data.values.length,
|
|
622
|
-
num_rows: rowCount,
|
|
623
|
-
encoding: _parquetThrift.Encoding[column.encoding],
|
|
624
|
-
definition_levels_byte_length: dLevelsBuf.length,
|
|
625
|
-
repetition_levels_byte_length: rLevelsBuf.length,
|
|
626
|
-
is_compressed: column.compression !== 'UNCOMPRESSED'
|
|
627
|
-
}),
|
|
628
|
-
uncompressed_page_size: rLevelsBuf.length + dLevelsBuf.length + valuesBuf.length,
|
|
629
|
-
compressed_page_size: rLevelsBuf.length + dLevelsBuf.length + compressedBuf.length
|
|
630
|
-
});
|
|
631
|
-
headerBuf = (0, _readUtils.serializeThrift)(header);
|
|
632
|
-
page = Buffer.concat([headerBuf, rLevelsBuf, dLevelsBuf, compressedBuf]);
|
|
633
|
-
return _context12.abrupt("return", {
|
|
634
|
-
header: header,
|
|
635
|
-
headerSize: headerBuf.length,
|
|
636
|
-
page: page
|
|
637
|
-
});
|
|
638
|
-
|
|
639
|
-
case 12:
|
|
640
|
-
case "end":
|
|
641
|
-
return _context12.stop();
|
|
642
|
-
}
|
|
643
|
-
}
|
|
644
|
-
}, _callee12);
|
|
645
|
-
}));
|
|
646
|
-
return _encodeDataPageV.apply(this, arguments);
|
|
647
|
-
}
|
|
242
|
+
if (column.dLevelMax > 0) {
|
|
243
|
+
dLevelsBuf = encodeValues(PARQUET_RDLVL_TYPE, PARQUET_RDLVL_ENCODING, data.dlevels, {
|
|
244
|
+
bitWidth: (0, _readUtils.getBitWidth)(column.dLevelMax)
|
|
245
|
+
});
|
|
246
|
+
}
|
|
648
247
|
|
|
649
|
-
|
|
650
|
-
|
|
248
|
+
const valuesBuf = encodeValues(column.primitiveType, column.encoding, data.values, {
|
|
249
|
+
typeLength: column.typeLength,
|
|
250
|
+
bitWidth: column.typeLength
|
|
251
|
+
});
|
|
252
|
+
const dataBuf = Buffer.concat([rLevelsBuf, dLevelsBuf, valuesBuf]);
|
|
253
|
+
const compressedBuf = await Compression.deflate(column.compression, dataBuf);
|
|
254
|
+
const header = new _parquetThrift.PageHeader({
|
|
255
|
+
type: _parquetThrift.PageType.DATA_PAGE,
|
|
256
|
+
data_page_header: new _parquetThrift.DataPageHeader({
|
|
257
|
+
num_values: data.count,
|
|
258
|
+
encoding: _parquetThrift.Encoding[column.encoding],
|
|
259
|
+
definition_level_encoding: _parquetThrift.Encoding[PARQUET_RDLVL_ENCODING],
|
|
260
|
+
repetition_level_encoding: _parquetThrift.Encoding[PARQUET_RDLVL_ENCODING]
|
|
261
|
+
}),
|
|
262
|
+
uncompressed_page_size: dataBuf.length,
|
|
263
|
+
compressed_page_size: compressedBuf.length
|
|
264
|
+
});
|
|
265
|
+
const headerBuf = (0, _readUtils.serializeThrift)(header);
|
|
266
|
+
const page = Buffer.concat([headerBuf, compressedBuf]);
|
|
267
|
+
return {
|
|
268
|
+
header,
|
|
269
|
+
headerSize: headerBuf.length,
|
|
270
|
+
page
|
|
271
|
+
};
|
|
651
272
|
}
|
|
652
273
|
|
|
653
|
-
function
|
|
654
|
-
|
|
655
|
-
|
|
656
|
-
|
|
657
|
-
|
|
658
|
-
|
|
659
|
-
|
|
660
|
-
|
|
661
|
-
|
|
662
|
-
|
|
663
|
-
|
|
664
|
-
|
|
665
|
-
|
|
666
|
-
|
|
667
|
-
|
|
668
|
-
|
|
669
|
-
|
|
670
|
-
|
|
671
|
-
|
|
672
|
-
|
|
673
|
-
|
|
674
|
-
|
|
675
|
-
|
|
676
|
-
|
|
677
|
-
|
|
678
|
-
|
|
679
|
-
|
|
680
|
-
|
|
681
|
-
|
|
682
|
-
|
|
683
|
-
|
|
684
|
-
|
|
685
|
-
|
|
686
|
-
|
|
687
|
-
|
|
688
|
-
|
|
689
|
-
|
|
690
|
-
|
|
691
|
-
|
|
692
|
-
|
|
693
|
-
|
|
694
|
-
|
|
695
|
-
|
|
696
|
-
|
|
697
|
-
|
|
698
|
-
codec: _parquetThrift.CompressionCodec[column.compression]
|
|
699
|
-
});
|
|
700
|
-
metadata.encodings.push(_parquetThrift.Encoding[PARQUET_RDLVL_ENCODING]);
|
|
701
|
-
metadata.encodings.push(_parquetThrift.Encoding[column.encoding]);
|
|
702
|
-
metadataOffset = baseOffset + pageBuf.length;
|
|
703
|
-
body = Buffer.concat([pageBuf, (0, _readUtils.serializeThrift)(metadata)]);
|
|
704
|
-
return _context13.abrupt("return", {
|
|
705
|
-
body: body,
|
|
706
|
-
metadata: metadata,
|
|
707
|
-
metadataOffset: metadataOffset
|
|
708
|
-
});
|
|
709
|
-
|
|
710
|
-
case 23:
|
|
711
|
-
case "end":
|
|
712
|
-
return _context13.stop();
|
|
713
|
-
}
|
|
714
|
-
}
|
|
715
|
-
}, _callee13);
|
|
716
|
-
}));
|
|
717
|
-
return _encodeColumnChunk.apply(this, arguments);
|
|
274
|
+
async function encodeDataPageV2(column, data, rowCount) {
|
|
275
|
+
const valuesBuf = encodeValues(column.primitiveType, column.encoding, data.values, {
|
|
276
|
+
typeLength: column.typeLength,
|
|
277
|
+
bitWidth: column.typeLength
|
|
278
|
+
});
|
|
279
|
+
const compressedBuf = await Compression.deflate(column.compression, valuesBuf);
|
|
280
|
+
let rLevelsBuf = Buffer.alloc(0);
|
|
281
|
+
|
|
282
|
+
if (column.rLevelMax > 0) {
|
|
283
|
+
rLevelsBuf = encodeValues(PARQUET_RDLVL_TYPE, PARQUET_RDLVL_ENCODING, data.rlevels, {
|
|
284
|
+
bitWidth: (0, _readUtils.getBitWidth)(column.rLevelMax),
|
|
285
|
+
disableEnvelope: true
|
|
286
|
+
});
|
|
287
|
+
}
|
|
288
|
+
|
|
289
|
+
let dLevelsBuf = Buffer.alloc(0);
|
|
290
|
+
|
|
291
|
+
if (column.dLevelMax > 0) {
|
|
292
|
+
dLevelsBuf = encodeValues(PARQUET_RDLVL_TYPE, PARQUET_RDLVL_ENCODING, data.dlevels, {
|
|
293
|
+
bitWidth: (0, _readUtils.getBitWidth)(column.dLevelMax),
|
|
294
|
+
disableEnvelope: true
|
|
295
|
+
});
|
|
296
|
+
}
|
|
297
|
+
|
|
298
|
+
const header = new _parquetThrift.PageHeader({
|
|
299
|
+
type: _parquetThrift.PageType.DATA_PAGE_V2,
|
|
300
|
+
data_page_header_v2: new _parquetThrift.DataPageHeaderV2({
|
|
301
|
+
num_values: data.count,
|
|
302
|
+
num_nulls: data.count - data.values.length,
|
|
303
|
+
num_rows: rowCount,
|
|
304
|
+
encoding: _parquetThrift.Encoding[column.encoding],
|
|
305
|
+
definition_levels_byte_length: dLevelsBuf.length,
|
|
306
|
+
repetition_levels_byte_length: rLevelsBuf.length,
|
|
307
|
+
is_compressed: column.compression !== 'UNCOMPRESSED'
|
|
308
|
+
}),
|
|
309
|
+
uncompressed_page_size: rLevelsBuf.length + dLevelsBuf.length + valuesBuf.length,
|
|
310
|
+
compressed_page_size: rLevelsBuf.length + dLevelsBuf.length + compressedBuf.length
|
|
311
|
+
});
|
|
312
|
+
const headerBuf = (0, _readUtils.serializeThrift)(header);
|
|
313
|
+
const page = Buffer.concat([headerBuf, rLevelsBuf, dLevelsBuf, compressedBuf]);
|
|
314
|
+
return {
|
|
315
|
+
header,
|
|
316
|
+
headerSize: headerBuf.length,
|
|
317
|
+
page
|
|
318
|
+
};
|
|
718
319
|
}
|
|
719
320
|
|
|
720
|
-
function
|
|
721
|
-
|
|
321
|
+
async function encodeColumnChunk(column, buffer, offset, opts) {
|
|
322
|
+
const data = buffer.columnData[column.path.join()];
|
|
323
|
+
const baseOffset = (opts.baseOffset || 0) + offset;
|
|
324
|
+
let pageBuf;
|
|
325
|
+
let total_uncompressed_size = 0;
|
|
326
|
+
let total_compressed_size = 0;
|
|
327
|
+
{
|
|
328
|
+
const result = opts.useDataPageV2 ? await encodeDataPageV2(column, data, buffer.rowCount) : await encodeDataPage(column, data);
|
|
329
|
+
pageBuf = result.page;
|
|
330
|
+
total_uncompressed_size += result.header.uncompressed_page_size + result.headerSize;
|
|
331
|
+
total_compressed_size += result.header.compressed_page_size + result.headerSize;
|
|
332
|
+
}
|
|
333
|
+
const metadata = new _parquetThrift.ColumnMetaData({
|
|
334
|
+
path_in_schema: column.path,
|
|
335
|
+
num_values: data.count,
|
|
336
|
+
data_page_offset: baseOffset,
|
|
337
|
+
encodings: [],
|
|
338
|
+
total_uncompressed_size,
|
|
339
|
+
total_compressed_size,
|
|
340
|
+
type: _parquetThrift.Type[column.primitiveType],
|
|
341
|
+
codec: _parquetThrift.CompressionCodec[column.compression]
|
|
342
|
+
});
|
|
343
|
+
metadata.encodings.push(_parquetThrift.Encoding[PARQUET_RDLVL_ENCODING]);
|
|
344
|
+
metadata.encodings.push(_parquetThrift.Encoding[column.encoding]);
|
|
345
|
+
const metadataOffset = baseOffset + pageBuf.length;
|
|
346
|
+
const body = Buffer.concat([pageBuf, (0, _readUtils.serializeThrift)(metadata)]);
|
|
347
|
+
return {
|
|
348
|
+
body,
|
|
349
|
+
metadata,
|
|
350
|
+
metadataOffset
|
|
351
|
+
};
|
|
722
352
|
}
|
|
723
353
|
|
|
724
|
-
function
|
|
725
|
-
|
|
726
|
-
|
|
727
|
-
|
|
728
|
-
|
|
729
|
-
|
|
730
|
-
|
|
731
|
-
|
|
732
|
-
|
|
733
|
-
|
|
734
|
-
|
|
735
|
-
|
|
736
|
-
|
|
737
|
-
|
|
738
|
-
|
|
739
|
-
|
|
740
|
-
|
|
741
|
-
|
|
742
|
-
|
|
743
|
-
|
|
744
|
-
|
|
745
|
-
|
|
746
|
-
|
|
747
|
-
|
|
748
|
-
|
|
749
|
-
|
|
750
|
-
|
|
751
|
-
if (!field.isNested) {
|
|
752
|
-
_context14.next = 9;
|
|
753
|
-
break;
|
|
754
|
-
}
|
|
755
|
-
|
|
756
|
-
return _context14.abrupt("continue", 16);
|
|
757
|
-
|
|
758
|
-
case 9:
|
|
759
|
-
_context14.next = 11;
|
|
760
|
-
return encodeColumnChunk(field, data, body.length, opts);
|
|
761
|
-
|
|
762
|
-
case 11:
|
|
763
|
-
cchunkData = _context14.sent;
|
|
764
|
-
cchunk = new _parquetThrift.ColumnChunk({
|
|
765
|
-
file_offset: cchunkData.metadataOffset,
|
|
766
|
-
meta_data: cchunkData.metadata
|
|
767
|
-
});
|
|
768
|
-
metadata.columns.push(cchunk);
|
|
769
|
-
metadata.total_byte_size = new _nodeInt.default(Number(metadata.total_byte_size) + cchunkData.body.length);
|
|
770
|
-
body = Buffer.concat([body, cchunkData.body]);
|
|
771
|
-
|
|
772
|
-
case 16:
|
|
773
|
-
_context14.next = 5;
|
|
774
|
-
break;
|
|
775
|
-
|
|
776
|
-
case 18:
|
|
777
|
-
_context14.next = 23;
|
|
778
|
-
break;
|
|
779
|
-
|
|
780
|
-
case 20:
|
|
781
|
-
_context14.prev = 20;
|
|
782
|
-
_context14.t0 = _context14["catch"](3);
|
|
783
|
-
|
|
784
|
-
_iterator2.e(_context14.t0);
|
|
785
|
-
|
|
786
|
-
case 23:
|
|
787
|
-
_context14.prev = 23;
|
|
788
|
-
|
|
789
|
-
_iterator2.f();
|
|
790
|
-
|
|
791
|
-
return _context14.finish(23);
|
|
792
|
-
|
|
793
|
-
case 26:
|
|
794
|
-
return _context14.abrupt("return", {
|
|
795
|
-
body: body,
|
|
796
|
-
metadata: metadata
|
|
797
|
-
});
|
|
798
|
-
|
|
799
|
-
case 27:
|
|
800
|
-
case "end":
|
|
801
|
-
return _context14.stop();
|
|
802
|
-
}
|
|
803
|
-
}
|
|
804
|
-
}, _callee14, null, [[3, 20, 23, 26]]);
|
|
805
|
-
}));
|
|
806
|
-
return _encodeRowGroup.apply(this, arguments);
|
|
354
|
+
async function encodeRowGroup(schema, data, opts) {
|
|
355
|
+
const metadata = new _parquetThrift.RowGroup({
|
|
356
|
+
num_rows: data.rowCount,
|
|
357
|
+
columns: [],
|
|
358
|
+
total_byte_size: 0
|
|
359
|
+
});
|
|
360
|
+
let body = Buffer.alloc(0);
|
|
361
|
+
|
|
362
|
+
for (const field of schema.fieldList) {
|
|
363
|
+
if (field.isNested) {
|
|
364
|
+
continue;
|
|
365
|
+
}
|
|
366
|
+
|
|
367
|
+
const cchunkData = await encodeColumnChunk(field, data, body.length, opts);
|
|
368
|
+
const cchunk = new _parquetThrift.ColumnChunk({
|
|
369
|
+
file_offset: cchunkData.metadataOffset,
|
|
370
|
+
meta_data: cchunkData.metadata
|
|
371
|
+
});
|
|
372
|
+
metadata.columns.push(cchunk);
|
|
373
|
+
metadata.total_byte_size = new _nodeInt.default(Number(metadata.total_byte_size) + cchunkData.body.length);
|
|
374
|
+
body = Buffer.concat([body, cchunkData.body]);
|
|
375
|
+
}
|
|
376
|
+
|
|
377
|
+
return {
|
|
378
|
+
body,
|
|
379
|
+
metadata
|
|
380
|
+
};
|
|
807
381
|
}
|
|
808
382
|
|
|
809
383
|
function encodeFooter(schema, rowCount, rowGroups, userMetadata) {
|
|
810
|
-
|
|
384
|
+
const metadata = new _parquetThrift.FileMetaData({
|
|
811
385
|
version: PARQUET_VERSION,
|
|
812
386
|
created_by: 'parquets',
|
|
813
387
|
num_rows: rowCount,
|
|
@@ -816,57 +390,47 @@ function encodeFooter(schema, rowCount, rowGroups, userMetadata) {
|
|
|
816
390
|
key_value_metadata: []
|
|
817
391
|
});
|
|
818
392
|
|
|
819
|
-
for (
|
|
393
|
+
for (const key in userMetadata) {
|
|
820
394
|
var _metadata$key_value_m, _metadata$key_value_m2;
|
|
821
395
|
|
|
822
|
-
|
|
823
|
-
key
|
|
396
|
+
const kv = new _parquetThrift.KeyValue({
|
|
397
|
+
key,
|
|
824
398
|
value: userMetadata[key]
|
|
825
399
|
});
|
|
826
400
|
(_metadata$key_value_m = metadata.key_value_metadata) === null || _metadata$key_value_m === void 0 ? void 0 : (_metadata$key_value_m2 = _metadata$key_value_m.push) === null || _metadata$key_value_m2 === void 0 ? void 0 : _metadata$key_value_m2.call(_metadata$key_value_m, kv);
|
|
827
401
|
}
|
|
828
402
|
|
|
829
403
|
{
|
|
830
|
-
|
|
404
|
+
const schemaRoot = new _parquetThrift.SchemaElement({
|
|
831
405
|
name: 'root',
|
|
832
406
|
num_children: Object.keys(schema.fields).length
|
|
833
407
|
});
|
|
834
408
|
metadata.schema.push(schemaRoot);
|
|
835
409
|
}
|
|
836
410
|
|
|
837
|
-
|
|
838
|
-
|
|
839
|
-
|
|
840
|
-
|
|
841
|
-
|
|
842
|
-
|
|
843
|
-
|
|
844
|
-
|
|
845
|
-
|
|
846
|
-
|
|
847
|
-
|
|
848
|
-
|
|
849
|
-
if (field.isNested) {
|
|
850
|
-
schemaElem.num_children = field.fieldCount;
|
|
851
|
-
} else {
|
|
852
|
-
schemaElem.type = _parquetThrift.Type[field.primitiveType];
|
|
853
|
-
}
|
|
854
|
-
|
|
855
|
-
if (field.originalType) {
|
|
856
|
-
schemaElem.converted_type = _parquetThrift.ConvertedType[field.originalType];
|
|
857
|
-
}
|
|
858
|
-
|
|
859
|
-
schemaElem.type_length = field.typeLength;
|
|
860
|
-
metadata.schema.push(schemaElem);
|
|
411
|
+
for (const field of schema.fieldList) {
|
|
412
|
+
const relt = _parquetThrift.FieldRepetitionType[field.repetitionType];
|
|
413
|
+
const schemaElem = new _parquetThrift.SchemaElement({
|
|
414
|
+
name: field.name,
|
|
415
|
+
repetition_type: relt
|
|
416
|
+
});
|
|
417
|
+
|
|
418
|
+
if (field.isNested) {
|
|
419
|
+
schemaElem.num_children = field.fieldCount;
|
|
420
|
+
} else {
|
|
421
|
+
schemaElem.type = _parquetThrift.Type[field.primitiveType];
|
|
861
422
|
}
|
|
862
|
-
|
|
863
|
-
|
|
864
|
-
|
|
865
|
-
|
|
423
|
+
|
|
424
|
+
if (field.originalType) {
|
|
425
|
+
schemaElem.converted_type = _parquetThrift.ConvertedType[field.originalType];
|
|
426
|
+
}
|
|
427
|
+
|
|
428
|
+
schemaElem.type_length = field.typeLength;
|
|
429
|
+
metadata.schema.push(schemaElem);
|
|
866
430
|
}
|
|
867
431
|
|
|
868
|
-
|
|
869
|
-
|
|
432
|
+
const metadataEncoded = (0, _readUtils.serializeThrift)(metadata);
|
|
433
|
+
const footerEncoded = Buffer.alloc(metadataEncoded.length + 8);
|
|
870
434
|
metadataEncoded.copy(footerEncoded);
|
|
871
435
|
footerEncoded.writeUInt32LE(metadataEncoded.length, metadataEncoded.length);
|
|
872
436
|
footerEncoded.write(PARQUET_MAGIC, metadataEncoded.length + 4);
|