@loaders.gl/parquet 5.0.0-alpha.2 → 5.0.0-alpha.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/LICENSE +1 -1
- package/dist/bundled.d.ts +1 -1
- package/dist/bundled.d.ts.map +1 -1
- package/dist/bundled.js.map +1 -1
- package/dist/delta-format.d.ts +9 -0
- package/dist/delta-format.d.ts.map +1 -0
- package/dist/delta-format.js +12 -0
- package/dist/delta-format.js.map +1 -0
- package/dist/delta-source-loader-types.d.ts +27 -0
- package/dist/delta-source-loader-types.d.ts.map +1 -0
- package/dist/delta-source-loader-types.js +28 -0
- package/dist/delta-source-loader-types.js.map +1 -0
- package/dist/delta-source.d.ts +48 -0
- package/dist/delta-source.d.ts.map +1 -0
- package/dist/delta-source.js +222 -0
- package/dist/delta-source.js.map +1 -0
- package/dist/delta-types.d.ts +38 -0
- package/dist/delta-types.d.ts.map +1 -0
- package/dist/delta-types.js +5 -0
- package/dist/delta-types.js.map +1 -0
- package/dist/geoparquet-loader.js +1 -1
- package/dist/iceberg-rest-catalog.d.ts +49 -0
- package/dist/iceberg-rest-catalog.d.ts.map +1 -0
- package/dist/iceberg-rest-catalog.js +78 -0
- package/dist/iceberg-rest-catalog.js.map +1 -0
- package/dist/iceberg-table-source.d.ts +63 -0
- package/dist/iceberg-table-source.d.ts.map +1 -0
- package/dist/iceberg-table-source.js +797 -0
- package/dist/iceberg-table-source.js.map +1 -0
- package/dist/iceberg-types.d.ts +150 -0
- package/dist/iceberg-types.d.ts.map +1 -0
- package/dist/iceberg-types.js +5 -0
- package/dist/iceberg-types.js.map +1 -0
- package/dist/index.cjs +15005 -3143
- package/dist/index.cjs.map +3 -3
- package/dist/index.d.ts +7 -3
- package/dist/index.d.ts.map +1 -1
- package/dist/index.js +5 -1
- package/dist/index.js.map +1 -1
- package/dist/lib/arrow/convert-schema-from-parquet.d.ts +3 -1
- package/dist/lib/arrow/convert-schema-from-parquet.d.ts.map +1 -1
- package/dist/lib/arrow/convert-schema-from-parquet.js +94 -19
- package/dist/lib/arrow/convert-schema-from-parquet.js.map +1 -1
- package/dist/lib/arrow/create-nested-arrow-vector.js +89 -0
- package/dist/lib/arrow/create-nested-arrow-vector.js.map +1 -1
- package/dist/lib/constants.js +1 -1
- package/dist/lib/encoders/encode-table-to-parquet-js.d.ts.map +1 -1
- package/dist/lib/encoders/encode-table-to-parquet-js.js +202 -7
- package/dist/lib/encoders/encode-table-to-parquet-js.js.map +1 -1
- package/dist/lib/geo/geoparquet-covering.d.ts +19 -0
- package/dist/lib/geo/geoparquet-covering.d.ts.map +1 -0
- package/dist/lib/geo/geoparquet-covering.js +137 -0
- package/dist/lib/geo/geoparquet-covering.js.map +1 -0
- package/dist/lib/geo/geospatial-metadata.d.ts.map +1 -1
- package/dist/lib/geo/geospatial-metadata.js +39 -13
- package/dist/lib/geo/geospatial-metadata.js.map +1 -1
- package/dist/lib/parquet-bloom-filter-planner.d.ts +16 -0
- package/dist/lib/parquet-bloom-filter-planner.d.ts.map +1 -0
- package/dist/lib/parquet-bloom-filter-planner.js +57 -0
- package/dist/lib/parquet-bloom-filter-planner.js.map +1 -0
- package/dist/lib/parquet-bloom-filter.d.ts +30 -0
- package/dist/lib/parquet-bloom-filter.d.ts.map +1 -0
- package/dist/lib/parquet-bloom-filter.js +289 -0
- package/dist/lib/parquet-bloom-filter.js.map +1 -0
- package/dist/lib/parquet-encryption.d.ts +80 -0
- package/dist/lib/parquet-encryption.d.ts.map +1 -0
- package/dist/lib/parquet-encryption.js +190 -0
- package/dist/lib/parquet-encryption.js.map +1 -0
- package/dist/lib/parquet-page-index.d.ts +34 -5
- package/dist/lib/parquet-page-index.d.ts.map +1 -1
- package/dist/lib/parquet-page-index.js +114 -28
- package/dist/lib/parquet-page-index.js.map +1 -1
- package/dist/lib/parquet-predicate.d.ts +5 -1
- package/dist/lib/parquet-predicate.d.ts.map +1 -1
- package/dist/lib/parquet-predicate.js +42 -14
- package/dist/lib/parquet-predicate.js.map +1 -1
- package/dist/lib/parquet-source-worker-decoder.d.ts +4 -1
- package/dist/lib/parquet-source-worker-decoder.d.ts.map +1 -1
- package/dist/lib/parquet-source-worker-decoder.js +83 -17
- package/dist/lib/parquet-source-worker-decoder.js.map +1 -1
- package/dist/lib/parquet-source-worker-types.d.ts +24 -0
- package/dist/lib/parquet-source-worker-types.d.ts.map +1 -1
- package/dist/lib/parsers/get-parquet-schema.d.ts.map +1 -1
- package/dist/lib/parsers/get-parquet-schema.js +3 -1
- package/dist/lib/parsers/get-parquet-schema.js.map +1 -1
- package/dist/lib/parsers/parse-parquet-to-arrow-js.d.ts.map +1 -1
- package/dist/lib/parsers/parse-parquet-to-arrow-js.js +327 -18
- package/dist/lib/parsers/parse-parquet-to-arrow-js.js.map +1 -1
- package/dist/lib/parsers/parse-parquet-to-arrow.d.ts.map +1 -1
- package/dist/lib/parsers/parse-parquet-to-arrow.js +5 -5
- package/dist/lib/parsers/parse-parquet-to-arrow.js.map +1 -1
- package/dist/lib/parsers/parse-parquet-to-columns.d.ts.map +1 -1
- package/dist/lib/parsers/parse-parquet-to-columns.js +3 -1
- package/dist/lib/parsers/parse-parquet-to-columns.js.map +1 -1
- package/dist/lib/parsers/parse-parquet-to-json.d.ts.map +1 -1
- package/dist/lib/parsers/parse-parquet-to-json.js +18 -2
- package/dist/lib/parsers/parse-parquet-to-json.js.map +1 -1
- package/dist/lib/slice-parquet-batch.d.ts +9 -0
- package/dist/lib/slice-parquet-batch.d.ts.map +1 -0
- package/dist/lib/slice-parquet-batch.js +28 -0
- package/dist/lib/slice-parquet-batch.js.map +1 -0
- package/dist/lib/sources/parquet-range-file.d.ts +1 -1
- package/dist/lib/sources/parquet-range-file.d.ts.map +1 -1
- package/dist/lib/sources/parquet-range-file.js +28 -22
- package/dist/lib/sources/parquet-range-file.js.map +1 -1
- package/dist/lib/utils/normalize-parquet-options.d.ts.map +1 -1
- package/dist/lib/utils/normalize-parquet-options.js +15 -1
- package/dist/lib/utils/normalize-parquet-options.js.map +1 -1
- package/dist/parquet-dataset-source.cjs +15766 -4896
- package/dist/parquet-dataset-source.cjs.map +3 -3
- package/dist/parquet-dataset-source.d.ts +31 -4
- package/dist/parquet-dataset-source.d.ts.map +1 -1
- package/dist/parquet-dataset-source.js +296 -74
- package/dist/parquet-dataset-source.js.map +1 -1
- package/dist/parquet-js-loader-types.d.ts +2 -0
- package/dist/parquet-js-loader-types.d.ts.map +1 -1
- package/dist/parquet-js-loader-types.js +2 -1
- package/dist/parquet-js-loader-types.js.map +1 -1
- package/dist/parquet-js-loader.d.ts +1 -0
- package/dist/parquet-js-loader.d.ts.map +1 -1
- package/dist/parquet-js-loader.js +2 -1
- package/dist/parquet-js-loader.js.map +1 -1
- package/dist/parquet-js-writer.d.ts +23 -2
- package/dist/parquet-js-writer.d.ts.map +1 -1
- package/dist/parquet-js-writer.js +1 -1
- package/dist/parquet-js-writer.js.map +1 -1
- package/dist/parquet-loader-base.js +1 -1
- package/dist/parquet-loader-options.d.ts +11 -0
- package/dist/parquet-loader-options.d.ts.map +1 -1
- package/dist/parquet-loader-options.js.map +1 -1
- package/dist/parquet-source-capabilities.d.ts +8 -1
- package/dist/parquet-source-capabilities.d.ts.map +1 -1
- package/dist/parquet-source-capabilities.js +10 -0
- package/dist/parquet-source-capabilities.js.map +1 -1
- package/dist/parquet-source-loader-types.d.ts +10 -0
- package/dist/parquet-source-loader-types.d.ts.map +1 -1
- package/dist/parquet-source-loader-types.js +11 -1
- package/dist/parquet-source-loader-types.js.map +1 -1
- package/dist/parquet-source-loader.cjs +14911 -4316
- package/dist/parquet-source-loader.cjs.map +3 -3
- package/dist/parquet-source-loader.d.ts +42 -4
- package/dist/parquet-source-loader.d.ts.map +1 -1
- package/dist/parquet-source-loader.js +585 -38
- package/dist/parquet-source-loader.js.map +1 -1
- package/dist/parquet-source-types.d.ts +180 -30
- package/dist/parquet-source-types.d.ts.map +1 -1
- package/dist/parquet-source-worker.js +7055 -6989
- package/dist/parquet-source-worker.js.map +4 -4
- package/dist/parquet-worker-node.cjs +8505 -1041
- package/dist/parquet-worker-node.cjs.map +4 -4
- package/dist/parquet-worker-url.js +5 -1
- package/dist/parquet-worker-url.js.map +1 -1
- package/dist/parquet-worker.js +289 -218
- package/dist/parquet-worker.js.map +4 -4
- package/dist/parquetjs/codecs/declare.d.ts +15 -0
- package/dist/parquetjs/codecs/declare.d.ts.map +1 -1
- package/dist/parquetjs/codecs/declare.js +10 -0
- package/dist/parquetjs/codecs/declare.js.map +1 -1
- package/dist/parquetjs/codecs/delta.d.ts.map +1 -1
- package/dist/parquetjs/codecs/delta.js +356 -7
- package/dist/parquetjs/codecs/delta.js.map +1 -1
- package/dist/parquetjs/codecs/index.d.ts.map +1 -1
- package/dist/parquetjs/codecs/index.js +4 -0
- package/dist/parquetjs/codecs/index.js.map +1 -1
- package/dist/parquetjs/codecs/plain.d.ts +1 -1
- package/dist/parquetjs/codecs/plain.d.ts.map +1 -1
- package/dist/parquetjs/codecs/plain.js +153 -12
- package/dist/parquetjs/codecs/plain.js.map +1 -1
- package/dist/parquetjs/codecs/rle.d.ts +4 -0
- package/dist/parquetjs/codecs/rle.d.ts.map +1 -1
- package/dist/parquetjs/codecs/rle.js +406 -3
- package/dist/parquetjs/codecs/rle.js.map +1 -1
- package/dist/parquetjs/compression.d.ts +9 -0
- package/dist/parquetjs/compression.d.ts.map +1 -1
- package/dist/parquetjs/compression.js +86 -67
- package/dist/parquetjs/compression.js.map +1 -1
- package/dist/parquetjs/encoder/dictionary-planner.d.ts.map +1 -1
- package/dist/parquetjs/encoder/dictionary-planner.js +5 -2
- package/dist/parquetjs/encoder/dictionary-planner.js.map +1 -1
- package/dist/parquetjs/encoder/parquet-encoder.d.ts +43 -1
- package/dist/parquetjs/encoder/parquet-encoder.d.ts.map +1 -1
- package/dist/parquetjs/encoder/parquet-encoder.js +691 -34
- package/dist/parquetjs/encoder/parquet-encoder.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/BoundingBox.d.ts +29 -0
- package/dist/parquetjs/parquet-thrift/BoundingBox.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/BoundingBox.js +85 -0
- package/dist/parquetjs/parquet-thrift/BoundingBox.js.map +1 -0
- package/dist/parquetjs/parquet-thrift/ColumnChunk.d.ts +5 -0
- package/dist/parquetjs/parquet-thrift/ColumnChunk.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/ColumnChunk.js +35 -0
- package/dist/parquetjs/parquet-thrift/ColumnChunk.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/ColumnCryptoMetaData.d.ts +35 -0
- package/dist/parquetjs/parquet-thrift/ColumnCryptoMetaData.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/ColumnCryptoMetaData.js +135 -0
- package/dist/parquetjs/parquet-thrift/ColumnCryptoMetaData.js.map +1 -0
- package/dist/parquetjs/parquet-thrift/ColumnMetaData.d.ts +10 -0
- package/dist/parquetjs/parquet-thrift/ColumnMetaData.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/ColumnMetaData.js +73 -0
- package/dist/parquetjs/parquet-thrift/ColumnMetaData.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/EncryptionAlgorithm.d.ts +22 -0
- package/dist/parquetjs/parquet-thrift/EncryptionAlgorithm.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/EncryptionAlgorithm.js +111 -0
- package/dist/parquetjs/parquet-thrift/EncryptionAlgorithm.js.map +1 -0
- package/dist/parquetjs/parquet-thrift/FileCryptoMetaData.d.ts +16 -0
- package/dist/parquetjs/parquet-thrift/FileCryptoMetaData.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/FileCryptoMetaData.js +57 -0
- package/dist/parquetjs/parquet-thrift/FileCryptoMetaData.js.map +1 -0
- package/dist/parquetjs/parquet-thrift/FileMetaData.d.ts +5 -0
- package/dist/parquetjs/parquet-thrift/FileMetaData.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/FileMetaData.js +35 -0
- package/dist/parquetjs/parquet-thrift/FileMetaData.js.map +1 -1
- package/dist/parquetjs/parquet-thrift/GeospatialStatistics.d.ts +18 -0
- package/dist/parquetjs/parquet-thrift/GeospatialStatistics.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/GeospatialStatistics.js +62 -0
- package/dist/parquetjs/parquet-thrift/GeospatialStatistics.js.map +1 -0
- package/dist/parquetjs/parquet-thrift/SizeStatistics.d.ts +26 -0
- package/dist/parquetjs/parquet-thrift/SizeStatistics.d.ts.map +1 -0
- package/dist/parquetjs/parquet-thrift/SizeStatistics.js +85 -0
- package/dist/parquetjs/parquet-thrift/SizeStatistics.js.map +1 -0
- package/dist/parquetjs/parquet-thrift/index.d.ts +6 -0
- package/dist/parquetjs/parquet-thrift/index.d.ts.map +1 -1
- package/dist/parquetjs/parquet-thrift/index.js +6 -0
- package/dist/parquetjs/parquet-thrift/index.js.map +1 -1
- package/dist/parquetjs/parser/decoders.d.ts +10 -1
- package/dist/parquetjs/parser/decoders.d.ts.map +1 -1
- package/dist/parquetjs/parser/decoders.js +318 -14
- package/dist/parquetjs/parser/decoders.js.map +1 -1
- package/dist/parquetjs/parser/parquet-reader.d.ts +78 -12
- package/dist/parquetjs/parser/parquet-reader.d.ts.map +1 -1
- package/dist/parquetjs/parser/parquet-reader.js +605 -66
- package/dist/parquetjs/parser/parquet-reader.js.map +1 -1
- package/dist/parquetjs/schema/declare.d.ts +22 -3
- package/dist/parquetjs/schema/declare.d.ts.map +1 -1
- package/dist/parquetjs/schema/declare.js.map +1 -1
- package/dist/parquetjs/schema/schema.d.ts.map +1 -1
- package/dist/parquetjs/schema/schema.js +10 -0
- package/dist/parquetjs/schema/schema.js.map +1 -1
- package/dist/parquetjs/schema/shred.d.ts.map +1 -1
- package/dist/parquetjs/schema/shred.js +219 -11
- package/dist/parquetjs/schema/shred.js.map +1 -1
- package/dist/parquetjs/schema/types.js +19 -21
- package/dist/parquetjs/schema/types.js.map +1 -1
- package/dist/parquetjs/schema/variant.d.ts +3 -0
- package/dist/parquetjs/schema/variant.d.ts.map +1 -0
- package/dist/parquetjs/schema/variant.js +248 -0
- package/dist/parquetjs/schema/variant.js.map +1 -0
- package/dist/parquetjs/utils/crc32.d.ts +3 -0
- package/dist/parquetjs/utils/crc32.d.ts.map +1 -0
- package/dist/parquetjs/utils/crc32.js +15 -0
- package/dist/parquetjs/utils/crc32.js.map +1 -0
- package/dist/parquetjs/utils/read-utils.d.ts +11 -1
- package/dist/parquetjs/utils/read-utils.d.ts.map +1 -1
- package/dist/parquetjs/utils/read-utils.js +21 -1
- package/dist/parquetjs/utils/read-utils.js.map +1 -1
- package/dist/unbundled.d.ts +2 -1
- package/dist/unbundled.d.ts.map +1 -1
- package/dist/unbundled.js +1 -0
- package/dist/unbundled.js.map +1 -1
- package/package.json +36 -11
- package/src/bundled.ts +3 -0
- package/src/delta-format.ts +12 -0
- package/src/delta-source-loader-types.ts +40 -0
- package/src/delta-source.ts +305 -0
- package/src/delta-types.ts +40 -0
- package/src/iceberg-rest-catalog.ts +122 -0
- package/src/iceberg-table-source.ts +1037 -0
- package/src/iceberg-types.ts +165 -0
- package/src/index.ts +50 -1
- package/src/lib/arrow/convert-schema-from-parquet.ts +113 -19
- package/src/lib/arrow/create-nested-arrow-vector.ts +116 -0
- package/src/lib/encoders/encode-table-to-parquet-js.ts +268 -7
- package/src/lib/geo/geoparquet-covering.ts +201 -0
- package/src/lib/geo/geospatial-metadata.ts +42 -15
- package/src/lib/parquet-bloom-filter-planner.ts +86 -0
- package/src/lib/parquet-bloom-filter.ts +345 -0
- package/src/lib/parquet-encryption.ts +344 -0
- package/src/lib/parquet-page-index.ts +180 -34
- package/src/lib/parquet-predicate.ts +52 -16
- package/src/lib/parquet-source-worker-decoder.ts +116 -20
- package/src/lib/parquet-source-worker-types.ts +25 -0
- package/src/lib/parsers/get-parquet-schema.ts +3 -1
- package/src/lib/parsers/parse-parquet-to-arrow-js.ts +397 -17
- package/src/lib/parsers/parse-parquet-to-arrow.ts +4 -5
- package/src/lib/parsers/parse-parquet-to-columns.ts +3 -1
- package/src/lib/parsers/parse-parquet-to-json.ts +18 -2
- package/src/lib/slice-parquet-batch.ts +32 -0
- package/src/lib/sources/parquet-range-file.ts +35 -29
- package/src/lib/utils/normalize-parquet-options.ts +17 -1
- package/src/parquet-dataset-source.ts +397 -85
- package/src/parquet-js-loader-types.ts +1 -0
- package/src/parquet-js-loader.ts +2 -1
- package/src/parquet-js-writer.ts +30 -0
- package/src/parquet-loader-options.ts +11 -0
- package/src/parquet-source-capabilities.ts +18 -1
- package/src/parquet-source-loader-types.ts +10 -0
- package/src/parquet-source-loader.ts +929 -52
- package/src/parquet-source-types.ts +197 -28
- package/src/parquet-worker-url.ts +6 -1
- package/src/parquetjs/codecs/declare.ts +27 -0
- package/src/parquetjs/codecs/delta.ts +466 -6
- package/src/parquetjs/codecs/index.ts +4 -0
- package/src/parquetjs/codecs/plain.ts +177 -11
- package/src/parquetjs/codecs/rle.ts +573 -3
- package/src/parquetjs/compression.ts +102 -76
- package/src/parquetjs/encoder/dictionary-planner.ts +7 -2
- package/src/parquetjs/encoder/parquet-encoder.ts +1016 -38
- package/src/parquetjs/parquet-thrift/BoundingBox.ts +108 -0
- package/src/parquetjs/parquet-thrift/ColumnChunk.ts +35 -0
- package/src/parquetjs/parquet-thrift/ColumnCryptoMetaData.ts +139 -0
- package/src/parquetjs/parquet-thrift/ColumnMetaData.ts +73 -0
- package/src/parquetjs/parquet-thrift/EncryptionAlgorithm.ts +127 -0
- package/src/parquetjs/parquet-thrift/FileCryptoMetaData.ts +60 -0
- package/src/parquetjs/parquet-thrift/FileMetaData.ts +35 -0
- package/src/parquetjs/parquet-thrift/GeospatialStatistics.ts +68 -0
- package/src/parquetjs/parquet-thrift/SizeStatistics.ts +98 -0
- package/src/parquetjs/parquet-thrift/index.ts +6 -0
- package/src/parquetjs/parser/decoders.ts +451 -26
- package/src/parquetjs/parser/parquet-reader.ts +935 -87
- package/src/parquetjs/schema/declare.ts +22 -2
- package/src/parquetjs/schema/schema.ts +10 -0
- package/src/parquetjs/schema/shred.ts +238 -11
- package/src/parquetjs/schema/types.ts +16 -25
- package/src/parquetjs/schema/variant.ts +313 -0
- package/src/parquetjs/utils/crc32.ts +15 -0
- package/src/parquetjs/utils/read-utils.ts +26 -1
- package/src/unbundled.ts +14 -0
|
@@ -3,7 +3,7 @@
|
|
|
3
3
|
// Copyright (c) vis.gl contributors
|
|
4
4
|
|
|
5
5
|
import type {ReadableFile, Stat} from '@loaders.gl/loader-utils';
|
|
6
|
-
import {RangeRequestScheduler} from '@loaders.gl/loader-utils';
|
|
6
|
+
import {RangeRequestCache, RangeRequestScheduler} from '@loaders.gl/loader-utils';
|
|
7
7
|
|
|
8
8
|
import type {ParquetObjectVersion, ParquetRangeRequestOptions} from '../../parquet-source-types';
|
|
9
9
|
|
|
@@ -47,6 +47,8 @@ type ContentRange = {
|
|
|
47
47
|
total: number;
|
|
48
48
|
};
|
|
49
49
|
|
|
50
|
+
const DEFAULT_RANGE_CACHE_BYTES = 65536;
|
|
51
|
+
|
|
50
52
|
/** Strict HTTP range-backed file with Parquet object-version validation. */
|
|
51
53
|
export class ParquetRangeFile implements ReadableFile {
|
|
52
54
|
/** URL used as the file handle. */
|
|
@@ -64,8 +66,8 @@ export class ParquetRangeFile implements ReadableFile {
|
|
|
64
66
|
private readonly scheduler: RangeRequestScheduler;
|
|
65
67
|
/** Aborts all active requests when the file is closed. */
|
|
66
68
|
private readonly closeController = new AbortController();
|
|
67
|
-
/**
|
|
68
|
-
private readonly cache
|
|
69
|
+
/** Bounded cache for repeated metadata and small range reads. */
|
|
70
|
+
private readonly cache: RangeRequestCache;
|
|
69
71
|
/** Remote object byte length discovered from `Content-Range`. */
|
|
70
72
|
private fileByteLength = 0;
|
|
71
73
|
/** Object validators captured by the opening range request. */
|
|
@@ -80,6 +82,14 @@ export class ParquetRangeFile implements ReadableFile {
|
|
|
80
82
|
this.fetch = options.fetch;
|
|
81
83
|
this.headers = options.headers;
|
|
82
84
|
this.onTelemetry = options.onTelemetry;
|
|
85
|
+
this.cache = new RangeRequestCache({
|
|
86
|
+
maxBytes: DEFAULT_RANGE_CACHE_BYTES,
|
|
87
|
+
onEvent: event => {
|
|
88
|
+
if (event.type === 'hit') {
|
|
89
|
+
this.onTelemetry?.({type: 'cache-hit', cacheHits: 1});
|
|
90
|
+
}
|
|
91
|
+
}
|
|
92
|
+
});
|
|
83
93
|
this.scheduler =
|
|
84
94
|
options.rangeRequests?.scheduler ||
|
|
85
95
|
new RangeRequestScheduler({
|
|
@@ -113,7 +123,7 @@ export class ParquetRangeFile implements ReadableFile {
|
|
|
113
123
|
const result = await this.fetchExactRange(0, 4, signal, false);
|
|
114
124
|
this.fileByteLength = result.contentRange.total;
|
|
115
125
|
this.version = getObjectVersion(result.response.headers);
|
|
116
|
-
this.cache.set(
|
|
126
|
+
this.cache.set(getVersionedSourceId(this.url, this.version), 0, result.arrayBuffer);
|
|
117
127
|
return this;
|
|
118
128
|
}
|
|
119
129
|
|
|
@@ -130,32 +140,33 @@ export class ParquetRangeFile implements ReadableFile {
|
|
|
130
140
|
return new ArrayBuffer(0);
|
|
131
141
|
}
|
|
132
142
|
|
|
133
|
-
const cached = this.cache.get(getRangeKey(offset, length));
|
|
134
|
-
if (cached) {
|
|
135
|
-
this.onTelemetry?.({type: 'cache-hit', cacheHits: 1});
|
|
136
|
-
return cached.slice(0);
|
|
137
|
-
}
|
|
138
|
-
|
|
139
143
|
const abortContext = createCombinedAbortSignal(signal, this.closeController.signal);
|
|
140
144
|
try {
|
|
141
|
-
return await this.
|
|
145
|
+
return await this.cache.read({
|
|
142
146
|
sourceId: getVersionedSourceId(this.url, this.version),
|
|
143
147
|
offset,
|
|
144
148
|
length,
|
|
145
149
|
signal: abortContext.signal,
|
|
146
|
-
fetchRange: async (
|
|
147
|
-
|
|
148
|
-
|
|
149
|
-
|
|
150
|
-
|
|
151
|
-
|
|
152
|
-
|
|
153
|
-
|
|
154
|
-
|
|
155
|
-
|
|
156
|
-
|
|
157
|
-
|
|
158
|
-
|
|
150
|
+
fetchRange: async (rangeOffset, rangeLength, cacheSignal) =>
|
|
151
|
+
await this.scheduler.scheduleRequest({
|
|
152
|
+
sourceId: getVersionedSourceId(this.url, this.version),
|
|
153
|
+
offset: rangeOffset,
|
|
154
|
+
length: rangeLength,
|
|
155
|
+
signal: cacheSignal,
|
|
156
|
+
fetchRange: async (transportOffset, transportLength, transportSignal) => {
|
|
157
|
+
const result = await this.fetchExactRange(
|
|
158
|
+
transportOffset,
|
|
159
|
+
transportLength,
|
|
160
|
+
transportSignal,
|
|
161
|
+
true
|
|
162
|
+
);
|
|
163
|
+
return {
|
|
164
|
+
arrayBuffer: result.arrayBuffer,
|
|
165
|
+
status: result.response.status,
|
|
166
|
+
transportBytes: result.arrayBuffer.byteLength
|
|
167
|
+
};
|
|
168
|
+
}
|
|
169
|
+
})
|
|
159
170
|
});
|
|
160
171
|
} finally {
|
|
161
172
|
abortContext.dispose();
|
|
@@ -323,11 +334,6 @@ function assertObjectVersion(expected: ParquetObjectVersion, actual: ParquetObje
|
|
|
323
334
|
}
|
|
324
335
|
}
|
|
325
336
|
|
|
326
|
-
/** Creates a cache key for one exact range. */
|
|
327
|
-
function getRangeKey(offset: number, length: number): string {
|
|
328
|
-
return `${offset}:${length}`;
|
|
329
|
-
}
|
|
330
|
-
|
|
331
337
|
/** Creates a scheduler identity that isolates object versions. */
|
|
332
338
|
function getVersionedSourceId(url: string, version: ParquetObjectVersion): string {
|
|
333
339
|
return `${url}#${version.etag || version.lastModified || 'unversioned'}`;
|
|
@@ -7,8 +7,24 @@ export function normalizeParquetOptions<OptionsT extends {parquet?: unknown}>(
|
|
|
7
7
|
options: OptionsT | undefined,
|
|
8
8
|
parquetDefaults: Record<string, unknown>
|
|
9
9
|
): OptionsT {
|
|
10
|
+
const scanOptions = (
|
|
11
|
+
options as {_scan?: {columns?: readonly string[]; limit?: number}} | undefined
|
|
12
|
+
)?._scan;
|
|
13
|
+
if (scanOptions) {
|
|
14
|
+
for (const key of Object.keys(scanOptions)) {
|
|
15
|
+
if (key !== 'columns' && key !== 'limit') {
|
|
16
|
+
throw new Error(`Parquet _scan option "${key}" is not supported`);
|
|
17
|
+
}
|
|
18
|
+
}
|
|
19
|
+
}
|
|
20
|
+
const parquetOptions = (options?.parquet as Record<string, unknown> | undefined) || {};
|
|
10
21
|
return {
|
|
11
22
|
...options,
|
|
12
|
-
parquet: {
|
|
23
|
+
parquet: {
|
|
24
|
+
...parquetDefaults,
|
|
25
|
+
...parquetOptions,
|
|
26
|
+
...(scanOptions?.columns !== undefined ? {columns: [...scanOptions.columns]} : {}),
|
|
27
|
+
...(scanOptions?.limit !== undefined ? {limit: scanOptions.limit} : {})
|
|
28
|
+
}
|
|
13
29
|
} as unknown as OptionsT;
|
|
14
30
|
}
|