Skip to content

CHANGELOG.md

Latest commit

 

History

History
263 lines (185 loc) · 6.71 KB

CHANGELOG.md

File metadata and controls

263 lines (185 loc) · 6.71 KB
Raw

Changelog

[1.30.0]

  • Add opt-in physical row index symbol to object rows (#176)

[1.29.2]

  • Add default exports to package.json for compatibility with CJS resolvers (#174)

[1.29.1]

  • Defer parquetScan column planning until a column is read

[1.29.0]

  • Add parquetScan for lazy physical column scans (#172)

[1.28.2]

  • Never match null values against range filters

[1.28.1]

  • Fix $nin filters when comparing Date values

[1.28.0]

  • Support $in filters on array values (#171)

[1.27.1]

  • Coalesce page index fetches into fewer byte range requests

[1.27.0]

  • Add usePageIndex option for page-level filter pushdown (#169)
  • Fix filtering on null and NaN row-group matches
  • Fix nested child column alignment by absolute row offsets
  • Compare date statistics by timestamp
  • Fix metadata decode of unsigned and decimal bounds
  • Fix node file slices (should be end-exclusive)

[1.26.2]

  • Custom jsonFromBytes parser option

[1.26.1]

  • Fix UUID statistics decode (#166)

[1.26.0]

  • Bloom filter support (#165)
  • Fix unhandled rejections (#164)

[1.25.8]

  • Fix shredded sparse timestamp

[1.25.7]

  • Fix shredded variant type mismatch

[1.25.6]

  • Fix struct assembly with skips

[1.25.5]

  • Fix polars missing dictionary_page_offset (#160)

[1.25.4]

  • Convert UUID logical type to string using custom parser

[1.25.3]

  • Fix include dictionary when reading with offset index (#158 thanks @tomquist)
  • Don't fetch offset index when reading entire rowgroup

[1.25.2]

  • Fix type exports for submodules

[1.25.1]

  • Fix nullable UINT types

[1.25.0]

  • Support dot-notation for filtering on nested struct fields (#153)

[1.24.1]

  • Fix handling of null definition levels (#150)

[1.24.0]

  • Variant support (#149)

[1.23.3]

  • Fix readColumn truncation in struct columns (#148)

[1.23.2]

  • Add option useOffsetIndex to control use of offset index for page filtering (#146)

[1.23.1]

  • Add option filterStrict to control strictness of filter equality (#145)
  • Apply filter not just best-effort
  • Combine runs of smaller column chunks when prefetching

[1.23.0]

  • Replace columnName: string with pathInSchema: string[] in onPage callback (#144)

[1.22.1]

  • Fix BYTE_STREAM_SPLIT with data page v2 and compression

[1.22.0]

  • Rename constants to plural, and remove LogicalTypeType

[1.21.1]

  • Add bigint to ParquetQueryValue type

[1.21.0]

  • Adds optional filter parameter for pushdown row group filtering (#141)

[1.20.3]

  • Fix extra fetch on the boundary of row groups

[1.20.2]

  • Support S3 presigned URLs in asyncBufferFromUrl (#137 thanks @EpsilonPrime)

[1.20.1]

  • Update geospatial and variant metadata

[1.20.0]

  • Mark geospatial columns in GeoParquet with geometry or geography data type (#133)
  • Add geoparquet option to opt out marking GeoParquet columns (#133)

[1.19.0]

  • Parse geometry and geography data types to geojson geometry objects (#131)

[1.18.1]

  • Fix geospatial metadata parsing
  • Custom stringFromBytes parser option (#129)

[1.18.0]

  • Refine onComplete types for rowFormat: 'array' and rowFormat: 'object' (#120)
  • Only object format for parquetReadObjects and parquetQuery (#120)
  • Parquet __index__ column overrides parquetQuery index annotation (#120)

[1.17.8]

  • Export readColumnIndex and readOffsetIndex (#122)

[1.17.7]

  • Fix early page termination for non-flat columns (#119)
  • Fix a bug in parquetQuery, when rowFormat is 'array' (#118 thanks @severo)

[1.17.6]

  • Revert "Fix onComplete return type (#104)" (#117)

[1.17.5]

  • Fix high-precision decimal parsing (#116)

[1.17.4]

  • Fix onComplete return type (#104 thanks @supermar1010)

[1.17.3]

  • Fix circular import (#111)

[1.17.2]

  • Export ParquetQueryFilter type (#105)
  • Remove a circular dependency (#108)

[1.17.1]

  • Fix zero row file (#98 thanks @kroche98)

[1.17.0]

  • Require explicit $eq operator for parquetQuery filters.

[1.16.2]

  • Fix readRleBitPackedHybrid when length is zero

[1.16.1]

  • Fix duckdb empty block

[1.16.0]

  • New parsers option for custom date parsing.
  • Breaking change: parquetMetadataAsync moved initialFetchSize into an options object.

[1.15.0]

  • Change packaging to have node-specific exports for asyncBufferFromFile (#80).

[1.14.0]

  • Refactor to use AsyncRowGroup and AsyncColumn abstractions for better performance and flexibility (#83).

[1.13.6]

  • Fix page continuation (#81 thanks @jpivarski).

[1.13.5]

  • Fast parquetQuery filter (#78)

[1.13.4]

  • parquetSchema more generic argument type.

[1.13.3]

  • Convert logical type 'STRING'.
  • Side-effect-free in package.json.

[1.13.2]

  • Fix duckdb delta encoding (#77 thanks @mike-iqmo).

[1.13.1]

  • Throw exception for unsupported file_path.

[1.13.0]

  • Query planner: pre-fetch byte ranges in parallel (#75).

[1.12.1]

  • Fix conversion of unsigned types.
  • Allow passing a custom fetch function to utilities (#73).

[1.12.0]

  • Add onPage callback to parquetRead.

[1.11.1]

  • Fix handling of dictionary pages from parquet.net.

[1.11.0]

  • Fix continued data pages
  • Skip decoding unnecessary pages.

[1.10.4]

  • Add type definitions for thrift.

[1.10.3]

  • Internal refactor split out readPage function.

[1.10.2]

  • Export additional internal constants.

[1.10.1]

  • Fix parsing of crypto_metadata in thrift.

[1.10.0]

  • Map src files to TypeScript types via package exports (#70).
  • Use defaultInitialFetchSize for both metadata and cachedAsyncBuffer.

[1.9.x]

  • Add minSize parameter to cachedAsyncBuffer for finer control.
  • Return typed arrays in onChunk callbacks.
  • Change readColumn to return an array of DecodedArray (#67).

[1.8.x]

  • Support endpoints without range requests in asyncBufferFromUrl (#57 thanks @swlynch99).
  • Enhance error messages for common parsing issues.
  • Mongo-style filter option in parquetQuery (#56 thanks @park-brian).

[1.7.0]

  • Enable readColumn to read all rows (#53 thanks @park-brian).
  • Validate url in asyncBufferFromUrl.

[1.6.x]

  • Fix timestamp conversion in metadata parsing (#45 thanks @cbardasano).
  • Build TypeScript types before publishing to npm.

[1.5.0]

  • Export cachedAsyncBuffer utility.

[1.4.0]

  • Add parquetQuery with orderBy option.

[1.3.0]

  • Promisify parquetReadObjects function.
  • Add support for parsing column and offset indexes (#29).

[1.2.0]

  • Return columns in the requested order (#27 thanks @cstranstrum).
  • Add option to return each row as an object keyed by column names (#25 thanks @cstranstrum).

[1.1.0]

  • Export asyncBufferFromFile and asyncBufferFromUrl utilities.

[1.0.0]

  • Initial stable release.