@loaders.gl/parquet 3.1.0-beta.5 → 3.1.2

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.
Files changed (139) hide show
  1. package/dist/dist.min.js +6 -6
  2. package/dist/dist.min.js.map +2 -2
  3. package/dist/es5/bundle.js +1 -1
  4. package/dist/es5/bundle.js.map +1 -1
  5. package/dist/es5/constants.js +5 -5
  6. package/dist/es5/constants.js.map +1 -1
  7. package/dist/es5/index.js +19 -10
  8. package/dist/es5/index.js.map +1 -1
  9. package/dist/es5/lib/convert-schema.js +13 -13
  10. package/dist/es5/lib/convert-schema.js.map +1 -1
  11. package/dist/es5/lib/parse-parquet.js +154 -19
  12. package/dist/es5/lib/parse-parquet.js.map +1 -1
  13. package/dist/es5/lib/read-array-buffer.js +43 -6
  14. package/dist/es5/lib/read-array-buffer.js.map +1 -1
  15. package/dist/es5/parquet-loader.js +4 -4
  16. package/dist/es5/parquet-loader.js.map +1 -1
  17. package/dist/es5/parquet-writer.js +4 -4
  18. package/dist/es5/parquet-writer.js.map +1 -1
  19. package/dist/es5/parquetjs/codecs/dictionary.js +10 -2
  20. package/dist/es5/parquetjs/codecs/dictionary.js.map +1 -1
  21. package/dist/es5/parquetjs/codecs/index.js +6 -4
  22. package/dist/es5/parquetjs/codecs/index.js.map +1 -1
  23. package/dist/es5/parquetjs/codecs/plain.js +43 -41
  24. package/dist/es5/parquetjs/codecs/plain.js.map +1 -1
  25. package/dist/es5/parquetjs/codecs/rle.js +35 -25
  26. package/dist/es5/parquetjs/codecs/rle.js.map +1 -1
  27. package/dist/es5/parquetjs/compression.js +110 -27
  28. package/dist/es5/parquetjs/compression.js.map +1 -1
  29. package/dist/es5/parquetjs/encoder/writer.js +737 -301
  30. package/dist/es5/parquetjs/encoder/writer.js.map +1 -1
  31. package/dist/es5/parquetjs/file.js +15 -15
  32. package/dist/es5/parquetjs/file.js.map +1 -1
  33. package/dist/es5/parquetjs/parquet-thrift/BoundaryOrder.js +1 -1
  34. package/dist/es5/parquetjs/parquet-thrift/BsonType.js +45 -31
  35. package/dist/es5/parquetjs/parquet-thrift/BsonType.js.map +1 -1
  36. package/dist/es5/parquetjs/parquet-thrift/ColumnChunk.js +152 -141
  37. package/dist/es5/parquetjs/parquet-thrift/ColumnChunk.js.map +1 -1
  38. package/dist/es5/parquetjs/parquet-thrift/ColumnIndex.js +160 -147
  39. package/dist/es5/parquetjs/parquet-thrift/ColumnIndex.js.map +1 -1
  40. package/dist/es5/parquetjs/parquet-thrift/ColumnMetaData.js +259 -248
  41. package/dist/es5/parquetjs/parquet-thrift/ColumnMetaData.js.map +1 -1
  42. package/dist/es5/parquetjs/parquet-thrift/ColumnOrder.js +79 -67
  43. package/dist/es5/parquetjs/parquet-thrift/ColumnOrder.js.map +1 -1
  44. package/dist/es5/parquetjs/parquet-thrift/CompressionCodec.js +1 -1
  45. package/dist/es5/parquetjs/parquet-thrift/ConvertedType.js +1 -1
  46. package/dist/es5/parquetjs/parquet-thrift/DataPageHeader.js +124 -113
  47. package/dist/es5/parquetjs/parquet-thrift/DataPageHeader.js.map +1 -1
  48. package/dist/es5/parquetjs/parquet-thrift/DataPageHeaderV2.js +169 -158
  49. package/dist/es5/parquetjs/parquet-thrift/DataPageHeaderV2.js.map +1 -1
  50. package/dist/es5/parquetjs/parquet-thrift/DateType.js +45 -31
  51. package/dist/es5/parquetjs/parquet-thrift/DateType.js.map +1 -1
  52. package/dist/es5/parquetjs/parquet-thrift/DecimalType.js +79 -68
  53. package/dist/es5/parquetjs/parquet-thrift/DecimalType.js.map +1 -1
  54. package/dist/es5/parquetjs/parquet-thrift/DictionaryPageHeader.js +94 -83
  55. package/dist/es5/parquetjs/parquet-thrift/DictionaryPageHeader.js.map +1 -1
  56. package/dist/es5/parquetjs/parquet-thrift/Encoding.js +1 -1
  57. package/dist/es5/parquetjs/parquet-thrift/EnumType.js +45 -31
  58. package/dist/es5/parquetjs/parquet-thrift/EnumType.js.map +1 -1
  59. package/dist/es5/parquetjs/parquet-thrift/FieldRepetitionType.js +1 -1
  60. package/dist/es5/parquetjs/parquet-thrift/FileMetaData.js +182 -170
  61. package/dist/es5/parquetjs/parquet-thrift/FileMetaData.js.map +1 -1
  62. package/dist/es5/parquetjs/parquet-thrift/IndexPageHeader.js +45 -31
  63. package/dist/es5/parquetjs/parquet-thrift/IndexPageHeader.js.map +1 -1
  64. package/dist/es5/parquetjs/parquet-thrift/IntType.js +79 -68
  65. package/dist/es5/parquetjs/parquet-thrift/IntType.js.map +1 -1
  66. package/dist/es5/parquetjs/parquet-thrift/JsonType.js +45 -31
  67. package/dist/es5/parquetjs/parquet-thrift/JsonType.js.map +1 -1
  68. package/dist/es5/parquetjs/parquet-thrift/KeyValue.js +79 -68
  69. package/dist/es5/parquetjs/parquet-thrift/KeyValue.js.map +1 -1
  70. package/dist/es5/parquetjs/parquet-thrift/ListType.js +45 -31
  71. package/dist/es5/parquetjs/parquet-thrift/ListType.js.map +1 -1
  72. package/dist/es5/parquetjs/parquet-thrift/LogicalType.js +343 -319
  73. package/dist/es5/parquetjs/parquet-thrift/LogicalType.js.map +1 -1
  74. package/dist/es5/parquetjs/parquet-thrift/MapType.js +45 -31
  75. package/dist/es5/parquetjs/parquet-thrift/MapType.js.map +1 -1
  76. package/dist/es5/parquetjs/parquet-thrift/MicroSeconds.js +45 -31
  77. package/dist/es5/parquetjs/parquet-thrift/MicroSeconds.js.map +1 -1
  78. package/dist/es5/parquetjs/parquet-thrift/MilliSeconds.js +45 -31
  79. package/dist/es5/parquetjs/parquet-thrift/MilliSeconds.js.map +1 -1
  80. package/dist/es5/parquetjs/parquet-thrift/NullType.js +45 -31
  81. package/dist/es5/parquetjs/parquet-thrift/NullType.js.map +1 -1
  82. package/dist/es5/parquetjs/parquet-thrift/OffsetIndex.js +75 -64
  83. package/dist/es5/parquetjs/parquet-thrift/OffsetIndex.js.map +1 -1
  84. package/dist/es5/parquetjs/parquet-thrift/PageEncodingStats.js +94 -83
  85. package/dist/es5/parquetjs/parquet-thrift/PageEncodingStats.js.map +1 -1
  86. package/dist/es5/parquetjs/parquet-thrift/PageHeader.js +169 -158
  87. package/dist/es5/parquetjs/parquet-thrift/PageHeader.js.map +1 -1
  88. package/dist/es5/parquetjs/parquet-thrift/PageLocation.js +94 -83
  89. package/dist/es5/parquetjs/parquet-thrift/PageLocation.js.map +1 -1
  90. package/dist/es5/parquetjs/parquet-thrift/PageType.js +1 -1
  91. package/dist/es5/parquetjs/parquet-thrift/RowGroup.js +124 -113
  92. package/dist/es5/parquetjs/parquet-thrift/RowGroup.js.map +1 -1
  93. package/dist/es5/parquetjs/parquet-thrift/SchemaElement.js +199 -188
  94. package/dist/es5/parquetjs/parquet-thrift/SchemaElement.js.map +1 -1
  95. package/dist/es5/parquetjs/parquet-thrift/SortingColumn.js +94 -83
  96. package/dist/es5/parquetjs/parquet-thrift/SortingColumn.js.map +1 -1
  97. package/dist/es5/parquetjs/parquet-thrift/Statistics.js +135 -124
  98. package/dist/es5/parquetjs/parquet-thrift/Statistics.js.map +1 -1
  99. package/dist/es5/parquetjs/parquet-thrift/StringType.js +45 -31
  100. package/dist/es5/parquetjs/parquet-thrift/StringType.js.map +1 -1
  101. package/dist/es5/parquetjs/parquet-thrift/TimeType.js +79 -68
  102. package/dist/es5/parquetjs/parquet-thrift/TimeType.js.map +1 -1
  103. package/dist/es5/parquetjs/parquet-thrift/TimeUnit.js +101 -88
  104. package/dist/es5/parquetjs/parquet-thrift/TimeUnit.js.map +1 -1
  105. package/dist/es5/parquetjs/parquet-thrift/TimestampType.js +79 -68
  106. package/dist/es5/parquetjs/parquet-thrift/TimestampType.js.map +1 -1
  107. package/dist/es5/parquetjs/parquet-thrift/Type.js +1 -1
  108. package/dist/es5/parquetjs/parquet-thrift/TypeDefinedOrder.js +45 -31
  109. package/dist/es5/parquetjs/parquet-thrift/TypeDefinedOrder.js.map +1 -1
  110. package/dist/es5/parquetjs/parquet-thrift/UUIDType.js +45 -31
  111. package/dist/es5/parquetjs/parquet-thrift/UUIDType.js.map +1 -1
  112. package/dist/es5/parquetjs/parquet-thrift/index.js +43 -43
  113. package/dist/es5/parquetjs/parser/decoders.js +391 -218
  114. package/dist/es5/parquetjs/parser/decoders.js.map +1 -1
  115. package/dist/es5/parquetjs/parser/parquet-cursor.js +180 -62
  116. package/dist/es5/parquetjs/parser/parquet-cursor.js.map +1 -1
  117. package/dist/es5/parquetjs/parser/parquet-envelope-reader.js +370 -125
  118. package/dist/es5/parquetjs/parser/parquet-envelope-reader.js.map +1 -1
  119. package/dist/es5/parquetjs/parser/parquet-reader.js +320 -91
  120. package/dist/es5/parquetjs/parser/parquet-reader.js.map +1 -1
  121. package/dist/es5/parquetjs/schema/declare.js +11 -9
  122. package/dist/es5/parquetjs/schema/declare.js.map +1 -1
  123. package/dist/es5/parquetjs/schema/schema.js +87 -73
  124. package/dist/es5/parquetjs/schema/schema.js.map +1 -1
  125. package/dist/es5/parquetjs/schema/shred.js +96 -56
  126. package/dist/es5/parquetjs/schema/shred.js.map +1 -1
  127. package/dist/es5/parquetjs/schema/types.js +40 -39
  128. package/dist/es5/parquetjs/schema/types.js.map +1 -1
  129. package/dist/es5/parquetjs/utils/buffer-utils.js +1 -1
  130. package/dist/es5/parquetjs/utils/buffer-utils.js.map +1 -1
  131. package/dist/es5/parquetjs/utils/file-utils.js +12 -8
  132. package/dist/es5/parquetjs/utils/file-utils.js.map +1 -1
  133. package/dist/es5/parquetjs/utils/read-utils.js +50 -22
  134. package/dist/es5/parquetjs/utils/read-utils.js.map +1 -1
  135. package/dist/esm/parquet-loader.js +1 -1
  136. package/dist/esm/parquet-loader.js.map +1 -1
  137. package/dist/esm/parquet-writer.js +1 -1
  138. package/dist/esm/parquet-writer.js.map +1 -1
  139. package/package.json +5 -5
@@ -1 +1 @@
1
- {"version":3,"sources":["../../../../src/parquetjs/parser/decoders.ts"],"names":["decodeDataPages","buffer","options","cursor","offset","size","length","data","rlevels","dlevels","values","pageHeaders","count","dictionary","numValues","Number","page","decodePage","map","value","index","push","undefined","pageHeader","pageType","PageType","type","decodeDataPage","decodeDataPageV2","decodeDictionaryPage","Error","decodeSchema","schemaElements","len","schema","next","i","schemaElement","repetitionType","FieldRepetitionType","repetition_type","optional","repeated","num_children","res","name","fields","Type","logicalType","converted_type","ConvertedType","typeLength","type_length","presision","precision","scale","decodeValues","encoding","opts","PARQUET_CODECS","header","cursorEnd","compressed_page_size","valueCount","data_page_header","num_values","dataCursor","compression","valuesBuf","slice","uncompressed_page_size","rLevelEncoding","Encoding","repetition_level_encoding","rLevels","Array","column","rLevelMax","PARQUET_RDLVL_TYPE","bitWidth","disableEnvelope","fill","dLevelEncoding","definition_level_encoding","dLevels","dLevelMax","valueCountNonNull","dlvl","valueEncoding","decodeOptions","primitiveType","data_page_header_v2","num_nulls","PARQUET_RDLVL_ENCODING","valuesBufCursor","is_compressed","dictCursor","dictionary_page_header","d","toString"],"mappings":";;;;;;;;;AAUA;;AACA;;AASA;;AACA;;AACA;;AASO,eAAeA,eAAf,CACLC,MADK,EAELC,OAFK,EAGiB;AACtB,QAAMC,MAAoB,GAAG;AAC3BF,IAAAA,MAD2B;AAE3BG,IAAAA,MAAM,EAAE,CAFmB;AAG3BC,IAAAA,IAAI,EAAEJ,MAAM,CAACK;AAHc,GAA7B;AAMA,QAAMC,IAAiB,GAAG;AACxBC,IAAAA,OAAO,EAAE,EADe;AAExBC,IAAAA,OAAO,EAAE,EAFe;AAGxBC,IAAAA,MAAM,EAAE,EAHgB;AAIxBC,IAAAA,WAAW,EAAE,EAJW;AAKxBC,IAAAA,KAAK,EAAE;AALiB,GAA1B;AAQA,MAAIC,UAAU,GAAGX,OAAO,CAACW,UAAR,IAAsB,EAAvC;;AAEA,SAEEV,MAAM,CAACC,MAAP,GAAgBD,MAAM,CAACE,IAAvB,KACC,CAACH,OAAO,CAACY,SAAT,IAAsBP,IAAI,CAACE,OAAL,CAAaH,MAAb,GAAsBS,MAAM,CAACb,OAAO,CAACY,SAAT,CADnD,CAFF,EAIE;AAEA,UAAME,IAAI,GAAG,MAAMC,UAAU,CAACd,MAAD,EAASD,OAAT,CAA7B;;AAEA,QAAIc,IAAI,CAACH,UAAT,EAAqB;AACnBA,MAAAA,UAAU,GAAGG,IAAI,CAACH,UAAlB;AAEA;AACD;;AAED,QAAIA,UAAU,CAACP,MAAf,EAAuB;AAErBU,MAAAA,IAAI,CAACN,MAAL,GAAcM,IAAI,CAACN,MAAL,CAAYQ,GAAZ,CAAiBC,KAAD,IAAWN,UAAU,CAACM,KAAD,CAArC,CAAd;AACD;;AAED,SAAK,IAAIC,KAAK,GAAG,CAAjB,EAAoBA,KAAK,GAAGJ,IAAI,CAACR,OAAL,CAAaF,MAAzC,EAAiDc,KAAK,EAAtD,EAA0D;AACxDb,MAAAA,IAAI,CAACC,OAAL,CAAaa,IAAb,CAAkBL,IAAI,CAACR,OAAL,CAAaY,KAAb,CAAlB;AACAb,MAAAA,IAAI,CAACE,OAAL,CAAaY,IAAb,CAAkBL,IAAI,CAACP,OAAL,CAAaW,KAAb,CAAlB;AACA,YAAMD,KAAK,GAAGH,IAAI,CAACN,MAAL,CAAYU,KAAZ,CAAd;;AAEA,UAAID,KAAK,KAAKG,SAAd,EAAyB;AACvBf,QAAAA,IAAI,CAACG,MAAL,CAAYW,IAAZ,CAAiBF,KAAjB;AACD;AACF;;AAEDZ,IAAAA,IAAI,CAACK,KAAL,IAAcI,IAAI,CAACJ,KAAnB;AACAL,IAAAA,IAAI,CAACI,WAAL,CAAiBU,IAAjB,CAAsBL,IAAI,CAACO,UAA3B;AACD;;AAED,SAAOhB,IAAP;AACD;;AAOM,eAAeU,UAAf,CACLd,MADK,EAELD,OAFK,EAGqB;AAC1B,MAAIc,IAAJ;AACA,QAAM;AAACO,IAAAA,UAAD;AAAajB,IAAAA;AAAb,MAAuB,iCAAiBH,MAAM,CAACF,MAAxB,EAAgCE,MAAM,CAACC,MAAvC,CAA7B;AACAD,EAAAA,MAAM,CAACC,MAAP,IAAiBE,MAAjB;AAEA,QAAMkB,QAAQ,GAAG,8BAAcC,uBAAd,EAAwBF,UAAU,CAACG,IAAnC,CAAjB;;AAEA,UAAQF,QAAR;AACE,SAAK,WAAL;AACER,MAAAA,IAAI,GAAG,MAAMW,cAAc,CAACxB,MAAD,EAASoB,UAAT,EAAqBrB,OAArB,CAA3B;AACA;;AACF,SAAK,cAAL;AACEc,MAAAA,IAAI,GAAG,MAAMY,gBAAgB,CAACzB,MAAD,EAASoB,UAAT,EAAqBrB,OAArB,CAA7B;AACA;;AACF,SAAK,iBAAL;AACEc,MAAAA,IAAI,GAAG;AACLH,QAAAA,UAAU,EAAE,MAAMgB,oBAAoB,CAAC1B,MAAD,EAASoB,UAAT,EAAqBrB,OAArB,CADjC;AAELqB,QAAAA;AAFK,OAAP;AAIA;;AACF;AACE,YAAM,IAAIO,KAAJ,8BAAgCN,QAAhC,EAAN;AAdJ;;AAiBA,SAAOR,IAAP;AACD;;AAYM,SAASe,YAAT,CACLC,cADK,EAEL5B,MAFK,EAGL6B,GAHK,EAQL;AACA,QAAMC,MAAwB,GAAG,EAAjC;AACA,MAAIC,IAAI,GAAG/B,MAAX;;AACA,OAAK,IAAIgC,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGH,GAApB,EAAyBG,CAAC,EAA1B,EAA8B;AAC5B,UAAMC,aAAa,GAAGL,cAAc,CAACG,IAAD,CAApC;AAEA,UAAMG,cAAc,GAClBH,IAAI,GAAG,CAAP,GAAW,8BAAcI,kCAAd,EAAmCF,aAAa,CAACG,eAAjD,CAAX,GAAgF,MADlF;AAGA,QAAIC,QAAQ,GAAG,KAAf;AACA,QAAIC,QAAQ,GAAG,KAAf;;AACA,YAAQJ,cAAR;AACE,WAAK,UAAL;AACE;;AACF,WAAK,UAAL;AACEG,QAAAA,QAAQ,GAAG,IAAX;AACA;;AACF,WAAK,UAAL;AACEC,QAAAA,QAAQ,GAAG,IAAX;AACA;;AACF;AACE,cAAM,IAAIZ,KAAJ,CAAU,kCAAV,CAAN;AAVJ;;AAaA,QAAIO,aAAa,CAACM,YAAd,GAA8B,CAAlC,EAAqC;AACnC,YAAMC,GAAG,GAAGb,YAAY,CAACC,cAAD,EAAiBG,IAAI,GAAG,CAAxB,EAA2BE,aAAa,CAACM,YAAzC,CAAxB;AACAR,MAAAA,IAAI,GAAGS,GAAG,CAACT,IAAX;AACAD,MAAAA,MAAM,CAACG,aAAa,CAACQ,IAAf,CAAN,GAA6B;AAE3BJ,QAAAA,QAF2B;AAG3BC,QAAAA,QAH2B;AAI3BI,QAAAA,MAAM,EAAEF,GAAG,CAACV;AAJe,OAA7B;AAMD,KATD,MASO;AACL,YAAMR,IAAI,GAAG,8BAAcqB,mBAAd,EAAoBV,aAAa,CAACX,IAAlC,CAAb;AACA,UAAIsB,WAAW,GAAGtB,IAAlB;;AAEA,UAAIW,aAAa,CAACY,cAAlB,EAAkC;AAChCD,QAAAA,WAAW,GAAG,8BAAcE,4BAAd,EAA6Bb,aAAa,CAACY,cAA3C,CAAd;AACD;;AAED,cAAQD,WAAR;AACE,aAAK,SAAL;AACEA,UAAAA,WAAW,aAAMA,WAAN,cAAqBtB,IAArB,CAAX;AACA;;AACF;AAJF;;AAOAQ,MAAAA,MAAM,CAACG,aAAa,CAACQ,IAAf,CAAN,GAA6B;AAC3BnB,QAAAA,IAAI,EAAEsB,WADqB;AAE3BG,QAAAA,UAAU,EAAEd,aAAa,CAACe,WAFC;AAG3BC,QAAAA,SAAS,EAAEhB,aAAa,CAACiB,SAHE;AAI3BC,QAAAA,KAAK,EAAElB,aAAa,CAACkB,KAJM;AAK3Bd,QAAAA,QAL2B;AAM3BC,QAAAA;AAN2B,OAA7B;AAQAP,MAAAA,IAAI;AACL;AACF;;AACD,SAAO;AAACD,IAAAA,MAAD;AAAS9B,IAAAA,MAAT;AAAiB+B,IAAAA;AAAjB,GAAP;AACD;;AAKD,SAASqB,YAAT,CACE9B,IADF,EAEE+B,QAFF,EAGEtD,MAHF,EAIES,KAJF,EAKE8C,IALF,EAMS;AACP,MAAI,EAAED,QAAQ,IAAIE,sBAAd,CAAJ,EAAmC;AACjC,UAAM,IAAI7B,KAAJ,6BAA+B2B,QAA/B,EAAN;AACD;;AACD,SAAOE,uBAAeF,QAAf,EAAyBD,YAAzB,CAAsC9B,IAAtC,EAA4CvB,MAA5C,EAAoDS,KAApD,EAA2D8C,IAA3D,CAAP;AACD;;AAQD,eAAe/B,cAAf,CACExB,MADF,EAEEyD,MAFF,EAGE1D,OAHF,EAI4B;AAAA;;AAC1B,QAAM2D,SAAS,GAAG1D,MAAM,CAACC,MAAP,GAAgBwD,MAAM,CAACE,oBAAzC;AACA,QAAMC,UAAU,4BAAGH,MAAM,CAACI,gBAAV,0DAAG,sBAAyBC,UAA5C;AAGA,MAAIC,UAAU,GAAG/D,MAAjB;;AAEA,MAAID,OAAO,CAACiE,WAAR,KAAwB,cAA5B,EAA4C;AAC1C,UAAMC,SAAS,GAAG,MAAM,6BACtBlE,OAAO,CAACiE,WADc,EAEtBhE,MAAM,CAACF,MAAP,CAAcoE,KAAd,CAAoBlE,MAAM,CAACC,MAA3B,EAAmCyD,SAAnC,CAFsB,EAGtBD,MAAM,CAACU,sBAHe,CAAxB;AAKAJ,IAAAA,UAAU,GAAG;AACXjE,MAAAA,MAAM,EAAEmE,SADG;AAEXhE,MAAAA,MAAM,EAAE,CAFG;AAGXC,MAAAA,IAAI,EAAE+D,SAAS,CAAC9D;AAHL,KAAb;AAKAH,IAAAA,MAAM,CAACC,MAAP,GAAgByD,SAAhB;AACD;;AAGD,QAAMU,cAAc,GAAG,8BACrBC,uBADqB,4BAErBZ,MAAM,CAACI,gBAFc,2DAErB,uBAAyBS,yBAFJ,CAAvB;AAKA,MAAIC,OAAO,GAAG,IAAIC,KAAJ,CAAUZ,UAAV,CAAd;;AAEA,MAAI7D,OAAO,CAAC0E,MAAR,CAAeC,SAAf,GAA2B,CAA/B,EAAkC;AAChCH,IAAAA,OAAO,GAAGlB,YAAY,CAACsB,6BAAD,EAAqBP,cAArB,EAAqCL,UAArC,EAAiDH,UAAjD,EAA8D;AAClFgB,MAAAA,QAAQ,EAAE,4BAAY7E,OAAO,CAAC0E,MAAR,CAAeC,SAA3B,CADwE;AAElFG,MAAAA,eAAe,EAAE;AAFiE,KAA9D,CAAtB;AAKD,GAND,MAMO;AACLN,IAAAA,OAAO,CAACO,IAAR,CAAa,CAAb;AACD;;AAGD,QAAMC,cAAc,GAAG,8BACrBV,uBADqB,4BAErBZ,MAAM,CAACI,gBAFc,2DAErB,uBAAyBmB,yBAFJ,CAAvB;AAKA,MAAIC,OAAO,GAAG,IAAIT,KAAJ,CAAUZ,UAAV,CAAd;;AACA,MAAI7D,OAAO,CAAC0E,MAAR,CAAeS,SAAf,GAA2B,CAA/B,EAAkC;AAChCD,IAAAA,OAAO,GAAG5B,YAAY,CAACsB,6BAAD,EAAqBI,cAArB,EAAqChB,UAArC,EAAiDH,UAAjD,EAA8D;AAClFgB,MAAAA,QAAQ,EAAE,4BAAY7E,OAAO,CAAC0E,MAAR,CAAeS,SAA3B,CADwE;AAElFL,MAAAA,eAAe,EAAE;AAFiE,KAA9D,CAAtB;AAKD,GAND,MAMO;AACLI,IAAAA,OAAO,CAACH,IAAR,CAAa,CAAb;AACD;;AACD,MAAIK,iBAAiB,GAAG,CAAxB;;AACA,OAAK,MAAMC,IAAX,IAAmBH,OAAnB,EAA4B;AAC1B,QAAIG,IAAI,KAAKrF,OAAO,CAAC0E,MAAR,CAAeS,SAA5B,EAAuC;AACrCC,MAAAA,iBAAiB;AAClB;AACF;;AAGD,QAAME,aAAa,GAAG,8BAAchB,uBAAd,4BAAwBZ,MAAM,CAACI,gBAA/B,2DAAwB,uBAAyBP,QAAjD,CAAtB;AACA,QAAMgC,aAAa,GAAG;AACpBtC,IAAAA,UAAU,EAAEjD,OAAO,CAAC0E,MAAR,CAAezB,UADP;AAEpB4B,IAAAA,QAAQ,EAAE7E,OAAO,CAAC0E,MAAR,CAAezB;AAFL,GAAtB;AAKA,QAAMzC,MAAM,GAAG8C,YAAY,CACzBtD,OAAO,CAAC0E,MAAR,CAAec,aADU,EAEzBF,aAFyB,EAGzBtB,UAHyB,EAIzBoB,iBAJyB,EAKzBG,aALyB,CAA3B;AAQA,SAAO;AACLhF,IAAAA,OAAO,EAAE2E,OADJ;AAEL5E,IAAAA,OAAO,EAAEkE,OAFJ;AAGLhE,IAAAA,MAHK;AAILE,IAAAA,KAAK,EAAEmD,UAJF;AAKLxC,IAAAA,UAAU,EAAEqC;AALP,GAAP;AAOD;;AASD,eAAehC,gBAAf,CACEzB,MADF,EAEEyD,MAFF,EAGEF,IAHF,EAI4B;AAAA;;AAC1B,QAAMG,SAAS,GAAG1D,MAAM,CAACC,MAAP,GAAgBwD,MAAM,CAACE,oBAAzC;AAEA,QAAMC,UAAU,6BAAGH,MAAM,CAAC+B,mBAAV,2DAAG,uBAA4B1B,UAA/C;AAEA,QAAMqB,iBAAiB,GAAGvB,UAAU,8BAAGH,MAAM,CAAC+B,mBAAV,2DAAG,uBAA4BC,SAA/B,CAApC;AACA,QAAMJ,aAAa,GAAG,8BACpBhB,uBADoB,4BAEpBZ,MAAM,CAAC+B,mBAFa,2DAEpB,uBAA4BlC,QAFR,CAAtB;AAOA,MAAIiB,OAAO,GAAG,IAAIC,KAAJ,CAAUZ,UAAV,CAAd;;AACA,MAAIL,IAAI,CAACkB,MAAL,CAAYC,SAAZ,GAAwB,CAA5B,EAA+B;AAC7BH,IAAAA,OAAO,GAAGlB,YAAY,CAACsB,6BAAD,EAAqBe,iCAArB,EAA6C1F,MAA7C,EAAqD4D,UAArD,EAAkE;AACtFgB,MAAAA,QAAQ,EAAE,4BAAYrB,IAAI,CAACkB,MAAL,CAAYC,SAAxB,CAD4E;AAEtFG,MAAAA,eAAe,EAAE;AAFqE,KAAlE,CAAtB;AAID,GALD,MAKO;AACLN,IAAAA,OAAO,CAACO,IAAR,CAAa,CAAb;AACD;;AAID,MAAIG,OAAO,GAAG,IAAIT,KAAJ,CAAUZ,UAAV,CAAd;;AACA,MAAIL,IAAI,CAACkB,MAAL,CAAYS,SAAZ,GAAwB,CAA5B,EAA+B;AAC7BD,IAAAA,OAAO,GAAG5B,YAAY,CAACsB,6BAAD,EAAqBe,iCAArB,EAA6C1F,MAA7C,EAAqD4D,UAArD,EAAkE;AACtFgB,MAAAA,QAAQ,EAAE,4BAAYrB,IAAI,CAACkB,MAAL,CAAYS,SAAxB,CAD4E;AAEtFL,MAAAA,eAAe,EAAE;AAFqE,KAAlE,CAAtB;AAID,GALD,MAKO;AACLI,IAAAA,OAAO,CAACH,IAAR,CAAa,CAAb;AACD;;AAGD,MAAIa,eAAe,GAAG3F,MAAtB;;AAEA,gCAAIyD,MAAM,CAAC+B,mBAAX,mDAAI,uBAA4BI,aAAhC,EAA+C;AAC7C,UAAM3B,SAAS,GAAG,MAAM,6BACtBV,IAAI,CAACS,WADiB,EAEtBhE,MAAM,CAACF,MAAP,CAAcoE,KAAd,CAAoBlE,MAAM,CAACC,MAA3B,EAAmCyD,SAAnC,CAFsB,EAGtBD,MAAM,CAACU,sBAHe,CAAxB;AAMAwB,IAAAA,eAAe,GAAG;AAChB7F,MAAAA,MAAM,EAAEmE,SADQ;AAEhBhE,MAAAA,MAAM,EAAE,CAFQ;AAGhBC,MAAAA,IAAI,EAAE+D,SAAS,CAAC9D;AAHA,KAAlB;AAMAH,IAAAA,MAAM,CAACC,MAAP,GAAgByD,SAAhB;AACD;;AAED,QAAM4B,aAAa,GAAG;AACpBtC,IAAAA,UAAU,EAAEO,IAAI,CAACkB,MAAL,CAAYzB,UADJ;AAEpB4B,IAAAA,QAAQ,EAAErB,IAAI,CAACkB,MAAL,CAAYzB;AAFF,GAAtB;AAKA,QAAMzC,MAAM,GAAG8C,YAAY,CACzBE,IAAI,CAACkB,MAAL,CAAYc,aADa,EAEzBF,aAFyB,EAGzBM,eAHyB,EAIzBR,iBAJyB,EAKzBG,aALyB,CAA3B;AAQA,SAAO;AACLhF,IAAAA,OAAO,EAAE2E,OADJ;AAEL5E,IAAAA,OAAO,EAAEkE,OAFJ;AAGLhE,IAAAA,MAHK;AAILE,IAAAA,KAAK,EAAEmD,UAJF;AAKLxC,IAAAA,UAAU,EAAEqC;AALP,GAAP;AAOD;;AAQD,eAAe/B,oBAAf,CACE1B,MADF,EAEEoB,UAFF,EAGErB,OAHF,EAIqB;AAAA;;AACnB,QAAM2D,SAAS,GAAG1D,MAAM,CAACC,MAAP,GAAgBmB,UAAU,CAACuC,oBAA7C;AAEA,MAAIkC,UAAU,GAAG;AACf5F,IAAAA,MAAM,EAAE,CADO;AAEfH,IAAAA,MAAM,EAAEE,MAAM,CAACF,MAAP,CAAcoE,KAAd,CAAoBlE,MAAM,CAACC,MAA3B,EAAmCyD,SAAnC,CAFO;AAGfxD,IAAAA,IAAI,EAAEwD,SAAS,GAAG1D,MAAM,CAACC;AAHV,GAAjB;AAMAD,EAAAA,MAAM,CAACC,MAAP,GAAgByD,SAAhB;;AAEA,MAAI3D,OAAO,CAACiE,WAAR,KAAwB,cAA5B,EAA4C;AAC1C,UAAMC,SAAS,GAAG,MAAM,6BACtBlE,OAAO,CAACiE,WADc,EAEtB6B,UAAU,CAAC/F,MAAX,CAAkBoE,KAAlB,CAAwB2B,UAAU,CAAC5F,MAAnC,EAA2CyD,SAA3C,CAFsB,EAGtBtC,UAAU,CAAC+C,sBAHW,CAAxB;AAMA0B,IAAAA,UAAU,GAAG;AACX/F,MAAAA,MAAM,EAAEmE,SADG;AAEXhE,MAAAA,MAAM,EAAE,CAFG;AAGXC,MAAAA,IAAI,EAAE+D,SAAS,CAAC9D;AAHL,KAAb;AAMAH,IAAAA,MAAM,CAACC,MAAP,GAAgByD,SAAhB;AACD;;AAED,QAAM/C,SAAS,GAAG,CAAAS,UAAU,SAAV,IAAAA,UAAU,WAAV,qCAAAA,UAAU,CAAE0E,sBAAZ,gFAAoChC,UAApC,KAAkD,CAApE;AAEA,SAAOT,YAAY,CACjBtD,OAAO,CAAC0E,MAAR,CAAec,aADE,EAEjBxF,OAAO,CAAC0E,MAAR,CAAenB,QAFE,EAGjBuC,UAHiB,EAIjBlF,SAJiB,EAKjBZ,OALiB,CAAZ,CAMLgB,GANK,CAMAgF,CAAD,IAAOA,CAAC,CAACC,QAAF,EANN,CAAP;AAOD","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\nimport {\n ParquetCodec,\n ParquetData,\n ParquetOptions,\n ParquetPageData,\n ParquetType,\n PrimitiveType,\n SchemaDefinition\n} from '../schema/declare';\nimport {CursorBuffer, ParquetCodecOptions, PARQUET_CODECS} from '../codecs';\nimport {\n ConvertedType,\n Encoding,\n FieldRepetitionType,\n PageHeader,\n PageType,\n SchemaElement,\n Type\n} from '../parquet-thrift';\nimport {decompress} from '../compression';\nimport {PARQUET_RDLVL_TYPE, PARQUET_RDLVL_ENCODING} from '../../constants';\nimport {decodePageHeader, getThriftEnum, getBitWidth} from '../utils/read-utils';\n\n/**\n * Decode data pages\n * @param buffer - input data\n * @param column - parquet column\n * @param compression - compression type\n * @returns parquet data page data\n */\nexport async function decodeDataPages(\n buffer: Buffer,\n options: ParquetOptions\n): Promise<ParquetData> {\n const cursor: CursorBuffer = {\n buffer,\n offset: 0,\n size: buffer.length\n };\n\n const data: ParquetData = {\n rlevels: [],\n dlevels: [],\n values: [],\n pageHeaders: [],\n count: 0\n };\n\n let dictionary = options.dictionary || [];\n\n while (\n // @ts-ignore size can be undefined\n cursor.offset < cursor.size &&\n (!options.numValues || data.dlevels.length < Number(options.numValues))\n ) {\n // Looks like we have to decode these in sequence due to cursor updates?\n const page = await decodePage(cursor, options);\n\n if (page.dictionary) {\n dictionary = page.dictionary;\n // eslint-disable-next-line no-continue\n continue;\n }\n\n if (dictionary.length) {\n // eslint-disable-next-line no-loop-func\n page.values = page.values.map((value) => dictionary[value]);\n }\n\n for (let index = 0; index < page.rlevels.length; index++) {\n data.rlevels.push(page.rlevels[index]);\n data.dlevels.push(page.dlevels[index]);\n const value = page.values[index];\n\n if (value !== undefined) {\n data.values.push(value);\n }\n }\n\n data.count += page.count;\n data.pageHeaders.push(page.pageHeader);\n }\n\n return data;\n}\n\n/**\n * Decode parquet page based on page type\n * @param cursor\n * @param options\n */\nexport async function decodePage(\n cursor: CursorBuffer,\n options: ParquetOptions\n): Promise<ParquetPageData> {\n let page;\n const {pageHeader, length} = decodePageHeader(cursor.buffer, cursor.offset);\n cursor.offset += length;\n\n const pageType = getThriftEnum(PageType, pageHeader.type);\n\n switch (pageType) {\n case 'DATA_PAGE':\n page = await decodeDataPage(cursor, pageHeader, options);\n break;\n case 'DATA_PAGE_V2':\n page = await decodeDataPageV2(cursor, pageHeader, options);\n break;\n case 'DICTIONARY_PAGE':\n page = {\n dictionary: await decodeDictionaryPage(cursor, pageHeader, options),\n pageHeader\n };\n break;\n default:\n throw new Error(`invalid page type: ${pageType}`);\n }\n\n return page;\n}\n\n/**\n * Decode parquet schema\n * @param schemaElements input schema elements data\n * @param offset offset to read from\n * @param len length of data\n * @returns result.offset\n * result.next - offset at the end of function\n * result.schema - schema read from the input data\n * @todo output offset is the same as input - possibly excess output field\n */\nexport function decodeSchema(\n schemaElements: SchemaElement[],\n offset: number,\n len: number\n): {\n offset: number;\n next: number;\n schema: SchemaDefinition;\n} {\n const schema: SchemaDefinition = {};\n let next = offset;\n for (let i = 0; i < len; i++) {\n const schemaElement = schemaElements[next];\n\n const repetitionType =\n next > 0 ? getThriftEnum(FieldRepetitionType, schemaElement.repetition_type!) : 'ROOT';\n\n let optional = false;\n let repeated = false;\n switch (repetitionType) {\n case 'REQUIRED':\n break;\n case 'OPTIONAL':\n optional = true;\n break;\n case 'REPEATED':\n repeated = true;\n break;\n default:\n throw new Error('parquet: unknown repetition type');\n }\n\n if (schemaElement.num_children! > 0) {\n const res = decodeSchema(schemaElements, next + 1, schemaElement.num_children!);\n next = res.next;\n schema[schemaElement.name] = {\n // type: undefined,\n optional,\n repeated,\n fields: res.schema\n };\n } else {\n const type = getThriftEnum(Type, schemaElement.type!);\n let logicalType = type;\n\n if (schemaElement.converted_type) {\n logicalType = getThriftEnum(ConvertedType, schemaElement.converted_type);\n }\n\n switch (logicalType) {\n case 'DECIMAL':\n logicalType = `${logicalType}_${type}` as ParquetType;\n break;\n default:\n }\n\n schema[schemaElement.name] = {\n type: logicalType as ParquetType,\n typeLength: schemaElement.type_length,\n presision: schemaElement.precision,\n scale: schemaElement.scale,\n optional,\n repeated\n };\n next++;\n }\n }\n return {schema, offset, next};\n}\n\n/**\n * Decode a consecutive array of data using one of the parquet encodings\n */\nfunction decodeValues(\n type: PrimitiveType,\n encoding: ParquetCodec,\n cursor: CursorBuffer,\n count: number,\n opts: ParquetCodecOptions\n): any[] {\n if (!(encoding in PARQUET_CODECS)) {\n throw new Error(`invalid encoding: ${encoding}`);\n }\n return PARQUET_CODECS[encoding].decodeValues(type, cursor, count, opts);\n}\n\n/**\n * Do decoding of parquet dataPage from column chunk\n * @param cursor\n * @param header\n * @param options\n */\nasync function decodeDataPage(\n cursor: CursorBuffer,\n header: PageHeader,\n options: ParquetOptions\n): Promise<ParquetPageData> {\n const cursorEnd = cursor.offset + header.compressed_page_size;\n const valueCount = header.data_page_header?.num_values;\n\n /* uncompress page */\n let dataCursor = cursor;\n\n if (options.compression !== 'UNCOMPRESSED') {\n const valuesBuf = await decompress(\n options.compression,\n cursor.buffer.slice(cursor.offset, cursorEnd),\n header.uncompressed_page_size\n );\n dataCursor = {\n buffer: valuesBuf,\n offset: 0,\n size: valuesBuf.length\n };\n cursor.offset = cursorEnd;\n }\n\n /* read repetition levels */\n const rLevelEncoding = getThriftEnum(\n Encoding,\n header.data_page_header?.repetition_level_encoding!\n ) as ParquetCodec;\n // tslint:disable-next-line:prefer-array-literal\n let rLevels = new Array(valueCount);\n\n if (options.column.rLevelMax > 0) {\n rLevels = decodeValues(PARQUET_RDLVL_TYPE, rLevelEncoding, dataCursor, valueCount!, {\n bitWidth: getBitWidth(options.column.rLevelMax),\n disableEnvelope: false\n // column: opts.column\n });\n } else {\n rLevels.fill(0);\n }\n\n /* read definition levels */\n const dLevelEncoding = getThriftEnum(\n Encoding,\n header.data_page_header?.definition_level_encoding!\n ) as ParquetCodec;\n // tslint:disable-next-line:prefer-array-literal\n let dLevels = new Array(valueCount);\n if (options.column.dLevelMax > 0) {\n dLevels = decodeValues(PARQUET_RDLVL_TYPE, dLevelEncoding, dataCursor, valueCount!, {\n bitWidth: getBitWidth(options.column.dLevelMax),\n disableEnvelope: false\n // column: opts.column\n });\n } else {\n dLevels.fill(0);\n }\n let valueCountNonNull = 0;\n for (const dlvl of dLevels) {\n if (dlvl === options.column.dLevelMax) {\n valueCountNonNull++;\n }\n }\n\n /* read values */\n const valueEncoding = getThriftEnum(Encoding, header.data_page_header?.encoding!) as ParquetCodec;\n const decodeOptions = {\n typeLength: options.column.typeLength,\n bitWidth: options.column.typeLength\n };\n\n const values = decodeValues(\n options.column.primitiveType!,\n valueEncoding,\n dataCursor,\n valueCountNonNull,\n decodeOptions\n );\n\n return {\n dlevels: dLevels,\n rlevels: rLevels,\n values,\n count: valueCount!,\n pageHeader: header\n };\n}\n\n/**\n * Do decoding of parquet dataPage in version 2 from column chunk\n * @param cursor\n * @param header\n * @param opts\n * @returns\n */\nasync function decodeDataPageV2(\n cursor: CursorBuffer,\n header: PageHeader,\n opts: any\n): Promise<ParquetPageData> {\n const cursorEnd = cursor.offset + header.compressed_page_size;\n\n const valueCount = header.data_page_header_v2?.num_values;\n // @ts-ignore\n const valueCountNonNull = valueCount - header.data_page_header_v2?.num_nulls;\n const valueEncoding = getThriftEnum(\n Encoding,\n header.data_page_header_v2?.encoding!\n ) as ParquetCodec;\n\n /* read repetition levels */\n // tslint:disable-next-line:prefer-array-literal\n let rLevels = new Array(valueCount);\n if (opts.column.rLevelMax > 0) {\n rLevels = decodeValues(PARQUET_RDLVL_TYPE, PARQUET_RDLVL_ENCODING, cursor, valueCount!, {\n bitWidth: getBitWidth(opts.column.rLevelMax),\n disableEnvelope: true\n });\n } else {\n rLevels.fill(0);\n }\n\n /* read definition levels */\n // tslint:disable-next-line:prefer-array-literal\n let dLevels = new Array(valueCount);\n if (opts.column.dLevelMax > 0) {\n dLevels = decodeValues(PARQUET_RDLVL_TYPE, PARQUET_RDLVL_ENCODING, cursor, valueCount!, {\n bitWidth: getBitWidth(opts.column.dLevelMax),\n disableEnvelope: true\n });\n } else {\n dLevels.fill(0);\n }\n\n /* read values */\n let valuesBufCursor = cursor;\n\n if (header.data_page_header_v2?.is_compressed) {\n const valuesBuf = await decompress(\n opts.compression,\n cursor.buffer.slice(cursor.offset, cursorEnd),\n header.uncompressed_page_size\n );\n\n valuesBufCursor = {\n buffer: valuesBuf,\n offset: 0,\n size: valuesBuf.length\n };\n\n cursor.offset = cursorEnd;\n }\n\n const decodeOptions = {\n typeLength: opts.column.typeLength,\n bitWidth: opts.column.typeLength\n };\n\n const values = decodeValues(\n opts.column.primitiveType!,\n valueEncoding,\n valuesBufCursor,\n valueCountNonNull,\n decodeOptions\n );\n\n return {\n dlevels: dLevels,\n rlevels: rLevels,\n values,\n count: valueCount!,\n pageHeader: header\n };\n}\n\n/**\n * Do decoding of dictionary page which helps to iterate over all indexes and get dataPage values.\n * @param cursor\n * @param pageHeader\n * @param options\n */\nasync function decodeDictionaryPage(\n cursor: CursorBuffer,\n pageHeader: PageHeader,\n options: ParquetOptions\n): Promise<string[]> {\n const cursorEnd = cursor.offset + pageHeader.compressed_page_size;\n\n let dictCursor = {\n offset: 0,\n buffer: cursor.buffer.slice(cursor.offset, cursorEnd),\n size: cursorEnd - cursor.offset\n };\n\n cursor.offset = cursorEnd;\n\n if (options.compression !== 'UNCOMPRESSED') {\n const valuesBuf = await decompress(\n options.compression,\n dictCursor.buffer.slice(dictCursor.offset, cursorEnd),\n pageHeader.uncompressed_page_size\n );\n\n dictCursor = {\n buffer: valuesBuf,\n offset: 0,\n size: valuesBuf.length\n };\n\n cursor.offset = cursorEnd;\n }\n\n const numValues = pageHeader?.dictionary_page_header?.num_values || 0;\n\n return decodeValues(\n options.column.primitiveType!,\n options.column.encoding!,\n dictCursor,\n numValues,\n options as ParquetCodecOptions\n ).map((d) => d.toString());\n}\n"],"file":"decoders.js"}
1
+ {"version":3,"sources":["../../../../src/parquetjs/parser/decoders.ts"],"names":["decodeDataPages","buffer","options","cursor","offset","size","length","data","rlevels","dlevels","values","pageHeaders","count","dictionary","numValues","Number","decodePage","page","map","value","index","push","undefined","pageHeader","pageType","PageType","type","decodeDataPage","decodeDataPageV2","decodeDictionaryPage","Error","decodeSchema","schemaElements","len","schema","next","i","schemaElement","repetitionType","FieldRepetitionType","repetition_type","optional","repeated","num_children","res","name","fields","Type","logicalType","converted_type","ConvertedType","typeLength","type_length","presision","precision","scale","decodeValues","encoding","opts","PARQUET_CODECS","header","cursorEnd","compressed_page_size","valueCount","data_page_header","num_values","dataCursor","compression","slice","uncompressed_page_size","valuesBuf","rLevelEncoding","Encoding","repetition_level_encoding","rLevels","Array","column","rLevelMax","PARQUET_RDLVL_TYPE","bitWidth","disableEnvelope","fill","dLevelEncoding","definition_level_encoding","dLevels","dLevelMax","valueCountNonNull","dlvl","valueEncoding","decodeOptions","primitiveType","data_page_header_v2","num_nulls","PARQUET_RDLVL_ENCODING","valuesBufCursor","is_compressed","dictCursor","dictionary_page_header","d","toString"],"mappings":";;;;;;;;;;;;;;;AAUA;;AACA;;AASA;;AACA;;AACA;;;;;;;;SASsBA,e;;;;;+EAAf,iBACLC,MADK,EAELC,OAFK;AAAA;AAAA;AAAA;AAAA;AAAA;AAICC,YAAAA,MAJD,GAIwB;AAC3BF,cAAAA,MAAM,EAANA,MAD2B;AAE3BG,cAAAA,MAAM,EAAE,CAFmB;AAG3BC,cAAAA,IAAI,EAAEJ,MAAM,CAACK;AAHc,aAJxB;AAUCC,YAAAA,IAVD,GAUqB;AACxBC,cAAAA,OAAO,EAAE,EADe;AAExBC,cAAAA,OAAO,EAAE,EAFe;AAGxBC,cAAAA,MAAM,EAAE,EAHgB;AAIxBC,cAAAA,WAAW,EAAE,EAJW;AAKxBC,cAAAA,KAAK,EAAE;AALiB,aAVrB;AAkBDC,YAAAA,UAlBC,GAkBYX,OAAO,CAACW,UAAR,IAAsB,EAlBlC;;AAAA;AAAA,kBAsBHV,MAAM,CAACC,MAAP,GAAgBD,MAAM,CAACE,IAAvB,KACC,CAACH,OAAO,CAACY,SAAT,IAAsBP,IAAI,CAACE,OAAL,CAAaH,MAAb,GAAsBS,MAAM,CAACb,OAAO,CAACY,SAAT,CADnD,CAtBG;AAAA;AAAA;AAAA;;AAAA;AAAA,mBA0BgBE,UAAU,CAACb,MAAD,EAASD,OAAT,CA1B1B;;AAAA;AA0BGe,YAAAA,IA1BH;;AAAA,iBA4BCA,IAAI,CAACJ,UA5BN;AAAA;AAAA;AAAA;;AA6BDA,YAAAA,UAAU,GAAGI,IAAI,CAACJ,UAAlB;AA7BC;;AAAA;AAkCH,gBAAIA,UAAU,CAACP,MAAf,EAAuB;AAErBW,cAAAA,IAAI,CAACP,MAAL,GAAcO,IAAI,CAACP,MAAL,CAAYQ,GAAZ,CAAgB,UAACC,KAAD;AAAA,uBAAWN,UAAU,CAACM,KAAD,CAArB;AAAA,eAAhB,CAAd;AACD;;AAED,iBAASC,KAAT,GAAiB,CAAjB,EAAoBA,KAAK,GAAGH,IAAI,CAACT,OAAL,CAAaF,MAAzC,EAAiDc,KAAK,EAAtD,EAA0D;AACxDb,cAAAA,IAAI,CAACC,OAAL,CAAaa,IAAb,CAAkBJ,IAAI,CAACT,OAAL,CAAaY,KAAb,CAAlB;AACAb,cAAAA,IAAI,CAACE,OAAL,CAAaY,IAAb,CAAkBJ,IAAI,CAACR,OAAL,CAAaW,KAAb,CAAlB;AACMD,cAAAA,KAHkD,GAG1CF,IAAI,CAACP,MAAL,CAAYU,KAAZ,CAH0C;;AAKxD,kBAAID,KAAK,KAAKG,SAAd,EAAyB;AACvBf,gBAAAA,IAAI,CAACG,MAAL,CAAYW,IAAZ,CAAiBF,KAAjB;AACD;AACF;;AAEDZ,YAAAA,IAAI,CAACK,KAAL,IAAcK,IAAI,CAACL,KAAnB;AACAL,YAAAA,IAAI,CAACI,WAAL,CAAiBU,IAAjB,CAAsBJ,IAAI,CAACM,UAA3B;AAlDG;AAAA;;AAAA;AAAA,6CAqDEhB,IArDF;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,G;;;;SA6DeS,U;;;;;0EAAf,kBACLb,MADK,EAELD,OAFK;AAAA;;AAAA;AAAA;AAAA;AAAA;AAAA,gCAKwB,iCAAiBC,MAAM,CAACF,MAAxB,EAAgCE,MAAM,CAACC,MAAvC,CALxB,EAKEmB,UALF,qBAKEA,UALF,EAKcjB,MALd,qBAKcA,MALd;AAMLH,YAAAA,MAAM,CAACC,MAAP,IAAiBE,MAAjB;AAEMkB,YAAAA,QARD,GAQY,8BAAcC,uBAAd,EAAwBF,UAAU,CAACG,IAAnC,CARZ;AAAA,2BAUGF,QAVH;AAAA,8CAWE,WAXF,wBAcE,cAdF,yBAiBE,iBAjBF;AAAA;;AAAA;AAAA;AAAA,mBAYYG,cAAc,CAACxB,MAAD,EAASoB,UAAT,EAAqBrB,OAArB,CAZ1B;;AAAA;AAYDe,YAAAA,IAZC;AAAA;;AAAA;AAAA;AAAA,mBAeYW,gBAAgB,CAACzB,MAAD,EAASoB,UAAT,EAAqBrB,OAArB,CAf5B;;AAAA;AAeDe,YAAAA,IAfC;AAAA;;AAAA;AAAA;AAAA,mBAmBmBY,oBAAoB,CAAC1B,MAAD,EAASoB,UAAT,EAAqBrB,OAArB,CAnBvC;;AAAA;AAAA;AAAA,2BAoBCqB,UApBD;AAkBDN,YAAAA,IAlBC;AAmBCJ,cAAAA,UAnBD;AAoBCU,cAAAA,UApBD;AAAA;AAAA;;AAAA;AAAA,kBAwBK,IAAIO,KAAJ,8BAAgCN,QAAhC,EAxBL;;AAAA;AAAA,8CA2BEP,IA3BF;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,G;;;;AAwCA,SAASc,YAAT,CACLC,cADK,EAEL5B,MAFK,EAGL6B,GAHK,EAQL;AACA,MAAMC,MAAwB,GAAG,EAAjC;AACA,MAAIC,IAAI,GAAG/B,MAAX;;AACA,OAAK,IAAIgC,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGH,GAApB,EAAyBG,CAAC,EAA1B,EAA8B;AAC5B,QAAMC,aAAa,GAAGL,cAAc,CAACG,IAAD,CAApC;AAEA,QAAMG,cAAc,GAClBH,IAAI,GAAG,CAAP,GAAW,8BAAcI,kCAAd,EAAmCF,aAAa,CAACG,eAAjD,CAAX,GAAgF,MADlF;AAGA,QAAIC,QAAQ,GAAG,KAAf;AACA,QAAIC,QAAQ,GAAG,KAAf;;AACA,YAAQJ,cAAR;AACE,WAAK,UAAL;AACE;;AACF,WAAK,UAAL;AACEG,QAAAA,QAAQ,GAAG,IAAX;AACA;;AACF,WAAK,UAAL;AACEC,QAAAA,QAAQ,GAAG,IAAX;AACA;;AACF;AACE,cAAM,IAAIZ,KAAJ,CAAU,kCAAV,CAAN;AAVJ;;AAaA,QAAIO,aAAa,CAACM,YAAd,GAA8B,CAAlC,EAAqC;AACnC,UAAMC,GAAG,GAAGb,YAAY,CAACC,cAAD,EAAiBG,IAAI,GAAG,CAAxB,EAA2BE,aAAa,CAACM,YAAzC,CAAxB;AACAR,MAAAA,IAAI,GAAGS,GAAG,CAACT,IAAX;AACAD,MAAAA,MAAM,CAACG,aAAa,CAACQ,IAAf,CAAN,GAA6B;AAE3BJ,QAAAA,QAAQ,EAARA,QAF2B;AAG3BC,QAAAA,QAAQ,EAARA,QAH2B;AAI3BI,QAAAA,MAAM,EAAEF,GAAG,CAACV;AAJe,OAA7B;AAMD,KATD,MASO;AACL,UAAMR,IAAI,GAAG,8BAAcqB,mBAAd,EAAoBV,aAAa,CAACX,IAAlC,CAAb;AACA,UAAIsB,WAAW,GAAGtB,IAAlB;;AAEA,UAAIW,aAAa,CAACY,cAAlB,EAAkC;AAChCD,QAAAA,WAAW,GAAG,8BAAcE,4BAAd,EAA6Bb,aAAa,CAACY,cAA3C,CAAd;AACD;;AAED,cAAQD,WAAR;AACE,aAAK,SAAL;AACEA,UAAAA,WAAW,aAAMA,WAAN,cAAqBtB,IAArB,CAAX;AACA;;AACF;AAJF;;AAOAQ,MAAAA,MAAM,CAACG,aAAa,CAACQ,IAAf,CAAN,GAA6B;AAC3BnB,QAAAA,IAAI,EAAEsB,WADqB;AAE3BG,QAAAA,UAAU,EAAEd,aAAa,CAACe,WAFC;AAG3BC,QAAAA,SAAS,EAAEhB,aAAa,CAACiB,SAHE;AAI3BC,QAAAA,KAAK,EAAElB,aAAa,CAACkB,KAJM;AAK3Bd,QAAAA,QAAQ,EAARA,QAL2B;AAM3BC,QAAAA,QAAQ,EAARA;AAN2B,OAA7B;AAQAP,MAAAA,IAAI;AACL;AACF;;AACD,SAAO;AAACD,IAAAA,MAAM,EAANA,MAAD;AAAS9B,IAAAA,MAAM,EAANA,MAAT;AAAiB+B,IAAAA,IAAI,EAAJA;AAAjB,GAAP;AACD;;AAKD,SAASqB,YAAT,CACE9B,IADF,EAEE+B,QAFF,EAGEtD,MAHF,EAIES,KAJF,EAKE8C,IALF,EAMS;AACP,MAAI,EAAED,QAAQ,IAAIE,sBAAd,CAAJ,EAAmC;AACjC,UAAM,IAAI7B,KAAJ,6BAA+B2B,QAA/B,EAAN;AACD;;AACD,SAAOE,uBAAeF,QAAf,EAAyBD,YAAzB,CAAsC9B,IAAtC,EAA4CvB,MAA5C,EAAoDS,KAApD,EAA2D8C,IAA3D,CAAP;AACD;;SAQc/B,c;;;;;8EAAf,kBACExB,MADF,EAEEyD,MAFF,EAGE1D,OAHF;AAAA;;AAAA;;AAAA;AAAA;AAAA;AAAA;AAKQ2D,YAAAA,SALR,GAKoB1D,MAAM,CAACC,MAAP,GAAgBwD,MAAM,CAACE,oBAL3C;AAMQC,YAAAA,UANR,4BAMqBH,MAAM,CAACI,gBAN5B,0DAMqB,sBAAyBC,UAN9C;AASMC,YAAAA,UATN,GASmB/D,MATnB;;AAAA,kBAWMD,OAAO,CAACiE,WAAR,KAAwB,cAX9B;AAAA;AAAA;AAAA;;AAAA;AAAA,mBAY4B,6BACtBjE,OAAO,CAACiE,WADc,EAEtBhE,MAAM,CAACF,MAAP,CAAcmE,KAAd,CAAoBjE,MAAM,CAACC,MAA3B,EAAmCyD,SAAnC,CAFsB,EAGtBD,MAAM,CAACS,sBAHe,CAZ5B;;AAAA;AAYUC,YAAAA,SAZV;AAiBIJ,YAAAA,UAAU,GAAG;AACXjE,cAAAA,MAAM,EAAEqE,SADG;AAEXlE,cAAAA,MAAM,EAAE,CAFG;AAGXC,cAAAA,IAAI,EAAEiE,SAAS,CAAChE;AAHL,aAAb;AAKAH,YAAAA,MAAM,CAACC,MAAP,GAAgByD,SAAhB;;AAtBJ;AA0BQU,YAAAA,cA1BR,GA0ByB,8BACrBC,uBADqB,4BAErBZ,MAAM,CAACI,gBAFc,2DAErB,uBAAyBS,yBAFJ,CA1BzB;AA+BMC,YAAAA,OA/BN,GA+BgB,IAAIC,KAAJ,CAAUZ,UAAV,CA/BhB;;AAiCE,gBAAI7D,OAAO,CAAC0E,MAAR,CAAeC,SAAf,GAA2B,CAA/B,EAAkC;AAChCH,cAAAA,OAAO,GAAGlB,YAAY,CAACsB,6BAAD,EAAqBP,cAArB,EAAqCL,UAArC,EAAiDH,UAAjD,EAA8D;AAClFgB,gBAAAA,QAAQ,EAAE,4BAAY7E,OAAO,CAAC0E,MAAR,CAAeC,SAA3B,CADwE;AAElFG,gBAAAA,eAAe,EAAE;AAFiE,eAA9D,CAAtB;AAKD,aAND,MAMO;AACLN,cAAAA,OAAO,CAACO,IAAR,CAAa,CAAb;AACD;;AAGKC,YAAAA,cA5CR,GA4CyB,8BACrBV,uBADqB,4BAErBZ,MAAM,CAACI,gBAFc,2DAErB,uBAAyBmB,yBAFJ,CA5CzB;AAiDMC,YAAAA,OAjDN,GAiDgB,IAAIT,KAAJ,CAAUZ,UAAV,CAjDhB;;AAkDE,gBAAI7D,OAAO,CAAC0E,MAAR,CAAeS,SAAf,GAA2B,CAA/B,EAAkC;AAChCD,cAAAA,OAAO,GAAG5B,YAAY,CAACsB,6BAAD,EAAqBI,cAArB,EAAqChB,UAArC,EAAiDH,UAAjD,EAA8D;AAClFgB,gBAAAA,QAAQ,EAAE,4BAAY7E,OAAO,CAAC0E,MAAR,CAAeS,SAA3B,CADwE;AAElFL,gBAAAA,eAAe,EAAE;AAFiE,eAA9D,CAAtB;AAKD,aAND,MAMO;AACLI,cAAAA,OAAO,CAACH,IAAR,CAAa,CAAb;AACD;;AACGK,YAAAA,iBA3DN,GA2D0B,CA3D1B;AAAA,mDA4DqBF,OA5DrB;;AAAA;AA4DE,kEAA4B;AAAjBG,gBAAAA,IAAiB;;AAC1B,oBAAIA,IAAI,KAAKrF,OAAO,CAAC0E,MAAR,CAAeS,SAA5B,EAAuC;AACrCC,kBAAAA,iBAAiB;AAClB;AACF;AAhEH;AAAA;AAAA;AAAA;AAAA;;AAmEQE,YAAAA,aAnER,GAmEwB,8BAAchB,uBAAd,4BAAwBZ,MAAM,CAACI,gBAA/B,2DAAwB,uBAAyBP,QAAjD,CAnExB;AAoEQgC,YAAAA,aApER,GAoEwB;AACpBtC,cAAAA,UAAU,EAAEjD,OAAO,CAAC0E,MAAR,CAAezB,UADP;AAEpB4B,cAAAA,QAAQ,EAAE7E,OAAO,CAAC0E,MAAR,CAAezB;AAFL,aApExB;AAyEQzC,YAAAA,MAzER,GAyEiB8C,YAAY,CACzBtD,OAAO,CAAC0E,MAAR,CAAec,aADU,EAEzBF,aAFyB,EAGzBtB,UAHyB,EAIzBoB,iBAJyB,EAKzBG,aALyB,CAzE7B;AAAA,8CAiFS;AACLhF,cAAAA,OAAO,EAAE2E,OADJ;AAEL5E,cAAAA,OAAO,EAAEkE,OAFJ;AAGLhE,cAAAA,MAAM,EAANA,MAHK;AAILE,cAAAA,KAAK,EAAEmD,UAJF;AAKLxC,cAAAA,UAAU,EAAEqC;AALP,aAjFT;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,G;;;;SAiGehC,gB;;;;;+EAAf,kBACEzB,MADF,EAEEyD,MAFF,EAGEF,IAHF;AAAA;;AAAA;AAAA;AAAA;AAAA;AAAA;AAKQG,YAAAA,SALR,GAKoB1D,MAAM,CAACC,MAAP,GAAgBwD,MAAM,CAACE,oBAL3C;AAOQC,YAAAA,UAPR,6BAOqBH,MAAM,CAAC+B,mBAP5B,2DAOqB,uBAA4B1B,UAPjD;AASQqB,YAAAA,iBATR,GAS4BvB,UAAU,8BAAGH,MAAM,CAAC+B,mBAAV,2DAAG,uBAA4BC,SAA/B,CATtC;AAUQJ,YAAAA,aAVR,GAUwB,8BACpBhB,uBADoB,4BAEpBZ,MAAM,CAAC+B,mBAFa,2DAEpB,uBAA4BlC,QAFR,CAVxB;AAiBMiB,YAAAA,OAjBN,GAiBgB,IAAIC,KAAJ,CAAUZ,UAAV,CAjBhB;;AAkBE,gBAAIL,IAAI,CAACkB,MAAL,CAAYC,SAAZ,GAAwB,CAA5B,EAA+B;AAC7BH,cAAAA,OAAO,GAAGlB,YAAY,CAACsB,6BAAD,EAAqBe,iCAArB,EAA6C1F,MAA7C,EAAqD4D,UAArD,EAAkE;AACtFgB,gBAAAA,QAAQ,EAAE,4BAAYrB,IAAI,CAACkB,MAAL,CAAYC,SAAxB,CAD4E;AAEtFG,gBAAAA,eAAe,EAAE;AAFqE,eAAlE,CAAtB;AAID,aALD,MAKO;AACLN,cAAAA,OAAO,CAACO,IAAR,CAAa,CAAb;AACD;;AAIGG,YAAAA,OA7BN,GA6BgB,IAAIT,KAAJ,CAAUZ,UAAV,CA7BhB;;AA8BE,gBAAIL,IAAI,CAACkB,MAAL,CAAYS,SAAZ,GAAwB,CAA5B,EAA+B;AAC7BD,cAAAA,OAAO,GAAG5B,YAAY,CAACsB,6BAAD,EAAqBe,iCAArB,EAA6C1F,MAA7C,EAAqD4D,UAArD,EAAkE;AACtFgB,gBAAAA,QAAQ,EAAE,4BAAYrB,IAAI,CAACkB,MAAL,CAAYS,SAAxB,CAD4E;AAEtFL,gBAAAA,eAAe,EAAE;AAFqE,eAAlE,CAAtB;AAID,aALD,MAKO;AACLI,cAAAA,OAAO,CAACH,IAAR,CAAa,CAAb;AACD;;AAGGa,YAAAA,eAxCN,GAwCwB3F,MAxCxB;;AAAA,4CA0CMyD,MAAM,CAAC+B,mBA1Cb,mDA0CM,uBAA4BI,aA1ClC;AAAA;AAAA;AAAA;;AAAA;AAAA,mBA2C4B,6BACtBrC,IAAI,CAACS,WADiB,EAEtBhE,MAAM,CAACF,MAAP,CAAcmE,KAAd,CAAoBjE,MAAM,CAACC,MAA3B,EAAmCyD,SAAnC,CAFsB,EAGtBD,MAAM,CAACS,sBAHe,CA3C5B;;AAAA;AA2CUC,YAAAA,SA3CV;AAiDIwB,YAAAA,eAAe,GAAG;AAChB7F,cAAAA,MAAM,EAAEqE,SADQ;AAEhBlE,cAAAA,MAAM,EAAE,CAFQ;AAGhBC,cAAAA,IAAI,EAAEiE,SAAS,CAAChE;AAHA,aAAlB;AAMAH,YAAAA,MAAM,CAACC,MAAP,GAAgByD,SAAhB;;AAvDJ;AA0DQ4B,YAAAA,aA1DR,GA0DwB;AACpBtC,cAAAA,UAAU,EAAEO,IAAI,CAACkB,MAAL,CAAYzB,UADJ;AAEpB4B,cAAAA,QAAQ,EAAErB,IAAI,CAACkB,MAAL,CAAYzB;AAFF,aA1DxB;AA+DQzC,YAAAA,MA/DR,GA+DiB8C,YAAY,CACzBE,IAAI,CAACkB,MAAL,CAAYc,aADa,EAEzBF,aAFyB,EAGzBM,eAHyB,EAIzBR,iBAJyB,EAKzBG,aALyB,CA/D7B;AAAA,8CAuES;AACLhF,cAAAA,OAAO,EAAE2E,OADJ;AAEL5E,cAAAA,OAAO,EAAEkE,OAFJ;AAGLhE,cAAAA,MAAM,EAANA,MAHK;AAILE,cAAAA,KAAK,EAAEmD,UAJF;AAKLxC,cAAAA,UAAU,EAAEqC;AALP,aAvET;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,G;;;;SAsFe/B,oB;;;;;oFAAf,kBACE1B,MADF,EAEEoB,UAFF,EAGErB,OAHF;AAAA;;AAAA;AAAA;AAAA;AAAA;AAAA;AAKQ2D,YAAAA,SALR,GAKoB1D,MAAM,CAACC,MAAP,GAAgBmB,UAAU,CAACuC,oBAL/C;AAOMkC,YAAAA,UAPN,GAOmB;AACf5F,cAAAA,MAAM,EAAE,CADO;AAEfH,cAAAA,MAAM,EAAEE,MAAM,CAACF,MAAP,CAAcmE,KAAd,CAAoBjE,MAAM,CAACC,MAA3B,EAAmCyD,SAAnC,CAFO;AAGfxD,cAAAA,IAAI,EAAEwD,SAAS,GAAG1D,MAAM,CAACC;AAHV,aAPnB;AAaED,YAAAA,MAAM,CAACC,MAAP,GAAgByD,SAAhB;;AAbF,kBAeM3D,OAAO,CAACiE,WAAR,KAAwB,cAf9B;AAAA;AAAA;AAAA;;AAAA;AAAA,mBAgB4B,6BACtBjE,OAAO,CAACiE,WADc,EAEtB6B,UAAU,CAAC/F,MAAX,CAAkBmE,KAAlB,CAAwB4B,UAAU,CAAC5F,MAAnC,EAA2CyD,SAA3C,CAFsB,EAGtBtC,UAAU,CAAC8C,sBAHW,CAhB5B;;AAAA;AAgBUC,YAAAA,SAhBV;AAsBI0B,YAAAA,UAAU,GAAG;AACX/F,cAAAA,MAAM,EAAEqE,SADG;AAEXlE,cAAAA,MAAM,EAAE,CAFG;AAGXC,cAAAA,IAAI,EAAEiE,SAAS,CAAChE;AAHL,aAAb;AAMAH,YAAAA,MAAM,CAACC,MAAP,GAAgByD,SAAhB;;AA5BJ;AA+BQ/C,YAAAA,SA/BR,GA+BoB,CAAAS,UAAU,SAAV,IAAAA,UAAU,WAAV,qCAAAA,UAAU,CAAE0E,sBAAZ,gFAAoChC,UAApC,KAAkD,CA/BtE;AAAA,8CAiCST,YAAY,CACjBtD,OAAO,CAAC0E,MAAR,CAAec,aADE,EAEjBxF,OAAO,CAAC0E,MAAR,CAAenB,QAFE,EAGjBuC,UAHiB,EAIjBlF,SAJiB,EAKjBZ,OALiB,CAAZ,CAMLgB,GANK,CAMD,UAACgF,CAAD;AAAA,qBAAOA,CAAC,CAACC,QAAF,EAAP;AAAA,aANC,CAjCT;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,G","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\nimport {\n ParquetCodec,\n ParquetData,\n ParquetOptions,\n ParquetPageData,\n ParquetType,\n PrimitiveType,\n SchemaDefinition\n} from '../schema/declare';\nimport {CursorBuffer, ParquetCodecOptions, PARQUET_CODECS} from '../codecs';\nimport {\n ConvertedType,\n Encoding,\n FieldRepetitionType,\n PageHeader,\n PageType,\n SchemaElement,\n Type\n} from '../parquet-thrift';\nimport {decompress} from '../compression';\nimport {PARQUET_RDLVL_TYPE, PARQUET_RDLVL_ENCODING} from '../../constants';\nimport {decodePageHeader, getThriftEnum, getBitWidth} from '../utils/read-utils';\n\n/**\n * Decode data pages\n * @param buffer - input data\n * @param column - parquet column\n * @param compression - compression type\n * @returns parquet data page data\n */\nexport async function decodeDataPages(\n buffer: Buffer,\n options: ParquetOptions\n): Promise<ParquetData> {\n const cursor: CursorBuffer = {\n buffer,\n offset: 0,\n size: buffer.length\n };\n\n const data: ParquetData = {\n rlevels: [],\n dlevels: [],\n values: [],\n pageHeaders: [],\n count: 0\n };\n\n let dictionary = options.dictionary || [];\n\n while (\n // @ts-ignore size can be undefined\n cursor.offset < cursor.size &&\n (!options.numValues || data.dlevels.length < Number(options.numValues))\n ) {\n // Looks like we have to decode these in sequence due to cursor updates?\n const page = await decodePage(cursor, options);\n\n if (page.dictionary) {\n dictionary = page.dictionary;\n // eslint-disable-next-line no-continue\n continue;\n }\n\n if (dictionary.length) {\n // eslint-disable-next-line no-loop-func\n page.values = page.values.map((value) => dictionary[value]);\n }\n\n for (let index = 0; index < page.rlevels.length; index++) {\n data.rlevels.push(page.rlevels[index]);\n data.dlevels.push(page.dlevels[index]);\n const value = page.values[index];\n\n if (value !== undefined) {\n data.values.push(value);\n }\n }\n\n data.count += page.count;\n data.pageHeaders.push(page.pageHeader);\n }\n\n return data;\n}\n\n/**\n * Decode parquet page based on page type\n * @param cursor\n * @param options\n */\nexport async function decodePage(\n cursor: CursorBuffer,\n options: ParquetOptions\n): Promise<ParquetPageData> {\n let page;\n const {pageHeader, length} = decodePageHeader(cursor.buffer, cursor.offset);\n cursor.offset += length;\n\n const pageType = getThriftEnum(PageType, pageHeader.type);\n\n switch (pageType) {\n case 'DATA_PAGE':\n page = await decodeDataPage(cursor, pageHeader, options);\n break;\n case 'DATA_PAGE_V2':\n page = await decodeDataPageV2(cursor, pageHeader, options);\n break;\n case 'DICTIONARY_PAGE':\n page = {\n dictionary: await decodeDictionaryPage(cursor, pageHeader, options),\n pageHeader\n };\n break;\n default:\n throw new Error(`invalid page type: ${pageType}`);\n }\n\n return page;\n}\n\n/**\n * Decode parquet schema\n * @param schemaElements input schema elements data\n * @param offset offset to read from\n * @param len length of data\n * @returns result.offset\n * result.next - offset at the end of function\n * result.schema - schema read from the input data\n * @todo output offset is the same as input - possibly excess output field\n */\nexport function decodeSchema(\n schemaElements: SchemaElement[],\n offset: number,\n len: number\n): {\n offset: number;\n next: number;\n schema: SchemaDefinition;\n} {\n const schema: SchemaDefinition = {};\n let next = offset;\n for (let i = 0; i < len; i++) {\n const schemaElement = schemaElements[next];\n\n const repetitionType =\n next > 0 ? getThriftEnum(FieldRepetitionType, schemaElement.repetition_type!) : 'ROOT';\n\n let optional = false;\n let repeated = false;\n switch (repetitionType) {\n case 'REQUIRED':\n break;\n case 'OPTIONAL':\n optional = true;\n break;\n case 'REPEATED':\n repeated = true;\n break;\n default:\n throw new Error('parquet: unknown repetition type');\n }\n\n if (schemaElement.num_children! > 0) {\n const res = decodeSchema(schemaElements, next + 1, schemaElement.num_children!);\n next = res.next;\n schema[schemaElement.name] = {\n // type: undefined,\n optional,\n repeated,\n fields: res.schema\n };\n } else {\n const type = getThriftEnum(Type, schemaElement.type!);\n let logicalType = type;\n\n if (schemaElement.converted_type) {\n logicalType = getThriftEnum(ConvertedType, schemaElement.converted_type);\n }\n\n switch (logicalType) {\n case 'DECIMAL':\n logicalType = `${logicalType}_${type}` as ParquetType;\n break;\n default:\n }\n\n schema[schemaElement.name] = {\n type: logicalType as ParquetType,\n typeLength: schemaElement.type_length,\n presision: schemaElement.precision,\n scale: schemaElement.scale,\n optional,\n repeated\n };\n next++;\n }\n }\n return {schema, offset, next};\n}\n\n/**\n * Decode a consecutive array of data using one of the parquet encodings\n */\nfunction decodeValues(\n type: PrimitiveType,\n encoding: ParquetCodec,\n cursor: CursorBuffer,\n count: number,\n opts: ParquetCodecOptions\n): any[] {\n if (!(encoding in PARQUET_CODECS)) {\n throw new Error(`invalid encoding: ${encoding}`);\n }\n return PARQUET_CODECS[encoding].decodeValues(type, cursor, count, opts);\n}\n\n/**\n * Do decoding of parquet dataPage from column chunk\n * @param cursor\n * @param header\n * @param options\n */\nasync function decodeDataPage(\n cursor: CursorBuffer,\n header: PageHeader,\n options: ParquetOptions\n): Promise<ParquetPageData> {\n const cursorEnd = cursor.offset + header.compressed_page_size;\n const valueCount = header.data_page_header?.num_values;\n\n /* uncompress page */\n let dataCursor = cursor;\n\n if (options.compression !== 'UNCOMPRESSED') {\n const valuesBuf = await decompress(\n options.compression,\n cursor.buffer.slice(cursor.offset, cursorEnd),\n header.uncompressed_page_size\n );\n dataCursor = {\n buffer: valuesBuf,\n offset: 0,\n size: valuesBuf.length\n };\n cursor.offset = cursorEnd;\n }\n\n /* read repetition levels */\n const rLevelEncoding = getThriftEnum(\n Encoding,\n header.data_page_header?.repetition_level_encoding!\n ) as ParquetCodec;\n // tslint:disable-next-line:prefer-array-literal\n let rLevels = new Array(valueCount);\n\n if (options.column.rLevelMax > 0) {\n rLevels = decodeValues(PARQUET_RDLVL_TYPE, rLevelEncoding, dataCursor, valueCount!, {\n bitWidth: getBitWidth(options.column.rLevelMax),\n disableEnvelope: false\n // column: opts.column\n });\n } else {\n rLevels.fill(0);\n }\n\n /* read definition levels */\n const dLevelEncoding = getThriftEnum(\n Encoding,\n header.data_page_header?.definition_level_encoding!\n ) as ParquetCodec;\n // tslint:disable-next-line:prefer-array-literal\n let dLevels = new Array(valueCount);\n if (options.column.dLevelMax > 0) {\n dLevels = decodeValues(PARQUET_RDLVL_TYPE, dLevelEncoding, dataCursor, valueCount!, {\n bitWidth: getBitWidth(options.column.dLevelMax),\n disableEnvelope: false\n // column: opts.column\n });\n } else {\n dLevels.fill(0);\n }\n let valueCountNonNull = 0;\n for (const dlvl of dLevels) {\n if (dlvl === options.column.dLevelMax) {\n valueCountNonNull++;\n }\n }\n\n /* read values */\n const valueEncoding = getThriftEnum(Encoding, header.data_page_header?.encoding!) as ParquetCodec;\n const decodeOptions = {\n typeLength: options.column.typeLength,\n bitWidth: options.column.typeLength\n };\n\n const values = decodeValues(\n options.column.primitiveType!,\n valueEncoding,\n dataCursor,\n valueCountNonNull,\n decodeOptions\n );\n\n return {\n dlevels: dLevels,\n rlevels: rLevels,\n values,\n count: valueCount!,\n pageHeader: header\n };\n}\n\n/**\n * Do decoding of parquet dataPage in version 2 from column chunk\n * @param cursor\n * @param header\n * @param opts\n * @returns\n */\nasync function decodeDataPageV2(\n cursor: CursorBuffer,\n header: PageHeader,\n opts: any\n): Promise<ParquetPageData> {\n const cursorEnd = cursor.offset + header.compressed_page_size;\n\n const valueCount = header.data_page_header_v2?.num_values;\n // @ts-ignore\n const valueCountNonNull = valueCount - header.data_page_header_v2?.num_nulls;\n const valueEncoding = getThriftEnum(\n Encoding,\n header.data_page_header_v2?.encoding!\n ) as ParquetCodec;\n\n /* read repetition levels */\n // tslint:disable-next-line:prefer-array-literal\n let rLevels = new Array(valueCount);\n if (opts.column.rLevelMax > 0) {\n rLevels = decodeValues(PARQUET_RDLVL_TYPE, PARQUET_RDLVL_ENCODING, cursor, valueCount!, {\n bitWidth: getBitWidth(opts.column.rLevelMax),\n disableEnvelope: true\n });\n } else {\n rLevels.fill(0);\n }\n\n /* read definition levels */\n // tslint:disable-next-line:prefer-array-literal\n let dLevels = new Array(valueCount);\n if (opts.column.dLevelMax > 0) {\n dLevels = decodeValues(PARQUET_RDLVL_TYPE, PARQUET_RDLVL_ENCODING, cursor, valueCount!, {\n bitWidth: getBitWidth(opts.column.dLevelMax),\n disableEnvelope: true\n });\n } else {\n dLevels.fill(0);\n }\n\n /* read values */\n let valuesBufCursor = cursor;\n\n if (header.data_page_header_v2?.is_compressed) {\n const valuesBuf = await decompress(\n opts.compression,\n cursor.buffer.slice(cursor.offset, cursorEnd),\n header.uncompressed_page_size\n );\n\n valuesBufCursor = {\n buffer: valuesBuf,\n offset: 0,\n size: valuesBuf.length\n };\n\n cursor.offset = cursorEnd;\n }\n\n const decodeOptions = {\n typeLength: opts.column.typeLength,\n bitWidth: opts.column.typeLength\n };\n\n const values = decodeValues(\n opts.column.primitiveType!,\n valueEncoding,\n valuesBufCursor,\n valueCountNonNull,\n decodeOptions\n );\n\n return {\n dlevels: dLevels,\n rlevels: rLevels,\n values,\n count: valueCount!,\n pageHeader: header\n };\n}\n\n/**\n * Do decoding of dictionary page which helps to iterate over all indexes and get dataPage values.\n * @param cursor\n * @param pageHeader\n * @param options\n */\nasync function decodeDictionaryPage(\n cursor: CursorBuffer,\n pageHeader: PageHeader,\n options: ParquetOptions\n): Promise<string[]> {\n const cursorEnd = cursor.offset + pageHeader.compressed_page_size;\n\n let dictCursor = {\n offset: 0,\n buffer: cursor.buffer.slice(cursor.offset, cursorEnd),\n size: cursorEnd - cursor.offset\n };\n\n cursor.offset = cursorEnd;\n\n if (options.compression !== 'UNCOMPRESSED') {\n const valuesBuf = await decompress(\n options.compression,\n dictCursor.buffer.slice(dictCursor.offset, cursorEnd),\n pageHeader.uncompressed_page_size\n );\n\n dictCursor = {\n buffer: valuesBuf,\n offset: 0,\n size: valuesBuf.length\n };\n\n cursor.offset = cursorEnd;\n }\n\n const numValues = pageHeader?.dictionary_page_header?.num_values || 0;\n\n return decodeValues(\n options.column.primitiveType!,\n options.column.encoding!,\n dictCursor,\n numValues,\n options as ParquetCodecOptions\n ).map((d) => d.toString());\n}\n"],"file":"decoders.js"}
@@ -7,16 +7,25 @@ Object.defineProperty(exports, "__esModule", {
7
7
  });
8
8
  exports.ParquetCursor = void 0;
9
9
 
10
+ var _regenerator = _interopRequireDefault(require("@babel/runtime/regenerator"));
11
+
12
+ var _asyncToGenerator2 = _interopRequireDefault(require("@babel/runtime/helpers/asyncToGenerator"));
13
+
14
+ var _classCallCheck2 = _interopRequireDefault(require("@babel/runtime/helpers/classCallCheck"));
15
+
16
+ var _createClass2 = _interopRequireDefault(require("@babel/runtime/helpers/createClass"));
17
+
10
18
  var _defineProperty2 = _interopRequireDefault(require("@babel/runtime/helpers/defineProperty"));
11
19
 
12
20
  var _shred = require("../schema/shred");
13
21
 
14
- let _Symbol$asyncIterator;
22
+ var _Symbol$asyncIterator;
15
23
 
16
24
  _Symbol$asyncIterator = Symbol.asyncIterator;
17
25
 
18
- class ParquetCursor {
19
- constructor(metadata, envelopeReader, schema, columnList) {
26
+ var ParquetCursor = function () {
27
+ function ParquetCursor(metadata, envelopeReader, schema, columnList) {
28
+ (0, _classCallCheck2.default)(this, ParquetCursor);
20
29
  (0, _defineProperty2.default)(this, "metadata", void 0);
21
30
  (0, _defineProperty2.default)(this, "envelopeReader", void 0);
22
31
  (0, _defineProperty2.default)(this, "schema", void 0);
@@ -30,68 +39,177 @@ class ParquetCursor {
30
39
  this.rowGroupIndex = 0;
31
40
  }
32
41
 
33
- async next() {
34
- if (this.rowGroup.length === 0) {
35
- if (this.rowGroupIndex >= this.metadata.row_groups.length) {
36
- return null;
42
+ (0, _createClass2.default)(ParquetCursor, [{
43
+ key: "next",
44
+ value: function () {
45
+ var _next = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee() {
46
+ var rowBuffer;
47
+ return _regenerator.default.wrap(function _callee$(_context) {
48
+ while (1) {
49
+ switch (_context.prev = _context.next) {
50
+ case 0:
51
+ if (!(this.rowGroup.length === 0)) {
52
+ _context.next = 8;
53
+ break;
54
+ }
55
+
56
+ if (!(this.rowGroupIndex >= this.metadata.row_groups.length)) {
57
+ _context.next = 3;
58
+ break;
59
+ }
60
+
61
+ return _context.abrupt("return", null);
62
+
63
+ case 3:
64
+ _context.next = 5;
65
+ return this.envelopeReader.readRowGroup(this.schema, this.metadata.row_groups[this.rowGroupIndex], this.columnList);
66
+
67
+ case 5:
68
+ rowBuffer = _context.sent;
69
+ this.rowGroup = (0, _shred.materializeRecords)(this.schema, rowBuffer);
70
+ this.rowGroupIndex++;
71
+
72
+ case 8:
73
+ return _context.abrupt("return", this.rowGroup.shift());
74
+
75
+ case 9:
76
+ case "end":
77
+ return _context.stop();
78
+ }
79
+ }
80
+ }, _callee, this);
81
+ }));
82
+
83
+ function next() {
84
+ return _next.apply(this, arguments);
37
85
  }
38
86
 
39
- const rowBuffer = await this.envelopeReader.readRowGroup(this.schema, this.metadata.row_groups[this.rowGroupIndex], this.columnList);
40
- this.rowGroup = (0, _shred.materializeRecords)(this.schema, rowBuffer);
41
- this.rowGroupIndex++;
87
+ return next;
88
+ }()
89
+ }, {
90
+ key: "rewind",
91
+ value: function rewind() {
92
+ this.rowGroup = [];
93
+ this.rowGroupIndex = 0;
42
94
  }
43
-
44
- return this.rowGroup.shift();
45
- }
46
-
47
- rewind() {
48
- this.rowGroup = [];
49
- this.rowGroupIndex = 0;
50
- }
51
-
52
- [_Symbol$asyncIterator]() {
53
- let done = false;
54
- return {
55
- next: async () => {
56
- if (done) {
57
- return {
58
- done,
59
- value: null
60
- };
61
- }
62
-
63
- const value = await this.next();
64
-
65
- if (value === null) {
66
- return {
67
- done: true,
68
- value
69
- };
70
- }
71
-
72
- return {
73
- done: false,
74
- value
75
- };
76
- },
77
- return: async () => {
78
- done = true;
79
- return {
80
- done,
81
- value: null
82
- };
83
- },
84
- throw: async () => {
85
- done = true;
86
- return {
87
- done: true,
88
- value: null
89
- };
90
- }
91
- };
92
- }
93
-
94
- }
95
+ }, {
96
+ key: _Symbol$asyncIterator,
97
+ value: function value() {
98
+ var _this = this;
99
+
100
+ var done = false;
101
+ return {
102
+ next: function () {
103
+ var _next2 = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee2() {
104
+ var value;
105
+ return _regenerator.default.wrap(function _callee2$(_context2) {
106
+ while (1) {
107
+ switch (_context2.prev = _context2.next) {
108
+ case 0:
109
+ if (!done) {
110
+ _context2.next = 2;
111
+ break;
112
+ }
113
+
114
+ return _context2.abrupt("return", {
115
+ done: done,
116
+ value: null
117
+ });
118
+
119
+ case 2:
120
+ _context2.next = 4;
121
+ return _this.next();
122
+
123
+ case 4:
124
+ value = _context2.sent;
125
+
126
+ if (!(value === null)) {
127
+ _context2.next = 7;
128
+ break;
129
+ }
130
+
131
+ return _context2.abrupt("return", {
132
+ done: true,
133
+ value: value
134
+ });
135
+
136
+ case 7:
137
+ return _context2.abrupt("return", {
138
+ done: false,
139
+ value: value
140
+ });
141
+
142
+ case 8:
143
+ case "end":
144
+ return _context2.stop();
145
+ }
146
+ }
147
+ }, _callee2);
148
+ }));
149
+
150
+ function next() {
151
+ return _next2.apply(this, arguments);
152
+ }
153
+
154
+ return next;
155
+ }(),
156
+ return: function () {
157
+ var _return2 = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee3() {
158
+ return _regenerator.default.wrap(function _callee3$(_context3) {
159
+ while (1) {
160
+ switch (_context3.prev = _context3.next) {
161
+ case 0:
162
+ done = true;
163
+ return _context3.abrupt("return", {
164
+ done: done,
165
+ value: null
166
+ });
167
+
168
+ case 2:
169
+ case "end":
170
+ return _context3.stop();
171
+ }
172
+ }
173
+ }, _callee3);
174
+ }));
175
+
176
+ function _return() {
177
+ return _return2.apply(this, arguments);
178
+ }
179
+
180
+ return _return;
181
+ }(),
182
+ throw: function () {
183
+ var _throw2 = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee4() {
184
+ return _regenerator.default.wrap(function _callee4$(_context4) {
185
+ while (1) {
186
+ switch (_context4.prev = _context4.next) {
187
+ case 0:
188
+ done = true;
189
+ return _context4.abrupt("return", {
190
+ done: true,
191
+ value: null
192
+ });
193
+
194
+ case 2:
195
+ case "end":
196
+ return _context4.stop();
197
+ }
198
+ }
199
+ }, _callee4);
200
+ }));
201
+
202
+ function _throw() {
203
+ return _throw2.apply(this, arguments);
204
+ }
205
+
206
+ return _throw;
207
+ }()
208
+ };
209
+ }
210
+ }]);
211
+ return ParquetCursor;
212
+ }();
95
213
 
96
214
  exports.ParquetCursor = ParquetCursor;
97
215
  //# sourceMappingURL=parquet-cursor.js.map
@@ -1 +1 @@
1
- {"version":3,"sources":["../../../../src/parquetjs/parser/parquet-cursor.ts"],"names":["Symbol","asyncIterator","ParquetCursor","constructor","metadata","envelopeReader","schema","columnList","rowGroupIndex","next","rowGroup","length","row_groups","rowBuffer","readRowGroup","shift","rewind","done","value","return","throw"],"mappings":";;;;;;;;;;;AAKA;;;;wBAiEGA,MAAM,CAACC,a;;AA5DH,MAAMC,aAAN,CAAmD;AAcxDC,EAAAA,WAAW,CACTC,QADS,EAETC,cAFS,EAGTC,MAHS,EAITC,UAJS,EAKT;AAAA;AAAA;AAAA;AAAA;AAAA,oDAdiC,EAcjC;AAAA;AACA,SAAKH,QAAL,GAAgBA,QAAhB;AACA,SAAKC,cAAL,GAAsBA,cAAtB;AACA,SAAKC,MAAL,GAAcA,MAAd;AACA,SAAKC,UAAL,GAAkBA,UAAlB;AACA,SAAKC,aAAL,GAAqB,CAArB;AACD;;AAMS,QAAJC,IAAI,GAAwB;AAChC,QAAI,KAAKC,QAAL,CAAcC,MAAd,KAAyB,CAA7B,EAAgC;AAC9B,UAAI,KAAKH,aAAL,IAAsB,KAAKJ,QAAL,CAAcQ,UAAd,CAAyBD,MAAnD,EAA2D;AAEzD,eAAO,IAAP;AACD;;AACD,YAAME,SAAS,GAAG,MAAM,KAAKR,cAAL,CAAoBS,YAApB,CACtB,KAAKR,MADiB,EAEtB,KAAKF,QAAL,CAAcQ,UAAd,CAAyB,KAAKJ,aAA9B,CAFsB,EAGtB,KAAKD,UAHiB,CAAxB;AAKA,WAAKG,QAAL,GAAgB,+BAAmB,KAAKJ,MAAxB,EAAgCO,SAAhC,CAAhB;AACA,WAAKL,aAAL;AACD;;AACD,WAAO,KAAKE,QAAL,CAAcK,KAAd,EAAP;AACD;;AAKDC,EAAAA,MAAM,GAAS;AACb,SAAKN,QAAL,GAAgB,EAAhB;AACA,SAAKF,aAAL,GAAqB,CAArB;AACD;;AAMD,4BAA2C;AACzC,QAAIS,IAAI,GAAG,KAAX;AACA,WAAO;AACLR,MAAAA,IAAI,EAAE,YAAY;AAChB,YAAIQ,IAAJ,EAAU;AACR,iBAAO;AAACA,YAAAA,IAAD;AAAOC,YAAAA,KAAK,EAAE;AAAd,WAAP;AACD;;AACD,cAAMA,KAAK,GAAG,MAAM,KAAKT,IAAL,EAApB;;AACA,YAAIS,KAAK,KAAK,IAAd,EAAoB;AAClB,iBAAO;AAACD,YAAAA,IAAI,EAAE,IAAP;AAAaC,YAAAA;AAAb,WAAP;AACD;;AACD,eAAO;AAACD,UAAAA,IAAI,EAAE,KAAP;AAAcC,UAAAA;AAAd,SAAP;AACD,OAVI;AAWLC,MAAAA,MAAM,EAAE,YAAY;AAClBF,QAAAA,IAAI,GAAG,IAAP;AACA,eAAO;AAACA,UAAAA,IAAD;AAAOC,UAAAA,KAAK,EAAE;AAAd,SAAP;AACD,OAdI;AAeLE,MAAAA,KAAK,EAAE,YAAY;AACjBH,QAAAA,IAAI,GAAG,IAAP;AACA,eAAO;AAACA,UAAAA,IAAI,EAAE,IAAP;AAAaC,UAAAA,KAAK,EAAE;AAApB,SAAP;AACD;AAlBI,KAAP;AAoBD;;AAlFuD","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\nimport {FileMetaData} from '../parquet-thrift';\nimport {ParquetEnvelopeReader} from './parquet-envelope-reader';\nimport {ParquetSchema} from '../schema/schema';\nimport {ParquetRecord} from '../schema/declare';\nimport {materializeRecords} from '../schema/shred';\n\n/**\n * A parquet cursor is used to retrieve rows from a parquet file in order\n */\nexport class ParquetCursor<T> implements AsyncIterable<T> {\n public metadata: FileMetaData;\n public envelopeReader: ParquetEnvelopeReader;\n public schema: ParquetSchema;\n public columnList: string[][];\n public rowGroup: ParquetRecord[] = [];\n public rowGroupIndex: number;\n\n /**\n * Create a new parquet reader from the file metadata and an envelope reader.\n * It is usually not recommended to call this constructor directly except for\n * advanced and internal use cases. Consider using getCursor() on the\n * ParquetReader instead\n */\n constructor(\n metadata: FileMetaData,\n envelopeReader: ParquetEnvelopeReader,\n schema: ParquetSchema,\n columnList: string[][]\n ) {\n this.metadata = metadata;\n this.envelopeReader = envelopeReader;\n this.schema = schema;\n this.columnList = columnList;\n this.rowGroupIndex = 0;\n }\n\n /**\n * Retrieve the next row from the cursor. Returns a row or NULL if the end\n * of the file was reached\n */\n async next<T = any>(): Promise<T> {\n if (this.rowGroup.length === 0) {\n if (this.rowGroupIndex >= this.metadata.row_groups.length) {\n // @ts-ignore\n return null;\n }\n const rowBuffer = await this.envelopeReader.readRowGroup(\n this.schema,\n this.metadata.row_groups[this.rowGroupIndex],\n this.columnList\n );\n this.rowGroup = materializeRecords(this.schema, rowBuffer);\n this.rowGroupIndex++;\n }\n return this.rowGroup.shift() as any;\n }\n\n /**\n * Rewind the cursor the the beginning of the file\n */\n rewind(): void {\n this.rowGroup = [];\n this.rowGroupIndex = 0;\n }\n\n /**\n * Implement AsyncIterable\n */\n // tslint:disable-next-line:function-name\n [Symbol.asyncIterator](): AsyncIterator<T> {\n let done = false;\n return {\n next: async () => {\n if (done) {\n return {done, value: null};\n }\n const value = await this.next();\n if (value === null) {\n return {done: true, value};\n }\n return {done: false, value};\n },\n return: async () => {\n done = true;\n return {done, value: null};\n },\n throw: async () => {\n done = true;\n return {done: true, value: null};\n }\n };\n }\n}\n"],"file":"parquet-cursor.js"}
1
+ {"version":3,"sources":["../../../../src/parquetjs/parser/parquet-cursor.ts"],"names":["Symbol","asyncIterator","ParquetCursor","metadata","envelopeReader","schema","columnList","rowGroupIndex","rowGroup","length","row_groups","readRowGroup","rowBuffer","shift","done","next","value","return","throw"],"mappings":";;;;;;;;;;;;;;;;;;;AAKA;;;;wBAiEGA,MAAM,CAACC,a;;IA5DGC,a;AAcX,yBACEC,QADF,EAEEC,cAFF,EAGEC,MAHF,EAIEC,UAJF,EAKE;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,oDAdiC,EAcjC;AAAA;AACA,SAAKH,QAAL,GAAgBA,QAAhB;AACA,SAAKC,cAAL,GAAsBA,cAAtB;AACA,SAAKC,MAAL,GAAcA,MAAd;AACA,SAAKC,UAAL,GAAkBA,UAAlB;AACA,SAAKC,aAAL,GAAqB,CAArB;AACD;;;;;4EAMD;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,sBACM,KAAKC,QAAL,CAAcC,MAAd,KAAyB,CAD/B;AAAA;AAAA;AAAA;;AAAA,sBAEQ,KAAKF,aAAL,IAAsB,KAAKJ,QAAL,CAAcO,UAAd,CAAyBD,MAFvD;AAAA;AAAA;AAAA;;AAAA,iDAIa,IAJb;;AAAA;AAAA;AAAA,uBAM4B,KAAKL,cAAL,CAAoBO,YAApB,CACtB,KAAKN,MADiB,EAEtB,KAAKF,QAAL,CAAcO,UAAd,CAAyB,KAAKH,aAA9B,CAFsB,EAGtB,KAAKD,UAHiB,CAN5B;;AAAA;AAMUM,gBAAAA,SANV;AAWI,qBAAKJ,QAAL,GAAgB,+BAAmB,KAAKH,MAAxB,EAAgCO,SAAhC,CAAhB;AACA,qBAAKL,aAAL;;AAZJ;AAAA,iDAcS,KAAKC,QAAL,CAAcK,KAAd,EAdT;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,O;;;;;;;;;;WAoBA,kBAAe;AACb,WAAKL,QAAL,GAAgB,EAAhB;AACA,WAAKD,aAAL,GAAqB,CAArB;AACD;;;WAMD,iBAA2C;AAAA;;AACzC,UAAIO,IAAI,GAAG,KAAX;AACA,aAAO;AACLC,QAAAA,IAAI;AAAA,iFAAE;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,yBACAD,IADA;AAAA;AAAA;AAAA;;AAAA,sDAEK;AAACA,sBAAAA,IAAI,EAAJA,IAAD;AAAOE,sBAAAA,KAAK,EAAE;AAAd,qBAFL;;AAAA;AAAA;AAAA,2BAIgB,KAAI,CAACD,IAAL,EAJhB;;AAAA;AAIEC,oBAAAA,KAJF;;AAAA,0BAKAA,KAAK,KAAK,IALV;AAAA;AAAA;AAAA;;AAAA,sDAMK;AAACF,sBAAAA,IAAI,EAAE,IAAP;AAAaE,sBAAAA,KAAK,EAALA;AAAb,qBANL;;AAAA;AAAA,sDAQG;AAACF,sBAAAA,IAAI,EAAE,KAAP;AAAcE,sBAAAA,KAAK,EAALA;AAAd,qBARH;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,WAAF;;AAAA;AAAA;AAAA;;AAAA;AAAA,WADC;AAWLC,QAAAA,MAAM;AAAA,mFAAE;AAAA;AAAA;AAAA;AAAA;AACNH,oBAAAA,IAAI,GAAG,IAAP;AADM,sDAEC;AAACA,sBAAAA,IAAI,EAAJA,IAAD;AAAOE,sBAAAA,KAAK,EAAE;AAAd,qBAFD;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,WAAF;;AAAA;AAAA;AAAA;;AAAA;AAAA,WAXD;AAeLE,QAAAA,KAAK;AAAA,kFAAE;AAAA;AAAA;AAAA;AAAA;AACLJ,oBAAAA,IAAI,GAAG,IAAP;AADK,sDAEE;AAACA,sBAAAA,IAAI,EAAE,IAAP;AAAaE,sBAAAA,KAAK,EAAE;AAApB,qBAFF;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,WAAF;;AAAA;AAAA;AAAA;;AAAA;AAAA;AAfA,OAAP;AAoBD","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\nimport {FileMetaData} from '../parquet-thrift';\nimport {ParquetEnvelopeReader} from './parquet-envelope-reader';\nimport {ParquetSchema} from '../schema/schema';\nimport {ParquetRecord} from '../schema/declare';\nimport {materializeRecords} from '../schema/shred';\n\n/**\n * A parquet cursor is used to retrieve rows from a parquet file in order\n */\nexport class ParquetCursor<T> implements AsyncIterable<T> {\n public metadata: FileMetaData;\n public envelopeReader: ParquetEnvelopeReader;\n public schema: ParquetSchema;\n public columnList: string[][];\n public rowGroup: ParquetRecord[] = [];\n public rowGroupIndex: number;\n\n /**\n * Create a new parquet reader from the file metadata and an envelope reader.\n * It is usually not recommended to call this constructor directly except for\n * advanced and internal use cases. Consider using getCursor() on the\n * ParquetReader instead\n */\n constructor(\n metadata: FileMetaData,\n envelopeReader: ParquetEnvelopeReader,\n schema: ParquetSchema,\n columnList: string[][]\n ) {\n this.metadata = metadata;\n this.envelopeReader = envelopeReader;\n this.schema = schema;\n this.columnList = columnList;\n this.rowGroupIndex = 0;\n }\n\n /**\n * Retrieve the next row from the cursor. Returns a row or NULL if the end\n * of the file was reached\n */\n async next<T = any>(): Promise<T> {\n if (this.rowGroup.length === 0) {\n if (this.rowGroupIndex >= this.metadata.row_groups.length) {\n // @ts-ignore\n return null;\n }\n const rowBuffer = await this.envelopeReader.readRowGroup(\n this.schema,\n this.metadata.row_groups[this.rowGroupIndex],\n this.columnList\n );\n this.rowGroup = materializeRecords(this.schema, rowBuffer);\n this.rowGroupIndex++;\n }\n return this.rowGroup.shift() as any;\n }\n\n /**\n * Rewind the cursor the the beginning of the file\n */\n rewind(): void {\n this.rowGroup = [];\n this.rowGroupIndex = 0;\n }\n\n /**\n * Implement AsyncIterable\n */\n // tslint:disable-next-line:function-name\n [Symbol.asyncIterator](): AsyncIterator<T> {\n let done = false;\n return {\n next: async () => {\n if (done) {\n return {done, value: null};\n }\n const value = await this.next();\n if (value === null) {\n return {done: true, value};\n }\n return {done: false, value};\n },\n return: async () => {\n done = true;\n return {done, value: null};\n },\n throw: async () => {\n done = true;\n return {done: true, value: null};\n }\n };\n }\n}\n"],"file":"parquet-cursor.js"}