@loaders.gl/parquet 3.0.12 → 3.1.0-alpha.4

package/dist/esm/parquetjs/parser/decoders.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../src/parquetjs/parser/decoders.ts"],"names":["PARQUET_CODECS","ConvertedType","Encoding","FieldRepetitionType","PageType","Type","decompress","PARQUET_RDLVL_TYPE","PARQUET_RDLVL_ENCODING","decodePageHeader","getThriftEnum","getBitWidth","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","type","decodeDataPage","decodeDataPageV2","decodeDictionaryPage","Error","decodeSchema","schemaElements","len","schema","next","i","schemaElement","repetitionType","repetition_type","optional","repeated","num_children","res","name","fields","logicalType","converted_type","typeLength","type_length","presision","precision","scale","decodeValues","encoding","opts","header","cursorEnd","compressed_page_size","valueCount","data_page_header","num_values","dataCursor","compression","valuesBuf","slice","uncompressed_page_size","rLevelEncoding","repetition_level_encoding","rLevels","Array","column","rLevelMax","bitWidth","disableEnvelope","fill","dLevelEncoding","definition_level_encoding","dLevels","dLevelMax","valueCountNonNull","dlvl","valueEncoding","decodeOptions","primitiveType","data_page_header_v2","num_nulls","valuesBufCursor","is_compressed","dictCursor","dictionary_page_header","d","toString"],"mappings":"AAUA,SAA2CA,cAA3C,QAAgE,WAAhE;AACA,SACEC,aADF,EAEEC,QAFF,EAGEC,mBAHF,EAKEC,QALF,EAOEC,IAPF,QAQO,mBARP;AASA,SAAQC,UAAR,QAAyB,gBAAzB;AACA,SAAQC,kBAAR,EAA4BC,sBAA5B,QAAyD,iBAAzD;AACA,SAAQC,gBAAR,EAA0BC,aAA1B,EAAyCC,WAAzC,QAA2D,qBAA3D;AASA,OAAO,eAAeC,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;AAOD,OAAO,eAAeU,UAAf,CACLd,MADK,EAELD,OAFK,EAGqB;AAC1B,MAAIc,IAAJ;AACA,QAAM;AAACO,IAAAA,UAAD;AAAajB,IAAAA;AAAb,MAAuB,MAAMT,gBAAgB,CAACM,MAAM,CAACF,MAAR,EAAgBE,MAAM,CAACC,MAAvB,CAAnD;AACAD,EAAAA,MAAM,CAACC,MAAP,IAAiBE,MAAjB;AAEA,QAAMkB,QAAQ,GAAG1B,aAAa,CAACN,QAAD,EAAW+B,UAAU,CAACE,IAAtB,CAA9B;;AAEA,UAAQD,QAAR;AACE,SAAK,WAAL;AACER,MAAAA,IAAI,GAAG,MAAMU,cAAc,CAACvB,MAAD,EAASoB,UAAT,EAAqBrB,OAArB,CAA3B;AACA;;AACF,SAAK,cAAL;AACEc,MAAAA,IAAI,GAAG,MAAMW,gBAAgB,CAACxB,MAAD,EAASoB,UAAT,EAAqBrB,OAArB,CAA7B;AACA;;AACF,SAAK,iBAAL;AACEc,MAAAA,IAAI,GAAG;AACLH,QAAAA,UAAU,EAAE,MAAMe,oBAAoB,CAACzB,MAAD,EAASoB,UAAT,EAAqBrB,OAArB,CADjC;AAELqB,QAAAA;AAFK,OAAP;AAIA;;AACF;AACE,YAAM,IAAIM,KAAJ,CAAW,sBAAqBL,QAAS,EAAzC,CAAN;AAdJ;;AAiBA,SAAOR,IAAP;AACD;AAYD,OAAO,SAASc,YAAT,CACLC,cADK,EAEL3B,MAFK,EAGL4B,GAHK,EAQL;AACA,QAAMC,MAAwB,GAAG,EAAjC;AACA,MAAIC,IAAI,GAAG9B,MAAX;;AACA,OAAK,IAAI+B,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,GAAWpC,aAAa,CAACP,mBAAD,EAAsB6C,aAAa,CAACE,eAApC,CAAxB,GAAgF,MADlF;AAGA,QAAIC,QAAQ,GAAG,KAAf;AACA,QAAIC,QAAQ,GAAG,KAAf;;AACA,YAAQH,cAAR;AACE,WAAK,UAAL;AACE;;AACF,WAAK,UAAL;AACEE,QAAAA,QAAQ,GAAG,IAAX;AACA;;AACF,WAAK,UAAL;AACEC,QAAAA,QAAQ,GAAG,IAAX;AACA;;AACF;AACE,cAAM,IAAIX,KAAJ,CAAU,kCAAV,CAAN;AAVJ;;AAaA,QAAIO,aAAa,CAACK,YAAd,GAA8B,CAAlC,EAAqC;AACnC,YAAMC,GAAG,GAAGZ,YAAY,CAACC,cAAD,EAAiBG,IAAI,GAAG,CAAxB,EAA2BE,aAAa,CAACK,YAAzC,CAAxB;AACAP,MAAAA,IAAI,GAAGQ,GAAG,CAACR,IAAX;AACAD,MAAAA,MAAM,CAACG,aAAa,CAACO,IAAf,CAAN,GAA6B;AAE3BJ,QAAAA,QAF2B;AAG3BC,QAAAA,QAH2B;AAI3BI,QAAAA,MAAM,EAAEF,GAAG,CAACT;AAJe,OAA7B;AAMD,KATD,MASO;AACL,YAAMR,IAAI,GAAG3B,aAAa,CAACL,IAAD,EAAO2C,aAAa,CAACX,IAArB,CAA1B;AACA,UAAIoB,WAAW,GAAGpB,IAAlB;;AAEA,UAAIW,aAAa,CAACU,cAAlB,EAAkC;AAChCD,QAAAA,WAAW,GAAG/C,aAAa,CAACT,aAAD,EAAgB+C,aAAa,CAACU,cAA9B,CAA3B;AACD;;AAED,cAAQD,WAAR;AACE,aAAK,SAAL;AACEA,UAAAA,WAAW,GAAI,GAAEA,WAAY,IAAGpB,IAAK,EAArC;AACA;;AACF;AAJF;;AAOAQ,MAAAA,MAAM,CAACG,aAAa,CAACO,IAAf,CAAN,GAA6B;AAC3BlB,QAAAA,IAAI,EAAEoB,WADqB;AAE3BE,QAAAA,UAAU,EAAEX,aAAa,CAACY,WAFC;AAG3BC,QAAAA,SAAS,EAAEb,aAAa,CAACc,SAHE;AAI3BC,QAAAA,KAAK,EAAEf,aAAa,CAACe,KAJM;AAK3BZ,QAAAA,QAL2B;AAM3BC,QAAAA;AAN2B,OAA7B;AAQAN,MAAAA,IAAI;AACL;AACF;;AACD,SAAO;AAACD,IAAAA,MAAD;AAAS7B,IAAAA,MAAT;AAAiB8B,IAAAA;AAAjB,GAAP;AACD;;AAKD,SAASkB,YAAT,CACE3B,IADF,EAEE4B,QAFF,EAGElD,MAHF,EAIES,KAJF,EAKE0C,IALF,EAMS;AACP,MAAI,EAAED,QAAQ,IAAIjE,cAAd,CAAJ,EAAmC;AACjC,UAAM,IAAIyC,KAAJ,CAAW,qBAAoBwB,QAAS,EAAxC,CAAN;AACD;;AACD,SAAOjE,cAAc,CAACiE,QAAD,CAAd,CAAyBD,YAAzB,CAAsC3B,IAAtC,EAA4CtB,MAA5C,EAAoDS,KAApD,EAA2D0C,IAA3D,CAAP;AACD;;AAQD,eAAe5B,cAAf,CACEvB,MADF,EAEEoD,MAFF,EAGErD,OAHF,EAI4B;AAAA;;AAC1B,QAAMsD,SAAS,GAAGrD,MAAM,CAACC,MAAP,GAAgBmD,MAAM,CAACE,oBAAzC;AACA,QAAMC,UAAU,4BAAGH,MAAM,CAACI,gBAAV,0DAAG,sBAAyBC,UAA5C;AAGA,MAAIC,UAAU,GAAG1D,MAAjB;;AAEA,MAAID,OAAO,CAAC4D,WAAR,KAAwB,cAA5B,EAA4C;AAC1C,UAAMC,SAAS,GAAG,MAAMrE,UAAU,CAChCQ,OAAO,CAAC4D,WADwB,EAEhC3D,MAAM,CAACF,MAAP,CAAc+D,KAAd,CAAoB7D,MAAM,CAACC,MAA3B,EAAmCoD,SAAnC,CAFgC,EAGhCD,MAAM,CAACU,sBAHyB,CAAlC;AAKAJ,IAAAA,UAAU,GAAG;AACX5D,MAAAA,MAAM,EAAE8D,SADG;AAEX3D,MAAAA,MAAM,EAAE,CAFG;AAGXC,MAAAA,IAAI,EAAE0D,SAAS,CAACzD;AAHL,KAAb;AAKAH,IAAAA,MAAM,CAACC,MAAP,GAAgBoD,SAAhB;AACD;;AAGD,QAAMU,cAAc,GAAGpE,aAAa,CAClCR,QADkC,4BAElCiE,MAAM,CAACI,gBAF2B,2DAElC,uBAAyBQ,yBAFS,CAApC;AAKA,MAAIC,OAAO,GAAG,IAAIC,KAAJ,CAAUX,UAAV,CAAd;;AAEA,MAAIxD,OAAO,CAACoE,MAAR,CAAeC,SAAf,GAA2B,CAA/B,EAAkC;AAChCH,IAAAA,OAAO,GAAGhB,YAAY,CAACzD,kBAAD,EAAqBuE,cAArB,EAAqCL,UAArC,EAAiDH,UAAjD,EAA8D;AAClFc,MAAAA,QAAQ,EAAEzE,WAAW,CAACG,OAAO,CAACoE,MAAR,CAAeC,SAAhB,CAD6D;AAElFE,MAAAA,eAAe,EAAE;AAFiE,KAA9D,CAAtB;AAKD,GAND,MAMO;AACLL,IAAAA,OAAO,CAACM,IAAR,CAAa,CAAb;AACD;;AAGD,QAAMC,cAAc,GAAG7E,aAAa,CAClCR,QADkC,4BAElCiE,MAAM,CAACI,gBAF2B,2DAElC,uBAAyBiB,yBAFS,CAApC;AAKA,MAAIC,OAAO,GAAG,IAAIR,KAAJ,CAAUX,UAAV,CAAd;;AACA,MAAIxD,OAAO,CAACoE,MAAR,CAAeQ,SAAf,GAA2B,CAA/B,EAAkC;AAChCD,IAAAA,OAAO,GAAGzB,YAAY,CAACzD,kBAAD,EAAqBgF,cAArB,EAAqCd,UAArC,EAAiDH,UAAjD,EAA8D;AAClFc,MAAAA,QAAQ,EAAEzE,WAAW,CAACG,OAAO,CAACoE,MAAR,CAAeQ,SAAhB,CAD6D;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,KAAK9E,OAAO,CAACoE,MAAR,CAAeQ,SAA5B,EAAuC;AACrCC,MAAAA,iBAAiB;AAClB;AACF;;AAGD,QAAME,aAAa,GAAGnF,aAAa,CAACR,QAAD,4BAAWiE,MAAM,CAACI,gBAAlB,2DAAW,uBAAyBN,QAApC,CAAnC;AACA,QAAM6B,aAAa,GAAG;AACpBnC,IAAAA,UAAU,EAAE7C,OAAO,CAACoE,MAAR,CAAevB,UADP;AAEpByB,IAAAA,QAAQ,EAAEtE,OAAO,CAACoE,MAAR,CAAevB;AAFL,GAAtB;AAKA,QAAMrC,MAAM,GAAG0C,YAAY,CACzBlD,OAAO,CAACoE,MAAR,CAAea,aADU,EAEzBF,aAFyB,EAGzBpB,UAHyB,EAIzBkB,iBAJyB,EAKzBG,aALyB,CAA3B;AAQA,SAAO;AACLzE,IAAAA,OAAO,EAAEoE,OADJ;AAELrE,IAAAA,OAAO,EAAE4D,OAFJ;AAGL1D,IAAAA,MAHK;AAILE,IAAAA,KAAK,EAAE8C,UAJF;AAKLnC,IAAAA,UAAU,EAAEgC;AALP,GAAP;AAOD;;AASD,eAAe5B,gBAAf,CACExB,MADF,EAEEoD,MAFF,EAGED,IAHF,EAI4B;AAAA;;AAC1B,QAAME,SAAS,GAAGrD,MAAM,CAACC,MAAP,GAAgBmD,MAAM,CAACE,oBAAzC;AAEA,QAAMC,UAAU,6BAAGH,MAAM,CAAC6B,mBAAV,2DAAG,uBAA4BxB,UAA/C;AAEA,QAAMmB,iBAAiB,GAAGrB,UAAU,8BAAGH,MAAM,CAAC6B,mBAAV,2DAAG,uBAA4BC,SAA/B,CAApC;AACA,QAAMJ,aAAa,GAAGnF,aAAa,CACjCR,QADiC,4BAEjCiE,MAAM,CAAC6B,mBAF0B,2DAEjC,uBAA4B/B,QAFK,CAAnC;AAOA,MAAIe,OAAO,GAAG,IAAIC,KAAJ,CAAUX,UAAV,CAAd;;AACA,MAAIJ,IAAI,CAACgB,MAAL,CAAYC,SAAZ,GAAwB,CAA5B,EAA+B;AAC7BH,IAAAA,OAAO,GAAGhB,YAAY,CAACzD,kBAAD,EAAqBC,sBAArB,EAA6CO,MAA7C,EAAqDuD,UAArD,EAAkE;AACtFc,MAAAA,QAAQ,EAAEzE,WAAW,CAACuD,IAAI,CAACgB,MAAL,CAAYC,SAAb,CADiE;AAEtFE,MAAAA,eAAe,EAAE;AAFqE,KAAlE,CAAtB;AAID,GALD,MAKO;AACLL,IAAAA,OAAO,CAACM,IAAR,CAAa,CAAb;AACD;;AAID,MAAIG,OAAO,GAAG,IAAIR,KAAJ,CAAUX,UAAV,CAAd;;AACA,MAAIJ,IAAI,CAACgB,MAAL,CAAYQ,SAAZ,GAAwB,CAA5B,EAA+B;AAC7BD,IAAAA,OAAO,GAAGzB,YAAY,CAACzD,kBAAD,EAAqBC,sBAArB,EAA6CO,MAA7C,EAAqDuD,UAArD,EAAkE;AACtFc,MAAAA,QAAQ,EAAEzE,WAAW,CAACuD,IAAI,CAACgB,MAAL,CAAYQ,SAAb,CADiE;AAEtFL,MAAAA,eAAe,EAAE;AAFqE,KAAlE,CAAtB;AAID,GALD,MAKO;AACLI,IAAAA,OAAO,CAACH,IAAR,CAAa,CAAb;AACD;;AAGD,MAAIY,eAAe,GAAGnF,MAAtB;;AAEA,gCAAIoD,MAAM,CAAC6B,mBAAX,mDAAI,uBAA4BG,aAAhC,EAA+C;AAC7C,UAAMxB,SAAS,GAAG,MAAMrE,UAAU,CAChC4D,IAAI,CAACQ,WAD2B,EAEhC3D,MAAM,CAACF,MAAP,CAAc+D,KAAd,CAAoB7D,MAAM,CAACC,MAA3B,EAAmCoD,SAAnC,CAFgC,EAGhCD,MAAM,CAACU,sBAHyB,CAAlC;AAMAqB,IAAAA,eAAe,GAAG;AAChBrF,MAAAA,MAAM,EAAE8D,SADQ;AAEhB3D,MAAAA,MAAM,EAAE,CAFQ;AAGhBC,MAAAA,IAAI,EAAE0D,SAAS,CAACzD;AAHA,KAAlB;AAMAH,IAAAA,MAAM,CAACC,MAAP,GAAgBoD,SAAhB;AACD;;AAED,QAAM0B,aAAa,GAAG;AACpBnC,IAAAA,UAAU,EAAEO,IAAI,CAACgB,MAAL,CAAYvB,UADJ;AAEpByB,IAAAA,QAAQ,EAAElB,IAAI,CAACgB,MAAL,CAAYvB;AAFF,GAAtB;AAKA,QAAMrC,MAAM,GAAG0C,YAAY,CACzBE,IAAI,CAACgB,MAAL,CAAYa,aADa,EAEzBF,aAFyB,EAGzBK,eAHyB,EAIzBP,iBAJyB,EAKzBG,aALyB,CAA3B;AAQA,SAAO;AACLzE,IAAAA,OAAO,EAAEoE,OADJ;AAELrE,IAAAA,OAAO,EAAE4D,OAFJ;AAGL1D,IAAAA,MAHK;AAILE,IAAAA,KAAK,EAAE8C,UAJF;AAKLnC,IAAAA,UAAU,EAAEgC;AALP,GAAP;AAOD;;AAQD,eAAe3B,oBAAf,CACEzB,MADF,EAEEoB,UAFF,EAGErB,OAHF,EAIqB;AAAA;;AACnB,QAAMsD,SAAS,GAAGrD,MAAM,CAACC,MAAP,GAAgBmB,UAAU,CAACkC,oBAA7C;AAEA,MAAI+B,UAAU,GAAG;AACfpF,IAAAA,MAAM,EAAE,CADO;AAEfH,IAAAA,MAAM,EAAEE,MAAM,CAACF,MAAP,CAAc+D,KAAd,CAAoB7D,MAAM,CAACC,MAA3B,EAAmCoD,SAAnC,CAFO;AAGfnD,IAAAA,IAAI,EAAEmD,SAAS,GAAGrD,MAAM,CAACC;AAHV,GAAjB;AAMAD,EAAAA,MAAM,CAACC,MAAP,GAAgBoD,SAAhB;;AAEA,MAAItD,OAAO,CAAC4D,WAAR,KAAwB,cAA5B,EAA4C;AAC1C,UAAMC,SAAS,GAAG,MAAMrE,UAAU,CAChCQ,OAAO,CAAC4D,WADwB,EAEhC0B,UAAU,CAACvF,MAAX,CAAkB+D,KAAlB,CAAwBwB,UAAU,CAACpF,MAAnC,EAA2CoD,SAA3C,CAFgC,EAGhCjC,UAAU,CAAC0C,sBAHqB,CAAlC;AAMAuB,IAAAA,UAAU,GAAG;AACXvF,MAAAA,MAAM,EAAE8D,SADG;AAEX3D,MAAAA,MAAM,EAAE,CAFG;AAGXC,MAAAA,IAAI,EAAE0D,SAAS,CAACzD;AAHL,KAAb;AAMAH,IAAAA,MAAM,CAACC,MAAP,GAAgBoD,SAAhB;AACD;;AAED,QAAM1C,SAAS,GAAG,CAAAS,UAAU,SAAV,IAAAA,UAAU,WAAV,qCAAAA,UAAU,CAAEkE,sBAAZ,gFAAoC7B,UAApC,KAAkD,CAApE;AAEA,SAAOR,YAAY,CACjBlD,OAAO,CAACoE,MAAR,CAAea,aADE,EAEjBjF,OAAO,CAACoE,MAAR,CAAejB,QAFE,EAGjBmC,UAHiB,EAIjB1E,SAJiB,EAKjBZ,OALiB,CAAZ,CAMLgB,GANK,CAMAwE,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} = await 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"}

package/dist/esm/parquetjs/parser/parquet-cursor.js

@@ -0,0 +1,90 @@
+import _defineProperty from "@babel/runtime/helpers/esm/defineProperty";
+
+let _Symbol$asyncIterator;
+
+import { materializeRecords } from '../schema/shred';
+_Symbol$asyncIterator = Symbol.asyncIterator;
+export class ParquetCursor {
+  constructor(metadata, envelopeReader, schema, columnList) {
+    _defineProperty(this, "metadata", void 0);
+
+    _defineProperty(this, "envelopeReader", void 0);
+
+    _defineProperty(this, "schema", void 0);
+
+    _defineProperty(this, "columnList", void 0);
+
+    _defineProperty(this, "rowGroup", []);
+
+    _defineProperty(this, "rowGroupIndex", void 0);
+
+    this.metadata = metadata;
+    this.envelopeReader = envelopeReader;
+    this.schema = schema;
+    this.columnList = columnList;
+    this.rowGroupIndex = 0;
+  }
+
+  async next() {
+    if (this.rowGroup.length === 0) {
+      if (this.rowGroupIndex >= this.metadata.row_groups.length) {
+        return null;
+      }
+
+      const rowBuffer = await this.envelopeReader.readRowGroup(this.schema, this.metadata.row_groups[this.rowGroupIndex], this.columnList);
+      this.rowGroup = materializeRecords(this.schema, rowBuffer);
+      this.rowGroupIndex++;
+    }
+
+    return this.rowGroup.shift();
+  }
+
+  rewind() {
+    this.rowGroup = [];
+    this.rowGroupIndex = 0;
+  }
+
+  [_Symbol$asyncIterator]() {
+    let done = false;
+    return {
+      next: async () => {
+        if (done) {
+          return {
+            done,
+            value: null
+          };
+        }
+
+        const value = await this.next();
+
+        if (value === null) {
+          return {
+            done: true,
+            value
+          };
+        }
+
+        return {
+          done: false,
+          value
+        };
+      },
+      return: async () => {
+        done = true;
+        return {
+          done,
+          value: null
+        };
+      },
+      throw: async () => {
+        done = true;
+        return {
+          done: true,
+          value: null
+        };
+      }
+    };
+  }
+
+}
+//# sourceMappingURL=parquet-cursor.js.map

package/dist/esm/parquetjs/parser/parquet-cursor.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../src/parquetjs/parser/parquet-cursor.ts"],"names":["materializeRecords","Symbol","asyncIterator","ParquetCursor","constructor","metadata","envelopeReader","schema","columnList","rowGroupIndex","next","rowGroup","length","row_groups","rowBuffer","readRowGroup","shift","rewind","done","value","return","throw"],"mappings":";;;;AAKA,SAAQA,kBAAR,QAAiC,iBAAjC;wBAiEGC,MAAM,CAACC,a;AA5DV,OAAO,MAAMC,aAAN,CAAmD;AAcxDC,EAAAA,WAAW,CACTC,QADS,EAETC,cAFS,EAGTC,MAHS,EAITC,UAJS,EAKT;AAAA;;AAAA;;AAAA;;AAAA;;AAAA,sCAdiC,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,GAAgBX,kBAAkB,CAAC,KAAKO,MAAN,EAAcO,SAAd,CAAlC;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"}

package/dist/esm/parquetjs/parser/parquet-envelope-reader.js

@@ -0,0 +1,164 @@
+import _defineProperty from "@babel/runtime/helpers/esm/defineProperty";
+import { PARQUET_MAGIC, PARQUET_MAGIC_ENCRYPTED } from '../../constants';
+import { CompressionCodec, Type } from '../parquet-thrift';
+import { fstat, fopen, fread, fclose } from '../utils/file-utils';
+import { decodeFileMetadata, getThriftEnum, fieldIndexOf } from '../utils/read-utils';
+import { decodeDataPages, decodePage } from './decoders';
+const DEFAULT_DICTIONARY_SIZE = 1e6;
+export class ParquetEnvelopeReader {
+  static async openFile(filePath) {
+    const fileStat = await fstat(filePath);
+    const fileDescriptor = await fopen(filePath);
+    const readFn = fread.bind(undefined, fileDescriptor);
+    const closeFn = fclose.bind(undefined, fileDescriptor);
+    return new ParquetEnvelopeReader(readFn, closeFn, fileStat.size);
+  }
+
+  static async openBuffer(buffer) {
+    const readFn = (position, length) => Promise.resolve(buffer.slice(position, position + length));
+
+    const closeFn = () => Promise.resolve();
+
+    return new ParquetEnvelopeReader(readFn, closeFn, buffer.length);
+  }
+
+  constructor(read, close, fileSize, options) {
+    _defineProperty(this, "read", void 0);
+
+    _defineProperty(this, "close", void 0);
+
+    _defineProperty(this, "fileSize", void 0);
+
+    _defineProperty(this, "defaultDictionarySize", void 0);
+
+    this.read = read;
+    this.close = close;
+    this.fileSize = fileSize;
+    this.defaultDictionarySize = (options === null || options === void 0 ? void 0 : options.defaultDictionarySize) || DEFAULT_DICTIONARY_SIZE;
+  }
+
+  async readHeader() {
+    const buffer = await this.read(0, PARQUET_MAGIC.length);
+    const magic = buffer.toString();
+
+    switch (magic) {
+      case PARQUET_MAGIC:
+        break;
+
+      case PARQUET_MAGIC_ENCRYPTED:
+        throw new Error('Encrypted parquet file not supported');
+
+      default:
+        throw new Error(`Invalid parquet file (magic=${magic})`);
+    }
+  }
+
+  async readRowGroup(schema, rowGroup, columnList) {
+    const buffer = {
+      rowCount: Number(rowGroup.num_rows),
+      columnData: {}
+    };
+
+    for (const colChunk of rowGroup.columns) {
+      const colMetadata = colChunk.meta_data;
+      const colKey = colMetadata === null || colMetadata === void 0 ? void 0 : colMetadata.path_in_schema;
+
+      if (columnList.length > 0 && fieldIndexOf(columnList, colKey) < 0) {
+        continue;
+      }
+
+      buffer.columnData[colKey.join()] = await this.readColumnChunk(schema, colChunk);
+    }
+
+    return buffer;
+  }
+
+  async readColumnChunk(schema, colChunk) {
+    var _colChunk$meta_data, _colChunk$meta_data2, _colChunk$meta_data3, _colChunk$meta_data4, _colChunk$meta_data5, _colChunk$meta_data7, _colChunk$meta_data8, _options$dictionary;
+
+    if (colChunk.file_path !== undefined && colChunk.file_path !== null) {
+      throw new Error('external references are not supported');
+    }
+
+    const field = schema.findField((_colChunk$meta_data = colChunk.meta_data) === null || _colChunk$meta_data === void 0 ? void 0 : _colChunk$meta_data.path_in_schema);
+    const type = getThriftEnum(Type, (_colChunk$meta_data2 = colChunk.meta_data) === null || _colChunk$meta_data2 === void 0 ? void 0 : _colChunk$meta_data2.type);
+
+    if (type !== field.primitiveType) {
+      throw new Error(`chunk type not matching schema: ${type}`);
+    }
+
+    const compression = getThriftEnum(CompressionCodec, (_colChunk$meta_data3 = colChunk.meta_data) === null || _colChunk$meta_data3 === void 0 ? void 0 : _colChunk$meta_data3.codec);
+    const pagesOffset = Number((_colChunk$meta_data4 = colChunk.meta_data) === null || _colChunk$meta_data4 === void 0 ? void 0 : _colChunk$meta_data4.data_page_offset);
+    let pagesSize = Number((_colChunk$meta_data5 = colChunk.meta_data) === null || _colChunk$meta_data5 === void 0 ? void 0 : _colChunk$meta_data5.total_compressed_size);
+
+    if (!colChunk.file_path) {
+      var _colChunk$meta_data6;
+
+      pagesSize = Math.min(this.fileSize - pagesOffset, Number((_colChunk$meta_data6 = colChunk.meta_data) === null || _colChunk$meta_data6 === void 0 ? void 0 : _colChunk$meta_data6.total_compressed_size));
+    }
+
+    const options = {
+      type,
+      rLevelMax: field.rLevelMax,
+      dLevelMax: field.dLevelMax,
+      compression,
+      column: field,
+      numValues: (_colChunk$meta_data7 = colChunk.meta_data) === null || _colChunk$meta_data7 === void 0 ? void 0 : _colChunk$meta_data7.num_values,
+      dictionary: []
+    };
+    let dictionary;
+    const dictionaryPageOffset = colChunk === null || colChunk === void 0 ? void 0 : (_colChunk$meta_data8 = colChunk.meta_data) === null || _colChunk$meta_data8 === void 0 ? void 0 : _colChunk$meta_data8.dictionary_page_offset;
+
+    if (dictionaryPageOffset) {
+      const dictionaryOffset = Number(dictionaryPageOffset);
+      dictionary = await this.getDictionary(dictionaryOffset, options, pagesOffset);
+    }
+
+    dictionary = (_options$dictionary = options.dictionary) !== null && _options$dictionary !== void 0 && _options$dictionary.length ? options.dictionary : dictionary;
+    const pagesBuf = await this.read(pagesOffset, pagesSize);
+    return await decodeDataPages(pagesBuf, { ...options,
+      dictionary
+    });
+  }
+
+  async getDictionary(dictionaryPageOffset, options, pagesOffset) {
+    if (dictionaryPageOffset === 0) {
+      return [];
+    }
+
+    const dictionarySize = Math.min(this.fileSize - dictionaryPageOffset, this.defaultDictionarySize);
+    const pagesBuf = await this.read(dictionaryPageOffset, dictionarySize);
+    const cursor = {
+      buffer: pagesBuf,
+      offset: 0,
+      size: pagesBuf.length
+    };
+    const decodedPage = await decodePage(cursor, options);
+    return decodedPage.dictionary;
+  }
+
+  async readFooter() {
+    const trailerLen = PARQUET_MAGIC.length + 4;
+    const trailerBuf = await this.read(this.fileSize - trailerLen, trailerLen);
+    const magic = trailerBuf.slice(4).toString();
+
+    if (magic !== PARQUET_MAGIC) {
+      throw new Error(`Not a valid parquet file (magic="${magic})`);
+    }
+
+    const metadataSize = trailerBuf.readUInt32LE(0);
+    const metadataOffset = this.fileSize - metadataSize - trailerLen;
+
+    if (metadataOffset < PARQUET_MAGIC.length) {
+      throw new Error(`Invalid metadata size ${metadataOffset}`);
+    }
+
+    const metadataBuf = await this.read(metadataOffset, metadataSize);
+    const {
+      metadata
+    } = decodeFileMetadata(metadataBuf);
+    return metadata;
+  }
+
+}
+//# sourceMappingURL=parquet-envelope-reader.js.map

package/dist/esm/parquetjs/parser/parquet-envelope-reader.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../src/parquetjs/parser/parquet-envelope-reader.ts"],"names":["PARQUET_MAGIC","PARQUET_MAGIC_ENCRYPTED","CompressionCodec","Type","fstat","fopen","fread","fclose","decodeFileMetadata","getThriftEnum","fieldIndexOf","decodeDataPages","decodePage","DEFAULT_DICTIONARY_SIZE","ParquetEnvelopeReader","openFile","filePath","fileStat","fileDescriptor","readFn","bind","undefined","closeFn","size","openBuffer","buffer","position","length","Promise","resolve","slice","constructor","read","close","fileSize","options","defaultDictionarySize","readHeader","magic","toString","Error","readRowGroup","schema","rowGroup","columnList","rowCount","Number","num_rows","columnData","colChunk","columns","colMetadata","meta_data","colKey","path_in_schema","join","readColumnChunk","file_path","field","findField","type","primitiveType","compression","codec","pagesOffset","data_page_offset","pagesSize","total_compressed_size","Math","min","rLevelMax","dLevelMax","column","numValues","num_values","dictionary","dictionaryPageOffset","dictionary_page_offset","dictionaryOffset","getDictionary","pagesBuf","dictionarySize","cursor","offset","decodedPage","readFooter","trailerLen","trailerBuf","metadataSize","readUInt32LE","metadataOffset","metadataBuf","metadata"],"mappings":";AAEA,SAAQA,aAAR,EAAuBC,uBAAvB,QAAqD,iBAArD;AACA,SAAqBC,gBAArB,EAA+DC,IAA/D,QAA0E,mBAA1E;AAQA,SAAQC,KAAR,EAAeC,KAAf,EAAsBC,KAAtB,EAA6BC,MAA7B,QAA0C,qBAA1C;AACA,SAAQC,kBAAR,EAA4BC,aAA5B,EAA2CC,YAA3C,QAA8D,qBAA9D;AACA,SAAQC,eAAR,EAAyBC,UAAzB,QAA0C,YAA1C;AAEA,MAAMC,uBAAuB,GAAG,GAAhC;AAQA,OAAO,MAAMC,qBAAN,CAA4B;AAUZ,eAARC,QAAQ,CAACC,QAAD,EAAmD;AACtE,UAAMC,QAAQ,GAAG,MAAMb,KAAK,CAACY,QAAD,CAA5B;AACA,UAAME,cAAc,GAAG,MAAMb,KAAK,CAACW,QAAD,CAAlC;AAEA,UAAMG,MAAM,GAAGb,KAAK,CAACc,IAAN,CAAWC,SAAX,EAAsBH,cAAtB,CAAf;AACA,UAAMI,OAAO,GAAGf,MAAM,CAACa,IAAP,CAAYC,SAAZ,EAAuBH,cAAvB,CAAhB;AAEA,WAAO,IAAIJ,qBAAJ,CAA0BK,MAA1B,EAAkCG,OAAlC,EAA2CL,QAAQ,CAACM,IAApD,CAAP;AACD;;AAEsB,eAAVC,UAAU,CAACC,MAAD,EAAiD;AACtE,UAAMN,MAAM,GAAG,CAACO,QAAD,EAAmBC,MAAnB,KACbC,OAAO,CAACC,OAAR,CAAgBJ,MAAM,CAACK,KAAP,CAAaJ,QAAb,EAAuBA,QAAQ,GAAGC,MAAlC,CAAhB,CADF;;AAEA,UAAML,OAAO,GAAG,MAAMM,OAAO,CAACC,OAAR,EAAtB;;AACA,WAAO,IAAIf,qBAAJ,CAA0BK,MAA1B,EAAkCG,OAAlC,EAA2CG,MAAM,CAACE,MAAlD,CAAP;AACD;;AAEDI,EAAAA,WAAW,CACTC,IADS,EAETC,KAFS,EAGTC,QAHS,EAITC,OAJS,EAKT;AAAA;;AAAA;;AAAA;;AAAA;;AACA,SAAKH,IAAL,GAAYA,IAAZ;AACA,SAAKC,KAAL,GAAaA,KAAb;AACA,SAAKC,QAAL,GAAgBA,QAAhB;AACA,SAAKE,qBAAL,GAA6B,CAAAD,OAAO,SAAP,IAAAA,OAAO,WAAP,YAAAA,OAAO,CAAEC,qBAAT,KAAkCvB,uBAA/D;AACD;;AAEe,QAAVwB,UAAU,GAAkB;AAChC,UAAMZ,MAAM,GAAG,MAAM,KAAKO,IAAL,CAAU,CAAV,EAAahC,aAAa,CAAC2B,MAA3B,CAArB;AAEA,UAAMW,KAAK,GAAGb,MAAM,CAACc,QAAP,EAAd;;AACA,YAAQD,KAAR;AACE,WAAKtC,aAAL;AACE;;AACF,WAAKC,uBAAL;AACE,cAAM,IAAIuC,KAAJ,CAAU,sCAAV,CAAN;;AACF;AACE,cAAM,IAAIA,KAAJ,CAAW,+BAA8BF,KAAM,GAA/C,CAAN;AANJ;AAQD;;AAEiB,QAAZG,YAAY,CAChBC,MADgB,EAEhBC,QAFgB,EAGhBC,UAHgB,EAIQ;AACxB,UAAMnB,MAAqB,GAAG;AAC5BoB,MAAAA,QAAQ,EAAEC,MAAM,CAACH,QAAQ,CAACI,QAAV,CADY;AAE5BC,MAAAA,UAAU,EAAE;AAFgB,KAA9B;;AAIA,SAAK,MAAMC,QAAX,IAAuBN,QAAQ,CAACO,OAAhC,EAAyC;AACvC,YAAMC,WAAW,GAAGF,QAAQ,CAACG,SAA7B;AACA,YAAMC,MAAM,GAAGF,WAAH,aAAGA,WAAH,uBAAGA,WAAW,CAAEG,cAA5B;;AACA,UAAIV,UAAU,CAACjB,MAAX,GAAoB,CAApB,IAAyBjB,YAAY,CAACkC,UAAD,EAAaS,MAAb,CAAZ,GAAoC,CAAjE,EAAoE;AAClE;AACD;;AACD5B,MAAAA,MAAM,CAACuB,UAAP,CAAkBK,MAAM,CAAEE,IAAR,EAAlB,IAAoC,MAAM,KAAKC,eAAL,CAAqBd,MAArB,EAA6BO,QAA7B,CAA1C;AACD;;AACD,WAAOxB,MAAP;AACD;;AAOoB,QAAf+B,eAAe,CAACd,MAAD,EAAwBO,QAAxB,EAAqE;AAAA;;AACxF,QAAIA,QAAQ,CAACQ,SAAT,KAAuBpC,SAAvB,IAAoC4B,QAAQ,CAACQ,SAAT,KAAuB,IAA/D,EAAqE;AACnE,YAAM,IAAIjB,KAAJ,CAAU,uCAAV,CAAN;AACD;;AAED,UAAMkB,KAAK,GAAGhB,MAAM,CAACiB,SAAP,wBAAiBV,QAAQ,CAACG,SAA1B,wDAAiB,oBAAoBE,cAArC,CAAd;AACA,UAAMM,IAAmB,GAAGnD,aAAa,CAACN,IAAD,0BAAO8C,QAAQ,CAACG,SAAhB,yDAAO,qBAAoBQ,IAA3B,CAAzC;;AAEA,QAAIA,IAAI,KAAKF,KAAK,CAACG,aAAnB,EAAkC;AAChC,YAAM,IAAIrB,KAAJ,CAAW,mCAAkCoB,IAAK,EAAlD,CAAN;AACD;;AAED,UAAME,WAA+B,GAAGrD,aAAa,CACnDP,gBADmD,0BAEnD+C,QAAQ,CAACG,SAF0C,yDAEnD,qBAAoBW,KAF+B,CAArD;AAKA,UAAMC,WAAW,GAAGlB,MAAM,yBAACG,QAAQ,CAACG,SAAV,yDAAC,qBAAoBa,gBAArB,CAA1B;AACA,QAAIC,SAAS,GAAGpB,MAAM,yBAACG,QAAQ,CAACG,SAAV,yDAAC,qBAAoBe,qBAArB,CAAtB;;AAEA,QAAI,CAAClB,QAAQ,CAACQ,SAAd,EAAyB;AAAA;;AACvBS,MAAAA,SAAS,GAAGE,IAAI,CAACC,GAAL,CACV,KAAKnC,QAAL,GAAgB8B,WADN,EAEVlB,MAAM,yBAACG,QAAQ,CAACG,SAAV,yDAAC,qBAAoBe,qBAArB,CAFI,CAAZ;AAID;;AAED,UAAMhC,OAAuB,GAAG;AAC9ByB,MAAAA,IAD8B;AAE9BU,MAAAA,SAAS,EAAEZ,KAAK,CAACY,SAFa;AAG9BC,MAAAA,SAAS,EAAEb,KAAK,CAACa,SAHa;AAI9BT,MAAAA,WAJ8B;AAK9BU,MAAAA,MAAM,EAAEd,KALsB;AAM9Be,MAAAA,SAAS,0BAAExB,QAAQ,CAACG,SAAX,yDAAE,qBAAoBsB,UAND;AAO9BC,MAAAA,UAAU,EAAE;AAPkB,KAAhC;AAUA,QAAIA,UAAJ;AAEA,UAAMC,oBAAoB,GAAG3B,QAAH,aAAGA,QAAH,+CAAGA,QAAQ,CAAEG,SAAb,yDAAG,qBAAqByB,sBAAlD;;AAEA,QAAID,oBAAJ,EAA0B;AACxB,YAAME,gBAAgB,GAAGhC,MAAM,CAAC8B,oBAAD,CAA/B;AAEAD,MAAAA,UAAU,GAAG,MAAM,KAAKI,aAAL,CAAmBD,gBAAnB,EAAqC3C,OAArC,EAA8C6B,WAA9C,CAAnB;AACD;;AAEDW,IAAAA,UAAU,GAAG,uBAAAxC,OAAO,CAACwC,UAAR,oEAAoBhD,MAApB,GAA6BQ,OAAO,CAACwC,UAArC,GAAkDA,UAA/D;AACA,UAAMK,QAAQ,GAAG,MAAM,KAAKhD,IAAL,CAAUgC,WAAV,EAAuBE,SAAvB,CAAvB;AACA,WAAO,MAAMvD,eAAe,CAACqE,QAAD,EAAW,EAAC,GAAG7C,OAAJ;AAAawC,MAAAA;AAAb,KAAX,CAA5B;AACD;;AASkB,QAAbI,aAAa,CACjBH,oBADiB,EAEjBzC,OAFiB,EAGjB6B,WAHiB,EAIE;AACnB,QAAIY,oBAAoB,KAAK,CAA7B,EAAgC;AAQ9B,aAAO,EAAP;AACD;;AAED,UAAMK,cAAc,GAAGb,IAAI,CAACC,GAAL,CACrB,KAAKnC,QAAL,GAAgB0C,oBADK,EAErB,KAAKxC,qBAFgB,CAAvB;AAIA,UAAM4C,QAAQ,GAAG,MAAM,KAAKhD,IAAL,CAAU4C,oBAAV,EAAgCK,cAAhC,CAAvB;AAEA,UAAMC,MAAM,GAAG;AAACzD,MAAAA,MAAM,EAAEuD,QAAT;AAAmBG,MAAAA,MAAM,EAAE,CAA3B;AAA8B5D,MAAAA,IAAI,EAAEyD,QAAQ,CAACrD;AAA7C,KAAf;AACA,UAAMyD,WAAW,GAAG,MAAMxE,UAAU,CAACsE,MAAD,EAAS/C,OAAT,CAApC;AAEA,WAAOiD,WAAW,CAACT,UAAnB;AACD;;AAEe,QAAVU,UAAU,GAA0B;AACxC,UAAMC,UAAU,GAAGtF,aAAa,CAAC2B,MAAd,GAAuB,CAA1C;AACA,UAAM4D,UAAU,GAAG,MAAM,KAAKvD,IAAL,CAAU,KAAKE,QAAL,GAAgBoD,UAA1B,EAAsCA,UAAtC,CAAzB;AAEA,UAAMhD,KAAK,GAAGiD,UAAU,CAACzD,KAAX,CAAiB,CAAjB,EAAoBS,QAApB,EAAd;;AACA,QAAID,KAAK,KAAKtC,aAAd,EAA6B;AAC3B,YAAM,IAAIwC,KAAJ,CAAW,oCAAmCF,KAAM,GAApD,CAAN;AACD;;AAED,UAAMkD,YAAY,GAAGD,UAAU,CAACE,YAAX,CAAwB,CAAxB,CAArB;AACA,UAAMC,cAAc,GAAG,KAAKxD,QAAL,GAAgBsD,YAAhB,GAA+BF,UAAtD;;AACA,QAAII,cAAc,GAAG1F,aAAa,CAAC2B,MAAnC,EAA2C;AACzC,YAAM,IAAIa,KAAJ,CAAW,yBAAwBkD,cAAe,EAAlD,CAAN;AACD;;AAED,UAAMC,WAAW,GAAG,MAAM,KAAK3D,IAAL,CAAU0D,cAAV,EAA0BF,YAA1B,CAA1B;AAGA,UAAM;AAACI,MAAAA;AAAD,QAAapF,kBAAkB,CAACmF,WAAD,CAArC;AACA,WAAOC,QAAP;AACD;;AAzLgC","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\nimport {ParquetSchema} from '../schema/schema';\nimport {PARQUET_MAGIC, PARQUET_MAGIC_ENCRYPTED} from '../../constants';\nimport {ColumnChunk, CompressionCodec, FileMetaData, RowGroup, Type} from '../parquet-thrift';\nimport {\n  ParquetBuffer,\n  ParquetCompression,\n  ParquetData,\n  PrimitiveType,\n  ParquetOptions\n} from '../schema/declare';\nimport {fstat, fopen, fread, fclose} from '../utils/file-utils';\nimport {decodeFileMetadata, getThriftEnum, fieldIndexOf} from '../utils/read-utils';\nimport {decodeDataPages, decodePage} from './decoders';\n\nconst DEFAULT_DICTIONARY_SIZE = 1e6;\n\n/**\n * The parquet envelope reader allows direct, unbuffered access to the individual\n * sections of the parquet file, namely the header, footer and the row groups.\n * This class is intended for advanced/internal users; if you just want to retrieve\n * rows from a parquet file use the ParquetReader instead\n */\nexport class ParquetEnvelopeReader {\n  public read: (position: number, length: number) => Promise<Buffer>;\n  /**\n   * Close this parquet reader. You MUST call this method once you're finished\n   * reading rows\n   */\n  public close: () => Promise<void>;\n  public fileSize: number;\n  public defaultDictionarySize: number;\n\n  static async openFile(filePath: string): Promise<ParquetEnvelopeReader> {\n    const fileStat = await fstat(filePath);\n    const fileDescriptor = await fopen(filePath);\n\n    const readFn = fread.bind(undefined, fileDescriptor);\n    const closeFn = fclose.bind(undefined, fileDescriptor);\n\n    return new ParquetEnvelopeReader(readFn, closeFn, fileStat.size);\n  }\n\n  static async openBuffer(buffer: Buffer): Promise<ParquetEnvelopeReader> {\n    const readFn = (position: number, length: number) =>\n      Promise.resolve(buffer.slice(position, position + length));\n    const closeFn = () => Promise.resolve();\n    return new ParquetEnvelopeReader(readFn, closeFn, buffer.length);\n  }\n\n  constructor(\n    read: (position: number, length: number) => Promise<Buffer>,\n    close: () => Promise<void>,\n    fileSize: number,\n    options?: any\n  ) {\n    this.read = read;\n    this.close = close;\n    this.fileSize = fileSize;\n    this.defaultDictionarySize = options?.defaultDictionarySize || DEFAULT_DICTIONARY_SIZE;\n  }\n\n  async readHeader(): Promise<void> {\n    const buffer = await this.read(0, PARQUET_MAGIC.length);\n\n    const magic = buffer.toString();\n    switch (magic) {\n      case PARQUET_MAGIC:\n        break;\n      case PARQUET_MAGIC_ENCRYPTED:\n        throw new Error('Encrypted parquet file not supported');\n      default:\n        throw new Error(`Invalid parquet file (magic=${magic})`);\n    }\n  }\n\n  async readRowGroup(\n    schema: ParquetSchema,\n    rowGroup: RowGroup,\n    columnList: string[][]\n  ): Promise<ParquetBuffer> {\n    const buffer: ParquetBuffer = {\n      rowCount: Number(rowGroup.num_rows),\n      columnData: {}\n    };\n    for (const colChunk of rowGroup.columns) {\n      const colMetadata = colChunk.meta_data;\n      const colKey = colMetadata?.path_in_schema;\n      if (columnList.length > 0 && fieldIndexOf(columnList, colKey!) < 0) {\n        continue; // eslint-disable-line no-continue\n      }\n      buffer.columnData[colKey!.join()] = await this.readColumnChunk(schema, colChunk);\n    }\n    return buffer;\n  }\n\n  /**\n   * Do reading of parquet file's column chunk\n   * @param schema\n   * @param colChunk\n   */\n  async readColumnChunk(schema: ParquetSchema, colChunk: ColumnChunk): Promise<ParquetData> {\n    if (colChunk.file_path !== undefined && colChunk.file_path !== null) {\n      throw new Error('external references are not supported');\n    }\n\n    const field = schema.findField(colChunk.meta_data?.path_in_schema!);\n    const type: PrimitiveType = getThriftEnum(Type, colChunk.meta_data?.type!) as any;\n\n    if (type !== field.primitiveType) {\n      throw new Error(`chunk type not matching schema: ${type}`);\n    }\n\n    const compression: ParquetCompression = getThriftEnum(\n      CompressionCodec,\n      colChunk.meta_data?.codec!\n    ) as any;\n\n    const pagesOffset = Number(colChunk.meta_data?.data_page_offset!);\n    let pagesSize = Number(colChunk.meta_data?.total_compressed_size!);\n\n    if (!colChunk.file_path) {\n      pagesSize = Math.min(\n        this.fileSize - pagesOffset,\n        Number(colChunk.meta_data?.total_compressed_size)\n      );\n    }\n\n    const options: ParquetOptions = {\n      type,\n      rLevelMax: field.rLevelMax,\n      dLevelMax: field.dLevelMax,\n      compression,\n      column: field,\n      numValues: colChunk.meta_data?.num_values,\n      dictionary: []\n    };\n\n    let dictionary;\n\n    const dictionaryPageOffset = colChunk?.meta_data?.dictionary_page_offset;\n\n    if (dictionaryPageOffset) {\n      const dictionaryOffset = Number(dictionaryPageOffset);\n      // Getting dictionary from column chunk to iterate all over indexes to get dataPage values.\n      dictionary = await this.getDictionary(dictionaryOffset, options, pagesOffset);\n    }\n\n    dictionary = options.dictionary?.length ? options.dictionary : dictionary;\n    const pagesBuf = await this.read(pagesOffset, pagesSize);\n    return await decodeDataPages(pagesBuf, {...options, dictionary});\n  }\n\n  /**\n   * Getting dictionary for allows to flatten values by indices.\n   * @param dictionaryPageOffset\n   * @param options\n   * @param pagesOffset\n   * @returns\n   */\n  async getDictionary(\n    dictionaryPageOffset: number,\n    options: ParquetOptions,\n    pagesOffset: number\n  ): Promise<string[]> {\n    if (dictionaryPageOffset === 0) {\n      // dictionarySize = Math.min(this.fileSize - pagesOffset, this.defaultDictionarySize);\n      // pagesBuf = await this.read(pagesOffset, dictionarySize);\n\n      // In this case we are working with parquet-mr files format. Problem is described below:\n      // https://stackoverflow.com/questions/55225108/why-is-dictionary-page-offset-0-for-plain-dictionary-encoding\n      // We need to get dictionary page from column chunk if it exists.\n      // Now if we use code commented above we don't get DICTIONARY_PAGE we get DATA_PAGE instead.\n      return [];\n    }\n\n    const dictionarySize = Math.min(\n      this.fileSize - dictionaryPageOffset,\n      this.defaultDictionarySize\n    );\n    const pagesBuf = await this.read(dictionaryPageOffset, dictionarySize);\n\n    const cursor = {buffer: pagesBuf, offset: 0, size: pagesBuf.length};\n    const decodedPage = await decodePage(cursor, options);\n\n    return decodedPage.dictionary!;\n  }\n\n  async readFooter(): Promise<FileMetaData> {\n    const trailerLen = PARQUET_MAGIC.length + 4;\n    const trailerBuf = await this.read(this.fileSize - trailerLen, trailerLen);\n\n    const magic = trailerBuf.slice(4).toString();\n    if (magic !== PARQUET_MAGIC) {\n      throw new Error(`Not a valid parquet file (magic=\"${magic})`);\n    }\n\n    const metadataSize = trailerBuf.readUInt32LE(0);\n    const metadataOffset = this.fileSize - metadataSize - trailerLen;\n    if (metadataOffset < PARQUET_MAGIC.length) {\n      throw new Error(`Invalid metadata size ${metadataOffset}`);\n    }\n\n    const metadataBuf = await this.read(metadataOffset, metadataSize);\n    // let metadata = new parquet_thrift.FileMetaData();\n    // parquet_util.decodeThrift(metadata, metadataBuf);\n    const {metadata} = decodeFileMetadata(metadataBuf);\n    return metadata;\n  }\n}\n"],"file":"parquet-envelope-reader.js"}

package/dist/esm/parquetjs/parser/parquet-reader.js

@@ -0,0 +1,133 @@
+import _defineProperty from "@babel/runtime/helpers/esm/defineProperty";
+
+let _Symbol$asyncIterator;
+
+import { ParquetEnvelopeReader } from './parquet-envelope-reader';
+import { ParquetSchema } from '../schema/schema';
+import { ParquetCursor } from './parquet-cursor';
+import { PARQUET_VERSION } from '../../constants';
+import { decodeSchema } from './decoders';
+_Symbol$asyncIterator = Symbol.asyncIterator;
+export class ParquetReader {
+  static async openBlob(blob) {
+    const readFn = async (start, length) => {
+      const arrayBuffer = await blob.slice(start, start + length).arrayBuffer();
+      return Buffer.from(arrayBuffer);
+    };
+
+    const closeFn = async () => {};
+
+    const size = blob.size;
+    const envelopeReader = new ParquetEnvelopeReader(readFn, closeFn, size);
+
+    try {
+      await envelopeReader.readHeader();
+      const metadata = await envelopeReader.readFooter();
+      return new ParquetReader(metadata, envelopeReader);
+    } catch (err) {
+      await envelopeReader.close();
+      throw err;
+    }
+  }
+
+  static async openArrayBuffer(arrayBuffer) {
+    const readFn = async (start, length) => Buffer.from(arrayBuffer, start, length);
+
+    const closeFn = async () => {};
+
+    const size = arrayBuffer.byteLength;
+    const envelopeReader = new ParquetEnvelopeReader(readFn, closeFn, size);
+
+    try {
+      await envelopeReader.readHeader();
+      const metadata = await envelopeReader.readFooter();
+      return new ParquetReader(metadata, envelopeReader);
+    } catch (err) {
+      await envelopeReader.close();
+      throw err;
+    }
+  }
+
+  static async openFile(filePath) {
+    const envelopeReader = await ParquetEnvelopeReader.openFile(filePath);
+
+    try {
+      await envelopeReader.readHeader();
+      const metadata = await envelopeReader.readFooter();
+      return new ParquetReader(metadata, envelopeReader);
+    } catch (err) {
+      await envelopeReader.close();
+      throw err;
+    }
+  }
+
+  static async openBuffer(buffer) {
+    const envelopeReader = await ParquetEnvelopeReader.openBuffer(buffer);
+
+    try {
+      await envelopeReader.readHeader();
+      const metadata = await envelopeReader.readFooter();
+      return new ParquetReader(metadata, envelopeReader);
+    } catch (err) {
+      await envelopeReader.close();
+      throw err;
+    }
+  }
+
+  constructor(metadata, envelopeReader) {
+    _defineProperty(this, "metadata", void 0);
+
+    _defineProperty(this, "envelopeReader", void 0);
+
+    _defineProperty(this, "schema", void 0);
+
+    if (metadata.version !== PARQUET_VERSION) {
+      throw new Error('invalid parquet version');
+    }
+
+    this.metadata = metadata;
+    this.envelopeReader = envelopeReader;
+    const root = this.metadata.schema[0];
+    const {
+      schema
+    } = decodeSchema(this.metadata.schema, 1, root.num_children);
+    this.schema = new ParquetSchema(schema);
+  }
+
+  async close() {
+    await this.envelopeReader.close();
+  }
+
+  getCursor(columnList) {
+    if (!columnList) {
+      columnList = [];
+    }
+
+    columnList = columnList.map(x => Array.isArray(x) ? x : [x]);
+    return new ParquetCursor(this.metadata, this.envelopeReader, this.schema, columnList);
+  }
+
+  getRowCount() {
+    return Number(this.metadata.num_rows);
+  }
+
+  getSchema() {
+    return this.schema;
+  }
+
+  getMetadata() {
+    const md = {};
+
+    for (const kv of this.metadata.key_value_metadata) {
+      md[kv.key] = kv.value;
+    }
+
+    return md;
+  }
+
+  [_Symbol$asyncIterator]() {
+    return this.getCursor()[Symbol.asyncIterator]();
+  }
+
+}
+//# sourceMappingURL=parquet-reader.js.map

package/dist/esm/parquetjs/parser/parquet-reader.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../src/parquetjs/parser/parquet-reader.ts"],"names":["ParquetEnvelopeReader","ParquetSchema","ParquetCursor","PARQUET_VERSION","decodeSchema","Symbol","asyncIterator","ParquetReader","openBlob","blob","readFn","start","length","arrayBuffer","slice","Buffer","from","closeFn","size","envelopeReader","readHeader","metadata","readFooter","err","close","openArrayBuffer","byteLength","openFile","filePath","openBuffer","buffer","constructor","version","Error","root","schema","num_children","getCursor","columnList","map","x","Array","isArray","getRowCount","Number","num_rows","getSchema","getMetadata","md","kv","key_value_metadata","key","value"],"mappings":";;;;AACA,SAAQA,qBAAR,QAAoC,2BAApC;AAEA,SAAQC,aAAR,QAA4B,kBAA5B;AACA,SAAQC,aAAR,QAA4B,kBAA5B;AACA,SAAQC,eAAR,QAA8B,iBAA9B;AACA,SAAQC,YAAR,QAA2B,YAA3B;wBAyKGC,MAAM,CAACC,a;AAhKV,OAAO,MAAMC,aAAN,CAAmD;AAInC,eAARC,QAAQ,CAAIC,IAAJ,EAA2C;AAC9D,UAAMC,MAAM,GAAG,OAAOC,KAAP,EAAsBC,MAAtB,KAAyC;AACtD,YAAMC,WAAW,GAAG,MAAMJ,IAAI,CAACK,KAAL,CAAWH,KAAX,EAAkBA,KAAK,GAAGC,MAA1B,EAAkCC,WAAlC,EAA1B;AACA,aAAOE,MAAM,CAACC,IAAP,CAAYH,WAAZ,CAAP;AACD,KAHD;;AAIA,UAAMI,OAAO,GAAG,YAAY,CAAE,CAA9B;;AACA,UAAMC,IAAI,GAAGT,IAAI,CAACS,IAAlB;AACA,UAAMC,cAAc,GAAG,IAAInB,qBAAJ,CAA0BU,MAA1B,EAAkCO,OAAlC,EAA2CC,IAA3C,CAAvB;;AACA,QAAI;AACF,YAAMC,cAAc,CAACC,UAAf,EAAN;AACA,YAAMC,QAAQ,GAAG,MAAMF,cAAc,CAACG,UAAf,EAAvB;AACA,aAAO,IAAIf,aAAJ,CAAkBc,QAAlB,EAA4BF,cAA5B,CAAP;AACD,KAJD,CAIE,OAAOI,GAAP,EAAY;AACZ,YAAMJ,cAAc,CAACK,KAAf,EAAN;AACA,YAAMD,GAAN;AACD;AACF;;AAK2B,eAAfE,eAAe,CAAIZ,WAAJ,EAAyD;AACnF,UAAMH,MAAM,GAAG,OAAOC,KAAP,EAAsBC,MAAtB,KAAyCG,MAAM,CAACC,IAAP,CAAYH,WAAZ,EAAyBF,KAAzB,EAAgCC,MAAhC,CAAxD;;AACA,UAAMK,OAAO,GAAG,YAAY,CAAE,CAA9B;;AACA,UAAMC,IAAI,GAAGL,WAAW,CAACa,UAAzB;AACA,UAAMP,cAAc,GAAG,IAAInB,qBAAJ,CAA0BU,MAA1B,EAAkCO,OAAlC,EAA2CC,IAA3C,CAAvB;;AACA,QAAI;AACF,YAAMC,cAAc,CAACC,UAAf,EAAN;AACA,YAAMC,QAAQ,GAAG,MAAMF,cAAc,CAACG,UAAf,EAAvB;AACA,aAAO,IAAIf,aAAJ,CAAkBc,QAAlB,EAA4BF,cAA5B,CAAP;AACD,KAJD,CAIE,OAAOI,GAAP,EAAY;AACZ,YAAMJ,cAAc,CAACK,KAAf,EAAN;AACA,YAAMD,GAAN;AACD;AACF;;AAMoB,eAARI,QAAQ,CAAIC,QAAJ,EAAiD;AACpE,UAAMT,cAAc,GAAG,MAAMnB,qBAAqB,CAAC2B,QAAtB,CAA+BC,QAA/B,CAA7B;;AACA,QAAI;AACF,YAAMT,cAAc,CAACC,UAAf,EAAN;AACA,YAAMC,QAAQ,GAAG,MAAMF,cAAc,CAACG,UAAf,EAAvB;AACA,aAAO,IAAIf,aAAJ,CAAqBc,QAArB,EAA+BF,cAA/B,CAAP;AACD,KAJD,CAIE,OAAOI,GAAP,EAAY;AACZ,YAAMJ,cAAc,CAACK,KAAf,EAAN;AACA,YAAMD,GAAN;AACD;AACF;;AAEsB,eAAVM,UAAU,CAAIC,MAAJ,EAA+C;AACpE,UAAMX,cAAc,GAAG,MAAMnB,qBAAqB,CAAC6B,UAAtB,CAAiCC,MAAjC,CAA7B;;AACA,QAAI;AACF,YAAMX,cAAc,CAACC,UAAf,EAAN;AACA,YAAMC,QAAQ,GAAG,MAAMF,cAAc,CAACG,UAAf,EAAvB;AACA,aAAO,IAAIf,aAAJ,CAAqBc,QAArB,EAA+BF,cAA/B,CAAP;AACD,KAJD,CAIE,OAAOI,GAAP,EAAY;AACZ,YAAMJ,cAAc,CAACK,KAAf,EAAN;AACA,YAAMD,GAAN;AACD;AACF;;AAYDQ,EAAAA,WAAW,CAACV,QAAD,EAAyBF,cAAzB,EAAgE;AAAA;;AAAA;;AAAA;;AACzE,QAAIE,QAAQ,CAACW,OAAT,KAAqB7B,eAAzB,EAA0C;AACxC,YAAM,IAAI8B,KAAJ,CAAU,yBAAV,CAAN;AACD;;AAED,SAAKZ,QAAL,GAAgBA,QAAhB;AACA,SAAKF,cAAL,GAAsBA,cAAtB;AACA,UAAMe,IAAI,GAAG,KAAKb,QAAL,CAAcc,MAAd,CAAqB,CAArB,CAAb;AACA,UAAM;AAACA,MAAAA;AAAD,QAAW/B,YAAY,CAAC,KAAKiB,QAAL,CAAcc,MAAf,EAAuB,CAAvB,EAA0BD,IAAI,CAACE,YAA/B,CAA7B;AACA,SAAKD,MAAL,GAAc,IAAIlC,aAAJ,CAAkBkC,MAAlB,CAAd;AACD;;AAMU,QAALX,KAAK,GAAkB;AAC3B,UAAM,KAAKL,cAAL,CAAoBK,KAApB,EAAN;AAGD;;AAeDa,EAAAA,SAAS,CAACC,UAAD,EAAgE;AACvE,QAAI,CAACA,UAAL,EAAiB;AAEfA,MAAAA,UAAU,GAAG,EAAb;AACD;;AAGDA,IAAAA,UAAU,GAAGA,UAAU,CAACC,GAAX,CAAgBC,CAAD,IAAQC,KAAK,CAACC,OAAN,CAAcF,CAAd,IAAmBA,CAAnB,GAAuB,CAACA,CAAD,CAA9C,CAAb;AAEA,WAAO,IAAItC,aAAJ,CACL,KAAKmB,QADA,EAEL,KAAKF,cAFA,EAGL,KAAKgB,MAHA,EAILG,UAJK,CAAP;AAMD;;AAMDK,EAAAA,WAAW,GAAW;AACpB,WAAOC,MAAM,CAAC,KAAKvB,QAAL,CAAcwB,QAAf,CAAb;AACD;;AAKDC,EAAAA,SAAS,GAAkB;AACzB,WAAO,KAAKX,MAAZ;AACD;;AAKDY,EAAAA,WAAW,GAA2B;AACpC,UAAMC,EAA0B,GAAG,EAAnC;;AACA,SAAK,MAAMC,EAAX,IAAiB,KAAK5B,QAAL,CAAc6B,kBAA/B,EAAoD;AAClDF,MAAAA,EAAE,CAACC,EAAE,CAACE,GAAJ,CAAF,GAAaF,EAAE,CAACG,KAAhB;AACD;;AACD,WAAOJ,EAAP;AACD;;AAMD,4BAA2C;AACzC,WAAO,KAAKX,SAAL,GAAiBhC,MAAM,CAACC,aAAxB,GAAP;AACD;;AAlKuD","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\nimport {ParquetEnvelopeReader} from './parquet-envelope-reader';\nimport {FileMetaData} from '../parquet-thrift';\nimport {ParquetSchema} from '../schema/schema';\nimport {ParquetCursor} from './parquet-cursor';\nimport {PARQUET_VERSION} from '../../constants';\nimport {decodeSchema} from './decoders';\n\n/**\n * A parquet reader allows retrieving the rows from a parquet file in order.\n * The basic usage is to create a reader and then retrieve a cursor/iterator\n * which allows you to consume row after row until all rows have been read. It is\n * important that you call close() after you are finished reading the file to\n * avoid leaking file descriptors.\n */\nexport class ParquetReader<T> implements AsyncIterable<T> {\n  /**\n   * return a new parquet reader initialized with a read function\n   */\n  static async openBlob<T>(blob: Blob): Promise<ParquetReader<T>> {\n    const readFn = async (start: number, length: number) => {\n      const arrayBuffer = await blob.slice(start, start + length).arrayBuffer();\n      return Buffer.from(arrayBuffer);\n    };\n    const closeFn = async () => {};\n    const size = blob.size;\n    const envelopeReader = new ParquetEnvelopeReader(readFn, closeFn, size);\n    try {\n      await envelopeReader.readHeader();\n      const metadata = await envelopeReader.readFooter();\n      return new ParquetReader(metadata, envelopeReader);\n    } catch (err) {\n      await envelopeReader.close();\n      throw err;\n    }\n  }\n\n  /**\n   * return a new parquet reader initialized with a read function\n   */\n  static async openArrayBuffer<T>(arrayBuffer: ArrayBuffer): Promise<ParquetReader<T>> {\n    const readFn = async (start: number, length: number) => Buffer.from(arrayBuffer, start, length);\n    const closeFn = async () => {};\n    const size = arrayBuffer.byteLength;\n    const envelopeReader = new ParquetEnvelopeReader(readFn, closeFn, size);\n    try {\n      await envelopeReader.readHeader();\n      const metadata = await envelopeReader.readFooter();\n      return new ParquetReader(metadata, envelopeReader);\n    } catch (err) {\n      await envelopeReader.close();\n      throw err;\n    }\n  }\n\n  /**\n   * Open the parquet file pointed to by the specified path and return a new\n   * parquet reader\n   */\n  static async openFile<T>(filePath: string): Promise<ParquetReader<T>> {\n    const envelopeReader = await ParquetEnvelopeReader.openFile(filePath);\n    try {\n      await envelopeReader.readHeader();\n      const metadata = await envelopeReader.readFooter();\n      return new ParquetReader<T>(metadata, envelopeReader);\n    } catch (err) {\n      await envelopeReader.close();\n      throw err;\n    }\n  }\n\n  static async openBuffer<T>(buffer: Buffer): Promise<ParquetReader<T>> {\n    const envelopeReader = await ParquetEnvelopeReader.openBuffer(buffer);\n    try {\n      await envelopeReader.readHeader();\n      const metadata = await envelopeReader.readFooter();\n      return new ParquetReader<T>(metadata, envelopeReader);\n    } catch (err) {\n      await envelopeReader.close();\n      throw err;\n    }\n  }\n\n  public metadata: FileMetaData;\n  public envelopeReader: ParquetEnvelopeReader;\n  public schema: ParquetSchema;\n\n  /**\n   * Create a new parquet reader from the file metadata and an envelope reader.\n   * It is not recommended to call this constructor directly except for advanced\n   * and internal use cases. Consider using one of the open{File,Buffer} methods\n   * instead\n   */\n  constructor(metadata: FileMetaData, envelopeReader: ParquetEnvelopeReader) {\n    if (metadata.version !== PARQUET_VERSION) {\n      throw new Error('invalid parquet version');\n    }\n\n    this.metadata = metadata;\n    this.envelopeReader = envelopeReader;\n    const root = this.metadata.schema[0];\n    const {schema} = decodeSchema(this.metadata.schema, 1, root.num_children!);\n    this.schema = new ParquetSchema(schema);\n  }\n\n  /**\n   * Close this parquet reader. You MUST call this method once you're finished\n   * reading rows\n   */\n  async close(): Promise<void> {\n    await this.envelopeReader.close();\n    // this.envelopeReader = null;\n    // this.metadata = null;\n  }\n\n  /**\n   * Return a cursor to the file. You may open more than one cursor and use\n   * them concurrently. All cursors become invalid once close() is called on\n   * the reader object.\n   *\n   * The required_columns parameter controls which columns are actually read\n   * from disk. An empty array or no value implies all columns. A list of column\n   * names means that only those columns should be loaded from disk.\n   */\n  getCursor(): ParquetCursor<T>;\n  // @ts-ignore\n  getCursor<K extends keyof T>(columnList: (K | K[])[]): ParquetCursor<Pick<T, K>>;\n  getCursor(columnList: (string | string[])[]): ParquetCursor<Partial<T>>;\n  getCursor(columnList?: (string | string[])[]): ParquetCursor<Partial<T>> {\n    if (!columnList) {\n      // tslint:disable-next-line:no-parameter-reassignment\n      columnList = [];\n    }\n\n    // tslint:disable-next-line:no-parameter-reassignment\n    columnList = columnList.map((x) => (Array.isArray(x) ? x : [x]));\n\n    return new ParquetCursor<T>(\n      this.metadata,\n      this.envelopeReader,\n      this.schema,\n      columnList as string[][]\n    );\n  }\n\n  /**\n   * Return the number of rows in this file. Note that the number of rows is\n   * not neccessarily equal to the number of rows in each column.\n   */\n  getRowCount(): number {\n    return Number(this.metadata.num_rows);\n  }\n\n  /**\n   * Returns the ParquetSchema for this file\n   */\n  getSchema(): ParquetSchema {\n    return this.schema;\n  }\n\n  /**\n   * Returns the user (key/value) metadata for this file\n   */\n  getMetadata(): Record<string, string> {\n    const md: Record<string, string> = {};\n    for (const kv of this.metadata.key_value_metadata!) {\n      md[kv.key] = kv.value!;\n    }\n    return md;\n  }\n\n  /**\n   * Implement AsyncIterable\n   */\n  // tslint:disable-next-line:function-name\n  [Symbol.asyncIterator](): AsyncIterator<T> {\n    return this.getCursor()[Symbol.asyncIterator]();\n  }\n}\n"],"file":"parquet-reader.js"}

package/dist/esm/parquetjs/schema/declare.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../../../src/parquetjs/schema/declare.ts"],"names":["ParquetBuffer","constructor","rowCount","columnData"],"mappings":";AAqFA,OAAO,MAAMA,aAAN,CAAoB;AAGzBC,EAAAA,WAAW,CAACC,QAAgB,GAAG,CAApB,EAAuBC,UAAuC,GAAG,EAAjE,EAAqE;AAAA;;AAAA;;AAC9E,SAAKD,QAAL,GAAgBA,QAAhB;AACA,SAAKC,UAAL,GAAkBA,UAAlB;AACD;;AANwB","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\n\nexport type ParquetCodec = 'PLAIN' | 'RLE';\nexport type ParquetCompression = 'UNCOMPRESSED' | 'GZIP' | 'SNAPPY' | 'LZO' | 'BROTLI' | 'LZ4';\nexport type RepetitionType = 'REQUIRED' | 'OPTIONAL' | 'REPEATED';\nexport type ParquetType = PrimitiveType | OriginalType;\n\nexport type PrimitiveType =\n  // Base Types\n  | 'BOOLEAN' // 0\n  | 'INT32' // 1\n  | 'INT64' // 2\n  | 'INT96' // 3\n  | 'FLOAT' // 4\n  | 'DOUBLE' // 5\n  | 'BYTE_ARRAY' // 6,\n  | 'FIXED_LEN_BYTE_ARRAY'; // 7\n\nexport type OriginalType =\n  // Converted Types\n  | 'UTF8' // 0\n  // | 'MAP' // 1\n  // | 'MAP_KEY_VALUE' // 2\n  // | 'LIST' // 3\n  // | 'ENUM' // 4\n  // | 'DECIMAL' // 5\n  | 'DATE' // 6\n  | 'TIME_MILLIS' // 7\n  | 'TIME_MICROS' // 8\n  | 'TIMESTAMP_MILLIS' // 9\n  | 'TIMESTAMP_MICROS' // 10\n  | 'UINT_8' // 11\n  | 'UINT_16' // 12\n  | 'UINT_32' // 13\n  | 'UINT_64' // 14\n  | 'INT_8' // 15\n  | 'INT_16' // 16\n  | 'INT_32' // 17\n  | 'INT_64' // 18\n  | 'JSON' // 19\n  | 'BSON' // 20\n  | 'INTERVAL'; // 21\n\nexport interface SchemaDefinition {\n  [string: string]: FieldDefinition;\n}\n\nexport interface FieldDefinition {\n  type?: ParquetType;\n  typeLength?: number;\n  encoding?: ParquetCodec;\n  compression?: ParquetCompression;\n  optional?: boolean;\n  repeated?: boolean;\n  fields?: SchemaDefinition;\n}\n\nexport interface ParquetField {\n  name: string;\n  path: string[];\n  key: string;\n  primitiveType?: PrimitiveType;\n  originalType?: OriginalType;\n  repetitionType: RepetitionType;\n  typeLength?: number;\n  encoding?: ParquetCodec;\n  compression?: ParquetCompression;\n  rLevelMax: number;\n  dLevelMax: number;\n  isNested?: boolean;\n  fieldCount?: number;\n  fields?: Record<string, ParquetField>;\n}\n\nexport interface ParquetData {\n  dlevels: number[];\n  rlevels: number[];\n  values: any[];\n  count: number;\n}\n\nexport interface ParquetRecord {\n  [key: string]: any;\n}\n\nexport class ParquetBuffer {\n  rowCount: number;\n  columnData: Record<string, ParquetData>;\n  constructor(rowCount: number = 0, columnData: Record<string, ParquetData> = {}) {\n    this.rowCount = rowCount;\n    this.columnData = columnData;\n  }\n}\n"],"file":"declare.js"}
+{"version":3,"sources":["../../../../src/parquetjs/schema/declare.ts"],"names":["ParquetBuffer","constructor","rowCount","columnData"],"mappings":";AAmIA,OAAO,MAAMA,aAAN,CAAoB;AAGzBC,EAAAA,WAAW,CAACC,QAAgB,GAAG,CAApB,EAAuBC,UAAuC,GAAG,EAAjE,EAAqE;AAAA;;AAAA;;AAC9E,SAAKD,QAAL,GAAgBA,QAAhB;AACA,SAAKC,UAAL,GAAkBA,UAAlB;AACD;;AANwB","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\nimport Int64 from 'node-int64';\nimport type {PageHeader} from '../parquet-thrift';\n\nexport type ParquetCodec = 'PLAIN' | 'RLE' | 'PLAIN_DICTIONARY';\nexport type ParquetCompression =\n  | 'UNCOMPRESSED'\n  | 'GZIP'\n  | 'SNAPPY'\n  | 'LZO'\n  | 'BROTLI'\n  | 'LZ4'\n  | 'LZ4_RAW'\n  | 'ZSTD';\nexport type RepetitionType = 'REQUIRED' | 'OPTIONAL' | 'REPEATED';\nexport type ParquetType = PrimitiveType | OriginalType;\n\n/**\n * Physical type\n */\nexport type PrimitiveType =\n  // Base Types\n  | 'BOOLEAN' // 0\n  | 'INT32' // 1\n  | 'INT64' // 2\n  | 'INT96' // 3\n  | 'FLOAT' // 4\n  | 'DOUBLE' // 5\n  | 'BYTE_ARRAY' // 6,\n  | 'FIXED_LEN_BYTE_ARRAY'; // 7\n\n/**\n * Logical type\n */\nexport type OriginalType =\n  // Converted Types\n  | 'UTF8' // 0\n  // | 'MAP' // 1\n  // | 'MAP_KEY_VALUE' // 2\n  // | 'LIST' // 3\n  // | 'ENUM' // 4\n  // | 'DECIMAL' // 5\n  | 'DECIMAL_INT32' // 5\n  | 'DECIMAL_INT64' // 5\n  | 'DECIMAL_BYTE_ARRAY' // 5\n  | 'DECIMAL_FIXED_LEN_BYTE_ARRAY' // 5\n  | 'DATE' // 6\n  | 'TIME_MILLIS' // 7\n  | 'TIME_MICROS' // 8\n  | 'TIMESTAMP_MILLIS' // 9\n  | 'TIMESTAMP_MICROS' // 10\n  | 'UINT_8' // 11\n  | 'UINT_16' // 12\n  | 'UINT_32' // 13\n  | 'UINT_64' // 14\n  | 'INT_8' // 15\n  | 'INT_16' // 16\n  | 'INT_32' // 17\n  | 'INT_64' // 18\n  | 'JSON' // 19\n  | 'BSON' // 20\n  | 'INTERVAL'; // 21\n\nexport type ParquetDictionary = string[];\n\nexport interface SchemaDefinition {\n  [string: string]: FieldDefinition;\n}\n\nexport interface FieldDefinition {\n  type?: ParquetType;\n  typeLength?: number;\n  presision?: number;\n  scale?: number;\n  encoding?: ParquetCodec;\n  compression?: ParquetCompression;\n  optional?: boolean;\n  repeated?: boolean;\n  fields?: SchemaDefinition;\n}\n\nexport interface ParquetField {\n  name: string;\n  path: string[];\n  key: string;\n  primitiveType?: PrimitiveType;\n  originalType?: OriginalType;\n  repetitionType: RepetitionType;\n  typeLength?: number;\n  presision?: number;\n  scale?: number;\n  encoding?: ParquetCodec;\n  compression?: ParquetCompression;\n  rLevelMax: number;\n  dLevelMax: number;\n  isNested?: boolean;\n  fieldCount?: number;\n  fields?: Record<string, ParquetField>;\n}\n\nexport interface ParquetOptions {\n  type: ParquetType;\n  rLevelMax: number;\n  dLevelMax: number;\n  compression: ParquetCompression;\n  column: ParquetField;\n  numValues?: Int64;\n  dictionary?: ParquetDictionary;\n}\n\nexport interface ParquetData {\n  dlevels: number[];\n  rlevels: number[];\n  values: any[];\n  count: number;\n  pageHeaders: PageHeader[];\n}\n\nexport interface ParquetPageData {\n  dlevels: number[];\n  rlevels: number[];\n  values: any[];\n  count: number;\n  dictionary?: ParquetDictionary;\n  pageHeader: PageHeader;\n}\n\nexport interface ParquetRecord {\n  [key: string]: any;\n}\n\nexport class ParquetBuffer {\n  rowCount: number;\n  columnData: Record<string, ParquetData>;\n  constructor(rowCount: number = 0, columnData: Record<string, ParquetData> = {}) {\n    this.rowCount = rowCount;\n    this.columnData = columnData;\n  }\n}\n"],"file":"declare.js"}

package/dist/esm/parquetjs/schema/schema.js

@@ -150,6 +150,8 @@ function buildFields(schema, rLevelParentMax, dLevelParentMax, path) {
       encoding: opts.encoding,
       compression: opts.compression,
       typeLength: opts.typeLength || typeDef.typeLength,
+      presision: opts.presision,
+      scale: opts.scale,
       rLevelMax,
       dLevelMax
     };

package/dist/esm/parquetjs/schema/schema.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../../../src/parquetjs/schema/schema.ts"],"names":["PARQUET_CODECS","PARQUET_COMPRESSION_METHODS","materializeRecords","shredBuffer","shredRecord","PARQUET_LOGICAL_TYPES","ParquetSchema","constructor","schema","fields","buildFields","fieldList","listFields","findField","path","split","slice","n","length","shift","findFieldBranch","branch","push","record","buffer","compress","type","setCompress","name","node","compression","rLevelParentMax","dLevelParentMax","opts","required","optional","repeated","Boolean","rLevelMax","dLevelMax","repetitionType","cpath","concat","key","join","isNested","fieldCount","Object","keys","typeDef","Error","encoding","primitiveType","originalType","typeLength","list","k"],"mappings":";AAEA,SAAQA,cAAR,QAA6B,WAA7B;AACA,SAAQC,2BAAR,QAA0C,gBAA1C;AAUA,SAAQC,kBAAR,EAA4BC,WAA5B,EAAyCC,WAAzC,QAA2D,SAA3D;AACA,SAAQC,qBAAR,QAAoC,SAApC;AAKA,OAAO,MAAMC,aAAN,CAAoB;AAQzBC,EAAAA,WAAW,CAACC,MAAD,EAA2B;AAAA;;AAAA;;AAAA;;AACpC,SAAKA,MAAL,GAAcA,MAAd;AACA,SAAKC,MAAL,GAAcC,WAAW,CAACF,MAAD,EAAS,CAAT,EAAY,CAAZ,EAAe,EAAf,CAAzB;AACA,SAAKG,SAAL,GAAiBC,UAAU,CAAC,KAAKH,MAAN,CAA3B;AACD;;AAKDI,EAAAA,SAAS,CAACC,IAAD,EAAwC;AAC/C,QAAI,OAAOA,IAAP,KAAgB,QAApB,EAA8B;AAE5BA,MAAAA,IAAI,GAAGA,IAAI,CAACC,KAAL,CAAW,GAAX,CAAP;AACD,KAHD,MAGO;AAELD,MAAAA,IAAI,GAAGA,IAAI,CAACE,KAAL,CAAW,CAAX,CAAP;AACD;;AAED,QAAIC,CAAC,GAAG,KAAKR,MAAb;;AACA,WAAOK,IAAI,CAACI,MAAL,GAAc,CAArB,EAAwBJ,IAAI,CAACK,KAAL,EAAxB,EAAsC;AACpCF,MAAAA,CAAC,GAAGA,CAAC,CAACH,IAAI,CAAC,CAAD,CAAL,CAAD,CAAWL,MAAf;AACD;;AAED,WAAOQ,CAAC,CAACH,IAAI,CAAC,CAAD,CAAL,CAAR;AACD;;AAKDM,EAAAA,eAAe,CAACN,IAAD,EAA0C;AACvD,QAAI,OAAOA,IAAP,KAAgB,QAApB,EAA8B;AAE5BA,MAAAA,IAAI,GAAGA,IAAI,CAACC,KAAL,CAAW,GAAX,CAAP;AACD;;AACD,UAAMM,MAAsB,GAAG,EAA/B;AACA,QAAIJ,CAAC,GAAG,KAAKR,MAAb;;AACA,WAAOK,IAAI,CAACI,MAAL,GAAc,CAArB,EAAwBJ,IAAI,CAACK,KAAL,EAAxB,EAAsC;AACpCE,MAAAA,MAAM,CAACC,IAAP,CAAYL,CAAC,CAACH,IAAI,CAAC,CAAD,CAAL,CAAb;;AACA,UAAIA,IAAI,CAACI,MAAL,GAAc,CAAlB,EAAqB;AACnBD,QAAAA,CAAC,GAAGA,CAAC,CAACH,IAAI,CAAC,CAAD,CAAL,CAAD,CAAWL,MAAf;AACD;AACF;;AACD,WAAOY,MAAP;AACD;;AAEDjB,EAAAA,WAAW,CAACmB,MAAD,EAAwBC,MAAxB,EAAqD;AAC9DpB,IAAAA,WAAW,CAAC,IAAD,EAAOmB,MAAP,EAAeC,MAAf,CAAX;AACD;;AAEDtB,EAAAA,kBAAkB,CAACsB,MAAD,EAAyC;AACzD,WAAOtB,kBAAkB,CAAC,IAAD,EAAOsB,MAAP,CAAzB;AACD;;AAEDC,EAAAA,QAAQ,CAACC,IAAD,EAAiC;AACvCC,IAAAA,WAAW,CAAC,KAAKnB,MAAN,EAAckB,IAAd,CAAX;AACAC,IAAAA,WAAW,CAAC,KAAKlB,MAAN,EAAciB,IAAd,CAAX;AACA,WAAO,IAAP;AACD;;AAEDF,EAAAA,MAAM,GAAkB;AACtB,WAAOrB,WAAW,CAAC,IAAD,CAAlB;AACD;;AArEwB;;AAwE3B,SAASwB,WAAT,CAAqBnB,MAArB,EAAkCkB,IAAlC,EAA4D;AAC1D,OAAK,MAAME,IAAX,IAAmBpB,MAAnB,EAA2B;AACzB,UAAMqB,IAAI,GAAGrB,MAAM,CAACoB,IAAD,CAAnB;;AACA,QAAIC,IAAI,CAACpB,MAAT,EAAiB;AACfkB,MAAAA,WAAW,CAACE,IAAI,CAACpB,MAAN,EAAciB,IAAd,CAAX;AACD,KAFD,MAEO;AACLG,MAAAA,IAAI,CAACC,WAAL,GAAmBJ,IAAnB;AACD;AACF;AACF;;AAGD,SAAShB,WAAT,CACEF,MADF,EAEEuB,eAFF,EAGEC,eAHF,EAIElB,IAJF,EAKgC;AAC9B,QAAMH,SAAuC,GAAG,EAAhD;;AAEA,OAAK,MAAMiB,IAAX,IAAmBpB,MAAnB,EAA2B;AACzB,UAAMyB,IAAI,GAAGzB,MAAM,CAACoB,IAAD,CAAnB;AAGA,UAAMM,QAAQ,GAAG,CAACD,IAAI,CAACE,QAAvB;AACA,UAAMC,QAAQ,GAAGC,OAAO,CAACJ,IAAI,CAACG,QAAN,CAAxB;AACA,QAAIE,SAAS,GAAGP,eAAhB;AACA,QAAIQ,SAAS,GAAGP,eAAhB;AAEA,QAAIQ,cAA8B,GAAG,UAArC;;AACA,QAAI,CAACN,QAAL,EAAe;AACbM,MAAAA,cAAc,GAAG,UAAjB;AACAD,MAAAA,SAAS;AACV;;AACD,QAAIH,QAAJ,EAAc;AACZI,MAAAA,cAAc,GAAG,UAAjB;AACAF,MAAAA,SAAS;AACT,UAAIJ,QAAJ,EAAcK,SAAS;AACxB;;AAGD,QAAIN,IAAI,CAACxB,MAAT,EAAiB;AACf,YAAMgC,KAAK,GAAG3B,IAAI,CAAC4B,MAAL,CAAY,CAACd,IAAD,CAAZ,CAAd;AACAjB,MAAAA,SAAS,CAACiB,IAAD,CAAT,GAAkB;AAChBA,QAAAA,IADgB;AAEhBd,QAAAA,IAAI,EAAE2B,KAFU;AAGhBE,QAAAA,GAAG,EAAEF,KAAK,CAACG,IAAN,EAHW;AAIhBJ,QAAAA,cAJgB;AAKhBF,QAAAA,SALgB;AAMhBC,QAAAA,SANgB;AAOhBM,QAAAA,QAAQ,EAAE,IAPM;AAQhBC,QAAAA,UAAU,EAAEC,MAAM,CAACC,IAAP,CAAYf,IAAI,CAACxB,MAAjB,EAAyBS,MARrB;AAShBT,QAAAA,MAAM,EAAEC,WAAW,CAACuB,IAAI,CAACxB,MAAN,EAAc6B,SAAd,EAAyBC,SAAzB,EAAoCE,KAApC;AATH,OAAlB;AAWA;AACD;;AAED,UAAMQ,OAAY,GAAG5C,qBAAqB,CAAC4B,IAAI,CAACP,IAAN,CAA1C;;AACA,QAAI,CAACuB,OAAL,EAAc;AACZ,YAAM,IAAIC,KAAJ,CAAW,yBAAwBjB,IAAI,CAACP,IAAK,EAA7C,CAAN;AACD;;AAEDO,IAAAA,IAAI,CAACkB,QAAL,GAAgBlB,IAAI,CAACkB,QAAL,IAAiB,OAAjC;;AACA,QAAI,EAAElB,IAAI,CAACkB,QAAL,IAAiBnD,cAAnB,CAAJ,EAAwC;AACtC,YAAM,IAAIkD,KAAJ,CAAW,iCAAgCjB,IAAI,CAACkB,QAAS,EAAzD,CAAN;AACD;;AAEDlB,IAAAA,IAAI,CAACH,WAAL,GAAmBG,IAAI,CAACH,WAAL,IAAoB,cAAvC;;AACA,QAAI,EAAEG,IAAI,CAACH,WAAL,IAAoB7B,2BAAtB,CAAJ,EAAwD;AACtD,YAAM,IAAIiD,KAAJ,CAAW,mCAAkCjB,IAAI,CAACH,WAAY,EAA9D,CAAN;AACD;;AAGD,UAAMW,KAAK,GAAG3B,IAAI,CAAC4B,MAAL,CAAY,CAACd,IAAD,CAAZ,CAAd;AACAjB,IAAAA,SAAS,CAACiB,IAAD,CAAT,GAAkB;AAChBA,MAAAA,IADgB;AAEhBwB,MAAAA,aAAa,EAAEH,OAAO,CAACG,aAFP;AAGhBC,MAAAA,YAAY,EAAEJ,OAAO,CAACI,YAHN;AAIhBvC,MAAAA,IAAI,EAAE2B,KAJU;AAKhBE,MAAAA,GAAG,EAAEF,KAAK,CAACG,IAAN,EALW;AAMhBJ,MAAAA,cANgB;AAOhBW,MAAAA,QAAQ,EAAElB,IAAI,CAACkB,QAPC;AAQhBrB,MAAAA,WAAW,EAAEG,IAAI,CAACH,WARF;AAShBwB,MAAAA,UAAU,EAAErB,IAAI,CAACqB,UAAL,IAAmBL,OAAO,CAACK,UATvB;AAUhBhB,MAAAA,SAVgB;AAWhBC,MAAAA;AAXgB,KAAlB;AAaD;;AACD,SAAO5B,SAAP;AACD;;AAED,SAASC,UAAT,CAAoBH,MAApB,EAA0E;AACxE,MAAI8C,IAAoB,GAAG,EAA3B;;AACA,OAAK,MAAMC,CAAX,IAAgB/C,MAAhB,EAAwB;AACtB8C,IAAAA,IAAI,CAACjC,IAAL,CAAUb,MAAM,CAAC+C,CAAD,CAAhB;;AACA,QAAI/C,MAAM,CAAC+C,CAAD,CAAN,CAAUX,QAAd,EAAwB;AACtBU,MAAAA,IAAI,GAAGA,IAAI,CAACb,MAAL,CAAY9B,UAAU,CAACH,MAAM,CAAC+C,CAAD,CAAN,CAAU/C,MAAX,CAAtB,CAAP;AACD;AACF;;AACD,SAAO8C,IAAP;AACD","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\n\nimport {PARQUET_CODECS} from '../codecs';\nimport {PARQUET_COMPRESSION_METHODS} from '../compression';\nimport {\n  FieldDefinition,\n  ParquetBuffer,\n  ParquetCompression,\n  ParquetField,\n  ParquetRecord,\n  RepetitionType,\n  SchemaDefinition\n} from './declare';\nimport {materializeRecords, shredBuffer, shredRecord} from './shred';\nimport {PARQUET_LOGICAL_TYPES} from './types';\n\n/**\n * A parquet file schema\n */\nexport class ParquetSchema {\n  public schema: Record<string, FieldDefinition>;\n  public fields: Record<string, ParquetField>;\n  public fieldList: ParquetField[];\n\n  /**\n   * Create a new schema from a JSON schema definition\n   */\n  constructor(schema: SchemaDefinition) {\n    this.schema = schema;\n    this.fields = buildFields(schema, 0, 0, []);\n    this.fieldList = listFields(this.fields);\n  }\n\n  /**\n   * Retrieve a field definition\n   */\n  findField(path: string | string[]): ParquetField {\n    if (typeof path === 'string') {\n      // tslint:disable-next-line:no-parameter-reassignment\n      path = path.split(',');\n    } else {\n      // tslint:disable-next-line:no-parameter-reassignment\n      path = path.slice(0); // clone array\n    }\n\n    let n = this.fields;\n    for (; path.length > 1; path.shift()) {\n      n = n[path[0]].fields as Record<string, ParquetField>;\n    }\n\n    return n[path[0]];\n  }\n\n  /**\n   * Retrieve a field definition and all the field's ancestors\n   */\n  findFieldBranch(path: string | string[]): ParquetField[] {\n    if (typeof path === 'string') {\n      // tslint:disable-next-line:no-parameter-reassignment\n      path = path.split(',');\n    }\n    const branch: ParquetField[] = [];\n    let n = this.fields;\n    for (; path.length > 0; path.shift()) {\n      branch.push(n[path[0]]);\n      if (path.length > 1) {\n        n = n[path[0]].fields as Record<string, ParquetField>;\n      }\n    }\n    return branch;\n  }\n\n  shredRecord(record: ParquetRecord, buffer: ParquetBuffer): void {\n    shredRecord(this, record, buffer);\n  }\n\n  materializeRecords(buffer: ParquetBuffer): ParquetRecord[] {\n    return materializeRecords(this, buffer);\n  }\n\n  compress(type: ParquetCompression): this {\n    setCompress(this.schema, type);\n    setCompress(this.fields, type);\n    return this;\n  }\n\n  buffer(): ParquetBuffer {\n    return shredBuffer(this);\n  }\n}\n\nfunction setCompress(schema: any, type: ParquetCompression) {\n  for (const name in schema) {\n    const node = schema[name];\n    if (node.fields) {\n      setCompress(node.fields, type);\n    } else {\n      node.compression = type;\n    }\n  }\n}\n\n// eslint-disable-next-line max-statements, complexity\nfunction buildFields(\n  schema: SchemaDefinition,\n  rLevelParentMax: number,\n  dLevelParentMax: number,\n  path: string[]\n): Record<string, ParquetField> {\n  const fieldList: Record<string, ParquetField> = {};\n\n  for (const name in schema) {\n    const opts = schema[name];\n\n    /* field repetition type */\n    const required = !opts.optional;\n    const repeated = Boolean(opts.repeated);\n    let rLevelMax = rLevelParentMax;\n    let dLevelMax = dLevelParentMax;\n\n    let repetitionType: RepetitionType = 'REQUIRED';\n    if (!required) {\n      repetitionType = 'OPTIONAL';\n      dLevelMax++;\n    }\n    if (repeated) {\n      repetitionType = 'REPEATED';\n      rLevelMax++;\n      if (required) dLevelMax++;\n    }\n\n    /* nested field */\n    if (opts.fields) {\n      const cpath = path.concat([name]);\n      fieldList[name] = {\n        name,\n        path: cpath,\n        key: cpath.join(),\n        repetitionType,\n        rLevelMax,\n        dLevelMax,\n        isNested: true,\n        fieldCount: Object.keys(opts.fields).length,\n        fields: buildFields(opts.fields, rLevelMax, dLevelMax, cpath)\n      };\n      continue; // eslint-disable-line no-continue\n    }\n\n    const typeDef: any = PARQUET_LOGICAL_TYPES[opts.type!];\n    if (!typeDef) {\n      throw new Error(`invalid parquet type: ${opts.type}`);\n    }\n\n    opts.encoding = opts.encoding || 'PLAIN';\n    if (!(opts.encoding in PARQUET_CODECS)) {\n      throw new Error(`unsupported parquet encoding: ${opts.encoding}`);\n    }\n\n    opts.compression = opts.compression || 'UNCOMPRESSED';\n    if (!(opts.compression in PARQUET_COMPRESSION_METHODS)) {\n      throw new Error(`unsupported compression method: ${opts.compression}`);\n    }\n\n    /* add to schema */\n    const cpath = path.concat([name]);\n    fieldList[name] = {\n      name,\n      primitiveType: typeDef.primitiveType,\n      originalType: typeDef.originalType,\n      path: cpath,\n      key: cpath.join(),\n      repetitionType,\n      encoding: opts.encoding,\n      compression: opts.compression,\n      typeLength: opts.typeLength || typeDef.typeLength,\n      rLevelMax,\n      dLevelMax\n    };\n  }\n  return fieldList;\n}\n\nfunction listFields(fields: Record<string, ParquetField>): ParquetField[] {\n  let list: ParquetField[] = [];\n  for (const k in fields) {\n    list.push(fields[k]);\n    if (fields[k].isNested) {\n      list = list.concat(listFields(fields[k].fields!));\n    }\n  }\n  return list;\n}\n"],"file":"schema.js"}
+{"version":3,"sources":["../../../../src/parquetjs/schema/schema.ts"],"names":["PARQUET_CODECS","PARQUET_COMPRESSION_METHODS","materializeRecords","shredBuffer","shredRecord","PARQUET_LOGICAL_TYPES","ParquetSchema","constructor","schema","fields","buildFields","fieldList","listFields","findField","path","split","slice","n","length","shift","findFieldBranch","branch","push","record","buffer","compress","type","setCompress","name","node","compression","rLevelParentMax","dLevelParentMax","opts","required","optional","repeated","Boolean","rLevelMax","dLevelMax","repetitionType","cpath","concat","key","join","isNested","fieldCount","Object","keys","typeDef","Error","encoding","primitiveType","originalType","typeLength","presision","scale","list","k"],"mappings":";AAEA,SAAQA,cAAR,QAA6B,WAA7B;AACA,SAAQC,2BAAR,QAA0C,gBAA1C;AAUA,SAAQC,kBAAR,EAA4BC,WAA5B,EAAyCC,WAAzC,QAA2D,SAA3D;AACA,SAAQC,qBAAR,QAAoC,SAApC;AAKA,OAAO,MAAMC,aAAN,CAAoB;AAQzBC,EAAAA,WAAW,CAACC,MAAD,EAA2B;AAAA;;AAAA;;AAAA;;AACpC,SAAKA,MAAL,GAAcA,MAAd;AACA,SAAKC,MAAL,GAAcC,WAAW,CAACF,MAAD,EAAS,CAAT,EAAY,CAAZ,EAAe,EAAf,CAAzB;AACA,SAAKG,SAAL,GAAiBC,UAAU,CAAC,KAAKH,MAAN,CAA3B;AACD;;AAKDI,EAAAA,SAAS,CAACC,IAAD,EAAwC;AAC/C,QAAI,OAAOA,IAAP,KAAgB,QAApB,EAA8B;AAE5BA,MAAAA,IAAI,GAAGA,IAAI,CAACC,KAAL,CAAW,GAAX,CAAP;AACD,KAHD,MAGO;AAELD,MAAAA,IAAI,GAAGA,IAAI,CAACE,KAAL,CAAW,CAAX,CAAP;AACD;;AAED,QAAIC,CAAC,GAAG,KAAKR,MAAb;;AACA,WAAOK,IAAI,CAACI,MAAL,GAAc,CAArB,EAAwBJ,IAAI,CAACK,KAAL,EAAxB,EAAsC;AACpCF,MAAAA,CAAC,GAAGA,CAAC,CAACH,IAAI,CAAC,CAAD,CAAL,CAAD,CAAWL,MAAf;AACD;;AAED,WAAOQ,CAAC,CAACH,IAAI,CAAC,CAAD,CAAL,CAAR;AACD;;AAKDM,EAAAA,eAAe,CAACN,IAAD,EAA0C;AACvD,QAAI,OAAOA,IAAP,KAAgB,QAApB,EAA8B;AAE5BA,MAAAA,IAAI,GAAGA,IAAI,CAACC,KAAL,CAAW,GAAX,CAAP;AACD;;AACD,UAAMM,MAAsB,GAAG,EAA/B;AACA,QAAIJ,CAAC,GAAG,KAAKR,MAAb;;AACA,WAAOK,IAAI,CAACI,MAAL,GAAc,CAArB,EAAwBJ,IAAI,CAACK,KAAL,EAAxB,EAAsC;AACpCE,MAAAA,MAAM,CAACC,IAAP,CAAYL,CAAC,CAACH,IAAI,CAAC,CAAD,CAAL,CAAb;;AACA,UAAIA,IAAI,CAACI,MAAL,GAAc,CAAlB,EAAqB;AACnBD,QAAAA,CAAC,GAAGA,CAAC,CAACH,IAAI,CAAC,CAAD,CAAL,CAAD,CAAWL,MAAf;AACD;AACF;;AACD,WAAOY,MAAP;AACD;;AAEDjB,EAAAA,WAAW,CAACmB,MAAD,EAAwBC,MAAxB,EAAqD;AAC9DpB,IAAAA,WAAW,CAAC,IAAD,EAAOmB,MAAP,EAAeC,MAAf,CAAX;AACD;;AAEDtB,EAAAA,kBAAkB,CAACsB,MAAD,EAAyC;AACzD,WAAOtB,kBAAkB,CAAC,IAAD,EAAOsB,MAAP,CAAzB;AACD;;AAEDC,EAAAA,QAAQ,CAACC,IAAD,EAAiC;AACvCC,IAAAA,WAAW,CAAC,KAAKnB,MAAN,EAAckB,IAAd,CAAX;AACAC,IAAAA,WAAW,CAAC,KAAKlB,MAAN,EAAciB,IAAd,CAAX;AACA,WAAO,IAAP;AACD;;AAEDF,EAAAA,MAAM,GAAkB;AACtB,WAAOrB,WAAW,CAAC,IAAD,CAAlB;AACD;;AArEwB;;AAwE3B,SAASwB,WAAT,CAAqBnB,MAArB,EAAkCkB,IAAlC,EAA4D;AAC1D,OAAK,MAAME,IAAX,IAAmBpB,MAAnB,EAA2B;AACzB,UAAMqB,IAAI,GAAGrB,MAAM,CAACoB,IAAD,CAAnB;;AACA,QAAIC,IAAI,CAACpB,MAAT,EAAiB;AACfkB,MAAAA,WAAW,CAACE,IAAI,CAACpB,MAAN,EAAciB,IAAd,CAAX;AACD,KAFD,MAEO;AACLG,MAAAA,IAAI,CAACC,WAAL,GAAmBJ,IAAnB;AACD;AACF;AACF;;AAGD,SAAShB,WAAT,CACEF,MADF,EAEEuB,eAFF,EAGEC,eAHF,EAIElB,IAJF,EAKgC;AAC9B,QAAMH,SAAuC,GAAG,EAAhD;;AAEA,OAAK,MAAMiB,IAAX,IAAmBpB,MAAnB,EAA2B;AACzB,UAAMyB,IAAI,GAAGzB,MAAM,CAACoB,IAAD,CAAnB;AAGA,UAAMM,QAAQ,GAAG,CAACD,IAAI,CAACE,QAAvB;AACA,UAAMC,QAAQ,GAAGC,OAAO,CAACJ,IAAI,CAACG,QAAN,CAAxB;AACA,QAAIE,SAAS,GAAGP,eAAhB;AACA,QAAIQ,SAAS,GAAGP,eAAhB;AAEA,QAAIQ,cAA8B,GAAG,UAArC;;AACA,QAAI,CAACN,QAAL,EAAe;AACbM,MAAAA,cAAc,GAAG,UAAjB;AACAD,MAAAA,SAAS;AACV;;AACD,QAAIH,QAAJ,EAAc;AACZI,MAAAA,cAAc,GAAG,UAAjB;AACAF,MAAAA,SAAS;AACT,UAAIJ,QAAJ,EAAcK,SAAS;AACxB;;AAGD,QAAIN,IAAI,CAACxB,MAAT,EAAiB;AACf,YAAMgC,KAAK,GAAG3B,IAAI,CAAC4B,MAAL,CAAY,CAACd,IAAD,CAAZ,CAAd;AACAjB,MAAAA,SAAS,CAACiB,IAAD,CAAT,GAAkB;AAChBA,QAAAA,IADgB;AAEhBd,QAAAA,IAAI,EAAE2B,KAFU;AAGhBE,QAAAA,GAAG,EAAEF,KAAK,CAACG,IAAN,EAHW;AAIhBJ,QAAAA,cAJgB;AAKhBF,QAAAA,SALgB;AAMhBC,QAAAA,SANgB;AAOhBM,QAAAA,QAAQ,EAAE,IAPM;AAQhBC,QAAAA,UAAU,EAAEC,MAAM,CAACC,IAAP,CAAYf,IAAI,CAACxB,MAAjB,EAAyBS,MARrB;AAShBT,QAAAA,MAAM,EAAEC,WAAW,CAACuB,IAAI,CAACxB,MAAN,EAAc6B,SAAd,EAAyBC,SAAzB,EAAoCE,KAApC;AATH,OAAlB;AAWA;AACD;;AAED,UAAMQ,OAAY,GAAG5C,qBAAqB,CAAC4B,IAAI,CAACP,IAAN,CAA1C;;AACA,QAAI,CAACuB,OAAL,EAAc;AACZ,YAAM,IAAIC,KAAJ,CAAW,yBAAwBjB,IAAI,CAACP,IAAK,EAA7C,CAAN;AACD;;AAEDO,IAAAA,IAAI,CAACkB,QAAL,GAAgBlB,IAAI,CAACkB,QAAL,IAAiB,OAAjC;;AACA,QAAI,EAAElB,IAAI,CAACkB,QAAL,IAAiBnD,cAAnB,CAAJ,EAAwC;AACtC,YAAM,IAAIkD,KAAJ,CAAW,iCAAgCjB,IAAI,CAACkB,QAAS,EAAzD,CAAN;AACD;;AAEDlB,IAAAA,IAAI,CAACH,WAAL,GAAmBG,IAAI,CAACH,WAAL,IAAoB,cAAvC;;AACA,QAAI,EAAEG,IAAI,CAACH,WAAL,IAAoB7B,2BAAtB,CAAJ,EAAwD;AACtD,YAAM,IAAIiD,KAAJ,CAAW,mCAAkCjB,IAAI,CAACH,WAAY,EAA9D,CAAN;AACD;;AAGD,UAAMW,KAAK,GAAG3B,IAAI,CAAC4B,MAAL,CAAY,CAACd,IAAD,CAAZ,CAAd;AACAjB,IAAAA,SAAS,CAACiB,IAAD,CAAT,GAAkB;AAChBA,MAAAA,IADgB;AAEhBwB,MAAAA,aAAa,EAAEH,OAAO,CAACG,aAFP;AAGhBC,MAAAA,YAAY,EAAEJ,OAAO,CAACI,YAHN;AAIhBvC,MAAAA,IAAI,EAAE2B,KAJU;AAKhBE,MAAAA,GAAG,EAAEF,KAAK,CAACG,IAAN,EALW;AAMhBJ,MAAAA,cANgB;AAOhBW,MAAAA,QAAQ,EAAElB,IAAI,CAACkB,QAPC;AAQhBrB,MAAAA,WAAW,EAAEG,IAAI,CAACH,WARF;AAShBwB,MAAAA,UAAU,EAAErB,IAAI,CAACqB,UAAL,IAAmBL,OAAO,CAACK,UATvB;AAUhBC,MAAAA,SAAS,EAAEtB,IAAI,CAACsB,SAVA;AAWhBC,MAAAA,KAAK,EAAEvB,IAAI,CAACuB,KAXI;AAYhBlB,MAAAA,SAZgB;AAahBC,MAAAA;AAbgB,KAAlB;AAeD;;AACD,SAAO5B,SAAP;AACD;;AAED,SAASC,UAAT,CAAoBH,MAApB,EAA0E;AACxE,MAAIgD,IAAoB,GAAG,EAA3B;;AACA,OAAK,MAAMC,CAAX,IAAgBjD,MAAhB,EAAwB;AACtBgD,IAAAA,IAAI,CAACnC,IAAL,CAAUb,MAAM,CAACiD,CAAD,CAAhB;;AACA,QAAIjD,MAAM,CAACiD,CAAD,CAAN,CAAUb,QAAd,EAAwB;AACtBY,MAAAA,IAAI,GAAGA,IAAI,CAACf,MAAL,CAAY9B,UAAU,CAACH,MAAM,CAACiD,CAAD,CAAN,CAAUjD,MAAX,CAAtB,CAAP;AACD;AACF;;AACD,SAAOgD,IAAP;AACD","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\n\nimport {PARQUET_CODECS} from '../codecs';\nimport {PARQUET_COMPRESSION_METHODS} from '../compression';\nimport {\n  FieldDefinition,\n  ParquetBuffer,\n  ParquetCompression,\n  ParquetField,\n  ParquetRecord,\n  RepetitionType,\n  SchemaDefinition\n} from './declare';\nimport {materializeRecords, shredBuffer, shredRecord} from './shred';\nimport {PARQUET_LOGICAL_TYPES} from './types';\n\n/**\n * A parquet file schema\n */\nexport class ParquetSchema {\n  public schema: Record<string, FieldDefinition>;\n  public fields: Record<string, ParquetField>;\n  public fieldList: ParquetField[];\n\n  /**\n   * Create a new schema from a JSON schema definition\n   */\n  constructor(schema: SchemaDefinition) {\n    this.schema = schema;\n    this.fields = buildFields(schema, 0, 0, []);\n    this.fieldList = listFields(this.fields);\n  }\n\n  /**\n   * Retrieve a field definition\n   */\n  findField(path: string | string[]): ParquetField {\n    if (typeof path === 'string') {\n      // tslint:disable-next-line:no-parameter-reassignment\n      path = path.split(',');\n    } else {\n      // tslint:disable-next-line:no-parameter-reassignment\n      path = path.slice(0); // clone array\n    }\n\n    let n = this.fields;\n    for (; path.length > 1; path.shift()) {\n      n = n[path[0]].fields as Record<string, ParquetField>;\n    }\n\n    return n[path[0]];\n  }\n\n  /**\n   * Retrieve a field definition and all the field's ancestors\n   */\n  findFieldBranch(path: string | string[]): ParquetField[] {\n    if (typeof path === 'string') {\n      // tslint:disable-next-line:no-parameter-reassignment\n      path = path.split(',');\n    }\n    const branch: ParquetField[] = [];\n    let n = this.fields;\n    for (; path.length > 0; path.shift()) {\n      branch.push(n[path[0]]);\n      if (path.length > 1) {\n        n = n[path[0]].fields as Record<string, ParquetField>;\n      }\n    }\n    return branch;\n  }\n\n  shredRecord(record: ParquetRecord, buffer: ParquetBuffer): void {\n    shredRecord(this, record, buffer);\n  }\n\n  materializeRecords(buffer: ParquetBuffer): ParquetRecord[] {\n    return materializeRecords(this, buffer);\n  }\n\n  compress(type: ParquetCompression): this {\n    setCompress(this.schema, type);\n    setCompress(this.fields, type);\n    return this;\n  }\n\n  buffer(): ParquetBuffer {\n    return shredBuffer(this);\n  }\n}\n\nfunction setCompress(schema: any, type: ParquetCompression) {\n  for (const name in schema) {\n    const node = schema[name];\n    if (node.fields) {\n      setCompress(node.fields, type);\n    } else {\n      node.compression = type;\n    }\n  }\n}\n\n// eslint-disable-next-line max-statements, complexity\nfunction buildFields(\n  schema: SchemaDefinition,\n  rLevelParentMax: number,\n  dLevelParentMax: number,\n  path: string[]\n): Record<string, ParquetField> {\n  const fieldList: Record<string, ParquetField> = {};\n\n  for (const name in schema) {\n    const opts = schema[name];\n\n    /* field repetition type */\n    const required = !opts.optional;\n    const repeated = Boolean(opts.repeated);\n    let rLevelMax = rLevelParentMax;\n    let dLevelMax = dLevelParentMax;\n\n    let repetitionType: RepetitionType = 'REQUIRED';\n    if (!required) {\n      repetitionType = 'OPTIONAL';\n      dLevelMax++;\n    }\n    if (repeated) {\n      repetitionType = 'REPEATED';\n      rLevelMax++;\n      if (required) dLevelMax++;\n    }\n\n    /* nested field */\n    if (opts.fields) {\n      const cpath = path.concat([name]);\n      fieldList[name] = {\n        name,\n        path: cpath,\n        key: cpath.join(),\n        repetitionType,\n        rLevelMax,\n        dLevelMax,\n        isNested: true,\n        fieldCount: Object.keys(opts.fields).length,\n        fields: buildFields(opts.fields, rLevelMax, dLevelMax, cpath)\n      };\n      continue; // eslint-disable-line no-continue\n    }\n\n    const typeDef: any = PARQUET_LOGICAL_TYPES[opts.type!];\n    if (!typeDef) {\n      throw new Error(`invalid parquet type: ${opts.type}`);\n    }\n\n    opts.encoding = opts.encoding || 'PLAIN';\n    if (!(opts.encoding in PARQUET_CODECS)) {\n      throw new Error(`unsupported parquet encoding: ${opts.encoding}`);\n    }\n\n    opts.compression = opts.compression || 'UNCOMPRESSED';\n    if (!(opts.compression in PARQUET_COMPRESSION_METHODS)) {\n      throw new Error(`unsupported compression method: ${opts.compression}`);\n    }\n\n    /* add to schema */\n    const cpath = path.concat([name]);\n    fieldList[name] = {\n      name,\n      primitiveType: typeDef.primitiveType,\n      originalType: typeDef.originalType,\n      path: cpath,\n      key: cpath.join(),\n      repetitionType,\n      encoding: opts.encoding,\n      compression: opts.compression,\n      typeLength: opts.typeLength || typeDef.typeLength,\n      presision: opts.presision,\n      scale: opts.scale,\n      rLevelMax,\n      dLevelMax\n    };\n  }\n  return fieldList;\n}\n\nfunction listFields(fields: Record<string, ParquetField>): ParquetField[] {\n  let list: ParquetField[] = [];\n  for (const k in fields) {\n    list.push(fields[k]);\n    if (fields[k].isNested) {\n      list = list.concat(listFields(fields[k].fields!));\n    }\n  }\n  return list;\n}\n"],"file":"schema.js"}

package/dist/esm/parquetjs/schema/shred.js

@@ -9,6 +9,7 @@ export function shredBuffer(schema) {
       dlevels: [],
       rlevels: [],
       values: [],
+      pageHeaders: [],
       count: 0
     };
   }
@@ -137,7 +138,7 @@ function materializeColumn(schema, buffer, key, records) {
     }
 
     if (dLevel === field.dLevelMax) {
-      const value = Types.fromPrimitive(field.originalType || field.primitiveType, data.values[vIndex]);
+      const value = Types.fromPrimitive(field.originalType || field.primitiveType, data.values[vIndex], field);
       vIndex++;
 
       if (field.repetitionType === 'REPEATED') {

package/dist/esm/parquetjs/schema/shred.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../../../src/parquetjs/schema/shred.ts"],"names":["ParquetBuffer","Types","shredBuffer","schema","columnData","field","fieldList","key","dlevels","rlevels","values","count","rowCount","shredRecord","record","buffer","data","shredRecordFields","fields","Array","prototype","push","apply","rLevel","dLevel","name","undefined","constructor","length","Boolean","repetitionType","Error","isNested","i","rlvl","rLevelMax","dLevelMax","toPrimitive","originalType","primitiveType","materializeRecords","records","materializeColumn","findField","branch","findFieldBranch","rLevels","fill","vIndex","rIndex","step","ix","value","fromPrimitive"],"mappings":"AAEA,SAAQA,aAAR,QAAsE,WAAtE;AAEA,OAAO,KAAKC,KAAZ,MAAuB,SAAvB;AAEA,SAAQD,aAAR;AAEA,OAAO,SAASE,WAAT,CAAqBC,MAArB,EAA2D;AAChE,QAAMC,UAAuC,GAAG,EAAhD;;AACA,OAAK,MAAMC,KAAX,IAAoBF,MAAM,CAACG,SAA3B,EAAsC;AACpCF,IAAAA,UAAU,CAACC,KAAK,CAACE,GAAP,CAAV,GAAwB;AACtBC,MAAAA,OAAO,EAAE,EADa;AAEtBC,MAAAA,OAAO,EAAE,EAFa;AAGtBC,MAAAA,MAAM,EAAE,EAHc;AAItBC,MAAAA,KAAK,EAAE;AAJe,KAAxB;AAMD;;AACD,SAAO;AAACC,IAAAA,QAAQ,EAAE,CAAX;AAAcR,IAAAA;AAAd,GAAP;AACD;AAwBD,OAAO,SAASS,WAAT,CAAqBV,MAArB,EAA4CW,MAA5C,EAAyDC,MAAzD,EAAsF;AAE3F,QAAMC,IAAI,GAAGd,WAAW,CAACC,MAAD,CAAX,CAAoBC,UAAjC;AAEAa,EAAAA,iBAAiB,CAACd,MAAM,CAACe,MAAR,EAAgBJ,MAAhB,EAAwBE,IAAxB,EAA8B,CAA9B,EAAiC,CAAjC,CAAjB;;AAGA,MAAID,MAAM,CAACH,QAAP,KAAoB,CAAxB,EAA2B;AACzBG,IAAAA,MAAM,CAACH,QAAP,GAAkB,CAAlB;AACAG,IAAAA,MAAM,CAACX,UAAP,GAAoBY,IAApB;AACA;AACD;;AACDD,EAAAA,MAAM,CAACH,QAAP,IAAmB,CAAnB;;AACA,OAAK,MAAMP,KAAX,IAAoBF,MAAM,CAACG,SAA3B,EAAsC;AACpCa,IAAAA,KAAK,CAACC,SAAN,CAAgBC,IAAhB,CAAqBC,KAArB,CAA2BP,MAAM,CAACX,UAAP,CAAkBC,KAAK,CAACE,GAAxB,EAA6BE,OAAxD,EAAiEO,IAAI,CAACX,KAAK,CAACE,GAAP,CAAJ,CAAgBE,OAAjF;AACAU,IAAAA,KAAK,CAACC,SAAN,CAAgBC,IAAhB,CAAqBC,KAArB,CAA2BP,MAAM,CAACX,UAAP,CAAkBC,KAAK,CAACE,GAAxB,EAA6BC,OAAxD,EAAiEQ,IAAI,CAACX,KAAK,CAACE,GAAP,CAAJ,CAAgBC,OAAjF;AACAW,IAAAA,KAAK,CAACC,SAAN,CAAgBC,IAAhB,CAAqBC,KAArB,CAA2BP,MAAM,CAACX,UAAP,CAAkBC,KAAK,CAACE,GAAxB,EAA6BG,MAAxD,EAAgEM,IAAI,CAACX,KAAK,CAACE,GAAP,CAAJ,CAAgBG,MAAhF;AACAK,IAAAA,MAAM,CAACX,UAAP,CAAkBC,KAAK,CAACE,GAAxB,EAA6BI,KAA7B,IAAsCK,IAAI,CAACX,KAAK,CAACE,GAAP,CAAJ,CAAgBI,KAAtD;AACD;AACF;;AAGD,SAASM,iBAAT,CACEC,MADF,EAEEJ,MAFF,EAGEE,IAHF,EAIEO,MAJF,EAKEC,MALF,EAME;AACA,OAAK,MAAMC,IAAX,IAAmBP,MAAnB,EAA2B;AACzB,UAAMb,KAAK,GAAGa,MAAM,CAACO,IAAD,CAApB;AAGA,QAAIf,MAAa,GAAG,EAApB;;AACA,QACEI,MAAM,IACNT,KAAK,CAACoB,IAAN,IAAcX,MADd,IAEAA,MAAM,CAACT,KAAK,CAACoB,IAAP,CAAN,KAAuBC,SAFvB,IAGAZ,MAAM,CAACT,KAAK,CAACoB,IAAP,CAAN,KAAuB,IAJzB,EAKE;AACA,UAAIX,MAAM,CAACT,KAAK,CAACoB,IAAP,CAAN,CAAmBE,WAAnB,KAAmCR,KAAvC,EAA8C;AAC5CT,QAAAA,MAAM,GAAGI,MAAM,CAACT,KAAK,CAACoB,IAAP,CAAf;AACD,OAFD,MAEO;AACLf,QAAAA,MAAM,CAACW,IAAP,CAAYP,MAAM,CAACT,KAAK,CAACoB,IAAP,CAAlB;AACD;AACF;;AAED,QAAIf,MAAM,CAACkB,MAAP,KAAkB,CAAlB,IAAuBC,OAAO,CAACf,MAAD,CAA9B,IAA0CT,KAAK,CAACyB,cAAN,KAAyB,UAAvE,EAAmF;AACjF,YAAM,IAAIC,KAAJ,CAAW,2BAA0B1B,KAAK,CAACoB,IAAK,EAAhD,CAAN;AACD;;AACD,QAAIf,MAAM,CAACkB,MAAP,GAAgB,CAAhB,IAAqBvB,KAAK,CAACyB,cAAN,KAAyB,UAAlD,EAA8D;AAC5D,YAAM,IAAIC,KAAJ,CAAW,8BAA6B1B,KAAK,CAACoB,IAAK,EAAnD,CAAN;AACD;;AAGD,QAAIf,MAAM,CAACkB,MAAP,KAAkB,CAAtB,EAAyB;AACvB,UAAIvB,KAAK,CAAC2B,QAAV,EAAoB;AAClBf,QAAAA,iBAAiB,CAACZ,KAAK,CAACa,MAAP,EAAgB,IAAhB,EAAsBF,IAAtB,EAA4BO,MAA5B,EAAoCC,MAApC,CAAjB;AACD,OAFD,MAEO;AACLR,QAAAA,IAAI,CAACX,KAAK,CAACE,GAAP,CAAJ,CAAgBI,KAAhB,IAAyB,CAAzB;AACAK,QAAAA,IAAI,CAACX,KAAK,CAACE,GAAP,CAAJ,CAAgBE,OAAhB,CAAwBY,IAAxB,CAA6BE,MAA7B;AACAP,QAAAA,IAAI,CAACX,KAAK,CAACE,GAAP,CAAJ,CAAgBC,OAAhB,CAAwBa,IAAxB,CAA6BG,MAA7B;AACD;;AACD;AACD;;AAGD,SAAK,IAAIS,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGvB,MAAM,CAACkB,MAA3B,EAAmCK,CAAC,EAApC,EAAwC;AACtC,YAAMC,IAAI,GAAGD,CAAC,KAAK,CAAN,GAAUV,MAAV,GAAmBlB,KAAK,CAAC8B,SAAtC;;AACA,UAAI9B,KAAK,CAAC2B,QAAV,EAAoB;AAClBf,QAAAA,iBAAiB,CAACZ,KAAK,CAACa,MAAP,EAAgBR,MAAM,CAACuB,CAAD,CAAtB,EAA2BjB,IAA3B,EAAiCkB,IAAjC,EAAuC7B,KAAK,CAAC+B,SAA7C,CAAjB;AACD,OAFD,MAEO;AACLpB,QAAAA,IAAI,CAACX,KAAK,CAACE,GAAP,CAAJ,CAAgBI,KAAhB,IAAyB,CAAzB;AACAK,QAAAA,IAAI,CAACX,KAAK,CAACE,GAAP,CAAJ,CAAgBE,OAAhB,CAAwBY,IAAxB,CAA6Ba,IAA7B;AACAlB,QAAAA,IAAI,CAACX,KAAK,CAACE,GAAP,CAAJ,CAAgBC,OAAhB,CAAwBa,IAAxB,CAA6BhB,KAAK,CAAC+B,SAAnC;AACApB,QAAAA,IAAI,CAACX,KAAK,CAACE,GAAP,CAAJ,CAAgBG,MAAhB,CAAuBW,IAAvB,CACEpB,KAAK,CAACoC,WAAN,CAAmBhC,KAAK,CAACiC,YAAN,IAAsBjC,KAAK,CAACkC,aAA/C,EAAgE7B,MAAM,CAACuB,CAAD,CAAtE,CADF;AAGD;AACF;AACF;AACF;;AAqBD,OAAO,SAASO,kBAAT,CAA4BrC,MAA5B,EAAmDY,MAAnD,EAA2F;AAChG,QAAM0B,OAAwB,GAAG,EAAjC;;AACA,OAAK,IAAIR,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGlB,MAAM,CAACH,QAA3B,EAAqCqB,CAAC,EAAtC,EAA0CQ,OAAO,CAACpB,IAAR,CAAa,EAAb;;AAC1C,OAAK,MAAMd,GAAX,IAAkBQ,MAAM,CAACX,UAAzB,EAAqC;AACnCsC,IAAAA,iBAAiB,CAACvC,MAAD,EAASY,MAAT,EAAiBR,GAAjB,EAAsBkC,OAAtB,CAAjB;AACD;;AACD,SAAOA,OAAP;AACD;;AAGD,SAASC,iBAAT,CACEvC,MADF,EAEEY,MAFF,EAGER,GAHF,EAIEkC,OAJF,EAKE;AACA,QAAMzB,IAAI,GAAGD,MAAM,CAACX,UAAP,CAAkBG,GAAlB,CAAb;AACA,MAAI,CAACS,IAAI,CAACL,KAAV,EAAiB;AAEjB,QAAMN,KAAK,GAAGF,MAAM,CAACwC,SAAP,CAAiBpC,GAAjB,CAAd;AACA,QAAMqC,MAAM,GAAGzC,MAAM,CAAC0C,eAAP,CAAuBtC,GAAvB,CAAf;AAGA,QAAMuC,OAAiB,GAAG,IAAI3B,KAAJ,CAAUd,KAAK,CAAC8B,SAAN,GAAkB,CAA5B,EAA+BY,IAA/B,CAAoC,CAApC,CAA1B;AACA,MAAIC,MAAM,GAAG,CAAb;;AACA,OAAK,IAAIf,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGjB,IAAI,CAACL,KAAzB,EAAgCsB,CAAC,EAAjC,EAAqC;AACnC,UAAMT,MAAM,GAAGR,IAAI,CAACR,OAAL,CAAayB,CAAb,CAAf;AACA,UAAMV,MAAM,GAAGP,IAAI,CAACP,OAAL,CAAawB,CAAb,CAAf;AACAa,IAAAA,OAAO,CAACvB,MAAD,CAAP;AACAuB,IAAAA,OAAO,CAACC,IAAR,CAAa,CAAb,EAAgBxB,MAAM,GAAG,CAAzB;AAEA,QAAI0B,MAAM,GAAG,CAAb;AACA,QAAInC,MAAM,GAAG2B,OAAO,CAACK,OAAO,CAACG,MAAM,EAAP,CAAP,GAAoB,CAArB,CAApB;;AAGA,SAAK,MAAMC,IAAX,IAAmBN,MAAnB,EAA2B;AACzB,UAAIM,IAAI,KAAK7C,KAAb,EAAoB;AACpB,UAAImB,MAAM,GAAG0B,IAAI,CAACd,SAAlB,EAA6B;;AAC7B,UAAIc,IAAI,CAACpB,cAAL,KAAwB,UAA5B,EAAwC;AACtC,YAAI,EAAEoB,IAAI,CAACzB,IAAL,IAAaX,MAAf,CAAJ,EAA4B;AAE1BA,UAAAA,MAAM,CAACoC,IAAI,CAACzB,IAAN,CAAN,GAAoB,EAApB;AACD;;AACD,cAAM0B,EAAE,GAAGL,OAAO,CAACG,MAAM,EAAP,CAAlB;;AACA,eAAOnC,MAAM,CAACoC,IAAI,CAACzB,IAAN,CAAN,CAAkBG,MAAlB,IAA4BuB,EAAnC,EAAuC;AAErCrC,UAAAA,MAAM,CAACoC,IAAI,CAACzB,IAAN,CAAN,CAAkBJ,IAAlB,CAAuB,EAAvB;AACD;;AACDP,QAAAA,MAAM,GAAGA,MAAM,CAACoC,IAAI,CAACzB,IAAN,CAAN,CAAkB0B,EAAlB,CAAT;AACD,OAXD,MAWO;AACLrC,QAAAA,MAAM,CAACoC,IAAI,CAACzB,IAAN,CAAN,GAAoBX,MAAM,CAACoC,IAAI,CAACzB,IAAN,CAAN,IAAqB,EAAzC;AACAX,QAAAA,MAAM,GAAGA,MAAM,CAACoC,IAAI,CAACzB,IAAN,CAAf;AACD;AACF;;AAGD,QAAID,MAAM,KAAKnB,KAAK,CAAC+B,SAArB,EAAgC;AAC9B,YAAMgB,KAAK,GAAGnD,KAAK,CAACoD,aAAN,CAEZhD,KAAK,CAACiC,YAAN,IAAsBjC,KAAK,CAACkC,aAFhB,EAGZvB,IAAI,CAACN,MAAL,CAAYsC,MAAZ,CAHY,CAAd;AAKAA,MAAAA,MAAM;;AACN,UAAI3C,KAAK,CAACyB,cAAN,KAAyB,UAA7B,EAAyC;AACvC,YAAI,EAAEzB,KAAK,CAACoB,IAAN,IAAcX,MAAhB,CAAJ,EAA6B;AAE3BA,UAAAA,MAAM,CAACT,KAAK,CAACoB,IAAP,CAAN,GAAqB,EAArB;AACD;;AACD,cAAM0B,EAAE,GAAGL,OAAO,CAACG,MAAD,CAAlB;;AACA,eAAOnC,MAAM,CAACT,KAAK,CAACoB,IAAP,CAAN,CAAmBG,MAAnB,IAA6BuB,EAApC,EAAwC;AAEtCrC,UAAAA,MAAM,CAACT,KAAK,CAACoB,IAAP,CAAN,CAAmBJ,IAAnB,CAAwB,IAAxB;AACD;;AACDP,QAAAA,MAAM,CAACT,KAAK,CAACoB,IAAP,CAAN,CAAmB0B,EAAnB,IAAyBC,KAAzB;AACD,OAXD,MAWO;AACLtC,QAAAA,MAAM,CAACT,KAAK,CAACoB,IAAP,CAAN,GAAqB2B,KAArB;AACD;AACF;AACF;AACF","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\n\nimport {ParquetBuffer, ParquetData, ParquetField, ParquetRecord} from './declare';\nimport {ParquetSchema} from './schema';\nimport * as Types from './types';\n\nexport {ParquetBuffer};\n\nexport function shredBuffer(schema: ParquetSchema): ParquetBuffer {\n  const columnData: Record<string, ParquetData> = {};\n  for (const field of schema.fieldList) {\n    columnData[field.key] = {\n      dlevels: [],\n      rlevels: [],\n      values: [],\n      count: 0\n    };\n  }\n  return {rowCount: 0, columnData};\n}\n\n/**\n * 'Shred' a record into a list of <value, repetition_level, definition_level>\n * tuples per column using the Google Dremel Algorithm..\n *\n * The buffer argument must point to an object into which the shredded record\n * will be returned. You may re-use the buffer for repeated calls to this function\n * to append to an existing buffer, as long as the schema is unchanged.\n *\n * The format in which the shredded records will be stored in the buffer is as\n * follows:\n *\n *   buffer = {\n *     columnData: [\n *       'my_col': {\n *          dlevels: [d1, d2, .. dN],\n *          rlevels: [r1, r2, .. rN],\n *          values: [v1, v2, .. vN],\n *        }, ...\n *      ],\n *      rowCount: X,\n *   }\n */\nexport function shredRecord(schema: ParquetSchema, record: any, buffer: ParquetBuffer): void {\n  /* shred the record, this may raise an exception */\n  const data = shredBuffer(schema).columnData;\n\n  shredRecordFields(schema.fields, record, data, 0, 0);\n\n  /* if no error during shredding, add the shredded record to the buffer */\n  if (buffer.rowCount === 0) {\n    buffer.rowCount = 1;\n    buffer.columnData = data;\n    return;\n  }\n  buffer.rowCount += 1;\n  for (const field of schema.fieldList) {\n    Array.prototype.push.apply(buffer.columnData[field.key].rlevels, data[field.key].rlevels);\n    Array.prototype.push.apply(buffer.columnData[field.key].dlevels, data[field.key].dlevels);\n    Array.prototype.push.apply(buffer.columnData[field.key].values, data[field.key].values);\n    buffer.columnData[field.key].count += data[field.key].count;\n  }\n}\n\n// eslint-disable-next-line max-statements, complexity\nfunction shredRecordFields(\n  fields: Record<string, ParquetField>,\n  record: any,\n  data: Record<string, ParquetData>,\n  rLevel: number,\n  dLevel: number\n) {\n  for (const name in fields) {\n    const field = fields[name];\n\n    // fetch values\n    let values: any[] = [];\n    if (\n      record &&\n      field.name in record &&\n      record[field.name] !== undefined &&\n      record[field.name] !== null\n    ) {\n      if (record[field.name].constructor === Array) {\n        values = record[field.name];\n      } else {\n        values.push(record[field.name]);\n      }\n    }\n    // check values\n    if (values.length === 0 && Boolean(record) && field.repetitionType === 'REQUIRED') {\n      throw new Error(`missing required field: ${field.name}`);\n    }\n    if (values.length > 1 && field.repetitionType !== 'REPEATED') {\n      throw new Error(`too many values for field: ${field.name}`);\n    }\n\n    // push null\n    if (values.length === 0) {\n      if (field.isNested) {\n        shredRecordFields(field.fields!, null, data, rLevel, dLevel);\n      } else {\n        data[field.key].count += 1;\n        data[field.key].rlevels.push(rLevel);\n        data[field.key].dlevels.push(dLevel);\n      }\n      continue; // eslint-disable-line no-continue\n    }\n\n    // push values\n    for (let i = 0; i < values.length; i++) {\n      const rlvl = i === 0 ? rLevel : field.rLevelMax;\n      if (field.isNested) {\n        shredRecordFields(field.fields!, values[i], data, rlvl, field.dLevelMax);\n      } else {\n        data[field.key].count += 1;\n        data[field.key].rlevels.push(rlvl);\n        data[field.key].dlevels.push(field.dLevelMax);\n        data[field.key].values.push(\n          Types.toPrimitive((field.originalType || field.primitiveType)!, values[i])\n        );\n      }\n    }\n  }\n}\n\n/**\n * 'Materialize' a list of <value, repetition_level, definition_level>\n * tuples back to nested records (objects/arrays) using the Google Dremel\n * Algorithm..\n *\n * The buffer argument must point to an object with the following structure (i.e.\n * the same structure that is returned by shredRecords):\n *\n *   buffer = {\n *     columnData: [\n *       'my_col': {\n *          dlevels: [d1, d2, .. dN],\n *          rlevels: [r1, r2, .. rN],\n *          values: [v1, v2, .. vN],\n *        }, ...\n *      ],\n *      rowCount: X,\n *   }\n */\nexport function materializeRecords(schema: ParquetSchema, buffer: ParquetBuffer): ParquetRecord[] {\n  const records: ParquetRecord[] = [];\n  for (let i = 0; i < buffer.rowCount; i++) records.push({});\n  for (const key in buffer.columnData) {\n    materializeColumn(schema, buffer, key, records);\n  }\n  return records;\n}\n\n// eslint-disable-next-line max-statements, complexity\nfunction materializeColumn(\n  schema: ParquetSchema,\n  buffer: ParquetBuffer,\n  key: string,\n  records: ParquetRecord[]\n) {\n  const data = buffer.columnData[key];\n  if (!data.count) return;\n\n  const field = schema.findField(key);\n  const branch = schema.findFieldBranch(key);\n\n  // tslint:disable-next-line:prefer-array-literal\n  const rLevels: number[] = new Array(field.rLevelMax + 1).fill(0);\n  let vIndex = 0;\n  for (let i = 0; i < data.count; i++) {\n    const dLevel = data.dlevels[i];\n    const rLevel = data.rlevels[i];\n    rLevels[rLevel]++;\n    rLevels.fill(0, rLevel + 1);\n\n    let rIndex = 0;\n    let record = records[rLevels[rIndex++] - 1];\n\n    // Internal nodes\n    for (const step of branch) {\n      if (step === field) break;\n      if (dLevel < step.dLevelMax) break;\n      if (step.repetitionType === 'REPEATED') {\n        if (!(step.name in record)) {\n          // eslint-disable max-depth\n          record[step.name] = [];\n        }\n        const ix = rLevels[rIndex++];\n        while (record[step.name].length <= ix) {\n          // eslint-disable max-depth\n          record[step.name].push({});\n        }\n        record = record[step.name][ix];\n      } else {\n        record[step.name] = record[step.name] || {};\n        record = record[step.name];\n      }\n    }\n\n    // Leaf node\n    if (dLevel === field.dLevelMax) {\n      const value = Types.fromPrimitive(\n        // @ts-ignore\n        field.originalType || field.primitiveType,\n        data.values[vIndex]\n      );\n      vIndex++;\n      if (field.repetitionType === 'REPEATED') {\n        if (!(field.name in record)) {\n          // eslint-disable max-depth\n          record[field.name] = [];\n        }\n        const ix = rLevels[rIndex];\n        while (record[field.name].length <= ix) {\n          // eslint-disable max-depth\n          record[field.name].push(null);\n        }\n        record[field.name][ix] = value;\n      } else {\n        record[field.name] = value;\n      }\n    }\n  }\n}\n"],"file":"shred.js"}
+{"version":3,"sources":["../../../../src/parquetjs/schema/shred.ts"],"names":["ParquetBuffer","Types","shredBuffer","schema","columnData","field","fieldList","key","dlevels","rlevels","values","pageHeaders","count","rowCount","shredRecord","record","buffer","data","shredRecordFields","fields","Array","prototype","push","apply","rLevel","dLevel","name","undefined","constructor","length","Boolean","repetitionType","Error","isNested","i","rlvl","rLevelMax","dLevelMax","toPrimitive","originalType","primitiveType","materializeRecords","records","materializeColumn","findField","branch","findFieldBranch","rLevels","fill","vIndex","rIndex","step","ix","value","fromPrimitive"],"mappings":"AAEA,SAAQA,aAAR,QAAsE,WAAtE;AAEA,OAAO,KAAKC,KAAZ,MAAuB,SAAvB;AAEA,SAAQD,aAAR;AAEA,OAAO,SAASE,WAAT,CAAqBC,MAArB,EAA2D;AAChE,QAAMC,UAAuC,GAAG,EAAhD;;AACA,OAAK,MAAMC,KAAX,IAAoBF,MAAM,CAACG,SAA3B,EAAsC;AACpCF,IAAAA,UAAU,CAACC,KAAK,CAACE,GAAP,CAAV,GAAwB;AACtBC,MAAAA,OAAO,EAAE,EADa;AAEtBC,MAAAA,OAAO,EAAE,EAFa;AAGtBC,MAAAA,MAAM,EAAE,EAHc;AAItBC,MAAAA,WAAW,EAAE,EAJS;AAKtBC,MAAAA,KAAK,EAAE;AALe,KAAxB;AAOD;;AACD,SAAO;AAACC,IAAAA,QAAQ,EAAE,CAAX;AAAcT,IAAAA;AAAd,GAAP;AACD;AAwBD,OAAO,SAASU,WAAT,CAAqBX,MAArB,EAA4CY,MAA5C,EAAyDC,MAAzD,EAAsF;AAE3F,QAAMC,IAAI,GAAGf,WAAW,CAACC,MAAD,CAAX,CAAoBC,UAAjC;AAEAc,EAAAA,iBAAiB,CAACf,MAAM,CAACgB,MAAR,EAAgBJ,MAAhB,EAAwBE,IAAxB,EAA8B,CAA9B,EAAiC,CAAjC,CAAjB;;AAGA,MAAID,MAAM,CAACH,QAAP,KAAoB,CAAxB,EAA2B;AACzBG,IAAAA,MAAM,CAACH,QAAP,GAAkB,CAAlB;AACAG,IAAAA,MAAM,CAACZ,UAAP,GAAoBa,IAApB;AACA;AACD;;AACDD,EAAAA,MAAM,CAACH,QAAP,IAAmB,CAAnB;;AACA,OAAK,MAAMR,KAAX,IAAoBF,MAAM,CAACG,SAA3B,EAAsC;AACpCc,IAAAA,KAAK,CAACC,SAAN,CAAgBC,IAAhB,CAAqBC,KAArB,CAA2BP,MAAM,CAACZ,UAAP,CAAkBC,KAAK,CAACE,GAAxB,EAA6BE,OAAxD,EAAiEQ,IAAI,CAACZ,KAAK,CAACE,GAAP,CAAJ,CAAgBE,OAAjF;AACAW,IAAAA,KAAK,CAACC,SAAN,CAAgBC,IAAhB,CAAqBC,KAArB,CAA2BP,MAAM,CAACZ,UAAP,CAAkBC,KAAK,CAACE,GAAxB,EAA6BC,OAAxD,EAAiES,IAAI,CAACZ,KAAK,CAACE,GAAP,CAAJ,CAAgBC,OAAjF;AACAY,IAAAA,KAAK,CAACC,SAAN,CAAgBC,IAAhB,CAAqBC,KAArB,CAA2BP,MAAM,CAACZ,UAAP,CAAkBC,KAAK,CAACE,GAAxB,EAA6BG,MAAxD,EAAgEO,IAAI,CAACZ,KAAK,CAACE,GAAP,CAAJ,CAAgBG,MAAhF;AACAM,IAAAA,MAAM,CAACZ,UAAP,CAAkBC,KAAK,CAACE,GAAxB,EAA6BK,KAA7B,IAAsCK,IAAI,CAACZ,KAAK,CAACE,GAAP,CAAJ,CAAgBK,KAAtD;AACD;AACF;;AAGD,SAASM,iBAAT,CACEC,MADF,EAEEJ,MAFF,EAGEE,IAHF,EAIEO,MAJF,EAKEC,MALF,EAME;AACA,OAAK,MAAMC,IAAX,IAAmBP,MAAnB,EAA2B;AACzB,UAAMd,KAAK,GAAGc,MAAM,CAACO,IAAD,CAApB;AAGA,QAAIhB,MAAa,GAAG,EAApB;;AACA,QACEK,MAAM,IACNV,KAAK,CAACqB,IAAN,IAAcX,MADd,IAEAA,MAAM,CAACV,KAAK,CAACqB,IAAP,CAAN,KAAuBC,SAFvB,IAGAZ,MAAM,CAACV,KAAK,CAACqB,IAAP,CAAN,KAAuB,IAJzB,EAKE;AACA,UAAIX,MAAM,CAACV,KAAK,CAACqB,IAAP,CAAN,CAAmBE,WAAnB,KAAmCR,KAAvC,EAA8C;AAC5CV,QAAAA,MAAM,GAAGK,MAAM,CAACV,KAAK,CAACqB,IAAP,CAAf;AACD,OAFD,MAEO;AACLhB,QAAAA,MAAM,CAACY,IAAP,CAAYP,MAAM,CAACV,KAAK,CAACqB,IAAP,CAAlB;AACD;AACF;;AAED,QAAIhB,MAAM,CAACmB,MAAP,KAAkB,CAAlB,IAAuBC,OAAO,CAACf,MAAD,CAA9B,IAA0CV,KAAK,CAAC0B,cAAN,KAAyB,UAAvE,EAAmF;AACjF,YAAM,IAAIC,KAAJ,CAAW,2BAA0B3B,KAAK,CAACqB,IAAK,EAAhD,CAAN;AACD;;AACD,QAAIhB,MAAM,CAACmB,MAAP,GAAgB,CAAhB,IAAqBxB,KAAK,CAAC0B,cAAN,KAAyB,UAAlD,EAA8D;AAC5D,YAAM,IAAIC,KAAJ,CAAW,8BAA6B3B,KAAK,CAACqB,IAAK,EAAnD,CAAN;AACD;;AAGD,QAAIhB,MAAM,CAACmB,MAAP,KAAkB,CAAtB,EAAyB;AACvB,UAAIxB,KAAK,CAAC4B,QAAV,EAAoB;AAClBf,QAAAA,iBAAiB,CAACb,KAAK,CAACc,MAAP,EAAgB,IAAhB,EAAsBF,IAAtB,EAA4BO,MAA5B,EAAoCC,MAApC,CAAjB;AACD,OAFD,MAEO;AACLR,QAAAA,IAAI,CAACZ,KAAK,CAACE,GAAP,CAAJ,CAAgBK,KAAhB,IAAyB,CAAzB;AACAK,QAAAA,IAAI,CAACZ,KAAK,CAACE,GAAP,CAAJ,CAAgBE,OAAhB,CAAwBa,IAAxB,CAA6BE,MAA7B;AACAP,QAAAA,IAAI,CAACZ,KAAK,CAACE,GAAP,CAAJ,CAAgBC,OAAhB,CAAwBc,IAAxB,CAA6BG,MAA7B;AACD;;AACD;AACD;;AAGD,SAAK,IAAIS,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGxB,MAAM,CAACmB,MAA3B,EAAmCK,CAAC,EAApC,EAAwC;AACtC,YAAMC,IAAI,GAAGD,CAAC,KAAK,CAAN,GAAUV,MAAV,GAAmBnB,KAAK,CAAC+B,SAAtC;;AACA,UAAI/B,KAAK,CAAC4B,QAAV,EAAoB;AAClBf,QAAAA,iBAAiB,CAACb,KAAK,CAACc,MAAP,EAAgBT,MAAM,CAACwB,CAAD,CAAtB,EAA2BjB,IAA3B,EAAiCkB,IAAjC,EAAuC9B,KAAK,CAACgC,SAA7C,CAAjB;AACD,OAFD,MAEO;AACLpB,QAAAA,IAAI,CAACZ,KAAK,CAACE,GAAP,CAAJ,CAAgBK,KAAhB,IAAyB,CAAzB;AACAK,QAAAA,IAAI,CAACZ,KAAK,CAACE,GAAP,CAAJ,CAAgBE,OAAhB,CAAwBa,IAAxB,CAA6Ba,IAA7B;AACAlB,QAAAA,IAAI,CAACZ,KAAK,CAACE,GAAP,CAAJ,CAAgBC,OAAhB,CAAwBc,IAAxB,CAA6BjB,KAAK,CAACgC,SAAnC;AACApB,QAAAA,IAAI,CAACZ,KAAK,CAACE,GAAP,CAAJ,CAAgBG,MAAhB,CAAuBY,IAAvB,CACErB,KAAK,CAACqC,WAAN,CAAmBjC,KAAK,CAACkC,YAAN,IAAsBlC,KAAK,CAACmC,aAA/C,EAAgE9B,MAAM,CAACwB,CAAD,CAAtE,CADF;AAGD;AACF;AACF;AACF;;AAqBD,OAAO,SAASO,kBAAT,CAA4BtC,MAA5B,EAAmDa,MAAnD,EAA2F;AAChG,QAAM0B,OAAwB,GAAG,EAAjC;;AACA,OAAK,IAAIR,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGlB,MAAM,CAACH,QAA3B,EAAqCqB,CAAC,EAAtC,EAA0CQ,OAAO,CAACpB,IAAR,CAAa,EAAb;;AAC1C,OAAK,MAAMf,GAAX,IAAkBS,MAAM,CAACZ,UAAzB,EAAqC;AACnCuC,IAAAA,iBAAiB,CAACxC,MAAD,EAASa,MAAT,EAAiBT,GAAjB,EAAsBmC,OAAtB,CAAjB;AACD;;AACD,SAAOA,OAAP;AACD;;AAGD,SAASC,iBAAT,CACExC,MADF,EAEEa,MAFF,EAGET,GAHF,EAIEmC,OAJF,EAKE;AACA,QAAMzB,IAAI,GAAGD,MAAM,CAACZ,UAAP,CAAkBG,GAAlB,CAAb;AACA,MAAI,CAACU,IAAI,CAACL,KAAV,EAAiB;AAEjB,QAAMP,KAAK,GAAGF,MAAM,CAACyC,SAAP,CAAiBrC,GAAjB,CAAd;AACA,QAAMsC,MAAM,GAAG1C,MAAM,CAAC2C,eAAP,CAAuBvC,GAAvB,CAAf;AAGA,QAAMwC,OAAiB,GAAG,IAAI3B,KAAJ,CAAUf,KAAK,CAAC+B,SAAN,GAAkB,CAA5B,EAA+BY,IAA/B,CAAoC,CAApC,CAA1B;AACA,MAAIC,MAAM,GAAG,CAAb;;AACA,OAAK,IAAIf,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGjB,IAAI,CAACL,KAAzB,EAAgCsB,CAAC,EAAjC,EAAqC;AACnC,UAAMT,MAAM,GAAGR,IAAI,CAACT,OAAL,CAAa0B,CAAb,CAAf;AACA,UAAMV,MAAM,GAAGP,IAAI,CAACR,OAAL,CAAayB,CAAb,CAAf;AACAa,IAAAA,OAAO,CAACvB,MAAD,CAAP;AACAuB,IAAAA,OAAO,CAACC,IAAR,CAAa,CAAb,EAAgBxB,MAAM,GAAG,CAAzB;AAEA,QAAI0B,MAAM,GAAG,CAAb;AACA,QAAInC,MAAM,GAAG2B,OAAO,CAACK,OAAO,CAACG,MAAM,EAAP,CAAP,GAAoB,CAArB,CAApB;;AAGA,SAAK,MAAMC,IAAX,IAAmBN,MAAnB,EAA2B;AACzB,UAAIM,IAAI,KAAK9C,KAAb,EAAoB;AACpB,UAAIoB,MAAM,GAAG0B,IAAI,CAACd,SAAlB,EAA6B;;AAC7B,UAAIc,IAAI,CAACpB,cAAL,KAAwB,UAA5B,EAAwC;AACtC,YAAI,EAAEoB,IAAI,CAACzB,IAAL,IAAaX,MAAf,CAAJ,EAA4B;AAE1BA,UAAAA,MAAM,CAACoC,IAAI,CAACzB,IAAN,CAAN,GAAoB,EAApB;AACD;;AACD,cAAM0B,EAAE,GAAGL,OAAO,CAACG,MAAM,EAAP,CAAlB;;AACA,eAAOnC,MAAM,CAACoC,IAAI,CAACzB,IAAN,CAAN,CAAkBG,MAAlB,IAA4BuB,EAAnC,EAAuC;AAErCrC,UAAAA,MAAM,CAACoC,IAAI,CAACzB,IAAN,CAAN,CAAkBJ,IAAlB,CAAuB,EAAvB;AACD;;AACDP,QAAAA,MAAM,GAAGA,MAAM,CAACoC,IAAI,CAACzB,IAAN,CAAN,CAAkB0B,EAAlB,CAAT;AACD,OAXD,MAWO;AACLrC,QAAAA,MAAM,CAACoC,IAAI,CAACzB,IAAN,CAAN,GAAoBX,MAAM,CAACoC,IAAI,CAACzB,IAAN,CAAN,IAAqB,EAAzC;AACAX,QAAAA,MAAM,GAAGA,MAAM,CAACoC,IAAI,CAACzB,IAAN,CAAf;AACD;AACF;;AAGD,QAAID,MAAM,KAAKpB,KAAK,CAACgC,SAArB,EAAgC;AAC9B,YAAMgB,KAAK,GAAGpD,KAAK,CAACqD,aAAN,CAEZjD,KAAK,CAACkC,YAAN,IAAsBlC,KAAK,CAACmC,aAFhB,EAGZvB,IAAI,CAACP,MAAL,CAAYuC,MAAZ,CAHY,EAIZ5C,KAJY,CAAd;AAMA4C,MAAAA,MAAM;;AACN,UAAI5C,KAAK,CAAC0B,cAAN,KAAyB,UAA7B,EAAyC;AACvC,YAAI,EAAE1B,KAAK,CAACqB,IAAN,IAAcX,MAAhB,CAAJ,EAA6B;AAE3BA,UAAAA,MAAM,CAACV,KAAK,CAACqB,IAAP,CAAN,GAAqB,EAArB;AACD;;AACD,cAAM0B,EAAE,GAAGL,OAAO,CAACG,MAAD,CAAlB;;AACA,eAAOnC,MAAM,CAACV,KAAK,CAACqB,IAAP,CAAN,CAAmBG,MAAnB,IAA6BuB,EAApC,EAAwC;AAEtCrC,UAAAA,MAAM,CAACV,KAAK,CAACqB,IAAP,CAAN,CAAmBJ,IAAnB,CAAwB,IAAxB;AACD;;AACDP,QAAAA,MAAM,CAACV,KAAK,CAACqB,IAAP,CAAN,CAAmB0B,EAAnB,IAAyBC,KAAzB;AACD,OAXD,MAWO;AACLtC,QAAAA,MAAM,CAACV,KAAK,CAACqB,IAAP,CAAN,GAAqB2B,KAArB;AACD;AACF;AACF;AACF","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\n\nimport {ParquetBuffer, ParquetData, ParquetField, ParquetRecord} from './declare';\nimport {ParquetSchema} from './schema';\nimport * as Types from './types';\n\nexport {ParquetBuffer};\n\nexport function shredBuffer(schema: ParquetSchema): ParquetBuffer {\n  const columnData: Record<string, ParquetData> = {};\n  for (const field of schema.fieldList) {\n    columnData[field.key] = {\n      dlevels: [],\n      rlevels: [],\n      values: [],\n      pageHeaders: [],\n      count: 0\n    };\n  }\n  return {rowCount: 0, columnData};\n}\n\n/**\n * 'Shred' a record into a list of <value, repetition_level, definition_level>\n * tuples per column using the Google Dremel Algorithm..\n *\n * The buffer argument must point to an object into which the shredded record\n * will be returned. You may re-use the buffer for repeated calls to this function\n * to append to an existing buffer, as long as the schema is unchanged.\n *\n * The format in which the shredded records will be stored in the buffer is as\n * follows:\n *\n *   buffer = {\n *     columnData: [\n *       'my_col': {\n *          dlevels: [d1, d2, .. dN],\n *          rlevels: [r1, r2, .. rN],\n *          values: [v1, v2, .. vN],\n *        }, ...\n *      ],\n *      rowCount: X,\n *   }\n */\nexport function shredRecord(schema: ParquetSchema, record: any, buffer: ParquetBuffer): void {\n  /* shred the record, this may raise an exception */\n  const data = shredBuffer(schema).columnData;\n\n  shredRecordFields(schema.fields, record, data, 0, 0);\n\n  /* if no error during shredding, add the shredded record to the buffer */\n  if (buffer.rowCount === 0) {\n    buffer.rowCount = 1;\n    buffer.columnData = data;\n    return;\n  }\n  buffer.rowCount += 1;\n  for (const field of schema.fieldList) {\n    Array.prototype.push.apply(buffer.columnData[field.key].rlevels, data[field.key].rlevels);\n    Array.prototype.push.apply(buffer.columnData[field.key].dlevels, data[field.key].dlevels);\n    Array.prototype.push.apply(buffer.columnData[field.key].values, data[field.key].values);\n    buffer.columnData[field.key].count += data[field.key].count;\n  }\n}\n\n// eslint-disable-next-line max-statements, complexity\nfunction shredRecordFields(\n  fields: Record<string, ParquetField>,\n  record: any,\n  data: Record<string, ParquetData>,\n  rLevel: number,\n  dLevel: number\n) {\n  for (const name in fields) {\n    const field = fields[name];\n\n    // fetch values\n    let values: any[] = [];\n    if (\n      record &&\n      field.name in record &&\n      record[field.name] !== undefined &&\n      record[field.name] !== null\n    ) {\n      if (record[field.name].constructor === Array) {\n        values = record[field.name];\n      } else {\n        values.push(record[field.name]);\n      }\n    }\n    // check values\n    if (values.length === 0 && Boolean(record) && field.repetitionType === 'REQUIRED') {\n      throw new Error(`missing required field: ${field.name}`);\n    }\n    if (values.length > 1 && field.repetitionType !== 'REPEATED') {\n      throw new Error(`too many values for field: ${field.name}`);\n    }\n\n    // push null\n    if (values.length === 0) {\n      if (field.isNested) {\n        shredRecordFields(field.fields!, null, data, rLevel, dLevel);\n      } else {\n        data[field.key].count += 1;\n        data[field.key].rlevels.push(rLevel);\n        data[field.key].dlevels.push(dLevel);\n      }\n      continue; // eslint-disable-line no-continue\n    }\n\n    // push values\n    for (let i = 0; i < values.length; i++) {\n      const rlvl = i === 0 ? rLevel : field.rLevelMax;\n      if (field.isNested) {\n        shredRecordFields(field.fields!, values[i], data, rlvl, field.dLevelMax);\n      } else {\n        data[field.key].count += 1;\n        data[field.key].rlevels.push(rlvl);\n        data[field.key].dlevels.push(field.dLevelMax);\n        data[field.key].values.push(\n          Types.toPrimitive((field.originalType || field.primitiveType)!, values[i])\n        );\n      }\n    }\n  }\n}\n\n/**\n * 'Materialize' a list of <value, repetition_level, definition_level>\n * tuples back to nested records (objects/arrays) using the Google Dremel\n * Algorithm..\n *\n * The buffer argument must point to an object with the following structure (i.e.\n * the same structure that is returned by shredRecords):\n *\n *   buffer = {\n *     columnData: [\n *       'my_col': {\n *          dlevels: [d1, d2, .. dN],\n *          rlevels: [r1, r2, .. rN],\n *          values: [v1, v2, .. vN],\n *        }, ...\n *      ],\n *      rowCount: X,\n *   }\n */\nexport function materializeRecords(schema: ParquetSchema, buffer: ParquetBuffer): ParquetRecord[] {\n  const records: ParquetRecord[] = [];\n  for (let i = 0; i < buffer.rowCount; i++) records.push({});\n  for (const key in buffer.columnData) {\n    materializeColumn(schema, buffer, key, records);\n  }\n  return records;\n}\n\n// eslint-disable-next-line max-statements, complexity\nfunction materializeColumn(\n  schema: ParquetSchema,\n  buffer: ParquetBuffer,\n  key: string,\n  records: ParquetRecord[]\n) {\n  const data = buffer.columnData[key];\n  if (!data.count) return;\n\n  const field = schema.findField(key);\n  const branch = schema.findFieldBranch(key);\n\n  // tslint:disable-next-line:prefer-array-literal\n  const rLevels: number[] = new Array(field.rLevelMax + 1).fill(0);\n  let vIndex = 0;\n  for (let i = 0; i < data.count; i++) {\n    const dLevel = data.dlevels[i];\n    const rLevel = data.rlevels[i];\n    rLevels[rLevel]++;\n    rLevels.fill(0, rLevel + 1);\n\n    let rIndex = 0;\n    let record = records[rLevels[rIndex++] - 1];\n\n    // Internal nodes\n    for (const step of branch) {\n      if (step === field) break;\n      if (dLevel < step.dLevelMax) break;\n      if (step.repetitionType === 'REPEATED') {\n        if (!(step.name in record)) {\n          // eslint-disable max-depth\n          record[step.name] = [];\n        }\n        const ix = rLevels[rIndex++];\n        while (record[step.name].length <= ix) {\n          // eslint-disable max-depth\n          record[step.name].push({});\n        }\n        record = record[step.name][ix];\n      } else {\n        record[step.name] = record[step.name] || {};\n        record = record[step.name];\n      }\n    }\n\n    // Leaf node\n    if (dLevel === field.dLevelMax) {\n      const value = Types.fromPrimitive(\n        // @ts-ignore\n        field.originalType || field.primitiveType,\n        data.values[vIndex],\n        field\n      );\n      vIndex++;\n      if (field.repetitionType === 'REPEATED') {\n        if (!(field.name in record)) {\n          // eslint-disable max-depth\n          record[field.name] = [];\n        }\n        const ix = rLevels[rIndex];\n        while (record[field.name].length <= ix) {\n          // eslint-disable max-depth\n          record[field.name].push(null);\n        }\n        record[field.name][ix] = value;\n      } else {\n        record[field.name] = value;\n      }\n    }\n  }\n}\n"],"file":"shred.js"}

package/dist/esm/parquetjs/schema/types.js

@@ -125,16 +125,40 @@ export const PARQUET_LOGICAL_TYPES = {
     typeLength: 12,
     toPrimitive: toPrimitive_INTERVAL,
     fromPrimitive: fromPrimitive_INTERVAL
+  },
+  DECIMAL_INT32: {
+    primitiveType: 'INT32',
+    originalType: 'DECIMAL_INT32',
+    toPrimitive: decimalToPrimitive_INT32,
+    fromPrimitive: decimalFromPrimitive_INT
+  },
+  DECIMAL_INT64: {
+    primitiveType: 'INT64',
+    originalType: 'DECIMAL_INT64',
+    toPrimitive: decimalToPrimitive_INT64,
+    fromPrimitive: decimalFromPrimitive_INT
+  },
+  DECIMAL_BYTE_ARRAY: {
+    primitiveType: 'BYTE_ARRAY',
+    originalType: 'DECIMAL_BYTE_ARRAY',
+    toPrimitive: decimalToPrimitive_BYTE_ARRAY,
+    fromPrimitive: decimalFromPrimitive_BYTE_ARRAY
+  },
+  DECIMAL_FIXED_LEN_BYTE_ARRAY: {
+    primitiveType: 'FIXED_LEN_BYTE_ARRAY',
+    originalType: 'DECIMAL_FIXED_LEN_BYTE_ARRAY',
+    toPrimitive: decimalToPrimitive_BYTE_ARRAY,
+    fromPrimitive: decimalFromPrimitive_BYTE_ARRAY
   }
 };
-export function toPrimitive(type, value) {
+export function toPrimitive(type, value, field) {
   if (!(type in PARQUET_LOGICAL_TYPES)) {
     throw new Error(`invalid type: ${type}`);
   }
 
-  return PARQUET_LOGICAL_TYPES[type].toPrimitive(value);
+  return PARQUET_LOGICAL_TYPES[type].toPrimitive(value, field);
 }
-export function fromPrimitive(type, value) {
+export function fromPrimitive(type, value, field) {
   if (!(type in PARQUET_LOGICAL_TYPES)) {
     throw new Error(`invalid type: ${type}`);
   }
@@ -142,7 +166,7 @@ export function fromPrimitive(type, value) {
   if ('fromPrimitive' in PARQUET_LOGICAL_TYPES[type]) {
     var _PARQUET_LOGICAL_TYPE, _PARQUET_LOGICAL_TYPE2;
 
-    return (_PARQUET_LOGICAL_TYPE = (_PARQUET_LOGICAL_TYPE2 = PARQUET_LOGICAL_TYPES[type]).fromPrimitive) === null || _PARQUET_LOGICAL_TYPE === void 0 ? void 0 : _PARQUET_LOGICAL_TYPE.call(_PARQUET_LOGICAL_TYPE2, value);
+    return (_PARQUET_LOGICAL_TYPE = (_PARQUET_LOGICAL_TYPE2 = PARQUET_LOGICAL_TYPES[type]).fromPrimitive) === null || _PARQUET_LOGICAL_TYPE === void 0 ? void 0 : _PARQUET_LOGICAL_TYPE.call(_PARQUET_LOGICAL_TYPE2, value, field);
   }
 
   return value;
@@ -226,6 +250,17 @@ function toPrimitive_INT32(value) {
   return v;
 }
 
+function decimalToPrimitive_INT32(value, field) {
+  const primitiveValue = value * 10 ** (field.scale || 0);
+  const v = Math.round(primitiveValue * 10 ** -field.presision % 1 * 10 ** field.presision);
+
+  if (v < -0x80000000 || v > 0x7fffffff || isNaN(v)) {
+    throw new Error(`invalid value for INT32: ${value}`);
+  }
+
+  return v;
+}
+
 function toPrimitive_UINT32(value) {
   const v = parseInt(value, 10);
 
@@ -246,6 +281,17 @@ function toPrimitive_INT64(value) {
   return v;
 }
 
+function decimalToPrimitive_INT64(value, field) {
+  const primitiveValue = value * 10 ** (field.scale || 0);
+  const v = Math.round(primitiveValue * 10 ** -field.presision % 1 * 10 ** field.presision);
+
+  if (isNaN(v)) {
+    throw new Error(`invalid value for INT64: ${value}`);
+  }
+
+  return v;
+}
+
 function toPrimitive_UINT64(value) {
   const v = parseInt(value, 10);
 
@@ -270,6 +316,10 @@ function toPrimitive_BYTE_ARRAY(value) {
   return Buffer.from(value);
 }
 
+function decimalToPrimitive_BYTE_ARRAY(value) {
+  return Buffer.from(value);
+}
+
 function toPrimitive_UTF8(value) {
   return Buffer.from(value, 'utf8');
 }
@@ -399,4 +449,28 @@ function fromPrimitive_INTERVAL(value) {
     milliseconds: millis
   };
 }
+
+function decimalFromPrimitive_INT(value, field) {
+  const presisionInt = Math.round(value * 10 ** -field.presision % 1 * 10 ** field.presision);
+  return presisionInt * 10 ** -(field.scale || 0);
+}
+
+function decimalFromPrimitive_BYTE_ARRAY(value, field) {
+  let number = 0;
+
+  if (value.length <= 4) {
+    for (let i = 0; i < value.length; i++) {
+      const component = value[i] << 8 * (value.length - i - 1);
+      number += component;
+    }
+  } else {
+    for (let i = 0; i < value.length; i++) {
+      const component = value[i] * 2 ** (8 * (value.length - 1 - i));
+      number += component;
+    }
+  }
+
+  const presisionInt = Math.round(number * 10 ** -field.presision % 1 * 10 ** field.presision);
+  return presisionInt * 10 ** -(field.scale || 0);
+}
 //# sourceMappingURL=types.js.map