@loaders.gl/parquet 3.1.3 → 4.0.0-alpha.5

package/dist/{es5/parquetjs → parquetjs}/schema/schema.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../../../src/parquetjs/schema/schema.ts"],"names":["ParquetSchema","schema","fields","buildFields","fieldList","listFields","path","split","slice","n","length","shift","branch","push","record","buffer","type","setCompress","name","node","compression","rLevelParentMax","dLevelParentMax","opts","required","optional","repeated","Boolean","rLevelMax","dLevelMax","repetitionType","cpath","concat","key","join","isNested","fieldCount","Object","keys","typeDef","PARQUET_LOGICAL_TYPES","Error","encoding","PARQUET_CODECS","PARQUET_COMPRESSION_METHODS","primitiveType","originalType","typeLength","presision","scale","list","k"],"mappings":";;;;;;;;;;;;;;;AAEA;;AACA;;AAUA;;AACA;;IAKaA,a;AAQX,yBAAYC,MAAZ,EAAsC;AAAA;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;;;;WAKD,mBAAUI,IAAV,EAAiD;AAC/C,UAAI,OAAOA,IAAP,KAAgB,QAApB,EAA8B;AAE5BA,QAAAA,IAAI,GAAGA,IAAI,CAACC,KAAL,CAAW,GAAX,CAAP;AACD,OAHD,MAGO;AAELD,QAAAA,IAAI,GAAGA,IAAI,CAACE,KAAL,CAAW,CAAX,CAAP;AACD;;AAED,UAAIC,CAAC,GAAG,KAAKP,MAAb;;AACA,aAAOI,IAAI,CAACI,MAAL,GAAc,CAArB,EAAwBJ,IAAI,CAACK,KAAL,EAAxB,EAAsC;AACpCF,QAAAA,CAAC,GAAGA,CAAC,CAACH,IAAI,CAAC,CAAD,CAAL,CAAD,CAAWJ,MAAf;AACD;;AAED,aAAOO,CAAC,CAACH,IAAI,CAAC,CAAD,CAAL,CAAR;AACD;;;WAKD,yBAAgBA,IAAhB,EAAyD;AACvD,UAAI,OAAOA,IAAP,KAAgB,QAApB,EAA8B;AAE5BA,QAAAA,IAAI,GAAGA,IAAI,CAACC,KAAL,CAAW,GAAX,CAAP;AACD;;AACD,UAAMK,MAAsB,GAAG,EAA/B;AACA,UAAIH,CAAC,GAAG,KAAKP,MAAb;;AACA,aAAOI,IAAI,CAACI,MAAL,GAAc,CAArB,EAAwBJ,IAAI,CAACK,KAAL,EAAxB,EAAsC;AACpCC,QAAAA,MAAM,CAACC,IAAP,CAAYJ,CAAC,CAACH,IAAI,CAAC,CAAD,CAAL,CAAb;;AACA,YAAIA,IAAI,CAACI,MAAL,GAAc,CAAlB,EAAqB;AACnBD,UAAAA,CAAC,GAAGA,CAAC,CAACH,IAAI,CAAC,CAAD,CAAL,CAAD,CAAWJ,MAAf;AACD;AACF;;AACD,aAAOU,MAAP;AACD;;;WAED,qBAAYE,MAAZ,EAAmCC,MAAnC,EAAgE;AAC9D,8BAAY,IAAZ,EAAkBD,MAAlB,EAA0BC,MAA1B;AACD;;;WAED,4BAAmBA,MAAnB,EAA2D;AACzD,aAAO,+BAAmB,IAAnB,EAAyBA,MAAzB,CAAP;AACD;;;WAED,kBAASC,IAAT,EAAyC;AACvCC,MAAAA,WAAW,CAAC,KAAKhB,MAAN,EAAce,IAAd,CAAX;AACAC,MAAAA,WAAW,CAAC,KAAKf,MAAN,EAAcc,IAAd,CAAX;AACA,aAAO,IAAP;AACD;;;WAED,kBAAwB;AACtB,aAAO,wBAAY,IAAZ,CAAP;AACD;;;;;;;AAGH,SAASC,WAAT,CAAqBhB,MAArB,EAAkCe,IAAlC,EAA4D;AAC1D,OAAK,IAAME,IAAX,IAAmBjB,MAAnB,EAA2B;AACzB,QAAMkB,IAAI,GAAGlB,MAAM,CAACiB,IAAD,CAAnB;;AACA,QAAIC,IAAI,CAACjB,MAAT,EAAiB;AACfe,MAAAA,WAAW,CAACE,IAAI,CAACjB,MAAN,EAAcc,IAAd,CAAX;AACD,KAFD,MAEO;AACLG,MAAAA,IAAI,CAACC,WAAL,GAAmBJ,IAAnB;AACD;AACF;AACF;;AAGD,SAASb,WAAT,CACEF,MADF,EAEEoB,eAFF,EAGEC,eAHF,EAIEhB,IAJF,EAKgC;AAC9B,MAAMF,SAAuC,GAAG,EAAhD;;AAEA,OAAK,IAAMc,IAAX,IAAmBjB,MAAnB,EAA2B;AACzB,QAAMsB,IAAI,GAAGtB,MAAM,CAACiB,IAAD,CAAnB;AAGA,QAAMM,QAAQ,GAAG,CAACD,IAAI,CAACE,QAAvB;AACA,QAAMC,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,CAACrB,MAAT,EAAiB;AACf,UAAM6B,MAAK,GAAGzB,IAAI,CAAC0B,MAAL,CAAY,CAACd,IAAD,CAAZ,CAAd;;AACAd,MAAAA,SAAS,CAACc,IAAD,CAAT,GAAkB;AAChBA,QAAAA,IAAI,EAAJA,IADgB;AAEhBZ,QAAAA,IAAI,EAAEyB,MAFU;AAGhBE,QAAAA,GAAG,EAAEF,MAAK,CAACG,IAAN,EAHW;AAIhBJ,QAAAA,cAAc,EAAdA,cAJgB;AAKhBF,QAAAA,SAAS,EAATA,SALgB;AAMhBC,QAAAA,SAAS,EAATA,SANgB;AAOhBM,QAAAA,QAAQ,EAAE,IAPM;AAQhBC,QAAAA,UAAU,EAAEC,MAAM,CAACC,IAAP,CAAYf,IAAI,CAACrB,MAAjB,EAAyBQ,MARrB;AAShBR,QAAAA,MAAM,EAAEC,WAAW,CAACoB,IAAI,CAACrB,MAAN,EAAc0B,SAAd,EAAyBC,SAAzB,EAAoCE,MAApC;AATH,OAAlB;AAWA;AACD;;AAED,QAAMQ,OAAY,GAAGC,6BAAsBjB,IAAI,CAACP,IAA3B,CAArB;;AACA,QAAI,CAACuB,OAAL,EAAc;AACZ,YAAM,IAAIE,KAAJ,iCAAmClB,IAAI,CAACP,IAAxC,EAAN;AACD;;AAEDO,IAAAA,IAAI,CAACmB,QAAL,GAAgBnB,IAAI,CAACmB,QAAL,IAAiB,OAAjC;;AACA,QAAI,EAAEnB,IAAI,CAACmB,QAAL,IAAiBC,sBAAnB,CAAJ,EAAwC;AACtC,YAAM,IAAIF,KAAJ,yCAA2ClB,IAAI,CAACmB,QAAhD,EAAN;AACD;;AAEDnB,IAAAA,IAAI,CAACH,WAAL,GAAmBG,IAAI,CAACH,WAAL,IAAoB,cAAvC;;AACA,QAAI,EAAEG,IAAI,CAACH,WAAL,IAAoBwB,wCAAtB,CAAJ,EAAwD;AACtD,YAAM,IAAIH,KAAJ,2CAA6ClB,IAAI,CAACH,WAAlD,EAAN;AACD;;AAGD,QAAMW,KAAK,GAAGzB,IAAI,CAAC0B,MAAL,CAAY,CAACd,IAAD,CAAZ,CAAd;AACAd,IAAAA,SAAS,CAACc,IAAD,CAAT,GAAkB;AAChBA,MAAAA,IAAI,EAAJA,IADgB;AAEhB2B,MAAAA,aAAa,EAAEN,OAAO,CAACM,aAFP;AAGhBC,MAAAA,YAAY,EAAEP,OAAO,CAACO,YAHN;AAIhBxC,MAAAA,IAAI,EAAEyB,KAJU;AAKhBE,MAAAA,GAAG,EAAEF,KAAK,CAACG,IAAN,EALW;AAMhBJ,MAAAA,cAAc,EAAdA,cANgB;AAOhBY,MAAAA,QAAQ,EAAEnB,IAAI,CAACmB,QAPC;AAQhBtB,MAAAA,WAAW,EAAEG,IAAI,CAACH,WARF;AAShB2B,MAAAA,UAAU,EAAExB,IAAI,CAACwB,UAAL,IAAmBR,OAAO,CAACQ,UATvB;AAUhBC,MAAAA,SAAS,EAAEzB,IAAI,CAACyB,SAVA;AAWhBC,MAAAA,KAAK,EAAE1B,IAAI,CAAC0B,KAXI;AAYhBrB,MAAAA,SAAS,EAATA,SAZgB;AAahBC,MAAAA,SAAS,EAATA;AAbgB,KAAlB;AAeD;;AACD,SAAOzB,SAAP;AACD;;AAED,SAASC,UAAT,CAAoBH,MAApB,EAA0E;AACxE,MAAIgD,IAAoB,GAAG,EAA3B;;AACA,OAAK,IAAMC,CAAX,IAAgBjD,MAAhB,EAAwB;AACtBgD,IAAAA,IAAI,CAACrC,IAAL,CAAUX,MAAM,CAACiD,CAAD,CAAhB;;AACA,QAAIjD,MAAM,CAACiD,CAAD,CAAN,CAAUhB,QAAd,EAAwB;AACtBe,MAAAA,IAAI,GAAGA,IAAI,CAAClB,MAAL,CAAY3B,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"}
+{"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,iCAAmCjB,IAAI,CAACP,IAAxC,EAAN;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,yCAA2CjB,IAAI,CAACkB,QAAhD,EAAN;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,2CAA6CjB,IAAI,CAACH,WAAlD,EAAN;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/parquetjs/schema/shred.js

@@ -1,225 +1,162 @@
-"use strict";
-// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)
-var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
-    if (k2 === undefined) k2 = k;
-    Object.defineProperty(o, k2, { enumerable: true, get: function() { return m[k]; } });
-}) : (function(o, m, k, k2) {
-    if (k2 === undefined) k2 = k;
-    o[k2] = m[k];
-}));
-var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
-    Object.defineProperty(o, "default", { enumerable: true, value: v });
-}) : function(o, v) {
-    o["default"] = v;
-});
-var __importStar = (this && this.__importStar) || function (mod) {
-    if (mod && mod.__esModule) return mod;
-    var result = {};
-    if (mod != null) for (var k in mod) if (k !== "default" && Object.prototype.hasOwnProperty.call(mod, k)) __createBinding(result, mod, k);
-    __setModuleDefault(result, mod);
-    return result;
-};
-Object.defineProperty(exports, "__esModule", { value: true });
-exports.materializeRecords = exports.shredRecord = exports.shredBuffer = exports.ParquetBuffer = void 0;
-const declare_1 = require("./declare");
-Object.defineProperty(exports, "ParquetBuffer", { enumerable: true, get: function () { return declare_1.ParquetBuffer; } });
-const Types = __importStar(require("./types"));
-function shredBuffer(schema) {
-    const columnData = {};
-    for (const field of schema.fieldList) {
-        columnData[field.key] = {
-            dlevels: [],
-            rlevels: [],
-            values: [],
-            pageHeaders: [],
-            count: 0
-        };
-    }
-    return { rowCount: 0, columnData };
+import { ParquetBuffer } from './declare';
+import * as Types from './types';
+export { ParquetBuffer };
+export function shredBuffer(schema) {
+  const columnData = {};
+
+  for (const field of schema.fieldList) {
+    columnData[field.key] = {
+      dlevels: [],
+      rlevels: [],
+      values: [],
+      pageHeaders: [],
+      count: 0
+    };
+  }
+
+  return {
+    rowCount: 0,
+    columnData
+  };
 }
-exports.shredBuffer = shredBuffer;
-/**
- * 'Shred' a record into a list of <value, repetition_level, definition_level>
- * tuples per column using the Google Dremel Algorithm..
- *
- * The buffer argument must point to an object into which the shredded record
- * will be returned. You may re-use the buffer for repeated calls to this function
- * to append to an existing buffer, as long as the schema is unchanged.
- *
- * The format in which the shredded records will be stored in the buffer is as
- * follows:
- *
- *   buffer = {
- *     columnData: [
- *       'my_col': {
- *          dlevels: [d1, d2, .. dN],
- *          rlevels: [r1, r2, .. rN],
- *          values: [v1, v2, .. vN],
- *        }, ...
- *      ],
- *      rowCount: X,
- *   }
- */
-function shredRecord(schema, record, buffer) {
-    /* shred the record, this may raise an exception */
-    const data = shredBuffer(schema).columnData;
-    shredRecordFields(schema.fields, record, data, 0, 0);
-    /* if no error during shredding, add the shredded record to the buffer */
-    if (buffer.rowCount === 0) {
-        buffer.rowCount = 1;
-        buffer.columnData = data;
-        return;
-    }
-    buffer.rowCount += 1;
-    for (const field of schema.fieldList) {
-        Array.prototype.push.apply(buffer.columnData[field.key].rlevels, data[field.key].rlevels);
-        Array.prototype.push.apply(buffer.columnData[field.key].dlevels, data[field.key].dlevels);
-        Array.prototype.push.apply(buffer.columnData[field.key].values, data[field.key].values);
-        buffer.columnData[field.key].count += data[field.key].count;
-    }
+export function shredRecord(schema, record, buffer) {
+  const data = shredBuffer(schema).columnData;
+  shredRecordFields(schema.fields, record, data, 0, 0);
+
+  if (buffer.rowCount === 0) {
+    buffer.rowCount = 1;
+    buffer.columnData = data;
+    return;
+  }
+
+  buffer.rowCount += 1;
+
+  for (const field of schema.fieldList) {
+    Array.prototype.push.apply(buffer.columnData[field.key].rlevels, data[field.key].rlevels);
+    Array.prototype.push.apply(buffer.columnData[field.key].dlevels, data[field.key].dlevels);
+    Array.prototype.push.apply(buffer.columnData[field.key].values, data[field.key].values);
+    buffer.columnData[field.key].count += data[field.key].count;
+  }
 }
-exports.shredRecord = shredRecord;
-// eslint-disable-next-line max-statements, complexity
+
 function shredRecordFields(fields, record, data, rLevel, dLevel) {
-    for (const name in fields) {
-        const field = fields[name];
-        // fetch values
-        let values = [];
-        if (record &&
-            field.name in record &&
-            record[field.name] !== undefined &&
-            record[field.name] !== null) {
-            if (record[field.name].constructor === Array) {
-                values = record[field.name];
-            }
-            else {
-                values.push(record[field.name]);
-            }
-        }
-        // check values
-        if (values.length === 0 && Boolean(record) && field.repetitionType === 'REQUIRED') {
-            throw new Error(`missing required field: ${field.name}`);
-        }
-        if (values.length > 1 && field.repetitionType !== 'REPEATED') {
-            throw new Error(`too many values for field: ${field.name}`);
-        }
-        // push null
-        if (values.length === 0) {
-            if (field.isNested) {
-                shredRecordFields(field.fields, null, data, rLevel, dLevel);
-            }
-            else {
-                data[field.key].count += 1;
-                data[field.key].rlevels.push(rLevel);
-                data[field.key].dlevels.push(dLevel);
-            }
-            continue; // eslint-disable-line no-continue
-        }
-        // push values
-        for (let i = 0; i < values.length; i++) {
-            const rlvl = i === 0 ? rLevel : field.rLevelMax;
-            if (field.isNested) {
-                shredRecordFields(field.fields, values[i], data, rlvl, field.dLevelMax);
-            }
-            else {
-                data[field.key].count += 1;
-                data[field.key].rlevels.push(rlvl);
-                data[field.key].dlevels.push(field.dLevelMax);
-                data[field.key].values.push(Types.toPrimitive((field.originalType || field.primitiveType), values[i]));
-            }
-        }
+  for (const name in fields) {
+    const field = fields[name];
+    let values = [];
+
+    if (record && field.name in record && record[field.name] !== undefined && record[field.name] !== null) {
+      if (record[field.name].constructor === Array) {
+        values = record[field.name];
+      } else {
+        values.push(record[field.name]);
+      }
     }
-}
-/**
- * 'Materialize' a list of <value, repetition_level, definition_level>
- * tuples back to nested records (objects/arrays) using the Google Dremel
- * Algorithm..
- *
- * The buffer argument must point to an object with the following structure (i.e.
- * the same structure that is returned by shredRecords):
- *
- *   buffer = {
- *     columnData: [
- *       'my_col': {
- *          dlevels: [d1, d2, .. dN],
- *          rlevels: [r1, r2, .. rN],
- *          values: [v1, v2, .. vN],
- *        }, ...
- *      ],
- *      rowCount: X,
- *   }
- */
-function materializeRecords(schema, buffer) {
-    const records = [];
-    for (let i = 0; i < buffer.rowCount; i++)
-        records.push({});
-    for (const key in buffer.columnData) {
-        materializeColumn(schema, buffer, key, records);
+
+    if (values.length === 0 && Boolean(record) && field.repetitionType === 'REQUIRED') {
+      throw new Error("missing required field: ".concat(field.name));
+    }
+
+    if (values.length > 1 && field.repetitionType !== 'REPEATED') {
+      throw new Error("too many values for field: ".concat(field.name));
+    }
+
+    if (values.length === 0) {
+      if (field.isNested) {
+        shredRecordFields(field.fields, null, data, rLevel, dLevel);
+      } else {
+        data[field.key].count += 1;
+        data[field.key].rlevels.push(rLevel);
+        data[field.key].dlevels.push(dLevel);
+      }
+
+      continue;
+    }
+
+    for (let i = 0; i < values.length; i++) {
+      const rlvl = i === 0 ? rLevel : field.rLevelMax;
+
+      if (field.isNested) {
+        shredRecordFields(field.fields, values[i], data, rlvl, field.dLevelMax);
+      } else {
+        data[field.key].count += 1;
+        data[field.key].rlevels.push(rlvl);
+        data[field.key].dlevels.push(field.dLevelMax);
+        data[field.key].values.push(Types.toPrimitive(field.originalType || field.primitiveType, values[i]));
+      }
     }
-    return records;
+  }
 }
-exports.materializeRecords = materializeRecords;
-// eslint-disable-next-line max-statements, complexity
+
+export function materializeRecords(schema, buffer) {
+  const records = [];
+
+  for (let i = 0; i < buffer.rowCount; i++) records.push({});
+
+  for (const key in buffer.columnData) {
+    materializeColumn(schema, buffer, key, records);
+  }
+
+  return records;
+}
+
 function materializeColumn(schema, buffer, key, records) {
-    const data = buffer.columnData[key];
-    if (!data.count)
-        return;
-    const field = schema.findField(key);
-    const branch = schema.findFieldBranch(key);
-    // tslint:disable-next-line:prefer-array-literal
-    const rLevels = new Array(field.rLevelMax + 1).fill(0);
-    let vIndex = 0;
-    for (let i = 0; i < data.count; i++) {
-        const dLevel = data.dlevels[i];
-        const rLevel = data.rlevels[i];
-        rLevels[rLevel]++;
-        rLevels.fill(0, rLevel + 1);
-        let rIndex = 0;
-        let record = records[rLevels[rIndex++] - 1];
-        // Internal nodes
-        for (const step of branch) {
-            if (step === field)
-                break;
-            if (dLevel < step.dLevelMax)
-                break;
-            if (step.repetitionType === 'REPEATED') {
-                if (!(step.name in record)) {
-                    // eslint-disable max-depth
-                    record[step.name] = [];
-                }
-                const ix = rLevels[rIndex++];
-                while (record[step.name].length <= ix) {
-                    // eslint-disable max-depth
-                    record[step.name].push({});
-                }
-                record = record[step.name][ix];
-            }
-            else {
-                record[step.name] = record[step.name] || {};
-                record = record[step.name];
-            }
+  const data = buffer.columnData[key];
+  if (!data.count) return;
+  const field = schema.findField(key);
+  const branch = schema.findFieldBranch(key);
+  const rLevels = new Array(field.rLevelMax + 1).fill(0);
+  let vIndex = 0;
+
+  for (let i = 0; i < data.count; i++) {
+    const dLevel = data.dlevels[i];
+    const rLevel = data.rlevels[i];
+    rLevels[rLevel]++;
+    rLevels.fill(0, rLevel + 1);
+    let rIndex = 0;
+    let record = records[rLevels[rIndex++] - 1];
+
+    for (const step of branch) {
+      if (step === field) break;
+      if (dLevel < step.dLevelMax) break;
+
+      if (step.repetitionType === 'REPEATED') {
+        if (!(step.name in record)) {
+          record[step.name] = [];
+        }
+
+        const ix = rLevels[rIndex++];
+
+        while (record[step.name].length <= ix) {
+          record[step.name].push({});
+        }
+
+        record = record[step.name][ix];
+      } else {
+        record[step.name] = record[step.name] || {};
+        record = record[step.name];
+      }
+    }
+
+    if (dLevel === field.dLevelMax) {
+      const value = Types.fromPrimitive(field.originalType || field.primitiveType, data.values[vIndex], field);
+      vIndex++;
+
+      if (field.repetitionType === 'REPEATED') {
+        if (!(field.name in record)) {
+          record[field.name] = [];
         }
-        // Leaf node
-        if (dLevel === field.dLevelMax) {
-            const value = Types.fromPrimitive(
-            // @ts-ignore
-            field.originalType || field.primitiveType, data.values[vIndex], field);
-            vIndex++;
-            if (field.repetitionType === 'REPEATED') {
-                if (!(field.name in record)) {
-                    // eslint-disable max-depth
-                    record[field.name] = [];
-                }
-                const ix = rLevels[rIndex];
-                while (record[field.name].length <= ix) {
-                    // eslint-disable max-depth
-                    record[field.name].push(null);
-                }
-                record[field.name][ix] = value;
-            }
-            else {
-                record[field.name] = value;
-            }
+
+        const ix = rLevels[rIndex];
+
+        while (record[field.name].length <= ix) {
+          record[field.name].push(null);
         }
+
+        record[field.name][ix] = value;
+      } else {
+        record[field.name] = value;
+      }
     }
+  }
 }
+//# sourceMappingURL=shred.js.map

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

@@ -0,0 +1 @@
+{"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,mCAAqC3B,KAAK,CAACqB,IAA3C,EAAN;AACD;;AACD,QAAIhB,MAAM,CAACmB,MAAP,GAAgB,CAAhB,IAAqBxB,KAAK,CAAC0B,cAAN,KAAyB,UAAlD,EAA8D;AAC5D,YAAM,IAAIC,KAAJ,sCAAwC3B,KAAK,CAACqB,IAA9C,EAAN;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"}