@loaders.gl/parquet 4.0.0-alpha.5 → 4.0.0-alpha.7

package/dist/parquetjs/parser/parquet-reader.d.ts

@@ -1,68 +1,58 @@
-/// <reference types="node" />
-import { ParquetEnvelopeReader } from './parquet-envelope-reader';
-import { FileMetaData } from '../parquet-thrift';
+import type { ReadableFile } from '@loaders.gl/loader-utils';
 import { ParquetSchema } from '../schema/schema';
-import { ParquetCursor } from './parquet-cursor';
+import { ColumnChunk, FileMetaData, RowGroup } from '../parquet-thrift';
+import { ParquetRowGroup, ParquetColumnChunk, ParquetOptions } from '../schema/declare';
+export type ParquetReaderProps = {
+    defaultDictionarySize?: number;
+};
+/** Properties for initializing a ParquetRowGroupReader */
+export type ParquetIterationProps = {
+    /** Filter allowing some columns to be dropped */
+    columnList?: string[] | string[][];
+};
 /**
- * A parquet reader allows retrieving the rows from a parquet file in order.
- * The basic usage is to create a reader and then retrieve a cursor/iterator
- * which allows you to consume row after row until all rows have been read. It is
- * important that you call close() after you are finished reading the file to
- * avoid leaking file descriptors.
+ * The parquet envelope reader allows direct, unbuffered access to the individual
+ * sections of the parquet file, namely the header, footer and the row groups.
+ * This class is intended for advanced/internal users; if you just want to retrieve
+ * rows from a parquet file use the ParquetReader instead
  */
-export declare class ParquetReader<T> implements AsyncIterable<T> {
+export declare class ParquetReader {
+    props: Required<ParquetReaderProps>;
+    file: ReadableFile;
+    metadata: Promise<FileMetaData> | null;
+    constructor(file: ReadableFile, props?: ParquetReaderProps);
+    close(): void;
+    /** Yield one row at a time */
+    rowIterator(props?: ParquetIterationProps): AsyncGenerator<import("../schema/declare").ParquetRow, void, unknown>;
+    /** Yield one batch of rows at a time */
+    rowBatchIterator(props?: ParquetIterationProps): AsyncGenerator<import("../schema/declare").ParquetRow[], void, unknown>;
+    /** Iterate over the raw row groups */
+    rowGroupIterator(props?: ParquetIterationProps): AsyncGenerator<ParquetRowGroup, void, unknown>;
+    getRowCount(): Promise<number>;
+    getSchema(): Promise<ParquetSchema>;
     /**
-     * return a new parquet reader initialized with a read function
-     */
-    static openBlob<T>(blob: Blob): Promise<ParquetReader<T>>;
-    /**
-     * return a new parquet reader initialized with a read function
-     */
-    static openArrayBuffer<T>(arrayBuffer: ArrayBuffer): Promise<ParquetReader<T>>;
-    static openBuffer<T>(buffer: Buffer): Promise<ParquetReader<T>>;
-    metadata: FileMetaData;
-    envelopeReader: ParquetEnvelopeReader;
-    schema: ParquetSchema;
-    /**
-     * Create a new parquet reader from the file metadata and an envelope reader.
-     * It is not recommended to call this constructor directly except for advanced
-     * and internal use cases. Consider using one of the open{File,Buffer} methods
-     * instead
-     */
-    constructor(metadata: FileMetaData, envelopeReader: ParquetEnvelopeReader);
-    /**
-     * Close this parquet reader. You MUST call this method once you're finished
-     * reading rows
-     */
-    close(): Promise<void>;
-    /**
-     * Return a cursor to the file. You may open more than one cursor and use
-     * them concurrently. All cursors become invalid once close() is called on
-     * the reader object.
-     *
-     * The required_columns parameter controls which columns are actually read
-     * from disk. An empty array or no value implies all columns. A list of column
-     * names means that only those columns should be loaded from disk.
-     */
-    getCursor(): ParquetCursor<T>;
-    getCursor<K extends keyof T>(columnList: (K | K[])[]): ParquetCursor<Pick<T, K>>;
-    getCursor(columnList: (string | string[])[]): ParquetCursor<Partial<T>>;
-    /**
-     * Return the number of rows in this file. Note that the number of rows is
-     * not neccessarily equal to the number of rows in each column.
-     */
-    getRowCount(): number;
-    /**
-     * Returns the ParquetSchema for this file
+     * Returns the user (key/value) metadata for this file
+     * In parquet this is not stored on the schema like it is in arrow
      */
-    getSchema(): ParquetSchema;
+    getSchemaMetadata(): Promise<Record<string, string>>;
+    getFileMetadata(): Promise<FileMetaData>;
+    /** Metadata is stored in the footer */
+    readHeader(): Promise<void>;
+    /** Metadata is stored in the footer */
+    readFooter(): Promise<FileMetaData>;
+    /** Data is stored in row groups (similar to Apache Arrow record batches) */
+    readRowGroup(schema: ParquetSchema, rowGroup: RowGroup, columnList: string[][]): Promise<ParquetRowGroup>;
     /**
-     * Returns the user (key/value) metadata for this file
+     * Each row group contains column chunks for all the columns.
      */
-    getMetadata(): Record<string, string>;
+    readColumnChunk(schema: ParquetSchema, colChunk: ColumnChunk): Promise<ParquetColumnChunk>;
     /**
-     * Implement AsyncIterable
+     * Getting dictionary for allows to flatten values by indices.
+     * @param dictionaryPageOffset
+     * @param options
+     * @param pagesOffset
+     * @returns
      */
-    [Symbol.asyncIterator](): AsyncIterator<T>;
+    getDictionary(dictionaryPageOffset: number, options: ParquetOptions, pagesOffset: number): Promise<string[]>;
 }
 //# sourceMappingURL=parquet-reader.d.ts.map

package/dist/parquetjs/parser/parquet-reader.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"parquet-reader.d.ts","sourceRoot":"","sources":["../../../src/parquetjs/parser/parquet-reader.ts"],"names":[],"mappings":";AACA,OAAO,EAAC,qBAAqB,EAAC,MAAM,2BAA2B,CAAC;AAChE,OAAO,EAAC,YAAY,EAAC,MAAM,mBAAmB,CAAC;AAC/C,OAAO,EAAC,aAAa,EAAC,MAAM,kBAAkB,CAAC;AAC/C,OAAO,EAAC,aAAa,EAAC,MAAM,kBAAkB,CAAC;AAI/C;;;;;;GAMG;AACH,qBAAa,aAAa,CAAC,CAAC,CAAE,YAAW,aAAa,CAAC,CAAC,CAAC;IACvD;;OAEG;WACU,QAAQ,CAAC,CAAC,EAAE,IAAI,EAAE,IAAI,GAAG,OAAO,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC;IAkB/D;;OAEG;WACU,eAAe,CAAC,CAAC,EAAE,WAAW,EAAE,WAAW,GAAG,OAAO,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC;WAevE,UAAU,CAAC,CAAC,EAAE,MAAM,EAAE,MAAM,GAAG,OAAO,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC;IAY9D,QAAQ,EAAE,YAAY,CAAC;IACvB,cAAc,EAAE,qBAAqB,CAAC;IACtC,MAAM,EAAE,aAAa,CAAC;IAE7B;;;;;OAKG;gBACS,QAAQ,EAAE,YAAY,EAAE,cAAc,EAAE,qBAAqB;IAYzE;;;OAGG;IACG,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC;IAM5B;;;;;;;;OAQG;IACH,SAAS,IAAI,aAAa,CAAC,CAAC,CAAC;IAE7B,SAAS,CAAC,CAAC,SAAS,MAAM,CAAC,EAAE,UAAU,EAAE,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,GAAG,aAAa,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;IAChF,SAAS,CAAC,UAAU,EAAE,CAAC,MAAM,GAAG,MAAM,EAAE,CAAC,EAAE,GAAG,aAAa,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;IAkBvE;;;OAGG;IACH,WAAW,IAAI,MAAM;IAIrB;;OAEG;IACH,SAAS,IAAI,aAAa;IAI1B;;OAEG;IACH,WAAW,IAAI,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC;IAQrC;;OAEG;IAEH,CAAC,MAAM,CAAC,aAAa,CAAC,IAAI,aAAa,CAAC,CAAC,CAAC;CAG3C"}
+{"version":3,"file":"parquet-reader.d.ts","sourceRoot":"","sources":["../../../src/parquetjs/parser/parquet-reader.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAC,YAAY,EAAC,MAAM,0BAA0B,CAAC;AAE3D,OAAO,EAAC,aAAa,EAAC,MAAM,kBAAkB,CAAC;AAK/C,OAAO,EAAC,WAAW,EAAoB,YAAY,EAAE,QAAQ,EAAO,MAAM,mBAAmB,CAAC;AAC9F,OAAO,EACL,eAAe,EAEf,kBAAkB,EAElB,cAAc,EACf,MAAM,mBAAmB,CAAC;AAI3B,MAAM,MAAM,kBAAkB,GAAG;IAC/B,qBAAqB,CAAC,EAAE,MAAM,CAAC;CAChC,CAAC;AAEF,0DAA0D;AAC1D,MAAM,MAAM,qBAAqB,GAAG;IAClC,iDAAiD;IACjD,UAAU,CAAC,EAAE,MAAM,EAAE,GAAG,MAAM,EAAE,EAAE,CAAC;CACpC,CAAC;AAMF;;;;;GAKG;AACH,qBAAa,aAAa;IACxB,KAAK,EAAE,QAAQ,CAAC,kBAAkB,CAAC,CAAC;IACpC,IAAI,EAAE,YAAY,CAAC;IACnB,QAAQ,EAAE,OAAO,CAAC,YAAY,CAAC,GAAG,IAAI,CAAQ;gBAElC,IAAI,EAAE,YAAY,EAAE,KAAK,CAAC,EAAE,kBAAkB;IAK1D,KAAK,IAAI,IAAI;IAOb,8BAA8B;IACvB,WAAW,CAAC,KAAK,CAAC,EAAE,qBAAqB;IAShD,wCAAwC;IACjC,gBAAgB,CAAC,KAAK,CAAC,EAAE,qBAAqB;IAOrD,sCAAsC;IAC/B,gBAAgB,CAAC,KAAK,CAAC,EAAE,qBAAqB;IAqB/C,WAAW,IAAI,OAAO,CAAC,MAAM,CAAC;IAK9B,SAAS,IAAI,OAAO,CAAC,aAAa,CAAC;IAQzC;;;OAGG;IACG,iBAAiB,IAAI,OAAO,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IASpD,eAAe,IAAI,OAAO,CAAC,YAAY,CAAC;IAU9C,uCAAuC;IACjC,UAAU,IAAI,OAAO,CAAC,IAAI,CAAC;IAajC,uCAAuC;IACjC,UAAU,IAAI,OAAO,CAAC,YAAY,CAAC;IAsBzC,4EAA4E;IACtE,YAAY,CAChB,MAAM,EAAE,aAAa,EACrB,QAAQ,EAAE,QAAQ,EAClB,UAAU,EAAE,MAAM,EAAE,EAAE,GACrB,OAAO,CAAC,eAAe,CAAC;IAgB3B;;OAEG;IACG,eAAe,CAAC,MAAM,EAAE,aAAa,EAAE,QAAQ,EAAE,WAAW,GAAG,OAAO,CAAC,kBAAkB,CAAC;IAoDhG;;;;;;OAMG;IACG,aAAa,CACjB,oBAAoB,EAAE,MAAM,EAC5B,OAAO,EAAE,cAAc,EACvB,WAAW,EAAE,MAAM,GAClB,OAAO,CAAC,MAAM,EAAE,CAAC;CAuBrB"}

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

@@ -1,120 +1,200 @@
-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;
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.ParquetReader = void 0;
+const schema_1 = require("../schema/schema");
+const decoders_1 = require("./decoders");
+const shred_1 = require("../schema/shred");
+const constants_1 = require("../../constants");
+const parquet_thrift_1 = require("../parquet-thrift");
+const read_utils_1 = require("../utils/read-utils");
+const decoders_2 = require("./decoders");
+const DEFAULT_PROPS = {
+    defaultDictionarySize: 1e6
+};
+/**
+ * The parquet envelope reader allows direct, unbuffered access to the individual
+ * sections of the parquet file, namely the header, footer and the row groups.
+ * This class is intended for advanced/internal users; if you just want to retrieve
+ * rows from a parquet file use the ParquetReader instead
+ */
+class ParquetReader {
+    constructor(file, props) {
+        this.metadata = null;
+        this.file = file;
+        this.props = { ...DEFAULT_PROPS, ...props };
     }
-  }
-
-  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;
+    close() {
+        // eslint-disable-next-line @typescript-eslint/no-floating-promises
+        this.file.close();
     }
-  }
-
-  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;
+    // HIGH LEVEL METHODS
+    /** Yield one row at a time */
+    async *rowIterator(props) {
+        for await (const rows of this.rowBatchIterator(props)) {
+            // yield *rows
+            for (const row of rows) {
+                yield row;
+            }
+        }
     }
-  }
-
-  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');
+    /** Yield one batch of rows at a time */
+    async *rowBatchIterator(props) {
+        const schema = await this.getSchema();
+        for await (const rowGroup of this.rowGroupIterator(props)) {
+            yield (0, shred_1.materializeRows)(schema, rowGroup);
+        }
     }
-
-    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 = [];
+    /** Iterate over the raw row groups */
+    async *rowGroupIterator(props) {
+        // Ensure strings are nested in arrays
+        const columnList = (props?.columnList || []).map((x) => Array.isArray(x) ? x : [x]);
+        const metadata = await this.getFileMetadata();
+        const schema = await this.getSchema();
+        const rowGroupCount = metadata?.row_groups.length || 0;
+        for (let rowGroupIndex = 0; rowGroupIndex < rowGroupCount; rowGroupIndex++) {
+            const rowGroup = await this.readRowGroup(schema, metadata.row_groups[rowGroupIndex], columnList);
+            yield rowGroup;
+        }
     }
-
-    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;
+    async getRowCount() {
+        const metadata = await this.getFileMetadata();
+        return Number(metadata.num_rows);
+    }
+    async getSchema() {
+        const metadata = await this.getFileMetadata();
+        const root = metadata.schema[0];
+        const { schema: schemaDefinition } = (0, decoders_1.decodeSchema)(metadata.schema, 1, root.num_children);
+        const schema = new schema_1.ParquetSchema(schemaDefinition);
+        return schema;
+    }
+    /**
+     * Returns the user (key/value) metadata for this file
+     * In parquet this is not stored on the schema like it is in arrow
+     */
+    async getSchemaMetadata() {
+        const metadata = await this.getFileMetadata();
+        const md = {};
+        for (const kv of metadata.key_value_metadata) {
+            md[kv.key] = kv.value;
+        }
+        return md;
+    }
+    async getFileMetadata() {
+        if (!this.metadata) {
+            await this.readHeader();
+            this.metadata = this.readFooter();
+        }
+        return this.metadata;
+    }
+    // LOW LEVEL METHODS
+    /** Metadata is stored in the footer */
+    async readHeader() {
+        const buffer = await this.file.read(0, constants_1.PARQUET_MAGIC.length);
+        const magic = buffer.toString();
+        switch (magic) {
+            case constants_1.PARQUET_MAGIC:
+                break;
+            case constants_1.PARQUET_MAGIC_ENCRYPTED:
+                throw new Error('Encrypted parquet file not supported');
+            default:
+                throw new Error(`Invalid parquet file (magic=${magic})`);
+        }
+    }
+    /** Metadata is stored in the footer */
+    async readFooter() {
+        const trailerLen = constants_1.PARQUET_MAGIC.length + 4;
+        const trailerBuf = await this.file.read(this.file.size - trailerLen, trailerLen);
+        const magic = trailerBuf.slice(4).toString();
+        if (magic !== constants_1.PARQUET_MAGIC) {
+            throw new Error(`Not a valid parquet file (magic="${magic})`);
+        }
+        const metadataSize = trailerBuf.readUInt32LE(0);
+        const metadataOffset = this.file.size - metadataSize - trailerLen;
+        if (metadataOffset < constants_1.PARQUET_MAGIC.length) {
+            throw new Error(`Invalid metadata size ${metadataOffset}`);
+        }
+        const metadataBuf = await this.file.read(metadataOffset, metadataSize);
+        // let metadata = new parquet_thrift.FileMetaData();
+        // parquet_util.decodeThrift(metadata, metadataBuf);
+        const { metadata } = (0, read_utils_1.decodeFileMetadata)(metadataBuf);
+        return metadata;
+    }
+    /** Data is stored in row groups (similar to Apache Arrow record batches) */
+    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?.path_in_schema;
+            if (columnList.length > 0 && (0, read_utils_1.fieldIndexOf)(columnList, colKey) < 0) {
+                continue; // eslint-disable-line no-continue
+            }
+            buffer.columnData[colKey.join()] = await this.readColumnChunk(schema, colChunk);
+        }
+        return buffer;
+    }
+    /**
+     * Each row group contains column chunks for all the columns.
+     */
+    async readColumnChunk(schema, colChunk) {
+        if (colChunk.file_path !== undefined && colChunk.file_path !== null) {
+            throw new Error('external references are not supported');
+        }
+        const field = schema.findField(colChunk.meta_data?.path_in_schema);
+        const type = (0, read_utils_1.getThriftEnum)(parquet_thrift_1.Type, colChunk.meta_data?.type);
+        if (type !== field.primitiveType) {
+            throw new Error(`chunk type not matching schema: ${type}`);
+        }
+        const compression = (0, read_utils_1.getThriftEnum)(parquet_thrift_1.CompressionCodec, colChunk.meta_data?.codec);
+        const pagesOffset = Number(colChunk.meta_data?.data_page_offset);
+        let pagesSize = Number(colChunk.meta_data?.total_compressed_size);
+        if (!colChunk.file_path) {
+            pagesSize = Math.min(this.file.size - pagesOffset, Number(colChunk.meta_data?.total_compressed_size));
+        }
+        const options = {
+            type,
+            rLevelMax: field.rLevelMax,
+            dLevelMax: field.dLevelMax,
+            compression,
+            column: field,
+            numValues: colChunk.meta_data?.num_values,
+            dictionary: []
+        };
+        let dictionary;
+        const dictionaryPageOffset = colChunk?.meta_data?.dictionary_page_offset;
+        if (dictionaryPageOffset) {
+            const dictionaryOffset = Number(dictionaryPageOffset);
+            // Getting dictionary from column chunk to iterate all over indexes to get dataPage values.
+            dictionary = await this.getDictionary(dictionaryOffset, options, pagesOffset);
+        }
+        dictionary = options.dictionary?.length ? options.dictionary : dictionary;
+        const pagesBuf = await this.file.read(pagesOffset, pagesSize);
+        return await (0, decoders_2.decodeDataPages)(pagesBuf, { ...options, dictionary });
+    }
+    /**
+     * Getting dictionary for allows to flatten values by indices.
+     * @param dictionaryPageOffset
+     * @param options
+     * @param pagesOffset
+     * @returns
+     */
+    async getDictionary(dictionaryPageOffset, options, pagesOffset) {
+        if (dictionaryPageOffset === 0) {
+            // dictionarySize = Math.min(this.fileSize - pagesOffset, this.defaultDictionarySize);
+            // pagesBuf = await this.read(pagesOffset, dictionarySize);
+            // In this case we are working with parquet-mr files format. Problem is described below:
+            // https://stackoverflow.com/questions/55225108/why-is-dictionary-page-offset-0-for-plain-dictionary-encoding
+            // We need to get dictionary page from column chunk if it exists.
+            // Now if we use code commented above we don't get DICTIONARY_PAGE we get DATA_PAGE instead.
+            return [];
+        }
+        const dictionarySize = Math.min(this.file.size - dictionaryPageOffset, this.props.defaultDictionarySize);
+        const pagesBuf = await this.file.read(dictionaryPageOffset, dictionarySize);
+        const cursor = { buffer: pagesBuf, offset: 0, size: pagesBuf.length };
+        const decodedPage = await (0, decoders_2.decodePage)(cursor, options);
+        return decodedPage.dictionary;
     }
-
-    return md;
-  }
-
-  [_Symbol$asyncIterator]() {
-    return this.getCursor()[Symbol.asyncIterator]();
-  }
-
 }
-//# sourceMappingURL=parquet-reader.js.map
+exports.ParquetReader = ParquetReader;

package/dist/parquetjs/schema/declare.d.ts

@@ -1,18 +1,18 @@
 import Int64 from 'node-int64';
 import type { PageHeader } from '../parquet-thrift';
-export declare type ParquetCodec = 'PLAIN' | 'RLE' | 'PLAIN_DICTIONARY';
-export declare type ParquetCompression = 'UNCOMPRESSED' | 'GZIP' | 'SNAPPY' | 'LZO' | 'BROTLI' | 'LZ4' | 'LZ4_RAW' | 'ZSTD';
-export declare type RepetitionType = 'REQUIRED' | 'OPTIONAL' | 'REPEATED';
-export declare type ParquetType = PrimitiveType | OriginalType;
+export type ParquetCodec = 'PLAIN' | 'RLE' | 'PLAIN_DICTIONARY';
+export type ParquetCompression = 'UNCOMPRESSED' | 'GZIP' | 'SNAPPY' | 'LZO' | 'BROTLI' | 'LZ4' | 'LZ4_RAW' | 'ZSTD';
+export type RepetitionType = 'REQUIRED' | 'OPTIONAL' | 'REPEATED';
+export type ParquetType = PrimitiveType | OriginalType;
 /**
  * Physical type
  */
-export declare type PrimitiveType = 'BOOLEAN' | 'INT32' | 'INT64' | 'INT96' | 'FLOAT' | 'DOUBLE' | 'BYTE_ARRAY' | 'FIXED_LEN_BYTE_ARRAY';
+export type PrimitiveType = 'BOOLEAN' | 'INT32' | 'INT64' | 'INT96' | 'FLOAT' | 'DOUBLE' | 'BYTE_ARRAY' | 'FIXED_LEN_BYTE_ARRAY';
 /**
  * Logical type
  */
-export declare type OriginalType = 'UTF8' | 'DECIMAL_INT32' | 'DECIMAL_INT64' | 'DECIMAL_BYTE_ARRAY' | 'DECIMAL_FIXED_LEN_BYTE_ARRAY' | 'DATE' | 'TIME_MILLIS' | 'TIME_MICROS' | 'TIMESTAMP_MILLIS' | 'TIMESTAMP_MICROS' | 'UINT_8' | 'UINT_16' | 'UINT_32' | 'UINT_64' | 'INT_8' | 'INT_16' | 'INT_32' | 'INT_64' | 'JSON' | 'BSON' | 'INTERVAL';
-export declare type ParquetDictionary = string[];
+export type OriginalType = 'UTF8' | 'DECIMAL_INT32' | 'DECIMAL_INT64' | 'DECIMAL_BYTE_ARRAY' | 'DECIMAL_FIXED_LEN_BYTE_ARRAY' | 'DATE' | 'TIME_MILLIS' | 'TIME_MICROS' | 'TIMESTAMP_MILLIS' | 'TIMESTAMP_MICROS' | 'UINT_8' | 'UINT_16' | 'UINT_32' | 'UINT_64' | 'INT_8' | 'INT_16' | 'INT_32' | 'INT_64' | 'JSON' | 'BSON' | 'INTERVAL';
+export type ParquetDictionary = string[];
 export interface SchemaDefinition {
     [string: string]: FieldDefinition;
 }
@@ -45,6 +45,7 @@ export interface ParquetField {
     fieldCount?: number;
     fields?: Record<string, ParquetField>;
 }
+/** @todo better name, this is an internal type? */
 export interface ParquetOptions {
     type: ParquetType;
     rLevelMax: number;
@@ -54,27 +55,34 @@ export interface ParquetOptions {
     numValues?: Int64;
     dictionary?: ParquetDictionary;
 }
-export interface ParquetData {
-    dlevels: number[];
-    rlevels: number[];
-    values: any[];
-    count: number;
-    pageHeaders: PageHeader[];
-}
 export interface ParquetPageData {
     dlevels: number[];
     rlevels: number[];
+    /** Actual column chunks */
     values: any[];
     count: number;
     dictionary?: ParquetDictionary;
+    /** The "raw" page header from the file */
     pageHeader: PageHeader;
 }
-export interface ParquetRecord {
+export interface ParquetRow {
     [key: string]: any;
 }
-export declare class ParquetBuffer {
+/** @
+ * Holds data for one row group (column chunks) */
+export declare class ParquetRowGroup {
+    /** Number of rows in this page */
     rowCount: number;
-    columnData: Record<string, ParquetData>;
-    constructor(rowCount?: number, columnData?: Record<string, ParquetData>);
+    /** Map of Column chunks */
+    columnData: Record<string, ParquetColumnChunk>;
+    constructor(rowCount?: number, columnData?: Record<string, ParquetColumnChunk>);
+}
+/** Holds the data for one column chunk */
+export interface ParquetColumnChunk {
+    dlevels: number[];
+    rlevels: number[];
+    values: any[];
+    count: number;
+    pageHeaders: PageHeader[];
 }
 //# sourceMappingURL=declare.d.ts.map

package/dist/parquetjs/schema/declare.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"declare.d.ts","sourceRoot":"","sources":["../../../src/parquetjs/schema/declare.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,MAAM,YAAY,CAAC;AAC/B,OAAO,KAAK,EAAC,UAAU,EAAC,MAAM,mBAAmB,CAAC;AAElD,oBAAY,YAAY,GAAG,OAAO,GAAG,KAAK,GAAG,kBAAkB,CAAC;AAChE,oBAAY,kBAAkB,GAC1B,cAAc,GACd,MAAM,GACN,QAAQ,GACR,KAAK,GACL,QAAQ,GACR,KAAK,GACL,SAAS,GACT,MAAM,CAAC;AACX,oBAAY,cAAc,GAAG,UAAU,GAAG,UAAU,GAAG,UAAU,CAAC;AAClE,oBAAY,WAAW,GAAG,aAAa,GAAG,YAAY,CAAC;AAEvD;;GAEG;AACH,oBAAY,aAAa,GAErB,SAAS,GACT,OAAO,GACP,OAAO,GACP,OAAO,GACP,OAAO,GACP,QAAQ,GACR,YAAY,GACZ,sBAAsB,CAAC;AAE3B;;GAEG;AACH,oBAAY,YAAY,GAEpB,MAAM,GAMN,eAAe,GACf,eAAe,GACf,oBAAoB,GACpB,8BAA8B,GAC9B,MAAM,GACN,aAAa,GACb,aAAa,GACb,kBAAkB,GAClB,kBAAkB,GAClB,QAAQ,GACR,SAAS,GACT,SAAS,GACT,SAAS,GACT,OAAO,GACP,QAAQ,GACR,QAAQ,GACR,QAAQ,GACR,MAAM,GACN,MAAM,GACN,UAAU,CAAC;AAEf,oBAAY,iBAAiB,GAAG,MAAM,EAAE,CAAC;AAEzC,MAAM,WAAW,gBAAgB;IAC/B,CAAC,MAAM,EAAE,MAAM,GAAG,eAAe,CAAC;CACnC;AAED,MAAM,WAAW,eAAe;IAC9B,IAAI,CAAC,EAAE,WAAW,CAAC;IACnB,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,QAAQ,CAAC,EAAE,YAAY,CAAC;IACxB,WAAW,CAAC,EAAE,kBAAkB,CAAC;IACjC,QAAQ,CAAC,EAAE,OAAO,CAAC;IACnB,QAAQ,CAAC,EAAE,OAAO,CAAC;IACnB,MAAM,CAAC,EAAE,gBAAgB,CAAC;CAC3B;AAED,MAAM,WAAW,YAAY;IAC3B,IAAI,EAAE,MAAM,CAAC;IACb,IAAI,EAAE,MAAM,EAAE,CAAC;IACf,GAAG,EAAE,MAAM,CAAC;IACZ,aAAa,CAAC,EAAE,aAAa,CAAC;IAC9B,YAAY,CAAC,EAAE,YAAY,CAAC;IAC5B,cAAc,EAAE,cAAc,CAAC;IAC/B,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,QAAQ,CAAC,EAAE,YAAY,CAAC;IACxB,WAAW,CAAC,EAAE,kBAAkB,CAAC;IACjC,SAAS,EAAE,MAAM,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,QAAQ,CAAC,EAAE,OAAO,CAAC;IACnB,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,MAAM,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,YAAY,CAAC,CAAC;CACvC;AAED,MAAM,WAAW,cAAc;IAC7B,IAAI,EAAE,WAAW,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,WAAW,EAAE,kBAAkB,CAAC;IAChC,MAAM,EAAE,YAAY,CAAC;IACrB,SAAS,CAAC,EAAE,KAAK,CAAC;IAClB,UAAU,CAAC,EAAE,iBAAiB,CAAC;CAChC;AAED,MAAM,WAAW,WAAW;IAC1B,OAAO,EAAE,MAAM,EAAE,CAAC;IAClB,OAAO,EAAE,MAAM,EAAE,CAAC;IAClB,MAAM,EAAE,GAAG,EAAE,CAAC;IACd,KAAK,EAAE,MAAM,CAAC;IACd,WAAW,EAAE,UAAU,EAAE,CAAC;CAC3B;AAED,MAAM,WAAW,eAAe;IAC9B,OAAO,EAAE,MAAM,EAAE,CAAC;IAClB,OAAO,EAAE,MAAM,EAAE,CAAC;IAClB,MAAM,EAAE,GAAG,EAAE,CAAC;IACd,KAAK,EAAE,MAAM,CAAC;IACd,UAAU,CAAC,EAAE,iBAAiB,CAAC;IAC/B,UAAU,EAAE,UAAU,CAAC;CACxB;AAED,MAAM,WAAW,aAAa;IAC5B,CAAC,GAAG,EAAE,MAAM,GAAG,GAAG,CAAC;CACpB;AAED,qBAAa,aAAa;IACxB,QAAQ,EAAE,MAAM,CAAC;IACjB,UAAU,EAAE,MAAM,CAAC,MAAM,EAAE,WAAW,CAAC,CAAC;gBAC5B,QAAQ,GAAE,MAAU,EAAE,UAAU,GAAE,MAAM,CAAC,MAAM,EAAE,WAAW,CAAM;CAI/E"}
+{"version":3,"file":"declare.d.ts","sourceRoot":"","sources":["../../../src/parquetjs/schema/declare.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,MAAM,YAAY,CAAC;AAC/B,OAAO,KAAK,EAAC,UAAU,EAAC,MAAM,mBAAmB,CAAC;AAElD,MAAM,MAAM,YAAY,GAAG,OAAO,GAAG,KAAK,GAAG,kBAAkB,CAAC;AAChE,MAAM,MAAM,kBAAkB,GAC1B,cAAc,GACd,MAAM,GACN,QAAQ,GACR,KAAK,GACL,QAAQ,GACR,KAAK,GACL,SAAS,GACT,MAAM,CAAC;AACX,MAAM,MAAM,cAAc,GAAG,UAAU,GAAG,UAAU,GAAG,UAAU,CAAC;AAClE,MAAM,MAAM,WAAW,GAAG,aAAa,GAAG,YAAY,CAAC;AAEvD;;GAEG;AACH,MAAM,MAAM,aAAa,GAErB,SAAS,GACT,OAAO,GACP,OAAO,GACP,OAAO,GACP,OAAO,GACP,QAAQ,GACR,YAAY,GACZ,sBAAsB,CAAC;AAE3B;;GAEG;AACH,MAAM,MAAM,YAAY,GAEpB,MAAM,GAMN,eAAe,GACf,eAAe,GACf,oBAAoB,GACpB,8BAA8B,GAC9B,MAAM,GACN,aAAa,GACb,aAAa,GACb,kBAAkB,GAClB,kBAAkB,GAClB,QAAQ,GACR,SAAS,GACT,SAAS,GACT,SAAS,GACT,OAAO,GACP,QAAQ,GACR,QAAQ,GACR,QAAQ,GACR,MAAM,GACN,MAAM,GACN,UAAU,CAAC;AAEf,MAAM,MAAM,iBAAiB,GAAG,MAAM,EAAE,CAAC;AAEzC,MAAM,WAAW,gBAAgB;IAC/B,CAAC,MAAM,EAAE,MAAM,GAAG,eAAe,CAAC;CACnC;AAED,MAAM,WAAW,eAAe;IAC9B,IAAI,CAAC,EAAE,WAAW,CAAC;IACnB,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,QAAQ,CAAC,EAAE,YAAY,CAAC;IACxB,WAAW,CAAC,EAAE,kBAAkB,CAAC;IACjC,QAAQ,CAAC,EAAE,OAAO,CAAC;IACnB,QAAQ,CAAC,EAAE,OAAO,CAAC;IACnB,MAAM,CAAC,EAAE,gBAAgB,CAAC;CAC3B;AAED,MAAM,WAAW,YAAY;IAC3B,IAAI,EAAE,MAAM,CAAC;IACb,IAAI,EAAE,MAAM,EAAE,CAAC;IACf,GAAG,EAAE,MAAM,CAAC;IACZ,aAAa,CAAC,EAAE,aAAa,CAAC;IAC9B,YAAY,CAAC,EAAE,YAAY,CAAC;IAC5B,cAAc,EAAE,cAAc,CAAC;IAC/B,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,QAAQ,CAAC,EAAE,YAAY,CAAC;IACxB,WAAW,CAAC,EAAE,kBAAkB,CAAC;IACjC,SAAS,EAAE,MAAM,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,QAAQ,CAAC,EAAE,OAAO,CAAC;IACnB,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,MAAM,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,YAAY,CAAC,CAAC;CACvC;AAED,mDAAmD;AACnD,MAAM,WAAW,cAAc;IAC7B,IAAI,EAAE,WAAW,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,SAAS,EAAE,MAAM,CAAC;IAClB,WAAW,EAAE,kBAAkB,CAAC;IAChC,MAAM,EAAE,YAAY,CAAC;IACrB,SAAS,CAAC,EAAE,KAAK,CAAC;IAClB,UAAU,CAAC,EAAE,iBAAiB,CAAC;CAChC;AAED,MAAM,WAAW,eAAe;IAC9B,OAAO,EAAE,MAAM,EAAE,CAAC;IAClB,OAAO,EAAE,MAAM,EAAE,CAAC;IAClB,2BAA2B;IAC3B,MAAM,EAAE,GAAG,EAAE,CAAC;IACd,KAAK,EAAE,MAAM,CAAC;IACd,UAAU,CAAC,EAAE,iBAAiB,CAAC;IAC/B,0CAA0C;IAC1C,UAAU,EAAE,UAAU,CAAC;CACxB;AAED,MAAM,WAAW,UAAU;IACzB,CAAC,GAAG,EAAE,MAAM,GAAG,GAAG,CAAC;CACpB;AAED;kDACkD;AAClD,qBAAa,eAAe;IAC1B,kCAAkC;IAClC,QAAQ,EAAE,MAAM,CAAC;IACjB,2BAA2B;IAC3B,UAAU,EAAE,MAAM,CAAC,MAAM,EAAE,kBAAkB,CAAC,CAAC;gBAEnC,QAAQ,GAAE,MAAU,EAAE,UAAU,GAAE,MAAM,CAAC,MAAM,EAAE,kBAAkB,CAAM;CAItF;AAED,0CAA0C;AAC1C,MAAM,WAAW,kBAAkB;IACjC,OAAO,EAAE,MAAM,EAAE,CAAC;IAClB,OAAO,EAAE,MAAM,EAAE,CAAC;IAClB,MAAM,EAAE,GAAG,EAAE,CAAC;IACd,KAAK,EAAE,MAAM,CAAC;IACd,WAAW,EAAE,UAAU,EAAE,CAAC;CAC3B"}

package/dist/parquetjs/schema/declare.js

@@ -1,13 +1,12 @@
-import _defineProperty from "@babel/runtime/helpers/esm/defineProperty";
-export class ParquetBuffer {
-  constructor(rowCount = 0, columnData = {}) {
-    _defineProperty(this, "rowCount", void 0);
-
-    _defineProperty(this, "columnData", void 0);
-
-    this.rowCount = rowCount;
-    this.columnData = columnData;
-  }
-
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.ParquetRowGroup = void 0;
+/** @
+ * Holds data for one row group (column chunks) */
+class ParquetRowGroup {
+    constructor(rowCount = 0, columnData = {}) {
+        this.rowCount = rowCount;
+        this.columnData = columnData;
+    }
 }
-//# sourceMappingURL=declare.js.map
+exports.ParquetRowGroup = ParquetRowGroup;

package/dist/parquetjs/schema/schema.d.ts

@@ -1,4 +1,4 @@
-import { FieldDefinition, ParquetBuffer, ParquetCompression, ParquetField, ParquetRecord, SchemaDefinition } from './declare';
+import { FieldDefinition, ParquetRowGroup, ParquetCompression, ParquetField, ParquetRow, SchemaDefinition } from './declare';
 /**
  * A parquet file schema
  */
@@ -18,9 +18,9 @@ export declare class ParquetSchema {
      * Retrieve a field definition and all the field's ancestors
      */
     findFieldBranch(path: string | string[]): ParquetField[];
-    shredRecord(record: ParquetRecord, buffer: ParquetBuffer): void;
-    materializeRecords(buffer: ParquetBuffer): ParquetRecord[];
+    shredRecord(row: ParquetRow, rowGroup: ParquetRowGroup): void;
+    materializeRows(rowGroup: ParquetRowGroup): ParquetRow[];
     compress(type: ParquetCompression): this;
-    buffer(): ParquetBuffer;
+    rowGroup(): ParquetRowGroup;
 }
 //# sourceMappingURL=schema.d.ts.map

package/dist/parquetjs/schema/schema.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"schema.d.ts","sourceRoot":"","sources":["../../../src/parquetjs/schema/schema.ts"],"names":[],"mappings":"AAIA,OAAO,EACL,eAAe,EACf,aAAa,EACb,kBAAkB,EAClB,YAAY,EACZ,aAAa,EAEb,gBAAgB,EACjB,MAAM,WAAW,CAAC;AAInB;;GAEG;AACH,qBAAa,aAAa;IACjB,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,eAAe,CAAC,CAAC;IACxC,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,YAAY,CAAC,CAAC;IACrC,SAAS,EAAE,YAAY,EAAE,CAAC;IAEjC;;OAEG;gBACS,MAAM,EAAE,gBAAgB;IAMpC;;OAEG;IACH,SAAS,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,GAAG,YAAY;IAiBhD;;OAEG;IACH,eAAe,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,GAAG,YAAY,EAAE;IAgBxD,WAAW,CAAC,MAAM,EAAE,aAAa,EAAE,MAAM,EAAE,aAAa,GAAG,IAAI;IAI/D,kBAAkB,CAAC,MAAM,EAAE,aAAa,GAAG,aAAa,EAAE;IAI1D,QAAQ,CAAC,IAAI,EAAE,kBAAkB,GAAG,IAAI;IAMxC,MAAM,IAAI,aAAa;CAGxB"}
+{"version":3,"file":"schema.d.ts","sourceRoot":"","sources":["../../../src/parquetjs/schema/schema.ts"],"names":[],"mappings":"AAIA,OAAO,EACL,eAAe,EACf,eAAe,EACf,kBAAkB,EAClB,YAAY,EACZ,UAAU,EAEV,gBAAgB,EACjB,MAAM,WAAW,CAAC;AAInB;;GAEG;AACH,qBAAa,aAAa;IACjB,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,eAAe,CAAC,CAAC;IACxC,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,YAAY,CAAC,CAAC;IACrC,SAAS,EAAE,YAAY,EAAE,CAAC;IAEjC;;OAEG;gBACS,MAAM,EAAE,gBAAgB;IAMpC;;OAEG;IACH,SAAS,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,GAAG,YAAY;IAiBhD;;OAEG;IACH,eAAe,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,GAAG,YAAY,EAAE;IAgBxD,WAAW,CAAC,GAAG,EAAE,UAAU,EAAE,QAAQ,EAAE,eAAe,GAAG,IAAI;IAI7D,eAAe,CAAC,QAAQ,EAAE,eAAe,GAAG,UAAU,EAAE;IAIxD,QAAQ,CAAC,IAAI,EAAE,kBAAkB,GAAG,IAAI;IAMxC,QAAQ,IAAI,eAAe;CAG5B"}