@pujansrt/data-genie/dist/readers/tsv-reader.js

High performant ETL engine written in TypeScript

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.TSVReader = void 0;
const fs_1 = require("fs");
const csv_parse_1 = require("csv-parse");
/**
 * TSVReader class for reading data records from a Tab-Separated Values (TSV) file.
 * It uses the 'csv-parse' library, configured specifically for tab delimiters.
 */
class TSVReader {
    /**
     * Constructs a new TSVReader.
     * @param filePath The path to the TSV file.
     */
    constructor(filePath) {
        this.hasFieldNamesInFirstRow = false;
        this.filePath = filePath;
    }
    /**
     * Specifies whether the first row of the TSV file contains field names (headers).
     * If true, the yielded records will be objects with these field names as keys.
     * If false, records will be arrays or objects with numerical keys (e.g., {0: 'value1', 1: 'value2'}).
     * @param value True if the first row is a header, false otherwise.
     * @returns The current TSVReader instance for chaining.
     */
    setFieldNamesInFirstRow(value) {
        this.hasFieldNamesInFirstRow = value;
        return this;
    }
    /**
     * Reads data records from the TSV file asynchronously, yielding each record as it's parsed.
     * @returns An AsyncIterableIterator of DataRecord objects.
     */
    async *read() {
        const parser = (0, fs_1.createReadStream)(this.filePath).pipe((0, csv_parse_1.parse)({
            columns: this.hasFieldNamesInFirstRow,
            delimiter: '\t', // Fixed delimiter for TSV
            skip_empty_lines: true
        }));
        for await (const record of parser) {
            yield record;
        }
    }
}
exports.TSVReader = TSVReader;