csv#

import { parseCsv, parseCsvRecords, stringifyCsv } from "velt/encoding/csv". RFC 4180 CSV. Quoted fields may contain delimiters, "" escapes and line breaks. Records end at LF, CRLF or a lone CR. A final line break doesn't add a record, and a byte order mark is skipped.

import { parseCsv, parseCsvRecords, stringifyCsv } from "velt/encoding/csv";

const text = "name,city\nAda,London\n\"Hopper, Grace\",\"New\nYork\"\n";
console.log(JSON.stringify(parseCsv(text)));
// [["name","city"],["Ada","London"],["Hopper, Grace","New\nYork"]]
for (const r of parseCsvRecords(text)) {
  console.log(r.get("name"), "|", r.get("city"));
}
console.log(JSON.stringify(stringifyCsv([["id", "note"], ["1", "say \"hi\""]], { crlf: true })));
// "id,note\r\n1,\"say \"\"hi\"\"\"\r\n"

Reading#

parseCsv(text, opts: CsvOptions = {}): string[][]: every record as an array of fields. CsvOptions { delimiter?; trim?; skipEmptyLines? }: delimiter is one ASCII character (default ,; ";" or "\t"), trim strips spaces and tabs around every field, quoted or not, and skipEmptyLines drops blank lines.
parseCsvRecords(text, opts): Map<string, string>[]: the first row is the header, and each later row a map from column name to field. Missing fields become "" and extra fields are ignored.

The parser reads the whole text; for a large file, read it with readFile first, or split it into lines yourself when the fields hold no line breaks.

Writing#

stringifyCsv(rows, opts: CsvWriteOptions = {}) writes string[][] rows, quoting a field only when it needs it, and ends every record with a line break. CsvWriteOptions { delimiter?; crlf? }: crlf: true ends records with CRLF, as RFC 4180 and spreadsheets expect.

Errors#

CsvError { message; line } is thrown for an unterminated quote, text after a closing quote, or a bad delimiter (line is 0 for the delimiter case).

import { parseCsv } from "velt/encoding/csv";

try {
  parseCsv("a\n\"open");
} catch (e) {
  console.log(e.line, e.message); // 2 Unterminated quoted field starting on line 2
}