CSV With Quoted Commas: The 30-Line Parser

// CSV done right: a tiny state machine over the input string.
// Handles: quoted fields, commas inside quotes, embedded newlines, escaped
// double-quotes (RFC 4180 "" -> "), and CRLF or LF row endings.
// Does NOT handle: streaming over chunks (see accordion 3) or alternate delimiters.

function parseCsv(input) {
    const rows = [];
    let row = [];
    let field = '';
    let inQuotes = false;
    let i = 0;
    while (i < input.length) {
        const c = input[i];
        if (inQuotes) {
            if (c === '"') {
                if (input[i + 1] === '"') { field += '"'; i += 2; continue; }
                inQuotes = false; i++; continue;
            }
            field += c; i++; continue;
        }
        if (c === '"') { inQuotes = true; i++; continue; }
        if (c === ',') { row.push(field); field = ''; i++; continue; }
        if (c === '\n' || c === '\r') {
            row.push(field); field = '';
            rows.push(row); row = [];
            if (c === '\r' && input[i + 1] === '\n') i += 2; else i++;
            continue;
        }
        field += c; i++;
    }
    if (field.length > 0 || row.length > 0) { row.push(field); rows.push(row); }
    return rows;
}

const sample = `name,note,score
Alice,"Says, hello",95
Bob,"He said ""hi"" \nback",87
Carol,plain text,73\r\n`;

console.log(parseCsv(sample));

// Most consumers want { name: 'Alice', score: 95 } records, not [['Alice','95']].
// Wrap the parser with a header pass.

function parseCsv(input) {
    const rows = [];
    let row = [];
    let field = '';
    let inQuotes = false;
    let i = 0;
    while (i < input.length) {
        const c = input[i];
        if (inQuotes) {
            if (c === '"') {
                if (input[i + 1] === '"') { field += '"'; i += 2; continue; }
                inQuotes = false; i++; continue;
            }
            field += c; i++; continue;
        }
        if (c === '"') { inQuotes = true; i++; continue; }
        if (c === ',') { row.push(field); field = ''; i++; continue; }
        if (c === '\n' || c === '\r') {
            row.push(field); field = '';
            rows.push(row); row = [];
            if (c === '\r' && input[i + 1] === '\n') i += 2; else i++;
            continue;
        }
        field += c; i++;
    }
    if (field.length > 0 || row.length > 0) { row.push(field); rows.push(row); }
    return rows;
}

function parseCsvAsRecords(input, { coerce = {} } = {}) {
    const rows = parseCsv(input);
    if (rows.length === 0) return [];
    const [header, ...rest] = rows;
    return rest.map((cells) => {
        const obj = {};
        for (let i = 0; i < header.length; i++) {
            const key = header[i];
            const raw = cells[i] ?? '';
            obj[key] = coerce[key] ? coerce[key](raw) : raw;
        }
        return obj;
    });
}

const sample = `name,score,active
Alice,95,true
Bob,87,false`;

const records = parseCsvAsRecords(sample, {
    coerce: { score: Number, active: (s) => s === 'true' },
});
console.log(records);

// When the file is multi-GB you cannot read it as one string.
// Drive the same state machine from a chunk feeder; emit complete records.

class CsvStreamParser {
    constructor() {
        this.row = [];
        this.field = '';
        this.inQuotes = false;
        this.prevWasCR = false;  // CRLF spanning chunk boundaries
        this.headers = null;
        this.records = [];
    }
    feed(chunk) {
        // The parser state itself (row, field, inQuotes, prevWasCR) IS the buffer.
        // An unterminated tail lives in this.field until the next chunk arrives.
        for (let i = 0; i < chunk.length; i++) {
            const c = chunk[i];
            // If the previous char was CR, swallow a leading LF as the same line ending.
            if (this.prevWasCR) {
                this.prevWasCR = false;
                if (c === '\n') continue;
            }
            if (this.inQuotes) {
                if (c === '"') {
                    if (chunk[i + 1] === '"') { this.field += '"'; i += 1; continue; }
                    this.inQuotes = false; continue;
                }
                this.field += c; continue;
            }
            if (c === '"') { this.inQuotes = true; continue; }
            if (c === ',') { this.row.push(this.field); this.field = ''; continue; }
            if (c === '\r') {
                this.row.push(this.field); this.field = '';
                this._commit();
                this.prevWasCR = true;
                continue;
            }
            if (c === '\n') {
                this.row.push(this.field); this.field = '';
                this._commit();
                continue;
            }
            this.field += c;
        }
    }
    end() {
        if (this.field.length > 0 || this.row.length > 0) {
            this.row.push(this.field); this.field = '';
            this._commit();
        }
        const records = this.records; this.records = [];
        return records;
    }
    _commit() {
        if (this.headers === null) {
            this.headers = this.row;
        } else {
            const obj = {};
            for (let i = 0; i < this.headers.length; i++) obj[this.headers[i]] = this.row[i] ?? '';
            this.records.push(obj);
        }
        this.row = [];
    }
}

// Simulate a chunked feed; chunks split mid-field and mid-CRLF.
const parser = new CsvStreamParser();
parser.feed('name,note\nAlice,"sa');
parser.feed('ys, hello"\r');
parser.feed('\nBob,plain');
console.log(parser.end());