CSV Parser - Final state machine - Javascript

csv-parser.js

const states = {
    startField: 'startField',
    startRow: 'startRow',
    escaped: 'escaped',
    nonQuoted: 'nonQuoted',
    quoted: 'quoted',
};
const EOF = 'eof';

module.exports.csvParser = (content) => {
    const data = [];
    let state = states.startRow;
    let fieldBuffer = [];

    const startRow = () => {
        data.push([]);
    };
    const endField = () => {
        data[data.length - 1].push(fieldBuffer.join(''));
        fieldBuffer = [];
    };
    const machineError = () => new Error('Invalid end of state');

    const machine = {
        [states.startRow](char) {
            if (char === EOF) {
                return;
            } else if (/\n/.test(char)) {
                return states.startRow;
            }

            startRow();

            if (/,/.test(char)) {
                endField();

                return states.startField;
            } else if (/"/.test(char)) {
                return states.quoted;
            } else if (char) {
                fieldBuffer.push(char);

                return states.nonQuoted;
            }
        },
        [states.startField](char) {
            if (char === EOF) {
                // no op
            } else if (/,/.test(char)) {
                endField();

                return states.startField;
            } else if (/"/.test(char)) {
                return states.quoted;
            } else if (/[\n\r]/.test(char)) {
                endField();

                return states.startRow;
            } else if (char) {
                fieldBuffer.push(char);

                return states.nonQuoted;
            } else {
                throw machineError();
            }
        },
        [states.escaped](char) {
            if (char === EOF) {
                endField();
            } else if (/"/.test(char)) {
                fieldBuffer.push(char);

                return states.quoted;
            } else if (/,/.test(char)) {
                endField();

                return states.startField;
            } else if (/[\n\r]/.test(char)) {
                endField();

                return states.startRow;
            } else {
                throw machineError();
            }
        },
        [states.quoted](char) {
            if (/"/.test(char)) {
                return states.escaped;
            } else if (char !== EOF) {
                fieldBuffer.push(char);

                return states.quoted;
            } else {
                throw machineError();
            }
        },
        [states.nonQuoted](char) {
            if (char === EOF) {
                endField();
            } else if (/,/.test(char)) {
                endField();

                return states.startField;
            } else if (/[\n\r]/.test(char)) {
                endField();

                return states.startRow;
            } else if (char) {
                fieldBuffer.push(char);

                return states.nonQuoted;
            } else {
                throw machineError();
            }
        },
    };

    for (const char of content) {
        state = machine[state](char);
    }

    machine[state](EOF);

    return data;
};

csv-parser.spec.js

// JEST

const {csvParser} = require('./index');

describe('parser', () => {
    it('should handle no content', () => {
        expect(csvParser('')).toEqual([]);
    });

    it('should handle one word', () => {
        expect(csvParser('hello')).toEqual([['hello']]);
        expect(csvParser('"hello"')).toEqual([['hello']]);
    });

    it('should handle two words', () => {
        expect(csvParser('hello,world')).toEqual([['hello', 'world']]);
        expect(csvParser('hello,"world"')).toEqual([['hello', 'world']]);
    });

    it('should handle escaped words', () => {
        expect(csvParser('hello,world""')).toEqual([['hello', 'world""']]);
        expect(csvParser('hello,"""world"""')).toEqual([['hello', '"world"']]);
    });

    it('should handle empty words', () => {
        expect(csvParser('hello,,world')).toEqual([['hello', '', 'world']]);
        expect(csvParser('"hello",,world')).toEqual([['hello', '', 'world']]);
        expect(csvParser(',world')).toEqual([['', 'world']]);
        expect(csvParser(',\n')).toEqual([['', '']]);
        expect(csvParser(',\r')).toEqual([['', '']]);
        expect(csvParser(',\r\n')).toEqual([['', '']]);
    });

    it('should handle lf', () => {
        expect(csvParser('hello\n"hello"')).toEqual([['hello'], ['hello']]);
        expect(csvParser('hello,world\nhello,world')).toEqual([['hello', 'world'], ['hello', 'world']]);
        expect(csvParser('hello,,world\n"hello",,world')).toEqual([['hello', '', 'world'], ['hello', '', 'world']]);
    });

    it('should handle cr', () => {
        expect(csvParser('hello\r"hello"')).toEqual([['hello'], ['hello']]);
        expect(csvParser('hello,world\rhello,world')).toEqual([['hello', 'world'], ['hello', 'world']]);
        expect(csvParser('hello,,world\r"hello",,world')).toEqual([['hello', '', 'world'], ['hello', '', 'world']]);
    });

    it('should handle crlf', () => {
        expect(csvParser('hello\r\n"hello"')).toEqual([['hello'], ['hello']]);
        expect(csvParser('hello,world\r\nhello,world')).toEqual([['hello', 'world'], ['hello', 'world']]);
        expect(csvParser('hello,,world\r\n"hello",,world')).toEqual([['hello', '', 'world'], ['hello', '', 'world']]);
    });

    it('should handle all', () => {
        const sampleRows = (lineBreak) => [
            'hello,world,"this,is,good"',
            'this,has,numbers,1234',
            'this,has,special,characters,!@#$%^&*()\'[]{}./\\|-=_+<>?',
            'hello,world,"this,is,escaped""quotation"""',
            'this,,has,,spaces',
            'this has spaces',
            `and,"trailing,spaces",with,eol,${lineBreak}`,
        ].join(lineBreak);

        const lfRows = sampleRows('\n');
        const crRows = sampleRows('\r');
        const crlfRows = sampleRows('\r\n');

        const expected = [
            ['hello', 'world', 'this,is,good'],
            ['this', 'has', 'numbers', '1234'],
            ['this', 'has', 'special', 'characters', '!@#$%^&*()\'[]{}./\\|-=_+<>?'],
            ['hello', 'world', 'this,is,escaped"quotation"'],
            ['this', '', 'has', '', 'spaces'],
            ['this has spaces'],
            ['and', 'trailing,spaces', 'with', 'eol', ''],
        ];

        expect(csvParser(lfRows)).toEqual(expected);
        expect(csvParser(crRows)).toEqual(expected);
        expect(csvParser(crlfRows)).toEqual(expected);
    });
});