stupid-coffee-lexer

coffee-lex

Stupid lexer for CoffeeScript.

Install

$ npm install [--save] coffee-lex

Usage

The main lex function simply returns a list of tokens:

import lex, { SourceType } from 'coffee-lex';

let source = 'a?(b: c)';
let tokens = lex(source);

// Print tokens along with their source.
tokens.forEach(token =>
  console.log(
    SourceType[token.type],
    JSON.stringify(source.slice(token.start, token.end)),
    `${token.start}→${token.end}`
  )
);
// IDENTIFIER "a" 0→1
// EXISTENCE "?" 1→2
// CALL_START "(" 2→3
// IDENTIFIER "b" 3→4
// COLON ":" 4→5
// IDENTIFIER "c" 6→7
// CALL_END ")" 7→8

You can also get more fine control of what you'd like to lex by using the stream function:

import { stream, SourceType } from 'coffee-lex';

let source = 'a?(b: c)';
let step = stream(source);
let location;

do {
  location = step();
  console.log(location.index, SourceType[location.type]);
} while (location.type !== SourceType.EOF);
// 0 IDENTIFIER
// 1 EXISTENCE
// 2 CALL_START
// 3 IDENTIFIER
// 4 COLON
// 5 SPACE
// 6 IDENTIFIER
// 7 CALL_END
// 8 EOF

This function not only lets you control how far into the source you'd like to go, it also gives you information about source code that wouldn't become part of a token, such as spaces.

Note that the input source code should have only UNIX line endings (LF). If you want to process a file with Windows line endings (CRLF), you should convert to UNIX line endings first, then use coffee-lex, then convert back if necessary.

Why?

The official CoffeeScript lexer does a lot of pre-processing, even with rewrite: false. That makes it good for building an AST, but bad for identifying parts of the source code that aren't part of the final AST, such as the location of operators. One good example of this is string interpolation. The official lexer turns it into a series of string tokens separated by (virtual) + tokens, but they have no reality in the original source code. Here's what the official lexer generates for "a#{b}c":

[ [ 'STRING_START',
    '(',
    { first_line: 0, first_column: 0, last_line: 0, last_column: 0 },
    origin: [ 'STRING', null, [Object] ] ],
  [ 'STRING',
    '"a"',
    { first_line: 0, first_column: 0, last_line: 0, last_column: 1 } ],
  [ '+',
    '+',
    { first_line: 0, first_column: 3, last_line: 0, last_column: 3 } ],
  [ '(',
    '(',
    { first_line: 0, first_column: 3, last_line: 0, last_column: 3 } ],
  [ 'IDENTIFIER',
    'b',
    { first_line: 0, first_column: 4, last_line: 0, last_column: 4 },
    variable: true ],
  [ ')',
    ')',
    { first_line: 0, first_column: 5, last_line: 0, last_column: 5 },
    origin: [ '', 'end of interpolation', [Object] ] ],
  [ '+',
    '+',
    { first_line: 0, first_column: 6, last_line: 0, last_column: 6 } ],
  [ 'STRING',
    '"c"',
    { first_line: 0, first_column: 6, last_line: 0, last_column: 7 } ],
  [ 'STRING_END',
    ')',
    { first_line: 0, first_column: 7, last_line: 0, last_column: 7 } ],
  [ 'TERMINATOR',
    '\n',
    { first_line: 0, first_column: 8, last_line: 0, last_column: 8 } ] ]

Here's what coffee-lex generates for the same source:

[ SourceToken { type: DSTRING_START, start: 0, end: 1 },
  SourceToken { type: STRING_CONTENT, start: 1, end: 2 },
  SourceToken { type: INTERPOLATION_START, start: 2, end: 4 },
  SourceToken { type: IDENTIFIER, start: 4, end: 5 },
  SourceToken { type: INTERPOLATION_END, start: 5, end: 6 },
  SourceToken { type: STRING_CONTENT, start: 6, end: 7 },
  SourceToken { type: DSTRING_END, start: 7, end: 8 } ]

License

MIT