Regex expression not match line of Node.js code

Question

I am looking for a regex that can match any line of code that contains a single reference to a core module)

Something like this:

const coreModuleMatches = /'^[var|const]{0,1}[a-z\$\_]{1,}=require([\'|"][assert|fs|path][\'|"])[;|,]{0,1}$/;

This should match all of these lines

var pth = require("path"); 
const asrt = require('assert'),
     fs = require('fs'),
     cp = require('child_process');

The problem is I can't get the simple regex to work, so my more complex regex currently has no hope.

I am stripping out all whitespace except newline characters before matching the code with regular expressions, and then splitting by newline so that I can go line by line through the code. Any ideas welcome.

Is string from `var. . . . . ('child_process');` one match or two different matches ? — , Mar 20 '16 at 10:03
http://stackoverflow.com/questions/399078/what-special-characters-must-be-escaped-in-regular-expressions and http://stackoverflow.com/questions/9801630/what-is-the-difference-between-square-brackets-and-parentheses-in-a-regex — Wiktor Stribiżew, Mar 20 '16 at 11:03
var. . . . . ('child_process'); is one match, and the subexpression I really care about is simply 'child_process', I just want an array of core-modules that are referenced in the code. — Alexander Mills, Mar 20 '16 at 18:50

score 1 · Answer 1 · 2016-03-20T19:16:47.383

1

Apart from forgetting to escape ( and ) there were few mistakes in your regex too.

Your Regex:

/'^[var|const]{0,1}[a-z\$\_]{1,}=require([\'|"][assert|fs|path][\'|"])[;|,]{0,1}$/

My Regex:

Explanation:

(?:var|const)\s*([a-z$_]+\s*=\s* This matches from var or const upto the variable name followed by = including all whitespaces.
require$('|")(?:assert|fs|path|child_process)\2$,? This matches the require() and whatever the module is inside it. As the first quote is captured using ('|"), \2 implies that it's the one repeated while closing too, so that mismatching of quotes does not takes place.
[\n\r\t\s]*)*; This matches all the whitespaces in your second variable consisting of newlines, tabs, spaces, carriage returns.

Regex101 Demo

edited Mar 20 '16 at 19:16

answered Mar 20 '16 at 09:57

thanks, I actually don't need to match whitespace. If you could simplify your regex by removing whitespace matches and explain how yours works that would help! – Alexander Mills Mar 20 '16 at 18:35
@AlexMills: Which whitespaces would you like to be removed ? In your question the match from `const.....` till end have many whitespaces. Besides that you haven't answered to my comment on your question. – Mar 20 '16 at 18:51
yeah, don't worry about that, it still works with the optional whitespace...I don't know what the '\2\' is doing. Your answer is unlikely to help anyone unless you can explain the regex a bit, thanks – Alexander Mills Mar 20 '16 at 18:52
@AlexMills: I have updated the answer with explanation. Let me know if it's clear or not. – Mar 20 '16 at 19:08

Shanoor · Answer 2 · 2016-03-20T10:18:16.173

1

You don't need a regex so complicated, something like /("|')(assert|fs|http)\1/ should be enough:

// don't list all modules by hand
var builtinModules = require('builtin-modules');

console.time('end');
var input =
`var pth = require("path");
var _ = require('lodash');
const asrt = require('assert'),
fs = require('fs'),
cp = require('child_process');`.split('\n');

// \1 is a reference to the matched beginning (double)quote
// to prevent something like "path'/'fs" to match
var rgxStr = `("|')(${ builtinModules.join('|') })\\1`;
var rgx = new RegExp(rgxStr);
// console.log(rgxStr); // uncomment to see how the Regex looks like

var output = input.filter((line) => line.match(rgx));

console.timeEnd('end');
console.log('input');
console.log(input);
console.log('--------------------------------------');
console.log('output');
console.log(output);

Output:

end: 0.428ms
input
[ 'var pth = require("path");',
  'var _ = require(\'lodash\');',
  'const asrt = require(\'assert\'),',
  'fs = require(\'fs\'),',
  'cp = require(\'child_process\');' ]
--------------------------------------
output
[ 'var pth = require("path");',
  'const asrt = require(\'assert\'),',
  'fs = require(\'fs\'),',
  'cp = require(\'child_process\');' ]

edited Mar 20 '16 at 10:18

answered Mar 20 '16 at 10:02

Shanoor

13,344
2
29
40

You don't want to mismatch **quotes**. Use back-reference to captured groups. – Mar 20 '16 at 10:04
I get this output, given this input: – Alexander Mills Mar 20 '16 at 18:48
input: var assert = require('assert'); const fs = require('fs'); – Alexander Mills Mar 20 '16 at 18:48
output: core module match: 'assert' core module match: ' core module match: assert core module match: 'fs' core module match: ' core module match: fs – Alexander Mills Mar 20 '16 at 18:48
so this regex is matching this literal: ' ... and also, it's matching 'assert' or assert , etc. – Alexander Mills Mar 20 '16 at 18:49
I have to capture '/" to check if the closing quote is the same as the opening, you could just grab match[1] instead of logging the entire array. Or, if you're sure the input is valid JS, you can use this regex: `var rgxStr = \`(?:"|')(${ builtinModules.join('|') })(?:"|')\`;`. And no, it doesn't match `assert` alone, only if it's between quotes. – Shanoor Mar 21 '16 at 13:34

Regex expression not match line of Node.js code

2 Answers2