Regex-tools NPM

Regular Expression Tools

I use regular expressions to do a lot of things, and managing a long regular expression could be painful. So I write this simple tool to manage long and complex regular expressions such as regular expressions of lex parser.

Install

npm install regex-tools --save-dev

Create Options File

Create a test-regex.js file and write something like this.

exports.options = {
  name: 'test',
  operation: 'combine',
  target: 'target.js', // support *.ts file, too
  flags: 'g',
  regexes: [/</, [/($term:\w+)/, /\d*/], />/],
};

Source File

Create a target.js and mark related code with /* /$test/ */ ($ followed by the name you configured in options file).

let testRegex = /* /$test/ */ /./;

let groups = testRegex.exec('<abc123>');

/* /$test/ */
let text = groups[0];

'<def456>'.replace(testRegex, function(/* /$test/ */ text) {
  return text;
});

A Basic Task

Create a task and run it.

let Gulp = require('gulp');
let RegexTools = require('regex-tools');

let glob = require('glob');

Gulp.task('update-regex', function() {
  let optionsFiles = glob.sync('*-regex.js');

  optionsFiles.forEach(function(path) {
    RegexTools.process(path);
  });
});

After that, target.js should look like:

let testRegex = /* /$test/ */ /<(\w+)\d*>/g;

let groups = testRegex.exec('<abc123>');

/* /$test/ */
let text = groups[0];
let term = groups[1];

'<def456>'.replace(testRegex, function(/* /$test/ */ text, term) {
  return text;
});

Problem solved! You may checkout demo for relatively more complex examples.

API References

An options file is a node module that exports options as default. The exported default value could be either RegexToolsOptions or RegexToolsOptions[].

And here's related type declarations:

interface RegexToolsOptions {
  /** name that will match related tag in target source file. */
  name: string;
  /** target source file. */
  target: string;
  /** only "combine" so far. */
  operation?: string;
  flags?: string;
  regexes: NestedRegexes;
}

type Lookahead = boolean | '=' | '!';

interface NestedRegexOptions {
  /** captured group name. */
  name?: string;
  /** whether to use `|`, default to false. */
  or?: boolean;
  /** whether to capture, default to false if `name` is not provided, otherwise true. */
  capture?: boolean;
  /** lookahead, `true` or `"="` for positive and `"!"` for negative. */
  lookahead?: Lookahead;
  /** ?, *, +, *?, +?, {1}, {1,}, {1,2} */
  repeat?: string;
  regexes: RegExp | NestedRegexArray | NestedRegexOptions;
}

interface NestedRegexArray
  extends Array<RegExp | NestedRegexArray | NestedRegexOptions> {}

type NestedRegexes = NestedRegexArray | NestedRegexOptions;

Back Reference Tracking

If you are using back reference, it will keep the index updated. That means you should write back references relative to the current part of regular expression.

The options below will result in /(distraction)(["'])\2/.

exports.options = {
  name: 'test',
  operation: 'combine',
  target: 'target.js',
  regexes: [/(distraction)/, /(["'])\1/],
};

However as \[number] can also be a char with code in octal form, it's kind of complex to deal with. Please avoid writing a char (instead of a back reference) in this form.

Named References

Another way to deal with back references is to use named references provided by the tool.

exports.options = {
  name: 'test',
  operation: 'combine',
  target: 'target.js',
  regexes: [/($quote:["'])/, /.*?/, /($quote)/],
};

When generating group/parameter/enumerator list, the name of named reference (as well as name specified in NestedRegexOptions) will be exported according to its capture index.

For groups and enumerator list, if you don't want some of them to show up, you may add a ~ between $ and name. Such as ($~quote:["']).