ses/src/get-source-url.js

Hardened JavaScript for Fearless Cooperation

import { FERAL_REG_EXP, regexpExec, stringSlice } from './commons.js';

// Captures a key and value of the form #key=value or @key=value
const sourceMetaEntryRegExp =
  '\\s*[@#]\\s*([a-zA-Z][a-zA-Z0-9]*)\\s*=\\s*([^\\s\\*]*)';
// Captures either a one-line or multi-line comment containing
// one #key=value or @key=value.
// Produces two pairs of capture groups, but the initial two may be undefined.
// On account of the mechanics of regular expressions, scanning from the end
// does not allow us to capture every pair, so getSourceURL must capture and
// trim until there are no matching comments.
const sourceMetaEntriesRegExp = new FERAL_REG_EXP(
  `(?:\\s*//${sourceMetaEntryRegExp}|/\\*${sourceMetaEntryRegExp}\\s*\\*/)\\s*$`,
);

/**
 * @param {string} src
 */
export const getSourceURL = src => {
  let sourceURL = '<unknown>';

  // Our regular expression matches the last one or two comments with key value
  // pairs at the end of the source, avoiding a scan over the entire length of
  // the string, but at the expense of being able to capture all the (key,
  // value) pair meta comments at the end of the source, which may include
  // sourceMapURL in addition to sourceURL.
  // So, we sublimate the comments out of the source until no source or no
  // comments remain.
  while (src.length > 0) {
    const match = regexpExec(sourceMetaEntriesRegExp, src);
    if (match === null) {
      break;
    }
    src = stringSlice(src, 0, src.length - match[0].length);

    // We skip $0 since it contains the entire match.
    // The match contains four capture groups,
    // two (key, value) pairs, the first of which
    // may be undefined.
    // On the off-chance someone put two sourceURL comments in their code with
    // different commenting conventions, the latter has precedence.
    if (match[3] === 'sourceURL') {
      sourceURL = match[4];
    } else if (match[1] === 'sourceURL') {
      sourceURL = match[2];
    }
  }

  return sourceURL;
};