1
0
mirror of https://github.com/JuanCanham/HackMyResume.git synced 2024-11-05 09:56:22 +00:00
HackMyResume/dist/inspectors/keyword-inspector.js

79 lines
2.7 KiB
JavaScript
Raw Normal View History

2016-01-27 10:29:26 +00:00
(function() {
2018-02-12 05:05:29 +00:00
/**
Keyword analysis for HackMyResume.
@license MIT. See LICENSE.md for details.
@module inspectors/keyword-inspector
*/
2016-01-27 10:29:26 +00:00
/**
Analyze the resume's use of keywords.
TODO: BUG: Keyword search regex is inaccurate, especially for one or two
letter keywords like "C" or "CLI".
@class keywordInspector
2018-02-12 05:05:29 +00:00
*/
var FluentDate, _, keywordInspector;
2016-01-27 10:29:26 +00:00
2018-02-12 05:05:29 +00:00
_ = require('underscore');
2016-01-27 10:29:26 +00:00
2018-02-12 05:05:29 +00:00
FluentDate = require('../core/fluent-date');
keywordInspector = module.exports = {
2016-01-27 10:29:26 +00:00
/** A unique name for this inspector. */
moniker: 'keyword-inspector',
/**
Run the Keyword Inspector on a resume.
@method run
@return An collection of statistical keyword data.
2018-02-12 05:05:29 +00:00
*/
2016-01-27 10:29:26 +00:00
run: function(rez) {
var prefix, regex_quote, searchable, suffix;
2018-02-12 05:05:29 +00:00
// "Quote" or safely escape a keyword so it can be used as a regex. For
// example, if the keyword is "C++", yield "C\+\+".
// http://stackoverflow.com/a/2593661/4942583
2016-01-27 10:29:26 +00:00
regex_quote = function(str) {
return (str + '').replace(/[.?*+^$[\]\\(){}|-]/ig, "\\$&");
};
2018-02-12 05:05:29 +00:00
// Create a searchable plain-text digest of the resume
// TODO: BUG: Don't search within keywords for other keywords. Job A
// declares the "foo" keyword. Job B declares the "foo & bar" keyword. Job
// B's mention of "foobar" should not count as a mention of "foo".
// To achieve this, remove keywords from the search digest and treat them
// separately.
2016-01-27 10:29:26 +00:00
searchable = '';
rez.transformStrings(['imp', 'computed', 'safe'], function(key, val) {
return searchable += ' ' + val;
});
2018-02-12 05:05:29 +00:00
// Assemble a regex skeleton we can use to test for keywords with a bit
// more
2016-01-27 10:29:26 +00:00
prefix = '(?:' + ['^', '\\s+', '[\\.,]+'].join('|') + ')';
suffix = '(?:' + ['$', '\\s+', '[\\.,]+'].join('|') + ')';
return rez.keywords().map(function(kw) {
var count, myArray, regex, regex_str;
2018-02-12 05:05:29 +00:00
// 1. Using word boundary or other regex class is inaccurate
// var regex = new RegExp( '\\b' + regex_quote( kw )/* + '\\b'*/, 'ig');
// 2. Searching for the raw keyword is inaccurate ("C" will match any
// word containing a 'c'!).
// var regex = new RegExp( regex_quote( kw ), 'ig');
// 3. Instead, use a custom regex with special delimeters.
2016-01-27 10:29:26 +00:00
regex_str = prefix + regex_quote(kw) + suffix;
regex = new RegExp(regex_str, 'ig');
myArray = null;
count = 0;
while ((myArray = regex.exec(searchable)) !== null) {
count++;
}
return {
name: kw,
count: count
};
});
}
};
}).call(this);
2016-02-02 02:14:36 +00:00
//# sourceMappingURL=keyword-inspector.js.map