275 lines
9.3 KiB
JavaScript
275 lines
9.3 KiB
JavaScript
|
/*
|
||
|
|
|
||
|
|
didYouMean.js - A simple JavaScript matching engine
|
||
|
|
===================================================
|
||
|
|
|
||
|
|
[Available on GitHub](https://github.com/dcporter/didyoumean.js).
|
||
|
|
|
||
|
|
A super-simple, highly optimized JS library for matching human-quality input to a list of potential
|
||
|
|
matches. You can use it to suggest a misspelled command-line utility option to a user, or to offer
|
||
|
|
links to nearby valid URLs on your 404 page. (The examples below are taken from a personal project,
|
||
|
|
my [HTML5 business card](http://dcporter.aws.af.cm/me), which uses didYouMean.js to suggest correct
|
||
|
|
URLs from misspelled ones, such as [dcporter.aws.af.cm/me/instagarm](http://dcporter.aws.af.cm/me/instagarm).)
|
||
|
|
Uses the [Levenshtein distance algorithm](https://en.wikipedia.org/wiki/Levenshtein_distance).
|
||
|
|
|
||
|
|
didYouMean.js works in the browser as well as in node.js. To install it for use in node:
|
||
|
|
|
||
|
|
```
|
||
|
|
npm install didyoumean
|
||
|
|
```
|
||
|
|
|
||
|
|
|
||
|
|
Examples
|
||
|
|
--------
|
||
|
|
|
||
|
|
Matching against a list of strings:
|
||
|
|
```
|
||
|
|
var input = 'insargrm'
|
||
|
|
var list = ['facebook', 'twitter', 'instagram', 'linkedin'];
|
||
|
|
console.log(didYouMean(input, list));
|
||
|
|
> 'instagram'
|
||
|
|
// The method matches 'insargrm' to 'instagram'.
|
||
|
|
|
||
|
|
input = 'google plus';
|
||
|
|
console.log(didYouMean(input, list));
|
||
|
|
> null
|
||
|
|
// The method was unable to find 'google plus' in the list of options.
|
||
|
|
```
|
||
|
|
|
||
|
|
Matching against a list of objects:
|
||
|
|
```
|
||
|
|
var input = 'insargrm';
|
||
|
|
var list = [ { id: 'facebook' }, { id: 'twitter' }, { id: 'instagram' }, { id: 'linkedin' } ];
|
||
|
|
var key = 'id';
|
||
|
|
console.log(didYouMean(input, list, key));
|
||
|
|
> 'instagram'
|
||
|
|
// The method returns the matching value.
|
||
|
|
|
||
|
|
didYouMean.returnWinningObject = true;
|
||
|
|
console.log(didYouMean(input, list, key));
|
||
|
|
> { id: 'instagram' }
|
||
|
|
// The method returns the matching object.
|
||
|
|
```
|
||
|
|
|
||
|
|
|
||
|
|
didYouMean(str, list, [key])
|
||
|
|
----------------------------
|
||
|
|
|
||
|
|
- str: The string input to match.
|
||
|
|
- list: An array of strings or objects to match against.
|
||
|
|
- key (OPTIONAL): If your list array contains objects, you must specify the key which contains the string
|
||
|
|
to match against.
|
||
|
|
|
||
|
|
Returns: the closest matching string, or null if no strings exceed the threshold.
|
||
|
|
|
||
|
|
|
||
|
|
Options
|
||
|
|
-------
|
||
|
|
|
||
|
|
Options are set on the didYouMean function object. You may change them at any time.
|
||
|
|
|
||
|
|
### threshold
|
||
|
|
|
||
|
|
By default, the method will only return strings whose edit distance is less than 40% (0.4x) of their length.
|
||
|
|
For example, if a ten-letter string is five edits away from its nearest match, the method will return null.
|
||
|
|
|
||
|
|
You can control this by setting the "threshold" value on the didYouMean function. For example, to set the
|
||
|
|
edit distance threshold to 50% of the input string's length:
|
||
|
|
|
||
|
|
```
|
||
|
|
didYouMean.threshold = 0.5;
|
||
|
|
```
|
||
|
|
|
||
|
|
To return the nearest match no matter the threshold, set this value to null.
|
||
|
|
|
||
|
|
### thresholdAbsolute
|
||
|
|
|
||
|
|
This option behaves the same as threshold, but instead takes an integer number of edit steps. For example,
|
||
|
|
if thresholdAbsolute is set to 20 (the default), then the method will only return strings whose edit distance
|
||
|
|
is less than 20. Both options apply.
|
||
|
|
|
||
|
|
### caseSensitive
|
||
|
|
|
||
|
|
By default, the method will perform case-insensitive comparisons. If you wish to force case sensitivity, set
|
||
|
|
the "caseSensitive" value to true:
|
||
|
|
|
||
|
|
```
|
||
|
|
didYouMean.caseSensitive = true;
|
||
|
|
```
|
||
|
|
|
||
|
|
### nullResultValue
|
||
|
|
|
||
|
|
By default, the method will return null if there is no sufficiently close match. You can change this value here.
|
||
|
|
|
||
|
|
### returnWinningObject
|
||
|
|
|
||
|
|
By default, the method will return the winning string value (if any). If your list contains objects rather
|
||
|
|
than strings, you may set returnWinningObject to true.
|
||
|
|
|
||
|
|
```
|
||
|
|
didYouMean.returnWinningObject = true;
|
||
|
|
```
|
||
|
|
|
||
|
|
This option has no effect on lists of strings.
|
||
|
|
|
||
|
|
### returnFirstMatch
|
||
|
|
|
||
|
|
By default, the method will search all values and return the closest match. If you're simply looking for a "good-
|
||
|
|
enough" match, you can set your thresholds appropriately and set returnFirstMatch to true to substantially speed
|
||
|
|
things up.
|
||
|
|
|
||
|
|
|
||
|
|
License
|
||
|
|
-------
|
||
|
|
|
||
|
|
didYouMean copyright (c) 2013-2014 Dave Porter.
|
||
|
|
|
||
|
|
Licensed under the Apache License, Version 2.0 (the "License");
|
||
|
|
you may not use this file except in compliance with the License.
|
||
|
|
You may obtain a copy of the License
|
||
|
|
[here](http://www.apache.org/licenses/LICENSE-2.0).
|
||
|
|
|
||
|
|
Unless required by applicable law or agreed to in writing, software
|
||
|
|
distributed under the License is distributed on an "AS IS" BASIS,
|
||
|
|
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||
|
|
See the License for the specific language governing permissions and
|
||
|
|
limitations under the License.
|
||
|
|
|
||
|
|
*/
|
||
|
|
(function() {
|
||
|
|
"use strict";
|
||
|
|
|
||
|
|
// The didYouMean method.
|
||
|
|
function didYouMean(str, list, key) {
|
||
|
|
if (!str) return null;
|
||
|
|
|
||
|
|
// If we're running a case-insensitive search, smallify str.
|
||
|
|
if (!didYouMean.caseSensitive) { str = str.toLowerCase(); }
|
||
|
|
|
||
|
|
// Calculate the initial value (the threshold) if present.
|
||
|
|
var thresholdRelative = didYouMean.threshold === null ? null : didYouMean.threshold * str.length,
|
||
|
|
thresholdAbsolute = didYouMean.thresholdAbsolute,
|
||
|
|
winningVal;
|
||
|
|
if (thresholdRelative !== null && thresholdAbsolute !== null) winningVal = Math.min(thresholdRelative, thresholdAbsolute);
|
||
|
|
else if (thresholdRelative !== null) winningVal = thresholdRelative;
|
||
|
|
else if (thresholdAbsolute !== null) winningVal = thresholdAbsolute;
|
||
|
|
else winningVal = null;
|
||
|
|
|
||
|
|
// Get the edit distance to each option. If the closest one is less than 40% (by default) of str's length,
|
||
|
|
// then return it.
|
||
|
|
var winner, candidate, testCandidate, val,
|
||
|
|
i, len = list.length;
|
||
|
|
for (i = 0; i < len; i++) {
|
||
|
|
// Get item.
|
||
|
|
candidate = list[i];
|
||
|
|
// If there's a key, get the candidate value out of the object.
|
||
|
|
if (key) { candidate = candidate[key]; }
|
||
|
|
// Gatekeep.
|
||
|
|
if (!candidate) { continue; }
|
||
|
|
// If we're running a case-insensitive search, smallify the candidate.
|
||
|
|
if (!didYouMean.caseSensitive) { testCandidate = candidate.toLowerCase(); }
|
||
|
|
else { testCandidate = candidate; }
|
||
|
|
// Get and compare edit distance.
|
||
|
|
val = getEditDistance(str, testCandidate, winningVal);
|
||
|
|
// If this value is smaller than our current winning value, OR if we have no winning val yet (i.e. the
|
||
|
|
// threshold option is set to null, meaning the caller wants a match back no matter how bad it is), then
|
||
|
|
// this is our new winner.
|
||
|
|
if (winningVal === null || val < winningVal) {
|
||
|
|
winningVal = val;
|
||
|
|
// Set the winner to either the value or its object, depending on the returnWinningObject option.
|
||
|
|
if (key && didYouMean.returnWinningObject) winner = list[i];
|
||
|
|
else winner = candidate;
|
||
|
|
// If we're returning the first match, return it now.
|
||
|
|
if (didYouMean.returnFirstMatch) return winner;
|
||
|
|
}
|
||
|
|
}
|
||
|
|
|
||
|
|
// If we have a winner, return it.
|
||
|
|
return winner || didYouMean.nullResultValue;
|
||
|
|
}
|
||
|
|
|
||
|
|
// Set default options.
|
||
|
|
didYouMean.threshold = 0.4;
|
||
|
|
didYouMean.thresholdAbsolute = 20;
|
||
|
|
didYouMean.caseSensitive = false;
|
||
|
|
didYouMean.nullResultValue = null;
|
||
|
|
didYouMean.returnWinningObject = null;
|
||
|
|
didYouMean.returnFirstMatch = false;
|
||
|
|
|
||
|
|
// Expose.
|
||
|
|
// In node...
|
||
|
|
if (typeof module !== 'undefined' && module.exports) {
|
||
|
|
module.exports = didYouMean;
|
||
|
|
}
|
||
|
|
// Otherwise...
|
||
|
|
else {
|
||
|
|
window.didYouMean = didYouMean;
|
||
|
|
}
|
||
|
|
|
||
|
|
var MAX_INT = Math.pow(2,32) - 1; // We could probably go higher than this, but for practical reasons let's not.
|
||
|
|
function getEditDistance(a, b, max) {
|
||
|
|
// Handle null or undefined max.
|
||
|
|
max = max || max === 0 ? max : MAX_INT;
|
||
|
|
|
||
|
|
var lena = a.length;
|
||
|
|
var lenb = b.length;
|
||
|
|
|
||
|
|
// Fast path - no A or B.
|
||
|
|
if (lena === 0) return Math.min(max + 1, lenb);
|
||
|
|
if (lenb === 0) return Math.min(max + 1, lena);
|
||
|
|
|
||
|
|
// Fast path - length diff larger than max.
|
||
|
|
if (Math.abs(lena - lenb) > max) return max + 1;
|
||
|
|
|
||
|
|
// Slow path.
|
||
|
|
var matrix = [],
|
||
|
|
i, j, colMin, minJ, maxJ;
|
||
|
|
|
||
|
|
// Set up the first row ([0, 1, 2, 3, etc]).
|
||
|
|
for (i = 0; i <= lenb; i++) { matrix[i] = [i]; }
|
||
|
|
|
||
|
|
// Set up the first column (same).
|
||
|
|
for (j = 0; j <= lena; j++) { matrix[0][j] = j; }
|
||
|
|
|
||
|
|
// Loop over the rest of the columns.
|
||
|
|
for (i = 1; i <= lenb; i++) {
|
||
|
|
colMin = MAX_INT;
|
||
|
|
minJ = 1;
|
||
|
|
if (i > max) minJ = i - max;
|
||
|
|
maxJ = lenb + 1;
|
||
|
|
if (maxJ > max + i) maxJ = max + i;
|
||
|
|
// Loop over the rest of the rows.
|
||
|
|
for (j = 1; j <= lena; j++) {
|
||
|
|
// If j is out of bounds, just put a large value in the slot.
|
||
|
|
if (j < minJ || j > maxJ) {
|
||
|
|
matrix[i][j] = max + 1;
|
||
|
|
}
|
||
|
|
|
||
|
|
// Otherwise do the normal Levenshtein thing.
|
||
|
|
else {
|
||
|
|
// If the characters are the same, there's no change in edit distance.
|
||
|
|
if (b.charAt(i - 1) === a.charAt(j - 1)) {
|
||
|
|
matrix[i][j] = matrix[i - 1][j - 1];
|
||
|
|
}
|
||
|
|
// Otherwise, see if we're substituting, inserting or deleting.
|
||
|
|
else {
|
||
|
|
matrix[i][j] = Math.min(matrix[i - 1][j - 1] + 1, // Substitute
|
||
|
|
Math.min(matrix[i][j - 1] + 1, // Insert
|
||
|
|
matrix[i - 1][j] + 1)); // Delete
|
||
|
|
}
|
||
|
|
}
|
||
|
|
|
||
|
|
// Either way, update colMin.
|
||
|
|
if (matrix[i][j] < colMin) colMin = matrix[i][j];
|
||
|
|
}
|
||
|
|
|
||
|
|
// If this column's minimum is greater than the allowed maximum, there's no point
|
||
|
|
// in going on with life.
|
||
|
|
if (colMin > max) return max + 1;
|
||
|
|
}
|
||
|
|
// If we made it this far without running into the max, then return the final matrix value.
|
||
|
|
return matrix[lenb][lena];
|
||
|
|
}
|
||
|
|
|
||
|
|
})();
|