UNPKG

2.77 kBJavaScriptView Raw
1const maxDistance = 3;
2
3function editDistance(a, b) {
4 // https://en.wikipedia.org/wiki/Damerau–Levenshtein_distance
5 // Calculating optimal string alignment distance, no substring is edited more than once.
6 // (Simple implementation.)
7
8 // Quick early exit, return worst case.
9 if (Math.abs(a.length - b.length) > maxDistance) return Math.max(a.length, b.length);
10
11 // distance between prefix substrings of a and b
12 const d = [];
13
14 // pure deletions turn a into empty string
15 for (let i = 0; i <= a.length; i++) {
16 d[i] = [i];
17 }
18 // pure insertions turn empty string into b
19 for (let j = 0; j <= b.length; j++) {
20 d[0][j] = j;
21 }
22
23 // fill matrix
24 for (let j = 1; j <= b.length; j++) {
25 for (let i = 1; i <= a.length; i++) {
26 let cost = 1;
27 if (a[i - 1] === b[j - 1]) {
28 cost = 0;
29 } else {
30 cost = 1;
31 }
32 d[i][j] = Math.min(
33 d[i - 1][j] + 1, // deletion
34 d[i][j - 1] + 1, // insertion
35 d[i - 1][j - 1] + cost // substitution
36 );
37 // transposition
38 if (i > 1 && j > 1 && a[i - 1] === b[j - 2] && a[i - 2] === b[j - 1]) {
39 d[i][j] = Math.min(d[i][j], d[i - 2][j - 2] + 1);
40 }
41 }
42 }
43
44 return d[a.length][b.length];
45}
46
47/**
48 * Find close matches, restricted to same number of edits.
49 *
50 * @param {string} word
51 * @param {string[]} candidates
52 * @returns {string}
53 */
54
55function suggestSimilar(word, candidates) {
56 if (!candidates || candidates.length === 0) return '';
57 // remove possible duplicates
58 candidates = Array.from(new Set(candidates));
59
60 const searchingOptions = word.startsWith('--');
61 if (searchingOptions) {
62 word = word.slice(2);
63 candidates = candidates.map(candidate => candidate.slice(2));
64 }
65
66 let similar = [];
67 let bestDistance = maxDistance;
68 const minSimilarity = 0.4;
69 candidates.forEach((candidate) => {
70 if (candidate.length <= 1) return; // no one character guesses
71
72 const distance = editDistance(word, candidate);
73 const length = Math.max(word.length, candidate.length);
74 const similarity = (length - distance) / length;
75 if (similarity > minSimilarity) {
76 if (distance < bestDistance) {
77 // better edit distance, throw away previous worse matches
78 bestDistance = distance;
79 similar = [candidate];
80 } else if (distance === bestDistance) {
81 similar.push(candidate);
82 }
83 }
84 });
85
86 similar.sort((a, b) => a.localeCompare(b));
87 if (searchingOptions) {
88 similar = similar.map(candidate => `--${candidate}`);
89 }
90
91 if (similar.length > 1) {
92 return `\n(Did you mean one of ${similar.join(', ')}?)`;
93 }
94 if (similar.length === 1) {
95 return `\n(Did you mean ${similar[0]}?)`;
96 }
97 return '';
98}
99
100exports.suggestSimilar = suggestSimilar;