fuzzy file finder
haldean
1 year, 5 months ago
| 0 | import algorithm | |
| 1 | import options | |
| 2 | import os | |
| 3 | import re | |
| 4 | import sets | |
| 5 | import sequtils | |
| 6 | ||
| 7 | type | |
| 8 | Found = seq[string] | |
| 9 | Score = (float, float) | |
| 10 | ||
| 11 | let exclude = toHashSet([".git", "gen", "inst"]) | |
| 12 | ||
| 13 | proc includef(fname: string): bool = | |
| 14 | var x = fname | |
| 15 | while true: | |
| 16 | let (dir, base) = splitPath(x) | |
| 17 | if exclude.contains(base): | |
| 18 | return false | |
| 19 | if base[0] == '.': | |
| 20 | return false | |
| 21 | if dir == "/" or dir == "": | |
| 22 | return true | |
| 23 | x = dir | |
| 24 | ||
| 25 | proc find*(root: string): Found = | |
| 26 | var found = newSeq[string]() | |
| 27 | for file in walkDirRec(root): | |
| 28 | if includef(file): | |
| 29 | found.add(file) | |
| 30 | return found | |
| 31 | ||
| 32 | proc lcss(s1, s2: string): int = | |
| 33 | let | |
| 34 | m1 = s1.len - 1 | |
| 35 | m2 = s2.len - 1 | |
| 36 | var | |
| 37 | cache = newSeq[seq[int]]() | |
| 38 | maxlen = 0 | |
| 39 | ||
| 40 | for i in 0..m1: | |
| 41 | var s = newSeq[int]() | |
| 42 | for i in 0..m2: | |
| 43 | s.add(0) | |
| 44 | cache.add(s) | |
| 45 | ||
| 46 | for i1 in 0..m1: | |
| 47 | for i2 in 0..m2: | |
| 48 | if s1[i1] == s2[i2]: | |
| 49 | var x = 0 | |
| 50 | if i1 == 0 or i2 == 0: | |
| 51 | x = 1 | |
| 52 | else: | |
| 53 | x = cache[i1-1][i2-1] + 1 | |
| 54 | cache[i1][i2] = x | |
| 55 | if x > maxlen: | |
| 56 | maxlen = x | |
| 57 | return maxlen | |
| 58 | ||
| 59 | proc score(path, search: string): Score = | |
| 60 | let | |
| 61 | (_, base) = splitPath(path) | |
| 62 | basescore = lcss(search, base).float | |
| 63 | fullscore = lcss(search, path).float | |
| 64 | # In the event of a tie, break the tie by which has the longest match in the | |
| 65 | # basename of the file. Negated so that lower scores are better, so we can | |
| 66 | # ascending-sort on score and file name | |
| 67 | return (-fullscore, -basescore) | |
| 68 | ||
| 69 | proc find*(root, search: string): Found = | |
| 70 | var patstr = "" | |
| 71 | for ch in search: | |
| 72 | patstr = patstr & ".*" & ch | |
| 73 | patstr = patstr & ".*" | |
| 74 | let pat = re(patstr, {reIgnoreCase}) | |
| 75 | ||
| 76 | let found = find(root).filter do (abspath: string) -> bool: | |
| 77 | return match(relativePath(abspath, root), pat) | |
| 78 | ||
| 79 | var scored = found.map do (abspath: string) -> (Score, string): | |
| 80 | let relpath = relativePath(abspath, root) | |
| 81 | return (score(relpath, search), relpath) | |
| 82 | ||
| 83 | scored.sort() | |
| 84 | return scored.map do (p: (Score, string)) -> string: p[1] | |
| 85 | ||
| 86 | when isMainModule: | |
| 87 | var found: Found | |
| 88 | if paramCount() < 1: | |
| 89 | found = find(getCurrentDir()) | |
| 90 | else: | |
| 91 | let search = paramstr(1) | |
| 92 | found = find(getCurrentDir(), search) | |
| 93 | for f in found: | |
| 94 | echo f |