pstream is dead; long live pstream
taciturnaxolotl.github.io/pstream-ng/
add fuzzy search
+79
-3
+79
-3
src/backend/metadata/search.ts
+79
-3
src/backend/metadata/search.ts
···
1
+
import Fuse from "fuse.js";
2
+
1
3
import { SimpleCache } from "@/utils/cache";
2
4
import { MediaItem } from "@/utils/mediaTypes";
3
5
···
8
10
getMediaPoster,
9
11
multiSearch,
10
12
} from "./tmdb";
11
-
import { TMDBContentTypes } from "./types/tmdb";
13
+
import {
14
+
TMDBContentTypes,
15
+
TMDBMovieSearchResult,
16
+
TMDBShowSearchResult,
17
+
} from "./types/tmdb";
12
18
13
19
export interface MWQuery {
14
20
searchQuery: string;
···
22
28
23
29
// detect "tmdb:123456" or "tmdb:123456:movie" or "tmdb:123456:tv"
24
30
const tmdbIdPattern = /^tmdb:(\d+)(?::(movie|tv))?$/i;
31
+
const trailingYearPattern = /\s+\b(19|20)\d{2}\b$/;
32
+
33
+
function normalizeQuery(input: string): string {
34
+
return input
35
+
.toLowerCase()
36
+
.replace(/[^\p{L}\p{N}\s]/gu, " ")
37
+
.replace(/\s+/g, " ")
38
+
.trim();
39
+
}
40
+
41
+
function getLenientQueries(searchQuery: string): string[] {
42
+
const base = searchQuery.trim();
43
+
const normalized = normalizeQuery(base);
44
+
const withoutTrailingYear = base.replace(trailingYearPattern, "").trim();
45
+
const normalizedWithoutYear = normalizeQuery(withoutTrailingYear);
46
+
47
+
return [
48
+
...new Set([base, normalized, withoutTrailingYear, normalizedWithoutYear]),
49
+
].filter((q) => q.length > 0);
50
+
}
51
+
52
+
function dedupeTMDBResults(
53
+
items: (TMDBMovieSearchResult | TMDBShowSearchResult)[],
54
+
): (TMDBMovieSearchResult | TMDBShowSearchResult)[] {
55
+
const deduped = new Map<
56
+
string,
57
+
TMDBMovieSearchResult | TMDBShowSearchResult
58
+
>();
59
+
60
+
items.forEach((item) => {
61
+
deduped.set(`${item.media_type}:${item.id}`, item);
62
+
});
63
+
64
+
return Array.from(deduped.values());
65
+
}
66
+
67
+
function rankTMDBResultsFuzzy(
68
+
items: (TMDBMovieSearchResult | TMDBShowSearchResult)[],
69
+
query: string,
70
+
): (TMDBMovieSearchResult | TMDBShowSearchResult)[] {
71
+
if (items.length <= 1) return items;
72
+
73
+
const fuse = new Fuse(items, {
74
+
includeScore: true,
75
+
ignoreLocation: true,
76
+
threshold: 0.45,
77
+
minMatchCharLength: 2,
78
+
keys: [
79
+
{ name: "title", weight: 0.6 },
80
+
{ name: "name", weight: 0.6 },
81
+
{ name: "original_title", weight: 0.2 },
82
+
{ name: "original_name", weight: 0.2 },
83
+
],
84
+
});
85
+
86
+
const ranked = fuse.search(query).map((result) => result.item);
87
+
const rankedSet = new Set(
88
+
ranked.map((item) => `${item.media_type}:${item.id}`),
89
+
);
90
+
const remainder = items.filter(
91
+
(item) => !rankedSet.has(`${item.media_type}:${item.id}`),
92
+
);
93
+
94
+
return ranked.concat(remainder);
95
+
}
25
96
26
97
export async function searchForMedia(query: MWQuery): Promise<MediaItem[]> {
27
98
if (cache.has(query)) return cache.get(query) as MediaItem[];
···
69
140
}
70
141
}
71
142
72
-
const data = await multiSearch(searchQuery);
143
+
const queryVariants = getLenientQueries(searchQuery);
144
+
const resultSets = await Promise.all(
145
+
queryVariants.map((q) => multiSearch(q)),
146
+
);
147
+
const data = dedupeTMDBResults(resultSets.flat());
148
+
const rankedData = rankTMDBResultsFuzzy(data, searchQuery);
73
149
74
-
const results = data.map((v) => {
150
+
const results = rankedData.map((v) => {
75
151
const formattedResult = formatTMDBSearchResult(v, v.media_type);
76
152
return formatTMDBMetaToMediaItem(formattedResult);
77
153
});