fix(match): Harden calculateWeightedSimilarity

1 files changed, 29 insertions, 3 deletions

diff --git a/src/lib/Media/Anime/Airing/Subtitled/match.ts b/src/lib/Media/Anime/Airing/Subtitled/match.ts
index fb494539..d6168d2c 100644
--- a/src/lib/Media/Anime/Airing/Subtitled/match.ts
+++ b/src/lib/Media/Anime/Airing/Subtitled/match.ts
@@ -42,9 +42,35 @@ const preprocessTitle = (title: string): string => {
     .join(' ');
 };
 
+const NON_DISTINCTIVE_TOKENS = new Set([
+  'a',
+  'and',
+  'de',
+  'e',
+  'for',
+  'ga',
+  'in',
+  'na',
+  'ni',
+  'no',
+  'o',
+  'of',
+  'on',
+  'the',
+  'to',
+  'wa',
+  'wo'
+]);
+
+const isMeaningfulToken = (token: string): boolean =>
+  /^\d+$/.test(token) || (token.length >= 3 && !NON_DISTINCTIVE_TOKENS.has(token));
+
 const calculateWeightedSimilarity = (title1: string, title2: string): number => {
-  const tokens1 = title1.split(' ');
-  const tokens2 = title2.split(' ');
+  const tokens1 = title1.split(' ').filter(isMeaningfulToken);
+  const tokens2 = title2.split(' ').filter(isMeaningfulToken);
+
+  if (tokens1.length === 0 || tokens2.length === 0) return 0;
+
   const set2 = new Set(tokens2);
   let score = 0;
 
@@ -54,7 +80,7 @@ const calculateWeightedSimilarity = (title1: string, title2: string): number =>
     }
   });
 
-  return score / (Math.max(tokens1.length, tokens2.length) * 2);
+  return score / ((Math.max(tokens1.length, tokens2.length) || 1) * 2);
 };
 
 export const findClosestMatch = (times: Time[], anime: Media): Time | null => {