fix(scoring): a Solr operator is not an acronym

src/tools/package.ts

@@ -224,13 +224,26 @@ const QUERY_STOPWORDS = new Set([
   "those"
 ]);
 
+/** Solr boolean keywords: all-caps by convention, never a term to score on. */
+const SOLR_OPERATORS = new Set(["AND", "OR", "NOT"]);
+
 export const extractQueryTerms = (query: string): string[] => {
-  const raw = query.normalize("NFC").match(/[\p{L}\p{N}]+/gu) ?? [];
+  const normalized = query.normalize("NFC");
+  const raw = normalized.match(/[\p{L}\p{N}]+/gu) ?? [];
+  // Inside double quotes Solr reads a keyword as a literal, so `"OR"` is a term to
+  // score on while a bare `OR` is syntax. Tokenisation has already dropped the quotes,
+  // so the quoted tokens are collected first.
+  const quoted = new Set<string>();
+  for (const m of normalized.matchAll(/"([^"]*)"/g)) {
+    for (const t of m[1].match(/[\p{L}\p{N}]+/gu) ?? []) quoted.add(t);
+  }
   // An all-caps token is an acronym, not an article: the stopword list is there for
   // `defibrillatori Comune di Lecce`, and must not swallow the `UN` of `UN population`
-  // on a catalog in another language.
+  // on a catalog in another language. Solr's own operators are the exception to the
+  // exception — `aria OR acqua` is a query, not a mention of an organisation called OR.
   const terms = raw
     .filter((token) => {
+      if (SOLR_OPERATORS.has(token) && !quoted.has(token)) return false;
       const term = token.toLowerCase();
       if (term.length <= 1) return false;
       if (!QUERY_STOPWORDS.has(term)) return true;

tests/unit/package-scoring.test.ts

@@ -637,6 +637,20 @@ describe('multilingual term extraction', () => {
     expect(extractQueryTerms('un comune di lecce')).toEqual(['comune', 'lecce']);
   });
 
+  it('never scores on a Solr operator, even though it is all-caps', () => {
+    // The acronym rule let `OR` through: `aria OR acqua` scored on three terms and
+    // a title carrying one of them got 4 × 1/3 instead of 4 × 1/2.
+    expect(extractQueryTerms('aria OR acqua')).toEqual(['aria', 'acqua']);
+    expect(extractQueryTerms('aria AND NOT rifiuti')).toEqual(['aria', 'rifiuti']);
+    // lowercase `or` was already a stopword; an English `Or` mid-sentence stays one too
+    expect(extractQueryTerms('water or sewage')).toEqual(['water', 'sewage']);
+  });
+
+  it('keeps a quoted operator: inside quotes Solr reads it as a literal', () => {
+    expect(extractQueryTerms('"OR" Oregon')).toEqual(['or', 'oregon']);
+    expect(extractQueryTerms('aria OR "AND"')).toEqual(['aria', 'and']);
+  });
+
   it('matches across Unicode normal forms', () => {
     const nfd = 'mobilita\u0300 urbana';          // decomposed
     expect(countMatchingTerms(nfd, ['mobilità'])).toBe(1);