feat(categories): add new categories for Kondis & fika and Kaffebröd, and Te & choklad

backend/src/receipt-import/receipt-import.service.ts

@@ -32,6 +32,10 @@ function tokenize(value: string): string[] {
     .filter((w) => w.length >= 3);
 }
 
+function normalizeToken(s: string): string {
+  return s.replace(/å/g, 'a').replace(/ä/g, 'a').replace(/ö/g, 'o').replace(/é/g, 'e').replace(/è/g, 'e');
+}
+
 function normalizeForRules(value: string): string {
   return value
     .toLowerCase()
@@ -153,6 +157,9 @@ export class ReceiptImportService {
     if (rawWords.length === 0) return undefined;
 
     const rawWordSet = new Set(rawWords);
+    // Normaliserade versioner (utan diakritik) för att hantera t.ex. gradde == grädde
+    const rawWordsNorm = rawWords.map(normalizeToken);
+    const rawWordSetNorm = new Set(rawWordsNorm);
 
     let best:
       | { product: { id: number; name: string; canonicalName: string | null }; score: number }
@@ -174,8 +181,9 @@ export class ReceiptImportService {
 
       for (const pw of productWords) {
         const isWeak = WEAK_DESCRIPTORS.has(pw);
+        const pwNorm = normalizeToken(pw);
 
-        if (rawWordSet.has(pw)) {
+        if (rawWordSet.has(pw) || rawWordSetNorm.has(pwNorm)) {
           exactAny += 1;
           if (isWeak) {
             score += 1;
@@ -189,7 +197,9 @@ export class ReceiptImportService {
         // Delmatchning tillåts bara för ord med minst 4 tecken.
         if (pw.length < 4) continue;
 
-        const hasPartial = rawWords.some((rw) => rw.includes(pw) || pw.includes(rw));
+        const hasPartial =
+          rawWords.some((rw) => rw.includes(pw) || pw.includes(rw)) ||
+          rawWordsNorm.some((rw) => rw.includes(pwNorm) || pwNorm.includes(rw));
         if (!hasPartial) continue;
 
         if (isWeak) {
@@ -202,7 +212,9 @@ export class ReceiptImportService {
       }
 
       // Kräv antingen minst ett starkt exakt ord, eller flera samverkande signaler.
-      const hasStrongSignal = exactStrong >= 1 || exactAny + partialStrong >= 2;
+      // Undantag: ett enstaka starkt partiellt ord (>=5 tecken) räcker, t.ex. vispgrädde → grädde.
+      const hasLongPartial = partialStrong >= 1 && productWords.some((pw) => pw.length >= 5);
+      const hasStrongSignal = exactStrong >= 1 || exactAny + partialStrong >= 2 || hasLongPartial;
       if (!hasStrongSignal) continue;
 
       // Tröskel för att undvika svaga enkelträffar.
@@ -253,6 +265,59 @@ export class ReceiptImportService {
   ): CategorySuggestion | null {
     const normalized = normalizeForRules(rawName);
 
+    // ── Regel: Te ────────────────────────────────────────────────────────
+    const isTea =
+      /\bte\b/.test(normalized) ||
+      /\btea\b/.test(normalized) ||
+      /\bchai\b/.test(normalized) ||
+      /\btepa(se|k|r)?\b/.test(normalized);
+
+    if (isTea) {
+      const l3Te = categories.find(
+        (c) => c.name.toLowerCase() === 'te' && c.path.toLowerCase().includes('te & choklad'),
+      );
+      if (l3Te) {
+        return { categoryId: l3Te.id, categoryName: l3Te.name, path: l3Te.path, confidence: 'high', usedFallback: false };
+      }
+      const l2TeChoklad = categories.find(
+        (c) => c.name.toLowerCase() === 'te & choklad' && c.path.toLowerCase().startsWith('dryck'),
+      );
+      if (l2TeChoklad) {
+        return { categoryId: l2TeChoklad.id, categoryName: l2TeChoklad.name, path: l2TeChoklad.path, confidence: 'medium', usedFallback: false };
+      }
+    }
+
+    // ── Regel: Kaffebröd ─────────────────────────────────────────────────
+    const isKaffebrod =
+      /\bwienerbrod\b/.test(normalized) ||
+      /\bdonut\b/.test(normalized) ||
+      /\bmunk\b/.test(normalized) ||
+      /\bcroissant\b/.test(normalized) ||
+      /\bkanelbulle\b/.test(normalized) ||
+      /\bbakelse\b/.test(normalized) ||
+      /\bsemla\b/.test(normalized) ||
+      /\bdammsugare\b/.test(normalized) ||
+      /\bkladdkaka\b/.test(normalized) ||
+      /\bmuffin\b/.test(normalized) ||
+      /\bcupcake\b/.test(normalized) ||
+      /\bchokladboll\b/.test(normalized);
+
+    if (isKaffebrod) {
+      const l3Kaffebrod = categories.find(
+        (c) => c.name.toLowerCase() === 'kaffebröd' && c.path.toLowerCase().includes('kondis & fika'),
+      );
+      if (l3Kaffebrod) {
+        return { categoryId: l3Kaffebrod.id, categoryName: l3Kaffebrod.name, path: l3Kaffebrod.path, confidence: 'high', usedFallback: false };
+      }
+      const l2Kondis = categories.find(
+        (c) => c.name.toLowerCase() === 'kondis & fika' && c.path.toLowerCase().startsWith('bröd & kakor'),
+      );
+      if (l2Kondis) {
+        return { categoryId: l2Kondis.id, categoryName: l2Kondis.name, path: l2Kondis.path, confidence: 'medium', usedFallback: false };
+      }
+    }
+
+    // ── Regel: Laktosfri/växtbaserad mejeri ──────────────────────────────
     const isCookingBase =
       /\bmatlagningsbas\b/.test(normalized) ||
       /\bmatlagnings\b/.test(normalized) ||