feat: Implement site-specific recipe parsers for ICA and generic fallback

2026-04-12 09:39:32 +02:00
parent 2c92e07d39
commit 4e2616fe2e
5 changed files with 414 additions and 86 deletions
@@ -0,0 +1,136 @@
+import { RecipeParser, ParsedRecipe } from './base.parser';
+
+/**
+ * Generisk parser för okända receptsidor
+ * Försöker JSON-LD först, sedan vanlig HTML-parsing
+ * Denna är mer permissiv än site-specifika parsers
+ */
+export class GenericRecipeParser extends RecipeParser {
+  canHandle(url: string): boolean {
+    // Denna parser hanterar alltid (är fallback)
+    return true;
+  }
+
+  parse(html: string): ParsedRecipe {
+    console.log('[GenericParser] Parsing recipe from unknown site...');
+
+    // Försöka extrahera JSON-LD recipe data
+    const jsonLdMatch = html.match(
+      /<script[^>]*type="application\/ld\+json"[^>]*>([\s\S]*?)<\/script>/i
+    );
+
+    if (jsonLdMatch) {
+      try {
+        const jsonData = JSON.parse(jsonLdMatch[1]);
+        const recipe =
+          jsonData['@type'] === 'Recipe'
+            ? jsonData
+            : jsonData['@graph']?.find((item) => item['@type'] === 'Recipe');
+
+        if (recipe) {
+          console.log('[GenericParser] ✓ JSON-LD data found');
+          return this.extractFromJsonLd(recipe);
+        }
+      } catch (err) {
+        console.log('[GenericParser] JSON-LD parsing failed');
+      }
+    }
+
+    console.log('[GenericParser] No JSON-LD found, using HTML parsing');
+    return this.parseFromHtml(html);
+  }
+
+  private extractFromJsonLd(recipe: any): ParsedRecipe {
+    const name = recipe.name || '';
+
+    const ingredients: Array<{ quantity: number; unit: string; name: string }> = [];
+    if (recipe.recipeIngredient && Array.isArray(recipe.recipeIngredient)) {
+      for (const ing of recipe.recipeIngredient) {
+        const parsed = this.parseIngredientLine(ing);
+        if (parsed) {
+          ingredients.push(parsed);
+        }
+      }
+    }
+
+    let instructions = '';
+    if (recipe.recipeInstructions) {
+      if (typeof recipe.recipeInstructions === 'string') {
+        instructions = recipe.recipeInstructions;
+      } else if (Array.isArray(recipe.recipeInstructions)) {
+        instructions = recipe.recipeInstructions
+          .map((step) => {
+            if (typeof step === 'string') return step;
+            if (step.text) return step.text;
+            return '';
+          })
+          .filter((s) => s)
+          .join('\n\n');
+      }
+    }
+
+    return {
+      name,
+      ingredients,
+      instructions,
+    };
+  }
+
+  private parseFromHtml(html: string): ParsedRecipe {
+    // Försöka hitta titel
+    let name = '';
+    
+    // Prova olika selector-mönster
+    let titleMatch =
+      html.match(/<h1[^>]*>([^<]+)<\/h1>/i) ||
+      html.match(/<meta\s+property="og:title"\s+content="([^"]+)"/i) ||
+      html.match(/<title>([^<]+)<\/title>/i);
+
+    if (titleMatch) {
+      name = titleMatch[1].trim();
+    }
+
+    // Försöka extrahera ingredienser från vanliga strukturer
+    const ingredients: Array<{ quantity: number; unit: string; name: string }> = [];
+    
+    // Testa olika ingredient-selectors
+    const ingredientPatterns = [
+      /<li[^>]*>(.*?)<\/li>/gi,
+      /<div[^>]*class="ingredient"[^>]*>(.*?)<\/div>/gi,
+      /<p[^>]*class="ingredient"[^>]*>(.*?)<\/p>/gi,
+    ];
+
+    for (const pattern of ingredientPatterns) {
+      let match;
+      while ((match = pattern.exec(html)) !== null) {
+        const parsed = this.parseIngredientLine(match[1]);
+        if (parsed && parsed.name.length > 2) {
+          // Undvik mycket korta ingredienser (troligen brus)
+          ingredients.push(parsed);
+        }
+      }
+      if (ingredients.length > 0) break; // Om vi hittat några, använd dessa
+    }
+
+    // Försöka hitta instruktioner
+    let instructions = '';
+    const instructionsPatterns = [
+      /<(?:div|section)[^>]*class="[^"]*(?:instruction|method|step)[^"]*"[^>]*>(.*?)<\/(?:div|section)>/is,
+      /<ol[^>]*>(.*?)<\/ol>/i,
+    ];
+
+    for (const pattern of instructionsPatterns) {
+      const match = html.match(pattern);
+      if (match) {
+        instructions = match[1].replace(/<[^>]+>/g, '').trim();
+        if (instructions.length > 10) break;
+      }
+    }
+
+    return {
+      name,
+      ingredients,
+      instructions,
+    };
+  }
+}