feat(import): enhance image URL handling and error reporting during recipe import

backend/src/common/utils/download-image.ts

@@ -20,19 +20,24 @@ export async function downloadAndOptimizeImage(
   sourceUrl: string,
   destDir: string,
 ): Promise<string> {
-  // Protokollvalidering
+  const raw = sourceUrl.trim();
-  if (!sourceUrl.startsWith('https://')) {
+  const protocolNormalized = raw.startsWith('//') ? `https:${raw}` : raw;
-    throw new Error('Bild-URL måste använda https://');
-  }
 
   // SSRF: blockera privata hostnames
-  let hostname: string;
+  let parsedUrl: URL;
   try {
-    hostname = new URL(sourceUrl).hostname;
+    parsedUrl = new URL(protocolNormalized);
   } catch {
     throw new Error('Ogiltig bild-URL');
   }
 
+  // Protokollvalidering
+  if (parsedUrl.protocol !== 'https:') {
+    throw new Error('Bild-URL måste använda https://');
+  }
+
+  const hostname = parsedUrl.hostname;
+
   if (BLOCKED_HOSTNAMES.test(hostname)) {
     throw new Error('Bild-URL pekar på ett blockerat nätverk');
   }
@@ -42,7 +47,7 @@ export async function downloadAndOptimizeImage(
   const timeout = setTimeout(() => controller.abort(), 10_000);
   let response: Response;
   try {
-    response = await fetch(sourceUrl, {
+    response = await fetch(parsedUrl.toString(), {
       signal: controller.signal,
       headers: { 'User-Agent': 'Mozilla/5.0 (compatible; RecipeApp/1.0)' },
     });

backend/src/quick-import/parsers/generic.parser.ts

@@ -19,24 +19,22 @@ export class GenericRecipeParser extends RecipeParser {
     // Extrahera og:image för bildurl-fallback
     const ogImage = this.extractOgImage(html);
 
-    // Försöka extrahera JSON-LD recipe data
+    // Försöka extrahera JSON-LD recipe data (flera script-taggar är vanligt)
-    const jsonLdMatch = html.match(
+    const jsonLdRegex =
-      /<script[^>]*type="application\/ld\+json"[^>]*>([\s\S]*?)<\/script>/i
+      /<script[^>]*type="application\/ld\+json"[^>]*>([\s\S]*?)<\/script>/gi;
-    );
+    let jsonLdMatch: RegExpExecArray | null;
+    while ((jsonLdMatch = jsonLdRegex.exec(html)) !== null) {
+      const rawJson = jsonLdMatch[1]?.trim();
+      if (!rawJson) continue;
 
-    if (jsonLdMatch) {
       try {
-        const jsonData = JSON.parse(jsonLdMatch[1]);
+        const parsedJson = JSON.parse(rawJson);
-        const recipe =
+        const recipe = this.findRecipeInJsonLd(parsedJson);
-          jsonData['@type'] === 'Recipe'
-            ? jsonData
-            : jsonData['@graph']?.find((item: any) => item['@type'] === 'Recipe');
-
         if (recipe) {
           this.logger.log('JSON-LD data found');
           return this.extractFromJsonLd(recipe, ogImage);
         }
-      } catch (err) {
+      } catch {
         this.logger.warn('JSON-LD parsing failed');
       }
     }
@@ -45,6 +43,37 @@ export class GenericRecipeParser extends RecipeParser {
     return this.parseFromHtml(html, ogImage);
   }
 
+  private findRecipeInJsonLd(jsonData: any): any {
+    if (!jsonData) return null;
+
+    if (Array.isArray(jsonData)) {
+      for (const item of jsonData) {
+        const recipe = this.findRecipeInJsonLd(item);
+        if (recipe) return recipe;
+      }
+      return null;
+    }
+
+    if (jsonData['@type'] === 'Recipe') {
+      return jsonData;
+    }
+
+    if (Array.isArray(jsonData['@type']) && jsonData['@type'].includes('Recipe')) {
+      return jsonData;
+    }
+
+    const graph = jsonData['@graph'];
+    if (Array.isArray(graph)) {
+      return graph.find(
+        (item: any) =>
+          item?.['@type'] === 'Recipe' ||
+          (Array.isArray(item?.['@type']) && item['@type'].includes('Recipe')),
+      ) ?? null;
+    }
+
+    return null;
+  }
+
   private extractOgImage(html: string): string | undefined {
     const match = html.match(/<meta[^>]+property="og:image"[^>]+content="([^"]+)"/i)
       || html.match(/<meta[^>]+content="([^"]+)"[^>]+property="og:image"/i);

backend/src/quick-import/parsers/ica.parser.ts

@@ -17,21 +17,17 @@ export class IcaRecipeParser extends RecipeParser {
     // Extrahera og:image för bildurl-fallback
     const ogImage = this.extractOgImage(html);
 
-    // Försöka extrahera JSON-LD recipe data (ICA använder detta)
+    // Försöka extrahera JSON-LD recipe data (ICA använder ofta flera script-taggar)
-    const jsonLdMatch = html.match(
+    const jsonLdRegex =
-      /<script[^>]*type="application\/ld\+json"[^>]*>([\s\S]*?)<\/script>/i
+      /<script[^>]*type="application\/ld\+json"[^>]*>([\s\S]*?)<\/script>/gi;
-    );
+    let jsonLdMatch: RegExpExecArray | null;
+    while ((jsonLdMatch = jsonLdRegex.exec(html)) !== null) {
+      const rawJson = jsonLdMatch[1]?.trim();
+      if (!rawJson) continue;
 
-    if (jsonLdMatch) {
       try {
-        const jsonData = JSON.parse(jsonLdMatch[1]);
+        const parsedJson = JSON.parse(rawJson);
-
+        const recipe = this.findRecipeInJsonLd(parsedJson);
-        // Hitta recipe-objektet
-        const recipe =
-          jsonData['@type'] === 'Recipe'
-            ? jsonData
-            : jsonData['@graph']?.find((item: any) => item['@type'] === 'Recipe');
-
         if (recipe) {
           this.logger.log('JSON-LD recipe found');
           return this.extractFromJsonLd(recipe, ogImage);
@@ -46,6 +42,37 @@ export class IcaRecipeParser extends RecipeParser {
     return this.parseFromHtml(html, ogImage);
   }
 
+  private findRecipeInJsonLd(jsonData: any): any {
+    if (!jsonData) return null;
+
+    if (Array.isArray(jsonData)) {
+      for (const item of jsonData) {
+        const recipe = this.findRecipeInJsonLd(item);
+        if (recipe) return recipe;
+      }
+      return null;
+    }
+
+    if (jsonData['@type'] === 'Recipe') {
+      return jsonData;
+    }
+
+    if (Array.isArray(jsonData['@type']) && jsonData['@type'].includes('Recipe')) {
+      return jsonData;
+    }
+
+    const graph = jsonData['@graph'];
+    if (Array.isArray(graph)) {
+      return graph.find(
+        (item: any) =>
+          item?.['@type'] === 'Recipe' ||
+          (Array.isArray(item?.['@type']) && item['@type'].includes('Recipe')),
+      ) ?? null;
+    }
+
+    return null;
+  }
+
   private extractOgImage(html: string): string | undefined {
     const match = html.match(/<meta[^>]+property="og:image"[^>]+content="([^"]+)"/i)
       || html.match(/<meta[^>]+content="([^"]+)"[^>]+property="og:image"/i);

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

@@ -20,6 +20,7 @@ export interface QuickImportResult {
   markdown: string;
   source: 'ica' | 'pdf' | 'image' | 'other';
   imageUrl?: string;
+  imageWarning?: string;
 }
 
 type UploadKind = 'pdf' | 'image';
@@ -317,12 +318,24 @@ export class QuickImportService {
 
       // Ladda ner och optimera bild om parser hittade en
       let imageUrl: string | undefined;
+      let imageWarning: string | undefined;
       if (recipe.imageUrl) {
+        const normalizedImageUrl = this.normalizeImageUrl(recipe.imageUrl, url);
+        if (!normalizedImageUrl) {
+          imageWarning = 'Receptbild kunde inte tolkas till en giltig URL.';
+          this.logger.warn(
+            `Kunde inte normalisera bild-URL: "${recipe.imageUrl}" (källsida: ${url})`,
+          );
+        } else {
           try {
-          imageUrl = await downloadAndOptimizeImage(recipe.imageUrl, IMAGE_DEST_DIR);
+            imageUrl = await downloadAndOptimizeImage(normalizedImageUrl, IMAGE_DEST_DIR);
             this.logger.log(`Bild optimerad och sparad: ${imageUrl}`);
           } catch (imgErr) {
-          this.logger.warn(`Kunde inte ladda ner bild: ${imgErr}`);
+            imageWarning = 'Receptbild kunde inte laddas ner.';
+            this.logger.warn(
+              `Kunde inte ladda ner bild: ${imgErr} (källa: ${normalizedImageUrl})`,
+            );
+          }
         }
       }
 
@@ -330,6 +343,7 @@ export class QuickImportService {
         markdown,
         source,
         imageUrl,
+        imageWarning,
       };
     } catch (err) {
       const message = err instanceof Error ? err.message : 'Okänt fel vid scraping';
@@ -340,6 +354,20 @@ export class QuickImportService {
     }
   }
 
+  private normalizeImageUrl(rawImageUrl: string, pageUrl: string): string | null {
+    const trimmed = rawImageUrl.trim();
+    if (!trimmed) return null;
+
+    const protocolNormalized =
+      trimmed.startsWith('//') ? `https:${trimmed}` : trimmed;
+
+    try {
+      return new URL(protocolNormalized, pageUrl).toString();
+    } catch {
+      return null;
+    }
+  }
+
   /**
    * Konvertera receptobjekt till Markdown-format
    */