watermark detection

foodie_utils.py

@@ -1086,7 +1086,7 @@ def get_flickr_image(search_query, relevance_keywords, main_topic):
     logger = logging.getLogger(__name__)
     
     def process_image(image_url, source_name, page_url):
-        """Download image, check for text with OCR, validate resolution, and exclude screenshots."""
+        """Download image, check for text with OCR, validate resolution, exclude screenshots and watermarks."""
         try:
             headers = {'User-Agent': 'InsiderFoodieBot/1.0 (https://insiderfoodie.com; contact@insiderfoodie.com)'}
             response = requests.get(image_url, headers=headers, timeout=10)
@@ -1095,7 +1095,7 @@ def get_flickr_image(search_query, relevance_keywords, main_topic):
             
             # Check image resolution
             width, height = img.size
-            min_dimension = 1280  # Minimum width or height for high quality
+            min_dimension = 1280
             if width < min_dimension and height < min_dimension:
                 logger.info(f"Skipping low-resolution image: {image_url} ({width}x{height})")
                 return None
@@ -1106,10 +1106,27 @@ def get_flickr_image(search_query, relevance_keywords, main_topic):
                 logger.info(f"Skipping potential screenshot: {image_url} (aspect ratio: {aspect_ratio})")
                 return None
             
-            # OCR to detect text (unchanged)
-            text = pytesseract.image_to_string(img).strip()
+            # Check for watermarks in URL or page URL
+            watermark_domains = [
+                'shutterstock.com', 'gettyimages.com', 'istockphoto.com', 'adobestock.com',
+                '123rf.com', 'dreamstime.com', 'alamy.com', 'stock.adobe.com'
+            ]
+            if any(domain in image_url.lower() or domain in page_url.lower() for domain in watermark_domains):
+                logger.info(f"Skipping image from stock photo site (potential watermark): {image_url}")
+                return None
+            
+            # OCR to detect text and watermarks
+            text = pytesseract.image_to_string(img).strip().lower()
+            watermark_phrases = [
+                'shutterstock', 'getty images', 'istock', 'adobe stock', 'watermark',
+                '123rf', 'dreamstime', 'alamy', 'preview', 'stock photo'
+            ]
+            if any(phrase in text for phrase in watermark_phrases):
+                logger.info(f"Skipping watermarked image: {image_url} (detected: {text})")
+                return None
+            
             word_count = len(text.split())
-            if word_count > 10:
+            if word_count > 5:  # Lowered threshold for stricter filtering
                 logger.info(f"Skipping image with too much text: {image_url} ({word_count} words)")
                 return None
             
@@ -1131,7 +1148,7 @@ def get_flickr_image(search_query, relevance_keywords, main_topic):
     logger.info(f"Searching DDG with query: '{ddg_query}'")
     try:
         with DDGS() as ddgs:
-            results = ddgs.images(ddg_query, safesearch="on", max_results=20)  # Increased to 20 for more options
+            results = ddgs.images(ddg_query, safesearch="on", max_results=20)
             for result in results:
                 image_url = result.get("image")
                 page_url = result.get("url")
@@ -1159,6 +1176,87 @@ def get_flickr_image(search_query, relevance_keywords, main_topic):
     
     logger.warning(f"No valid images found for query '{search_query}'")
     return None, None, None, None
+
+def get_image(search_query):
+    headers = {'User-Agent': 'InsiderFoodieBot/1.0 (https://insiderfoodie.com; contact@insiderfoodie.com)'}
+    
+    def process_image(image_url, source_name, page_url):
+        """Helper to process Pixabay images for watermarks and resolution."""
+        try:
+            response = requests.get(image_url, headers=headers, timeout=10)
+            response.raise_for_status()
+            img = Image.open(io.BytesIO(response.content))
+            
+            # Check resolution
+            width, height = img.size
+            min_dimension = 1280
+            if width < min_dimension and height < min_dimension:
+                logger.info(f"Skipping low-resolution Pixabay image: {image_url} ({width}x{height})")
+                return None
+            
+            # Check for watermarks via OCR
+            text = pytesseract.image_to_string(img).strip().lower()
+            watermark_phrases = [
+                'shutterstock', 'getty images', 'istock', 'adobe stock', 'watermark',
+                '123rf', 'dreamstime', 'alamy', 'preview', 'stock photo'
+            ]
+            if any(phrase in text for phrase in watermark_phrases):
+                logger.info(f"Skipping watermarked Pixabay image: {image_url} (detected: {text})")
+                return None
+            
+            word_count = len(text.split())
+            if word_count > 5:
+                logger.info(f"Skipping Pixabay image with too much text: {image_url} ({word_count} words)")
+                return None
+            
+            return img_url, source_name, uploader, page_url
+        except Exception as e:
+            logger.warning(f"Failed to process Pixabay image {image_url}: {e}")
+            return None
+    
+    def fetch_pixabay_image(query):
+        try:
+            pixabay_url = f"https://pixabay.com/api/?key={PIXABAY_API_KEY}&q={quote(query)}&image_type=photo&per_page=20"
+            response = requests.get(pixabay_url, headers=headers, timeout=10)
+            response.raise_for_status()
+            data = response.json()
+            
+            for hit in data.get('hits', []):
+                img_url = hit.get('largeImageURL')
+                if not img_url or img_url in used_images:
+                    continue
+                
+                uploader = hit.get('user', 'Unknown')
+                page_url = hit.get('pageURL', img_url)
+                
+                # Process the image for watermarks and resolution
+                result = process_image(img_url, "Pixabay", page_url)
+                if result:
+                    used_images.add(img_url)
+                    save_used_images()
+                    logger.info(f"Selected Pixabay image: {img_url} by {uploader} for query '{query}' ({result[0].split('x')[0]}x{result[0].split('x')[1]})")
+                    return result
+                
+            logger.info(f"No valid Pixabay image found for query '{query}'. Trying fallback query.")
+            return None, None, None, None
+        
+        except Exception as e:
+            logger.warning(f"Pixabay image fetch failed for query '{query}': {e}")
+            return None, None, None, None
+    
+    # Try with the original query
+    image_url, source_name, uploader, page_url = fetch_pixabay_image(search_query)
+    if image_url:
+        return image_url, source_name, uploader, page_url
+    
+    # Fallback to a generic query
+    fallback_query = "food dining"
+    image_url, source_name, uploader, page_url = fetch_pixabay_image(fallback_query)
+    if image_url:
+        return image_url, source_name, uploader, page_url
+    
+    logger.error(f"All image fetch attempts failed for query '{search_query}'. Returning None.")
+    return None, None, None, None
     
 def select_best_author(content, interest_score):
     try: