Update app.py

app.py

@@ -140,39 +140,69 @@ def identify_headers_with_openrouter(pdf_path, model, LLM_prompt, pages_to_check
         # if pno in toc_pages:
         #     logger.debug(f"Skipping TOC page {pno}")
         #     continue
-        
         page = doc.load_page(pno)
         page_height = page.rect.height
-        lines_on_page = 0
+        
         text_dict = page.get_text("dict")
-        lines = []
-        y_tolerance = 0.2  # tweak if needed (1–3 usually works)
-        for block in page.get_text("dict").get('blocks', []):
-            if block.get('type') != 0:
+        lines_for_prompt = []
+        lines_on_page = 0
+        
+        for block in text_dict.get("blocks", []):
+            if block.get("type") != 0:  # text blocks only
                 continue
-            for line in block.get('lines', []):
-                spans = line.get('spans', [])
+        
+            for line in block.get("lines", []):
+                spans = line.get("spans", [])
                 if not spans:
                     continue
-                y0 = spans[0]['bbox'][1]
-                y1 = spans[0]['bbox'][3]
-                if y0 < top_margin or y1 > (page_height - bottom_margin):
-                    continue
-                for s in spans:
-                    # text,font,size,flags,color
-                    # ArrayofTextWithFormat={'Font':s.get('font')},{'Size':s.get('size')},{'Flags':s.get('flags')},{'Color':s.get('color')},{'Text':s.get('text')}
+        
+                # Use first span to check vertical position
+                y0 = spans[0]["bbox"][1]
+                y1 = spans[0]["bbox"][3]
+        
+                # if y0 < top_margin or y1 > (page_height - bottom_margin):
+                #     continue
+        
+                for span in spans:
+                    text = span.get("text", "").strip()
+                    if not text:
+                        continue
+        
+                    lines_for_prompt.append(f"PAGE {pno + 1}: {text}")
+                    lines_on_page += 1
+
+        # page = doc.load_page(pno)
+        # page_height = page.rect.height
+        # lines_on_page = 0
+        # text_dict = page.get_text("dict")
+        # lines = []
+        # y_tolerance = 0.2  # tweak if needed (1–3 usually works)
+        # for block in page.get_text("dict").get('blocks', []):
+        #     if block.get('type') != 0:
+        #         continue
+        #     for line in block.get('lines', []):
+        #         spans = line.get('spans', [])
+        #         if not spans:
+        #             continue
+        #         y0 = spans[0]['bbox'][1]
+        #         y1 = spans[0]['bbox'][3]
+        #         if y0 < top_margin or y1 > (page_height - bottom_margin):
+        #             continue
+        #         for s in spans:
+        #             # text,font,size,flags,color
+        #             # ArrayofTextWithFormat={'Font':s.get('font')},{'Size':s.get('size')},{'Flags':s.get('flags')},{'Color':s.get('color')},{'Text':s.get('text')}
                 
-                    # prefix with page for easier mapping back
-                    text = s["text"].strip()
-                    lines_for_prompt.append(f"PAGE {pno+1}: {text}")
+        #             # prefix with page for easier mapping back
+        #             text = s["text"].strip()
+        #             lines_for_prompt.append(f"PAGE {pno+1}: {text}")
 
-            # if not lines_for_prompt:
-            #     return []       
+        #     # if not lines_for_prompt:
+        #     #     return []       
 
-            if text:
-                # prefix with page for easier mapping back
-                # lines_for_prompt.append(f"PAGE {pno+1}: {line}")
-                lines_on_page += 1
+        #     if text:
+        #         # prefix with page for easier mapping back
+        #         # lines_for_prompt.append(f"PAGE {pno+1}: {line}")
+        #         lines_on_page += 1
 
         
         if lines_on_page > 0: