add margin

src/controllers/_file_controller.py

@@ -28,6 +28,7 @@ class FileController:
         try:
             async with self.service() as service:
                 result = await service.extract_file(file=file)
+            result.update({"file_name": file.filename})
             return result
         except Exception as e:
             logger.error(e)

src/utils/_file_client.py

@@ -28,28 +28,132 @@ class FileClient:
             "pages": []
         }
 
+        all_headers = []
+        all_footers = []
+
         for page_num, page in enumerate(doc):
             blocks = page.get_text("dict")["blocks"]
             page_content = []
+            text_blocks = []
+            header_blocks = []
+            footer_blocks = []
+            body_blocks = []
+
+            page_rect = page.rect
+            media_box = page.mediabox if hasattr(page, 'mediabox') else page_rect
+            crop_box = page.cropbox
+
+            header_region = (page_rect.y0, page_rect.y0 + (page_rect.height * 0.2))
+            footer_region = (page_rect.y1 - (page_rect.height * 0.2), page_rect.y1)
 
             for block in blocks:
-                if "lines" in block:
+                if "lines" in block and block.get("bbox"):
+                    block_bbox = block["bbox"]
+                    
+                    is_header = block_bbox[1] >= header_region[0] and block_bbox[3] <= header_region[1]
+                    is_footer = block_bbox[1] >= footer_region[0] and block_bbox[3] <= footer_region[1]
+
+                    if is_header:
+                        header_blocks.append(block_bbox)
+                    elif is_footer:
+                        footer_blocks.append(block_bbox)
+                    else:
+                        body_blocks.append(block_bbox)
+                        text_blocks.append(block_bbox)
+
                     for line in block["lines"]:
                         for span in line["spans"]:
-                            page_content.append({
+                            content_item = {
                                 "text": span["text"],
                                 "font": span["font"],
                                 "size": span["size"],
                                 "color": span.get("color", None),
-                                "flags": span.get("flags", None)
-                            })
-
-            layout_data["pages"].append({
+                                "flags": span.get("flags", None),
+                                "is_header": is_header,
+                                "is_footer": is_footer
+                            }
+                            page_content.append(content_item)
+
+            if body_blocks:
+                min_x = min(block[0] for block in body_blocks)
+                min_y = min(block[1] for block in body_blocks)
+                max_x = max(block[2] for block in body_blocks)
+                max_y = max(block[3] for block in body_blocks)
+                
+                margin_left = (min_x - page_rect.x0) / 72
+                margin_top = (min_y - page_rect.y0) / 72
+                margin_right = (page_rect.x1 - max_x) / 72
+                margin_bottom = (page_rect.y1 - max_y) / 72
+            else:
+                margin_left = margin_top = margin_right = margin_bottom = 0
+
+            crop_margin_left = (crop_box.x0 - media_box.x0) / 72
+            crop_margin_top = (crop_box.y0 - media_box.y0) / 72
+            crop_margin_right = (media_box.x1 - crop_box.x1) / 72
+            crop_margin_bottom = (media_box.y1 - crop_box.y1) / 72
+
+            header_height = 0
+            footer_height = 0
+            if header_blocks:
+                header_min_y = min(block[1] for block in header_blocks)
+                header_max_y = max(block[3] for block in header_blocks)
+                header_height = (header_max_y - header_min_y) / 72
+            if footer_blocks:
+                footer_min_y = min(block[1] for block in footer_blocks)
+                footer_max_y = max(block[3] for block in footer_blocks)
+                footer_height = (footer_max_y - footer_min_y) / 72
+
+            header_text = ""
+            footer_text = ""
+            for item in page_content:
+                if item["is_header"]:
+                    header_text += item["text"] + " "
+                elif item["is_footer"]:
+                    footer_text += item["text"] + " "
+
+            header_text = header_text.strip()
+            footer_text = footer_text.strip()
+
+            if header_text:
+                all_headers.append(header_text)
+            if footer_text:
+                all_footers.append(footer_text)
+
+            page_data = {
                 "page_number": page_num + 1,
-                "width": page.rect.width,
-                "height": page.rect.height,
+                "width": page_rect.width,
+                "height": page_rect.height,
+                "margin_top": f"{round(margin_top, 1)} inches",
+                "margin_left": f"{round(margin_left, 1)} inches",
+                "margin_right": f"{round(margin_right, 1)} inches",
+                "margin_bottom": f"{round(margin_bottom, 1)} inches",
+                "header_height": f"{header_height} inches",
+                "footer_height": f"{footer_height} inches",
+                "has_header": len(header_blocks) > 0,
+                "has_footer": len(footer_blocks) > 0,
                 "content": page_content
-            })
+            }
+
+            if page_num == 0:
+                page_data["is_first_page"] = True
+
+            layout_data["pages"].append(page_data)
+
+        if all_headers:
+            unique_headers = set(all_headers)
+            layout_data["header_analysis"] = {
+                "total_pages_with_headers": len(all_headers),
+                "unique_headers": len(unique_headers),
+                "is_header_consistent": len(unique_headers) == 1 if all_headers else False
+            }
+
+        if all_footers:
+            unique_footers = set(all_footers)
+            layout_data["footer_analysis"] = {
+                "total_pages_with_footers": len(all_footers),
+                "unique_footers": len(unique_footers),
+                "is_footer_consistent": len(unique_footers) == 1 if all_footers else False
+            }
 
         return layout_data