Update InitialMarkups.py

InitialMarkups.py

@@ -527,44 +527,41 @@ def print_tree_with_numbers(headers, indent=0):
               f"(Level {header['level']}, p:{header['page']+1}, {size_info})")
         print_tree_with_numbers(header["children"], indent + 1)
 
-def highlight_boxes(doc, highlights):
-    for page_num, bbox in highlights.items():
 
+def highlight_boxes(doc, highlights, stringtowrite, fixed_width=500):  # Set your desired width here
+    for page_num, bbox in highlights.items():
         page = doc.load_page(page_num)
         page_width = page.rect.width
-        rect = fitz.Rect(bbox)
-        # Get the original bounding box
+
+        # Get original rect for vertical coordinates
         orig_rect = fitz.Rect(bbox)
-        rect_width = orig_rect.width
         rect_height = orig_rect.height
-        if rect_width>10:
-          annot = page.add_rect_annot(rect)
-
-          annot.set_colors(stroke=(1,1,0), fill=(1,1,0))
-          annot.set_opacity(0.3)
-          annot.update()
-
-
-          # Calculate new x coordinates so the rect is centered on the page width
-          center_x = page_width / 2
-          new_x0 = center_x - rect_width / 2
-          new_x1 = center_x + rect_width / 2
-
-          # Create new rect centered on the page's x-center, preserving y-coordinates
-          new_rect = fitz.Rect(new_x0, orig_rect.y0, new_x1, orig_rect.y1)
-
-          # Add centered text annotation
-          text = "[To be billed]"
-          annot1 = page.add_freetext_annot(
-              new_rect,
-              text,
-              fontsize=15,
-              fontname='helv',
-              text_color=(1, 0, 0),
-              rotate=page.rotation,
-              align=1  # centered alignment
-          )
-          annot1.update()
+
+        if orig_rect.width > 10:
+            # Center horizontally using fixed width
+            center_x = page_width / 2
+            new_x0 = center_x - fixed_width / 2
+            new_x1 = center_x + fixed_width / 2
+            new_rect = fitz.Rect(new_x0, orig_rect.y0, new_x1, orig_rect.y1)
+
+            # Add highlight rectangle
+            annot = page.add_rect_annot(new_rect)
+            annot.set_colors(stroke=(1, 1, 0), fill=(1, 1, 0))
+            annot.set_opacity(0.3)
+            annot.update()
+
+            # Add right-aligned freetext annotation inside the fixed-width box
+            text = '['+stringtowrite +']'
+            annot1 = page.add_freetext_annot(
+                new_rect,
+                text,
+                fontsize=15,
+                fontname='helv',
+                text_color=(1, 0, 0),
+                rotate=page.rotation,
+                align=2  # right alignment
+            )
+            annot1.update()
 
 def get_leaf_headers_with_paths(listtoloop, path=None, output=None):
     if path is None:
@@ -742,7 +739,10 @@ def extract_section_under_header(pdf_path):
                     if combined_line_norm and combined_line_norm in paths[-2]:
                         print(combined_line_norm)
                         headertoContinue2 = combined_line_norm
-
+                    if  'installation' in paths[-2].lower() or 'execution' in paths[-2].lower() or 'miscellaneous items' in paths[-2].lower() :
+                      stringtowrite='Not to be billed'
+                    else:
+                      stringtowrite='To be billed'
                     # Optimized header matching
                     existsfull = (
                         ( combined_line_norm in allchildrenheaders_set or
@@ -839,7 +839,7 @@ def extract_section_under_header(pdf_path):
                                         "Author": "ADR",
                                         "Creation Date": formatted_time,
                                         "Layer": "Initial",
-                                        "Code": "to be added",
+                                        "Code": stringtowrite,
                                         "head above 1":  paths[-2],
                                         "head above 2":  paths[0]
                                     }
@@ -935,7 +935,7 @@ def extract_section_under_header(pdf_path):
                                             "Author": "ADR",
                                             "Creation Date": formatted_time,
                                             "Layer": "Initial",
-                                            "Code": "to be added",
+                                            "Code": stringtowrite,
                                             "head above 1":  paths[-2],
                                             "head above 2":  paths[0]
                                         }
@@ -976,7 +976,7 @@ def extract_section_under_header(pdf_path):
                                   for page_num, bbox in current_bbox.items():
                                       bbox[3] = last_y1s.get(page_num, bbox[3])
                                       page_highlights[page_num] = bbox
-                                  highlight_boxes(docHighlights, page_highlights)
+                                  highlight_boxes(docHighlights, page_highlights,stringtowrite)
 
                                   break_collecting = True
                                   break
@@ -1012,7 +1012,11 @@ def extract_section_under_header(pdf_path):
             for page_num, bbox in current_bbox.items():
                 bbox[3] = last_y1s.get(page_num, bbox[3])
                 page_highlights[page_num] = bbox
-            highlight_boxes(docHighlights, page_highlights)
+            if  'installation' in paths[-2].lower() or 'execution' in paths[-2].lower() or 'miscellaneous items' in paths[-2].lower() :
+                stringtowrite='Not to be billed'
+            else:
+                stringtowrite='To be billed'
+            highlight_boxes(docHighlights, page_highlights,stringtowrite)
 
     # docHighlights.save("highlighted_output.pdf", garbage=4, deflate=True)