Use per-page annotation spacing with inferred margin fallback

copienator/annotation_layout.py

@@ -61,13 +61,26 @@ def move_answer_pdf(source, destination):
     metadata_path(source).unlink(missing_ok=True)
 
 
+def added_margin(margin_mm):
+    """Interpolate the configured (millimetres, pixels) points, clamping at ends."""
+    if margin_mm is None:
+        return configuration.ANNOTATION_MARGIN_LEFT
+    points = configuration.ANNOTATION_MARGIN_POINTS
+    if (not points or any(x < 0 or type(y) is not int or y < 0 for x, y in points)
+            or any(a[0] >= b[0] for a, b in zip(points, points[1:]))):
+        raise ValueError('Annotation margin points must have increasing mm and non-negative integer pixels')
+    if margin_mm <= points[0][0]:
+        return points[0][1]
+    for (x0, y0), (x1, y1) in zip(points, points[1:]):
+        if margin_mm <= x1:
+            return round(y0 + (y1 - y0) * (margin_mm - x0) / (x1 - x0))
+    return points[-1][1]
+
+
 def layout_for(pdf):
-    low = any(part.get('margin_status') == 'low' for part in answer_parts(pdf))
-    margin = (configuration.ANNOTATION_LOW_MARGIN_LEFT if low
-              else configuration.ANNOTATION_MARGIN_LEFT)
-    if type(margin) is not int or margin < 0:
-        raise ValueError('Annotation left margins must be non-negative integers')
-    return {'version': 1, 'margin_left': margin, 'narrow_left_pad_limit': 300}
+    margins = [added_margin(part.get('margin_mm')) for part in answer_parts(pdf)]
+    return {'version': 2, 'margin_left': max(margins, default=added_margin(None)),
+            'page_margins': margins, 'narrow_left_pad_limit': 0}
 
 
 def legacy_layout():

copienator/commands/annotating.py

@@ -63,9 +63,13 @@ def make_base_image(pdf_path):
     base_img = Image.new("RGBA", (max_w, total_h), "white")
 
     current_y = 0
+    page_regions = []
     for page in pages:
-        base_img.paste(page.convert("RGBA"), ((max_w - page.width) // 2, current_y))
+        x = (max_w - page.width) // 2
+        base_img.paste(page.convert("RGBA"), (x, current_y))
+        page_regions.append({'x': x, 'y': current_y, 'width': page.width, 'height': page.height})
         current_y += page.height
+    base_img.info['page_regions'] = page_regions
     return (base_img, total_h, max_w)
 
 def normalize_mathtext(text):
@@ -352,6 +356,10 @@ def compose_label_image(base_img, label, result, hmin,
 
     layout = dict(layout if layout is not None else layout_for(source_pdf))
     margin_left = layout['margin_left']
+    page_regions = base_img.info.get('page_regions', [
+        {'x': 0, 'y': 0, 'width': base_img.width, 'height': base_img.height}])
+    page_margins = layout.get('page_margins', [])
+    per_page = layout.get('version', 1) >= 2
     left_pad = 0
     if base_img.width < TARGET_MIN_WIDTH:
         total_missing = TARGET_MIN_WIDTH - base_img.width
@@ -411,7 +419,11 @@ def compose_label_image(base_img, label, result, hmin,
     total_height = base_img.height + header_height
 
     # Create Canvas
-    final_img = Image.new("RGB", (base_img.width + margin_left, total_height), "white")
+    final_width = (max(TARGET_MIN_WIDTH + margin_left,
+                      max(region['width'] + (page_margins[i] if i < len(page_margins) else margin_left)
+                          for i, region in enumerate(page_regions)))
+                   if per_page else base_img.width + margin_left)
+    final_img = Image.new("RGB", (final_width, total_height), "white")
 
     # Draw Headers
     current_y = 0
@@ -434,7 +446,13 @@ def compose_label_image(base_img, label, result, hmin,
 
     # Paste Base Image
     image_offset_y = current_y
-    final_img.paste(base_img, (margin_left, image_offset_y))
+    if per_page:
+        for i, region in enumerate(page_regions):
+            x, y, w, h = (region[key] for key in ('x', 'y', 'width', 'height'))
+            margin = page_margins[i] if i < len(page_margins) else margin_left
+            final_img.paste(base_img.crop((x, y, x + w, y + h)), (margin, image_offset_y + y))
+    else:
+        final_img.paste(base_img, (margin_left, image_offset_y))
 
     # 2. Draw Local Annotations
     draw = ImageDraw.Draw(final_img, "RGBA") # Refresh draw object
@@ -451,12 +469,24 @@ def compose_label_image(base_img, label, result, hmin,
 
         # Draw Rectangle (if not suppressed)
         if "norectangle" not in fb:
-            draw.rectangle([target_xmin, target_ymin, target_xmax, target_ymax], outline="red", width=3)
+            boxes = [[target_xmin, target_ymin, target_xmax, target_ymax]]
+            if per_page:
+                boxes = []
+                for i, region in enumerate(page_regions):
+                    top = max(ymin - hmin, region['y'])
+                    bottom = min(ymax - hmin, region['y'] + region['height'])
+                    if bottom <= top:
+                        continue
+                    margin = page_margins[i] if i < len(page_margins) else margin_left
+                    shift = margin - region['x']
+                    boxes.append([xmin + shift, top + image_offset_y,
+                                  xmax + shift, bottom + image_offset_y])
+            for box in boxes:
+                draw.rectangle(box, outline="red", width=3)
 
-            if draw_callback:
-                callback_items.append(("local_rect",
-                                       {"box": [target_xmin, target_ymin, target_xmax, target_ymax]},
-                                       {"data": fb, "index": idx}))
+                if draw_callback:
+                    callback_items.append(("local_rect", {"box": box},
+                                           {"data": fb, "index": idx}))
 
         # Render Text
         txt_img = render_fn(fb['text'], width_px=ANNOT_WIDTH,

copienator/commands/cutleft.py

@@ -428,6 +428,11 @@ class ImageReviewer:
                                for group in analysis['groups'])
             margin_info = (f"Auto margin: {offsets} mm from left"
                            + (' | Check uncertain pages' if needs_review else '') + '\n')
+            for page in analysis.get('pages', []):
+                if page.get('status') == 'unknown':
+                    value = page.get('effective_edge_mm')
+                    fallback = f" (copy average: {value:.1f} mm)" if value is not None else ''
+                    margin_info += f"Page {page['page']}: unknown margin{fallback}\n"
             if analysis.get('no_margin_from_page') is not None:
                 margin_info += f"No margin from page {analysis['no_margin_from_page']} onwards\n"
         self.label_info.configure(

copienator/commands/splitting_int.py

@@ -23,6 +23,7 @@ from copienator import (
 )
 from copienator.filesystem import staged_directory
 from copienator.annotation_layout import fingerprint, metadata_path, save_parts
+from copienator.left_margin import margins_with_fallback
 
 SQUARE = 1000 // 38
 ANSWER_TOP_PADDING_POINTS = 4 * 72 / 25.4
@@ -133,7 +134,7 @@ def _render_split_outputs(
         provenance: defaultdict[str, list[dict]] = defaultdict(list)
         schema = read_json(input_pdf.parent.parent / 'Cutleft' / f'{input_pdf.stem}_schema.json', default={})
         analysis = schema.get('auto_margin') or {}
-        measurements = ({item['page']: item for item in analysis.get('pages', [])}
+        measurements = ({item['page']: item for item in margins_with_fallback(analysis.get('pages', []))}
                         if schema.get('source_sha256') == fingerprint(input_pdf) else {})
         with tempfile.TemporaryDirectory(prefix="copienator-split-") as temp_directory:
             temporary = Path(temp_directory)
@@ -206,13 +207,14 @@ def _render_split_outputs(
                         part_path,
                     )
                     parts_by_label[clean_label].append(part_path)
-                    edge = measurements.get(page_number + 1, {}).get('edge_mm')
+                    measurement = measurements.get(page_number + 1, {})
+                    edge = measurement.get('effective_edge_mm')
                     remaining = None if edge is None else max(0, edge - clip.x0 * 25.4 / 72)
                     provenance[clean_label].append({
                         'source': input_pdf.name, 'source_sha256': schema.get('source_sha256'),
                         'page': page_number + 1, 'crop_points': list(clip),
                         'margin_mm': remaining,
-                        'margin_status': ('unknown' if remaining is None else
+                        'margin_status': ('unknown' if measurement.get('edge_mm') is None else
                                           'low' if remaining < 20 else 'normal'),
                     })
 

copienator/configuration.py

@@ -40,7 +40,7 @@ THINKING_LEVEL_FLASH = "medium"
 THINKING_LEVEL_PRO = "high"
 ALWAYS_CROP = False
 ANNOTATION_MARGIN_LEFT = 300
-ANNOTATION_LOW_MARGIN_LEFT = 400
+ANNOTATION_MARGIN_POINTS = [(10, 450), (20, 400), (35, 300)]
 ANNOTATION_MAX_HEIGHT_PX = 33000
 ANNOTATION_MAX_PIXELS = 67_108_864
 SHOW_PERSONAL_STEPS = False

copienator/left_margin.py

@@ -27,6 +27,15 @@ STANDARD_LINE_HEIGHT_MM = 8
 MAX_EXCEPTION_LINES = 4
 
 
+def margins_with_fallback(pages: list[dict]) -> list[dict]:
+    """Keep measurements intact; unknown pages borrow the copy's measured mean."""
+    known = [page['edge_mm'] for page in pages if page.get('edge_mm') is not None]
+    average = sum(known) / len(known) if known else None
+    return [{**page, 'effective_edge_mm': (
+        page['edge_mm'] if page.get('edge_mm') is not None else average
+    )} for page in pages]
+
+
 def trimmed_left_edge(starts: list[float]) -> float | None:
     """Discard the leftmost 10% of occupied bands; retain the next minimum."""
     if not starts:
@@ -142,9 +151,13 @@ def analyse_margins(images: list[Image.Image], render_dpi: int, reference_width:
                                      if name_localized else estimate_page(images[index], render_dpi))
         return page_estimates[index]
 
+    page_estimates = dict(enumerate(margins_with_fallback(
+        [page_estimate(index) for index in range(len(images))])))
+
     for indices in format_groups(images, reference_width):
         estimates = {index: page_estimate(index) for index in indices}
-        values = [item['edge_mm'] for item in estimates.values() if item['edge_mm'] is not None]
+        values = [item['effective_edge_mm'] for item in estimates.values()
+                  if item['effective_edge_mm'] is not None]
         band_starts = [start for item in estimates.values() for start in item['band_starts_mm']]
         status = 'insufficient-ink'
         edge = round(float(np.median(values)), 1) if values else None
@@ -178,10 +191,15 @@ def analyse_margins(images: list[Image.Image], render_dpi: int, reference_width:
         previous_coherent = 0
         for index in indices:
             estimate = page_estimate(index)
-            low = estimate['edge_mm'] is not None and estimate['edge_mm'] < MIN_MARGIN_MM
+            effective_edge = estimate['effective_edge_mm']
+            low = effective_edge is not None and effective_edge < MIN_MARGIN_MM
             tolerated = (status == 'margin' and low and previous_coherent >= 2
                          and short_margin_exception(estimate, edge))
             starts[index] = 0 if low and not tolerated else start
+            if estimate['edge_mm'] is None and status not in {'narrow-page', 'labels-in-margin'}:
+                fallback_left = (max(effective_edge / 2, effective_edge - MARGIN_CLEARANCE_MM)
+                                 if effective_edge is not None and effective_edge >= MIN_MARGIN_MM else 0)
+                starts[index] = round(fallback_left * render_dpi / 25.4)
             estimate['short_margin_exception'] = tolerated
             estimate['decision'] = ('short-exception' if tolerated else
                                     'page-near-edge' if low else status)
@@ -209,7 +227,7 @@ def analyse_margins(images: list[Image.Image], render_dpi: int, reference_width:
                       'status': 'unknown' if edge is None else
                       ('low' if edge < MIN_MARGIN_MM else 'normal')})
 
-    return {'version': 13, 'pages': pages, 'starts_px': starts, 'groups': groups,
+    return {'version': 14, 'pages': pages, 'starts_px': starts, 'groups': groups,
             'no_margin_from_page': None,
             'discard_left_fraction': DISCARD_LEFT_FRACTION, 'margin_clearance_mm': MARGIN_CLEARANCE_MM,
             'cut_fraction_of_margin': .5, 'cut_rule': 'maximum',