Initial support for user general remarks

copienator/left_margin.py

@@ -13,6 +13,10 @@ WIDTH_TOLERANCE = .08
 # Start below it with clearance, including on later pages with repeated names.
 BODY_TOP = .30
 BODY_BOTTOM = .82
+LABEL_BOTTOM = .92
+LABEL_LEFT_FRACTION = .05
+COHERENT_MARGIN_TOLERANCE_MM = 10
+LABEL_MARGIN_GAP_MM = 10
 DISCARD_LEFT_FRACTION = .10
 STANDARD_MARGIN_MM = 40
 MARGIN_CLEARANCE_MM = 20
@@ -47,21 +51,28 @@ def handwriting_mask(rgb: np.ndarray) -> np.ndarray:
     return keep[labels] & diagnostics['masks']['foreground']
 
 
-def body_edge(mask: np.ndarray) -> dict:
-    """Estimate a left edge from occupied 4 mm bands within 30–82% height."""
-    height = mask.shape[0]
+def occupied_band_starts(mask: np.ndarray, bottom_fraction: float) -> list[float]:
+    """Ignore empty bands and writing starting beyond the page midpoint."""
+    height, width = mask.shape
     band_height = max(1, round(4 * ANALYSIS_DPI / 25.4))
-    bottom = round(BODY_BOTTOM * height)
+    bottom = round(bottom_fraction * height)
     starts = []
     for y in range(round(BODY_TOP * height), bottom, band_height):
         band = mask[y:min(y + band_height, bottom)]
         xs = np.flatnonzero(band.sum(axis=0) >= 2)
-        if band.sum() >= 40 and len(xs) >= 12:
+        if band.sum() >= 40 and len(xs) >= 12 and xs[0] <= width / 2:
             starts.append(float(xs[0]) * 25.4 / ANALYSIS_DPI)
+    return starts
+
+
+def body_edge(mask: np.ndarray) -> dict:
+    """Keep the body decile; collect a wider 30–92% label-search window."""
+    starts = occupied_band_starts(mask, BODY_BOTTOM)
     return {
         'edge_mm': trimmed_left_edge(starts) if len(starts) >= 6 else None,
         'bands': len(starts),
         'band_starts_mm': starts,
+        'label_band_starts_mm': occupied_band_starts(mask, LABEL_BOTTOM),
     }
 
 
@@ -90,7 +101,7 @@ def format_groups(images: list[Image.Image], reference_width: float) -> list[lis
 
 
 def analyse_margins(images: list[Image.Image], render_dpi: int, reference_width: float) -> dict:
-    """Pool occupied bands and use the larger half-margin or 20 mm-clearance cut."""
+    """Estimate body margins, then retain left labels supported across coherent pages."""
     starts = [0] * len(images)
     groups = []
     page_estimates = {}
@@ -116,6 +127,24 @@ def analyse_margins(images: list[Image.Image], render_dpi: int, reference_width:
             status = 'near-edge' if edge < MIN_MARGIN_MM else 'margin'
             left_mm = (0.0 if edge < MIN_MARGIN_MM else
                        max(edge / 2, edge - MARGIN_CLEARANCE_MM))
+        # Use ALL coherent pages, not just the four body-margin samples.
+        # A sparse/low/outlying page cannot supply evidence for this format's
+        # established body margin. Its normal fallback behaviour remains below.
+        coherent = [index for index in indices
+                    if edge is not None and page_estimate(index)['edge_mm'] is not None
+                    and page_estimate(index)['edge_mm'] >= MIN_MARGIN_MM
+                    and abs(page_estimate(index)['edge_mm'] - edge) <= COHERENT_MARGIN_TOLERANCE_MM]
+        label_starts = [start for index in coherent for start in
+                        page_estimate(index).get('label_band_starts_mm',
+                                                 page_estimate(index)['band_starts_mm'])]
+        label_edge = (round(sorted(label_starts)[int(len(label_starts) * LABEL_LEFT_FRACTION)], 1)
+                      if label_starts else None)
+        labels_in_margin = (status == 'margin' and len(coherent) >= min(2, len(indices))
+                            and label_edge is not None and edge - label_edge >= LABEL_MARGIN_GAP_MM)
+        body_left_mm = left_mm
+        if labels_in_margin:
+            status = 'labels-in-margin'
+            left_mm = 0.0
         # Secondary subpages can start in the middle of an answer. Preserve
         # their left edge even when the only visible writing is indented.
         if images[indices[0]].width < .8 * reference_width:
@@ -129,6 +158,11 @@ def analyse_margins(images: list[Image.Image], render_dpi: int, reference_width:
             'status': status,
             'edge_mm': edge,
             'left_mm': round(left_mm, 1),
+            'body_left_mm': round(body_left_mm, 1),
+            'labels_in_margin': labels_in_margin,
+            'coherent_pages': [index + 1 for index in coherent],
+            'label_edge_mm': label_edge,
+            'label_bands': len(label_starts),
             'bands': len(band_starts),
             'discarded_bands': int(len(band_starts) * DISCARD_LEFT_FRACTION),
             'samples': [{'page': index + 1, **item} for index, item in estimates.items()],
@@ -154,9 +188,14 @@ 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': 8, 'pages': pages, 'starts_px': starts, 'groups': groups,
+    return {'version': 9, 'pages': pages, 'starts_px': starts, 'groups': groups,
             'no_margin_from_page': no_margin_from_page,
             'discard_left_fraction': DISCARD_LEFT_FRACTION, 'margin_clearance_mm': MARGIN_CLEARANCE_MM,
             'cut_fraction_of_margin': .5, 'cut_rule': 'maximum',
             'standard_margin_mm': STANDARD_MARGIN_MM, 'minimum_margin_mm': MIN_MARGIN_MM,
-            'body_height_fraction': [BODY_TOP, BODY_BOTTOM]}
+            'body_height_fraction': [BODY_TOP, BODY_BOTTOM],
+            'label_height_fraction': [BODY_TOP, LABEL_BOTTOM],
+            'label_left_fraction': LABEL_LEFT_FRACTION,
+            'coherent_margin_tolerance_mm': COHERENT_MARGIN_TOLERANCE_MM,
+            'label_margin_gap_mm': LABEL_MARGIN_GAP_MM,
+            'maximum_band_start_width_fraction': .5}