miscs

copienator/left_margin.py

@@ -15,7 +15,8 @@ BODY_TOP = .30
 BODY_BOTTOM = .82
 DISCARD_LEFT_FRACTION = .10
 STANDARD_MARGIN_MM = 40
-MIN_MARGIN_MM = STANDARD_MARGIN_MM / 2
+MARGIN_CLEARANCE_MM = 20
+MIN_MARGIN_MM = MARGIN_CLEARANCE_MM
 
 
 def trimmed_left_edge(starts: list[float]) -> float | None:
@@ -89,7 +90,7 @@ 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 per format, trim outliers, and crop at half the edge."""
+    """Pool occupied bands and use the larger half-margin or 20 mm-clearance cut."""
     starts = [0] * len(images)
     groups = []
     page_estimates = {}
@@ -113,7 +114,8 @@ def analyse_margins(images: list[Image.Image], render_dpi: int, reference_width:
         left_mm = 0.0
         if len(values) >= min(2, len(indices)):
             status = 'near-edge' if edge < MIN_MARGIN_MM else 'margin'
-            left_mm = 0.0 if edge < MIN_MARGIN_MM else edge / 2
+            left_mm = (0.0 if edge < MIN_MARGIN_MM else
+                       max(edge / 2, edge - MARGIN_CLEARANCE_MM))
         # 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:
@@ -152,8 +154,9 @@ 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': 6, 'pages': pages, 'starts_px': starts, 'groups': groups,
+    return {'version': 8, 'pages': pages, 'starts_px': starts, 'groups': groups,
             'no_margin_from_page': no_margin_from_page,
-            'discard_left_fraction': DISCARD_LEFT_FRACTION, 'cut_fraction_of_margin': .5,
+            '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]}