Initial support for user general remarks

tests/test_left_margin.py

@@ -118,6 +118,81 @@ class LeftMarginTests(unittest.TestCase):
         mask[300:960] = False
         self.assertIsNone(left_margin.body_edge(mask)['edge_mm'])
 
+    def test_bands_starting_beyond_midpoint_are_empty_for_both_windows(self):
+        mask = np.zeros((1169, 827), bool)
+        for y in range(370, 1050, 35):
+            mask[y:y + 5, 450:550] = True
+        result = left_margin.body_edge(mask)
+        self.assertIsNone(result['edge_mm'])
+        self.assertEqual(result['band_starts_mm'], [])
+        self.assertEqual(result['label_band_starts_mm'], [])
+        mask[400:410, 100:200] = True
+        self.assertTrue(left_margin.body_edge(mask)['band_starts_mm'])
+
+    def test_label_window_keeps_lower_body_but_excludes_names_and_bottom_eight_percent(self):
+        mask = np.zeros((1169, 827), bool)
+        mask[100:300, 5:80] = True  # Name/header; exclude from both pools.
+        mask[1100:1150, 5:80] = True  # Footer; below 92%.
+        for y in range(370, 920, 35):
+            mask[y:y + 5, 170:270] = True
+        mask[1000:1010, 40:140] = True  # Label at 85.5%; new window only.
+        result = left_margin.body_edge(mask)
+        self.assertGreater(min(result['band_starts_mm']), 40)
+        self.assertAlmostEqual(min(result['label_band_starts_mm']), 10.16)
+
+    def test_fifth_percentile_detects_labels_while_page_decile_stays_at_body(self):
+        images = [Image.new('RGB', (1654, 200)) for _ in range(6)]
+        bands = [5.] * 7 + [42.] * 93
+        estimate = {'edge_mm': left_margin.trimmed_left_edge(bands),
+                    'bands': len(bands), 'band_starts_mm': bands,
+                    'label_band_starts_mm': bands}
+        with patch.object(left_margin, 'estimate_page', return_value=estimate):
+            result = left_margin.analyse_margins(images, 200, 1654)
+        group = result['groups'][0]
+        self.assertEqual(group['edge_mm'], 42)
+        self.assertEqual(group['body_left_mm'], 22)
+        self.assertEqual(group['label_edge_mm'], 5)
+        self.assertTrue(group['labels_in_margin'])
+        self.assertEqual(result['starts_px'], [0] * 6)
+
+    def test_unsampled_coherent_page_can_move_the_whole_group_left(self):
+        images = [Image.new('RGB', (1654, 200)) for _ in range(6)]
+        def estimate(image, _dpi):
+            labels = [5.] * 20 if image is images[2] else []
+            return {'edge_mm': 42, 'bands': 12, 'band_starts_mm': [42.] * 12,
+                    'label_band_starts_mm': labels + [42.] * 12}
+        with patch.object(left_margin, 'estimate_page', side_effect=estimate):
+            result = left_margin.analyse_margins(images, 200, 1654)
+        group = result['groups'][0]
+        self.assertNotIn(3, [p['page'] for p in group['samples']])
+        self.assertIn(3, group['coherent_pages'])
+        self.assertEqual(result['starts_px'], [0] * 6)
+
+    def test_sparse_or_incoherent_pages_do_not_supply_label_evidence(self):
+        images = [Image.new('RGB', (1654, 200)) for _ in range(6)]
+        edges = [42, 42, None, 42, 65, 42]
+        def estimate(image, _dpi):
+            edge = edges[next(i for i, candidate in enumerate(images) if candidate is image)]
+            return {'edge_mm': edge, 'bands': 12, 'band_starts_mm': [edge or 42.] * 12,
+                    'label_band_starts_mm': [1.] * 20 if edge in (None, 65) else [42.] * 12}
+        with patch.object(left_margin, 'estimate_page', side_effect=estimate):
+            result = left_margin.analyse_margins(images, 200, 1654)
+        group = result['groups'][0]
+        self.assertEqual(group['coherent_pages'], [1, 2, 4, 6])
+        self.assertFalse(group['labels_in_margin'])
+        self.assertGreater(result['starts_px'][0], 0)
+
+    def test_less_than_five_percent_noise_or_small_gap_does_not_trigger_labels(self):
+        for bands in ([1.] * 4 + [42.] * 96, [35.] * 7 + [42.] * 93):
+            with self.subTest(bands=bands[:8]):
+                images = [Image.new('RGB', (1654, 200)) for _ in range(2)]
+                estimate = {'edge_mm': 42, 'bands': 100, 'band_starts_mm': [42.] * 100,
+                            'label_band_starts_mm': bands}
+                with patch.object(left_margin, 'estimate_page', return_value=estimate):
+                    result = left_margin.analyse_margins(images, 200, 1654)
+                self.assertFalse(result['groups'][0]['labels_in_margin'])
+                self.assertGreater(result['starts_px'][0], 0)
+
     def test_real_mask_rejects_ruling_and_preserves_handwriting(self):
         image = Image.new('RGB', (827, 1169), 'white')
         draw = ImageDraw.Draw(image)