Merge branch 'feat/document-engine' into feat/degradations

Author: hBouanane

src/ocrsmith/core/__init__.py

@@ -1,7 +1,7 @@
 from .augmentation import NoiseAugmentation
 from .AugmentationManager import AugmentationManager
 from .BackgroundManager import BackgroundManager
+from .engine import OCRSmithEngine
 from .FontManager import FontManager
-from .OCRSmithEngine import OCRSmithEngine
 from .TextPlacementManager import TextPlacementManager
 from .TextRenderingManager import TextRenderingManager

src/ocrsmith/core/app.py

@@ -1,7 +1,7 @@
 import argparse
 
 from ocrsmith.config.loader import load_config
-from ocrsmith.core.OCRSmithEngine import OCRSmithEngine
+from ocrsmith.core.engine import OCRSmithEngine
 
 
 def parse_overrides(pairs):

src/ocrsmith/core/OCRSmithEngine.py src/ocrsmith/core/engine.pysrc/ocrsmith/core/OCRSmithEngine.py renamed to src/ocrsmith/core/engine.py

@@ -253,15 +253,27 @@ def test_no_text_is_lost_across_page_breaks(self, renderer, typography, rng, lat
         # Bullet markers and page furniture can add tokens; nothing may go missing.
         assert rendered_words >= expected
 
-    def test_two_columns_hold_more_per_page_than_one(self, renderer, typography, rng, latin_source):
+    def test_columns_are_narrower_and_both_get_used(self, renderer, typography, rng, latin_source):
+        # Not "two columns need fewer pages": narrower columns wrap more, and with orphan
+        # control a two-column layout can legitimately need one page more. What the
+        # feature actually guarantees is that blocks are confined to a column and that a
+        # page with enough content uses both of them.
         content = ArticleTemplate(min_sections=5, max_sections=5).build(latin_source, rng)
-        one = PageSpec.from_paper("a5", dpi=100, columns=1)
-        two = PageSpec.from_paper("a5", dpi=100, columns=2)
+        spec = PageSpec.from_paper("a5", dpi=100, columns=2)
 
-        single = renderer.render(content, one, typography, rng=random.Random(1), max_pages=40)
-        double = renderer.render(content, two, typography, rng=random.Random(1), max_pages=40)
+        pages = renderer.render(content, spec, typography, rng=random.Random(1), max_pages=40)
 
-        assert len(double) <= len(single)
+        column_width = spec.column_width
+        midpoint = spec.content_box.x0 + spec.content_box.width / 2
+        used_left = used_right = False
+        for page in pages:
+            for region in page.page.regions:
+                assert region.bbox.width <= column_width + 2
+                if region.bbox.center[0] < midpoint:
+                    used_left = True
+                else:
+                    used_right = True
+        assert used_left and used_right
 
     def test_right_to_left_documents_fill_the_right_column_first(
         self, renderer, typography, rng, arabic_source

tests/test_documents_layout.py

@@ -9,7 +9,7 @@
 
 # --- Import all necessary classes ---
 from ocrsmith.config.schema import AppConfig
-from ocrsmith.core.OCRSmithEngine import OCRSmithEngine
+from ocrsmith.core.engine import OCRSmithEngine
 from ocrsmith.core.text_placement import PlacementResult
 
 # --- Fixtures ---
@@ -66,12 +66,12 @@ def temp_dir():
 
 # Patch all managers at the class level to ensure they are mocked for every test.
 # This prevents real managers from being initialized with mock configs.
-@patch("ocrsmith.core.OCRSmithEngine.BackgroundManager")
-@patch("ocrsmith.core.OCRSmithEngine.TextRenderingManager")
-@patch("ocrsmith.core.OCRSmithEngine.TextPlacementManager")
-@patch("ocrsmith.core.OCRSmithEngine.AugmentationManager")
-@patch("ocrsmith.core.OCRSmithEngine.FontManager")
-@patch("ocrsmith.core.OCRSmithEngine.TextDataManager")
+@patch("ocrsmith.core.engine.BackgroundManager")
+@patch("ocrsmith.core.engine.TextRenderingManager")
+@patch("ocrsmith.core.engine.TextPlacementManager")
+@patch("ocrsmith.core.engine.AugmentationManager")
+@patch("ocrsmith.core.engine.FontManager")
+@patch("ocrsmith.core.engine.TextDataManager")
 class TestOCRSmithEngine:
     """Tests for the main OCRSmithEngine class with all dependencies mocked."""
 

tests/test_engine.py

@@ -3,7 +3,7 @@
 import pytest
 
 from ocrsmith.config.schema import AppConfig
-from ocrsmith.core.OCRSmithEngine import OCRSmithEngine
+from ocrsmith.core.engine import OCRSmithEngine
 
 
 @pytest.fixture