feat: pdf2cad

tests/test_dimension_extractor.py

@@ -0,0 +1,130 @@
+"""Tests for dimension extraction."""
+
+import pytest
+import pymupdf
+from pathlib import Path
+
+from pdf2imos.extract.geometry import extract_geometry
+from pdf2imos.extract.text import extract_text
+from pdf2imos.interpret.title_block import detect_title_block
+from pdf2imos.interpret.view_segmenter import segment_views
+from pdf2imos.interpret.line_classifier import classify_lines
+from pdf2imos.parse.dimensions import extract_dimensions
+from pdf2imos.models import (
+    PageExtraction,
+    ViewType,
+    DimensionAnnotation,
+    DimensionDirection,
+)
+
+
+def make_pipeline(pdf_path):
+    """Run full pipeline up to dimension extraction."""
+    doc = pymupdf.open(str(pdf_path))
+    page = doc[0]
+    page_height = page.rect.height
+
+    geo = extract_geometry(page)
+    texts = extract_text(page)
+    extraction = PageExtraction(
+        paths=geo.paths,
+        texts=tuple(texts),
+        page_width=geo.page_width,
+        page_height=page_height,
+    )
+    _, filtered = detect_title_block(extraction)
+    views = segment_views(filtered)
+
+    return views, page_height
+
+
+class TestExtractDimensions:
+    def test_returns_list(self, simple_panel_pdf):
+        views, page_height = make_pipeline(simple_panel_pdf)
+        if not views:
+            pytest.skip("No views detected")
+        view = views[0]
+        classified = classify_lines(list(view.paths))
+        result = extract_dimensions(view, classified, page_height)
+        assert isinstance(result, list)
+
+    def test_dimension_annotations_type(self, simple_panel_pdf):
+        views, page_height = make_pipeline(simple_panel_pdf)
+        if not views:
+            pytest.skip("No views detected")
+        view = views[0]
+        classified = classify_lines(list(view.paths))
+        result = extract_dimensions(view, classified, page_height)
+        assert all(isinstance(d, DimensionAnnotation) for d in result)
+
+    def test_finds_dimensions_in_largest_view(self, simple_panel_pdf):
+        """The largest view (by text count) should have dimension values."""
+        views, page_height = make_pipeline(simple_panel_pdf)
+        if not views:
+            pytest.skip("No views detected")
+        # Pick the view with the most texts (most likely the main dimensioned view)
+        main_view = max(views, key=lambda v: len(v.texts))
+        if not main_view.texts:
+            pytest.skip("No texts in any view")
+        classified = classify_lines(list(main_view.paths))
+        result = extract_dimensions(main_view, classified, page_height)
+        assert len(result) > 0, (
+            f"No dimensions found in {main_view.view_type.value} view "
+            f"({len(main_view.texts)} texts, {len(main_view.paths)} paths)"
+        )
+
+    def test_dimension_values_reasonable(self, simple_panel_pdf):
+        """Dimension values should be positive and reasonable (1-3000mm range)."""
+        views, page_height = make_pipeline(simple_panel_pdf)
+        for view in views:
+            classified = classify_lines(list(view.paths))
+            dims = extract_dimensions(view, classified, page_height)
+            for d in dims:
+                assert d.value_mm > 0, f"Negative dimension: {d.value_mm}"
+                assert d.value_mm < 10000, f"Unreasonably large dimension: {d.value_mm}"
+
+    def test_direction_is_enum(self, simple_panel_pdf):
+        """Direction field is a DimensionDirection enum value."""
+        views, page_height = make_pipeline(simple_panel_pdf)
+        for view in views:
+            classified = classify_lines(list(view.paths))
+            dims = extract_dimensions(view, classified, page_height)
+            for d in dims:
+                assert isinstance(d.direction, DimensionDirection)
+
+    def test_finds_600mm_or_720mm_dimension(self, simple_panel_pdf):
+        """simple_panel.pdf front view should have 600 or 720mm dimensions."""
+        views, page_height = make_pipeline(simple_panel_pdf)
+        all_dims = []
+        for view in views:
+            classified = classify_lines(list(view.paths))
+            all_dims.extend(extract_dimensions(view, classified, page_height))
+
+        values = {d.value_mm for d in all_dims}
+        # At least one of the main panel dimensions should be found
+        assert any(
+            580 <= v <= 620 or 700 <= v <= 740 or 15 <= v <= 21 for v in values
+        ), f"No expected dimension found in: {sorted(values)}"
+
+    def test_all_fixtures_processable(self, all_fixture_pdfs):
+        """All fixture PDFs process without error."""
+        for pdf_path in all_fixture_pdfs:
+            views, page_height = make_pipeline(pdf_path)
+            for view in views:
+                classified = classify_lines(list(view.paths))
+                dims = extract_dimensions(view, classified, page_height)
+                assert isinstance(dims, list)
+
+    def test_horizontal_vertical_present(self, simple_panel_pdf):
+        """Both H and V dimensions expected in a panel drawing."""
+        views, page_height = make_pipeline(simple_panel_pdf)
+        all_dims = []
+        for view in views:
+            classified = classify_lines(list(view.paths))
+            all_dims.extend(extract_dimensions(view, classified, page_height))
+
+        if not all_dims:
+            pytest.skip("No dimensions extracted")
+        directions = {d.direction for d in all_dims}
+        # Should have at least one direction type
+        assert len(directions) > 0