assertions for all segmentation tests

OCR-D · Apr 30, 2024 · 10a21ff · 10a21ff
1 parent b70e864
commit 10a21ff
Show file tree

Hide file tree

Showing 3 changed files with 56 additions and 0 deletions.
diff --git a/test/test_segment_line.py b/test/test_segment_line.py
@@ -1,6 +1,8 @@
 from ocrd_tesserocr import TesserocrSegmentRegion
 from ocrd_tesserocr import TesserocrSegmentLine
 from ocrd_tesserocr import TesserocrSegment
+from ocrd_modelfactory import page_from_file
+from ocrd_utils import MIMETYPE_PAGE
 
 def test_run_modular(workspace_herold_small):
     TesserocrSegmentRegion(
@@ -14,6 +16,13 @@ def test_run_modular(workspace_herold_small):
         input_file_grp="OCR-D-SEG-BLOCK",
         output_file_grp="OCR-D-SEG-LINE"
     ).process()
+    out_files = list(workspace_herold_small.find_files(
+        fileGrp="OCR-D-SEG-LINE", pageId="PHYS_0001", mimetype=MIMETYPE_PAGE))
+    assert len(out_files)
+    out_pcgts = page_from_file(out_files[0])
+    assert out_pcgts is not None
+    out_lines = out_pcgts.get_Page().get_AllTextLines()
+    assert len(out_lines)
     workspace_herold_small.save_mets()
 
 def test_run_allinone(workspace_herold_small):
@@ -22,4 +31,11 @@ def test_run_allinone(workspace_herold_small):
         input_file_grp="OCR-D-IMG",
         output_file_grp="OCR-D-SEG"
     ).process()
+    out_files = list(workspace_herold_small.find_files(
+        fileGrp="OCR-D-SEG", pageId="PHYS_0001", mimetype=MIMETYPE_PAGE))
+    assert len(out_files)
+    out_pcgts = page_from_file(out_files[0])
+    assert out_pcgts is not None
+    out_lines = out_pcgts.get_Page().get_AllTextLines()
+    assert len(out_lines)
     workspace_herold_small.save_mets()
diff --git a/test/test_segment_region.py b/test/test_segment_region.py
@@ -1,11 +1,20 @@
 from ocrd_tesserocr import TesserocrSegmentRegion
+from ocrd_modelfactory import page_from_file
+from ocrd_utils import MIMETYPE_PAGE
 
 def test_run(workspace_herold_small):
     TesserocrSegmentRegion(
         workspace_herold_small,
         input_file_grp="OCR-D-IMG",
         output_file_grp="OCR-D-SEG-BLOCK"
     ).process()
+    out_files = list(workspace_herold_small.find_files(
+        fileGrp="OCR-D-SEG-BLOCK", pageId="PHYS_0001", mimetype=MIMETYPE_PAGE))
+    assert len(out_files)
+    out_pcgts = page_from_file(out_files[0])
+    assert out_pcgts is not None
+    out_blocks = out_pcgts.get_Page().get_AllRegions(classes=['Text'])
+    assert len(out_blocks)
     workspace_herold_small.save_mets()
 
 def test_run_shrink(workspace_herold_small):
@@ -15,6 +24,13 @@ def test_run_shrink(workspace_herold_small):
         output_file_grp="OCR-D-SEG-BLOCK",
         parameter={'shrink_polygons': True}
     ).process()
+    out_files = list(workspace_herold_small.find_files(
+        fileGrp="OCR-D-SEG-BLOCK", pageId="PHYS_0001", mimetype=MIMETYPE_PAGE))
+    assert len(out_files)
+    out_pcgts = page_from_file(out_files[0])
+    assert out_pcgts is not None
+    out_blocks = out_pcgts.get_Page().get_AllRegions(classes=['Text'])
+    assert len(out_blocks)
     workspace_herold_small.save_mets()
 
 def test_run_sparse(workspace_herold_small):
@@ -24,6 +40,13 @@ def test_run_sparse(workspace_herold_small):
         output_file_grp="OCR-D-SEG-BLOCK",
         parameter={'sparse_text': True}
     ).process()
+    out_files = list(workspace_herold_small.find_files(
+        fileGrp="OCR-D-SEG-BLOCK", pageId="PHYS_0001", mimetype=MIMETYPE_PAGE))
+    assert len(out_files)
+    out_pcgts = page_from_file(out_files[0])
+    assert out_pcgts is not None
+    out_blocks = out_pcgts.get_Page().get_AllRegions(classes=['Text'])
+    assert len(out_blocks)
     workspace_herold_small.save_mets()
 
 def test_run_staves(workspace_herold_small):
@@ -33,4 +56,11 @@ def test_run_staves(workspace_herold_small):
         output_file_grp="OCR-D-SEG-BLOCK",
         parameter={'find_staves': True, 'find_tables': False}
     ).process()
+    out_files = list(workspace_herold_small.find_files(
+        fileGrp="OCR-D-SEG-BLOCK", pageId="PHYS_0001", mimetype=MIMETYPE_PAGE))
+    assert len(out_files)
+    out_pcgts = page_from_file(out_files[0])
+    assert out_pcgts is not None
+    out_blocks = out_pcgts.get_Page().get_AllRegions(classes=['Text'])
+    assert len(out_blocks)
     workspace_herold_small.save_mets()
diff --git a/test/test_segment_word.py b/test/test_segment_word.py
@@ -1,6 +1,8 @@
 from ocrd_tesserocr import TesserocrSegmentRegion
 from ocrd_tesserocr import TesserocrSegmentLine
 from ocrd_tesserocr import TesserocrSegmentWord
+from ocrd_modelfactory import page_from_file
+from ocrd_utils import MIMETYPE_PAGE
 
 def test_run_modular(workspace_kant_binarized):
     TesserocrSegmentRegion(
@@ -18,4 +20,12 @@ def test_run_modular(workspace_kant_binarized):
         input_file_grp="OCR-D-SEG-LINE",
         output_file_grp="OCR-D-SEG-WORD"
     ).process()
+    out_files = list(workspace_kant_binarized.find_files(
+        fileGrp="OCR-D-SEG-WORD", pageId="P_0017", mimetype=MIMETYPE_PAGE))
+    assert len(out_files)
+    out_pcgts = page_from_file(out_files[0])
+    assert out_pcgts is not None
+    out_lines = out_pcgts.get_Page().get_AllTextLines()
+    assert len(out_lines)
+    assert all(len(line.get_Word()) for line in out_lines)
     workspace_kant_binarized.save_mets()