More fixes to the font-understanding

Signed-off-by: Peter Staar <[email protected]>
DS4SD · Jan 23, 2025 · 9d71602 · 9d71602
1 parent 1d2eece
commit 9d71602
Show file tree

Hide file tree

Showing 2 changed files with 25 additions and 14 deletions.
diff --git a/docling_parse/document.py b/docling_parse/document.py
@@ -153,7 +153,7 @@ class PdfColoredElement(PdfBaseElement):
 class PdfCell(PdfColoredElement):
 
     rect: BoundingRectangle
-    
+
     rect_fontbbox: Optional[BoundingRectangle] = None
     rect_capheight: Optional[BoundingRectangle] = None
 
@@ -366,7 +366,7 @@ def export_to_textlines(
                 line += f"({cell.rect.r_x1:03.02f}, {cell.rect.r_y1:03.02f}) "
                 line += f"({cell.rect.r_x2:03.02f}, {cell.rect.r_y2:03.02f}) "
                 line += f"({cell.rect.r_x3:03.02f}, {cell.rect.r_y3:03.02f}) "
-            
+
             if add_fontkey:
                 line += f"{cell.font_key} "
 

diff --git a/tests/test_parse.py b/tests/test_parse.py
@@ -54,7 +54,7 @@ def verify_bitmap_resources(
 
 
 def verify_cells(
-        true_cells: List[PdfCell], pred_cells: List[PdfCell], eps: float, filename: str
+    true_cells: List[PdfCell], pred_cells: List[PdfCell], eps: float, filename: str
 ) -> bool:
 
     assert len(true_cells) == len(pred_cells), "len(true_cells)==len(pred_cells)"
@@ -67,13 +67,16 @@ def verify_cells(
             true_cell.ordering == pred_cell.ordering
         ), "true_cell.ordering == pred_cell.ordering"
 
-        assert true_cell.text == pred_cell.text, f"true_cell.text == pred_cell.text => {true_cell.text} == {pred_cell.text} for {filename}"
-        assert true_cell.orig == pred_cell.orig, f"true_cell.orig == pred_cell.orig => {true_cell.orig} == {pred_cell.orig} for {filename}"
-
+        assert (
+            true_cell.text == pred_cell.text
+        ), f"true_cell.text == pred_cell.text => {true_cell.text} == {pred_cell.text} for {filename}"
+        assert (
+            true_cell.orig == pred_cell.orig
+        ), f"true_cell.orig == pred_cell.orig => {true_cell.orig} == {pred_cell.orig} for {filename}"
+
         true_rect = true_cell.rect.to_polygon()
         pred_rect = pred_cell.rect.to_polygon()
 
-
         for l in range(0, 4):
             assert (
                 abs(true_rect[l][0] - pred_rect[l][0]) < eps
@@ -82,7 +85,7 @@ def verify_cells(
             assert (
                 abs(true_rect[l][1] - pred_rect[l][1]) < eps
             ), f"abs(true_rect[{l}][1]-pred_rect[{l}][1])<eps -> abs({true_rect[l][1]}-{pred_rect[l][1]})<{eps} for {filename}"
-            
+
         # print("true-text: ", true_cell.text)
         # print("pred-text: ", pred_cell.text)
 
@@ -162,7 +165,9 @@ def verify_lines(
     return True
 
 
-def verify_SegmentedPdfPage(true_page: SegmentedPdfPage, pred_page: SegmentedPdfPage, filename: str):
+def verify_SegmentedPdfPage(
+    true_page: SegmentedPdfPage, pred_page: SegmentedPdfPage, filename: str
+):
 
     eps = max(true_page.dimension.width / 100.0, true_page.dimension.height / 100.0)
 
@@ -175,10 +180,12 @@ def verify_SegmentedPdfPage(true_page: SegmentedPdfPage, pred_page: SegmentedPdf
     verify_lines(true_page.lines, pred_page.lines, eps=eps)
 
 
-def verify_ParsedPdfPage(true_page: ParsedPdfPage, pred_page: ParsedPdfPage):
+def verify_ParsedPdfPage(
+    true_page: ParsedPdfPage, pred_page: ParsedPdfPage, filename: str = ""
+):
 
-    verify_SegmentedPdfPage(true_page.original, pred_page.original)
-    verify_SegmentedPdfPage(true_page.sanitized, pred_page.sanitized)
+    verify_SegmentedPdfPage(true_page.original, pred_page.original, filename=filename)
+    verify_SegmentedPdfPage(true_page.sanitized, pred_page.sanitized, filename=filename)
 
 
 def test_reference_documents_from_filenames():
@@ -216,7 +223,9 @@ def test_reference_documents_from_filenames():
                     print(f"loading from {fname}")
 
                     true_page = SegmentedPdfPage.load_from_json(fname)
-                    verify_SegmentedPdfPage(true_page, pred_page.original, filename=pdf_doc_path)
+                    verify_SegmentedPdfPage(
+                        true_page, pred_page.original, filename=pdf_doc_path
+                    )
 
             if True:
                 rname = os.path.basename(pdf_doc_path)
@@ -230,7 +239,9 @@ def test_reference_documents_from_filenames():
                     print(f"loading from {fname}")
 
                     true_page = SegmentedPdfPage.load_from_json(fname)
-                    verify_SegmentedPdfPage(true_page, pred_page.sanitized, filename=fname)
+                    verify_SegmentedPdfPage(
+                        true_page, pred_page.sanitized, filename=fname
+                    )
 
             pred_page.original.render()
             # res.show()