Fix remaining tests

Author: VikParuchuri

marker/builders/line.py

@@ -53,7 +53,7 @@ class LineBuilder(BaseBuilder):
     provider_line_provider_line_min_overlap_pct: Annotated[
         float,
         "The percentage of a provider line that has to be covered by a detected line",
-    ] = 0.15
+    ] = 0.1
     excluded_for_coverage: Annotated[
         Tuple[BlockTypes],
         "A list of block types to exclude from the layout coverage check.",

tests/builders/test_document_builder.py

@@ -38,6 +38,5 @@ def test_document_builder_inline_eq(pdf_document):
 
     first_span = first_page.get_block(first_text_block.structure[0])
     assert first_span.block_type == BlockTypes.Span
-    assert first_span.text == "Subspace Adversarial Training"
-    assert first_span.font == "NimbusRomNo9L-Medi"
-    assert first_span.formats == ["plain"]
+    assert first_span.text.strip() == "Subspace Adversarial Training"
+    assert "bold" in first_span.formats

tests/builders/test_pdf_links.py

@@ -11,6 +11,7 @@
 
 @pytest.mark.filename("arxiv_test.pdf")
 @pytest.mark.output_format("markdown")
+@pytest.mark.config({"disable_ocr": True})
 def test_pdf_links(pdf_document: Document, config, renderer, model_dict, temp_doc):
     first_page = pdf_document.pages[1]
 
@@ -19,26 +20,35 @@ def test_pdf_links(pdf_document: Document, config, renderer, model_dict, temp_do
         artifact_dict=model_dict,
         processor_list=processors,
         renderer=classes_to_strings([renderer])[0],
-        config=config
+        config=config,
     )
 
-    for section_header_span in first_page.contained_blocks(pdf_document, (BlockTypes.Span,)):
+    for section_header_span in first_page.contained_blocks(
+        pdf_document, (BlockTypes.Span,)
+    ):
         if "II." in section_header_span.text:
             assert section_header_span.url == "#page-1-0"
             break
     else:
         raise ValueError("Could not find II. in the first page")
 
-    section_header_block = first_page.contained_blocks(pdf_document, (BlockTypes.SectionHeader,))[0]
-    assert section_header_block.raw_text(pdf_document) == 'II. THEORETICAL FRAMEWORK\n'
+    section_header_block = first_page.contained_blocks(
+        pdf_document, (BlockTypes.SectionHeader,)
+    )[0]
+    assert section_header_block.raw_text(pdf_document) == "II. THEORETICAL FRAMEWORK\n"
 
     assert first_page.refs[0].ref == "page-1-0"
 
     markdown_output: MarkdownOutput = pdf_converter(temp_doc.name)
     markdown = markdown_output.markdown
 
-    assert '[II.](#page-1-0)' in markdown
+    assert "[II.](#page-1-0)" in markdown
     assert '<span id="page-1-0"></span>II. THEORETICAL FRAMEWORK' in markdown
 
-    for ref in set([f'<span id="page-{m[0]}-{m[1]}">' for m in re.findall(r'\]\(#page-(\d+)-(\d+)\)', markdown)]):
+    for ref in set(
+        [
+            f'<span id="page-{m[0]}-{m[1]}">'
+            for m in re.findall(r"\]\(#page-(\d+)-(\d+)\)", markdown)
+        ]
+    ):
         assert ref in markdown, f"Reference {ref} not found in markdown"

tests/builders/test_rotated_bboxes.py

@@ -13,7 +13,7 @@ def test_rotated_bboxes(pdf_document):
     text_blocks = first_page.contained_blocks(
         pdf_document, (BlockTypes.Text, BlockTypes.TextInlineMath)
     )
-    assert len(text_lines) == 85
+    assert len(text_lines) == 84
 
     # Ensure the bbox sizes match up
     max_line_position = max([line.polygon.x_end for line in text_lines])