Misc fixes

VikParuchuri · Nov 19, 2024 · a72a508 · a72a508
1 parent 7e449a1
commit a72a508
Show file tree

Hide file tree

Showing 3 changed files with 3 additions and 2 deletions.
diff --git a/marker/v2/converters/pdf.py b/marker/v2/converters/pdf.py
@@ -97,6 +97,7 @@ def main(fpath: str, output_dir: str, debug: bool, output_format: str, pages: st
         config["debug_pdf_images"] = True
         config["debug_layout_images"] = True
         config["debug_json"] = True
+        config["debug_data_folder"] = output_dir
     if force_ocr:
         config["force_ocr"] = True
 

diff --git a/marker/v2/schema/blocks/listitem.py b/marker/v2/schema/blocks/listitem.py
@@ -11,7 +11,7 @@ def replace_bullets(child_blocks):
         first_block = child_blocks[0]
         child_blocks = first_block.children
 
-    if first_block.id.block_type == BlockTypes.Line:
+    if first_block is not None and first_block.id.block_type == BlockTypes.Line:
         bullet_pattern = r"(^|[\n ]|<[^>]*>)[•●○■▪▫–—-]( )"
         first_block.html = re.sub(bullet_pattern, r"\1\2", first_block.html)
 

diff --git a/tests/processors/test_document_toc_processor.py b/tests/processors/test_document_toc_processor.py
@@ -9,4 +9,4 @@ def test_table_processor(pdf_document, detection_model, recognition_model, table
     processor(pdf_document)
 
     assert len(pdf_document.table_of_contents) == 3
-    assert pdf_document.table_of_contents[0].text == "Subspace Adversarial Training"
+    assert pdf_document.table_of_contents[0]["text"] == "Subspace Adversarial Training"