Merge pull request #15 from krrome/bug-reduce-min-req-numpy

drop numpy min requirement to 1.19.5. Other minor changes.

Author: krrome

pyproject.toml

@@ -19,9 +19,9 @@ classifiers = [
 ]
 dependencies = [
     "docling>=2.53.0",
-    "numpy>=2.0.2",
+    "numpy>=1.19.5",
     "pymupdf>=1.26.4",
-    "scikit-learn>=1.6.1",
+    "scikit-learn>=1.6.0",
 ]
 
 [project.urls]

tests/samples/sample_document.pdf …samples/sample_document_hierarchical.pdftests/samples/sample_document.pdf renamed to tests/samples/sample_document_hierarchical.pdf tests/samples/sample_document.pdf renamed to tests/samples/sample_document_hierarchical.pdf

@@ -22,7 +22,7 @@ def test_convert():
   3. Conclusion
 """
 
-    source = sample_path / "sample_document.pdf"  # document per local path or URL
+    source = sample_path / "sample_document_hierarchical.pdf"  # document per local path or URL
     converter = DocumentConverter()
     result = converter.convert(source)
     hbm = HierarchyBuilderMetadata(result)

…samples/sample_document_no_bookmarks.pdf …e_document_hierarchical_no_bookmarks.pdftests/samples/sample_document_no_bookmarks.pdf renamed to tests/samples/sample_document_hierarchical_no_bookmarks.pdf tests/samples/sample_document_no_bookmarks.pdf renamed to tests/samples/sample_document_hierarchical_no_bookmarks.pdf

@@ -21,7 +21,7 @@ def compare(res_text, fn):
 
 
 def test_result_postprocessor_textpdf_no_bookmarks():
-    source = sample_path / "sample_document_no_bookmarks.pdf"  # document per local path or URL
+    source = sample_path / "sample_document_hierarchical_no_bookmarks.pdf"  # document per local path or URL
     converter = DocumentConverter()
     result = converter.convert(source)
     ResultPostprocessor(result).process()
@@ -47,7 +47,7 @@ def test_result_postprocessor_textpdf_no_bookmarks():
 
 
 def test_result_postprocessor_textpdf():
-    source = sample_path / "sample_document.pdf"  # document per local path or URL
+    source = sample_path / "sample_document_hierarchical.pdf"  # document per local path or URL
     converter = DocumentConverter()
     result = converter.convert(source)
     ResultPostprocessor(result).process()
@@ -76,7 +76,7 @@ def test_result_postprocessor_textpdf():
 
 
 def test_result_postprocessor_textpdf_stream():
-    source_path = sample_path / "sample_document.pdf"  # document per local path or URL
+    source_path = sample_path / "sample_document_hierarchical.pdf"  # document per local path or URL
     with source_path.open("rb") as fh:
         source = DocumentStream(name=source_path.name, stream=BytesIO(fh.read()))
     converter = DocumentConverter()
@@ -120,7 +120,7 @@ def test_result_postprocessor_textpdf_stream():
 
 
 def test_result_postprocessor_textpdf_string():
-    source_path = sample_path / "sample_document.pdf"  # document per local path or URL
+    source_path = sample_path / "sample_document_hierarchical.pdf"  # document per local path or URL
     source = str(source_path)
     converter = DocumentConverter()
     result = converter.convert(source)

tests/test_metadata_toc.py

@@ -1928,7 +1928,7 @@
 
 
 def compare(res_text, i):
-    p = results_path / f"sample_{i}.txt"
+    p = results_path / f"hierarch_header_sample_{i}.txt"
     if p.exists():
         assert res_text == p.read_text()
     else: