html-to-markdown 2.1.2__cp310-abi3-win_amd64.whl → 2.2.0__cp310-abi3-win_amd64.whl

html_to_markdown/__init__.py

@@ -39,4 +39,4 @@ __all__ = [
     "markdownify",
 ]
 
-__version__ = "2.1.1"
+__version__ = "2.2.0"

html_to_markdown/_rust.pyi

@@ -11,6 +11,7 @@ class ConversionOptions:
     autolinks: bool
     default_title: bool
     br_in_tables: bool
+    hocr_spatial_tables: bool
     highlight_style: str
     extract_metadata: bool
     whitespace_mode: str
@@ -39,6 +40,7 @@ class ConversionOptions:
         autolinks: bool = True,
         default_title: bool = False,
         br_in_tables: bool = False,
+        hocr_spatial_tables: bool = True,
         highlight_style: str = "double-equal",
         extract_metadata: bool = True,
         whitespace_mode: str = "normalized",

html_to_markdown/api.py

@@ -51,6 +51,7 @@ def convert(
         autolinks=options.autolinks,
         default_title=options.default_title,
         br_in_tables=options.br_in_tables,
+        hocr_spatial_tables=options.hocr_spatial_tables,
         highlight_style=options.highlight_style,
         extract_metadata=options.extract_metadata,
         whitespace_mode=options.whitespace_mode,

html_to_markdown/options.py

@@ -71,6 +71,9 @@ class ConversionOptions:
     br_in_tables: bool = False
     """Use <br> tags for line breaks in table cells instead of spaces."""
 
+    hocr_spatial_tables: bool = True
+    """Reconstruct tables in hOCR documents using spatial heuristics."""
+
     highlight_style: Literal["double-equal", "html", "bold"] = "double-equal"
     """Style for highlighting <mark> elements."""
 

html_to_markdown/v1_compat.py

@@ -120,8 +120,11 @@ def convert_to_markdown(
     if custom_converters is not None:
         raise NotImplementedError("custom_converters is not yet implemented in v2")
     if not hocr_extract_tables:
-        raise NotImplementedError(
-            "hocr_extract_tables toggle was removed in v2. hOCR tables are always reconstructed when detected."
+        warnings.warn(
+            "hocr_extract_tables is deprecated and will be removed in a future release. "
+            "Use ConversionOptions(hocr_spatial_tables=False) to disable spatial table reconstruction.",
+            DeprecationWarning,
+            stacklevel=2,
         )
     if hocr_table_column_threshold != 50 or hocr_table_row_threshold_ratio != 0.5:
         raise NotImplementedError(
@@ -146,6 +149,7 @@ def convert_to_markdown(
         autolinks=autolinks,
         default_title=default_title,
         br_in_tables=br_in_tables,
+        hocr_spatial_tables=hocr_extract_tables,
         highlight_style=highlight_style,  # type: ignore[arg-type]
         extract_metadata=extract_metadata,
         whitespace_mode=whitespace_mode,  # type: ignore[arg-type]

{html_to_markdown-2.1.2.dist-info → html_to_markdown-2.2.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: html-to-markdown
-Version: 2.1.2
+Version: 2.2.0
 Classifier: Development Status :: 5 - Production/Stable
 Classifier: Environment :: Console
 Classifier: Intended Audience :: Developers

html_to_markdown-2.2.0.dist-info/RECORD

@@ -0,0 +1,17 @@
+html_to_markdown-2.2.0.data/scripts/html-to-markdown.exe,sha256=aMAtVnodlg1vMH3l9sQOHHpRJN6c9D0nEnBqjsjz0Cg,4465664
+html_to_markdown-2.2.0.dist-info/METADATA,sha256=dpiwXABx23YyxSxDe9hiHcspTwNLtCO35TMXLt7KAYQ,7233
+html_to_markdown-2.2.0.dist-info/WHEEL,sha256=4EDp_7DiFfWl1yYv5M4wSosAn5L_xgD1dyrQxQxfCx8,95
+html_to_markdown-2.2.0.dist-info/licenses/LICENSE,sha256=QhKFMkQLa4mSUlOsyG9VElzC7GYbAKtiS_EwOCyH-b4,1107
+html_to_markdown/__init__.py,sha256=qNK0cpaGcb8H2zl9BFO9sG_RQs4hMp_GrihUuIHqV10,1191
+html_to_markdown/__main__.py,sha256=5objj9lB7hhpSpZsDok5tv9o9yztVR63Ccww-pXsAyY,343
+html_to_markdown/_html_to_markdown.pyd,sha256=44hglf_qiVnD8-bCpsScL_qxKiWRlNjl1ai_Hpko26U,4211200
+html_to_markdown/_rust.pyi,sha256=JP8tvcjYDfFJeJkbLpQ4qeK-5jl0hzIVT3Sa0daTkyo,2171
+html_to_markdown/api.py,sha256=U7-Tu8TaVa32vveCtiOhTwoEojklkDV2e-6ItAiP3d4,2858
+html_to_markdown/bin/html-to-markdown.exe,sha256=aMAtVnodlg1vMH3l9sQOHHpRJN6c9D0nEnBqjsjz0Cg,4465664
+html_to_markdown/cli.py,sha256=z59l8sF8wIRRzJtUd-tXgqiC0WTqkTjzl-df8Ey_oQ0,67
+html_to_markdown/cli_proxy.py,sha256=J2Qk9MnnkFKIroxc0wn79nzI0dXqXDDNEAF9o9hth9Y,3829
+html_to_markdown/exceptions.py,sha256=31VqpPi4JLGv7lI2481Z4f2s5ejYmq97c3s-WFFkXVU,2443
+html_to_markdown/options.py,sha256=jna7fx9bHhx8N7u5IYtMXganFFzdJSVVgLZW0tYk3GA,5054
+html_to_markdown/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+html_to_markdown/v1_compat.py,sha256=aVt9cVTBfYcrS8EfBsrC6HQwWc3Kz9-65-LB9foN6Jk,8227
+html_to_markdown-2.2.0.dist-info/RECORD,,

html_to_markdown-2.1.2.dist-info/RECORD

@@ -1,17 +0,0 @@
-html_to_markdown-2.1.2.data/scripts/html-to-markdown.exe,sha256=SJCBlZp0uTo6_wzS7KEgXariZtABUVc64o5TVxOwVZo,4414976
-html_to_markdown-2.1.2.dist-info/METADATA,sha256=TgFH9djK4HzJF_vDFVZCm7EDXYscA4v9t31DuXCujIE,7233
-html_to_markdown-2.1.2.dist-info/WHEEL,sha256=4EDp_7DiFfWl1yYv5M4wSosAn5L_xgD1dyrQxQxfCx8,95
-html_to_markdown-2.1.2.dist-info/licenses/LICENSE,sha256=QhKFMkQLa4mSUlOsyG9VElzC7GYbAKtiS_EwOCyH-b4,1107
-html_to_markdown/__init__.py,sha256=3_Egcf46oNcEam7rc7zAHx8lfOj1eVNO1p0kErVf_fs,1191
-html_to_markdown/__main__.py,sha256=5objj9lB7hhpSpZsDok5tv9o9yztVR63Ccww-pXsAyY,343
-html_to_markdown/_html_to_markdown.pyd,sha256=ES7QEe9lTb2ZK3yvC2-vNHng__U7HB3CY5p2wJ0IuNQ,4159488
-html_to_markdown/_rust.pyi,sha256=SHrrT8opJd5kcRYycooR4AS9is5tr1beSGtpoUWqzNc,2097
-html_to_markdown/api.py,sha256=YQQuJoO1OQnXpuOLk8TbdQDTARcKYFbf_zSA44BeHCM,2800
-html_to_markdown/bin/html-to-markdown.exe,sha256=SJCBlZp0uTo6_wzS7KEgXariZtABUVc64o5TVxOwVZo,4414976
-html_to_markdown/cli.py,sha256=z59l8sF8wIRRzJtUd-tXgqiC0WTqkTjzl-df8Ey_oQ0,67
-html_to_markdown/cli_proxy.py,sha256=J2Qk9MnnkFKIroxc0wn79nzI0dXqXDDNEAF9o9hth9Y,3829
-html_to_markdown/exceptions.py,sha256=31VqpPi4JLGv7lI2481Z4f2s5ejYmq97c3s-WFFkXVU,2443
-html_to_markdown/options.py,sha256=ijjRBTwrbESbwmYTOXV_ZO1A1GAmOzzILiFoPeC-jZk,4940
-html_to_markdown/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-html_to_markdown/v1_compat.py,sha256=Lb3pppLfVH9EyAYGbOfpcO3vYkof4SIYDMI-CBEbh-A,8045
-html_to_markdown-2.1.2.dist-info/RECORD,,