html-to-markdown 2.1.0__cp310-abi3-macosx_11_0_arm64.whl → 2.2.0__cp310-abi3-macosx_11_0_arm64.whl

html_to_markdown/__init__.py

@@ -39,4 +39,4 @@ __all__ = [
     "markdownify",
 ]
 
-__version__ = "2.1.0"
+__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-2.1.0.dist-info → html_to_markdown-2.2.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: html-to-markdown
-Version: 2.1.0
+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,sha256=_jgp22NNhPsmx9MgSb11wrTWiYpwnzZ7qZtpdXrudls,3817824
+html_to_markdown-2.2.0.dist-info/RECORD,,
+html_to_markdown-2.2.0.dist-info/WHEEL,sha256=HtAbUhtjhH1WdiDuIy2CapdoAiKCwe6bij_Tlxr1lEg,131
+html_to_markdown-2.2.0.dist-info/METADATA,sha256=v8jDHZrq6_Vuwry7Dn3e8r9QlAZDj3-8kU_mFrSA9og,7071
+html_to_markdown-2.2.0.dist-info/licenses/LICENSE,sha256=oQvPC-0UWvfg0WaeUBe11OJMtX60An-TW1ev_oaAA0k,1086
+html_to_markdown/options.py,sha256=N1orEtzXiqgV88Y7eqRjw3ilxDi7N97FBkBa79hdfzo,4913
+html_to_markdown/_html_to_markdown.abi3.so,sha256=xI0Wr26xhNoPGzlrlngo49vWvg61X_4d0xdMd9de8Rk,3668288
+html_to_markdown/__init__.py,sha256=QZ314Edod-PD-v5CvRtJ-Lw7lqXqHP6o36UtyzUQrg8,1149
+html_to_markdown/api.py,sha256=f0jhD003p8Kz5jHe0BdaKN8Uh9mvGDo2Sl9eNw06VAY,2784
+html_to_markdown/_rust.pyi,sha256=pi6C_qAdB81qUlC89Dy1ZKC1JrpdnqVce-caJx3ekPA,2098
+html_to_markdown/v1_compat.py,sha256=5DZA-fPMqZ5hYiA43rFaOAqshLS8MScbBnivDXuvQII,8034
+html_to_markdown/cli.py,sha256=Rn-s3FZPea1jgCJtDzH_TFvOEiA_uZFVfgjhr6xyL_g,64
+html_to_markdown/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+html_to_markdown/exceptions.py,sha256=aTASOzbywgfqOYjlw18ZkOWSxKff4EbUbmMua_73TGA,2370
+html_to_markdown/cli_proxy.py,sha256=MbDRZdmQMCDI9cruy1vifc__FsjNPRdvBXKFU9GaAZE,3695
+html_to_markdown/__main__.py,sha256=3Ic_EbOt2h6W88q084pkz5IKU6iY5z_woBygH6u9aw0,327
+html_to_markdown/bin/html-to-markdown,sha256=_jgp22NNhPsmx9MgSb11wrTWiYpwnzZ7qZtpdXrudls,3817824

html_to_markdown-2.1.0.dist-info/RECORD

@@ -1,17 +0,0 @@
-html_to_markdown-2.1.0.dist-info/RECORD,,
-html_to_markdown-2.1.0.dist-info/WHEEL,sha256=HtAbUhtjhH1WdiDuIy2CapdoAiKCwe6bij_Tlxr1lEg,131
-html_to_markdown-2.1.0.dist-info/METADATA,sha256=FPqk_oGX2HOKE-VW5m8w4lfuUm4oxqQTXPanby6jwVU,7071
-html_to_markdown-2.1.0.dist-info/licenses/LICENSE,sha256=oQvPC-0UWvfg0WaeUBe11OJMtX60An-TW1ev_oaAA0k,1086
-html_to_markdown-2.1.0.data/scripts/html-to-markdown,sha256=U7CnxO8PkyRby25ZWYC1jbC0LQepZMzZbidDcDp6yBM,3784640
-html_to_markdown/options.py,sha256=oV-_GFEKuL3RBu27RD1AhVruEh-bNuY3c8ATRbzcos0,4802
-html_to_markdown/_html_to_markdown.abi3.so,sha256=mGBBfwO_d5IbJW2bY0fvJIJhuLKn2Yb3zShn3y-m9eE,3618704
-html_to_markdown/__init__.py,sha256=ePkhgm_2JevNaqTGchhqkUVNLtxkFJiBJ9EE4gUXpBw,1149
-html_to_markdown/api.py,sha256=HuM6RZg064VxrTvwcY-OmraS-hsGM9Bt1tIaM0_w7F8,2727
-html_to_markdown/_rust.pyi,sha256=An3Wlvedlr_2XgzqmXulLi5AzMx3HTqOJWH11M5cgcY,2026
-html_to_markdown/v1_compat.py,sha256=VQq1wv8OedkESpCFaUpaSUBh6vJNkByylVUbY6EPIZ8,7856
-html_to_markdown/cli.py,sha256=Rn-s3FZPea1jgCJtDzH_TFvOEiA_uZFVfgjhr6xyL_g,64
-html_to_markdown/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-html_to_markdown/exceptions.py,sha256=aTASOzbywgfqOYjlw18ZkOWSxKff4EbUbmMua_73TGA,2370
-html_to_markdown/cli_proxy.py,sha256=MbDRZdmQMCDI9cruy1vifc__FsjNPRdvBXKFU9GaAZE,3695
-html_to_markdown/__main__.py,sha256=3Ic_EbOt2h6W88q084pkz5IKU6iY5z_woBygH6u9aw0,327
-html_to_markdown/bin/html-to-markdown,sha256=U7CnxO8PkyRby25ZWYC1jbC0LQepZMzZbidDcDp6yBM,3784640