html-to-markdown 1.10.0__py3-none-any.whl → 1.11.0__py3-none-any.whl

html_to_markdown/converters.py

@@ -578,8 +578,11 @@ def _convert_semantic_block(*, text: str, convert_as_inline: bool) -> str:
     return f"{text}\n\n" if text.strip() else ""
 
 
-def _convert_div(*, text: str, convert_as_inline: bool) -> str:  # noqa: ARG001
-    return text
+def _convert_div(*, text: str, convert_as_inline: bool) -> str:
+    if convert_as_inline:
+        return text
+
+    return _format_block_element(text)
 
 
 def _convert_details(*, text: str, convert_as_inline: bool) -> str:

html_to_markdown/processing.py

@@ -258,6 +258,18 @@ def _process_tag(
             if n_eol_to_add > 0:
                 prefix = "\n" * n_eol_to_add
                 return f"{prefix}{rendered}"
+
+        from html_to_markdown.whitespace import BLOCK_ELEMENTS  # noqa: PLC0415
+
+        is_block_element = tag.name.lower() in BLOCK_ELEMENTS
+        if (
+            is_block_element
+            and not convert_as_inline
+            and context_before
+            and not context_before.endswith("\n")
+            and rendered.strip()
+        ):
+            return f"\n\n{rendered}"
         return rendered
 
     return text
@@ -358,7 +370,7 @@ def _as_optional_set(value: str | Iterable[str] | None) -> set[str] | None:
     if value is None:
         return None
     if isinstance(value, str):
-        return set(",".split(value))
+        return set(value.split(","))
     return {*chain(*[v.split(",") for v in value])}
 
 
@@ -836,15 +848,6 @@ def _process_html_core(
 
     try:
         if isinstance(source, str):
-            if (
-                heading_style == UNDERLINED
-                and "Header" in source
-                and "\n------\n\n" in source
-                and "Next paragraph" in source
-            ):
-                sink.write(source)
-                return
-
             if strip_newlines:
                 source = source.replace("\n", " ").replace("\r", " ")
 

html_to_markdown/whitespace.py

@@ -171,13 +171,13 @@ class WhitespaceHandler:
         if not text:
             return ""
 
-        text = self.normalize_unicode_spaces(text)
-
         if in_pre or self.should_preserve_whitespace(element):
             return text
 
         if self.mode == "strict":
             return text
+
+        text = self.normalize_unicode_spaces(text)
         return self._process_normalized(text, element)
 
     def _process_normalized(self, text: str, element: NavigableString) -> str:
@@ -242,6 +242,14 @@ class WhitespaceHandler:
         prev_sibling = element.previous_sibling
         next_sibling = element.next_sibling
 
+        multiple_newlines_before_block = (
+            original
+            and original.count("\n") >= 2
+            and self.is_block_element(next_sibling)
+            and text.strip()
+            and (self.is_inline_element(prev_sibling) or prev_sibling is None)
+        )
+
         has_leading = (
             has_lead_space
             and original[0] == " "
@@ -268,6 +276,9 @@ class WhitespaceHandler:
         if has_trailing and not (original and original[-1] in "\n\t"):
             text = text + " "
 
+        if multiple_newlines_before_block:
+            text = text + "\n\n"
+
         return text
 
     def get_block_spacing(self, tag: Tag, next_sibling: PageElement | None = None) -> str:
@@ -286,7 +297,7 @@ class WhitespaceHandler:
             return "\n"
         if tag_name in single_newline_elements:
             return "\n"
-        if tag_name.startswith("h") and len(tag_name) == 2:
+        if tag_name.startswith("h") and len(tag_name) == 2 and tag_name[1].isdigit():
             return "\n\n"
 
         return ""

{html_to_markdown-1.10.0.dist-info → html_to_markdown-1.11.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: html-to-markdown
-Version: 1.10.0
+Version: 1.11.0
 Summary: A modern, type-safe Python library for converting HTML to Markdown with comprehensive tag support and customizable options
 Author-email: Na'aman Hirschfeld <nhirschfeld@gmail.com>
 License: MIT
@@ -33,7 +33,7 @@ License-File: LICENSE
 Requires-Dist: beautifulsoup4>=4.13.5
 Requires-Dist: nh3>=0.3
 Provides-Extra: lxml
-Requires-Dist: lxml>=6.0.1; extra == "lxml"
+Requires-Dist: beautifulsoup4[lxml]>=4.13.5; extra == "lxml"
 Dynamic: license-file
 
 # html-to-markdown

{html_to_markdown-1.10.0.dist-info → html_to_markdown-1.11.0.dist-info}/RECORD

@@ -2,16 +2,16 @@ html_to_markdown/__init__.py,sha256=TzZzhZDJHeXW_3B9zceYehz2zlttqdLsDr5un8stZLM,
 html_to_markdown/__main__.py,sha256=E9d62nVceR_5TUWgVu5L5CnSZxKcnT_7a6ScWZUGE-s,292
 html_to_markdown/cli.py,sha256=ilnrJN2XMhPDQ4UkkG4cjLXTvglu_ZJj-bBsohVF3fw,8541
 html_to_markdown/constants.py,sha256=CKFVHjUZKgi8-lgU6AHPic7X5ChlTkbZt4Jv6VaVjjs,665
-html_to_markdown/converters.py,sha256=ewdKUwkQXuwgzwCBhxZ1AJufX90jR_aGLr02GkdB2So,32443
+html_to_markdown/converters.py,sha256=CbChkRIlOPe0d1MK5-txDE56IG4Ea_dcCV6KRCTjeKY,32497
 html_to_markdown/exceptions.py,sha256=YjfwVCWE_oZakr9iy0E-_aPSYHNaocJZgWeQ9Enty7Q,1212
 html_to_markdown/preprocessor.py,sha256=acmuJJvx1RaXE3c0F_aWsartQE0cEpa3AOnJYGnPzqw,9708
-html_to_markdown/processing.py,sha256=tqrBfXKqbN_rQbFOY4pGhDjY9fHyj_E1gOlhqE1ywK0,34214
+html_to_markdown/processing.py,sha256=sOIIFNyRkRYAH8Q4ehrh66RY71bkvttSuqzXYsMC5JM,34334
 html_to_markdown/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 html_to_markdown/utils.py,sha256=4Vzk2cCjxN0LAZ1DXQCufYtxE7a6739TYgPbje-VM_E,1086
-html_to_markdown/whitespace.py,sha256=b8Vf_AWhIvGFqka4Au0GsxsOYeYRO9XBpD4DxW99Pg0,7806
-html_to_markdown-1.10.0.dist-info/licenses/LICENSE,sha256=3J_HR5BWvUM1mlIrlkF32-uC1FM64gy8JfG17LBuheQ,1122
-html_to_markdown-1.10.0.dist-info/METADATA,sha256=LlFYc0EDFdfapqLacVQ9Da12SjEWKExW-L-5j55bicM,17797
-html_to_markdown-1.10.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-html_to_markdown-1.10.0.dist-info/entry_points.txt,sha256=xmFijrTfgYW7lOrZxZGRPciicQHa5KiXKkUhBCmICtQ,116
-html_to_markdown-1.10.0.dist-info/top_level.txt,sha256=Ev6djb1c4dSKr_-n4K-FpEGDkzBigXY6LuZ5onqS7AE,17
-html_to_markdown-1.10.0.dist-info/RECORD,,
+html_to_markdown/whitespace.py,sha256=EJ0gEsfLB_wZAk5d5qP4UPhPg0pJJ8LZLRRr_QoL01o,8186
+html_to_markdown-1.11.0.dist-info/licenses/LICENSE,sha256=3J_HR5BWvUM1mlIrlkF32-uC1FM64gy8JfG17LBuheQ,1122
+html_to_markdown-1.11.0.dist-info/METADATA,sha256=Cej6bnqT9JVFzACZvND6Z5-kD0QoabiLi46opAaC11U,17814
+html_to_markdown-1.11.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+html_to_markdown-1.11.0.dist-info/entry_points.txt,sha256=xmFijrTfgYW7lOrZxZGRPciicQHa5KiXKkUhBCmICtQ,116
+html_to_markdown-1.11.0.dist-info/top_level.txt,sha256=Ev6djb1c4dSKr_-n4K-FpEGDkzBigXY6LuZ5onqS7AE,17
+html_to_markdown-1.11.0.dist-info/RECORD,,