html-to-markdown 1.3.1__py3-none-any.whl → 1.3.3__py3-none-any.whl

html_to_markdown/__main__.py

@@ -1,6 +1,6 @@
 import sys
 
-if __name__ == "__main__":
+def cli():
     from html_to_markdown.cli import main
 
     try:
@@ -9,3 +9,6 @@ if __name__ == "__main__":
     except ValueError as e:
         print(str(e), file=sys.stderr)  # noqa: T201
         sys.exit(1)
+
+if __name__ == "__main__":
+    cli()

html_to_markdown/processing.py

@@ -89,6 +89,7 @@ def _process_tag(
     escape_misc: bool,
     escape_underscores: bool,
     strip: set[str] | None,
+    context_before: str = "",
 ) -> str:
     should_convert_tag = _should_convert_tag(tag_name=tag.name, strip=strip, convert=convert)
     tag_name: SupportedTag | None = (
@@ -129,12 +130,21 @@ def _process_tag(
                 escape_misc=escape_misc,
                 escape_underscores=escape_underscores,
                 strip=strip,
+                context_before=(context_before + text)[-2:],
             )
 
     if tag_name and should_convert_tag:
-        return converters_map[tag_name](  # type: ignore[call-arg]
+        rendered = converters_map[tag_name](  # type: ignore[call-arg]
             tag=tag, text=text, convert_as_inline=convert_as_inline
         )
+        # For headings, ensure two newlines before if not already present
+        # Edge case where the document starts with a \n and then a heading
+        if is_heading and context_before not in {"", "\n"}:
+            n_eol_to_add = 2 - (len(context_before) - len(context_before.rstrip("\n")))
+            if n_eol_to_add > 0:
+                prefix = "\n" * n_eol_to_add
+                return f"{prefix}{rendered}"
+        return rendered
 
     return text
 
@@ -275,13 +285,25 @@ def convert_to_markdown(
     if custom_converters:
         converters_map.update(cast("ConvertersMap", custom_converters))
 
-    return _process_tag(
-        source,
-        converters_map,
-        convert=_as_optional_set(convert),
-        convert_as_inline=convert_as_inline,
-        escape_asterisks=escape_asterisks,
-        escape_misc=escape_misc,
-        escape_underscores=escape_underscores,
-        strip=_as_optional_set(strip),
-    )
+    text = ""
+    for el in filter(lambda value: not isinstance(value, (Comment, Doctype)), source.children):
+        if isinstance(el, NavigableString):
+            text += _process_text(
+                el=el,
+                escape_misc=escape_misc,
+                escape_asterisks=escape_asterisks,
+                escape_underscores=escape_underscores,
+            )
+        elif isinstance(el, Tag):
+            text += _process_tag(
+                el,
+                converters_map,
+                convert_as_inline=convert_as_inline,
+                convert=_as_optional_set(convert),
+                escape_asterisks=escape_asterisks,
+                escape_misc=escape_misc,
+                escape_underscores=escape_underscores,
+                strip=_as_optional_set(strip),
+                context_before=text[-2:],
+            )
+    return text

{html_to_markdown-1.3.1.dist-info → html_to_markdown-1.3.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: html-to-markdown
-Version: 1.3.1
+Version: 1.3.3
 Summary: Convert HTML to markdown
 Author-email: Na'aman Hirschfeld <nhirschfeld@gmail.com>
 License: MIT

html_to_markdown-1.3.3.dist-info/RECORD

@@ -0,0 +1,14 @@
+html_to_markdown/__init__.py,sha256=95S7_7mR_g88uTnFI0FaRNykrtAaSKb6sJbwSea2zjk,145
+html_to_markdown/__main__.py,sha256=_EeKI8veMWZO7xsl-mBHBE-OmH1vnkVyXuExsOfduFI,286
+html_to_markdown/cli.py,sha256=HVnzmcyrYwah_yWhZ87mZcG0VgnKYp6y89fJh2R-Rlw,4532
+html_to_markdown/constants.py,sha256=Usk67k18tuRovJpKDsiEXdgH20KgqI9KOnK4Fbx-M5c,547
+html_to_markdown/converters.py,sha256=p8arBdejEeuAp9_wIYvp5PuWNBB0M699CgLSEkW3v88,11910
+html_to_markdown/processing.py,sha256=JNCjDgbfuW3YI7mfsj9aHlk2-KriQXJHU8Eo5D9Qj1E,10280
+html_to_markdown/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+html_to_markdown/utils.py,sha256=HJUDej5HSpXRtYv-OkCyD0hwnPnVfQCwY6rBRlIOt9s,1989
+html_to_markdown-1.3.3.dist-info/licenses/LICENSE,sha256=3J_HR5BWvUM1mlIrlkF32-uC1FM64gy8JfG17LBuheQ,1122
+html_to_markdown-1.3.3.dist-info/METADATA,sha256=BcfsHs0cSG8Y1BScMsGFOGfN5mrxiu-HA_fJC6DrtFg,7653
+html_to_markdown-1.3.3.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+html_to_markdown-1.3.3.dist-info/entry_points.txt,sha256=jhMqXDYvIyzQDLKjCn4xCyzCCbAMl94tzQx_HiG5Qi0,67
+html_to_markdown-1.3.3.dist-info/top_level.txt,sha256=Ev6djb1c4dSKr_-n4K-FpEGDkzBigXY6LuZ5onqS7AE,17
+html_to_markdown-1.3.3.dist-info/RECORD,,

{html_to_markdown-1.3.1.dist-info → html_to_markdown-1.3.3.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (78.1.0)
+Generator: setuptools (80.9.0)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

html_to_markdown-1.3.3.dist-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+html_to_markdown = html_to_markdown.__main__:cli

html_to_markdown-1.3.1.dist-info/RECORD

@@ -1,13 +0,0 @@
-html_to_markdown/__init__.py,sha256=95S7_7mR_g88uTnFI0FaRNykrtAaSKb6sJbwSea2zjk,145
-html_to_markdown/__main__.py,sha256=u5xevySlT5eIGyLUaethdDQIKJygaKnc3F2sHWoz75g,264
-html_to_markdown/cli.py,sha256=HVnzmcyrYwah_yWhZ87mZcG0VgnKYp6y89fJh2R-Rlw,4532
-html_to_markdown/constants.py,sha256=Usk67k18tuRovJpKDsiEXdgH20KgqI9KOnK4Fbx-M5c,547
-html_to_markdown/converters.py,sha256=p8arBdejEeuAp9_wIYvp5PuWNBB0M699CgLSEkW3v88,11910
-html_to_markdown/processing.py,sha256=ZYp4sMsC2Plb0iyGTFmyCKWc7lSHHFYc3S46UrlfOHw,9199
-html_to_markdown/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-html_to_markdown/utils.py,sha256=HJUDej5HSpXRtYv-OkCyD0hwnPnVfQCwY6rBRlIOt9s,1989
-html_to_markdown-1.3.1.dist-info/licenses/LICENSE,sha256=3J_HR5BWvUM1mlIrlkF32-uC1FM64gy8JfG17LBuheQ,1122
-html_to_markdown-1.3.1.dist-info/METADATA,sha256=CR__rjsnqp1XncpI9oWUTHxetKQY1wX6sxVfl_U1fEo,7653
-html_to_markdown-1.3.1.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
-html_to_markdown-1.3.1.dist-info/top_level.txt,sha256=Ev6djb1c4dSKr_-n4K-FpEGDkzBigXY6LuZ5onqS7AE,17
-html_to_markdown-1.3.1.dist-info/RECORD,,