html-to-markdown 2.3.0__cp310-abi3-win_amd64.whl → 2.3.4__cp310-abi3-win_amd64.whl

html_to_markdown/__init__.py

@@ -15,7 +15,13 @@ V1 API (backward compatibility):
     markdown = convert_to_markdown(html, heading_style="atx")
 """
 
-from html_to_markdown.api import convert
+from html_to_markdown.api import (
+    InlineImage,
+    InlineImageConfig,
+    InlineImageWarning,
+    convert,
+    convert_with_inline_images,
+)
 from html_to_markdown.exceptions import (
     ConflictingOptionsError,
     EmptyHtmlError,
@@ -31,12 +37,16 @@ __all__ = [
     "ConversionOptions",
     "EmptyHtmlError",
     "HtmlToMarkdownError",
+    "InlineImage",
+    "InlineImageConfig",
+    "InlineImageWarning",
     "InvalidParserError",
     "MissingDependencyError",
     "PreprocessingOptions",
     "convert",
     "convert_to_markdown",
+    "convert_with_inline_images",
     "markdownify",
 ]
 
-__version__ = "2.3.0"
+__version__ = "2.3.4"

html_to_markdown/api.py

@@ -6,38 +6,52 @@ using the Rust backend for conversion.
 
 from __future__ import annotations
 
+from typing import TYPE_CHECKING, Literal, TypedDict, cast
+
 import html_to_markdown._html_to_markdown as _rust  # type: ignore[import-not-found]
 from html_to_markdown.options import ConversionOptions, PreprocessingOptions
 
+if TYPE_CHECKING:
+    from html_to_markdown._html_to_markdown import InlineImageConfig
+else:
+    InlineImageConfig = _rust.InlineImageConfig  # type: ignore[misc, assignment]
 
-def convert(
-    html: str,
-    options: ConversionOptions | None = None,
-    preprocessing: PreprocessingOptions | None = None,
-) -> str:
-    """Convert HTML to Markdown using the Rust backend.
 
-    Args:
-        html: HTML string to convert.
-        options: Conversion configuration options (defaults to ConversionOptions()).
-        preprocessing: HTML preprocessing options (defaults to PreprocessingOptions()).
+class InlineImage(TypedDict):
+    """Inline image extracted during conversion."""
+
+    data: bytes
+    format: str
+    filename: str | None
+    description: str | None
+    dimensions: tuple[int, int] | None
+    source: Literal["img_data_uri", "svg_element"]
+    attributes: dict[str, str]
 
-    Returns:
-        Converted Markdown string.
-    """
-    if options is None:
-        options = ConversionOptions()
-    if preprocessing is None:
-        preprocessing = PreprocessingOptions()
 
-    rust_preprocessing = _rust.PreprocessingOptions(
-        enabled=preprocessing.enabled,
-        preset=preprocessing.preset,
-        remove_navigation=preprocessing.remove_navigation,
-        remove_forms=preprocessing.remove_forms,
+class InlineImageWarning(TypedDict):
+    """Warning produced during inline image extraction."""
+
+    index: int
+    message: str
+
+
+def _to_rust_preprocessing(options: PreprocessingOptions) -> _rust.PreprocessingOptions:
+    """Convert high-level preprocessing options to the Rust bindings."""
+    return _rust.PreprocessingOptions(
+        enabled=options.enabled,
+        preset=options.preset,
+        remove_navigation=options.remove_navigation,
+        remove_forms=options.remove_forms,
     )
 
-    rust_options = _rust.ConversionOptions(
+
+def _to_rust_options(
+    options: ConversionOptions,
+    preprocessing: PreprocessingOptions,
+) -> _rust.ConversionOptions:
+    """Convert high-level conversion options to the Rust bindings."""
+    return _rust.ConversionOptions(
         heading_style=options.heading_style,
         list_indent_type=options.list_indent_type,
         list_indent_width=options.list_indent_width,
@@ -64,11 +78,66 @@ def convert(
         newline_style=options.newline_style,
         code_block_style=options.code_block_style,
         keep_inline_images_in=list(options.keep_inline_images_in) if options.keep_inline_images_in else [],
-        preprocessing=rust_preprocessing,
+        preprocessing=_to_rust_preprocessing(preprocessing),
         encoding=options.encoding,
         debug=options.debug,
         strip_tags=list(options.strip_tags) if options.strip_tags else [],
     )
 
-    result: str = _rust.convert(html, rust_options)
-    return result
+
+def convert(
+    html: str,
+    options: ConversionOptions | None = None,
+    preprocessing: PreprocessingOptions | None = None,
+) -> str:
+    """Convert HTML to Markdown using the Rust backend.
+
+    Args:
+        html: HTML string to convert.
+        options: Conversion configuration options (defaults to ConversionOptions()).
+        preprocessing: HTML preprocessing options (defaults to PreprocessingOptions()).
+
+    Returns:
+        Converted Markdown string.
+    """
+    if options is None:
+        options = ConversionOptions()
+    if preprocessing is None:
+        preprocessing = PreprocessingOptions()
+
+    rust_options = _to_rust_options(options, preprocessing)
+    return cast("str", _rust.convert(html, rust_options))
+
+
+def convert_with_inline_images(
+    html: str,
+    options: ConversionOptions | None = None,
+    preprocessing: PreprocessingOptions | None = None,
+    image_config: InlineImageConfig | None = None,
+) -> tuple[str, list[InlineImage], list[InlineImageWarning]]:
+    """Convert HTML and extract inline images.
+
+    Returns Markdown along with extracted inline images and any warnings.
+    """
+    if options is None:
+        options = ConversionOptions()
+    if preprocessing is None:
+        preprocessing = PreprocessingOptions()
+    if image_config is None:
+        image_config = InlineImageConfig()
+
+    rust_options = _to_rust_options(options, preprocessing)
+    markdown, images, warnings = cast(
+        "tuple[str, list[InlineImage], list[InlineImageWarning]]",
+        _rust.convert_with_inline_images(html, rust_options, image_config),
+    )
+    return markdown, list(images), list(warnings)
+
+
+__all__ = [
+    "InlineImage",
+    "InlineImageConfig",
+    "InlineImageWarning",
+    "convert",
+    "convert_with_inline_images",
+]

{html_to_markdown-2.3.0.dist-info → html_to_markdown-2.3.4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: html-to-markdown
-Version: 2.3.0
+Version: 2.3.4
 Classifier: Development Status :: 5 - Production/Stable
 Classifier: Environment :: Console
 Classifier: Intended Audience :: Developers

html_to_markdown-2.3.4.dist-info/RECORD

@@ -0,0 +1,17 @@
+html_to_markdown-2.3.4.data/scripts/html-to-markdown.exe,sha256=_d3wckq2iLwVEUENRr5yDwmXSqRkhI2_umhtuaKPBbI,4470272
+html_to_markdown-2.3.4.dist-info/METADATA,sha256=t4Tw_UzLr7ewVN7deNNtkyUh_8XGD3SoafiGxlRSJDk,9012
+html_to_markdown-2.3.4.dist-info/WHEEL,sha256=4EDp_7DiFfWl1yYv5M4wSosAn5L_xgD1dyrQxQxfCx8,95
+html_to_markdown-2.3.4.dist-info/licenses/LICENSE,sha256=QhKFMkQLa4mSUlOsyG9VElzC7GYbAKtiS_EwOCyH-b4,1107
+html_to_markdown/__init__.py,sha256=P7YA3QFT6Le-BF1c-thxQuKzXFHaQEWhZCHPmkirYWU,1410
+html_to_markdown/__main__.py,sha256=5objj9lB7hhpSpZsDok5tv9o9yztVR63Ccww-pXsAyY,343
+html_to_markdown/_html_to_markdown.pyd,sha256=B6k-CqYNPzJ0rorjOl-Rcox8iDXk7LYshHY15B8g1_A,4214784
+html_to_markdown/_rust.pyi,sha256=JP8tvcjYDfFJeJkbLpQ4qeK-5jl0hzIVT3Sa0daTkyo,2171
+html_to_markdown/api.py,sha256=jDb8PT1cS3KqipT4m_rKBE0R20UKOU85rH-7M6P6Owk,5003
+html_to_markdown/bin/html-to-markdown.exe,sha256=_d3wckq2iLwVEUENRr5yDwmXSqRkhI2_umhtuaKPBbI,4470272
+html_to_markdown/cli.py,sha256=z59l8sF8wIRRzJtUd-tXgqiC0WTqkTjzl-df8Ey_oQ0,67
+html_to_markdown/cli_proxy.py,sha256=JGOuINBI8OMYLxojXGz8DdzMHo8eqgdINstOZWrdw-8,3816
+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-2.3.4.dist-info/RECORD,,

html_to_markdown-2.3.0.dist-info/RECORD

@@ -1,17 +0,0 @@
-html_to_markdown-2.3.0.data/scripts/html-to-markdown.exe,sha256=NoG5btr57ihL85-URLcreFRZTiAUfO7km_FiEUMa7xI,4469760
-html_to_markdown-2.3.0.dist-info/METADATA,sha256=wcmX8lAc-dZZp8ETrpbXqu58Kf19n3fE6ubBPd0OfKU,9012
-html_to_markdown-2.3.0.dist-info/WHEEL,sha256=4EDp_7DiFfWl1yYv5M4wSosAn5L_xgD1dyrQxQxfCx8,95
-html_to_markdown-2.3.0.dist-info/licenses/LICENSE,sha256=QhKFMkQLa4mSUlOsyG9VElzC7GYbAKtiS_EwOCyH-b4,1107
-html_to_markdown/__init__.py,sha256=KgR9V82EqdL5S7dzK_USOv6STjyhVRJubDYScHxOJS0,1191
-html_to_markdown/__main__.py,sha256=5objj9lB7hhpSpZsDok5tv9o9yztVR63Ccww-pXsAyY,343
-html_to_markdown/_html_to_markdown.pyd,sha256=Peez6o-WS_O6Cc3YqbxjJzRVw0mLY5HsbJyfeoYj7BY,4214784
-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=NoG5btr57ihL85-URLcreFRZTiAUfO7km_FiEUMa7xI,4469760
-html_to_markdown/cli.py,sha256=z59l8sF8wIRRzJtUd-tXgqiC0WTqkTjzl-df8Ey_oQ0,67
-html_to_markdown/cli_proxy.py,sha256=JGOuINBI8OMYLxojXGz8DdzMHo8eqgdINstOZWrdw-8,3816
-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-2.3.0.dist-info/RECORD,,