html_to_markdown_rs/
lib.rs

1//! High-performance HTML to Markdown converter.
2//!
3//! Built with html5ever for fast, memory-efficient HTML parsing.
4//!
5//! ## Optional inline image extraction
6//!
7//! Enable the `inline-images` Cargo feature to collect embedded data URI images and inline SVG
8//! assets alongside the produced Markdown.
9
10use std::borrow::Cow;
11
12pub mod converter;
13pub mod error;
14pub mod hocr;
15#[cfg(feature = "inline-images")]
16mod inline_images;
17pub mod options;
18pub mod safety;
19pub mod text;
20pub mod wrapper;
21
22pub use error::{ConversionError, Result};
23#[cfg(feature = "inline-images")]
24pub use inline_images::{
25    HtmlExtraction, InlineImage, InlineImageConfig, InlineImageFormat, InlineImageSource, InlineImageWarning,
26};
27pub use options::{
28    CodeBlockStyle, ConversionOptions, HeadingStyle, HighlightStyle, ListIndentType, NewlineStyle,
29    PreprocessingOptions, PreprocessingPreset, WhitespaceMode,
30};
31
32/// Convert HTML to Markdown.
33///
34/// # Arguments
35///
36/// * `html` - The HTML string to convert
37/// * `options` - Optional conversion options (defaults to ConversionOptions::default())
38///
39/// # Example
40///
41/// ```
42/// use html_to_markdown_rs::{convert, ConversionOptions};
43///
44/// let html = "<h1>Hello World</h1>";
45/// let markdown = convert(html, None).unwrap();
46/// assert!(markdown.contains("Hello World"));
47/// ```
48pub fn convert(html: &str, options: Option<ConversionOptions>) -> Result<String> {
49    let options = options.unwrap_or_default();
50
51    let normalized_html = if html.contains('\r') {
52        Cow::Owned(html.replace("\r\n", "\n").replace('\r', "\n"))
53    } else {
54        Cow::Borrowed(html)
55    };
56
57    let markdown = converter::convert_html(normalized_html.as_ref(), &options)?;
58
59    if options.wrap {
60        Ok(wrapper::wrap_markdown(&markdown, &options))
61    } else {
62        Ok(markdown)
63    }
64}
65
66#[cfg(feature = "inline-images")]
67/// Convert HTML to Markdown while collecting inline image assets (requires the `inline-images` feature).
68///
69/// Extracts inline image data URIs and inline `<svg>` elements alongside Markdown conversion.
70///
71/// # Arguments
72///
73/// * `html` - The HTML string to convert
74/// * `options` - Optional conversion options (defaults to ConversionOptions::default())
75/// * `image_cfg` - Configuration controlling inline image extraction
76pub fn convert_with_inline_images(
77    html: &str,
78    options: Option<ConversionOptions>,
79    image_cfg: InlineImageConfig,
80) -> Result<HtmlExtraction> {
81    use std::cell::RefCell;
82    use std::rc::Rc;
83
84    let options = options.unwrap_or_default();
85
86    let normalized_html = if html.contains('\r') {
87        Cow::Owned(html.replace("\r\n", "\n").replace('\r', "\n"))
88    } else {
89        Cow::Borrowed(html)
90    };
91
92    let collector = Rc::new(RefCell::new(inline_images::InlineImageCollector::new(image_cfg)?));
93
94    let markdown =
95        converter::convert_html_with_inline_collector(normalized_html.as_ref(), &options, Rc::clone(&collector))?;
96
97    let markdown = if options.wrap {
98        wrapper::wrap_markdown(&markdown, &options)
99    } else {
100        markdown
101    };
102
103    let collector = Rc::try_unwrap(collector)
104        .map_err(|_| ConversionError::Other("failed to recover inline image state".to_string()))?
105        .into_inner();
106    let (inline_images, warnings) = collector.finish();
107
108    Ok(HtmlExtraction {
109        markdown,
110        inline_images,
111        warnings,
112    })
113}
html_to_markdown_rs/lib.rs

html_to_markdown_rs/
lib.rs
