44 releases (11 breaking)

0.12.4 Mar 3, 2024
0.12.2 Feb 23, 2024
0.9.4 Dec 31, 2023
0.6.0 May 23, 2023
0.1.3 Dec 31, 2016

#34 in Text processing

Download history 1478/week @ 2023-12-24 2266/week @ 2023-12-31 2300/week @ 2024-01-07 2314/week @ 2024-01-14 2183/week @ 2024-01-21 1857/week @ 2024-01-28 1888/week @ 2024-02-04 2259/week @ 2024-02-11 2411/week @ 2024-02-18 2656/week @ 2024-02-25 3029/week @ 2024-03-03 2935/week @ 2024-03-10 3040/week @ 2024-03-17 3025/week @ 2024-03-24 3759/week @ 2024-03-31 2196/week @ 2024-04-07

12,313 downloads per month
Used in 72 crates (44 directly)

MIT license

245KB
6.5K SLoC

jugglerchris

html2text

html2text is a Rust crate which converts HTML to plain text.

It makes use of the Servo project's HTML parser, html5ever, using the DOM to generate text (which can optionally include annotations for some features such as hyperlinks).

The project aims to do a reasonable job of rendering reasonable HTML in a terminal or other places where HTML needs to be converted to text (for example the text/plain fallback in HTML e-mails).

With features (see below) some CSS/colour support is available.

Examples

The simple functions like from_read() return formatted text (in various formats including plain text).

use html2text::from_read;
let html = b"
       <ul>
         <li>Item one</li>
         <li>Item two</li>
         <li>Item three</li>
       </ul>";
assert_eq!(from_read(&html[..], 20),
           "\
* Item one
* Item two
* Item three
");

A lower level API gives a bit more control. This give the same result (except for returning errors as Result instead of panicking):

use html2text::config;

let html = b"
       <ul>
         <li>Item one</li>
         <li>Item two</li>
         <li>Item three</li>
       </ul>";

assert_eq!(
    config::plain()
           .string_from_read(&html[..], 20)
           .unwrap(),
    "\
* Item one
* Item two
* Item three
");

A couple of simple demonstration programs are included as examples:

html2text

The simplest example uses from_read to convert HTML on stdin into plain text:

$ cargo run --example html2text < foo.html
[...]

html2term

A very simple example of using the rich interface (from_read_rich) for a slightly interactive console HTML viewer is provided as html2term.

$ cargo run --example html2term foo.html
[...]

Note that this example takes the HTML file as a parameter so that it can read keys from stdin.

Cargo Features

Feature Description
css Limited handling of CSS, adding Coloured nodes to the render tree.
html_trace Add verbose internal logging (not recommended)
html_trace_bt Add backtraces to the verbose internal logging

CSS support

When the css feature is enabled, some simple CSS handling is available.

Style rules are taken from:

  • If Config::use_doc_css() is called, then style from the document:
    • <style> elements
    • Inline style attributes (<div style="...">)
    • <font color=...>
  • Independently of use_doc_css, extra rules can be added with Config::add_css(...)

The following CSS features are implemented:

  • Basic selector matching (including child and descendents, classes and element types).
  • CSS colors (color/background-color) will add Coloured(...)/BgColoured(...) nodes to the render tree.
  • Rules with display: none will cause matching elements to be removed from the render tree.

The CSS handling is expected to improve in future (PRs welcome), but not to a full- blown browser style system, which would be overkill for terminal output.

There are two ways to make use of the colours:

  • Use from_read_rich() or one of its variants. One of the annotations you may get back is Colour(..).
  • Use from_read_coloured(). This is similar to from_read(), but you provide a function to add terminal colours (or other styling) based on the same RichAnnotations. See examples/html2text.rs for an example using termion.

Dependencies

~3–13MB
~118K SLoC