Rust implementation of CSS Syntax Level 3
Rust Python
Latest commit ee735e8 Nov 1, 2016 @bors-servo bors-servo committed on GitHub Auto merge of #110 - emilio:data-uri, r=SimonSapin
Don't use utf8 logic to parse unquoted urls.

Since it's not needed, and it's noticeable when parsing large data-uris, like servo/servo#13778.

<!-- Reviewable:start -->
---

This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/rust-cssparser/110)

<!-- Reviewable:end -->

README.md

rust-cssparser

Build Status

Documentation

Rust implementation of CSS Syntax Module Level 3

Overview

Parsing CSS involves a series of steps:

  • When parsing from bytes, (e.g. reading a file or fetching an URL from the network,) detect the character encoding (based on a Content-Type HTTP header, an @charset rule, a BOM, etc.) and decode to Unicode text.

    rust-cssparser does not do this yet and just assumes UTF-8.

    This step is skipped when parsing from Unicode, e.g. in an HTML <style> element.

  • Tokenization, a.k.a. lexing. The input, a stream of Unicode text, is transformed into a stream of tokens. Tokenization never fails, although the output may contain error tokens.

  • This flat stream of tokens is then transformed into a tree of component values, which are either preserved tokens, or blocks/functions ({ … }, [ … ], ( … ), foo( … )) that contain more component values.

    rust-cssparser does this at the same time as tokenization: raw tokens are never materialized, you only get component values.

  • Component values can then be parsed into generic rules or declarations. The header and body of rules as well as the value of declarations are still just lists of component values at this point. See the ast module for the data structures.

  • The last step of a full CSS parser is parsing the remaining component values into Selectors, specific CSS properties, etc.

    By design, rust-cssparser does not do this last step which depends a lot on what you want to do: which properties you want to support, what you want to do with selectors, etc.

    It does however provide some helper functions to parse CSS colors and An+B (the argument to :nth-child() and related selectors.

    See Servo’s style crate for an example of a parser based on rust-cssparser.

TODO

  • Figure out float and integer overflow in parsing. (Clamp instead?)
  • Make it fast! (Add a fast path in identifier tokenization?)