This library is a fork of the tokenizer from [html5ever] with the following changes:
The dependencies on [markup5ever], [tendril], [mac] and [log] were removed.
This spares you about 40 build dependencies and the unsafe code from Tendril.
The dependency on [phf] was made optional: if you don't need to resolve named
entities like &, you can disable the named-entities feature, in which
case this library does not have any dependencies (other than the standard
library).
This library takes care of appropriately switching tokenizer states based on
tag names (e.g. for script and styles) ... with the html5ever tokenizer
you had to do this yourself.
An optional spans feature has been added to make the tokenizer report the
source code spans for parser errors, tag names and attributes.
The feature is disabled by default.
The API has been polished, e.g. the internal tokenizer state enums are no longer public and errors are no longer stringly typed.
If you want to parse HTML into a tree (DOM) you should by all means use html5ever, this crate is merely for those who only want an HTML5 tokenizer and seek to minimize their build dependencies (html5ever pulls in 56).
Thanks to the developers of html5ever for their awesome parser!