parse5

parse5 was caught up with the HTML spec, and parsing results might differ in edge-cases (#442, #451)

The parse5-serializer-stream package was removed #481

The rewriting stream now splits very long text sections (#434) and doesn’t escape text in special tags anymore (#434). If you worked around these issues before, you might have to update your code.

The htmlparser2 adapter now uses domhandler’s node interface (#327

The tokenizer now uses the state machine pattern from htmlparser2 5d7a780 (#362)

The token queue was replaced with callbacks (#404, #405, #419)

The OpenElementStack now uses callbacks #429

getNextToken was removed #461

The parser’s _bootstrap method was removed #384

We now drop chunks from the tokenizer right after they are emitted #432