fix(parser/html): fix whitespace being lexed as html literal #3908
Conversation
dyc3
force-pushed
the
09-12-fix_parser_html_fix_whitespace_being_lexed_as_html_literal
branch
from
272e65f to
fb07a79
Compare
CodSpeed Performance ReportMerging #3908 will not alter performanceComparing Summary
|
| @@ -150,17 +153,11 @@ fn element() { | |||
| } | |||
|
|
|||
| #[test] | |||
| fn element_with_text() { | |||
There was a problem hiding this comment.
Was this test changed for some reason?
There was a problem hiding this comment.
This test was removed because lexing it correctly requires changing contexts, which is not supported by the assert_lex! macro. It was replaced with other tests that specify the context upfront.
| /// Consume HTML text literals outside of tags. | ||
| /// | ||
| /// This includes text and single spaces between words. If newline or a second | ||
| /// consecutive space is found, this will stop consuming and to allow the lexer to | ||
| /// switch to `consume_whitespace`. | ||
| fn consume_html_text(&mut self) -> HtmlSyntaxKind { |
There was a problem hiding this comment.
Is there a link to the spec that justify having newlines and double spaces parsed differently? It would be useful to have that link in the docs of the function
There was a problem hiding this comment.
I found these and added them to the doc comment:
There was a problem hiding this comment.
My reasoning is that trivia means that the token doesn't affect the functionality of the code. The primary function of an HTML document is to render in a web browser. If you remove a space between words in HTML text, it does not render the document invalid, and it changes how the document is rendered (aka its functionality), and is therefore not trivia. Multiple spaces are collapsed into a single space, so all those extra spaces are trivia.
dyc3
force-pushed
the
09-12-fix_parser_html_fix_whitespace_being_lexed_as_html_literal
branch
from
fb07a79 to
96929a5
Compare
dyc3
force-pushed
the
09-12-fix_parser_html_fix_whitespace_being_lexed_as_html_literal
branch
from
96929a5 to
7df8723
Compare
dyc3
deleted the
09-12-fix_parser_html_fix_whitespace_being_lexed_as_html_literal
branch
Summary
This fixes the HTML lexer treating whitespace trivia as
HTML_LITERALwhen it shouldn't.We end up creating less syntax nodes by treating all HTML text that would affect what is rendered on the screen as part of the same
HTML_LITERAL. See the updated snapshots for examples of how this affects lexing and parsing.This also significantly restructures the HTML lexer, and it should be a bit more straightforward.
Test Plan
Added/updated tests