HTML escape codes are not handled properly for pango markup.
crocket opened this issue · 11 comments
& is converted to &.
< is not converted to <. < is converted to &lt;.
Refer to https://www.freeformatter.com/html-escape.html for details.
<, >, and " aren't converted because they are used as part of <span> tags which allow you to specifically colorize parts of the module's text.
I think i3pystatus shouldn't convert & to & because I want to insert < and so on.
That's fair, I think that the replace code can be altered to skip replacing that text when the & is part of an html escape.
I think it's better and simpler to not expand &.
It is only expanded when the text is processed by pango, because if it is not the processing fails.
I think it's better for users to handle HTML escape codes... because expanding only & is confusing, and this behavior is not even documented.
expanding only
&is confusing
Maybe that's the case for you, but I don't think you're considering the fact that not everyone understands HTML escapes. Many people will simply enable pango for a module because it enables specific colorization effects. For these users, adding a & to the format string will make the entire module's text disappear due to a failure of pango to process the text.
The pull request that I opened:
- Expands
&to prevent confusion for those who don't have a more thorough understanding of pango / HTML escapes - Fixes the bug in which
&is expanded when part of an HTML escape code - Does not prevent one from using
&,<,>, or"
I'm not convinced that it's a great idea to make i3pystatus less intuitive for less-knowledgeable users, but it would be worth it IMO if there were good reasons for & not being expanded (such as it being needed as part of an HTML tag, like <, >, and "). I'm not aware of any such uses, but HTML and CSS are not my strong points. Are ampersands perhaps valid within, for example, a style attribute?
@enkore thoughts?
We're not using something like markupsafe, so we can't easily distinguish what came from the format string and what is e.g. the title of a song that's now playing (and happens to contain a &). I think escaping & when it is not part of an HTML entity is reasonable in this situation.
Oh yeah, I was so focused on the format string itself that I hadn't considered the module output (which is harder to control for).
Whatever. If it works, I'm okay.