222193 - links with variables named similar to HTML entities translated into entities, causing link to be invalid when clicked

Status: RESOLVED INVALID

(Core :: Layout, defect)

Description

[Matt Magoffin]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20030925
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20030925

I have an application that generates URLs with variables named similar to HTML
entities, namely the ® entity. Mozilla seems to translate them into
entities, which causes the links to go to the wrong URL. For example,
"?foo=bar&reg.id=1" gets translated as if the &reg were "®" and the URL
becomes "?foo=bar%AE.id=1".

See http://msqr.us/reg-link.html for example.

Reproducible: Always

Steps to Reproduce:
1. Go to http://msqr.us/reg-link.html for example
2.
3.



Expected Results:  
URL parameter names should not be translated into HTML entities.

Comment 1

[dolphinling]

Not sure about whether or not this is a bug, but I /do/ know you're supposed to
use & when you have an ampersand in a url. The browser then converts that to
the correct url.

Comment 2

[Anne (:annevk)]

Entities in HTML should be escaped. & becomes &. Your URI would become:
http://no.where/something?one=1&two=2&reg.one=r1&reg.two=r2&amp.foo=bar

(I hope this doesn't get converted, shouldn't there be a preview button within
Bugzilla?)

Comment 3

[Boris Zbarsky [:bzbarsky]]

Entity conversion must happen in all attribute values.  Attribute values that
happen to be URLs are not excluded from this.

Comment 4

[Matt Magoffin]

These are *not* entities in the URL. They do not have the final semicolon, and
they are not supposed to be entities. The URL variable "reg.id" is just that, a
variable name, and should not change to "®". Perhaps Mozilla sees the period
and treats that as the closing entity semicolon? Take the example if the
"reg.id" variable was first in the URL, then Mozilla does not translate it:

http://foo.bar/?reg.id=1

But if it is second in the URL, Mozilla will translate it because it follows an
ampersand:

http://foo.bar/?foo=bar&reg.id=1

Comment 5

[Hixie (not reading bugmail)]

Semicolon is optional according to SGML parsing rules.

Comment 6

[Matt Magoffin]

Ugh, sorry I missed Anne'a comment... thanks for clearing up!

Comment 7

[Adam Guthrie]

*** Bug 352811 has been marked as a duplicate of this bug. ***