The format of a URI value is 'url(' followed by ...
だからやっぱだめ?
[17]Firefox 2.0's implementation:
Let u be an empty string.
Let invalid be false.
Unless the string starts by an ident with value
url (case-insensitive,
escape is allowed,
\ followed by a newline
is ignored), immediately followed by a
(, it is not a URI.
Stop these steps.
(WS) Skip any white space characters.
If the next two characters are /*,
Skip any character until */ is found.
If the substring */ is found,
If the following characters are not a white space
character or /*,
set invalid as true. Don't consume these
characters yet.
Go to the previous step in the overall steps (i.e. WS).
Otherwise, i.e. if the next character is EOF,
the URI is invalid. Stop the entire steps.
If the next character is ",
Appends any character until "
to u, with any escape decoded and any
\ followed by a newline
ignored.
(DWS) If the next character is a white space,
set u as that character (!).
If next few characters are white spaces, skip them.
If the next character is /*,
Skip any character until */.
If the next character is EOF,
the URI is inlvaid. Stop the entire steps.
Otherwise, go to the step named as DWS.
If the next character is ),
go to the step named as CP.
Otherwise, set invalid as true.
Skip any character until )
(including ) following \
but excluding ) after "
or ' but before "
or ' matching it).
If the next character is EOF,
the URI is invalid. Stop the entire steps.
Otherwise, go to step named as CP.
If the next character is ',
Appends any character until '
to u, with any escape decoded and any
\ followed by a newline
ignored.
Go to the step named as DWS.
(CP) If the next character is ),
If invalid is true, then the
URI is invalid. Stop the entire steps.
Otherwise, return a URI whose
value is u]].
If the next character is " or ',
set invalid as true.
If the next character is a white space,
Skip any white space characters.
If the next character is ),
go to the previous step in the overall steps (i.e. CP).
Otherwise, set invalid as true.
If the next character is \,
decode a few character as escape
and append it to u, or skip whitespace
if it immediately follows the \.
If the next character is EOF,
the URI is invalid. Stop the steps.
Go to the step named as CP.
[18]Opera 9.00 seems to parse URI
as specified in CSS 2.1.
For example,
http://suika.fam.cx/gate/2007/cssom/viewer?c=p%20%7B%0D%0A%20%20background-image%3A%20url(%20aa%3B%20font-size%3A%2012px%3B;h=;p=n;x=style-element
results in a font-size property
specified as 12px.
However, it post-process URI values as:
Drop any #]] and following characters.
Drop any preceding and following white space characters.