--- In jslint_com@yahoogroups.com, "Cheney, Edward A SSG RES USAR USARC"

<austin.cheney@...> wrote:

>

>

> > Try this regular expression:

> >

> > /url\(((?:[^\)\\]|\\.)*)\)/

> >

> > You would use it like so:

> >

> > string.replace(regexp, "url(\"$1\")");

> >

>

> There are some problems with that.

>

> 1) The regexp does not account for the prior existence of single or double

quote characters that may or may not be applied in a valid manner.



Correct. I attempted to do that with a regular expression, and I'm not sure if

it's even possible. Additional logic will likely be required to check for

already-existing quotes.



> 2) The regexp does not appear to be accounting for open parenthesis with or

without proper escapement.



With the above regular expression, the following would match:



url(foo(bar)



url(foo\(bar)



It could probably be changed to make the above invalid.



> 3) There are many characters that have valid syntax representation in URI and

like wise have valid syntax representation in CSS.  These characters have to be

accounted for in a manner that is not conflicting with CSS syntax and

simultaneously does not alter their application in a URI instance.



The above expression matches any character except a non-escaped ). That means,

that it is extremely liberal in what it accepts. If you want it to be stricter,

that would be possible as well, though it would require more work. Thus, this is

valid (as it should be):



url(ftp://www.foo.com/search?bar=%F0%9D%95%A2%F0%9D%95%A6%F0%9D%95%A9#%E2%81%98)



> As a result I have chosen to use regexp to escape certain necessary characters

into an arbitrary string and perform the next stages of processing using

standard string methods in a loop.



That is fine, just be careful because it is possible (albeit very unlikely) for

your arbitrary string to be found in the URI.

> Try this regular expression:

>

> /url\(((?:[^\)\\]|\\.)*)\)/

>

> You would use it like so:

>

> string.replace(regexp, "url(\"$1\")");

>



There are some problems with that.



1) The regexp does not account for the prior existence of single or double quote

characters that may or may not be applied in a valid manner.

2) The regexp does not appear to be accounting for open parenthesis with or

without proper escapement.

3) There are many characters that have valid syntax representation in URI and

like wise have valid syntax representation in CSS.  These characters have to be

accounted for in a manner that is not conflicting with CSS syntax and

simultaneously does not alter their application in a URI instance.



Regular expression execution is supposed to be one of the fastest forms of

execution in JavaScript against large string blocks because the execution is

performed below the JavaScript interpretation.  However, I have found that

regexp execution is only fast when it is performed outside of any loop

conditions.  When regexp execution is performed inside a loop then you suffer

dire consequences since the interpreter must fire on each loop index and halts

each time for return of independently executed regexp.  The complexity of the

regexp also plays a very big part in this lag as well.  This said if any logic

has to be applied to a regexp match where that logic requires any sort of

looping or if a regexp executes only in response to conditions in a loop then it

is best to not use regexp at all and just use standard string methods in normal

JavaScript loops.  As a result I have chosen to use regexp to escape certain

necessary characters into an arbitrary string and perform the next stages of

processing using standard string methods in a loop.  Another alternative is

nested regexp execution where a replace method searches for a match and the

replacement value fires a function that executes another regexp instruction, but

this method is less flexible and opens complexity to the depth of known test

cases.

Try this regular expression:



/url\(((?:[^\)\\]|\\.)*)\)/



You would use it like so:



string.replace(regexp, "url(\"$1\")");





--- In jslint_com@yahoogroups.com, "Cheney, Edward A SSG RES USAR USARC"

<austin.cheney@...> wrote:

>

> Using JavaScript I am attempting to forcefully inject double quote characters

into the parentheses of url fragments in CSS, so that url(jslint.com) becomes

url("jslint.com").  To keep from breaking complex URIs I am replacing

parenthesis characters that are escaped with a forward slash with an arbitrary

string:

>

> y = y.replace(/\\\)/g, "~PDpar~");

>

> When I am done with my function I need to return the escaped parentheses back

into the code, so logically I am using:

>

> y = y.replace(/~PDpar~/g, "\\\)");

>

> JSLint throws an error when "\)" is present in the second argument of a

replace method.  Is there a string equivalent that can be used which will not

throw a fit?  I understand I could simply throw a function in there which

returns a static string to prevent JSLint from directly associating the

problematic string with the replacement method, but arbitrary function calls is

less efficient for the interpreter.

>

> Thanks,

>

> Austin Cheney, CISSP

> http://prettydiff.com/

>

On 5 Nov 2010, at 14:12:22, Cheney, Edward A SSG RES USAR USARC wrote:



> Using JavaScript I am attempting to forcefully inject double quote characters

into the parentheses of url fragments in CSS, so that url(jslint.com) becomes

url("jslint.com"). To keep from breaking complex URIs I am replacing parenthesis

characters that are escaped with a forward slash with an arbitrary string:

>

> y = y.replace(/\\\)/g, "~PDpar~");

>

> When I am done with my function I need to return the escaped parentheses back

into the code, so logically I am using:

>

> y = y.replace(/~PDpar~/g, "\\\)");

>

> JSLint throws an error when "\)" is present in the second argument of a

replace method. Is there a string equivalent that can be used which will not

throw a fit? I understand I could simply throw a function in there which returns

a static string to prevent JSLint from directly associating the problematic

string with the replacement method, but arbitrary function calls is less

efficient for the interpreter.



var y = "\\)abc\\)fgh";

print(y);

y = y.replace(/\\\)/g, "~PDpar~");

print(y);

y = y.replace(/~PDpar~/g, "\\)"); // leave out the third backslash

print(y);



Output is:



\)abc\)fgh

~PDpar~abc~PDpar~fgh

\)abc\)fgh



Harry.

Using JavaScript I am attempting to forcefully inject double quote characters

into the parentheses of url fragments in CSS, so that url(jslint.com) becomes

url("jslint.com").  To keep from breaking complex URIs I am replacing

parenthesis characters that are escaped with a forward slash with an arbitrary

string:



y = y.replace(/\\\)/g, "~PDpar~");



When I am done with my function I need to return the escaped parentheses back

into the code, so logically I am using:



y = y.replace(/~PDpar~/g, "\\\)");



JSLint throws an error when "\)" is present in the second argument of a replace

method.  Is there a string equivalent that can be used which will not throw a

fit?  I understand I could simply throw a function in there which returns a

static string to prevent JSLint from directly associating the problematic string

with the replacement method, but arbitrary function calls is less efficient for

the interpreter.



Thanks,



Austin Cheney, CISSP

http://prettydiff.com/