Web Inspector: Allow retrieval of all transmitted cookies

[Alexander Pavlov (apavlov)]

Currently there is no way to retrieve all cookies in Web Inspector, which is required for audits.

[Alexander Pavlov (apavlov)]

Created attachment 44318 [details]
patch

The approach has been discussed with [e-mail]

[WebKit Review Bot]

style-queue ran check-webkit-style on attachment 44318 [details] without any errors.

[Pavel Feldman]

Comment on attachment 44318 [details]
patch

> +        for (var c = 0; c < allCookies.length; ++c) {

c is a strange index name.

> +            for (var id in WebInspector.resources) {
> +                var resource = WebInspector.resources[id];
> +                var match = resource.documentURL.match(WebInspector.URLRegExp);

You don't need to do matching numCookeis x numResources times. It is easier to go though resources and store their domains into a set first (numResources steps), then you can get all cookies for given frame within numCookies x numDomains steps.

>  
> +WebInspector.Cookies.cookieMatchesResourceURL = function(cookie, resourceURL)
> +{

I don't see this method used.

> +
> +WebInspector.Cookies.cookieDomainMatchesResourceDomain = function(cookieDomain, resourceDomain)
> +{
> +    if (cookieDomain.charAt(0) !== '.')
> +        return resourceDomain === cookieDomain;
> +    return !!resourceDomain.match(new RegExp("^([^\\.]+\\.)?" + cookieDomain.substring(1).escapeForRegExp() + "$"), "i");

A test for this would be nice. You need to do a inspector test that would match a bunch of URLs against given domain on the frontend side.

[Alexander Pavlov (apavlov)]

(In reply to comment #3)
> (From update of attachment 44318 [details])
> > +        for (var c = 0; c < allCookies.length; ++c) {
> 
> c is a strange index name.

i now

> > +            for (var id in WebInspector.resources) {
> > +                var resource = WebInspector.resources[id];
> > +                var match = resource.documentURL.match(WebInspector.URLRegExp);
> 
> You don't need to do matching numCookeis x numResources times. It is easier to
> go though resources and store their domains into a set first (numResources
> steps), then you can get all cookies for given frame within numCookies x
> numDomains steps.

We should store full URLs rather than domains (since the cookies are matched using the protocol, hostname, port, and path.) I have changed the impl, so we can gain some benefit when there are several cookie domains in the page.

> >  
> > +WebInspector.Cookies.cookieMatchesResourceURL = function(cookie, resourceURL)
> > +{
> 
> I don't see this method used.

Thanks for the catch. For some reason I was invoking domain matching rather than full matching.

> > +
> > +WebInspector.Cookies.cookieDomainMatchesResourceDomain = function(cookieDomain, resourceDomain)
> > +{
> > +    if (cookieDomain.charAt(0) !== '.')
> > +        return resourceDomain === cookieDomain;
> > +    return !!resourceDomain.match(new RegExp("^([^\\.]+\\.)?" + cookieDomain.substring(1).escapeForRegExp() + "$"), "i");
> 
> A test for this would be nice. You need to do a inspector test that would match
> a bunch of URLs against given domain on the frontend side.

Done.

[Alexander Pavlov (apavlov)]

Created attachment 44417 [details]
patch (fixed)

[WebKit Review Bot]

style-queue ran check-webkit-style on attachment 44417 [details] without any errors.

[Alexander Pavlov (apavlov)]

Created attachment 44421 [details]
patch (add comment + use match group)

[WebKit Review Bot]

style-queue ran check-webkit-style on attachment 44421 [details] without any errors.

[Alexander Pavlov (apavlov)]

Created attachment 44467 [details]
patch (test cleanup)

[WebKit Review Bot]

style-queue ran check-webkit-style on attachment 44467 [details] without any errors.

[WebKit Commit Bot]

Comment on attachment 44467 [details]
patch (test cleanup)

Clearing flags on attachment: 44467

Committed r51846: <http://trac.webkit.org/changeset/51846>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.