May 06 2005
Fun With Google Accelerator
Damien has just introduced me to the Google Web Accelerator and by chance it appears that we have stumbled upon some hidden features. When Damien tried to access the Performance Data for the web accelerator it failed to load the page in Firefox (http://localhost:9100/races) so he tried typing the address manually into Firefox, only he typed http://localhost:9100/race (missing the s off the end).
This typo seemed to take him to a page that allows for entering a url that is then loaded up, on the next page, in two frames. Looking at the src attributes for the frames it appears that the top half is being loaded via the Google Web Accelerator and the lower half is being loaded directly (bypassing the Web Accelerator). Unfortunately there doesn’t seem to be any more information given about this race and the url entry page does say this feature is still under development.
This chance find got me thinking… what other features are being hidden in the black box? And so I did the only sensible thing and started poking around in the files that got installed for the Web Accelerator. Six files were installed into Program Files\Google\Web Accelerator:
- GoogleWebAccClient.exe
- GoogleWebAccFirefox.dll
- GoogleWebAccFirefox.jar
- GoogleWebAccFirefox.xpt
- GoogleWebAccToolbar.dll
- GoogleWebAccWarden.exe
Most of these files are rather boring. GoogleWebAccWarden.exe seems to be just the system tray icon which gives you access to the preferences, performance data, etc. GoogleWebAccToolbar.dll is the plugin for Internet Explorer to display the same menu as the system tray icon and also display the time saved by using the Web Accelerator. GoogleWebAccFirefox.* are the plugin files for Firefox and do exactly the same as the Internet Explorer plugin.
Which leaves us with GoogleWebAccClient.exe. This is where I struck gold. Armed with Editplus I opened up GoogleWebAccClient.exe to see what I could find in the way of resource strings. I already knew what two of the available special urls were for the Preferences and Performace Data so that’s where I started my search. The first hit for preferences took me straight to line 1880, on the same line was races (the url for Performance Data) and race (the url that Damien stumbled upon). On this same line there were also some other urls (along with what I think they’re for/do:
- /proxy.pac (proxy configuration file)
- /mousehover
- /_pagestats
- /afternavigation
- /beforenavigation
- /preferences (proxy preferences configuration)
- /cache
- /preloadingpages
- /pingz
- /races (performance data)
- /advraces (more advanced performance data - including showing which files have loaded slower because of Web Accelerator!!!)
- /benefit (last page access or total benefit from using Web Accelerator)
- /resetbenefit (resets the benefit data)
- /ticker
- /race (comparison between direct and Web Accelerator loading of a web page
- /reconnect
- /exclude
- /is_excluded
I sure there is a whole host of things that can be done with this information but this is all that I’ve managed to find out so far.
Nice. Did you tweak anything? I mean, did GWAClient.exe work after the tweak at all, better or worse?
Now I’m into prototyping a next generation of Web browser. I call it Googler, and I want to base it on Google Earth. Patrick of splintered.co.uk made the first step in that prototyping, namely GeoURL-to-GoogleEarth transformation, very cool utility.
Next obvious step would be hack of Google Earth address input form, so it could accept URLs/IP addresses, not just postal addresses.
This is where your skills at reverse engineering GWA might shine, could you tweak Google Earth in this manner?
Some details of the Googler project can be found at my blog http://goolocalizations.blogspot.com , have a look.
Please answer or email me with any questions.
I’m afraid that I got no further with Google Accelerator than what you see on this page. Shortly after installing Google Accelerator I got hacked off with it as it has a tendancy to kill websites.
I’m actually very inexperience when it comes to reverse engineering so I fear that I am unable to help you with your project ideas. I must say, however, that it sounds very interesting and I shall be keeping an eye on it to see what you make of it.
tendency to kill websites? gee, that’s something new for a chain of gzipping cacheing proxies. I doubt that, though yes, they practice some killing, but exclusively on themselves: US SF/BayArea proxies get chocked several times a day.
Several of my friends that run websites off their home DSL connections have had problems with Google eating up their connections.
dddd
i think Google’s Accelerator is useless. http://www.mytailorstore.com is my site.the time is installed google Accelerator i was unable to open my site but from other computer yes.but then i uninstall that then i again i was able to open my site.so if it works like that then i think the people who have this program installed they are unable to open not only mine site but of others too.
Useless. I’ve been using it and its save me a whopping 2.9 minutes for 2169 pages and 1.7 hrs or 2.8%. I’m removing it.