Google PageRank Checksum Algorithm - Page 2

Unregistered · 06-24-2004, 09:55 AM

I intend to reduce it down to something vaguely more understandable as I get to understand the algorithm better. But this way should at least produce the right output. The problem with doing it this way is that if i make a mistake it's going to a bitch to correct.

Unregistered · 06-24-2004, 10:00 AM

What decompiler did you use? Do you know what flavour assembly it was interpreting in?

--alex (i shoudl register probably)

Unregistered · 06-24-2004, 10:01 AM

...AND what do you wan't it in PHP for anyway?

doctorow · 06-24-2004, 10:16 AM

Decompiler: IDA (Interactive Disassembler)

What do you mean by "flavour assembly"?

PHP can be used on a server, so we can check page's pageranks with a single click. Also, I hate seeing how people sell their "secret" script for hundreds of $$$, considering that Google could any time update their checksum algorithm with releasing a new toolbar.

doctorow · 06-24-2004, 10:16 AM

Btw, you really should register here! Looks better than this "Unregistered" thing...

alexstapleton · 06-24-2004, 04:42 PM

consider me registered. a PHP extension could be made using just the ASM stuff you realise. although that might be complex as it would mean having to hunt down my VS.NET CDs

Unregistered · 06-25-2004, 12:33 PM

Erm. Slight problem. The resulting number appears to be 36 bits long according to the . Yet we only have 32 bit registers. However the first digit is always a 6 (first 4 bits are always 1110) so im guessing theres some extra stuff going on not included in that assembly code (unless i'm being stupid.) For now i'm just going to add 60129542144 (1110 with 32 0's after it) to it, which i'm hoping should work. however if you wouldn't mind investigating the bit which actually produces the URL to see if anything else is going on it might be useful.

alexstapleton · 06-25-2004, 01:54 PM

upon further investigation they do not all start with 1110. they are however all 36 bits long. this is weird isnt it?

doctorow · 06-25-2004, 02:21 PM

The assembler routine above is only the CORE of the algorith used. When the routines returns, you must first convert the content of eax from hex to decimal. After that, you must PREPEND "ch=6" (used by the latest toolbar version) to the number. So if eax return 12345678, converted to is 305419896, and the final checksum is ch=6305419896.

doctorow · 06-25-2004, 02:21 PM

Quote:

Originally Posted by Unregistered

Erm. Slight problem. The resulting number appears to be 36 bits long according to the . Yet we only have 32 bit registers. However the first digit is always a 6 (first 4 bits are always 1110) so im guessing theres some extra stuff going on not included in that assembly code (unless i'm being stupid.) .

You are definitely not stupid

Great guesswork. See my post above!

alexstapleton · 06-25-2004, 02:27 PM

I think i've figured it out. Cunningly, they shuv a 6 on the start of the string. Problem solved.

doctorow · 06-25-2004, 02:45 PM

Cool!

Will you post the finished script here? Best is to use the [ PHP ] [/ PHP ] tags (without spaces) and put the code inbetween. I am sure those people selling the algorithm will be somewhat upset

alexstapleton · 06-25-2004, 04:01 PM

lol its not ready yet! i think some code might be missing. can i talk to you on AIM or anything like that?

doctorow · 06-25-2004, 04:18 PM

I PM you my AIM... will be online for the next 45 mins or so.

alexstapleton · 06-25-2004, 08:31 PM

OK i've kinda hit a wall and I don't know whats wrong. I have implemented a single codepath for the algorithm. I have implemented everything for ciphertexts which are a multiple of 12 in length. I think the answer may be additional post processing which isn't in the ASM code but I can't be sure. If people would please examine my source to see if i'm missing anything (other than the large empty switch) I would be grateful.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Seriously thoughtful Dekker's Algorithm help.	Catire	Lounge	13	03-19-2010 10:03 AM
Bulk Pagerank Checker Script?	SNaRe	Lounge	2	10-22-2006 04:36 PM
Google Toolbar Pagerank Checksum Revealed!	Alexander Turcic	Lounge	5	02-17-2006 08:09 AM
Google Checksum CH calculator	cyberax	Lounge	2	08-17-2004 09:37 PM

06-24-2004, 09:55 AM	#16
Unregistered Nameless Being	I intend to reduce it down to something vaguely more understandable as I get to understand the algorithm better. But this way should at least produce the right output. The problem with doing it this way is that if i make a mistake it's going to a bitch to correct.

06-24-2004, 10:00 AM	#17
Unregistered Nameless Being	What decompiler did you use? Do you know what flavour assembly it was interpreting in? --alex (i shoudl register probably)

06-24-2004, 10:01 AM	#18
Unregistered Nameless Being	...AND what do you wan't it in PHP for anyway?

06-24-2004, 10:16 AM	#19
doctorow Guru Posts: 914 Karma: 3410461 Join Date: May 2004 Device: Kindle Touch	Decompiler: IDA (Interactive Disassembler) What do you mean by "flavour assembly"? PHP can be used on a server, so we can check page's pageranks with a single click. Also, I hate seeing how people sell their "secret" script for hundreds of $$$, considering that Google could any time update their checksum algorithm with releasing a new toolbar.

06-24-2004, 04:42 PM	#21
alexstapleton Enthusiast Posts: 49 Karma: 123 Join Date: Jun 2004	consider me registered. a PHP extension could be made using just the ASM stuff you realise. although that might be complex as it would mean having to hunt down my VS.NET CDs

06-25-2004, 12:33 PM	#22
Unregistered Nameless Being	Erm. Slight problem. The resulting number appears to be 36 bits long according to the . Yet we only have 32 bit registers. However the first digit is always a 6 (first 4 bits are always 1110) so im guessing theres some extra stuff going on not included in that assembly code (unless i'm being stupid.) For now i'm just going to add 60129542144 (1110 with 32 0's after it) to it, which i'm hoping should work. however if you wouldn't mind investigating the bit which actually produces the URL to see if anything else is going on it might be useful.

06-25-2004, 01:54 PM	#23
alexstapleton Enthusiast Posts: 49 Karma: 123 Join Date: Jun 2004	upon further investigation they do not all start with 1110. they are however all 36 bits long. this is weird isnt it?

06-25-2004, 02:21 PM	#24
doctorow Guru Posts: 914 Karma: 3410461 Join Date: May 2004 Device: Kindle Touch	The assembler routine above is only the CORE of the algorith used. When the routines returns, you must first convert the content of eax from hex to decimal. After that, you must PREPEND "ch=6" (used by the latest toolbar version) to the number. So if eax return 12345678, converted to is 305419896, and the final checksum is ch=6305419896.

06-25-2004, 02:27 PM	#26
alexstapleton Enthusiast Posts: 49 Karma: 123 Join Date: Jun 2004	I think i've figured it out. Cunningly, they shuv a 6 on the start of the string. Problem solved.

06-25-2004, 02:45 PM	#27
doctorow Guru Posts: 914 Karma: 3410461 Join Date: May 2004 Device: Kindle Touch	Cool! Will you post the finished script here? Best is to use the [ PHP ] [/ PHP ] tags (without spaces) and put the code inbetween. I am sure those people selling the algorithm will be somewhat upset

Advert

Advert

06-25-2004, 04:01 PM	#28
alexstapleton Enthusiast Posts: 49 Karma: 123 Join Date: Jun 2004	lol its not ready yet! i think some code might be missing. can i talk to you on AIM or anything like that?

06-25-2004, 04:18 PM	#29
doctorow Guru Posts: 914 Karma: 3410461 Join Date: May 2004 Device: Kindle Touch	I PM you my AIM... will be online for the next 45 mins or so.

06-25-2004, 08:31 PM	#30
alexstapleton Enthusiast Posts: 49 Karma: 123 Join Date: Jun 2004	OK i've kinda hit a wall and I don't know whats wrong. I have implemented a single codepath for the algorithm. I have implemented everything for ciphertexts which are a multiple of 12 in length. I think the answer may be additional post processing which isn't in the ASM code but I can't be sure. If people would please examine my source to see if i'm missing anything (other than the large empty switch) I would be grateful.