Questions & Answer Captchas

AlexR

In GSA SER - (Not CB!) I've been sorting out over 1 million of these things today.

I'm not sure if these features are present or not:

I believe it can handle maths captchas (as mentioned on another thread). I see 10 000's of captchas in format "four + 4 =" most of them are the words for numbers 1 to 10, but sometimes they go a little higher. 

To solve 100 000's of more captchas.

FEATURE 1

Can it convert the first 100 letters into words and test for them in applying a maths captcha. 

i.e. convert the "four" to "4" so it can solve "four + 4 ="

FEATURE 2 

Can it convert 

• "plus" to "+" 
• "minus" to "-" 
• "divide" to "/" 
• "subtract" to "-"
• "multiply" to "x" 
• "and" to "+" (like in "4 and 4 = ")
• "multipliziert mit" to "x" (to solve "2 multipliziert mit 8")

i.e. convert the "plus" to "+" so it can solve "4 plus 4 ="

FEATURE 3

Check where the number is missing and then solve. 

+ 1 = 5

1 +  = 5

FEATURE 4

Switch out words for numbers, then check where the number is missing and solve. (i.e. switch out five for 5):

+ 1 = five

4.a - also have top 100 spanish numbers in words to convert.

4.b.- also have top 100 german numbers as words. There are many German captchas. "zwei+zwei=" or "zwei × = 16"

FEATURE 5

If feature 2 works, will it be able to solve, "What is the sum of 9 and 6?"

FEATURE 6

• What number comes before 8? (Change "What number comes before" to "-1+" so it reads "-1+8=")
• What number comes after 9? (Change "What number comes after" to "1+" so it reads "1+9=")
• What is the square root of 81?

Sven

send me the captchas to play with

AlexR

@sven will do..... 

Let me work on a few files and separate out per feature to make it easier. 

AlexR

@sven - here they are:

Feature 1,3,4:

http://pastebin.com/tSm81jvE

Feature 2:

http://pastebin.com/3jqMpQ5s

Feature 5:

http://pastebin.com/Mzw8r3xk

@anyone - can you help me with a regular expression to remove all the url's out of my .txt file using notepad++.

I.e. strip out URL in :

1 × = seven  http://www.wheelsdirect2u.com/saab-95-sedan.htm

theorbital

this is regex that will capture just the url but it won't work in notepad++ for some reason. I think they have some weird things going on with their regex engine.

/(((f|ht){1}(tp|tps)://)[\w\d\S]+)/g

AlexR

@theorbital - for interest I've placed solution here. 

Use notepad++ and regex 

http://.*

Bytefaker

Great input Thanks a lot!

Sven

Oh wait, thats all handled already. Have a look in the file "calc_replace.dat". That holds all replacements done before checking for a math question.

AlexR

@Sven - can it solve the below? I've seen quite a few of these in the unsolved file. 

FEATURE 3

Check where the number is missing and then solve. 

+ 1 = 5

1 +  = 5

I cleared my unknown and found this today which is why I ask???

5 - = oneÿhttp://www.wild-about-travel.com/know-wild-about-travel/

ALSO 

I see very very many in the unsolved that have:

six - = 5 Commentÿhttp://mylifeinapyramid.com/fitness/

It's like the "comment" tab is giving it an issue. 

AlexR

@sven - if I edit this calc_replace file, does it get overwritten on SER updates?

Sven

X + 1 = 5 is not solved as i would have to code some more complex stuff to solve this. No priority for it now...but you are right, it might already be solvable when editing the calc_replace file and add the missing things on my own.

yes the file is overwritten, give me your corrections instead.