CB hangs when editing drupal 10 chars captcha and other drupal captcha problems/bugs

hans51

When editing drupal 10 chars captcha, CB usually needs a minute or longer for every single step of the work = EACH of following steps=minutes or hanging:
edit, use mask = calculate, test, save=closing window at the end and repeatedly hangs.
then even win7 task-manager unable to shutdown CB.

hence ONE single captcha editing requires 5-10 minutes or hang with reboot of machine

furthermore, I wanted to test connection from SER to CB during one such hanger, then SER hangs as well.

recently I have done repeated drupal captcha and never noticed such behavior for smaller sizes but ALWAYS for the most frequent captcha type I see in my drupal submission = 10 chars captcha. CB same work for 9 chars captcha absolutely normal - hanging or excessive time only for 10 chars drupal type.

all above is from using CB load/test captcha from "unsolved" folder

above in addition to above problems, there also are auto-saved unsolved captcha in drupal section that belong to a totally different category (for example plogger image comment, php link directory, guestbook, ...)
see my recent post about "contamination" of captcha folders
https://forum.gsa-online.de/discussion/5899/a-feature-to-prevet-captcha-folder-contamination#latest

and some of the "drupal 10 chars" subfolders only have captcha with 8 chars ALL

in addition, when doing CB load/test of a whole unsolved drupal 10 chars sub-folder, some of the 10 chars captcha are "solved" using "drupal 4 chars" or "drupal 5 chars"  = may be resulting in login problems to drupal sites ???

Sven

Well drupal is different, it has many variations and sizes. 

The lag might be due to the program solving captchas for SER while you edit it? For the other problem I might need you to send me the captchas.

hans51

to do such work I have to stop ALL activity on that machine, = NO other work going on
hence the extreme lag up to repeated machine hanging ONLY is for 10 char drupal
and is for EVERY 10 char drupal w/o exception (updated win7 OS)

the "other" being what ?
wrong captcha in size-folder
or wrong captcha type in drupal folder ?

it appears to be in unsolved section as well as in folders of "captcha_systems"
thus I am quiet sure you also find it on your own machine

hans51

Up to drupal 6 chars captcha, the time to process = solve or NOT-solve is typically far below 1 second
but
7 chars and up are FAR slower
7 chars = 9 second
10 chars = 19-20 second

with 3 retries  (total 4 captcha to solve by CB)
for a 10 chars captcha that would occupy CB for 80 second

many of the good quality drupal submission have 7-10 chars

hence a drupal submission or multiple within a minute, using large 7-10 chars captcha eventually may occupy the system or CB to the extent that all other submissions MAY BE (?) FAIL or time out ...

I would assume that a captcha type entirely solvable with mask to a high % should run below 1 second on all lengths of captcha offered.

hence there may be a bug in the configuration settings (file Drupal 7-10) ?? or else where?

hans51

one of the many examples of wrong processing of drupal captcha sizes is
http://tinpahar.com/?q=user/register

CB tries repeatedly to solve this as 5 chars (dirty) drupal captcha
which of course always leads to wrong result - CB sometimes just invents a 5 captcha using OCR or simply fails using mask

if you have a look into your various unsolved drupal subfolders, there are some large size (many chars) subfolders with ample wrong sizes (in one case lots of 3 chars) in it

see for exaqmple YOUR unsolved > drupal > 8 chars > 216x72
contains (mostly/many) drupal 4, 5 r 6 chars and almost none 8 chars captcha

while the folder unsolved > drupal > 8 chars > 360x60
contains (mostly) 10 chars catpcha

or another
unsolved > drupal > 6 chars > 135x60
contains approx half 3 chars captcha besides 6 chars

hans51

2 questions @sven:

If for drupal captcha a near 100% of captcha can be solved by MASK of chars only (with exception of may be "fake drupal" captcha),
can I DISABLE use of OCR in the Drupal config file present in each drupal sub-folder for each size ??

Is it possible that ONE single chars mask may be valid and working for ALL different sizes - 2-10 chars ??

one of the main work is to get mask values for all characters [a-z,, A-Z] and all numeric and special chars @#$%&*() and others used by drupal NOT even in my keyboard

if drupal accidentally uses same mask for all sizes of captcha from 2-10 chars,
then we could either change config or copy/past one complete mask into each Drupal config file ... ??
just a idea from a non-pro / simple user

Sven

Hmm usually I edit the files by hand and copy/paste the masks into the other files. 

hans51

@sven

I see some HUGE difference in mask sizes between lower chars sizes and large ones such as 10 chars.

10 chars appear to be multiple time the size of 5 chars masks

and

I see that each Drupal.ini has some size specific values as listed below


what happens if we create ONE Drupal.ini for 3-10 chars
and adapt config values such as

;how short/long is the word that the captcha is delivering
min length=3
max length=10


MAY BE the following has to do with the use mask ??

;instead of using OCR we make a crc32 of the image and compare it with the one we have stored
use fixed results=0
use min size=0
min size=1098
use max size=0
max size=4321

and needs to be adapted as well - may be something like

;instead of using OCR we make a crc32 of the image and compare it with the one we have stored
use fixed results=0
use min size=0
min size=1098
use max size=0
max size=13496
use mask chars=1


and other modifications such as image dimension= etc to put ALL drupal captcha into ONE Drupal.ini

???

The last 3-4 days I have spent substantial time to change, try (YOU are the creator and know all about your creation = me just kind of reverse-enginiring and trial and error.)

the result so far seems a substantially higher successful drupal article submission.
still the vast MAJORITY of all drupal sites FAIL only because of the a.m. documented size-mix up of various drupal captcha types.

one other question about the mask values for all characters and symbols:

1. s there a technical way to calculate the values WITHOUT the use of CB to speed up as to avoid mistakes due to poorly readable captcha??
2. why is a 10 char mask for each char MUCH longer than for example a 5 char mask??
3. if captcha are solved by mask, can we completely omit the time and resource consuming use of OCR and such processes as often listed at botom of each Drupal.ini
   

[PROCESSING1]
...

[PROCESSING2]
...

[PROCESSING3]
...

or is OCR still needed for solving by mask?

4. in order to have better quality and success control, I would love to EMPTY all subflders of UNSOLVED > drupal
to see what is currently still unsolved.

is CB recreating all subfolders of unsolved > drupal as needed
or
do I have to keep all subfolders as is and just delete folder content???

since drupal captcha of newer generation include all special characters (including many NOT on my keyboard and on NO online docu found such as similar to a mirror of 

3

(NOT EURO symbol!!!) = I never found that symbol online but it exists in newer drupal captcha

it may be advised to add by default all new chars to all Drupal.ini PLUS the a.m. "mirror" of 3 whatever that is

charset=123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz!#$%*+-?@:;(){}&§


Your help and replies to above open question sincerely appreciated for further testing and captcha improvements

Sven

Hard to read post but I try my best...

Making one drupal ini out of the severals might work as long as the image looks the same but sizes and captcha length differ. 

>EURO:

True, there are probably others that are not in the charset box. But I prefer not to add the once that are too uncoman and not used much. That keeps the false detection of OCR low.

>subfolders:

CB creates them for you...no need to do that

>EMPTY all subflders

next version has this added

hans51

@sven
I am aware that above hard to read = it is about problems created when making drupal engine and Drupal.ini system = too many mistakes in entire current drupal processing = too cost and resources expensive

the last 2 days I have removed drupal from most projects for that reason.

too many (MOST)  of open questions above without answer,

best is to postpone solving above until more ppl NEED or WANT drupal and until time available to study a.m. outlined built-in problems until understood.

RayBan

@hans51 have you found any solution for this ? I scraped large list of drupal sites, but only very few of them can be solved by captchabraker.

Not sure, if I should apply for additional 3rd party service or just try to scrape other footprints.

hans51

@rayban
no - not yet
had no time to make more efforts
I have fairly good results on drupal but limited to 7 or less chars only

it is either a CB issue on how CB calculates
or
it simply is an issue of lacking CPU power

unfortunately I no longer have my powerful workstation but only a small acer aspire one with atom

the plain CPU processing time of10-20 seconds is simply too long for any testing while working on same machine
a total of many minutes for ONE single captcha solving NEW captchas requires a powerful machine with ultra fast quad CPU

however I may work on this matter again in near future as soon as other more important SER efficiency issues on my end of the world are solved.

meanwhile I use antigate as human captcha solving for drupal
it costs $ but brings results until CB can solve more
drupal is an excellent blog platform for quality posts and it surely is worth to make some efforts in solving more by CB