Found old paper with password hints... btcrecover help w/ python errors

[b2352351]

I found an old paper with some hints on it. There's numbers, words, letters, sequence of digits etc.. I also left my self a hint that part of the password will be caps lock(or only one letter in the sequence is caps).

I'm trying to put together lists of possible combinations. The password is broken down into 3 sections. 4 letters, 3 numbers/symbols, and 4 letters. So I want to have btcrecover guess all possible combinations. I'm an idiot and "encrypted" myself a message on a piece of paper. I have quite a few ideas on ways I could have encrypted the message in 2013 though.

I am trying to run the tests with GPU enabled. The open CL tests were skipped even though I've installed requirements for the GPU acceleration.

I tested the install by using

Code:

python run-all-tests.py -vv 

Code:

FAILED (errors=20, passes=319, skips=75)

Skips and errors consist of

Code:

-- requires OpenCL and a compatible device and 

-- requires coincurve 

-- requires ecdsa 

//and a few others.

Since I'm using bitcoin core and wallet.dat I'm pretty sure I just need the OpenCL and compatible device tests to pass correct? I believe everything required for bitcoin core has been installed correctly. So how do I enable GPU and get it to test and pass?

Using Windows 10 and bitcoin core. Any help would be appreciated.

[PX-Z]

I dont have any technical knowledge on how you can do it with the tools but I can only suggest you to a right person or service that might help you. You can try this service of walletrecoveryservices^[1], this service have lots of good reviews. Just take note he charge fees for X% of the total funds you want to recover.

[1] https://bitcointalk.org/index.php?topic=240779.0

[b2352351]

I dont have any technical knowledge on how you can do it with the tools but I can only suggest you to a right person or service that might help you. You can try this service of walletrecoveryservices^[1], this service have lots of good reviews. Just take note he charge fees for X% of the total funds you want to recover.

[1] https://bitcointalk.org/index.php?topic=240779.0

I did reach out again but I have not heard back after notifying him I found some password hints. I also read somewhere that he is only accepting wallets with a certain amount of coin in them but I'm not sure if it's still up to date.

[o_e_l_e_o]

It looks like you just haven't been successful in installing the necessary dependencies. You are right that you don't need coincurve or ecdsa libraries for recovering a Bitcoin Core wallet. Then OpenCL libraries aren't installed by default when you install btcrecover, and you need to install them separately. What did you do to try to install them? Are you sure you have installed the right versions for your version of Python?

I'm not really familiar with Windows,* but this section of the documentation should talk you through getting it installed: https://btcrecover.readthedocs.io/en/latest/GPU_Acceleration/#pyopencl-installation-for-windows

*Do you have a Linux machine you can run this on instead? In my experience it generally works far more smoothly.

[PawGo]

btcrecover in gpu mode works on Windows 10 without any problems, just follow the instruction how to install additional dependencies.
But let's move that aside. Are you sure you know how to read/understand the information you have?
Are you able to configure btcrecover to decrypt the data you have? I would first focus on proper configuration for your problem (to create a token file etc), not on GPU.
First try to launch in on CPU, maybe the performance will be good enough.

[b2352351]

It looks like you just haven't been successful in installing the necessary dependencies. You are right that you don't need coincurve or ecdsa libraries for recovering a Bitcoin Core wallet. Then OpenCL libraries aren't installed by default when you install btcrecover, and you need to install them separately. What did you do to try to install them? Are you sure you have installed the right versions for your version of Python?

I'm not really familiar with Windows,* but this section of the documentation should talk you through getting it installed: https://btcrecover.readthedocs.io/en/latest/GPU_Acceleration/#pyopencl-installation-for-windows

*Do you have a Linux machine you can run this on instead? In my experience it generally works far more smoothly.

I could spin up a linux VM or run it on my main PC but if its possible on windows i'd like to avoid all that

I have had a few errors throught the proccess... This is what happens why trying to install the OpenCL libraries.. I might have forgot to include this in original post..

Code:

C:\REDACTED\REDACTED\Documents\btcrecover-master (2)\btcrecover-master>pip3 install[b] pyopencl-2021.2.9+cl12-cp310-cp310-win_amd64.whl[/b]
Processing c:\REDACTED\REDACTED\documents\btcrecover-master (2)\btcrecover-master\pyopencl-2021.2.9+cl12-cp310-cp310-win_amd64.whl
Requirement already satisfied: appdirs>=1.4.0 in c:\REDACTED\REDACTED\appdata\local\programs\python\python310\lib\site-packages (from pyopencl==2021.2.9+cl12) (1.4.4)
Collecting pytools>=2021.2.7
  Using cached pytools-2022.1.4.tar.gz (68 kB)
  Preparing metadata (setup.py) ... done
Requirement already satisfied: numpy in c:\REDACTED\REDACTED\appdata\local\programs\python\python310\lib\site-packages (from pyopencl==2021.2.9+cl12) (1.22.3)
Requirement already satisfied: platformdirs>=2.2.0 in c:\REDACTED\REDACTED\appdata\local\programs\python\python310\lib\site-packages (from pytools>=2021.2.7->pyopencl==2021.2.9+cl12) (2.5.1)
Requirement already satisfied: typing_extensions in c:\REDACTED\REDACTED\appdata\local\programs\python\python310\lib\site-packages (from pytools>=2021.2.7->pyopencl==2021.2.9+cl12) (4.2.0)
[b]Using legacy 'setup.py install' for pytools, since package 'wheel' is not installed.[/b]
Installing collected packages: pytools, pyopencl
  [b]Running setup.py install for pytools ... error
  error: subprocess-exited-with-error[/b]

  × Running setup.py install for pytools did not run successfully.
  │ exit code: 33947949
  ╰─> [28 lines of output]
      running install
      running build
      running build_py
      creating build
      creating build\lib
      creating build\lib\pytools
      copying pytools\batchjob.py -> build\lib\pytools
      copying pytools\codegen.py -> build\lib\pytools
      copying pytools\convergence.py -> build\lib\pytools
      copying pytools\datatable.py -> build\lib\pytools
      copying pytools\debug.py -> build\lib\pytools
      copying pytools\graph.py -> build\lib\pytools
      copying pytools\lex.py -> build\lib\pytools
      copying pytools\log.py -> build\lib\pytools
      copying pytools\mpi.py -> build\lib\pytools
      copying pytools\mpiwrap.py -> build\lib\pytools
      copying pytools\obj_array.py -> build\lib\pytools
      copying pytools\persistent_dict.py -> build\lib\pytools
      copying pytools\prefork.py -> build\lib\pytools
      copying pytools\py_codegen.py -> build\lib\pytools
      copying pytools\spatial_btree.py -> build\lib\pytools
      copying pytools\stopwatch.py -> build\lib\pytools
      copying pytools\tag.py -> build\lib\pytools
      copying pytools\version.py -> build\lib\pytools
      copying pytools\__init__.py -> build\lib\pytools
      copying pytools\py.typed -> build\lib\pytools
      running install_lib
      running install_egg_info
      [end of output]

[b]  note: This error originates from a subprocess, and is likely not a problem with pip.
error: legacy-install-failure[/b]
[b]
× Encountered error while trying to install package.
╰─> pytools
[/b]
note: This is an issue with the package mentioned above, not pip.
hint: See above for output from the failure.

Few question,
1. Is it right to assume you've followed these installation guide (https://btcrecover.readthedocs.io/en/latest/INSTALL/ and https://btcrecover.readthedocs.io/en/latest/GPU_Acceleration/)? Did you install PyOpenCL which match your computer environment?

Yes but see above for the error I forgot to include while installing OpenCL

2. What GPU do you use?

Nvidia 1660

Are you able to configure btcrecover to decrypt the data you have? I would first focus on proper configuration for your problem (to create a token file etc), not on GPU.
First try to launch in on CPU, maybe the performance will be good enough.

It seems like I am missing a complete understanding of the usage commands of btcrecover. I saw one example but didn't really understand it and how to apply it to my instance.

I did try this test though and go the following errors:

Code:

C:\REDACTED\REDACTED\Documents\btcrecover-master (2)\btcrecover-master>python3 -m btcrecover.test.test_passwords -v GPUTests
Traceback (most recent call last):
  File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.10_3.10.1264.0_x64__qbz5n2kfra8p0\lib\runpy.py", line 187, in _run_module_as_main
    mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
  File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.10_3.10.1264.0_x64__qbz5n2kfra8p0\lib\runpy.py", line 110, in _get_module_details
    __import__(pkg_name)
  File "C:\REDACTED\REDACTED\Documents\btcrecover-master (2)\btcrecover-master\btcrecover\__init__.py", line 2, in <module>
    from .btcrpass import *
  File "C:\REDACTED\REDACTED\Documents\btcrecover-master (2)\btcrecover-master\btcrecover\btcrpass.py", line 47, in <module>
    from Crypto.Cipher import AES
ModuleNotFoundError: No module named 'Crypto'

How would I go about launching a crack attempt with cpu? I dont exactly have a world list of thousands.. It will probably be 3 to 8 world lists with a couple dozen words in each. Not sure how many cracks that would take.

[b2352351]

UPDATE:

Read somewhere that the legacy pytools error installing was due to antivirus.. I disabled antivirus and pytools seems to be installed correctly now!

Code:

  note: This error originates from a subprocess, and is likely not a problem with pip.
error: legacy-install-failure

× Encountered error while trying to install package.
╰─> pytools

note: This is an issue with the package mentioned above, not pip.
hint: See above for output from the failure.

C:\Users\REDACTED\Documents\btcrecover-master (2)\btcrecover-master>pip3 install pyopencl-2021.2.9+cl12-cp310-cp310-win_amd64.whl
Processing c:\users\REDACTED\documents\btcrecover-master (2)\btcrecover-master\pyopencl-2021.2.9+cl12-cp310-cp310-win_amd64.whl
Collecting pytools>=2021.2.7
  Using cached pytools-2022.1.4.tar.gz (68 kB)
  Preparing metadata (setup.py) ... done
Requirement already satisfied: numpy in c:\users\REDACTED\appdata\local\programs\python\python310\lib\site-packages (from pyopencl==2021.2.9+cl12) (1.22.3)
Requirement already satisfied: appdirs>=1.4.0 in c:\users\REDACTED\appdata\local\programs\python\python310\lib\site-packages (from pyopencl==2021.2.9+cl12) (1.4.4)
Requirement already satisfied: platformdirs>=2.2.0 in c:\users\REDACTED\appdata\local\programs\python\python310\lib\site-packages (from pytools>=2021.2.7->pyopencl==2021.2.9+cl12) (2.5.1)
Requirement already satisfied: typing_extensions in c:\users\REDACTED\appdata\local\programs\python\python310\lib\site-packages (from pytools>=2021.2.7->pyopencl==2021.2.9+cl12) (4.2.0)
Using legacy 'setup.py install' for pytools, since package 'wheel' is not installed.
Installing collected packages: pytools, pyopencl
  Running setup.py install for pytools ... done
Successfully installed pyopencl-2021.2.9+cl12 pytools-2022.1.4

Time to test the rest...

EDIT:

It looks like its doing something with opencl.. It's stuck on the   test_blockchain_second_OpenCL_Brute Test..

EDIT:

I restarted the machine and reinstalled pip3 install pyopencl-2021.2.9+cl12-cp310-cp310-win_amd64.whl AND pip3 install -r requirements.txt


Now running python run-all-tests.py -vvand it looks like it passed the test_blockchain_second_OpenCL_Brute test notating green text!

I'm onto something boys! Now I just don't know how to configure tokenlist and get that running...

[PawGo]

Forget about GPU for now, focus on correct configuration.
We may help you build the token file, but you should tell clearly what do you know / what do you suspect.
Do you know address for which you want to find private key?
How the private key is build - is it kind of "brain wallet"? Or what you have is a password for custodial service/exchange?
How the text is composed? From your first post I understand it could be:

Code:

Uppercase letter
Uppercase or lowercase letter
Uppercase or lowercase letter
Uppercase or lowercase letter
1234567890 or symbols : !#$%&^*? etc...
1234567890 or symbols : !#$%&^*? etc...
1234567890 or symbols : !#$%&^*? etc...
Uppercase letter
Uppercase or lowercase letter
Uppercase or lowercase letter
Uppercase or lowercase letter

Right?
just 11 characters or something extra in the middle, like dash (-) between sections?

[b2352351]

Forget about GPU for now, focus on correct configuration.
We may help you build the token file, but you should tell clearly what do you know / what do you suspect.
Do you know address for which you want to find private key?
How the private key is build - is it kind of "brain wallet"? Or what you have is a password for custodial service/exchange?
How the text is composed? From your first post I understand it could be:

Code:

Uppercase letter
Uppercase or lowercase letter
Uppercase or lowercase letter
Uppercase or lowercase letter
1234567890 or symbols : !#$%&^*? etc...
1234567890 or symbols : !#$%&^*? etc...
1234567890 or symbols : !#$%&^*? etc...
Uppercase letter
Uppercase or lowercase letter
Uppercase or lowercase letter
Uppercase or lowercase letter

Right?
just 11 characters or something extra in the middle, like dash (-) between sections?

UPDATE:

I restarted the machine and reinstalled pip3 install pyopencl-2021.2.9+cl12-cp310-cp310-win_amd64.whl AND pip3 install -r requirements.txt


Now running python run-all-tests.py -vvand it looks like it passed the test_blockchain_second_OpenCL_Brute test notating green text!

I'm onto something boys! Now I just don't know how to configure tokenlist and get that running...

But on the tokenlist subject. Yes I do need help with that and deploying those commands

The password hints I have are as follows:

3 letters, followed by 4 characters and 3 letters or numbers.. a little complicated I know.

for example

xxx 1:23 xxx

I believe I found another piece of paper that has even more hints, or its just the correct piece of paper and the other had decoy hints(such as the first letter of certain words were capitalized in a sequence of words, this could be relevant or irrelevant but I will program this in once i figure out how to make the token list and deploy that command.

In the end I believe both pieces of paper do have relevant parts of the puzzle(caps or holding shift on the middle sequence of characters etc in addition to the other sequences of 10 characters).

This second piece of paper also has a 10 digit sequence of numbers and letters. 10 digits I believe was the minimum length for bitcoin core at the time. This 10 digit sequence also is somewhat related to the other piece of paper because one character is notated in such a way that it possibly relates to a certain character in the xxx 1:23 xxx 10 digit sequence on the first piece of paper

Also, there is 2 digits that really say 23 in the first part of the puzzle and then on the second piece of the paper if you line up the two codes on top of each other it changes to 33... then repeats at the end... 33 which makes me think I have to correct this part of the password.


I have a few things I believe could be the "xxx" or 1:23 part of the password

It's either the first capitalized letters in a word sequence I wrote down... Or on the new piece of paper it's 3 letters either Abc or abc followed by the 1:23 middle part (this part may be all caps or partial caps). then followed by D11.

I have redacted actual characters and numbers from these hints. but the numbers are where numbers should be and letters where letters should be in appropriate Capitals in each place. in the middle of 1:23 the : is actually what is encoded in my password though. But this : could represented by another special character noted in such a way on the first piece of paper.

To make things a little more complicated I was messing around with some basic encryption algorithms at the time. So a second attempt could possible be made by adding +1 +2 +3 to some of the number/letter sequences. So +1 would turn 1 into 2 and +2 would turn 2 into 4 and +3 would turn 3 into 6.

[o_e_l_e_o]

Good job getting it working. We can help you make the tokenlist, but you are going to have to be much more specific. From what you've said it still isn't clear what format your password might take.

3 letters, followed by 4 characters and 3 letters or numbers.. a little complicated I know.

In your first post you said 4-3-4. Now you are saying 3-4-3. You'll need to decide which it is.

Picking out the important points from the post you just made:

xxx 1:23 xxx
It's either the first capitalized letters in a word sequence I wrote down... Or on the new piece of paper it's 3 letters either Abc or abc
followed by the 1:23 middle part
then followed by D11
But this : could represented by another special character

From this, I deduce your password looks like this:

3 letters, the first of which may or may not be capitalized
1 number
1 special character
2 numbers
1 uppercase letter
2 numbers

Please correct me if this is wrong. Further, are there spaces in your password, or have you just put them in your post for formatting/clarity purposes?

[b2352351]

Good job getting it working. We can help you make the tokenlist, but you are going to have to be much more specific. From what you've said it still isn't clear what format your password might take.

3 letters, followed by 4 characters and 3 letters or numbers.. a little complicated I know.

In your first post you said 4-3-4. Now you are saying 3-4-3. You'll need to decide which it is.

Picking out the important points from the post you just made:

xxx 1:23 xxx
It's either the first capitalized letters in a word sequence I wrote down... Or on the new piece of paper it's 3 letters either Abc or abc
followed by the 1:23 middle part
then followed by D11
But this : could represented by another special character

From this, I deduce your password looks like this:

3 letters, the first of which may or may not be capitalized
1 number
1 special character
2 numbers
1 uppercase letter
2 numbers

Please correct me if this is wrong. Further, are there spaces in your password, or have you just put them in your post for formatting/clarity purposes?

There shouldn't be any spaces. I was just using a space to separate the sequences as an example..

Its actual 3-4-3 not 4-3-4. My bad.

first three letters of passwords should be 3 letters. either Abc, abc or the first three capitals of each word in the word sequence such as EFG. This could also correlate to the xxx on the last part of the password.

Next 4 characters of the password are either 1:23, 1:33 or Z:33 or Z33... There could also be the shift/caps key applied to this part of the sequence..

Last 3 characters of the password could be Abc, abc, EFG or D33.. Also the 33 in the D33 could be while holding the shift key ex - D##

I know its kind of complicated and it makes it a little more complicated that I have to change the letters and numbers around all while attempting to keep the same format. Some parts of my password such as the 23, 33 and : are original components though.

If you could maybe give me some formatting tips for the tokenlist then I could work more in depth on modifying that to my password rules.

I am more focused now on trying to deploy the crack command after passing opencl tests and getting pytools installed correctly.

[o_e_l_e_o]

Ok, that's helpful. However, the additional uncertainty will greatly increase the search space.

So the first part we are now saying is 3 letters, which could either be all uppercase, all lowercase, or one uppercase followed by two lowercase.

The second part becomes tricky:

Next 4 characters of the password are either 1:23, 1:33 or Z:33 or Z33... There could also be the shift/caps key applied to this part of the sequence..

So definitely 4 characters? Or it might only be 3? The first character could be a number or a letter (either case)? This might be followed by a symbol, but might not be? Then any two numbers or symbols? Or is it specifically one of the following:

Code:

23
33
@#
##

Or perhaps your keyboard doesn't have the @ and # symbols above 2 and 3?

Then the last three characters could be 3 letters (all uppercase, all lowercase, or one uppercase followed by two lowercase) OR an uppercase letter followed by 2 numbers or 2 symbols? Again, is this definitely 33 or ##, or could it be any two numbers or symbols?

[b2352351]

Ok, that's helpful. However, the additional uncertainty will greatly increase the search space.

So the first part we are now saying is 3 letters, which could either be all uppercase, all lowercase, or one uppercase followed by two lowercase.

The second part becomes tricky:

Next 4 characters of the password are either 1:23, 1:33 or Z:33 or Z33... There could also be the shift/caps key applied to this part of the sequence..

So definitely 4 characters? Or it might only be 3? The first character could be a number or a letter (either case)? This might be followed by a symbol, but might not be? Then any two numbers or symbols? Or is it specifically one of the following:

Code:

23
33
@#
##

Or perhaps your keyboard doesn't have the @ and # symbols above 2 and 3?

Then the last three characters could be 3 letters (all uppercase, all lowercase, or one uppercase followed by two lowercase) OR an uppercase letter followed by 2 numbers or 2 symbols? Again, is this definitely 33 or ##, or could it be any two numbers or symbols?

First part, The last part could also be Abc abc or EFG as in the first cap of each word hint. It could also be Hij or hij lower case which is another hint for the first part of the password..

Second part, the middle sequence is definitely 4 characters such as 1:23, 1:22, Z:23 or Z:33 the Z represents a notation I did to suggest a different character but I am redacting that part. It is deff either 23, 33, @# or ## Also another possibility for this part could be Z339. Where as the 339 part starts the : is replaced by the 3. 9 also represents another number I am redacting..

Sorry, I don't mean to make this more complicated that it has to be...

Third part, the last 3 could have a few options. Z33 or Z## where Z represents another letter I am redacting.

Hopefully this clears some things up, trying my best to not forget certain aspects of the puzzle but it turns out to be somewhat complex lol.

Not sure how tokenlist works with btcrecover but another cracker I read about uses multiple word lists and tries all possible combinations. Something like that makes a little more sense to me. I can start with 3 word lists and put all possible 3 character sequence in the first word list, 4 character sequence in the 2nd word list, and last 3 character sequence in the 3rd word list and start from there.

[o_e_l_e_o]

Ok, so based on all of that, your token file will look like this:

Code:

^1^%3A ^1^%3a ^1^%A%2a
^2^%in%y23 ^2^%in%y33 ^2^%in%y@# ^2^%in%y## ^2^%ia33%d
^3^%3A ^3^%3a ^3^%A%2a ^3^%A33 ^3^%A##

You can find the instructions on how I constructed this here: https://btcrecover.readthedocs.io/en/latest/tokenlist_file/


So for the first part it will try the following, where A/a represents any letter (case sensitive):

Code:

AAA
aaa
Aaa

For the second part it will try the following:

Code:

Any single digit, lowercase, or uppercase letter, followed by
Any single symbol, followed by
23, 33, @#, or ##

OR

Any single lowercase or uppercase letter, followed by 33, followed by any single digit

For the third part it will try the same as the first part, but also any uppercase letter followed by 33 or ##.


However, before you try to run this, you will need to try to narrow it down a bit. If you can replace any of the wildcards with something more specific (e.g. instead of "Any single digit, lowercase, or uppercase letter" at the start of the second part with "Definitely either a 1 or an uppercase Z"), then it would reduce the search space a lot. As it stands, that tokensfile will generate 23.5 trillion possible passwords, which on the provided benchmark for a 1660 of 6750 passwords/second, would take ~110 years to check.

[b2352351]

Ok, so based on all of that, your token file will look like this:

Code:

^1^%3A ^1^%3a ^1^%A%2a
^2^%in%y23 ^2^%in%y33 ^2^%in%y@# ^2^%in%y## ^2^%ia33%d
^3^%3A ^3^%3a ^3^%A%2a ^3^%A33 ^3^%A##

You can find the instructions on how I constructed this here: https://btcrecover.readthedocs.io/en/latest/tokenlist_file/


So for the first part it will try the following, where A/a represents any letter (case sensitive):

Code:

AAA
aaa
Aaa

For the second part it will try the following:

Code:

Any single digit, lowercase, or uppercase letter, followed by
Any single symbol, followed by
23, 33, @#, or ##

OR

Any single lowercase or uppercase letter, followed by 33, followed by any single digit

For the third part it will try the same as the first part, but also any uppercase letter followed by 33 or ##.


However, before you try to run this, you will need to try to narrow it down a bit. If you can replace any of the wildcards with something more specific (e.g. instead of "Any single digit, lowercase, or uppercase letter" at the start of the second part with "Definitely either a 1 or an uppercase Z"), then it would reduce the search space a lot. As it stands, that tokensfile will generate 23.5 trillion possible passwords, which on the provided benchmark for a 1660 of 6750 passwords/second, would take ~110 years to check.

A little confused on the structures and how tokenlist works with btcrecover but another cracker I read about uses multiple word lists and tries all possible combinations. Something like that makes a little more sense to me. I can start with 3 word lists and put all possible 3 character sequence in the first word list, 4 character sequence in the 2nd word list, and last 3 character sequence in the 3rd word list and start from there.

Anyway to format the tokenlist like into 3 seperate documents but on the same tokenlist? I could work from there and input my stuff.

[o_e_l_e_o]

If you think you can manage that then no harm in trying. It is difficult to understand exactly what you are saying when you say you are redacting some information - whether the "Z" represents any uppercase letter or one specific uppercase letter, for example.

Here is a much more narrow token file based on some assumptions about what you have said above:

Code:

^1^Abc ^1^abc ^1^EFG ^1^Hij ^1^hij
^2^%[1Z]:23 ^2^%[1Z]:33 ^2^%[1Z]:@# ^2^%[1Z]:## ^2^Z339
^3^Abc ^3^abc ^3^EFG ^3^Hij ^3^hij ^3^Z33 ^3^Z##

This will only generate 365 passwords.

A little confused on the structures and how tokenlist works

It will try one token from each line. The ^1^ before each entry simply tells it to try that token in the first position, ^2^ in the second position, and ^3^ in the third position. So essentially the file above will taken one token from each line in the order the lines are presented.

You can run btcrecover with the argument --listpass and it will simply print all the passwords in your terminal window so you can see what it is generating to make sure it is doing what you think it should be.

[b2352351]

If you think you can manage that then no harm in trying. It is difficult to understand exactly what you are saying when you say you are redacting some information - whether the "Z" represents any uppercase letter or one specific uppercase letter, for example.

Here is a much more narrow token file based on some assumptions about what you have said above:

Code:

^1^Abc ^1^abc ^1^EFG ^1^Hij ^1^hij
^2^%[1Z]:23 ^2^%[1Z]:33 ^2^%[1Z]:@# ^2^%[1Z]:## ^2^Z339
^3^Abc ^3^abc ^3^EFG ^3^Hij ^3^hij ^3^Z33 ^3^Z##

This will only generate 365 passwords.

A little confused on the structures and how tokenlist works

It will try one token from each line. The ^1^ before each entry simply tells it to try that token in the first position, ^2^ in the second position, and ^3^ in the third position. So essentially the file above will taken one token from each line in the order the lines are presented.

You can run btcrecover with the argument --listpass and it will simply print all the passwords in your terminal window so you can see what it is generating to make sure it is doing what you think it should be.

Ok let me dig into that a little more now. I am kind of dumb so it still looking a little daunting to me. But I will try to work through it slowly. Now, how do I go about deploying the crack command? Which token list in which directory do I update? Oh and I'm not sure how to load the wallet.dat... Thanks again for all the help!

[b2352351]

I got it cracking the tokenlist.txt file.. My first pass was unsuccessful. Said about 3,000 attempts. It is only using the CPU though. I am wondering how to enable GPU?

Also on the bottom of the password  list I added some old random passwords I had saved. I added 4 and it changed it to over 200,000 guesses now. How is that working?

^1^3characters ^1^3characters^1^3characters^1^3characters^1^3characters^1^3characters
^2^4characters ^2^4characters ^2^4characters ^2^4characters ^2^4characters ^2^4characters
^3^3characters ^3^3characters ^3^3characters ^3^3characters ^3^3characters ^3^3characters
oldpassword1
oldpassword2
oldpassword3
oldpassword4

[nc50lc]

-snip- I am wondering how to enable GPU?

Add --enable-gpu argument to your command.
That's assuming that you've successfully installed pyopencl and your GPU is supported.

note: You can check all available commands by entering: python btcrecover.py -h

[o_e_l_e_o]

Also on the bottom of the password  list I added some old random passwords I had saved. I added 4 and it changed it to over 200,000 guesses now. How is that working?

As I explained above, the tokens file will take a maximum of one entry from each line and combine them in all allowed permutations. So by including four passwords at the end like that, it will now be combining those passwords in every possible permutation with all the entries from the first three lines. If you want to simply try the four old passwords, then use the command --passwordlist instead of --tokenlist. That will just try each line of your file verbatim without combining it with other lines.

[PawGo]

-snip- I am wondering how to enable GPU?

Add --enable-gpu argument to your command.
That's assuming that you've successfully installed pyopencl and your GPU is supported.

b2352351, could you paste here the command you launch? In which mode you use btcrecover? Maybe in the one where GPU is not supported (or gives very limited performance improvement).
What is "password"? Is it for a seed or brainwallet?

[b2352351]

-snip- I am wondering how to enable GPU?

Add --enable-gpu argument to your command.
That's assuming that you've successfully installed pyopencl and your GPU is supported.

b2352351, could you paste here the command you launch? In which mode you use btcrecover? Maybe in the one where GPU is not supported (or gives very limited performance improvement).
What is "password"? Is it for a seed or brainwallet?

Code:

python btcrecover.py --wallet wallet.dat --tokenlist tokenlist.txt

But now I am going to add

Code:

python btcrecover.py --enable-gpu --wallet wallet.dat --tokenlist tokenlist.txt

and expand my password/token list.

Also I'm pretty sure my wallet.dat file is intact. It still says modified in 2013 in the properties and opens fine inside of the latest version of bitcoin core. Has there ever been any cases of wallet.dat files saying wrong password after being copied from the original machine? Or would you get a file corruption error within bitcoin core long before that? I really thought I'd be able to get it on the first pass. Now I just gotta sit down and expand the tokenlist a little bit.

[PawGo]

If the problem is to find wallet.dat password, I think hashcat is a much better solution.
I think it is easier to configure search parameters (at least for me)...
https://hashcat.net/hashcat/

You would simply prepare a configuration using masks:
https://hashcat.net/wiki/doku.php?id=mask_attack

[b2352351]

If the problem is to find wallet.dat password, I think hashcat is a much better solution.
I think it is easier to configure search parameters (at least for me)...
https://hashcat.net/hashcat/

You would simply prepare a configuration using masks:
https://hashcat.net/wiki/doku.php?id=mask_attack

Well I somewhat understand how to configure tokenlist and am cracking at about 10,000 checks per second now on the GPU so I'm not sure why I'd need to switch to hashcat. I think I just need to refine my tokenlist now.

[PawGo]

Well I somewhat understand how to configure tokenlist and am cracking at about 10,000 checks per second now on the GPU so I'm not sure why I'd need to switch to hashcat. I think I just need to refine my tokenlist now.

Ah, ok, good luck then!
I think you have many hints, so your chances are really big!

[b2352351]

Dang so i'm up to about a million guesses. No go yet

I have about 90 words on each line. I duplicated 1, 2 and 3 into the first line, 2nd line and 3rd line so my guesses have tripled. Still no go. I really thought I was going to get it.

I'll keep refining the list and thinking of more possibilities.

Would there be a way to input single characters? like break it down even more? for example first line ^1^ would have every character/letter/number/symbol I think is possible in the password. that would be three characters long, the first part of the password. Then do the same thing for the middle 4 characters, and then again for the last 3 characters. Would this be a huge password list? I don't mind letting it run, it took about 1 or 2 min for 1 million guesses.

Also, it says its using GPU but now I'm not seeing my GPU get maxed out. I'm pretty sure the first time I was doing it I was seeing GPU take up a lot more resources like 80 or 100% usage in task manager.

[o_e_l_e_o]

Would there be a way to input single characters?

Absolutely. Just add more lines with the same ^1^ at the start of each entry, changing the 1 to the number of the character position in the password. I would also use the argument %[chars] rather than list every character as its own entry, as it will be much easier for you to write and edit. So for example, if you think the first character might be A, B, C, a, b, or c, then rather than writing this:

Code:

^1^A ^1^B ^1^C ^1^a ^1^b ^1^c

You would just write this:

Code:

^1^%[ABCabc]

[b2352351]

Would there be a way to input single characters?

Absolutely. Just add more lines with the same ^1^ at the start of each entry, changing the 1 to the number of the character position in the password. I would also use the argument %[chars] rather than list every character as its own entry, as it will be much easier for you to write and edit. So for example, if you think the first character might be A, B, C, a, b, or c, then rather than writing this:

Code:

^1^A ^1^B ^1^C ^1^a ^1^b ^1^c

You would just write this:

Code:

^1^%[ABCabc]

Thank you Mr. Wizard

[b2352351]

I'm actually a little confused at this point. I've ran a few variations of the commands you've given me and it seems to be starting with 3 characters and testing that out to 10 characters.

I want it to test only 10 character possibilities, but with every character involved in the 1st position, 2nd and 3rd(last 1/3) last part of the 10 digit character sequence.

Any ideas?

[o_e_l_e_o]

Ahh, my bad. Yeah, when you have a list of tokens like this, it will try all possible ways of combining them, including selecting no tokens from a given line. First of all, make sure that you have ^1^ up to ^10^ in front of every token. Then, add --min-tokens 10 to your command in the terminal.

[b2352351]

Ahh, my bad. Yeah, when you have a list of tokens like this, it will try all possible ways of combining them, including selecting no tokens from a given line. First of all, make sure that you have ^1^ up to ^10^ in front of every token. Then, add --min-tokens 10 to your command in the terminal.

Okay now it's guessing ABCXXXXXX

I want it to guess all characters in the first three spaces, combined with all possible characters in the 2nd space, and all possible characters in the 3rd or last part of the password. Brake it down into 3 sections, hopefully this makes sense?

Thanks again for all the help. I really do appreciate it.

[o_e_l_e_o]

Not really following you. Can you share your tokens file? PM me if you would prefer (and I can provide a PGP key for encryption if you would prefer).

[b2352351]

password format is

xxx1:23xxx

and if there's 20 different characters that could represent xxx, and 50 characters that could represent 1:23, and 20 different characters that could represent the last part, xxx I want to be able to guess 20 characters for each x x x and then 20 possible characters in each middle position of 1:32, and 20 characters for the each of the characters in the last x x x

hopefully this makes a little more sense.. I'm thinking this is going to generate a lot of guesses.. But I am willing to wait.

I do have a moderate GPU but I'm not sure if it's using it correctly. The guesses haven't seemed to slow down, but now in task manager it is not reporting any usage on the GPU while attempting to crack. I'm pretty sure when I first started it stated around 80% usage on the GPU.

[PawGo]

Do you mean something like that?

Code:

^1^%[ABC...Zabc....z]
^2^%[ABC...Zabc....z]
^3^%[ABC...Zabc....z]
^4^%[0123456789]
^5^%[:!@#;]
^6^%[0123456789]
^7^%[0123456789]
^8^%[ABC...Zabc....z]
^9^%[ABC...Zabc....z]
^10^%[ABC...Zabc....z]

depends what are you "special characters"
And you should fill correctly "a..z" part

[b2352351]

Do you mean something like that?

Code:

^1^%[ABC...Zabc....z]
^2^%[ABC...Zabc....z]
^3^%[ABC...Zabc....z]
^4^%[0123456789]
^5^%[:!@#;]
^6^%[0123456789]
^7^%[0123456789]
^8^%[ABC...Zabc....z]
^9^%[ABC...Zabc....z]
^10^%[ABC...Zabc....z]

depends what are you "special characters"
And you should fill correctly "a..z" part

I'm gonna try this in combination with the  --min-tokens 10 command o_e_l_e_o suggested

EDIT: It now looks like it's doing what I was intending! Now I'm still not sure if GPU is being utilized. I think i'm going to brake it down a lot simpler using this method to reduce password list size. I use enablegpu command but task manager shows 0% usage on GPU now.. But it doesn't seemed like it has slowed down.

[b2352351]

I do have a moderate GPU but I'm not sure if it's using it correctly. The guesses haven't seemed to slow down, but now in task manager it is not reporting any usage on the GPU while attempting to crack. I'm pretty sure when I first started it stated around 80% usage on the GPU.

I don't know how accurate is task manager to monitor GPU usage, but it might be because known CPU bottleneck on btcrecover[1-3]. Have you tried different software such as GPU-Z[4] to monitor GPU usage?

[1] https://btcrecover.readthedocs.io/en/latest/GPU_Acceleration/#performance-tuning-background
[2] https://github.com/3rdIteration/btcrecover/issues/184
[3] https://github.com/3rdIteration/btcrecover/issues/119
[4] https://www.techpowerup.com/download/techpowerup-gpu-z/

I actually tried hardware monitor and it's actually reporting a higher percentage and higher temp. I cranked up the fans. I've done 2 or three new password lists and still nothing but I have some more to generate.

I was running one list for 4 hrs at 10k/Ps but it started to eat up like 20gb of ram lol.

I've seen some people using the --global-ws and --local-ws commands when they have larger lists. I'm guessing this helps allocate resources? Can anyone describe how these two commands work and the numbers associated with them? Thanks again everyone.

[b2352351]

Ok I'm running into some issues.. I guess I didn't really understand how the syntax worked, but i'm running some password lists and double checking if certain combos are being spit out but theyre not..

So I've gone ahead and broke it down into the 3 basic parts of the password.

If I have 40 guesses for each third of the password, then how would I write it to test all 40 combinations in each slot against the previous?

example

Password part 1 line 1: possibilities = 123 or 4567 or 8910
Password part 2 line 2: possibilities = 123 or 4567 or 8910
Password part 3 line 3: possibilities = 123 or 4567 or 8910

I want to test every possible combination available on each line

So.. ex.
123 123 123
123 123 4567
123 4567 4567
123 4567 123

But in all reality I have 40 total 3-4 character sequence combinations for each "line" of the password.

[PawGo]

Please see that documentation page, especially part about anchors:
https://btcrecover.readthedocs.io/en/latest/tokenlist_file/

There’s very similar case, where they specify tokens for the end and beginning (and even for the middle)
I guess it would be:

Code:

^1^abc
^1^cde
^2^987
^2^654
Etc.

[b2352351]

Please see that documentation page, especially part about anchors:
https://btcrecover.readthedocs.io/en/latest/tokenlist_file/

There’s very similar case, where they specify tokens for the end and beginning (and even for the middle)
I guess it would be:

Code:

^1^abc
^1^cde
^2^987
^2^654
Etc.

Ok I'll try this, I had my token list setup where all ^1^'s were on the same line and only had 3 lines but 40 guesses on each line..

edit: I now have 29 lines total ^1^ through ^10^ with 2-5 possibilities for each ^1-10^ notation.

Are you sure this format wont do something funky though? It's not even starting to count or crack after a while. Usually it starts counting right away. Maybe it counts # of password variations with cpu before cracking with gpu? Because listpass does not use gpu, list pass is not counting passwords either - also stuck right before stuff starts to load. The number must be really big, it is taking about 20% cpu right now.

How does --global-ws and --local-ws work? I heard some people use that with larger lists... Wondering if I'll need that. One of my old larger lists was like more than 100 million I think and it almost started taking up 20gb ram after 4 hours so I'm guessing global and local commands prevent that much usage? I can afford to give it 20gb ram but I'm afraid it would have kept climbing at only 4 hrs before I quit the run.

[b2352351]

How does --global-ws and --local-ws work?

You need to learn a bit about OpenCL to understand how those works. One thing i know, it's related with how much btcrecover use GPU VRAM (which also affect brute force speed).

I heard some people use that with larger lists... Wondering if I'll need that. One of my old larger lists was like more than 100 million I think and it almost started taking up 20gb ram after 4 hours so I'm guessing global and local commands prevent that much usage? I can afford to give it 20gb ram but I'm afraid it would have kept climbing at only 4 hrs before I quit the run.

I don't remember about RAM usage, but --global-ws and --local-ws is strongly recommended to improve brute force speed for large list password/token. You'll need to test best number for your device though.

For some reason my most recent word list took forever to launch. It got stuck after the

Code:

* * * * * * * * * * * * * * * * * * * *
*          Security: Warning                *
* * * * * * * * * * * * * * * * * * * *

Wallet Type: btcrpass.WalletBitcoinCore
Wallet difficulty: 40,829 SHA-512 iterations

part. Before it even says "counting passwords" I come back hours later to find it only guessed 24,000 passwords out of 93,000 possibilities. I was zipping through 100k lists in 10 seconds before. I will restart my machine maybe..

Also, for the global and local commands. Do I need to associated numbers with each of those commands? I don't recall reading anything that could actually explain what the numbers were, just that you need to adjust them in tandem.

[o_e_l_e_o]

There’s very similar case, where they specify tokens for the end and beginning (and even for the middle)
I guess it would be:

Code:

^1^abc
^1^cde
^2^987
^2^654
Etc.

Are you sure this format wont do something funky though? It's not even starting to count or crack after a while.

Provided you have a positional anchor (^x^) before every entry, then it will make no difference if you put every entry on its own line or you group all the entries in the same position together on the same line.

I have just created a test file which looks like this:

Code:

^1^1
^1^One
^2^2
^2^Two
^3^3
^3^Three
^4^4
^4^Four

And run with the argument --min-tokens 4 and it spat out the exact 16 combinations I would expect it to. Similarly, I changed it to this:

Code:

^1^1 ^1^One
^2^2 ^2^Two
^3^3 ^3^Three
^4^4 ^4^Four

And it spat out the exact same 16 combinations. If your tokens file is not working there must be something else wrong with it.

[b2352351]

There’s very similar case, where they specify tokens for the end and beginning (and even for the middle)
I guess it would be:

Code:

^1^abc
^1^cde
^2^987
^2^654
Etc.

Are you sure this format wont do something funky though? It's not even starting to count or crack after a while.

Provided you have a positional anchor (^x^) before every entry, then it will make no difference if you put every entry on its own line or you group all the entries in the same position together on the same line.

I have just created a test file which looks like this:

Code:

^1^1
^1^One
^2^2
^2^Two
^3^3
^3^Three
^4^4
^4^Four

And run with the argument --min-tokens 4 and it spat out the exact 16 combinations I would expect it to. Similarly, I changed it to this:

Code:

^1^1 ^1^One
^2^2 ^2^Two
^3^3 ^3^Three
^4^4 ^4^Four

And it spat out the exact same 16 combinations. If your tokens file is not working there must be something else wrong with it.

For some reason stacking ^1^'s and ^2^'s on top of eachother with 30 total lines took 12 hours to guess 100,000 passwords. The other format I was doing took 10 seconds. Weird. I wonder what is going on here.

[o_e_l_e_o]

For some reason stacking ^1^'s and ^2^'s on top of eachother with 30 total lines took 12 hours to guess 100,000 passwords. The other format I was doing took 10 seconds. Weird. I wonder what is going on here.

So, in a few local tests it seems the putting all the same numbers on the same line does make things run a little faster for me when compared to listing them all on their own line, but the difference is ~10%, and nowhere near the discrepancy you have seen.

Still, as I said above, if you are putting a positional anchor on every entry, then whether or not all the same positions are on the same line or on their own line is irrelevant to the passwords it will generate. There is no reason to run the one which is so slow when the other option will generate the same list in a fraction of the time.