Problem converting binary data to text using base64

**MartinR** · 01-25-2014

Hello, I have two C functions, one is RC4 which succesfull produce binary data from any text, second is base64 which also works great. The problem occur when I want to combine these two method to gain simple method to convert and save binary data to text using base64.

Usage of program is very simple, you threw some text into it and it's returning encoded and decoded version. Even if it works for very short text like "123" completely fails with longer like "Hello World"

I belive it's not hard to resolve but I have not enought C knowledge to do this. Could you please help ?

http://codepad.org/CrAgF6dU

Thanks for help

**laserlight** · 01-25-2014

How does it not work?

**MartinR** · 01-25-2014

Originally Posted by laserlight

How does it not work?

Have you tried it with even simple text like " Hello world !" ? If so, take a look at decoded string, unfortunately it's wrong.

**tabstop** · 01-25-2014

Originally Posted by MartinR

Hello, I have two C functions, one is RC4 which succesfull produce binary data from any text, second is base64 which also works great. The problem occur when I want to combine these two method to gain simple method to convert and save binary data to text using base64.

Usage of program is very simple, you threw some text into it and it's returning encoded and decoded version. Even if it works for very short text like "123" completely fails with longer like "Hello World"

I belive it's not hard to resolve but I have not enought C knowledge to do this. Could you please help ?

C code - 226 lines - codepad

Thanks for help

That suggests that either one or the other does not "work great". Compiling your code gives the following:

Code:

crypt.c: In function 'Base64Decode':
crypt.c:41:38: warning: initializer-string for array of chars is too long [enabled by default]
crypt.c: In function 'encodeblock':
crypt.c:71:38: warning: initializer-string for array of chars is too long [enabled by default]
crypt.c: In function 'main':
crypt.c:165:9: warning: unused variable 'input' [-Wunused-variable]
crypt.c:164:9: warning: variable 'rc' set but not used [-Wunused-but-set-variable]
crypt.c:162:59: warning: unused variable 'pch' [-Wunused-variable]

The unused variables aren't going to matter, but the fact that you have 66 characters in your base64 very well might.

**MartinR** · 01-25-2014

Originally Posted by tabstop

The unused variables aren't going to matter, but the fact that you have 66 characters in your base64 very well might.

Yes, there was indeed two more charasters but if you repair it, function still doesn't work as it should.

I think there is something wrong in proces of encoding and decoding binary data, as I said before it works with very short 4 - letters string but fails with longer.
It sounds bad since I need to convert 300 lines of pure text this way ;/

So, could you please look closely on this problem ?

**stahta01** · 01-25-2014

Code:

#define BASE64CHARSET   "ABCDEFGHIJKLMNOPQRSTUVWXYZ"\
                        "abcdefghijklmnopqrstuvwxyz"\
                        "0123456789"\
                        "+/?!"\
                        "";

Creating a macro with a ";" semicolon at the end is almost always a mistake.

Tim S.

**laserlight** · 01-25-2014

Here's what I suggest: write a comprehensive suite of unit tests for each of the 4 functions. Test with small input, large input, input with all sorts of different characters. This will either show you that they don't actually work all that great, or give you confidence that the problem probably lies elsewhere.

**MartinR** · 01-25-2014

If they are separated they works good, both RC4 encode and decode as they should with short and long text. As I said the problem occur when I joined them together and here I did something wrong. I am not a C guru and have problem to figure it out. Could you look at this ?

**oogabooga** · 01-25-2014

Try using unsigned char instead of char. Make a typedef:

Code:

typedef unsigned char uchar;

This code

Code:

        for (ix = strlen(buf) - 1; (ix >= 0) && (buf[ix] == 10); ix--)
            buf[ix] = 0;

removes multiple newlines from the end of buf, but fgets will only
deposit a single newline at the most. If you mean to remove whitespace characters generally, use isspace() (prototyped in ctype.h). And when you want to indicate a newline, use '\n' instead of 10.

And the whole "Hungarian notation" thing (e.g., pszKey) is generally frowned upon outside the Windows world.

**MartinR** · 01-25-2014

Thanks for answer, have you tested it and those changes solved problem ?

Originally Posted by oogabooga

Try using unsigned char instead of char. Make a typedef:

Code:

typedef unsigned char uchar;

Code:

typedef unsigned char uchar;

this code instead oof what ?

**stahta01** · 01-25-2014

Originally Posted by MartinR

Thanks for answer, have you tested it and those changes solved problem ?

Code:

typedef unsigned char uchar;

this code instead oof what ?

FYI: Bitwise operators should only be used on unsigned types in most cases.
You tend to get weird results on signed values on some systems. Yours might be such a system.

Tim S.

**MartinR** · 01-25-2014

Originally Posted by stahta01

FYI: Bitwise operators should only be used on unsigned types in most cases.
You tend to get weird results on signed values on some systems. Yours might be such a system.

Tim S.

Are you trying to say that this program works good on your system ? I am on Windows anyway. Once again, this program works good only for couple of first letters in my case. After first 3/4 letters it prints something weird.

**tabstop** · 01-25-2014

Originally Posted by MartinR

Are you trying to say that this program works good on your system ? I am on Windows anyway. Once again, this program works good only for couple of first letters in my case. After first 3/4 letters it prints something weird.

It doesn't even properly work on one letter, so it's not as though any of it works.

**oogabooga** · 01-25-2014

Originally Posted by MartinR

Are you trying to say that this program works good on your system ? I am on Windows anyway.

It probably wouldn't work properly on his system either because his system is likely similar to yours.

The problem is that when you right-shift a signed integer that has the sign bit set, it is up to the implementation to "sign-extend" the result (an "arithmetic" shift) or not (a "logical" shift). It's also up to the implementation whether or not char is signed or unsigned.

On your system, char is probably signed, so if any of your manipulations set the sign bit and this is followed by an arithmetic right-shift, the result will not be what you want (since, essentially, 1's come in from the left instead of 0's).

However, I haven't bothered to test the unsigned char idea (because I don't feel like it) so, although that is probably a problem, it may not be the specific one you're currently hunting.

**MartinR** · 01-25-2014

Could you please try to fix this code, I am unable to do this since I lack knowledge in C programming.

I expect to use this program from external file with aprox 300 lins to convert it to base64.

Thread: Problem converting binary data to text using base64

Thread Tools

Search Thread

Display

Problem converting binary data to text using base64

Similar Threads

Converting binary data in a txt file into HEX

converting a binary file to a text array

Converting text file to binary

converting text to binary

Converting Text to Binary (And Back)