dynamicCall and QByteArray - strange characters

[franco.amato]

1. Determine what the data structure actually is. My best guess is in post #37 and its content may be a serialised C# object.
1a. Match bytes from the byte array against candidates for the data structure and see if they make sense. A hex dump would be useful here.
1b. If you can see recognisable bits in the data (like your name) then look at the bytes around it to see if they make any sense, e.g. is the preceding byte, short, or long value the length of the string?
1c. Anything else that might help break the code (this is not unlike cracking weak encryption).
2. If you match the structure and it has a name then search for a wrapper that can handle that data structure.
3. Failing 2. write one yourself.

You are still at step 1 and the road is rocky.

Hi I got the data structure from the constructor.
There still is something I don't understand. Why I need to convert to hex? I did it and noticed tha the size is changed. It's twice respect the original bytearray:

ba =  m_axobj->dynamicCall("GetUserData(QString)", "1234567890" ).toByteArray(); //<-- size 441
QByteArray baHex = ba.toHex(); //<-- size 882

ba =  m_axobj->dynamicCall("GetUserData(QString)", "1234567890" ).toByteArray(); //<-- size 441
QByteArray baHex = ba.toHex(); //<-- size 882

To copy to clipboard, switch view to plain text mode 

[wysota]

You were supposed to print it on the screen to see what is inside. Though I still think an ascii dump would be better if the structure holds mostly character-based content.

[franco.amato]

You were supposed to print it on the screen to see what is inside. Though I still think an ascii dump would be better if the structure holds mostly character-based content.

I'll write a byte-to-ascii converter.
The structure contains both character-based content (as name and surname) and not character-based ( as finger print data )
Meanwhile my final goal is to parse the byte array and pack another structure to write back to another finger print terminal

[franco.amato]

I'll write a byte-to-ascii converter.
The structure contains both character-based content (as name and surname) and not character-based ( as finger print data )
Meanwhile my final goal is to parse the byte array and pack another structure to write back to another finger print terminal

Hi,
how could I do for example to read the first 4 bytes of the bytearray data and convert them to an integer?
I have to copy with the memcpy?

[wysota]

It depends on the byte order of the architecture but memcpy will work too.

[franco.amato]

It depends on the byte order of the architecture.

Hi wysota, could you do more specific?
I would convert the first 4 bytes of the bytearray in a number, the cardId in my case that it's a number of maximum 10 digits and that I can represent with 4 bytes.
I tried this

QByteArray ba;		
ba =  m_axobj->dynamicCall("GetUserData(QString)", "1234567890" ).toByteArray();
QByteArray cardId = ba.left(4); //now cardId contains the first 4 bytes of ba
qint32 v = qFromBigEndian( (const uchar*)cardId.constData() );

QByteArray ba;		
ba =  m_axobj->dynamicCall("GetUserData(QString)", "1234567890" ).toByteArray();
QByteArray cardId = ba.left(4); //now cardId contains the first 4 bytes of ba
qint32 v = qFromBigEndian( (const uchar*)cardId.constData() );

To copy to clipboard, switch view to plain text mode 

without success.
Where I'm wrong?

[wysota]

Hi wysota, could you do more specific?

It depends whether the architecture is big-endian or little-endian.

int size = ba.size();
QByteArray cardId = ba.left(4); //now cardId contains the first 4 bytes of ba
qint32 v = qFromBigEndian( (const uchar*)cardId.constData() );

int size = ba.size();
QByteArray cardId = ba.left(4); //now cardId contains the first 4 bytes of ba
qint32 v = qFromBigEndian( (const uchar*)cardId.constData() );

To copy to clipboard, switch view to plain text mode 

without success.

Maybe your machine is little-endian.

[franco.amato]

It depends whether the architecture is big-endian or little-endian.


Maybe your machine is little-endian.

wysota, the problem is in compilation. Compiler says I cannot convert from 'const unsigned char *' to 'qint32'

so I changed

qint32 v = qFromBigEndian( (const uchar*)cardId.constData() );

qint32 v = qFromBigEndian( (const uchar*)cardId.constData() );

To copy to clipboard, switch view to plain text mode 

to

qint32 v = qFromBigEndian( (const uchar*)(cardId.constData()) );

qint32 v = qFromBigEndian( (const uchar*)(cardId.constData()) );

To copy to clipboard, switch view to plain text mode 

for the operator precedence but nothing change.
I still get such error. This sounds strange as the documentation says
T qFromBigEndian ( const uchar * src )where T can be a qint32

[ChrisW67]

Why I need to convert to hex? I did it and noticed that the size is changed.

Google Hexadecimal. Seriously. If you printed each byte out in decimal you'd expect the size to be different... hex is just another human readable format that's longer than what it represents.

There were clearly a good number of non-printable characters in the 441 byte data block. Just looking at the gibberish printed doesn't concretely tell you that the printed sequence "I??" is actually a series of bytes with decimal values: 49, 63, 2, 20 (example only). The "?" characters might be the result of a byte matching the ASCII for "?" (first one) or a byte that matches nothing the machine can print so it substitutes a "?" (the second). Dumping the bytes as a series of decimal or hexadecimal values allows you to see the actual values of the bytes. Hex was easier because there's already method to do that.

I agree with Wysota that straight character dump is also needed.

how could I do for example to read the first 4 bytes of the bytearray data and convert them to an integer?

Think about how four bytes are arranged into a 32-bit integer. Extract the four bytes and combine them. Also think about what order the four bytes may be in the data stream versus what order they may be in your 32-bit integer. Search for "Endianness". You could also try casting a pointer to the byte array's data block but that has downsides.

[franco.amato]

I agree with Wysota that straight character dump is also needed.

what does mean?

Think about how four bytes are arranged into a 32-bit integer. Extract the four bytes and combine them. Also think about what order the four bytes may be in the data stream versus what order they may be in your 32-bit integer. Search for "Endianness". You could also try casting a pointer to the byte array's data block but that has downsides.

I extracted and copied first 4 bytes to a new bytearray and tried to convert to a number. See post above with the code I wrote but I get a compiler error that I don't understand

[wysota]

what does mean?

Please read the whole thread again and Google for every term you are not sure you understand. We have already discussed this and at the end at least I came to an impression that you implied you knew how to print the contents of the byte array in ascii. I suggest you do it now and post here the output you get as for now you are only chasing ghosts.

I extracted and copied first 4 bytes to a new bytearray and tried to convert to a number. See post above with the code I wrote but I get a compiler error that I don't understand

The compiler says you can't cast a non-mutable pointer to a charater to 32 bit signed integer. Even if you managed to do that, I doubt you would be using the value you wanted to use, especially if your machine is 64bit. It'd be best if you checked whether qFromBigEndian() actually does what you expect it to. Moreover it seems you are using Intel based Windows machine so I can hardly believe some driver would return some internal number in big-endian encoding.

[franco.amato]

Please read the whole thread again and Google for every term you are not sure you understand. We have already discussed this and at the end at least I came to an impression that you implied you knew how to print the contents of the byte array in ascii. I suggest you do it now and post here the output you get as for now you are only chasing ghosts.


The compiler says you can't cast a non-mutable pointer to a charater to 32 bit signed integer. Even if you managed to do that, I doubt you would be using the value you wanted to use, especially if your machine is 64bit. It'd be best if you checked whether qFromBigEndian() actually does what you expect it to. Moreover it seems you are using Intel based Windows machine so I can hardly believe some driver would return some internal number in big-endian encoding.

Hi wysota yes my machine is little endian ( Intel ).
Meanwhile I'm trying to store the first 4 bytes (32 bits) into an integer to check the value. I should get the number 1234567890 that can be stored with 4 bytes,
but seems an impossible task for me

[wysota]

I will not post any more posts in this thread until I see the hexadecimal and ascii dump (or something that at least remotely looks like one) of the byte array contents you get from the COM method.

[franco.amato]

I will not post any more posts in this thread until I see the hexadecimal and ascii dump (or something that at least remotely looks like one) of the byte array contents you get from the COM method.

here the hexadecimal dump

499602d20000686f72697a6174694672616e636f0000000000 00000000416d61746f00000000000000000000010103000000 00323232323232323232323232323232632003010a04150a04 1530393837363534333231010160451710138d22554620c210 a90c144351aa861983c04f06224420a90b3204909f062845d1 a30a32468044850c4831080b1548f1b2890a496165882e89d1 9e0628cb603e862fcd703c893b4dc098a01dcdf09e883acec0 9598220f0138881e8f603a88210fe095831b50519e0d159081 04602791603288189251290affff1123333ffffffff0112333 34ffffffe011223334fffffde0112233344fffdee011223334 4fffeee01122233344ffeee01112233444ffeeee0112233444 ffddee0112233444ffddde0112333444ffcdde0122334444ff ccdd0123444445ffcccd0134445445ffbbcc02345555456fbb cc02455555556fabbbe4566666556faaaaa6666666667faaaa 977777666687f99998777776678f0000000000000000000000 00000000000000000000000000000000000000000000000000 00000000000000000000000000000000000000000000000000 00000000000000000000000000000000

[wysota]

And ascii?

[franco.amato]

And ascii?

"I??\0\0xml;q=0.Franco\0\0\0\0\0\0\0\0\0Amato \0\0 \0\0\0\0\0\0\0\0

\0\0\0\0AAAAAAAAAAAAAAA=0\0\n \n0987654321

`E?\"UF ??\fCQ????O\"D ?\v2??(E??\n2F?D?\fH1\b\vH???\nIae?.???(?`>?/?p<?;M???????:????\"
8??`:?!???PQ?\r??`'?`2 ??Q)\n??#3?????#34????\"34????\"34O???\"34O? ??\"#3D???#4D???
#4D???
#4D???
34D???
\"3DD? ??
#DDE???
4DTE???4UUEo??EUUUo???VffUo???ffff?? ?wwff????wwvg?\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0 \0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0 \0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0 \0\0\0\0\0\0\0\0\0"

from the C# code

[wysota]

I don't want C# code. How do you know ActiveX returns data packaged in the same way to C++ and C#?

Anyway I can tell you the first four bytes in your blob are not a number. Unless you are expecting a huge value there... like 47335830

[franco.amato]

I don't want C# code. How do you know ActiveX returns data packaged in the same way to C++ and C#?

Anyway I can tell you the first four bytes in your blob are not a number. Unless you are expecting a huge value there... like 47335830

yes I'm expecting 1234567890

[wysota]

In that case it's there as big endian.

[franco.amato]

In that case it's there as big endian.

I didn't understand.
You found the value 1234567890 in the first 4 bytes??? Really?
If yes please help me to extract and show it as I'm becoming crazy.
Meanwhile the big endian you mentioned has sense has the stream come from a fingerprint reader that's not based on
intel cpu.