Add __repr__ for capstone.CsInsn

[disconnect3d]

Currently, a print(instruction) displays a not very useful string like:

<capstone.CsInsn object at 0x7f3759d88128>

This PR enhances the display by adding a __repr__ magic method to the capstone.CsInsn class so it displays as follows:

<cs.CsInsn: address=0x5555555545fa, size=1, mnemonic=push, op_str=rbp>

[disconnect3d]

Let me know if you want any other/better format there. The one I choose here explicitly shows the properties one could use.

Some other ideas:

• We could change cs.CsInsn to capstone.CsInsn to be more verbose, or, drop the namespace and leave it with CsInsn (since there is already the Cs prefix)
• We could drop the property names and stay with something very compact like: '<CsInsn %s [%dB]: %s %s>' % (self.address, self.size, self.mnemonic, self.op_str) . Examples:

<CsInsn 0x5555555545fa [1B]: push rbp>
<CsInsn 0x7ffff7b93ae8 [4B]: adc byte ptr [r8 + 9], r8b>

[aquynh]

nice, the second idea looks fine to me, but maybe we can also print all bytes in bytes field?

[disconnect3d]

but maybe we can also print all bytes in bytes field?

Yup, that's a good idea. Below are some example outputs.

Version Alpha

This version shows hex bytes.

    def __repr__(self):
        return '<CsInsn 0x%x [%dB]: %s %s %s>' % (self.address, self.size, self.bytes.hex(), self.mnemonic, self.op_str)

That gives:

<CsInsn 0x5555555545fa [1B]: 55 push rbp>
<CsInsn 0x555555554600 [2B]: 4157 push r15>
<CsInsn 0x7ffff7b93ae8 [4B]: 4d104009 adc byte ptr [r8 + 9], r8b>
<CsInsn 0x5555555544f0 [2B]: 31ed xor ebp, ebp>
<CsInsn 0x5555555545fe [2B]: ebfe jmp 0x5555555545fe>
<CsInsn 0x7ffff7a05b97 [2B]: 89c7 mov edi, eax>

Version Bravo

This version displays the hex bytes in [] brackets.

    def __repr__(self):
        return '<CsInsn 0x%x [%dB] [%s]: %s %s>' % (self.address, self.size, self.bytes.hex(), self.mnemonic, self.op_str)

<CsInsn 0x5555555545fa [1B] [55]: push rbp>
<CsInsn 0x555555554600 [2B] [4157]: push r15>
<CsInsn 0x7ffff7b93ae8 [4B] [4d104009]: adc byte ptr [r8 + 9], r8b>
<CsInsn 0x5555555544f0 [2B] [31ed]: xor ebp, ebp>
<CsInsn 0x5555555545fe [2B] [ebfe]: jmp 0x5555555545fe>
<CsInsn 0x7ffff7a05b97 [2B] [89c7]: mov edi, eax>

Version Charlie

This version skips displaying the size in bytes.

    def __repr__(self):
        return '<CsInsn 0x%x [%s]: %s %s>' % (self.address, self.bytes.hex(), self.mnemonic, self.op_str)

<CsInsn 0x5555555545fa [55]: push rbp>
<CsInsn 0x555555554600 [4157]: push r15>
<CsInsn 0x7ffff7b93ae8 [4d104009]: adc byte ptr [r8 + 9], r8b>
<CsInsn 0x5555555544f0 [31ed]: xor ebp, ebp>
<CsInsn 0x5555555545fe [ebfe]: jmp 0x5555555545fe>
<CsInsn 0x7ffff7a05b97 [89c7]: mov edi, eax>

[disconnect3d]

@aquynh see some proposals above :)

[aquynh]

looks nice now, but why do you want to print the size? and with "B", which seems abundant?

[disconnect3d]

looks nice now, but why do you want to print the size? and with "B", which
seems abundant?

I initially printed the size to give more context but since we go towards displaying hex encoded bytes of the instruction, I don't think it is relevant anymore to do that.

So, shall we go with version charlie? (I updated the PR to contain the charlie version).

[aquynh]

merged, thanks.