zerotypic / ipyida Goto Github PK

In the binary ninja console, there's a few 'helper' functions / variables they alias (c.f. https://docs.binary.ninja/getting-started.html#script-python-console)

I think we should inject at least one or two into the iPython scope.

current_address / here()
current_function

Where current_address or here() returns the result of idaapi.get_screen_ea() and current_function returns idaapi.get_func(idaapi.get_screen_ea))

I am not sure where the best place is to integrate / alias these, but I don't think it should be too hard.

I've noticed that the iPython console seems to break/change at least some of IDA's console functionality. Specifically, I've noticed that the default IDA console/output window no longer prints the result of a statement being evaluated.

eg, this is what IDA does normally:

Python>idaapi.get_screen_ea()
0x1801c5977

But with iPython running/installed, the address does not get printed out. It seems that iPython is swallowing these prints or something. Also, related (or maybe a side effect?) is that the IDA console will no longer print integers as hex by default. I think this was maybe some hook they had on STDOUT but I am not sure.

There's still some issues with crashing on close that need to be investigated with older versions of IDA. 6.6 seems to intermittently crash on close on Windows.

I'm not sure if we can do anything about this? But it's not a pleasant thing to run into...

Maybe we can unbind IDA's hijacking of the help() command?

The way that IDA applies themeing makes it hard (maybe impossible) to change the colors of the 'call hint' dialog that the iPython console pops up through simple CSS. I actually had to tweak the plugin code.

As such, I was thinking it might be better to just make some simple theme/palette file kind of like lighthouse. It also gives us an opportunity to ship better default (bright) and dark themes for the console.

IPythonConsole keeps track of the widget that was in focus before it was activated, and when "Esc" is pressed, returns focus to that widget. However, if that widget gets deleted, e.g. when it is closed, IDA will crash upon trying to access the saved SWIG TWidget* object.

To fix this, we either need a way to determine if a TWidget* is valid or not, or we need to hook onto the closing code of the associated widget, such that we remove our reference to it if it's being deleted. Alternatively, we could avoid saving a reference to the TWidget*, and just save the title of the widget, which we could then look up using find_widget().

We should hook the iPython console output to print integers that fall within IDB segments as addresses.

eg:

In [4]: idaapi.get_screen_ea()
Out[4]: 5372018224

should print out 0x140327e30 instead.

It would also be nice to make these clickable, like the default IDA console.