Pure Python parser for IDA Pro databases (.idb files)


License
Apache-2.0
Install
pip install python-idb==0.7.1

Documentation

Python IDB

python-idb

python-idb is a library for accessing the contents of IDA Pro databases (.idb files). It provides read-only access to internal structures such as the B-tree (ID0 section), name address index (NAM section), flags index (ID2 section), and types (TIL section). The library also provides analysis of B-tree entries to expose logical structures like functions, cross references, bytes, and disassembly (via Capstone). An example use for python-idb might be to run IDA scripts in a pure-Python environment.

Willem Hengeveld (mailto:itsme@xs4all.nl) provided the initial research into the low-level structures in his projects pyidbutil and idbutil. Willem deserves substantial credit for reversing the .idb file format and publishing his results online. This project heavily borrows from his knowledge, though there is little code overlap.

example use:

example: list function names

In this example, we list the effective addresses and names of functions:

In [4]: import idb
   ...: with idb.from_file('./data/kernel32/kernel32.idb') as db:
   ...:     api = idb.IDAPython(db)
   ...:     for ea in api.idautils.Functions():
   ...:         print('%x: %s' % (ea, api.idc.GetFunctionName(ea)))

Out [4]: 68901010: GetStartupInfoA
   ....: 689011df: Sleep
   ....: 68901200: MulDiv
   ....: 68901320: SwitchToFiber
   ....: 6890142c: GetTickCount
   ....: 6890143a: ReleaseMutex
   ....: 68901445: WaitForSingleObject
   ....: 68901450: GetCurrentThreadId
        ...

Note that we create an emulated instance of the IDAPython scripting interface, and use this to invoke idc and idautils routines to fetch data.

example: run an existing IDAPython script

In this example, we run the yara_fn.py IDAPython script to generate a YARA rule for the function at effective address 0x68901695 in kernel32.idb:

asciicast

The target script yara_fn.py has only been slightly modified:

  • to make it Python 3.x compatible, and
  • to use the modern IDAPython modules, such as ida_bytes.GetManyBytes rather than idc.GetManyBytes.

what works

  • ~250 unit tests that demonstrate functionality including file format, B-tree, analysis, and idaapi features.
  • read-only parsing of .idb and .i64 files from IDA Pro v5.0 to v7.5
    • extraction of file sections
    • B-tree lookups and queries (ID0 section)
    • flag enumeration (ID1 section)
    • named address listing (NAM section)
    • types parsing (TIL section)
  • analysis of artifacts that reconstructs logical elements, including:
    • root metadata
    • loader metadata
    • entry points
    • functions
    • structures
    • cross references
    • fixups
    • segments
  • partial implementation of the IDAPython API, including:
    • Names
    • Heads
    • Segs
    • GetMnem (via Capstone)
    • Functions
    • FlowChart (basic blocks)
    • lots and lots of flags
  • Python 2.7 & 3.x compatibility
  • zlib-packed idb/i64 files

what will never work

  • write access

getting started

python-idb is a pure-Python library, with the exception of Capstone (required only when calling disassembly APIs). You can install it via pip or setup.py install, both of which should handle depedency resolution:

 $ cd ~/Downloads/python-idb/
 $ python setup.py install
 $ python scripts/run_ida_script.py  ~/tools/yara_fn.py  ~/Downloads/kernel32.idb
   ... profit! ...

While most python-idb function have meaningful docstrings, there is not yet a comprehensive documentation website. However, the unit tests demonstrate functionality that you'll probably find useful.

Someone interested in learning the file format and contributing to the project should review the idb.fileformat module & tests. Those that are looking to extract meaningful information from existing .idb files probably should look at the idb.analysis and idb.idapython modules & tests.

Please report issues or feature requests through Github's bug tracker associated with the project.

license

python-idb is licensed under the Apache License, Version 2.0. This means it is freely available for use and modification in a personal and professional capacity.