Applies sym file information to idb

apply_sym.py

import re
import idc
import ida_nalt
import ida_kernwin
import ida_name


MODULE = re.compile(r'^MODULE \w+ \w+ \w+ .+$')
INFO = re.compile(r'^INFO CODE_ID \w+ .+$')
FILE = re.compile(r'^FILE (\d+) (.+)$')
FUNC = re.compile(r'^FUNC (\w+) (\w+) (\w+) (.+)$')
LINE = re.compile(r'^([0-9a-f]+) (\w+) (\d+) (\d+)$')
PUBLIC = re.compile(r'^PUBLIC (\w+) (\w+) (.+)$')
STACK = re.compile(r'^STACK CFI .+$')


def is_module(line):
    return MODULE.match(line)


def parse_module(res):
    return None


def is_info(line):
    return INFO.match(line)


def parse_info(res):
    return None


def is_file(line):
    return FILE.match(line)


def parse_file(res):
    fnum = int(res.group(1))
    name = res.group(2)
    return fnum, name


def is_func(line):
    return FUNC.match(line)


def parse_func(res):
    addr = int(res.group(1), 16)
    size = int(res.group(2), 16)
    psize = int(res.group(3), 16)
    name = res.group(4)

    return addr, size, name


def is_line(line):
    return LINE.match(line)


def parse_line(res):
    addr = int(res.group(1), 16)
    size = int(res.group(2), 16)
    line = int(res.group(3))
    fnum = int(res.group(4))

    return addr, size, line, fnum


def is_public(line):
    return PUBLIC.match(line)


def parse_public(res):
    addr = int(res.group(1), 16)
    psize = int(res.group(2), 16)
    name = res.group(3)

    return addr, name


def is_stack(line):
    return STACK.match(line)


def parse_stack(res):
    return None


def parse_lines(line):
    res = is_module(line)
    if res:
        return 'md', parse_module(res)

    res = is_info(line)
    if res:
        return 'in', parse_info(res)

    res = is_file(line)
    if res:
        return 'fl', parse_file(res)

    res = is_func(line)
    if res:
        return 'fn', parse_func(res)

    res = is_line(line)
    if res:
        return 'ln', parse_line(res)

    res = is_public(line)
    if res:
        return 'pb', parse_public(res)

    res = is_stack(line)
    if res:
        return 'st', parse_stack(res)

    return 'nn', None


def use_files(files, funcs, lines, pubs):
    base = ida_nalt.get_imagebase()

    ida_kernwin.show_wait_box('Apply funcs')
    sz = len(funcs)
    for i, addr in enumerate(funcs.keys()):
        size, name = funcs[addr]
        name = ida_name.validate_name(name, ida_name.VNT_IDENT)
        idc.set_name(base + addr, name, idc.SN_NOWARN)
        ida_kernwin.replace_wait_box('%d/%d, 0x%X' % (i, sz, base + addr))
    ida_kernwin.hide_wait_box()

    ida_kernwin.show_wait_box('Apply pubs')
    sz = len(pubs)
    for i, addr in enumerate(pubs.keys()):
        name = pubs[addr]
        name = ida_name.validate_name(name, ida_name.VNT_IDENT)
        idc.set_name(base + addr, name, idc.SN_NOWARN)
        ida_kernwin.replace_wait_box('%d/%d, 0x%X' % (i, sz, base + addr))
    ida_kernwin.hide_wait_box()

    ida_kernwin.show_wait_box('Apply files')
    sz1 = len(files)
    for i, fnum in enumerate(files.keys()):
        fname = files[fnum]

        if fnum not in lines:
            continue

        addrs = lines[fnum]

        sz2 = len(addrs)
        for j, addr in enumerate(addrs.keys()):
            size, line = addrs[addr]

            idc.add_sourcefile(base + addr, base + addr + size, fname)
            idc.set_source_linnum(base + addr, line)
            ida_kernwin.replace_wait_box('%d/%d - %d/%d, 0x%X (0x%x)' % (i, sz1, j, sz2, base + addr, size))
    ida_kernwin.hide_wait_box()


def main(path):
    files = dict()
    funcs = dict()
    lines = dict()
    pubs = dict()

    with open(path) as f:
        for line in f.readlines():
            line = line.rstrip()

            t, args = parse_lines(line)

            if t in ['md', 'in', 'st']:
                continue
            elif t == 'fl':
                fnum, name = args
                files[fnum] = name
            elif t == 'fn':
                addr, size, name = args
                funcs[addr] = size, name
            elif t == 'ln':
                addr, size, line, fnum = args

                if fnum not in lines:
                    lines[fnum] = dict()

                lines[fnum][addr] = size, line
            elif t == 'pb':
                addr, name = args
                pubs[addr] = name
            else:
                print('wrong')

    print('total:\nfiles = %d, funcs = %d, lines = %d, pubs = %d' % (len(files), len(funcs), len(lines), len(pubs)))

    use_files(files, funcs, lines, pubs)


main('nw.dll.sym')