def get_data(segs, cpp_in):
'''[str] -> str -> [c_ast]'''
def gen(((ti, ea, declstr), node)):
'''(tinfo_t, ea_t) -> c_ast'''
# NOTE mutates node
has_data = ida.seg_name(ea) != '.bss'
ptr = ida.can_be_off32(ea)
if ptr not in [ida.BADADDR, 0]:
ptr_ti = ida.get_or_guess_tinfo(ptr)
ti_node = cdecl.get_decls(declstr).decls[utils.cpp_decomp_tag][0]
return gen(((ptr_ti, ptr, ''), ti_node))
# XXX our data segment has no strings, but that's probably specific to
# the one binary we're dealing with
elif ti.is_array():
ai = ida.array_type_data_t()
ti.get_array_details(ai)
item_size = ai.elem_type.get_size()
typename = get_type_for_c_ast_constant(ai.elem_type)
length = (ida.item_end(ea) - ida.item_head(ea)) / item_size
if has_data is True:
if ai.elem_type.is_char():
node.init = ep_ct.constant(typename, c_stringify(
ida.get_string(ea)))
else:
items = list(get_item(ea + j * item_size, ti)
for j in xrange(0, length))
node.init = ep_ct.initlist(
[ep_ct.constant(typename, str(x)) for x in items])
else:
if has_data is True:
typename = get_type_for_c_ast_constant(ti)
node.init = ep_ct.constant(typename,
str(get_item(ea, ti)))
return node
def get_str_literals(segs):
'''[str] -> {str : str}'''
# XXX making a bad assumption that we only need a string if it's referred to
# in .text, but this avoids some junk in .rodata (e.g. jump tables)
return get_literals(segs, ida.get_string,
lambda x: cdecl.is_c_str(ida.guess_type(x)) and
'.text' in list(ida.seg_name(y) for y in
(ida.data_refs_to(x))))
def has_got_seg_in_data_refs(refs):
# filter out data not referenced by GOT
return '.got' in [ida.seg_name(x) for x in refs]
