generalmimon
commented
3 years ago @hello-adam First of all, the reason why you don't see values of instances or their _debug info is simply that they haven't been parsed at all. Unlike seq fields that all get parsed just by calling the _read method (which is done automatically by default), one of the fundamental properties of instances is that they are lazy. See https://doc.kaitai.io/user_guide.html#_instances_data_beyond_the_sequence:
Another very important difference between the
seqattribute and theinstancesattribute is that instances are lazy by default. What does that mean? Unless someone would call thatbodygetter method programmatically, no actual parsing ofbodywould be done.
So if you want to read values of all of them, you need to eventually invoke them all. I guess the easiest method for Python is to use reflection on the generated parser classes to get the instance names for each subtype so that you can access them afterwards. I suppose that this is going to be quite easy to do, just find how to use reflection in Python - probably there is some single function that gives you all property names when you call it with the struct object as an argument. You will need to read all instances recursively, though - start with the top-level object and while you iterate over the properties, check the value of each one if it isn't a nested KaitaiStruct object (i.e. something like isinstance(struct[property], kaitaistruct.KaitaiStruct) and if it is, you need to recurse into it and do the same for properties of this nested object.
Second, if you want to get _debug info for instances, check the generated code about how the instance program_headers is parsed and what _debug info it stores - I marked the lines saving info to _debug map with an asterisk * (https://github.com/Mahlet-Inc/hobbits/blob/c51f39f/src/hobbits-plugins/analyzers/KaitaiStruct/ksy_py/executable/elf.py#L1572-L1610):
@property
def program_headers(self):
if hasattr(self, '_m_program_headers'):
return self._m_program_headers if hasattr(self, '_m_program_headers') else None
_pos = self._io.pos()
self._io.seek(self.program_header_offset)
* self._debug['_m_program_headers']['start'] = self._io.pos()
if self._is_le:
self._raw__m_program_headers = [None] * (self.qty_program_header)
self._m_program_headers = [None] * (self.qty_program_header)
for i in range(self.qty_program_header):
* if not 'arr' in self._debug['_m_program_headers']:
* self._debug['_m_program_headers']['arr'] = []
* self._debug['_m_program_headers']['arr'].append({'start': self._io.pos()})
self._raw__m_program_headers[i] = self._io.read_bytes(self.program_header_entry_size)
_io__raw__m_program_headers = KaitaiStream(BytesIO(self._raw__m_program_headers[i]))
_t__m_program_headers = Elf.EndianElf.ProgramHeader(_io__raw__m_program_headers, self, self._root, self._is_le)
_t__m_program_headers._read()
self._m_program_headers[i] = _t__m_program_headers
* self._debug['_m_program_headers']['arr'][i]['end'] = self._io.pos()
else:
# duplicate code from `if self._is_le` branch - I know the compiler
# could do a better job of eliminating this, but it's not anywhere
# high on our priorities I'd say, as long as the code works
* self._debug['_m_program_headers']['end'] = self._io.pos()
self._io.seek(_pos)
return self._m_program_headers if hasattr(self, '_m_program_headers') else NoneThe notable change from seq fields is that the key in the _debug map has the _m_ prefix, which is something you'll need to adapt your code to.
KOLANICH
hello-adam
I'm using the Kaitai Python runtime in https://github.com/Mahlet-Inc/hobbits to make a Kaitai runner and viewer plugin. A user pointed out that when running kaitai's
executable/elf.ksyon something likelibc.so, my viewer is missing theprogramHeaders,sectionHeaders, andstringsparts of theheaderthat show up in the kaitai web IDEMy issue is that I can't seem to find those fields anywhere in the
_debugmetadata produced by by the parser:The root parsed object:
The header object:
Am I missing something somewhere?