mirror of
https://github.com/isledecomp/isle.git
synced 2025-10-23 00:14:22 +00:00
Ghidra Import: Support virtual inheritance (#1071)
* Implement/fix Ghidra imports for multiple and virtual inheritance Unfortunately, the handling in Ghidra is still far from perfect. This is a good place to start, though. * Support offsets in vbase pointers * Support `this adjust` * minor stylistic improvement * Improvements to documentation --------- Co-authored-by: jonschz <jonschz@users.noreply.github.com>
This commit is contained in:
jonschz
@@ -1,3 +1,4 @@
|
||||
from dataclasses import dataclass
|
||||
import re
|
||||
import logging
|
||||
from typing import Any, Dict, List, NamedTuple, Optional
|
||||
@@ -26,6 +27,19 @@ class FieldListItem(NamedTuple):
|
||||
type: str
|
||||
|
||||
|
||||
@dataclass
|
||||
class VirtualBaseClass:
|
||||
type: str
|
||||
index: int
|
||||
direct: bool
|
||||
|
||||
|
||||
@dataclass
|
||||
class VirtualBasePointer:
|
||||
vboffset: int
|
||||
bases: list[VirtualBaseClass]
|
||||
|
||||
|
||||
class ScalarType(NamedTuple):
|
||||
offset: int
|
||||
name: Optional[str]
|
||||
@@ -157,6 +171,16 @@ class CvdumpTypesParser:
|
||||
r"^\s+list\[\d+\] = LF_BCLASS, (?P<scope>\w+), type = (?P<type>.*), offset = (?P<offset>\d+)"
|
||||
)
|
||||
|
||||
# LF_FIELDLIST virtual direct/indirect base pointer, line 1/2
|
||||
VBCLASS_RE = re.compile(
|
||||
r"^\s+list\[\d+\] = LF_(?P<indirect>I?)VBCLASS, .* base type = (?P<type>.*)$"
|
||||
)
|
||||
|
||||
# LF_FIELDLIST virtual direct/indirect base pointer, line 2/2
|
||||
VBCLASS_LINE_2_RE = re.compile(
|
||||
r"^\s+virtual base ptr = .+, vbpoff = (?P<vboffset>\d+), vbind = (?P<vbindex>\d+)$"
|
||||
)
|
||||
|
||||
# LF_FIELDLIST member name (2/2)
|
||||
MEMBER_RE = re.compile(r"^\s+member name = '(?P<name>.*)'$")
|
||||
|
||||
@@ -206,7 +230,7 @@ class CvdumpTypesParser:
|
||||
re.compile(r"\s*Arg list type = (?P<arg_list_type>[\w()]+)$"),
|
||||
re.compile(
|
||||
r"\s*This adjust = (?P<this_adjust>[\w()]+)$"
|
||||
), # TODO: figure out the meaning
|
||||
), # By how much the incoming pointers are shifted in virtual inheritance; hex value without `0x` prefix
|
||||
re.compile(
|
||||
r"\s*Func attr = (?P<func_attr>[\w()]+)$"
|
||||
), # Only for completeness, is always `none`
|
||||
@@ -282,12 +306,12 @@ class CvdumpTypesParser:
|
||||
|
||||
members: List[FieldListItem] = []
|
||||
|
||||
super_id = field_obj.get("super")
|
||||
if super_id is not None:
|
||||
super_ids = field_obj.get("super", [])
|
||||
for super_id in super_ids:
|
||||
# May need to resolve forward ref.
|
||||
superclass = self.get(super_id)
|
||||
if superclass.members is not None:
|
||||
members = superclass.members
|
||||
members += superclass.members
|
||||
|
||||
raw_members = field_obj.get("members", [])
|
||||
members += [
|
||||
@@ -526,7 +550,57 @@ class CvdumpTypesParser:
|
||||
|
||||
# Superclass is set here in the fieldlist rather than in LF_CLASS
|
||||
elif (match := self.SUPERCLASS_RE.match(line)) is not None:
|
||||
self._set("super", normalize_type_id(match.group("type")))
|
||||
superclass_list: dict[str, int] = self.keys[self.last_key].setdefault(
|
||||
"super", {}
|
||||
)
|
||||
superclass_list[normalize_type_id(match.group("type"))] = int(
|
||||
match.group("offset")
|
||||
)
|
||||
|
||||
# virtual base class (direct or indirect)
|
||||
elif (match := self.VBCLASS_RE.match(line)) is not None:
|
||||
virtual_base_pointer = self.keys[self.last_key].setdefault(
|
||||
"vbase",
|
||||
VirtualBasePointer(
|
||||
vboffset=-1, # default to -1 until we parse the correct value
|
||||
bases=[],
|
||||
),
|
||||
)
|
||||
assert isinstance(
|
||||
virtual_base_pointer, VirtualBasePointer
|
||||
) # type checker only
|
||||
|
||||
virtual_base_pointer.bases.append(
|
||||
VirtualBaseClass(
|
||||
type=match.group("type"),
|
||||
index=-1, # default to -1 until we parse the correct value
|
||||
direct=match.group("indirect") != "I",
|
||||
)
|
||||
)
|
||||
|
||||
elif (match := self.VBCLASS_LINE_2_RE.match(line)) is not None:
|
||||
virtual_base_pointer = self.keys[self.last_key].get("vbase", None)
|
||||
assert isinstance(
|
||||
virtual_base_pointer, VirtualBasePointer
|
||||
), "Parsed the second line of an (I)VBCLASS without the first one"
|
||||
vboffset = int(match.group("vboffset"))
|
||||
|
||||
if virtual_base_pointer.vboffset == -1:
|
||||
# default value
|
||||
virtual_base_pointer.vboffset = vboffset
|
||||
elif virtual_base_pointer.vboffset != vboffset:
|
||||
# vboffset is always equal to 4 in our examples. We are not sure if there can be multiple
|
||||
# virtual base pointers, and if so, how the layout is supposed to look.
|
||||
# We therefore assume that there is always only one virtual base pointer.
|
||||
logger.error(
|
||||
"Unhandled: Found multiple virtual base pointers at offsets %d and %d",
|
||||
virtual_base_pointer.vboffset,
|
||||
vboffset,
|
||||
)
|
||||
|
||||
virtual_base_pointer.bases[-1].index = int(match.group("vbindex"))
|
||||
# these come out of order, and the lists are so short that it's fine to sort them every time
|
||||
virtual_base_pointer.bases.sort(key=lambda x: x.index)
|
||||
|
||||
# Member offset and type given on the first of two lines.
|
||||
elif (match := self.LIST_RE.match(line)) is not None:
|
||||
@@ -579,7 +653,7 @@ class CvdumpTypesParser:
|
||||
else:
|
||||
logger.error("Unmatched line in arglist: %s", line[:-1])
|
||||
|
||||
def read_pointer_line(self, line):
|
||||
def read_pointer_line(self, line: str):
|
||||
if (match := self.LF_POINTER_ELEMENT.match(line)) is not None:
|
||||
self._set("element_type", match.group("element_type"))
|
||||
else:
|
||||
|
||||
@@ -6,6 +6,9 @@ from isledecomp.cvdump.types import (
|
||||
CvdumpTypesParser,
|
||||
CvdumpKeyError,
|
||||
CvdumpIntegrityError,
|
||||
FieldListItem,
|
||||
VirtualBaseClass,
|
||||
VirtualBasePointer,
|
||||
)
|
||||
|
||||
TEST_LINES = """
|
||||
@@ -245,10 +248,111 @@ NESTED, enum name = JukeBox::JukeBoxScript, UDT(0x00003cc2)
|
||||
list[12] = LF_MEMBER, private, type = T_USHORT(0021), offset = 12
|
||||
member name = 'm_length'
|
||||
|
||||
|
||||
0x4dee : Length = 406, Leaf = 0x1203 LF_FIELDLIST
|
||||
list[0] = LF_VBCLASS, public, direct base type = 0x15EA
|
||||
virtual base ptr = 0x43E9, vbpoff = 4, vbind = 3
|
||||
list[1] = LF_IVBCLASS, public, indirect base type = 0x1183
|
||||
virtual base ptr = 0x43E9, vbpoff = 4, vbind = 1
|
||||
list[2] = LF_IVBCLASS, public, indirect base type = 0x1468
|
||||
virtual base ptr = 0x43E9, vbpoff = 4, vbind = 2
|
||||
list[3] = LF_VFUNCTAB, type = 0x2B95
|
||||
list[4] = LF_ONEMETHOD, public, VANILLA, index = 0x15C2, name = 'LegoRaceMap'
|
||||
list[5] = LF_ONEMETHOD, public, VIRTUAL, index = 0x15C3, name = '~LegoRaceMap'
|
||||
list[6] = LF_ONEMETHOD, public, VIRTUAL, index = 0x15C5, name = 'Notify'
|
||||
list[7] = LF_ONEMETHOD, public, VIRTUAL, index = 0x15C4, name = 'ParseAction'
|
||||
list[8] = LF_ONEMETHOD, public, VIRTUAL, index = 0x4DED, name = 'VTable0x70'
|
||||
list[9] = LF_ONEMETHOD, public, INTRODUCING VIRTUAL, index = 0x15C2,
|
||||
vfptr offset = 0, name = 'FUN_1005d4b0'
|
||||
list[10] = LF_MEMBER, private, type = T_UCHAR(0020), offset = 8
|
||||
member name = 'm_parentClass2Field1'
|
||||
list[11] = LF_MEMBER, private, type = T_32PVOID(0403), offset = 12
|
||||
member name = 'm_parentClass2Field2'
|
||||
|
||||
0x4def : Length = 34, Leaf = 0x1504 LF_CLASS
|
||||
# members = 21, field list type 0x4dee, CONSTRUCTOR,
|
||||
Derivation list type 0x0000, VT shape type 0x12a0
|
||||
Size = 436, class name = LegoRaceMap, UDT(0x00004def)
|
||||
|
||||
0x4db6 : Length = 30, Leaf = 0x1504 LF_CLASS
|
||||
# members = 16, field list type 0x4db5, CONSTRUCTOR, OVERLOAD,
|
||||
Derivation list type 0x0000, VT shape type 0x1266
|
||||
Size = 16, class name = MxString, UDT(0x00004db6)
|
||||
|
||||
0x5591 : Length = 570, Leaf = 0x1203 LF_FIELDLIST
|
||||
list[0] = LF_VBCLASS, public, direct base type = 0x15EA
|
||||
virtual base ptr = 0x43E9, vbpoff = 4, vbind = 3
|
||||
list[1] = LF_IVBCLASS, public, indirect base type = 0x1183
|
||||
virtual base ptr = 0x43E9, vbpoff = 4, vbind = 1
|
||||
list[2] = LF_IVBCLASS, public, indirect base type = 0x1468
|
||||
virtual base ptr = 0x43E9, vbpoff = 4, vbind = 2
|
||||
list[3] = LF_VFUNCTAB, type = 0x4E11
|
||||
list[4] = LF_ONEMETHOD, public, VANILLA, index = 0x1ABD, name = 'LegoCarRaceActor'
|
||||
list[5] = LF_ONEMETHOD, public, VIRTUAL, index = 0x1AE0, name = 'ClassName'
|
||||
list[6] = LF_ONEMETHOD, public, VIRTUAL, index = 0x1AE1, name = 'IsA'
|
||||
list[7] = LF_ONEMETHOD, public, VIRTUAL, index = 0x1ADD, name = 'VTable0x6c'
|
||||
list[8] = LF_ONEMETHOD, public, VIRTUAL, index = 0x1ADB, name = 'VTable0x70'
|
||||
list[9] = LF_ONEMETHOD, public, VIRTUAL, index = 0x1ADA, name = 'SwitchBoundary'
|
||||
list[10] = LF_ONEMETHOD, public, VIRTUAL, index = 0x1ADC, name = 'VTable0x9c'
|
||||
list[11] = LF_ONEMETHOD, public, INTRODUCING VIRTUAL, index = 0x558E,
|
||||
vfptr offset = 0, name = 'FUN_10080590'
|
||||
list[12] = LF_ONEMETHOD, public, INTRODUCING VIRTUAL, index = 0x1AD8,
|
||||
vfptr offset = 4, name = 'FUN_10012bb0'
|
||||
list[13] = LF_ONEMETHOD, public, INTRODUCING VIRTUAL, index = 0x1AD9,
|
||||
vfptr offset = 8, name = 'FUN_10012bc0'
|
||||
list[14] = LF_ONEMETHOD, public, INTRODUCING VIRTUAL, index = 0x1AD8,
|
||||
vfptr offset = 12, name = 'FUN_10012bd0'
|
||||
list[15] = LF_ONEMETHOD, public, INTRODUCING VIRTUAL, index = 0x1AD9,
|
||||
vfptr offset = 16, name = 'FUN_10012be0'
|
||||
list[16] = LF_ONEMETHOD, public, INTRODUCING VIRTUAL, index = 0x1AD8,
|
||||
vfptr offset = 20, name = 'FUN_10012bf0'
|
||||
list[17] = LF_ONEMETHOD, public, INTRODUCING VIRTUAL, index = 0x1AD9,
|
||||
vfptr offset = 24, name = 'FUN_10012c00'
|
||||
list[18] = LF_ONEMETHOD, public, INTRODUCING VIRTUAL, index = 0x1ABD,
|
||||
vfptr offset = 28, name = 'VTable0x1c'
|
||||
list[19] = LF_MEMBER, protected, type = T_REAL32(0040), offset = 8
|
||||
member name = 'm_parentClass1Field1'
|
||||
list[25] = LF_ONEMETHOD, public, VIRTUAL, (compgenx), index = 0x15D1, name = '~LegoCarRaceActor'
|
||||
|
||||
0x5592 : Length = 38, Leaf = 0x1504 LF_CLASS
|
||||
# members = 26, field list type 0x5591, CONSTRUCTOR,
|
||||
Derivation list type 0x0000, VT shape type 0x34c7
|
||||
Size = 416, class name = LegoCarRaceActor, UDT(0x00005592)
|
||||
|
||||
0x5593 : Length = 638, Leaf = 0x1203 LF_FIELDLIST
|
||||
list[0] = LF_BCLASS, public, type = 0x5592, offset = 0
|
||||
list[1] = LF_BCLASS, public, type = 0x4DEF, offset = 32
|
||||
list[2] = LF_IVBCLASS, public, indirect base type = 0x1183
|
||||
virtual base ptr = 0x43E9, vbpoff = 4, vbind = 1
|
||||
list[3] = LF_IVBCLASS, public, indirect base type = 0x1468
|
||||
virtual base ptr = 0x43E9, vbpoff = 4, vbind = 2
|
||||
list[4] = LF_IVBCLASS, public, indirect base type = 0x15EA
|
||||
virtual base ptr = 0x43E9, vbpoff = 4, vbind = 3
|
||||
list[5] = LF_ONEMETHOD, public, VANILLA, index = 0x15CD, name = 'LegoRaceCar'
|
||||
list[6] = LF_ONEMETHOD, public, VIRTUAL, index = 0x15CE, name = '~LegoRaceCar'
|
||||
list[7] = LF_ONEMETHOD, public, VIRTUAL, index = 0x15D2, name = 'Notify'
|
||||
list[8] = LF_ONEMETHOD, public, VIRTUAL, index = 0x15E8, name = 'ClassName'
|
||||
list[9] = LF_ONEMETHOD, public, VIRTUAL, index = 0x15E9, name = 'IsA'
|
||||
list[10] = LF_ONEMETHOD, public, VIRTUAL, index = 0x15D5, name = 'ParseAction'
|
||||
list[11] = LF_ONEMETHOD, public, VIRTUAL, index = 0x15D3, name = 'SetWorldSpeed'
|
||||
list[12] = LF_ONEMETHOD, public, VIRTUAL, index = 0x15DF, name = 'VTable0x6c'
|
||||
list[13] = LF_ONEMETHOD, public, VIRTUAL, index = 0x15D3, name = 'VTable0x70'
|
||||
list[14] = LF_ONEMETHOD, public, VIRTUAL, index = 0x15DC, name = 'VTable0x94'
|
||||
list[15] = LF_ONEMETHOD, public, VIRTUAL, index = 0x15E5, name = 'SwitchBoundary'
|
||||
list[16] = LF_ONEMETHOD, public, VIRTUAL, index = 0x15DD, name = 'VTable0x9c'
|
||||
list[17] = LF_ONEMETHOD, public, INTRODUCING VIRTUAL, index = 0x15D4,
|
||||
vfptr offset = 32, name = 'SetMaxLinearVelocity'
|
||||
list[18] = LF_ONEMETHOD, public, INTRODUCING VIRTUAL, index = 0x15D4,
|
||||
vfptr offset = 36, name = 'FUN_10012ff0'
|
||||
list[19] = LF_ONEMETHOD, public, INTRODUCING VIRTUAL, index = 0x5588,
|
||||
vfptr offset = 40, name = 'HandleSkeletonKicks'
|
||||
list[20] = LF_MEMBER, private, type = T_UCHAR(0020), offset = 84
|
||||
member name = 'm_childClassField'
|
||||
|
||||
0x5594 : Length = 34, Leaf = 0x1504 LF_CLASS
|
||||
# members = 30, field list type 0x5593, CONSTRUCTOR,
|
||||
Derivation list type 0x0000, VT shape type 0x2d1e
|
||||
Size = 512, class name = LegoRaceCar, UDT(0x000055bb)
|
||||
"""
|
||||
|
||||
|
||||
@@ -309,6 +413,31 @@ def test_members(parser: CvdumpTypesParser):
|
||||
(12, "m_length", "T_USHORT"),
|
||||
]
|
||||
|
||||
# LegoRaceCar with multiple superclasses
|
||||
assert parser.get("0x5594").members == [
|
||||
FieldListItem(offset=0, name="vftable", type="T_32PVOID"),
|
||||
FieldListItem(offset=0, name="vftable", type="T_32PVOID"),
|
||||
FieldListItem(offset=8, name="m_parentClass1Field1", type="T_REAL32"),
|
||||
FieldListItem(offset=8, name="m_parentClass2Field1", type="T_UCHAR"),
|
||||
FieldListItem(offset=12, name="m_parentClass2Field2", type="T_32PVOID"),
|
||||
FieldListItem(offset=84, name="m_childClassField", type="T_UCHAR"),
|
||||
]
|
||||
|
||||
|
||||
def test_virtual_base_classes(parser: CvdumpTypesParser):
|
||||
"""Make sure that virtual base classes are parsed correctly."""
|
||||
|
||||
lego_car_race_actor = parser.keys.get("0x5591")
|
||||
assert lego_car_race_actor is not None
|
||||
assert lego_car_race_actor["vbase"] == VirtualBasePointer(
|
||||
vboffset=4,
|
||||
bases=[
|
||||
VirtualBaseClass(type="0x1183", index=1, direct=False),
|
||||
VirtualBaseClass(type="0x1468", index=2, direct=False),
|
||||
VirtualBaseClass(type="0x15EA", index=3, direct=True),
|
||||
],
|
||||
)
|
||||
|
||||
|
||||
def test_members_recursive(parser: CvdumpTypesParser):
|
||||
"""Make sure that we unwrap the dependency tree correctly."""
|
||||
|
||||
Reference in New Issue
Block a user