From 70dfbebb3acbc71a3d7829bf437e182aea5ff2d5 Mon Sep 17 00:00:00 2001
From: pompolic <pompolic@special-circumstanc.es>
Date: Tue, 6 Jul 2021 23:31:09 +0200
Subject: [PATCH] WIP commit
Dependencies between script files are kind of a mess at the moment
---
gdb-port/gui.py | 6 ++++--
gdb-port/parser-name-instrumentation-gdb.py | 12 +++++++++++-
gdb-port/parser-type-instrumentation-gdb.py | 13 +++++++++----
3 files changed, 24 insertions(+), 7 deletions(-)
diff --git a/gdb-port/gui.py b/gdb-port/gui.py
index 34089ef..429621f 100644
--- a/gdb-port/gui.py
+++ b/gdb-port/gui.py
@@ -46,6 +46,7 @@ class PresentationLayer():
#self.set_input_chunk(top_level_parse.get_input_chunk())
self.parser_env = StringVar()
+ self.parser_env.set(top_level_parse.get_parser_env())
#"Current parser" widgets
ttk.Label(self.frame, text="Current parser").grid(column=1, row=1)
@@ -59,8 +60,8 @@ class PresentationLayer():
# ParserHierarchy widget (sequence members, parsers this one contains, etc)
# Showing parser decisions is probably possible by capturing the return value of h_do_parse
ttk.Label(self.frame, text="Parser hierarchy").grid(column=2, row=3)
- #ttk.Label(self.frame, textvariable=self.parser_env).grid(column=2, row=4)
- ttk.Label(self.frame, text="[Placeholder, Placeholder, Placeholder]").grid(column=2, row=4)
+ ttk.Label(self.frame, textvariable=self.parser_env).grid(column=2, row=4)
+ #ttk.Label(self.frame, text="[Placeholder, Placeholder, Placeholder]").grid(column=2, row=4)
ttk.Button(self.frame, text="Step", command=step).grid(column=1, row=4)
@@ -71,6 +72,7 @@ def step(*args):
gdb.execute("hammer-parse-step 1")
presentation_layer.set_top_parser(top_level_parse.peek_parser())
presentation_layer.set_input_chunk(top_level_parse.get_input_chunk())
+ presentation_layer.set_parser_env(top_level_parse.get_parser_env())
if top_level_parse is None:
print("Please import parser name script")
diff --git a/gdb-port/parser-name-instrumentation-gdb.py b/gdb-port/parser-name-instrumentation-gdb.py
index c612002..6d55b62 100644
--- a/gdb-port/parser-name-instrumentation-gdb.py
+++ b/gdb-port/parser-name-instrumentation-gdb.py
@@ -121,6 +121,7 @@ class TopLevelParse:
self.unclaimed_mem_use = 0
# Holds 32 characters starting at state->input_stream[index], used by the GUI
self.current_input_chunk = ''
+ self.current_parser_env = ''
# Called from h_packrat_parse()'s handler, where the parse state and arena get initialized
def enter_h_packrat_parse(self, parser):
@@ -159,6 +160,9 @@ class TopLevelParse:
parser_stack = self.peek_parserstack()
parser_stack.push(parser_obj)
+ if parser_decombinator:
+ p_env = parser_decombinator.decompose_parser(parser_obj, self)
+ self.set_parser_env(str(p_env))
return parser_obj
def return_from_perform_lowlevel_parse(self):
@@ -223,6 +227,12 @@ class TopLevelParse:
def get_input_chunk(self):
return self.input_chunk
+ def set_parser_env(self, parser_env):
+ self.current_parser_env = parser_env
+
+ def get_parser_env(self):
+ return self.current_parser_env
+
def add_or_get_parser(self, parser_addr):
try:
parser_obj = self.parser_objs[int(parser_addr)]
@@ -275,8 +285,8 @@ class HDoParseBreakpoint(gdb.Breakpoint):
top_level_parse.enter_h_do_parse(state, None, parser)
input_chunk = input_ptr + index
- top_level_parse.set_input_chunk(input_chunk.string('UTF-8','replace',32))
#print(input_chunk.string('ascii','backslashreplace',10))
+ top_level_parse.set_input_chunk(input_chunk.string('UTF-8','replace',32))
# Check if we need to stop after a number of steps
diff --git a/gdb-port/parser-type-instrumentation-gdb.py b/gdb-port/parser-type-instrumentation-gdb.py
index fd46ec3..59c4b26 100644
--- a/gdb-port/parser-type-instrumentation-gdb.py
+++ b/gdb-port/parser-type-instrumentation-gdb.py
@@ -152,7 +152,7 @@ class AndEnv(HParserEnv):
self.member_parser = parser_obj
def __str__(self):
- return str([str(self.member_parser)]))
+ return str([str(self.member_parser)])
class ChEnv(HParserEnv):
def __init__(self, parser, top_level_parse):
@@ -270,7 +270,7 @@ class IgnoreSeqEnv(HParserEnv):
self.member_parsers.append(parser_obj)
def __str__(self):
- return str([str(self.which), str([str(p) for p in self.member_parsers]))
+ return str([str(self.which), str([str(p) for p in self.member_parsers])])
class IndirectEnv(HParserEnv):
def __init__(self, parser, top_level_parse):
@@ -515,7 +515,7 @@ vtable_to_env = {
'charset_vt': CharsetEnv,
'choice_vt': ChoiceEnv,
'difference_vt': DifferenceEnv,
- 'end_vt', EndEnv,
+ 'end_vt': EndEnv,
'endianness_vt': EndiannessEnv,
'epsilon_vt': EpsilonEnv,
'ignore_vt': IgnoreEnv,
@@ -540,6 +540,8 @@ vtable_to_env = {
'xor_vt': XorEnv
}
+# TODO: runtime address -> env class map
+
# When given a Parser object, decompose_parser() deduces its type from the vtable, and returns the appropriate HParserEnv subclass, containing member parsers and args
# Not sure what to name it. Alternatives:
# ParserDisassembler?
@@ -556,9 +558,12 @@ class ParserDecombinator:
parser_addr = parser.address
vtable_p = gdb.parse_and_eval("((HParser*) " + str(parser_addr) + ")->vtable")
try:
- envClass = vtable_to_env[vtable_p.name]
+ vtable_sym = vt_types.lookup_by_address(vtable_p)
+ envClass = vtable_to_env[vtable_sym.name]
except KeyError:
print("Unknown vtable: " + str(vtable_p))
return None
return envClass(parser, top_level_parse)
+
+parser_decombinator = ParserDecombinator()
--
GitLab