From 8d1d5a8bbd7240490113a9900799f1ffbffe2afe Mon Sep 17 00:00:00 2001
From: pompolic <pompolic@special-circumstanc.es>
Date: Mon, 5 Jul 2021 20:08:01 +0200
Subject: [PATCH] Env classes for all known parser types
---
gdb-port/parser-type-instrumentation-gdb.py | 373 +++++++++++++++++---
1 file changed, 323 insertions(+), 50 deletions(-)
diff --git a/gdb-port/parser-type-instrumentation-gdb.py b/gdb-port/parser-type-instrumentation-gdb.py
index 62e6310..e234409 100644
--- a/gdb-port/parser-type-instrumentation-gdb.py
+++ b/gdb-port/parser-type-instrumentation-gdb.py
@@ -15,6 +15,7 @@ parser_name_defaults = {
'difference_vt': '(Unnamed difference)',
'end_vt': '(Unnamed end)',
'endianness_vt': '(Unnamed endianness)',
+ 'epsilon_vt': '(Unnamed epsilon)',
'ignore_vt': '(Unnamed ignore)',
'ignoreseq_vt': '(Unnamed ignoreseq)',
'indirect_vt': '(Unnamed indirect)',
@@ -84,69 +85,46 @@ class AttrBoolEnv(HParserEnv):
def __init__(self, parser, top_level_parse):
self.parser = parser
- member_parser_p = gdb.parse_and_eval("((HAttrBool *) parser->env)->p")
+ member_parser_p = gdb.parse_and_eval("((HAttrBool *) parser->env)->p")
self.member_parser = top_level_parse.add_or_get_parser(member_parser_p)
- self.predicate_p = gdb.parse_and_eval("((HAttrBool *) parser->env)->pred")
- self.user_data_p = gdb.parse_and_eval("((HAttrBool *) parser->env)->user_data")
-
+ self.predicate_p = gdb.parse_and_eval("((HAttrBool *) parser->env)->pred")
+ self.user_data_p = gdb.parse_and_eval("((HAttrBool *) parser->env)->user_data")
+
def __str__(self):
return str(self.member_parser)
-class BitsEnv(HParserEnv):
- def __init__(self, parser, top_level_parse):
- self.parser = parser
-
- self.length = gdb.parse_and_eval("((HBitsEnv *) parser->env)->length")
- self.signedp = gdb.parse_and_eval("((HBitsEnv *) parser->env)->signedp")
-
-class SequenceEnv(HParserEnv):
+class BindEnv(HParserEnv):
def __init__(self, parser, top_level_parse):
- super().__init__(parser, top_level_parse) # TODO: maybe move self.parser to base class. otherwise, is this needed?
self.parser = parser
- self.member_parsers = []
-
- #frame = gdb.selected_frame()
- h_sequence_p = gdb.parse_and_eval("(HSequence*) parser->env")
- num_parsers = gdb.parse_and_eval("((HSequence *) parser->env)->len")
- # TODO: should GDB do the array indexing operation, or should the Python code?
- # TODO: top_level_parse.create_or_get_parser(address)
- #[gdb.parse_and_eval("((HSequence*) parser->env)->p_array[" + str(index) + "]") for index in range(0, num_parsers)]
- # TODO: maybe the loop can be replaced with a list comprehension
- #[top_level_parse.add_or_get_parser(gdb.parse_and_eval("((HSequence*) parser->env)->p_array[" + str(index) +"]")) for index in range(0, num_parsers)]
- for index in range(0, num_parsers):
- parser_p = gdb.parse_and_eval("((HSequence*) parser->env)->p_array[" + str(index) + "]")
- parser_obj = top_level_parse.add_or_get_parser(parser_p)
- if parser_obj.name is None:
- parser_obj.name_parser( self.name_from_vtable(parser_obj) )
- self.member_parsers.append(parser_obj)
+ member_parser_p = gdb.parse_and_eval("((HBindEnv *) parser->env)->p")
+ self.member_parser = top_level_parse.add_or_get_parser(member_parser_p)
+ continuation_p = gdb.parse_and_eval("((HBindEnv *) parser->env)->k")
+ self.continuation = int(continuation_p)
+ env_p = gdb.parse_and_eval("((HBindEnv *) parser->env)->env")
+ self.env = int(env_p)
def __str__(self):
- return str([str(parser) for parser in self.member_parsers])
+ return str([str(self.member_parser), str(hex(self.continuation)), str(hex(self.env))])
-
-class IgnoreEnv:
+class BitsEnv(HParserEnv):
def __init__(self, parser, top_level_parse):
self.parser = parser
- self.member_parser_pointers = []
- ignored_p = gdb.parse_and_eval("(HParser*) parser->env")
+ self.length = gdb.parse_and_eval("((HBitsEnv *) parser->env)->length")
+ self.signedp = gdb.parse_and_eval("((HBitsEnv *) parser->env)->signedp")
- # Ideally we could look up ignored_p using top_level_parse, but it will not find a result if perform_lowlevel_parse(ignored_p) wasn't called yet
- #TODO: method to add parser in TopLevelParse
+class ButNotEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
- '''
- try:
- parser_obj = top_level_parse.parser_objs[ignored_p]
- except KeyError:
- # Create a parser object with no name and the address of the parser
- parser_obj = Parser(None, int(ignored_p))
- top_level_parse.parser_objs[int(ignored_p)] = parser_obj
+ p1_p = gdb.parse_and_eval("((HTwoParsers *) parser->env)->p1")
+ self.p1 = top_level_parse.add_or_get_parser(p1_p)
+ p2_p = gdb.parse_and_eval("((HTwoParsers *) parser->env)->p2")
+ self.p2 = top_level_parse.add_or_get_parser(p2_p)
- self.member_parsers.append(parser_obj)
- '''
- ignore_obj = top_level_parse.add_or_get_parser(ignored_p)
- self.member_parsers.append(ignore_p)
+ def __str__(self):
+ return str([str(self.p1), str(self.p2)])
class ActionEnv:
def __init__(self, parser, top_level_parse):
@@ -180,6 +158,16 @@ class ChEnv(HParserEnv):
def __str__(self):
return str(self.ch_value)
+class CharsetEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ self.charset_arg = gdb.parse_and_eval("(HCharset) parser->env")
+
+ def __str__(self):
+ return str(self.charset_arg)
+
+# TODO: numeric parameter passed to parse_and_eval
class ChoiceEnv(HParserEnv):
def __init__(self, parser, top_level_parse):
super().__init__(parser, top_level_parse)
@@ -205,6 +193,119 @@ class ChoiceEnv(HParserEnv):
def __str__(self):
return str([str(parser) for parser in self.member_parsers])
+class DifferenceEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ p1_p = gdb.parse_and_eval("((HTwoParsers *) parser->env)->p1")
+ self.p1 = top_level_parse.add_or_get_parser(p1_p)
+ p2_p = gdb.parse_and_eval("((HTwoParsers *) parser->env)->p2")
+ self.p2 = top_level_parse.add_or_get_parser(p2_p)
+
+ def __str__(self):
+ return str([str(self.p1), str(self.p2)])
+
+class EndEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ def __str__(self):
+ return "[]"
+
+class EndiannessEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ member_parser_p = gdb.parse_and_eval("((HParseEndianness *) parser->env)->p")
+ self.member_parser = top_level_parse.add_or_get_parser(member_parser_p)
+ endianness_val = gdb.parse_and_eval("((HParseEndianness *) parser->env)->endianness")
+ self.endianness = int(endianness_val)
+
+ def __str__(self):
+ return str([str(self.member_parser), str(self.endianness)])
+
+class EpsilonEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ def __str__(self):
+ return "[]"
+
+class IgnoreEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ ignored_p = gdb.parse_and_eval("(HParser*) parser->env")
+
+ ignore_obj = top_level_parse.add_or_get_parser(ignored_p)
+ self.member_parser = ignore_obj
+
+ def __str__(self):
+ return str([str(self.member_parser)])
+
+# TODO: numeric parameter passed to parse_and_eval
+class IgnoreSeqEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+ self.member_parsers = []
+
+ ignoreseq_p = gdb.parse_and_eval("(HIgnoreSeq*) parser->env")
+ seq_len = gdb.parse_and_eval("((HIgnoreSeq*) parser->env)->len")
+ which = gdb.parse_and_eval("((HIgnoreSeq*) parser->env)->which")
+ self.which = int(which)
+
+ for index in range(0, seq_len):
+ parser_p = gdb.parse_and_eval("((HIgnoreSeq*) parser->env)->parsers[" + str(index) + "]")
+ parser_obj = top_level_parse.add_or_get_parser(parser_p)
+ # TODO: name parsers on add?
+ # TODO: move updates to parsers to TopLevelParse?
+ if parser_obj.name is None:
+ parser_obj.name_parser( self.name_from_vtable(parser_obj) )
+ self.member_parsers.append(parser_obj)
+
+ def __str__(self):
+ return str([str(self.which), str([str(p) for p in self.member_parsers]))
+
+class IndirectEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ member_parser_p = gdb.parse_and_eval("((HIndirectEnv*) parser->env)->parser")
+ self.member_parser = top_level_parse.add_or_get_parser(member_parser_p)
+ touched = gdb.parse_and_eval("((HIndirectEnv*) parser->env)->touched")
+ self.touched = bool(touched)
+
+ def __str__(self):
+ return str([str(self.member_parser), str(self.touched)])
+
+class IntRangeEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ member_parser_p = gdb.parse_and_eval("((HRange*) parser->env)->p")
+ self.member_parser = top_level_parse.add_or_get_parser(member_parser_p)
+ lower_val = gdb.parse_and_eval("((HRange*) parser->env)->lower")
+ self.lower = int(lower_val)
+ upper_val = gdb.parse_and_eval("((HRange*) parser->env)->upper")
+ self.upper = int(upper_val)
+
+
+ def __str__(self):
+ return str([str(self.member_parser), str(self.lower), str(self.upper)])
+
+class LengthValueEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ length_parser_p = gdb.parse_and_eval("((HLenVal *) parser->env)->length")
+ self.length_parser = top_level_parse.add_or_get_parser(length_parser_p)
+
+ value_parser_p = gdb.parse_and_eval("((HLenVal *) parser->env)->value")
+ self.value_parser = top_level_parse.add_or_get_parser(value_parser_p)
+
+ def __str__(self):
+ return str([str(self.length_parser), str(self.value_parser)])
+
class ManyEnv(HParserEnv):
def __init__(self, parser, top_level_parse):
self.parser = parser
@@ -247,18 +348,190 @@ class OptionalEnv(HParserEnv):
def __str__(self):
return str(self.member_parser)
+# TODO: numeric parameter passed to parse_and_eval
+class PermutationEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ super().__init__(parser, top_level_parse) # TODO: maybe move self.parser to base class. otherwise, is this needed?
+ self.parser = parser
+ self.member_parsers = []
+
+ #frame = gdb.selected_frame()
+ h_sequence_p = gdb.parse_and_eval("(HSequence*) parser->env")
+ num_parsers = gdb.parse_and_eval("((HSequence *) parser->env)->len")
+ # TODO: should GDB do the array indexing operation, or should the Python code?
+ # TODO: top_level_parse.create_or_get_parser(address)
+ #[gdb.parse_and_eval("((HSequence*) parser->env)->p_array[" + str(index) + "]") for index in range(0, num_parsers)]
+ # TODO: maybe the loop can be replaced with a list comprehension
+ #[top_level_parse.add_or_get_parser(gdb.parse_and_eval("((HSequence*) parser->env)->p_array[" + str(index) +"]")) for index in range(0, num_parsers)]
+
+ for index in range(0, num_parsers):
+ parser_p = gdb.parse_and_eval("((HSequence*) parser->env)->p_array[" + str(index) + "]")
+ parser_obj = top_level_parse.add_or_get_parser(parser_p)
+ if parser_obj.name is None:
+ parser_obj.name_parser( self.name_from_vtable(parser_obj) )
+ self.member_parsers.append(parser_obj)
+
+ def __str__(self):
+ return str([str(parser) for parser in self.member_parsers])
+
+class SeekEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ offset_val = gdb.parse_and_eval("((HSeek *) parser->env)->offset")
+ self.offset = int(offset_val)
+ whence_val = gdb.parse_and_eval("((HSeek *) parser->env)->whence")
+ self.whence = int(whence_val)
+ #TODO: enum for whence?
+
+ def __str__(self, parser, top_level_parse):
+ return str([str(self.offset), str(self.whence)])
+
+class SkipEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ skip_val = gdb.parse_and_eval("((uintptr_t) parser->env)")
+ self.skip = int(skip_val)
+
+ def __str__(self):
+ return str(self.skip)
+
+class TellEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ def __str__(self):
+ return "[]"
+
+# TODO: numeric parameter passed to parse_and_eval
+class SequenceEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ super().__init__(parser, top_level_parse) # TODO: maybe move self.parser to base class. otherwise, is this needed?
+ self.parser = parser
+ self.member_parsers = []
+
+ #frame = gdb.selected_frame()
+ h_sequence_p = gdb.parse_and_eval("(HSequence*) parser->env")
+ num_parsers = gdb.parse_and_eval("((HSequence *) parser->env)->len")
+ # TODO: should GDB do the array indexing operation, or should the Python code?
+ # TODO: top_level_parse.create_or_get_parser(address)
+ #[gdb.parse_and_eval("((HSequence*) parser->env)->p_array[" + str(index) + "]") for index in range(0, num_parsers)]
+ # TODO: maybe the loop can be replaced with a list comprehension
+ #[top_level_parse.add_or_get_parser(gdb.parse_and_eval("((HSequence*) parser->env)->p_array[" + str(index) +"]")) for index in range(0, num_parsers)]
+
+ for index in range(0, num_parsers):
+ parser_p = gdb.parse_and_eval("((HSequence*) parser->env)->p_array[" + str(index) + "]")
+ parser_obj = top_level_parse.add_or_get_parser(parser_p)
+ if parser_obj.name is None:
+ parser_obj.name_parser( self.name_from_vtable(parser_obj) )
+ self.member_parsers.append(parser_obj)
+
+ def __str__(self):
+ return str([str(parser) for parser in self.member_parsers])
+
+class TokenEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ str_p = gdb.parse_and_eval("((HToken *) parser->env)->str")
+ self.token = str_p
+ str_len = gdb.parse_and_eval("((HToken *) parser->env)->len")
+ self.str_len = int(str_len)
+ #TODO: get str_len bytes from memory and save it
+
+ def __str__(self):
+ return str(hex(self.token))
+
+class UnimplemenetedEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ def __str__(self):
+ return "[]"
+
+# TODO: print key as string if applicable
+class GetEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ member_parser_p = gdb.parse_and_eval("((HStoredValue *) parse->env)->p")
+ self.member_parser = top_level_parse.add_or_get_parser(member_parser_p)
+ key_p = gdb.parse_and_eval("((HStoredValue *) parse->env)->key")
+ self.key = int(key_p)
+
+ def __str__(self):
+ return str([str(self.member_parser), str(hex(self.key))])
+
+class PutEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ member_parser_p = gdb.parse_and_eval("((HStoredValue *) parse->env)->p")
+ self.member_parser = top_level_parse.add_or_get_parser(member_parser_p)
+ key_p = gdb.parse_and_eval("((HStoredValue *) parse->env)->key")
+ self.key = int(key_p)
+
+ def __str__(self):
+ return str([str(self.member_parser), str(hex(self.key))])
+
+class WhitespaceEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ member_parser_p = gdb.parse_and_eval("(HParser *) parse->env")
+ self.member_parser = top_level_parse.add_or_get_parser(member_parser_p)
+
+ def __str__(self):
+ return str([str(self.member_parser)])
+
+class XorEnv(HParserEnv):
+ def __init__(self, parser, top_level_parse):
+ self.parser = parser
+
+ p1_p = gdb.parse_and_eval("((HTwoParsers *) parser->env)->p1")
+ self.p1 = top_level_parse.add_or_get_parser(p1_p)
+ p2_p = gdb.parse_and_eval("((HTwoParsers *) parser->env)->p2")
+ self.p2 = top_level_parse.add_or_get_parser(p2_p)
+
+ def __str__(self):
+ return str([str(self.p1), str(self.p2)])
+
vtable_to_env = {
- 'sequence_vt': SequenceEnv,
- 'ignore_vt': IgnoreEnv,
'action_vt': ActionEnv,
'and_vt': AndEnv,
'attr_bool_vt': AttrBoolEnv,
+ 'bind_vt': BindEnv,
+ 'bits_vt': BitsEnv,
+ 'butnot_vt': ButNotEnv,
'ch_vt': ChEnv,
+ 'charset_vt': CharsetEnv,
+ 'choice_vt': ChoiceEnv,
+ 'difference_vt': DifferenceEnv,
+ 'end_vt', EndEnv,
+ 'endianness_vt': EndiannessEnv,
+ 'epsilon_vt': EpsilonEnv,
+ 'ignore_vt': IgnoreEnv,
+ 'ignoreseq_vt': IgnoreSeqEnv,
+ 'indirect_vt': IndirectEnv,
+ 'int_range_vt': IntRangeEnv,
+ 'length_value_vt': LengthValueEnv,
'many_vt': ManyEnv,
'not_vt': NotEnv,
'nothing_vt': NothingEnv,
- 'optional_vt': OptionalEnv
+ 'optional_vt': OptionalEnv,
+ 'permutation_vt': PermutationEnv,
+ 'seek_vt': SeekEnv,
+ 'skip_vt': SkipEnv,
+ 'tell_vt': TellEnv,
+ 'sequence_vt': SequenceEnv,
+ 'token_vt': TokenEnv,
+ 'unimplemented_vt': UnimplemenetedEnv,
+ 'get_vt': GetEnv,
+ 'put_vt': PutEnv,
+ 'whitespace_vt': WhitespaceEnv,
+ 'xor_vt': XorEnv
}
# When given a Parser object, decompose_parser() deduces its type from the vtable, and returns the appropriate HParserEnv subclass, containing member parsers and args
--
GitLab