def cppcheck_ast(sourcefile):
subprocess.call([CPPCHECK, '--dump', '--max-configs=1', sourcefile])
data = cppcheckdata.parsedump(sourcefile + '.dump')
cfg = data.configurations[0]
ret = []
for func in cfg.functions:
name = func.name
if func.type == 'Destructor':
name = '~' + name
s = '<Function'
s = s + ' name="' + name + '"'
s = s + ' filename="' + func.tokenDef.file + '"'
s = s + ' line="' + str(func.tokenDef.linenr) + '"'
s = s + '/>'
ret.append(s)
for scope in cfg.scopes:
if scope.type != 'Function':
continue
argStart = scope.bodyStart
while argStart and argStart.str != '(':
argStart = argStart.previous
s = '<Function'
s = s + ' name="' + scope.className + '"'
s = s + ' filename="' + argStart.file + '"'
s = s + ' line="' + str(argStart.linenr) + '"'
s = s + '/>'
if s not in ret:
ret.append(s)
ret.sort()
return ret
def main():
print(sys.argv)
function_calls = defaultdict(set)
files_deps = defaultdict(set)
func_decl_file = {} # type: Dict[str, str]
for dirpath, _, files in walk('.'):
for basename in files:
file = os.path.join(dirpath, basename)
if file[-2:] == '.c':
print(file)
call([CPP_CHECK, '--dump', file] + sys.argv[1:])
dump_file = file + '.dump'
d = parsedump(dump_file)
os.remove(dump_file)
try:
# functions = d.configurations[0].functions
scopes = d.configurations[0].scopes
tokens = d.configurations[0].tokenlist # type: List[Token]
for scope in scopes:
scope.tokens = [
t for t in tokens if t.scopeId == scope.Id
]
for scope in scopes:
if scope.type == 'Function':
func_decl_file[scope.function.name] = file
for t in scope.tokens:
if t.function is not None:
function_calls[scope.function.name].add(
t.function.name)
except Exception as e:
print(e)
# Dépendences entre fichiers .c
for caller, callees in function_calls.items():
for callee in callees:
if caller in func_decl_file and callee in func_decl_file:
files_deps[func_decl_file[caller]].add(func_decl_file[callee])
pprint(function_calls, indent=2)
pprint(func_decl_file, indent=2)
pprint(files_deps, indent=2)
dot_graph('calls', function_calls)
dot_graph('files', files_deps)
dot_graph('full', function_calls, reverse_dict(func_decl_file))
def main_run_collect(self, dump_file, source_file=''):
''' input: a cppcheck 'dump' file containing an Abstract Syntax Tree (AST), symbol table, and token list.
returns: None
side-effects: updates datbase with information about this unit analysis
'''
self.source_file = source_file
self.current_file_under_analysis = dump_file
# PARSE INPUT
data = cppcheckdata.parsedump(dump_file)
analysis_unit_dict = {}
# GIVE TREE WALKER ACCESS TO SOURCE FILE FOR DEBUG PRINT
if self.source_file and os.path.exists(self.source_file):
with open(self.source_file) as f:
self.source_file_lines = f.readlines()
#print "yes"
else:
print "no %s %s" % (self.source_file, self.debug)
# for c in data.configurations: #todo: what is a data configuration? -- Check for multiple
for c in data.configurations[:
1]: # MODIFIED TO ONLY TEST THE FIRST CONFIGURATION
# ADD AST DECORATION PLACEHOLDERS
c = self.init_cppcheck_config_data_structures(c)
c = self.init_cppcheck_config_functions(c)
# REFRESH VARIABLES
self.function_graph = nx.DiGraph()
# GET DICT OF ALL GLOBALLY SCOPED FUNCTIONS
analysis_unit_dict = self.find_functions(c)
sorted_analysis_unit_dict = analysis_unit_dict
# WILL BECOME AN ORDERED DICT IF self.should_sort_by_function_graph
# FIND ORDER FOR FUNCTION GRAPH EXPLORATION ( topo sort, if possible, otherwise todo ??)
if self.should_sort_by_function_graph:
self.build_function_graph(
analysis_unit_dict) # WILL USE DAG SUBGRAPH
sorted_analysis_unit_dict = self.make_sorted_analysis_unit_dict_from_function_graph(
analysis_unit_dict) # RETURNS ORDERED DICT
self.all_sorted_analysis_unit_dicts.append(
sorted_analysis_unit_dict)
# COLLECT ALL TOKEN PARSE TREES FOR EACH FUNCTION
for function_dict in sorted_analysis_unit_dict.values():
self.collect_constraints(function_dict)
if self.SHOULD_PRINT_CONSTRAINTS:
self.print_all_computed_unit_constraints()
self.print_all_df_constraints()
self.print_all_conversion_factor_constraints()
self.print_all_known_symbol_constraints()
self.print_all_naming_constraints()
self.configurations.append(c)
def process_cppcheck_dump_file(file):
""" Process the dump file data.
This processes variables and scopes configurations from cppcheckdata.parsedump.
The variables configuration contains only variables.
The scopes configuration contain the following types:
type="Namespace", type="Function", type="Class", type="Struct", type="Enum"
All scopes configuration types are checked except type="Namespace"
"""
print('Checking ' + file + '...')
data = cppcheckdata.parsedump(file)
for cfg in data.configurations:
if len(data.configurations) > 1:
print('Checking ' + file + ', config "' + cfg.name + '"...')
# process variables configuration
if RE_VARNAME:
for var in cfg.variables:
res = re.match(RE_VARNAME, var.nameToken.str)
if not res:
report_error(var.typeStartToken,
'style',
'Variable '
+ var.nameToken.str
+ ' violates naming convention')
# processes scope configuration type="Function"
if RE_FUNCTIONNAME:
classes = create_class_list(cfg.scopes)
for scope in cfg.scopes:
if scope.type == 'Function':
# delete constructors and destructors from function check
if scope.className in classes:
continue
res = re.match(RE_FUNCTIONNAME, scope.className)
if not res:
report_error(scope.classStart, 'style',
'Function '
+ scope.className
+ ' violates naming convention')
# process scope configuration type="Class", type="Struct", and type="Enum"
if RE_CLASSNAME:
for scope in cfg.scopes:
if (scope.type == 'Class' or scope.type == 'Struct'
or scope.type == 'Enum'):
res = re.match(RE_CLASSNAME, scope.className)
if not res:
report_error(scope.classStart, 'style',
scope.type + ' '
+ scope.className
+ ' violates naming convention')
# parse command line
args = parser.parse_args()
# now operate on each file in turn
dumpfiles = find_dump_files(args.paths)
for dumpfile in dumpfiles:
if not args.quiet:
print('Checking ' + dumpfile + '...')
srcfile = dumpfile.rstrip('.dump')
# at the start of the check, we don't know if code is Y2038 safe
y2038safe = False
# load XML from .dump file
data = cppcheckdata.parsedump(dumpfile)
# go through each configuration
for cfg in data.configurations:
if not args.quiet:
print('Checking ' + dumpfile + ', config "' + cfg.name + '"...')
safe_ranges = []
safe = -1
time_bits_defined = False
for directive in cfg.directives:
# track source line number
if directive.file == srcfile:
srclinenr = directive.linenr
# check for correct _TIME_BITS if present
if re_define_time_bits_64.match(directive.str):
time_bits_defined = True
elif re_define_time_bits.match(directive.str):
def process(dumpfiles, configfile, debugprint=False):
errors = []
conf = loadConfig(configfile)
for afile in dumpfiles:
if not afile[-5:] == '.dump':
continue
print('Checking ' + afile + '...')
data = cppcheckdata.parsedump(afile)
## Check File naming
if "RE_FILE" in conf and conf["RE_FILE"]:
mockToken = dataStruct(afile[:-5], "0", afile[afile.rfind('/')+1:-5])
msgType = 'File name'
for exp in conf["RE_FILE"]:
evalExpr(conf["RE_FILE"], exp, mockToken, msgType, errors)
## Check Namespace naming
if "RE_NAMESPACE" in conf and conf["RE_NAMESPACE"]:
for tk in data.rawTokens:
if (tk.str == 'namespace'):
mockToken = dataStruct(tk.next.file, tk.next.linenr, tk.next.str)
msgType = 'Namespace'
for exp in conf["RE_NAMESPACE"]:
evalExpr(conf["RE_NAMESPACE"], exp, mockToken, msgType, errors)
for cfg in data.configurations:
if len(data.configurations) > 1:
print('Checking ' + afile + ', config "' + cfg.name + '"...')
if "RE_VARNAME" in conf and conf["RE_VARNAME"]:
for var in cfg.variables:
if var.nameToken and var.access != 'Global' and var.access != 'Public' and var.access != 'Private':
prev = var.nameToken.previous
varType = prev.str
while "*" in varType and len(varType.replace("*", "")) == 0:
prev = prev.previous
varType = prev.str + varType
if debugprint:
print("Variable Name: " + str(var.nameToken.str))
print("original Type Name: " + str(var.nameToken.valueType.originalTypeName))
print("Type Name: " + var.nameToken.valueType.type)
print("Sign: " + str(var.nameToken.valueType.sign))
print("variable type: " + varType)
print("\n")
print("\t-- {} {}".format(varType, str(var.nameToken.str)))
if conf["skip_one_char_variables"] and len(var.nameToken.str) == 1:
continue
if varType in conf["var_prefixes"]:
if not var.nameToken.str.startswith(conf["var_prefixes"][varType]):
errors.append(reportError(
var.typeStartToken.file,
var.typeStartToken.linenr,
'style',
'Variable ' +
var.nameToken.str +
' violates naming convention'))
mockToken = dataStruct(var.typeStartToken.file, var.typeStartToken.linenr, var.nameToken.str)
msgType = 'Variable'
for exp in conf["RE_VARNAME"]:
evalExpr(conf["RE_VARNAME"], exp, mockToken, msgType, errors)
## Check Private Variable naming
if "RE_PRIVATE_MEMBER_VARIABLE" in conf and conf["RE_PRIVATE_MEMBER_VARIABLE"]:
# TODO: Not converted yet
for var in cfg.variables:
if (var.access is None) or var.access != 'Private':
continue
mockToken = dataStruct(var.typeStartToken.file, var.typeStartToken.linenr, var.nameToken.str)
msgType = 'Private member variable'
for exp in conf["RE_PRIVATE_MEMBER_VARIABLE"]:
evalExpr(conf["RE_PRIVATE_MEMBER_VARIABLE"], exp, mockToken, msgType, errors)
## Check Public Member Variable naming
if "RE_PUBLIC_MEMBER_VARIABLE" in conf and conf["RE_PUBLIC_MEMBER_VARIABLE"]:
for var in cfg.variables:
if (var.access is None) or var.access != 'Public':
continue
mockToken = dataStruct(var.typeStartToken.file, var.typeStartToken.linenr, var.nameToken.str)
msgType = 'Public member variable'
for exp in conf["RE_PUBLIC_MEMBER_VARIABLE"]:
evalExpr(conf["RE_PUBLIC_MEMBER_VARIABLE"], exp, mockToken, msgType, errors)
## Check Global Variable naming
if "RE_GLOBAL_VARNAME" in conf and conf["RE_GLOBAL_VARNAME"]:
for var in cfg.variables:
if (var.access is None) or var.access != 'Global':
continue
mockToken = dataStruct(var.typeStartToken.file, var.typeStartToken.linenr, var.nameToken.str)
msgType = 'Public member variable'
for exp in conf["RE_GLOBAL_VARNAME"]:
evalExpr(conf["RE_GLOBAL_VARNAME"], exp, mockToken, msgType, errors)
## Check Functions naming
if "RE_FUNCTIONNAME" in conf and conf["RE_FUNCTIONNAME"]:
for token in cfg.tokenlist:
if token.function:
if token.function.type == 'Constructor' or token.function.type == 'Destructor':
continue
retval = token.previous.str
prev = token.previous
while "*" in retval and len(retval.replace("*", "")) == 0:
prev = prev.previous
retval = prev.str + retval
if debugprint:
print("\t:: {} {}".format(retval, token.function.name))
if retval and retval in conf["function_prefixes"]:
if not token.function.name.startswith(conf["function_prefixes"][retval]):
errors.append(reportError(
token.file, token.linenr, 'style', 'Function ' + token.function.name + ' violates naming convention'))
mockToken = dataStruct(token.file, token.linenr, token.function.name)
msgType = 'Function'
for exp in conf["RE_FUNCTIONNAME"]:
evalExpr(conf["RE_FUNCTIONNAME"], exp, mockToken, msgType, errors)
## Check Class naming
if "RE_CLASS_NAME" in conf and conf["RE_CLASS_NAME"]:
for fnc in cfg.functions:
#Check if it is Constructor/Destructor
if (fnc.type == 'Constructor' or fnc.type == 'Destructor'):
mockToken = dataStruct(fnc.tokenDef.file, fnc.tokenDef.linenr, fnc.name)
msgType = 'Class ' + fnc.type
for exp in conf["RE_CLASS_NAME"]:
evalExpr(conf["RE_CLASS_NAME"], exp, mockToken, msgType, errors)
return errors
def get_cppcheck_config_data_structure(self, dump_file):
data = cppcheckdata.parsedump(dump_file)
for c in data.configurations[:1]:
return c
if __name__ == '__main__':
args = get_args()
if args.verify:
VERIFY = True
if not args.dumpfile:
if not args.quiet:
print("no input files.")
sys.exit(0)
for dumpfile in args.dumpfile:
if not args.quiet:
print('Checking %s...' % dumpfile)
data = cppcheckdata.parsedump(dumpfile)
if VERIFY:
VERIFY_ACTUAL = []
VERIFY_EXPECTED = []
for tok in data.rawTokens:
if tok.str.startswith('//') and 'TODO' not in tok.str:
for word in tok.str[2:].split(' '):
if re.match(r'cert-[A-Z][A-Z][A-Z][0-9][0-9].*', word):
VERIFY_EXPECTED.append(
str(tok.linenr) + ':' + word)
for cfg in data.configurations:
if (len(data.configurations) > 1) and (not args.quiet):
print('Checking %s, config %s...' % (dumpfile, cfg.name))
exp05(cfg)
def process_dump_file(file):
""" Process the dump file data.
This processes variables and scopes configurations from cppcheckdata.parsedump.
The variables configuration contains only variables.
The scopes configuration contain the following types:
type="Namespace", type="Function", type="Class", type="Struct", type="Enum"
All scopes configuration types are checked except type="Namespace"
This function is adapted from the cppcheck naming.py file.
"""
variable_names = 0
function_names = 0
class_names = 0
filepath = __astyle_src_dir + file
data = cppcheckdata.parsedump(filepath)
for cfg in data.configurations:
if len(data.configurations) > 1:
print('Checking ' + file + ', config "' + cfg.name + '"...')
# process variables configuration
if RE_VARNAME:
for var in cfg.variables:
# the following 'if' from version 1.88 corrects a bug
# python aborted when there is no variable name in a function argument
if var.nameToken:
variable_names += 1
res = re.match(RE_VARNAME, var.nameToken.str)
if not res:
report_error(
var.typeStartToken, 'style', 'Variable ' +
var.nameToken.str + ' violates naming convention')
# processes scope configuration type="Function"
if RE_FUNCTIONNAME:
# get class names to eliminate constructors and destructors
class_list = create_class_list(cfg.scopes)
for scope in cfg.scopes:
if scope.type == 'Function':
# delete constructors and destructors from function check
# they have class names not function names
if scope.className in class_list:
continue
function_names += 1
res = re.match(RE_FUNCTIONNAME, scope.className)
if not res:
report_error(
scope.bodyStart, 'style', 'Function ' +
scope.className + ' violates naming convention')
# process scope configuration type="Class", type="Struct", and type="Enum"
if RE_CLASSNAME:
for scope in cfg.scopes:
if scope.type in ('Class', 'Struct', 'Enum'):
class_names += 1
res = re.match(RE_CLASSNAME, scope.className)
if not res:
report_error(
scope.bodyStart, 'style', scope.type + ' ' +
scope.className + ' violates naming convention')
else:
# verify scope types
if scope.type in ('Global', 'Namespace', 'Function'):
continue
if scope.type in ('If', 'Else', 'While', 'For', 'Do',
'Switch'):
continue
print('Unrecognized scope.type: ' + scope.type)
# end of top 'for' loop
print('variables:', variable_names, ' functions:', function_names,
' classes:', class_names)
def run_checks(self):
num_raw_tokens = 0
# Remove duplicates from dump file list
self.args.dumpfile = list(dict.fromkeys(self.args.dumpfile))
# Run metric checks for each dump file
for dumpfile in self.args.dumpfile:
if not self.args.quiet:
printf("Checking %s...\n", dumpfile)
self.statistics_list.append(dumpfile)
data = cppcheckdata.parsedump(dumpfile)
if self.args.verify:
for token in data.rawTokens[num_raw_tokens:]:
if token.str.startswith('//') and 'TODO' not in token.str:
for word in token.str[2:].split(' '):
if word.startswith("HIS-"):
self.verify_expected.append(token.file + ':' +
str(token.linenr) +
':' + word)
cfg_idx = 0
for cfg in data.configurations:
if (cfg_idx < 1):
self.execute_metric_check("COMF", self.his_comf, cfg,
data.rawTokens[num_raw_tokens:])
self.execute_metric_check("PATH", self.his_path, cfg)
self.execute_metric_check("GOTO", self.his_goto, cfg)
self.execute_metric_check("STCYC", self.his_stcyc, cfg)
self.execute_metric_check("CALLING", self.his_calling, cfg)
self.execute_metric_check("CALLS", self.his_calls, cfg)
self.execute_metric_check("PARAM", self.his_param, cfg)
self.execute_metric_check("STMT", self.his_stmt, cfg)
self.execute_metric_check("LEVEL", self.his_level, cfg)
self.execute_metric_check("RETURN", self.his_return, cfg)
self.execute_metric_check("VOCF", self.his_vocf, cfg)
cfg_idx = cfg_idx + 1
# Since Cppcheck 2.4 rawTokens has been moved from class to instance level.
# It will be initialized for each dump file analysis.
if 'rawTokens' not in data.__dict__:
num_raw_tokens = len(data.rawTokens)
if not self.args.quiet:
printf("Checking metrics for all dump files...\n")
# Check for violations of HIS-CALLING after all dump files have been analyzed.
self.execute_metric_check("CALLING", self.his_calling_result)
# Check for violation of HIS-VOCF after all dump files have been analyzed.
self.execute_metric_check("VOCF", self.his_vocf_result)
# Check for violations of HIS-NRECUR after all dump files have been analyzed.
self.execute_metric_check("NRECUR", self.his_num_recursions)
if self.args.verify:
for expected in self.verify_expected:
if expected not in self.verify_actual:
printf("Expected but not seen: %s\n", expected)
for actual in self.verify_actual:
if actual not in self.verify_expected:
printf("Not expected: %s\n", actual)
# Print summary if not suppressed by command line
if not self.args.no_summary and not self.args.verify:
printf("\n---------------------------\n")
printf("--- Summary of violations\n")
printf("---------------------------\n")
for key in self.his_stats:
if (self.his_stats[key] == "Suppressed"):
printf("HIS-%s: %s\n", key.ljust(10), self.his_stats[key])
else:
printf("HIS-%s: %d\n", key.ljust(10), self.his_stats[key])
printf("\n")
if self.args.statistics and not self.args.verify:
printf("\n---------------------------\n")
printf("--- Statistics information\n")
printf("---------------------------\n")
for item in self.statistics_list:
printf("%s\n", item)
printf("\n")
def process(dumpfiles, configfile, debugprint=False):
errors = []
conf = loadConfig(configfile)
for afile in dumpfiles:
if not afile[-5:] == '.dump':
continue
print('Checking ' + afile + '...')
data = cppcheckdata.parsedump(afile)
for cfg in data.configurations:
if len(data.configurations) > 1:
print('Checking ' + afile + ', config "' + cfg.name + '"...')
if conf["RE_VARNAME"]:
for var in cfg.variables:
if var.nameToken:
prev = var.nameToken.previous
varType = prev.str
while "*" in varType and len(varType.replace("*",
"")) == 0:
prev = prev.previous
varType = prev.str + varType
if debugprint:
print("Variable Name: " + str(var.nameToken.str))
print(
"original Type Name: " +
str(var.nameToken.valueType.originalTypeName))
print("Type Name: " + var.nameToken.valueType.type)
print("Sign: " + str(var.nameToken.valueType.sign))
print("variable type: " + varType)
print("\n")
print("\t-- {} {}".format(varType,
str(var.nameToken.str)))
if conf["skip_one_char_variables"] and len(
var.nameToken.str) == 1:
continue
if varType in conf["var_prefixes"]:
if not var.nameToken.str.startswith(
conf["var_prefixes"][varType]):
errors.append(
reportError(
var.typeStartToken.file,
var.typeStartToken.linenr, 'style',
'Variable ' + var.nameToken.str +
' violates naming convention'))
res = re.match(conf["RE_VARNAME"], var.nameToken.str)
if not res:
errors.append(
reportError(
var.typeStartToken.file,
var.typeStartToken.linenr, 'style',
'Variable ' + var.nameToken.str +
' violates naming convention'))
if conf["RE_PRIVATE_MEMBER_VARIABLE"]:
# TODO: Not converted yet
for var in cfg.variables:
if (var.access is None) or var.access != 'Private':
continue
res = re.match(conf["RE_PRIVATE_MEMBER_VARIABLE"],
var.nameToken.str)
if not res:
errors.append(
reportError(
var.typeStartToken.file,
var.typeStartToken.linenr, 'style',
'Private member variable ' +
var.nameToken.str +
' violates naming convention'))
if conf["RE_FUNCTIONNAME"]:
for token in cfg.tokenlist:
if token.function:
retval = token.previous.str
prev = token.previous
while "*" in retval and len(retval.replace("*",
"")) == 0:
prev = prev.previous
retval = prev.str + retval
if debugprint:
print("\t:: {} {}".format(retval,
token.function.name))
if retval and retval in conf["function_prefixes"]:
if not token.function.name.startswith(
conf["function_prefixes"][retval]):
errors.append(
reportError(
token.file, token.linenr, 'style',
'Function ' + token.function.name +
' violates naming convention'))
res = re.match(conf["RE_FUNCTIONNAME"],
token.function.name)
if not res:
errors.append(
reportError(
token.file, token.linenr, 'style',
'Function ' + token.function.name +
' violates naming convention'))
return errors
def check_errors_with_low_confidence_when_top3_units(
self, cppcheck_configuration_unit, sorted_analysis_unit_dict):
# need to work on another copy of cppcheckdata
# check after all errors are collected
con.print_known_unit_variables()
data = cppcheckdata.parsedump(self.dump_file)
for c in data.configurations[:1]:
break
# copy token and function data from original config
tokenlist = {}
for t in cppcheck_configuration_unit.tokenlist:
tokenlist[t.Id] = (t.isRoot, t.isDimensionless)
functionlist = {}
returnexprtokenlist = {}
for f in cppcheck_configuration_unit.functions:
functionlist[f.Id] = (
f.return_units, f.arg_units, f.return_arg_var_nr,
f.return_expr_root_token,
f.is_unit_propagation_based_on_constants,
f.is_unit_propagation_based_on_unknown_variable,
f.is_unit_propagation_based_on_weak_inference,
f.maybe_generic_function)
if f.return_expr_root_token:
returnexprtokenlist[f.return_expr_root_token.Id] = None
returntokenlist = {}
for t in c.tokenlist:
(isRoot, isDimensionless) = tokenlist[t.Id]
t.units = []
t.isKnown = False
t.is_unit_propagation_based_on_constants = False
t.is_unit_propagation_based_on_unknown_variable = False
t.is_unit_propagation_based_on_weak_inference = False
t.isRoot = isRoot
t.isDimensionless = isDimensionless
if t.str == "return":
returntokenlist[t.Id] = t
if t.Id in returnexprtokenlist:
returnexprtokenlist[t.Id] = t
for f in c.functions:
(return_units, arg_units, return_arg_var_nr,
return_expr_root_token, is_unit_propagation_based_on_constants,
is_unit_propagation_based_on_unknown_variable,
is_unit_propagation_based_on_weak_inference,
maybe_generic_function) = functionlist[f.Id]
if return_expr_root_token:
return_expr_root_token = returnexprtokenlist[
return_expr_root_token.Id]
f.return_units = [] #return_units
f.arg_units = [] #arg_units
f.return_arg_var_nr = return_arg_var_nr
f.return_expr_root_token = return_expr_root_token
f.is_unit_propagation_based_on_constants = is_unit_propagation_based_on_constants
f.is_unit_propagation_based_on_unknown_variable = is_unit_propagation_based_on_unknown_variable
f.is_unit_propagation_based_on_weak_inference = is_unit_propagation_based_on_weak_inference
f.maybe_generic_function = maybe_generic_function
for arg_number in f.argument.keys():
f.arg_units.append([])
# collect return units of all functions
returnlist = {}
for function_dict in sorted_analysis_unit_dict.values():
if not function_dict['scopeObject'].function:
continue
if function_dict['scopeObject'].function.return_arg_var_nr:
continue
if function_dict['scopeObject'].function.maybe_generic_function:
continue
returnlist[function_dict['scopeObject'].function.Id] = []
for root_token in function_dict['root_tokens']:
if root_token.str == 'return':
t = returntokenlist[root_token.Id]
self.check_error_when_top3_units(t)
#RETURN STATEMENT WITH UNITS - STORE UNITS
if t.units:
for u in t.units:
if u not in returnlist[
function_dict['scopeObject'].function.Id]:
returnlist[function_dict['scopeObject'].
function.Id].append(u)
tw = TreeWalker(None)
tw.generic_recurse_and_apply_function(t, tw.reset_tokens)
for f in c.functions:
return_units = returnlist.get(f.Id)
if return_units:
f.return_units = return_units
# check all errors
for e in self.all_errors:
con.FOUND_DERIVED_CU_VARIABLE = False
if e.is_warning:
continue
if e.dont_check_for_warning:
continue
if e.ERROR_TYPE == UnitErrorTypes.ADDITION_OF_INCOMPATIBLE_UNITS or \
e.ERROR_TYPE == UnitErrorTypes.COMPARISON_INCOMPATIBLE_UNITS:
root_token = None
# find token in the copy
for t in c.tokenlist:
if t.Id == e.token.Id:
root_token = t
break
if not root_token:
continue
self.check_error_when_top3_units(root_token)
if e.ERROR_TYPE == UnitErrorTypes.ADDITION_OF_INCOMPATIBLE_UNITS:
if con.FOUND_DERIVED_CU_VARIABLE:
if len(root_token.units) > 2:
e.is_warning = True
elif root_token.units:
e.is_warning = True
else:
units = []
if root_token.astOperand1 and root_token.astOperand2:
left_units = root_token.astOperand1.units
right_units = root_token.astOperand2.units
if not left_units:
pass #units = right_units
elif not right_units:
pass #units = left_units
else:
for lu in left_units:
if lu in right_units:
units.append(lu)
if con.FOUND_DERIVED_CU_VARIABLE:
if len(units) > 2:
e.is_warning = True
elif units:
e.is_warning = True
tw = TreeWalker(None)
tw.generic_recurse_and_apply_function(root_token,
tw.reset_tokens)
elif e.ERROR_TYPE == UnitErrorTypes.VARIABLE_MULTIPLE_UNITS:
i = 0
root_token = None
left_token = None
# find token in the copy
for t in c.tokenlist:
if t.Id == e.token.Id:
root_token = t
i = i + 1
elif t.Id == e.token_left.Id:
left_token = t
i = i + 1
if i == 2:
break
if (not root_token) or (not left_token):
continue
elif not root_token.astOperand2:
continue
elif not root_token.astOperand1:
continue
self.check_error_when_top3_units(root_token.astOperand1)
self.check_error_when_top3_units(root_token.astOperand2)
if (not left_token.isKnown
): #and root_token.astOperand2.units:
if con.FOUND_DERIVED_CU_VARIABLE:
if len(root_token.astOperand2.units) > 2:
e.is_warning = True
elif root_token.astOperand2.units:
e.is_warning = True
else:
if root_token.astOperand2.units and (
root_token.astOperand1.units
== root_token.astOperand2.units):
e.is_warning = True
if (not e.is_warning) and (
root_token.astOperand1.
is_unit_propagation_based_on_weak_inference):
units = []
for lu in root_token.astOperand1.units:
if lu in root_token.astOperand2.units:
units.append(lu)
if con.FOUND_DERIVED_CU_VARIABLE:
if len(units) > 2:
e.is_warning = True
elif units:
e.is_warning = True
#if not con.is_df_constraint_present(e.token_left, e.var_name):
# if root_token.astOperand2.units: #and (root_token.astOperand1.units == root_token.astOperand2.units):
# units = []
# for lu in root_token.astOperand1.units:
# if lu in root_token.astOperand2.units:
# units.append(lu)
# if con.FOUND_DERIVED_CU_VARIABLE:
# if len(units) > 2:
# e.is_warning = True
# elif units:
# e.is_warning = True
tw = TreeWalker(None)
tw.generic_recurse_and_apply_function(root_token,
tw.reset_tokens)
tokens = tokens[1:]
if not token:
continue
if token.str == '(':
if (token.previous.str[0].isalpha()):
inputs.append(token.previous.str)
elif token.str[0].isalpha():
inputs.append(token.str)
elif token.str[0].isdigit():
inputs.append(token.str)
else:
tokens.append(token.astOperand1)
tokens.append(token.astOperand2)
return str(inputs)
data = cppcheckdata.parsedump('test/1.c.dump')
for scope in data.scopes:
if scope.type == 'Function':
print scope.className
tok = scope.classStart
while tok and tok != scope.classEnd:
if tok.astOperand1:
astTop = tok
while astTop.astParent:
astTop = astTop.astParent
if astTop.str == '=' and astTop.astOperand1.variable:
print(astTop.Id + ' ' + astTop.astOperand1.str + ":=" + expr(astTop.astOperand2))
if astTop.str == 'return':
def main_run_check(self, dump_file, source_file=''):
''' PEFORM UNITS CHECKING
todo: pep8
input: a cppcheck 'dump' file containing an Abstract Syntax Tree (AST), symbol table, and token list.
returns: None
side-effects: updates datbase with information about this unit analysis
'''
if self.debug_verbose:
print inspect.stack()[0][3]
self.source_file = source_file
self.current_file_under_analysis = dump_file
# PARSE INPUT
data = cppcheckdata.parsedump(dump_file)
analysis_unit_dict = {}
# INITIALIZE ERROR CHECKING OBJECT
self.error_checker = ErrorChecker(self.debug, source_file, self)
# GIVE TREE WALKER ACCESS TO SOURCE FILE FOR DEBUG PRINT
if self.source_file and self.debug and os.path.exists(
self.source_file):
with open(self.source_file) as f:
self.source_file_lines = f.readlines()
print "yes"
else:
if self.debug:
print "no %s %s" % (self.source_file, self.debug)
# for c in data.configurations: #todo: what is a data configuration? -- Check for multiple
for c in data.configurations[:
1]: # MODIFIED TO ONLY TEST THE FIRST CONFIGURATION
self.current_configuration = c
# ADD AST DECORATION PLACEHOLDERS
c = self.init_cppcheck_config_data_structures(c)
# REFRESH VARIABLES
self.function_graph = nx.DiGraph()
# GET DICT OF ALL GLOBALLY SCOPED FUNCTIONS
analysis_unit_dict = self.find_functions(c)
sorted_analysis_unit_dict = analysis_unit_dict
# WILL BECOME AN ORDERED DICT IF self.should_sort_by_function_graph
# FIND ORDER FOR FUNCTION GRAPH EXPLORATION ( topo sort, if possible, otherwise todo ??)
if self.should_sort_by_function_graph:
self.build_function_graph(
analysis_unit_dict) # WILL USE DAG SUBGRAPH
sorted_analysis_unit_dict = self.make_sorted_analysis_unit_dict_from_function_graph(
analysis_unit_dict) # RETURNS ORDERED DICT
self.all_sorted_analysis_unit_dicts.append(
sorted_analysis_unit_dict)
# SPECIAL MODE FOR COUNTING / IDENTIFYING FILES WITH ROS UNITS
if self.SHOULD_ONLY_FIND_FILES_WITH_UNITS and self.found_ros_units_in_this_file:
return
##print "G nodes:%d edges:%d" % (G.order(), G.size())
if self.debug_print_function_topo_sort:
self.debug_print_function_graph(sorted_analysis_unit_dict)
# COLLECT ALL TOKEN PARSE TREES FOR EACH FUNCTION
for function_dict in sorted_analysis_unit_dict.values():
self.analyze_function(function_dict)
# SAVE OFF A POINTER TO THE TREE WALKERS USED IN THIS ANALYSIS FOR TESTING
self.all_tree_walkers.append(self.tw)
self.debug_function_count += 1
# if self.debug:
# for e in self.errors:
# print e['error_msg']
# self.error_checker.error_check_function_args_consistent(c)
# for e in self.error_checker.all_errors:
# self.errors.append(e)
# ADD ERROR OBJECTS FOUND BY ERROR CHECKER TO MY ERRORS
if self.debug:
self.error_checker.pretty_print()
# DEBUG COUNTERS
self.debug_configuration_count += 1
self.debug_function_count = 0
## - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
## OPTIONAL FIND UNITS ONLY AND WRITE TO DATABASE
## - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
if self.SHOULD_ONLY_FIND_FILES_WITH_UNITS and self.SHOULD_FIND_ALL_UNITS:
self.insert_file_unit_class_records()
return
## - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
## MAIN ERROR CHECKING
## - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
# for c in data.configurations:
for c in data.configurations[:1]: # FIRST ONLY
# CHECK THIS CONFIGURATION FOR ERRORS
self.error_checker.error_check_multiple_units(c)
self.error_checker.error_check_function_args_consistent(c)
# self.error_checker.error_check_unit_smell(c)
# todo: when the same error occurs in multiple data configs, where is a good place to catch that
# todo: add checking of addition on RH
for sorted_analysis_unit_dict in self.all_sorted_analysis_unit_dicts:
self.error_checker.error_check_comparisons(
sorted_analysis_unit_dict)
self.error_checker.error_check_logical_operators(
sorted_analysis_unit_dict)
self.error_checker.error_check_addition_of_incompatible_units(
sorted_analysis_unit_dict)
# GET ERRORS FROM ERROR CHECKER OBJECT (add to list of pointers)
for e in self.error_checker.all_errors:
self.errors.append(e)
# COUNTS ERRORS BY TYPE
# self.list_of_error_counts_by_type[e.ERROR_TYPE] += 1
if self.SHOULD_WRITE_RESULTS_TO_DATABASE:
if self.errors:
self.update_database(self.errors)
arg1 = None
arg2 = None
if token.astOperand2 and token.astOperand2.str == ',':
if token.astOperand2.astOperand1 and token.astOperand2.astOperand1.str == ',':
arg1 = token.astOperand2.astOperand1.astOperand1
arg2 = token.astOperand2.astOperand1.astOperand2
if token.astOperand1.str == 'memcmp' and (isLocalUnpackedStruct(arg1) or isLocalUnpackedStruct(arg2)):
reportError(
token, 'style', "Comparison of struct padding data " +
"(fix either by packing the struct using '#pragma pack' or by rewriting the comparison)", 'cert-EXP42-C')
# EXP46-C
# Do not use a bitwise operator with a Boolean-like operand
# int x = (a == b) & c;
def exp46(data):
for token in data.tokenlist:
if isBitwiseOp(token) and (isComparisonOp(token.astOperand1) or isComparisonOp(token.astOperand2)):
reportError(
token, 'style', 'Bitwise operator is used with a Boolean-like operand', 'cert-EXP46-c')
for arg in sys.argv[1:]:
print('Checking ' + arg + '...')
data = cppcheckdata.parsedump(arg)
for cfg in data.configurations:
if len(data.configurations) > 1:
print('Checking ' + arg + ', config "' + cfg.name + '"...')
exp42(cfg)
exp46(cfg)
def check_y2038_safe(dumpfile, quiet=False):
# at the start of the check, we don't know if code is Y2038 safe
y2038safe = False
# load XML from .dump file
data = cppcheckdata.parsedump(dumpfile)
# Convert dump file path to source file in format generated by cppcheck.
# For example after the following call:
# cppcheck ./src/my-src.c --dump
# We got 'src/my-src.c' value for 'file' field in cppcheckdata.
srcfile = dumpfile.rstrip('.dump')
srcfile = os.path.expanduser(srcfile)
srcfile = os.path.normpath(srcfile)
# go through each configuration
for cfg in data.configurations:
cfg = data.Configuration(cfg)
if not quiet:
print('Checking ' + srcfile + ', config "' + cfg.name + '"...')
safe_ranges = []
safe = -1
time_bits_defined = False
srclinenr = '0'
for directive in cfg.directives:
# track source line number
if directive.file == srcfile:
srclinenr = directive.linenr
# check for correct _TIME_BITS if present
if re_define_time_bits_64.match(directive.str):
time_bits_defined = True
elif re_define_time_bits.match(directive.str):
cppcheckdata.reportError(
directive, 'error',
'_TIME_BITS must be defined equal to 64', 'y2038',
'type-bits-not-64')
time_bits_defined = False
y2038safe = False
elif re_undef_time_bits.match(directive.str):
time_bits_defined = False
# check for _USE_TIME_BITS64 (un)definition
if re_define_use_time_bits64.match(directive.str):
safe = int(srclinenr)
# warn about _TIME_BITS not being defined
if not time_bits_defined:
cppcheckdata.reportError(
directive, 'warning',
'_USE_TIME_BITS64 is defined but _TIME_BITS was not',
'y2038', 'type-bits-undef')
elif re_undef_use_time_bits64.match(directive.str):
unsafe = int(srclinenr)
# do we have a safe..unsafe area?
if unsafe > safe > 0:
safe_ranges.append((safe, unsafe))
safe = -1
# check end of source beyond last directive
if len(cfg.tokenlist) > 0:
unsafe = int(cfg.tokenlist[-1].linenr)
if unsafe > safe > 0:
safe_ranges.append((safe, unsafe))
# go through all tokens
for token in cfg.tokenlist:
if token.str in id_Y2038:
if not any(lower <= int(token.linenr) <= upper
for (lower, upper) in safe_ranges):
cppcheckdata.reportError(token, 'warning',
token.str + ' is Y2038-unsafe',
'y2038', 'unsafe-call')
y2038safe = False
token = token.next
return y2038safe
print('Fatal error: file is not found: ' + filename)
sys.exit(1)
loadRuleTexts(filename)
elif ".dump" in arg:
continue
elif arg == "-generate-table":
generateTable()
else:
print('Fatal error: unhandled argument ' + arg)
sys.exit(1)
for arg in sys.argv[1:]:
if not arg.endswith('.dump'):
continue
data = cppcheckdata.parsedump(arg)
CHAR_BIT = data.platform.char_bit
SHORT_BIT = data.platform.short_bit
INT_BIT = data.platform.int_bit
LONG_BIT = data.platform.long_bit
LONG_LONG_BIT = data.platform.long_long_bit
POINTER_BIT = data.platform.pointer_bit
if VERIFY:
VERIFY_ACTUAL = []
VERIFY_EXPECTED = []
for tok in data.rawTokens:
if tok.str.startswith('//') and 'TODO' not in tok.str:
compiled = re.compile(r'[0-9]+\.[0-9]+')
for word in tok.str[2:].split(' '):
def process(dumpfiles, configfile, debugprint=False):
errors = []
conf = loadConfig(configfile)
for afile in dumpfiles:
if not afile[-5:] == '.dump':
continue
print('Checking ' + afile + '...')
data = cppcheckdata.parsedump(afile)
for cfg in data.configurations:
if len(data.configurations) > 1:
print('Checking ' + afile + ', config "' + cfg.name + '"...')
if conf["RE_VARNAME"]:
for var in cfg.variables:
if var.nameToken:
prev = var.nameToken.previous
varType = prev.str
while "*" in varType and len(varType.replace("*", "")) == 0:
prev = prev.previous
varType = prev.str + varType
if debugprint:
print("Variable Name: " + str(var.nameToken.str))
print("original Type Name: " + str(var.nameToken.valueType.originalTypeName))
print("Type Name: " + var.nameToken.valueType.type)
print("Sign: " + str(var.nameToken.valueType.sign))
print("variable type: " + varType)
print("\n")
print("\t-- {} {}".format(varType, str(var.nameToken.str)))
if conf["skip_one_char_variables"] and len(var.nameToken.str) == 1:
continue
if varType in conf["var_prefixes"]:
if not var.nameToken.str.startswith(conf["var_prefixes"][varType]):
errors.append(reportError(
var.typeStartToken.file,
var.typeStartToken.linenr,
'style',
'Variable ' +
var.nameToken.str +
' violates naming convention'))
res = re.match(conf["RE_VARNAME"], var.nameToken.str)
if not res:
errors.append(reportError(var.typeStartToken.file, var.typeStartToken.linenr, 'style', 'Variable ' +
var.nameToken.str + ' violates naming convention'))
if conf["RE_PRIVATE_MEMBER_VARIABLE"]:
# TODO: Not converted yet
for var in cfg.variables:
if (var.access is None) or var.access != 'Private':
continue
res = re.match(conf["RE_PRIVATE_MEMBER_VARIABLE"], var.nameToken.str)
if not res:
errors.append(reportError(var.typeStartToken.file, var.typeStartToken.linenr, 'style', 'Private member variable ' +
var.nameToken.str + ' violates naming convention'))
if conf["RE_FUNCTIONNAME"]:
for token in cfg.tokenlist:
if token.function:
retval = token.previous.str
prev = token.previous
while "*" in retval and len(retval.replace("*", "")) == 0:
prev = prev.previous
retval = prev.str + retval
if debugprint:
print("\t:: {} {}".format(retval, token.function.name))
if retval and retval in conf["function_prefixes"]:
if not token.function.name.startswith(conf["function_prefixes"][retval]):
errors.append(reportError(
token.file, token.linenr, 'style', 'Function ' + token.function.name + ' violates naming convention'))
res = re.match(conf["RE_FUNCTIONNAME"], token.function.name)
if not res:
errors.append(reportError(
token.file, token.linenr, 'style', 'Function ' + token.function.name + ' violates naming convention'))
return errors
