def get_clang_TranslationUnit(path="t.cpp", in_args=[], in_str="", options=0):
"""
Get the TranslationalUnit for the input fule listed:
Parameters ::
- path: The path of the file to parse (not read if in_str is set)
- in_args: additional arguments to pass to clang
- in_str: input string to parse instead of the file path.
- options: clang.cindex.Options
Returns ::
- A TranslationalUnits
"""
# Make sure we are parsing as std c++11
args = '-x c++ --std=c++11'.split()
# Add the include files for the standard library.
syspath = ccsyspath.system_include_paths('clang++')
incargs = [b'-I' + inc for inc in syspath]
# turn args into a list of args (in_args may contain more includes)
args = args + incargs + in_args
# Create a clang index to parse into
index = cl.Index.create()
unsaved_files = None
# If we are parsing from a string instead
if in_str:
unsaved_files = [(path, in_str)]
return index.parse(path,
args=args,
options=options,
unsaved_files=unsaved_files)
def get_args(self):
import ccsyspath
syspath = ccsyspath.system_include_paths('clang++')
incargs = [b'-I' + inc for inc in syspath]
args = '-x c++'.split() + incargs
logging.debug(
'[INFO]: Count of dependency args by syspath: {0}.'.format(
len(args)))
return args
def parse_ast(self,
program: str,
imports: list = None,
thread_nr: int = 0):
os.makedirs('temp', exist_ok=True)
# Create temp file path for each trhead for clang to save in memory contents
temp_file_path = os.path.join('temp', f'tmp{thread_nr}.cpp')
# Set arguments and add compiler system include paths (with ccsyspath)
args = '-x c++ --std=c++20'.split()
syspath = ccsyspath.system_include_paths('clang')
incargs = [b'-I' + inc for inc in syspath]
args = args + incargs
if imports is None:
program, imports = extract_imports(program)
else:
imports = [
ele for ele in imports[1:-1].split("'")
if ele != '' and ele != ', '
]
# Preprocess the program, expand the macros
preprocessed_program = self.preprocess_program(program, temp_file_path,
imports)
# print(preprocessed_program)
# Parse the program to a clang AST
tu = self.index.parse(temp_file_path,
unsaved_files=[(temp_file_path,
preprocessed_program)],
args=args,
options=0)
# Retrieve only the cursor items (children) that contain the program code (no import code)
cursor_items = self.get_cursor_items(tu.cursor, temp_file_path)
# Create a root node
root_node = Node(self.tokenizers['RES'].get_token('root'),
is_reserved=True)
# for cursor_item in cursor_items:
# for c in cursor_item.walk_preorder():
# print(f'spelling: {c.spelling}, kind: {c.kind.name}, type spelling: {c.type.spelling}, return type: {c.type.get_result().spelling}, type kind: {c.type.kind}')
# Parse each cursor item
for cursor_item in cursor_items:
self.parse_item(cursor_item, root_node, program)
shutil.rmtree('temp')
# Return the root node filled with children to form the AST
return root_node
def main():
syspath = ccsyspath.system_include_paths('clang++')
incargs = [b'-I' + inc for inc in syspath]
args = '-x c++'.split() + incargs
source_code = [r'placeholder']
for path_item in source_code:
for file_item in filter_file_by_suffixation(path_item, '.c', '.cxx',
'.cc', '.c++', '.cpp'):
index = clang.cindex.Index.create()
tu = index.parse(file_item, args=args)
root = tu.cursor
_find_bare_threads(file_item, root, 0)
def parse(path):
index = clang.cindex.Index.create()
# Clang can't parse files with missing definitions, add static library definition or not?
args = ['-x', 'c++', '-std=c++11', '-fparse-all-comments', '-DIGL_STATIC_LIBRARY']
args.append('-I/usr/include/eigen3/') # TODO Properly add all needed includes
syspath = ccsyspath.system_include_paths('clang++') # Add the system libraries
incargs = [(b'-I' + inc).decode("utf-8") for inc in syspath]
args.extend(incargs)
tu = index.parse(path, args)
objects = {"functions": [], "enums": [], "namespaces": [], "classes": [], "structs": []}
traverse(tu.cursor, path, objects)
return objects
def parse(path):
index = clang.cindex.Index.create()
# Clang can't parse files with missing definitions, add static library definition or not?
args = ['-x', 'c++', '-std=c++11', '-fparse-all-comments', '-DIGL_STATIC_LIBRARY']
args.append('-I/usr/include/eigen3/') # TODO Properly add all needed includes
syspath = ccsyspath.system_include_paths('clang++') # Add the system libraries
incargs = [(b'-I' + inc).decode("utf-8") for inc in syspath]
args.extend(incargs)
tu = index.parse(path, args)
objects = {"functions": [], "enums": [], "namespaces": [], "classes": [], "structs": []}
traverse(tu.cursor, path, objects)
return objects
def parse(src):
"""Parse a string of source code and return the translation unit
"""
# Find the clang compiler
clang = os.path.join(os.environ['LLVM_HOME'], 'bin', 'clang++')
# Find the system paths for the compiler
syspath = ccsyspath.system_include_paths(clang)
syspath = [ p.decode('utf-8') for p in syspath ]
# Build up an appropriate set of compiler flags
args = '-x c++ --std=c++11'.split()
args += llvm_config('--cppflags')
args += [ '-I' + inc for inc in syspath ]
# Parse the source passed in
return glud.parse_string(src, name='input.cpp', args=args)
def parse(src):
"""Parse a string of source code and return the translation unit
"""
# Find the clang compiler
clang = os.path.join(os.environ['LLVM_HOME'], 'bin', 'clang++')
# Find the system paths for the compiler
syspath = ccsyspath.system_include_paths(clang)
syspath = [p.decode('utf-8') for p in syspath]
# Build up an appropriate set of compiler flags
args = '-x c++ --std=c++11'.split()
args += llvm_config('--cppflags')
args += ['-I' + inc for inc in syspath]
# Parse the source passed in
return glud.parse_string(src, name='input.cpp', args=args)
libclang_path = '/Library/Developer/CommandLineTools/usr/lib/libclang.dylib'
pipe = os.popen('mdfind -name "libclang.dylib"').readlines()
if not pipe:
raise Exception("Please install clang with libclang.dylib")
elif platform.system() == "Linux": # IS_LINUX
slash = '/'
libclang_path = '/usr/lib/llvm-7.0/lib/libclang-7.0.so.1'
if not os.path.exists(libclang_path):
raise Exception("Please install clang with libclang.so")
if Config.loaded == True:
pass
else:
Config.set_library_file(libclang_path)
syspath = ccsyspath.system_include_paths('clang++')
incargs = [b'-I' + inc for inc in syspath]
args = '-x c++'.split() + incargs
parser = OptionParser()
parser.add_option("-s", "--source_dir", action="store", dest="source_dir", help="read input data from source directory")
parser.add_option("-t", "--target_file", action="store", dest="target_file", help="parse data to target file")
(options, _) = parser.parse_args()
INPUT_SOURCE_ORIGIN = []
INPUT_SOURCE_ORIGIN.append(options.source_dir)
_VAR_DECL = [] # record variable offsets
_DECL_REF_EXPR = [] # record variable reference offsets
_NODE_DICT = {} # record node: domain
_GLOBAL_VAR_COUNT = 0 # records the number of global variables
_BARE_THREAD_COUNT = 0 # records the number of bare thread
def translation_units(compile_commands: cindex.CompilationDatabase, cache_path: Optional[str]) -> Iterator[typing.Union[cindex.TranslationUnit, SerializedTU]]:
'''Returns an iterator over a translation unit for each file in the compilation database.'''
compile_command: cindex.CompileCommand
for compile_command in compile_commands.getAllCompileCommands():
if cache_path:
full_path = os.path.join(
compile_command.directory,
compile_command.filename,
)
serialized_tu = read_tu(
cache_path,
full_path,
)
modified_time = os.path.getmtime(full_path)
if serialized_tu.serialization_time >= modified_time:
log(LogLevel.INFO, f'Using cached analysis for {full_path}')
yield serialized_tu
continue
log(
LogLevel.INFO,
f'parsing {compile_command.filename}'
)
try:
if 'lua' in compile_command.filename:
continue
os.chdir(compile_command.directory)
translation_unit = cindex.TranslationUnit.from_source(
os.path.join(compile_command.directory,
compile_command.filename),
args=[arg for arg in compile_command.arguments
if arg != compile_command.filename] + ['-I' + inc.decode() for inc in ccsyspath.system_include_paths('clang')],
)
for diag in translation_unit.diagnostics:
log(
LogLevel.WARNING,
f'Parsing: {compile_command.filename}: {diag}'
)
yield translation_unit
except cindex.TranslationUnitLoadError:
log(
LogLevel.WARNING,
f'could not parse {os.path.join(compile_command.directory, compile_command.filename)}',
)
def add_required_include_paths(extra_include_paths=[], compiler_path='clang'):
include_paths = ccsyspath.system_include_paths(
'clang') + extra_include_paths
for p in include_paths:
compiler_args.extend(['-I', p])
#
# You should have received a copy of the GNU Affero General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
import os
import sys
import argparse
import logging
from collections import defaultdict
tob_path = os.environ['TRAILOFBITS_LIBRARIES']
cc_path = tob_path + "/llvm/bin/clang"
try:
import ccsyspath
syspath = ccsyspath.system_include_paths(cc_path)
print(syspath)
except ImportError:
syspath = list()
SUPPORTED_ARCH = ["x86", "amd64"]
SUPPORTED_LIBRARY_TYPE = ["c", "cpp"]
ARCH_NAME = ""
ABI_LIBRARY_TYPE = "c"
logging.basicConfig(filename="debug.log", level=logging.DEBUG)
cc_pragma = """
def test_c_compiler(self):
lines = ccsyspath.system_include_paths('clang', cpp=False)
self.assertEqual(list, type(lines))
self.assertTrue(len(lines) > 0)
