C++ linter in Python using libclang

clang_lint.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
""" C++ linter using libclang. Call with [filenames] """
from __future__ import print_function

from clang.cindex import Config, TypeKind, CursorKind, Index
from pprint import pprint
import platform
import sys

if platform.system() == "Linux":
    Config.set_library_file("/usr/lib/llvm-3.4/lib/libclang.so")
else:
    Config.set_library_file("/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/lib/libclang.dylib")


def error(out_errors, node, msg):
    error_msg = "{}:{}:  {}".format(node.location.file, node.location.line, msg)
    out_errors.append(error_msg)
    print(error_msg, file=sys.stderr)


def is_mut_ref(arg_type):
    if arg_type.kind in [TypeKind.POINTER, TypeKind.LVALUEREFERENCE,
                         TypeKind.INCOMPLETEARRAY, TypeKind.CONSTANTARRAY]:

        if arg_type.kind in [TypeKind.POINTER, TypeKind.LVALUEREFERENCE]:
            pointee_type = arg_type.get_pointee()
        else:
            pointee_type = arg_type.get_array_element_type()

        # print("pointee_type.kind: {}".format(pointee_type.kind))
        # print("pointee_type.is_const_qualified(): {}".format(pointee_type.is_const_qualified()))

        if not pointee_type.is_const_qualified():
            return True

    return False


def check_argument(out_errors, function, node, function_parse_progress):
    assert node.kind == CursorKind.PARM_DECL

    if function.kind == CursorKind.FUNCTION_DECL and function.spelling == "main":
        # Ignore main function
        return

    # print("")
    # print("node.spelling: {}".format(node.spelling))
    # print("node.type.kind: {}".format(node.type.kind))
    # print("node.type.get_ref_qualifier(): {}".format(node.type.get_ref_qualifier()))
    # print("node.type.is_const_qualified(): {}".format(node.type.is_const_qualified()))
    # pprint(dir(node))

    name = node.spelling
    if not name:
        # Ignore nameless arguments (e.g. Foo(Foo&))
        return

    if is_mut_ref(node.type):
        if name.startswith("o_"):
            function_parse_progress["state"] = "out"
        elif name.startswith("io_"):
            if function_parse_progress["state"] != "io":
                error(out_errors, node, "io_ arguments should be first")
            function_parse_progress["state"] = "io"

        else:
            error(out_errors, node,
                  "Non-const reference/pointer/array argument should be prefixed with " \
                  "either o_ (for out) or io_ (for in-out), e.g. 'o_{}'".format(name))
    else:
        if function_parse_progress["state"] == "out":
            error(out_errors, node, "input arguments should come before output arguments")
        function_parse_progress["state"] = "in"


def do_lint(out_errors, node, root_file):
    if node.location.file and node.location.file.name != root_file.name:
        # This is ugly, but works.
        return

    # print("{}:{} node.kind: {}".format(node.location.file, node.location.line, node.kind))
    # print("node.translation_unit: {}".format(node.translation_unit))
    # # pprint(dir(node.translation_unit))
    # print("node.location.file: {}".format(node.location.file))
    # pprint(dir(node.location))
    # exit()

    # CursorKind.CONSTRUCTOR excluded: references there are often stored, so not o_ or io_
    if node.kind in [CursorKind.FUNCTION_DECL, CursorKind.CXX_METHOD]:
        # print("Found a function!")
        # print("node.spelling: {}".format(node.spelling))
        # print("node.displayname: {}".format(node.displayname))

        function_parse_progress = {
            "state": "io", # "io", "in" or "out"
        }

        for arg in node.get_arguments():
            check_argument(out_errors, node, arg, function_parse_progress)

    # Recurse for children of this node
    for c in node.get_children():
        do_lint(out_errors, c, root_file)


def lint_file(filepath):
    index = Index.create()
    tu = index.parse(filepath)
    root_file = tu.get_file(tu.spelling)

    errors = []
    do_lint(errors, tu.cursor, root_file)
    return errors


def main():
    if len(sys.argv) == 1:
        print("Usage: {} [filenames]".format(sys.argv[0]))
        return

    for filepath in sys.argv[1:]:
        lint_file(filepath)


if __name__ == "__main__":
    main()