1#!/usr/bin/env python
2# SPDX-License-Identifier: GPL-2.0
3#
4# Copyright (C) Google LLC, 2018
5#
6# Author: Tom Roeder <tmroeder@google.com>
7#
8"""A tool for generating compile_commands.json in the Linux kernel."""
9
10import argparse
11import json
12import logging
13import os
14import re
15
16_DEFAULT_OUTPUT = 'compile_commands.json'
17_DEFAULT_LOG_LEVEL = 'WARNING'
18
19_FILENAME_PATTERN = r'^\..*\.cmd$'
20_LINE_PATTERN = r'^cmd_[^ ]*\.o := (.* )([^ ]*\.c)$'
21_VALID_LOG_LEVELS = ['DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL']
22
23# A kernel build generally has over 2000 entries in its compile_commands.json
24# database. If this code finds 300 or fewer, then warn the user that they might
25# not have all the .cmd files, and they might need to compile the kernel.
26_LOW_COUNT_THRESHOLD = 300
27
28
29def parse_arguments():
30    """Sets up and parses command-line arguments.
31
32    Returns:
33        log_level: A logging level to filter log output.
34        directory: The directory to search for .cmd files.
35        output: Where to write the compile-commands JSON file.
36    """
37    usage = 'Creates a compile_commands.json database from kernel .cmd files'
38    parser = argparse.ArgumentParser(description=usage)
39
40    directory_help = ('Path to the kernel source directory to search '
41                      '(defaults to the working directory)')
42    parser.add_argument('-d', '--directory', type=str, help=directory_help)
43
44    output_help = ('The location to write compile_commands.json (defaults to '
45                   'compile_commands.json in the search directory)')
46    parser.add_argument('-o', '--output', type=str, help=output_help)
47
48    log_level_help = ('The level of log messages to produce (one of ' +
49                      ', '.join(_VALID_LOG_LEVELS) + '; defaults to ' +
50                      _DEFAULT_LOG_LEVEL + ')')
51    parser.add_argument(
52        '--log_level', type=str, default=_DEFAULT_LOG_LEVEL,
53        help=log_level_help)
54
55    args = parser.parse_args()
56
57    log_level = args.log_level
58    if log_level not in _VALID_LOG_LEVELS:
59        raise ValueError('%s is not a valid log level' % log_level)
60
61    directory = args.directory or os.getcwd()
62    output = args.output or os.path.join(directory, _DEFAULT_OUTPUT)
63    directory = os.path.abspath(directory)
64
65    return log_level, directory, output
66
67
68def process_line(root_directory, file_directory, command_prefix, relative_path):
69    """Extracts information from a .cmd line and creates an entry from it.
70
71    Args:
72        root_directory: The directory that was searched for .cmd files. Usually
73            used directly in the "directory" entry in compile_commands.json.
74        file_directory: The path to the directory the .cmd file was found in.
75        command_prefix: The extracted command line, up to the last element.
76        relative_path: The .c file from the end of the extracted command.
77            Usually relative to root_directory, but sometimes relative to
78            file_directory and sometimes neither.
79
80    Returns:
81        An entry to append to compile_commands.
82
83    Raises:
84        ValueError: Could not find the extracted file based on relative_path and
85            root_directory or file_directory.
86    """
87    # The .cmd files are intended to be included directly by Make, so they
88    # escape the pound sign '#', either as '\#' or '$(pound)' (depending on the
89    # kernel version). The compile_commands.json file is not interepreted
90    # by Make, so this code replaces the escaped version with '#'.
91    prefix = command_prefix.replace('\#', '#').replace('$(pound)', '#')
92
93    cur_dir = root_directory
94    expected_path = os.path.join(cur_dir, relative_path)
95    if not os.path.exists(expected_path):
96        # Try using file_directory instead. Some of the tools have a different
97        # style of .cmd file than the kernel.
98        cur_dir = file_directory
99        expected_path = os.path.join(cur_dir, relative_path)
100        if not os.path.exists(expected_path):
101            raise ValueError('File %s not in %s or %s' %
102                             (relative_path, root_directory, file_directory))
103    return {
104        'directory': cur_dir,
105        'file': relative_path,
106        'command': prefix + relative_path,
107    }
108
109
110def main():
111    """Walks through the directory and finds and parses .cmd files."""
112    log_level, directory, output = parse_arguments()
113
114    level = getattr(logging, log_level)
115    logging.basicConfig(format='%(levelname)s: %(message)s', level=level)
116
117    filename_matcher = re.compile(_FILENAME_PATTERN)
118    line_matcher = re.compile(_LINE_PATTERN)
119
120    compile_commands = []
121    for dirpath, _, filenames in os.walk(directory):
122        for filename in filenames:
123            if not filename_matcher.match(filename):
124                continue
125            filepath = os.path.join(dirpath, filename)
126
127            with open(filepath, 'rt') as f:
128                for line in f:
129                    result = line_matcher.match(line)
130                    if not result:
131                        continue
132
133                    try:
134                        entry = process_line(directory, dirpath,
135                                             result.group(1), result.group(2))
136                        compile_commands.append(entry)
137                    except ValueError as err:
138                        logging.info('Could not add line from %s: %s',
139                                     filepath, err)
140
141    with open(output, 'wt') as f:
142        json.dump(compile_commands, f, indent=2, sort_keys=True)
143
144    count = len(compile_commands)
145    if count < _LOW_COUNT_THRESHOLD:
146        logging.warning(
147            'Found %s entries. Have you compiled the kernel?', count)
148
149
150if __name__ == '__main__':
151    main()
152