tools/validate-commit.py

   1 #!/usr/bin/env python3
   2 # Verifies whether commit messages adhere to the standards.
   3 # Checks the author name and email and invokes the tools/commit-msg script.
   4 # Copy this into .git/hooks/post-commit
   5 #
   6 # Copyright (c) 2018 Peter Wu <peter@lekensteyn.nl>
   7 #
   8 # Wireshark - Network traffic analyzer
   9 # By Gerald Combs <gerald@wireshark.org>
  10 # Copyright 1998 Gerald Combs
  11 #
  12 # SPDX-License-Identifier: GPL-2.0-or-later
  13
  14 from __future__ import print_function
  15
  16 import argparse
  17 import difflib
  18 import json
  19 import os
  20 import subprocess
  21 import sys
  22 import urllib.request
  23 import re
  24
  25
  26 parser = argparse.ArgumentParser()
  27 parser.add_argument('commit', nargs='?', default='HEAD',
  28                     help='Commit ID to be checked (default %(default)s)')
  29 parser.add_argument('--commitmsg', help='commit-msg check', action='store')
  30
  31
  32 def print_git_user_instructions():
  33     print('To configure your name and email for git, run:')
  34     print('')
  35     print('  git config --global user.name "Your Name"')
  36     print('  git config --global user.email "you@example.com"')
  37     print('')
  38     print('After that update the author of your latest commit with:')
  39     print('')
  40     print('  git commit --amend --reset-author --no-edit')
  41     print('')
  42
  43
  44 def verify_name(name):
  45     name = name.lower().strip()
  46     forbidden_names = ('unknown', 'root', 'user', 'your name')
  47     if name in forbidden_names:
  48         return False
  49     # Warn about names without spaces. Sometimes it is a mistake where the
  50     # developer accidentally committed using the system username.
  51     if ' ' not in name:
  52         print("WARNING: name '%s' does not contain a space." % (name,))
  53         print_git_user_instructions()
  54     return True
  55
  56
  57 def verify_email(email):
  58     email = email.lower().strip()
  59     try:
  60         user, host = email.split('@')
  61     except ValueError:
  62         # Lacks a '@' (e.g. a plain domain or "foo[AT]example.com")
  63         return False
  64     tld = host.split('.')[-1]
  65
  66     # localhost, localhost.localdomain, my.local etc.
  67     if 'local' in tld:
  68         return False
  69
  70     # Possibly an IP address
  71     if tld.isdigit():
  72         return False
  73
  74     # forbid code.wireshark.org. Submissions could be submitted by other
  75     # addresses if one would like to remain anonymous.
  76     if host.endswith('.wireshark.org'):
  77         return False
  78
  79     # For documentation purposes only.
  80     if host == 'example.com':
  81         return False
  82
  83     # 'peter-ubuntu32.(none)'
  84     if '(none)' in host:
  85         return False
  86
  87     return True
  88
  89
  90 def tools_dir():
  91     if __file__.endswith('.py'):
  92         # Assume direct invocation from tools directory
  93         return os.path.dirname(__file__)
  94     # Otherwise it is a git hook. To support git worktrees, do not manually look
  95     # for the .git directory, but query the actual top level instead.
  96     cmd = ['git', 'rev-parse', '--show-toplevel']
  97     srcdir = subprocess.check_output(cmd, universal_newlines=True).strip()
  98     return os.path.join(srcdir, 'tools')
  99
 100
 101 def extract_subject(subject):
 102     '''Extracts the original subject (ignoring the Revert prefix).'''
 103     subject = subject.rstrip('\r\n')
 104     prefix = 'Revert "'
 105     suffix = '"'
 106     while subject.startswith(prefix) and subject.endswith(suffix):
 107         subject = subject[len(prefix):-len(suffix)]
 108     return subject
 109
 110
 111 def verify_body(body):
 112     bodynocomments = re.sub('^#.*$', '', body, flags=re.MULTILINE)
 113     old_lines = bodynocomments.splitlines(True)
 114     is_good = True
 115     if len(old_lines) >= 2 and old_lines[1].strip():
 116         print('ERROR: missing blank line after the first subject line.')
 117         is_good = False
 118     cleaned_subject = extract_subject(old_lines[0])
 119     if len(cleaned_subject) > 80:
 120         # Note that this check is also invoked by the commit-msg hook.
 121         print('Warning: keep lines in the commit message under 80 characters.')
 122         is_good = False
 123     if not is_good:
 124         print('''
 125 Please rewrite your commit message to our standards, matching this format:
 126
 127     component: a very brief summary of the change
 128
 129     A commit message should start with a brief summary, followed by a single
 130     blank line and an optional longer description. If the change is specific to
 131     a single protocol, start the summary line with the abbreviated name of the
 132     protocol and a colon.
 133
 134     Use paragraphs to improve readability. Limit each line to 80 characters.
 135
 136 ''')
 137     if any(line.startswith('Bug:') or line.startswith('Ping-Bug:') for line in old_lines):
 138         sys.stderr.write('''
 139 To close an issue, use "Closes #1234" or "Fixes #1234" instead of "Bug: 1234".
 140 To reference an issue, use "related to #1234" instead of "Ping-Bug: 1234". See
 141 https://docs.gitlab.com/ee/user/project/issues/managing_issues.html#closing-issues-automatically
 142 for details.
 143 ''')
 144         return False
 145
 146     # Cherry-picking can add an extra newline, which we'll allow.
 147     cp_line = '\n(cherry picked from commit'
 148     body = body.replace('\n' + cp_line, cp_line)
 149
 150     try:
 151         cmd = ['git', 'stripspace']
 152         newbody = subprocess.check_output(cmd, input=body, universal_newlines=True)
 153     except OSError as ex:
 154         print('Warning: unable to invoke git stripspace: %s' % (ex,))
 155         return is_good
 156     if newbody != body:
 157         new_lines = newbody.splitlines(True)
 158         diff = difflib.unified_diff(old_lines, new_lines,
 159                                     fromfile='OLD/.git/COMMIT_EDITMSG',
 160                                     tofile='NEW/.git/COMMIT_EDITMSG')
 161         # Clearly mark trailing whitespace (GNU patch supports such comments).
 162         diff = [
 163             '# NOTE: trailing space on the next line\n%s' % (line,)
 164             if len(line) > 2 and line[-2].isspace() else line
 165             for line in diff
 166         ]
 167         print('The commit message does not follow our standards.')
 168         print('Please rewrite it (there are likely whitespace issues):')
 169         print('')
 170         print(''.join(diff))
 171         return False
 172     return is_good
 173
 174
 175
 176 def verify_merge_request():
 177     # Not needed if/when https://gitlab.com/gitlab-org/gitlab/-/issues/23308 is fixed.
 178     gitlab_api_pfx = "https://gitlab.com/api/v4"
 179     # gitlab.com/wireshark/wireshark = 7898047
 180     project_id = os.getenv('CI_MERGE_REQUEST_PROJECT_ID')
 181     ansi_csi = '\x1b['
 182     ansi_codes = {
 183         'black_white': ansi_csi + '30;47m',
 184         'bold_red': ansi_csi + '31;1m', # gitlab-runner errors
 185         'reset': ansi_csi + '0m'
 186     }
 187     m_r_iid = os.getenv('CI_MERGE_REQUEST_IID')
 188     if project_id is None or m_r_iid is None:
 189         print("This doesn't appear to be a merge request. CI_MERGE_REQUEST_PROJECT_ID={}, CI_MERGE_REQUEST_IID={}".format(project_id, m_r_iid))
 190         return True
 191
 192     m_r_sb_protected = os.getenv('CI_MERGE_REQUEST_SOURCE_BRANCH_PROTECTED')
 193     if m_r_sb_protected == 'true':
 194         print(f'''\
 195 You're pushing from a protected branch ({os.getenv('CI_MERGE_REQUEST_SOURCE_BRANCH_NAME')}). You will probably
 196 have to close this merge request and push from a different branch.\n
 197 ''')
 198         # Assume that the "Allow commits" test is about to fail.
 199
 200     m_r_url = '{}/projects/{}/merge_requests/{}'.format(gitlab_api_pfx, project_id, m_r_iid)
 201     req = urllib.request.Request(m_r_url)
 202     # print('req', repr(req), m_r_url)
 203     with urllib.request.urlopen(req) as resp:
 204         resp_json = resp.read().decode('utf-8')
 205         # print('resp', resp_json)
 206         m_r_attrs = json.loads(resp_json)
 207         try:
 208             if not m_r_attrs['allow_collaboration']:
 209                 print('''\
 210 {bold_red}ERROR:{reset} Please edit your merge request and make sure the setting
 211     {black_white}✅ Allow commits from members who can merge to the target branch{reset}
 212 is checked so that maintainers can rebase your change and make minor edits.\
 213 '''.format(**ansi_codes))
 214                 return False
 215         except KeyError:
 216             sys.stderr.write('This appears to be a merge request, but we were not able to fetch the "Allow commits" status\n')
 217     return True
 218
 219
 220 def main():
 221     args = parser.parse_args()
 222     commit = args.commit
 223
 224     # If called from commit-msg script, just validate that part and return.
 225     if args.commitmsg:
 226         try:
 227             with open(args.commitmsg) as f:
 228                 return 0 if verify_body(f.read()) else 1
 229         except Exception:
 230             print("Couldn't verify body of message from file '", + args.commitmsg + "'")
 231             return 1
 232
 233
 234     if(os.getenv('CI_MERGE_REQUEST_EVENT_TYPE') == 'merge_train'):
 235         print("If we were on the love train, people all over the world would be joining hands for this merge request.\nInstead, we're on a merge train so we're skipping commit validation checks. ")
 236         return 0
 237
 238     cmd = ['git', 'show', '--no-patch',
 239            '--format=%h%n%an%n%ae%n%B', commit, '--']
 240     output = subprocess.check_output(cmd, universal_newlines=True)
 241     # For some reason there is always an additional LF in the output, drop it.
 242     if output.endswith('\n\n'):
 243         output = output[:-1]
 244     abbrev, author_name, author_email, body = output.split('\n', 3)
 245     subject = body.split('\n', 1)[0]
 246
 247     # If called directly (from the tools directory), print the commit that was
 248     # being validated. If called from a git hook (without .py extension), try to
 249     # remain silent unless there are issues.
 250     if __file__.endswith('.py'):
 251         print('Checking commit: %s %s' % (abbrev, subject))
 252
 253     exit_code = 0
 254     if not verify_name(author_name):
 255         print('Disallowed author name: {}'.format(author_name))
 256         exit_code = 1
 257
 258     if not verify_email(author_email):
 259         print('Disallowed author email address: {}'.format(author_email))
 260         exit_code = 1
 261
 262     if exit_code:
 263         print_git_user_instructions()
 264
 265     if not verify_body(body):
 266         exit_code = 1
 267
 268     if not verify_merge_request():
 269         exit_code = 1
 270
 271     return exit_code
 272
 273
 274 if __name__ == '__main__':
 275     try:
 276         sys.exit(main())
 277     except subprocess.CalledProcessError as ex:
 278         print('\n%s' % ex)
 279         sys.exit(ex.returncode)
 280     except KeyboardInterrupt:
 281         sys.exit(130)