tools/generate_authors.py

   1 #!/usr/bin/env python3
   2
   3 #
   4 # Generate the AUTHORS file combining existing AUTHORS file with
   5 # git commit log.
   6 #
   7 # Usage: generate_authors.py AUTHORS.src
   8
   9 # Copyright 2022 Moshe Kaplan
  10 # Based on generate_authors.pl by Michael Mann
  11 #
  12 # Wireshark - Network traffic analyzer
  13 # By Gerald Combs <gerald@wireshark.org>
  14 # Copyright 1998 Gerald Combs
  15 #
  16 # SPDX-License-Identifier: GPL-2.0-or-later
  17
  18 import argparse
  19 import re
  20 import subprocess
  21
  22 have_pyuca = False
  23 try:
  24     from pyuca import Collator
  25     have_pyuca = True
  26 except ModuleNotFoundError:
  27     import sys
  28     sys.stderr.write('pyuca module not found. Sorting names using the built-in locale module.\n')
  29     import locale
  30
  31 def get_git_authors():
  32     '''
  33     Sample line:
  34     #  4321     Navin R. Johnson <nrjohnson@example.com>
  35     '''
  36     GIT_LINE_REGEX = r"^\s*\d+\s+([^<]*)\s*<([^>]*)>"
  37     cmd = "git --no-pager shortlog --email --summary HEAD".split(' ')
  38     # check_output is used for Python 3.4 compatibility
  39     git_cmd_output = subprocess.check_output(cmd, universal_newlines=True, encoding='utf-8')
  40
  41     git_authors = []
  42     for line in git_cmd_output.splitlines():
  43         # Check if this is needed:
  44         line = line.strip()
  45         match = re.match(GIT_LINE_REGEX, line)
  46         name = match.group(1).strip()
  47         email = match.group(2).strip()
  48         # Try to lower how much spam people get:
  49         email = email.replace('@', '[AT]')
  50         git_authors.append((name, email))
  51     if have_pyuca:
  52         c = Collator()
  53         return sorted(git_authors, key=lambda x: c.sort_key(x[0]))
  54     return sorted(git_authors, key=lambda x: locale.strxfrm(x[0].casefold()))
  55
  56
  57 def extract_contributors(authors_content):
  58     # Extract names and email addresses from the AUTHORS file Contributors
  59     contributors_content = authors_content.split("= Contributors =", 1)[1]
  60     CONTRIBUTOR_LINE_REGEX = r"^([\w\.\-\'\x80-\xff]+(\s*[\w+\.\-\'\x80-\xff])*)\s+<([^>]*)>"
  61     contributors = []
  62     state = ""
  63     for line in contributors_content.splitlines():
  64         contributor_match = re.match(CONTRIBUTOR_LINE_REGEX, line)
  65         if re.search(r'([^\{]*)\{', line):
  66             if contributor_match:
  67                 name = contributor_match.group(1)
  68                 email = contributor_match.group(3)
  69                 contributors.append((name, email))
  70             state = "s_in_bracket"
  71         elif state == "s_in_bracket":
  72             if re.search(r'([^\}]*)\}', line):
  73                 state = ""
  74         elif re.search('<', line):
  75             if contributor_match:
  76                 name = contributor_match.group(1)
  77                 email = contributor_match.group(3)
  78                 contributors.append((name, email))
  79         elif re.search(r"(e-mail address removed at contributor's request)", line):
  80             if contributor_match:
  81                 name = contributor_match.group(1)
  82                 email = contributor_match.group(3)
  83                 contributors.append((name, email))
  84         else:
  85             pass
  86     return contributors
  87
  88
  89 def generate_git_contributors_text(contributors_emails, git_authors_emails):
  90     # Track the email addresses seen to avoid including the same email address twice
  91     emails_addresses_seen = set()
  92     for name, email in contributors_emails:
  93         emails_addresses_seen.add(email.lower())
  94
  95     output_lines = []
  96     for name, email in git_authors_emails:
  97         if email.lower() in emails_addresses_seen:
  98             continue
  99
 100         # Skip Gerald, since he's part of the header:
 101         if email == "gerald[AT]wireshark.org":
 102             continue
 103
 104         ntab = 3
 105         if len(name) >= 8*ntab:
 106             line = "{name} <{email}>".format(name=name, email=email)
 107         else:
 108             ntab -= len(name)/8
 109             if len(name) % 8:
 110                 ntab += 1
 111             tabs = '\t'*int(ntab)
 112             line = "{name}{tabs}<{email}>".format(name=name, tabs=tabs, email=email)
 113
 114         emails_addresses_seen.add(email.lower())
 115         output_lines += [line]
 116     return "\n".join(output_lines)
 117
 118
 119 # Read authors file until we find gitlog entries, then stop
 120 def read_authors(parsed_args):
 121     lines = []
 122     with open(parsed_args.authors[0], 'r', encoding='utf-8') as fh:
 123         for line in fh.readlines():
 124             if '= From git log =' in line:
 125                 break
 126             lines.append(line)
 127     return ''.join(lines)
 128
 129
 130 def main():
 131     parser = argparse.ArgumentParser(description="Generate the AUTHORS file combining existing AUTHORS file with git commit log.")
 132     parser.add_argument("authors", metavar='authors', nargs=1, help="path to AUTHORS file")
 133     parsed_args = parser.parse_args()
 134
 135     author_content = read_authors(parsed_args)
 136
 137     # Collect the listed contributors emails so that we don't duplicate them
 138     # in the listing of git contributors
 139     contributors_emails = extract_contributors(author_content)
 140     git_authors_emails = get_git_authors()
 141     # Then generate the text output for git contributors
 142     git_contributors_text = generate_git_contributors_text(contributors_emails, git_authors_emails)
 143
 144     # Now we can write our output:
 145     git_contributor_header = '= From git log =\n\n'
 146     output = author_content + git_contributor_header + git_contributors_text + '\n'
 147
 148     with open(parsed_args.authors[0], 'w', encoding='utf-8') as fh:
 149         fh.write(output)
 150
 151
 152 if __name__ == '__main__':
 153     main()