scripts/commits-to-changelog.py

   1 #!/usr/bin/python
   2
   3 from __future__ import print_function
   4 from optparse import OptionParser
   5 import subprocess
   6 import re
   7 import os.path
   8 import fnmatch
   9 import os
  10 import sys
  11
  12 # subtract 8 for the leading tabstop
  13 fill_column = 74 - 8
  14
  15 path_to_root = None
  16
  17 all_changelogs = {}
  18
  19 def git (command, *args):
  20     popen = subprocess.Popen (["git", command] + list (args), stdout = subprocess.PIPE)
  21     output = popen.communicate () [0]
  22     if popen.returncode != 0:
  23         print ("Error: git failed", file=sys.stderr)
  24         exit (1)
  25     return output
  26
  27 def changelog_path (changelog):
  28     global path_to_root
  29     if not path_to_root:
  30         path_to_root = git ("rev-parse", "--show-cdup").strip ()
  31     (pathname, filename) = changelog
  32     return path_to_root + "./" + pathname + "/" + filename
  33
  34 def changelog_for_file (filename):
  35     while filename != "":
  36         dirname = os.path.dirname (filename)
  37         if dirname in all_changelogs:
  38             return (dirname, all_changelogs [dirname])
  39         filename = dirname
  40     assert False
  41
  42 def changelogs_for_file_pattern (pattern, changed_files):
  43     changelogs = set ()
  44     for filename in changed_files:
  45         suffix = filename
  46         while suffix != "":
  47             # FIXME: fnmatch doesn't support the {x,y} pattern
  48             if fnmatch.fnmatch (suffix, pattern):
  49                 changelogs.add (changelog_for_file (filename))
  50             (_, _, suffix) = suffix.partition ("/")
  51     return changelogs
  52
  53 def format_paragraph (paragraph):
  54     lines = []
  55     words = paragraph.split ()
  56     if len (words) == 0:
  57         return lines
  58     current = words [0]
  59     for word in words [1:]:
  60         if len (current) + 1 + len (word) <= fill_column:
  61             current += " " + word
  62         else:
  63             lines.append ("\t" + current)
  64             current = word
  65     lines.append ("\t" + current)
  66     return lines
  67
  68 def format_changelog_paragraph (files, paragraph):
  69     files_string = ""
  70     for (filename, entity) in files:
  71         if len (files_string) > 0:
  72             files_string += ", "
  73         files_string += filename
  74         if entity:
  75             files_string += " (" + entity + ")"
  76     return format_paragraph ("* " + files_string + ": " + paragraph)
  77
  78 def append_paragraph (lines, paragraph):
  79     if len (lines):
  80         lines.append ("")
  81     lines += paragraph
  82
  83 def format_changelog_entries (commit, changed_files, prefix, file_entries, all_paragraphs):
  84     changelogs = set ()
  85     for f in changed_files:
  86         changelogs.add (changelog_for_file (f))
  87     marked_changelogs = set ()
  88
  89     author_line = git ("log", "-n1", "--date=short", "--format=%ad  %an  <%ae>", commit).strip ()
  90
  91     paragraphs = {}
  92     for changelog in changelogs:
  93         paragraphs [changelog] = [author_line]
  94
  95     for (files, comments) in file_entries:
  96         changelog_entries = {}
  97         for (filename, entity) in files:
  98             entry_changelogs = changelogs_for_file_pattern (filename, changed_files)
  99             if len (entry_changelogs) == 0:
 100                 print ("Warning: could not match file {0} in commit {1}".format (filename, commit))
 101             for changelog in entry_changelogs:
 102                 if changelog not in changelog_entries:
 103                     changelog_entries [changelog] = []
 104                 changelog_entries [changelog].append ((filename, entity))
 105                 marked_changelogs.add (changelog)
 106
 107         for (changelog, files) in changelog_entries.items ():
 108             append_paragraph (paragraphs [changelog], format_changelog_paragraph (files, comments [0]))
 109             for paragraph in comments [1:]:
 110                 append_paragraph (paragraphs [changelog], format_paragraph (paragraph))
 111
 112     unmarked_changelogs = changelogs - marked_changelogs
 113     for changelog in unmarked_changelogs:
 114         if len (prefix) == 0:
 115             print ("Warning: empty entry in {0} for commit {1}".format (changelog_path (changelog), commit))
 116             insert_paragraphs = all_paragraphs
 117         else:
 118             insert_paragraphs = prefix
 119         for paragraph in insert_paragraphs:
 120             append_paragraph (paragraphs [changelog], format_paragraph (paragraph))
 121
 122     return paragraphs
 123
 124 def debug_print_commit (commit, raw_message, prefix, file_entries, changed_files, changelog_entries):
 125     print ("===================== Commit")
 126     print (commit)
 127     print ("--------------------- RAW")
 128     print (raw_message)
 129     print ("--------------------- Prefix")
 130     for line in prefix:
 131         print (line)
 132     print ("--------------------- File entries")
 133     for (files, comments) in file_entries:
 134         files_str = ""
 135         for (filename, entity) in files:
 136             if len (files_str):
 137                 files_str = files_str + ", "
 138             files_str = files_str + filename
 139             if entity:
 140                 files_str = files_str + " (" + entity + ")"
 141         print files_str
 142         for line in comments:
 143             print ("  " + line)
 144     print ("--------------------- Files touched")
 145     for f in changed_files:
 146         print (f)
 147     print ("--------------------- ChangeLog entries")
 148     for ((dirname, filename), lines) in changelog_entries.items ():
 149         print ("{0}/{1}:".format (dirname, filename))
 150         for line in lines:
 151             print (line)
 152
 153 def process_commit (commit):
 154     changed_files = map (lambda l: l.split () [2], git ("diff-tree", "--numstat", commit).splitlines () [1:])
 155     if len (filter (lambda f: re.search ("(^|/)Change[Ll]og$", f), changed_files)):
 156         return None
 157     raw_message = git ("log", "-n1", "--format=%B", commit)
 158     # filter SVN migration message
 159     message = re.sub ("(^|\n)svn path=[^\n]+revision=\d+(?=$|\n)", "", raw_message)
 160     # filter ChangeLog headers
 161     message = re.sub ("(^|\n)\d+-\d+-\d+[ \t]+((\w|[.-])+[ \t]+)+<[^\n>]+>(?=$|\n)", "", message)
 162     # filter leading whitespace
 163     message = re.sub ("^\s+", "", message)
 164     # filter trailing whitespace
 165     message = re.sub ("\s+$", "", message)
 166     # paragraphize - first remove whitespace at beginnings and ends of lines
 167     message = re.sub ("[ \t]*\n[ \t]*", "\n", message)
 168     # paragraphize - now replace three or more consecutive newlines with two
 169     message = re.sub ("\n\n\n+", "\n\n", message)
 170     # paragraphize - replace single newlines with a space
 171     message = re.sub ("(?<!\n)\n(?!\n)", " ", message)
 172     # paragraphize - finally, replace double newlines with single ones
 173     message = re.sub ("\n\n", "\n", message)
 174
 175     # A list of paragraphs (strings without newlines) that occur
 176     # before the first file comments
 177     prefix = []
 178
 179     # A list of tuples of the form ([(filename, entity), ...], [paragraph, ...]).
 180     #
 181     # Each describes a file comment, containing multiple paragraphs.
 182     # Those paragraphs belong to a list of files, each with an
 183     # optional entity (usually a function name).
 184     file_entries = []
 185
 186     current_files = None
 187     current_files_comments = None
 188
 189     message_lines = message.splitlines ()
 190     for line in message_lines:
 191         if re.match ("\*\s[^:]+:", line):
 192             if current_files:
 193                 file_entries.append ((current_files, current_files_comments))
 194
 195             (files, _, comments) = line.partition (":")
 196
 197             current_files_comments = [comments.strip ()]
 198
 199             current_files = []
 200             for f in re.split ("\s*,\s*", files [1:].strip ()):
 201                 m = re.search ("\(([^()]+)\)$", f)
 202                 if m:
 203                     filename = f [:m.start (0)].strip ()
 204                     entity = m.group (1).strip ()
 205                 else:
 206                     filename = f
 207                     entity = None
 208                 current_files.append ((filename, entity))
 209         else:
 210             if current_files:
 211                 current_files_comments.append (line)
 212             else:
 213                 prefix.append (line)
 214     if current_files:
 215         file_entries.append ((current_files, current_files_comments))
 216
 217     changelog_entries = format_changelog_entries (commit, changed_files, prefix, file_entries, message_lines)
 218
 219     #debug_print_commit (commit, raw_message, prefix, file_entries, changed_files, changelog_entries)
 220
 221     return changelog_entries
 222
 223 def start_changelog (changelog):
 224     full_path = changelog_path (changelog)
 225     old_name = full_path + ".old"
 226     os.rename (full_path, old_name)
 227     return open (full_path, "w")
 228
 229 def finish_changelog (changelog, file):
 230     old_file = open (changelog_path (changelog) + ".old")
 231     file.write (old_file.read ())
 232     old_file.close ()
 233     file.close ()
 234
 235 def append_lines (file, lines):
 236     for line in lines:
 237         file.write (line + "\n")
 238     file.write ("\n")
 239
 240 def main ():
 241     usage = "usage: %prog [options] <start-commit>"
 242     parser = OptionParser (usage)
 243     parser.add_option ("-r", "--root", dest = "root", help = "Root directory of the working tree to be changed")
 244     (options, args) = parser.parse_args ()
 245     if len (args) != 1:
 246         parser.error ("incorrect number of arguments")
 247     start_commit = args [0]
 248
 249     if options.root:
 250         global path_to_root
 251         path_to_root = options.root + "/"
 252
 253     # MonkeyWrench uses a shared git repo but sets BUILD_REVISION,
 254     # if present we use it instead of HEAD
 255     HEAD = "HEAD"
 256     if 'BUILD_REVISION' in os.environ:
 257         HEAD = os.environ['BUILD_REVISION']
 258
 259     #see if git supports %B in --format
 260     output = git ("log", "-n1", "--format=%B", HEAD)
 261     if output.startswith ("%B"):
 262         print ("Error: git doesn't support %B in --format - install version 1.7.2 or newer", file=sys.stderr)
 263         exit (1)
 264
 265     for filename in git ("ls-tree", "-r", "--name-only", HEAD).splitlines ():
 266         if re.search ("(^|/)Change[Ll]og$", filename):
 267             (path, name) = os.path.split (filename)
 268             all_changelogs [path] = name
 269
 270     commits = git ("rev-list", "--no-merges", HEAD, "^{0}".format (start_commit)).splitlines ()
 271
 272     touched_changelogs = {}
 273     for commit in commits:
 274         entries = process_commit (commit)
 275         if entries == None:
 276             continue
 277         for (changelog, lines) in entries.items ():
 278             if not os.path.exists (changelog_path (changelog)):
 279                 continue
 280             if changelog not in touched_changelogs:
 281                 touched_changelogs [changelog] = start_changelog (changelog)
 282             append_lines (touched_changelogs [changelog], lines)
 283     for (changelog, file) in touched_changelogs.items ():
 284         finish_changelog (changelog, file)
 285
 286 if __name__ == "__main__":
 287     main ()