scripts/commits-to-changelog.py

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