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