gitlog.py

   1 #
   2 # Stuff for dealing with the git log output.
   3 #
   4 # Someday this will be the only version of grabpatch, honest.
   5 #
   6 import re, datetime
   7 from email.utils import parsedate
   8 from patterns import patterns
   9 import database
  10
  11
  12 #
  13 # Input file handling.  Someday it would be good to make this smarter
  14 # so that it handles running git with the right options and such.
  15 #
  16 # Someday.
  17 #
  18 SavedLine = ''
  19
  20 def getline(input):
  21     global SavedLine
  22     if SavedLine:
  23         ret = SavedLine
  24         SavedLine = ''
  25         return ret
  26     l = input.readline()
  27     if not l:
  28         return None
  29     #
  30     # In theory everything coming out of git is utf8.  In practice,
  31     # there's some funky stuff in the kernel repo.  So...fall back
  32     # to latin1 if a utf8 decode fails; that doesn't work for everything
  33     # but it's about as good as we can do.
  34     #
  35     try:
  36         l = l.decode('utf8')
  37     except UnicodeDecodeError:
  38         l = l.decode('latin1')
  39     return l.rstrip()
  40
  41 def SaveLine(line):
  42     global SavedLine
  43     SavedLine = line
  44
  45 #
  46 # A simple state machine based on where we are in the patch.  The
  47 # first stuff we get is the header.
  48 #
  49 S_HEADER = 0
  50 #
  51 # Then comes the single-line description.
  52 #
  53 S_DESC = 1
  54 #
  55 # ...the full changelog...
  56 #
  57 S_CHANGELOG = 2
  58 #
  59 # ...the tag section....
  60 #
  61 S_TAGS = 3
  62 #
  63 # ...the numstat section.
  64 #
  65 S_NUMSTAT = 4
  66
  67 S_DONE = 5
  68
  69 #
  70 # The functions to handle each of these states.
  71 #
  72 def get_header(patch, line, input):
  73     if line == '':
  74         if patch.author == '':
  75             print('Funky auth line in', patch.commit)
  76             patch.author = database.LookupStoreHacker('Unknown',
  77                                                       'unknown@hacker.net')
  78         return S_DESC
  79     m = patterns['author'].match(line)
  80     if m:
  81         patch.email = database.RemapEmail(m.group(2))
  82         patch.author = database.LookupStoreHacker(m.group(1), patch.email)
  83     else:
  84         m = patterns['date'].match(line)
  85         if m:
  86             dt = parsedate(m.group(2))
  87             patch.date = datetime.date(dt[0], dt[1], dt[2])
  88     return S_HEADER
  89
  90 def get_desc(patch, line, input):
  91     if not line:
  92         print('Missing desc in', patch.commit)
  93         return S_CHANGELOG
  94     patch.desc = line
  95     line = getline(input)
  96     while line:
  97         patch.desc += line
  98         line = getline(input)
  99     return S_CHANGELOG
 100
 101 tagline = re.compile(r'^\s+(([-a-z]+-by)|cc|fixes):.*$', re.I)
 102 def get_changelog(patch, line, input):
 103     if not line:
 104         if patch.templog:
 105             patch.changelog += patch.templog
 106             patch.templog = ''
 107     if patterns['commit'].match(line):
 108         # No changelog at all - usually a Linus tag
 109         SaveLine(line)
 110         return S_DONE
 111     elif tagline.match(line):
 112         if patch.templog:
 113             patch.changelog += patch.templog
 114         return get_tag(patch, line, input)
 115     else:
 116         patch.templog += line + '\n'
 117     return S_CHANGELOG
 118
 119 def get_tag(patch, line, input):
 120     #
 121     # Some people put blank lines in the middle of tags.
 122     #
 123     if not line:
 124         return S_TAGS
 125     #
 126     # A new commit line says we've gone too far.
 127     #
 128     if patterns['commit'].match(line):
 129         SaveLine(line)
 130         return S_DONE
 131     #
 132     # Check for a numstat line
 133     #
 134     if patterns['numstat'].match(line):
 135         return get_numstat(patch, line, input)
 136     #
 137     # Look for interesting tags
 138     #
 139     m = patterns['signed-off-by'].match(line)
 140     if m:
 141         patch.signoffs.append(m.group(2))
 142     else:
 143         #
 144         # Look for other tags indicating that somebody at least
 145         # looked at the patch.
 146         #
 147         for tag in ('acked-by', 'reviewed-by', 'tested-by'):
 148             if patterns[tag].match(line):
 149                 patch.othertags += 1
 150                 break
 151     patch.taglines.append(line)
 152     return S_TAGS
 153
 154 def get_numstat(patch, line, input):
 155     m = patterns['numstat'].match(line)
 156     if not m:
 157         return S_DONE
 158     try:
 159         patch.addfile(int(m.group(1)), int(m.group(2)), m.group(3))
 160     #
 161     # Binary files just have "-" in the line fields.  In this case, set
 162     # the counts to zero so that we at least track that the file was
 163     # touched.
 164     #
 165     except ValueError:
 166         patch.addfile(0, 0, m.group(3))
 167     return S_NUMSTAT
 168
 169 grabbers = [ get_header, get_desc, get_changelog, get_tag, get_numstat ]
 170
 171
 172 #
 173 # A variant on the gitdm patch class.
 174 #
 175 class patch:
 176     def __init__(self, commit):
 177         self.commit = commit
 178         self.desc = ''
 179         self.changelog = ''
 180         self.templog = ''
 181         self.author = ''
 182         self.signoffs = [ ]
 183         self.othertags = 0
 184         self.added = self.removed = 0
 185         self.files = [ ]
 186         self.taglines = [ ]
 187
 188     def addfile(self, added, removed, file):
 189         self.added += added
 190         self.removed += removed
 191         self.files.append(file)
 192
 193 tag = re.compile('\(tag: (v[.\d]+)\)')
 194 def grabpatch(input):
 195     #
 196     # If it's not a patch something is screwy.
 197     #
 198     line = getline(input)
 199     if line is None:
 200         return None
 201     m = patterns['commit'].match(line)
 202     if not m:
 203         print('noncommit', line)
 204         return None
 205     p = patch(m.group(1))
 206     state = S_HEADER
 207     #
 208     # Look for a tag on this line.  BUG fails with two tags
 209     #
 210     m = tag.search(line)
 211     if m:
 212         patch.tag = m.group(1)
 213     else:
 214         patch.tag = None
 215     #
 216     # Crank through the patch.
 217     #
 218     while state != S_DONE:
 219         line = getline(input)
 220         if line is None:
 221             if state != S_NUMSTAT:
 222                 print('Ran out of patch', state)
 223                 return None
 224             return p
 225         state = grabbers[state](p, line, input)
 226     return p