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, rfc822, datetime
   7 from patterns import patterns
   8 import database
   9
  10
  11 #
  12 # Input file handling.  Someday it would be good to make this smarter
  13 # so that it handles running git with the right options and such.
  14 #
  15 # Someday.
  16 #
  17 SavedLine = ''
  18
  19 def getline(input):
  20     global SavedLine
  21     if SavedLine:
  22         ret = SavedLine
  23         SavedLine = ''
  24         return ret
  25     l = input.readline()
  26     if l:
  27         return l.rstrip()
  28     return None
  29
  30 def SaveLine(line):
  31     global SavedLine
  32     SavedLine = line
  33
  34 #
  35 # A simple state machine based on where we are in the patch.  The
  36 # first stuff we get is the header.
  37 #
  38 S_HEADER = 0
  39 #
  40 # Then comes the single-line description.
  41 #
  42 S_DESC = 1
  43 #
  44 # ...the full changelog...
  45 #
  46 S_CHANGELOG = 2
  47 #
  48 # ...the tag section....
  49 #
  50 S_TAGS = 3
  51 #
  52 # ...the numstat section.
  53 #
  54 S_NUMSTAT = 4
  55
  56 S_DONE = 5
  57
  58 #
  59 # The functions to handle each of these states.
  60 #
  61 def get_header(patch, line, input):
  62     if line == '':
  63         if patch.author == '':
  64             print 'Funky auth line in', patch.commit
  65             patch.author = database.LookupStoreHacker('Unknown',
  66                                                       'unknown@hacker.net')
  67         return S_DESC
  68     m = patterns['author'].match(line)
  69     if m:
  70         patch.email = database.RemapEmail(m.group(2))
  71         patch.author = database.LookupStoreHacker(m.group(1), patch.email)
  72     else:
  73         m = patterns['date'].match(line)
  74         if m:
  75             dt = rfc822.parsedate(m.group(2))
  76             patch.date = datetime.date(dt[0], dt[1], dt[2])
  77     return S_HEADER
  78
  79 def get_desc(patch, line, input):
  80     if not line:
  81         print 'Missing desc in', patch.commit
  82         return S_CHANGELOG
  83     patch.desc = line
  84     line = getline(input)
  85     while line:
  86         patch.desc += line
  87         line = getline(input)
  88     return S_CHANGELOG
  89
  90 tagline = re.compile(r'^\s+(([-a-z]+-by)|cc|fixes):.*$', re.I)
  91 def get_changelog(patch, line, input):
  92     if not line:
  93         if patch.templog:
  94             patch.changelog += patch.templog
  95             patch.templog = ''
  96     if patterns['commit'].match(line):
  97         # No changelog at all - usually a Linus tag
  98         SaveLine(line)
  99         return S_DONE
 100     elif tagline.match(line):
 101         if patch.templog:
 102             patch.changelog += patch.templog
 103         return get_tag(patch, line, input)
 104     else:
 105         patch.templog += line + '\n'
 106     return S_CHANGELOG
 107
 108 def get_tag(patch, line, input):
 109     #
 110     # Some people put blank lines in the middle of tags.
 111     #
 112     if not line:
 113         return S_TAGS
 114     #
 115     # A new commit line says we've gone too far.
 116     #
 117     if patterns['commit'].match(line):
 118         SaveLine(line)
 119         return S_DONE
 120     #
 121     # Check for a numstat line
 122     #
 123     if patterns['numstat'].match(line):
 124         return get_numstat(patch, line, input)
 125     #
 126     # Look for interesting tags
 127     #
 128     m = patterns['signed-off-by'].match(line)
 129     if m:
 130         patch.signoffs.append(m.group(2))
 131     else:
 132         #
 133         # Look for other tags indicating that somebody at least
 134         # looked at the patch.
 135         #
 136         for tag in ('acked-by', 'reviewed-by', 'tested-by'):
 137             if patterns[tag].match(line):
 138                 patch.othertags += 1
 139                 break
 140     patch.taglines.append(line)
 141     return S_TAGS
 142
 143 def get_numstat(patch, line, input):
 144     m = patterns['numstat'].match(line)
 145     if not m:
 146         return S_DONE
 147     try:
 148         patch.addfile(int(m.group(1)), int(m.group(2)), m.group(3))
 149     #
 150     # Binary files just have "-" in the line fields.  In this case, set
 151     # the counts to zero so that we at least track that the file was
 152     # touched.
 153     #
 154     except ValueError:
 155         patch.addfile(0, 0, m.group(3))
 156     return S_NUMSTAT
 157
 158 grabbers = [ get_header, get_desc, get_changelog, get_tag, get_numstat ]
 159
 160
 161 #
 162 # A variant on the gitdm patch class.
 163 #
 164 class patch:
 165     def __init__(self, commit):
 166         self.commit = commit
 167         self.desc = ''
 168         self.changelog = ''
 169         self.templog = ''
 170         self.author = ''
 171         self.signoffs = [ ]
 172         self.othertags = 0
 173         self.added = self.removed = 0
 174         self.files = [ ]
 175         self.taglines = [ ]
 176
 177     def addfile(self, added, removed, file):
 178         self.added += added
 179         self.removed += removed
 180         self.files.append(file)
 181
 182 def grabpatch(input):
 183     #
 184     # If it's not a patch something is screwy.
 185     #
 186     line = getline(input)
 187     if line is None:
 188         return None
 189     m = patterns['commit'].match(line)
 190     if not m:
 191         print 'noncommit', line
 192         return None
 193     p = patch(m.group(1))
 194     state = S_HEADER
 195     #
 196     # Crank through the patch.
 197     #
 198     while state != S_DONE:
 199         line = getline(input)
 200         if line is None:
 201             if state != S_NUMSTAT:
 202                 print 'Ran out of patch', state
 203                 return None
 204             return p
 205         state = grabbers[state](p, line, input)
 206     return p