├── README.md
└── git-score


/README.md:
--------------------------------------------------------------------------------
 1 | # git-score
 2 | 
 3 | *git-score* is a script that computes 'scores' for authors who have contributed
 4 | to a git repository. It displays aggregate stats in a concise format for human
 5 | consumption and provides a rough estimation of developer involvement in a
 6 | project.
 7 | 
 8 | ![git-score screenshot](http://farm8.staticflickr.com/7180/6900456317_0d237aba81.jpg)
 9 | 
10 | ## Installation
11 | 
12 | Copy *git-score* to somewhere in your `PATH`. A good solution is to place a
13 | git-score symlink in *~/bin* that points to your git-score repository.
14 | 
15 | ## Usage
16 | 
17 | Inside a git repository, just run
18 | 
19 |     % git-score
20 | 
21 | It takes the following arguments directly:
22 | 
23 |     -e, --exclude=<pattern>
24 |       Tallies scores per-file, and if <pattern> matches the filename.
25 |       This is useful to exclude directories or particular file extensions.
26 |       Warning: may run much slower. Still untested on very large repos.
27 | 
28 |     -i, --include=<pattern>
29 |       Same as exclude, but the reverse (<pattern> must match).
30 |       Include and exclude can be used in conjunction.
31 | 
32 |     -o, --order=[+-]<field>
33 |       Sorts by chosen <field>. Prepend [+,-] for [ascending, descending]
34 |       <field> must be one of: author, files, commits, delta, added, removed
35 |       All fields (except author) default to descending.
36 | 
37 |     -r, --repo=<path>
38 |       Specify a path to the git repository.  Can be specified multiple times to
39 |       aggregate stats across multiple repositories.
40 | 
41 |     -h, --help
42 |       Shows available options.
43 | 
44 | All other arguments not beginning with a hypen ('-') will be passed to `git
45 | log`. This way, you can specify revision ranges to narrow scoring to a
46 | particular set of commits.
47 | 
48 |     % git-score 61bf126e..HEAD
49 | 
50 | Note: The hyphen ('-') is optional: `git score` works too on recent versions of
51 | git.
52 | 
53 | ## Examples
54 | 
55 | Note: Emails obscured here.
56 | 
57 |     % git score
58 |     author                          commits files  delta  (+)    (-)
59 |     Matt Sparks <ms@------------->  13      2      143    184    41
60 |     jbenet <jbenet@---------------> 8       2      76     110    34
61 | 
62 |     % git score b210b0..HEAD
63 |     author                          commits files  delta  (+)    (-)
64 |     jbenet <jbenet@---------------> 7       2      46     77     31
65 |     Matt Sparks <ms@------------->  6       2      14     31     17
66 | 
67 |     % git score --include=.*\.[hm]$
68 |     author                          commits files  delta  (+)    (-)
69 |     jbenet <jbenet@---------------> 267     363    24664  47848  23184
70 |     chris <------------------->     46      126    12908  17156  4248
71 |     Rico <----------------------->  39      119    8659   10880  2221
72 | 
73 |     % git score --include=.*/src/.*$ --exclude=.*\.m$
74 |     author                          commits files  delta  (+)    (-)
75 |     jbenet <jbenet@---------------> 267     124    1884   3568   1684
76 |     Rico <----------------------->  39      46     738    918    180
77 |     chris <------------------->     46      44     696    936    240
78 | 
79 |     % git score --order=author
80 |     author                          commits files  delta  (+)    (-)
81 |     jbenet <jbenet@---------------> 12      2      136    183    47
82 |     Matt Sparks <ms@------------->  13      2      143    184    41
83 | 
84 |     % git score --order=commits
85 |     author                          commits files  delta  (+)    (-)
86 |     Matt Sparks <ms@------------->  13      2      143    184    41
87 |     jbenet <jbenet@---------------> 12      2      136    183    47
88 | 
89 |     % git score --order=-delta
90 |     author                          commits files  delta  (+)    (-)
91 |     jbenet <jbenet@---------------> 13      2      151    198    47
92 |     Matt Sparks <ms@------------->  13      2      143    184    41
93 | 
94 |     % git score --order=+added
95 |     author                          commits files  delta  (+)    (-)
96 |     Matt Sparks <ms@------------->  13      2      143    184    41
97 |     jbenet <jbenet@---------------> 13      2      151    198    47
98 | 


--------------------------------------------------------------------------------
/git-score:
--------------------------------------------------------------------------------
  1 | #!/usr/bin/env python2
  2 | #
  3 | # git-score -- aggregate git commit statistics
  4 | #
  5 | # Author: Juan Batiz-Benet
  6 | # Author: Matt Sparks
  7 | import operator
  8 | import optparse
  9 | import os
 10 | import re
 11 | import subprocess
 12 | import sys
 13 | 
 14 | 
 15 | STAT_PATTERN = (r'^\s*(\d+) files changed, (\d+) insertions\(\+\), '
 16 |                 r'(\d+) deletions\(\-\)$')
 17 | 
 18 | FILE_PATTERN =  r'^\s*(\d+)\t(\d+)\t(.*)$'
 19 | 
 20 | # Truncate long author names to this size.
 21 | MAX_NAME_LENGTH = 45
 22 | 
 23 | 
 24 | class CommitStat(object):
 25 |   '''Represents the statistics of one commit.'''
 26 |   def __init__(self, author, added, removed, files):
 27 |     self._author = author
 28 |     self._added = added
 29 |     self._removed = removed
 30 |     self._files = files
 31 | 
 32 |   def __str__(self):
 33 |     '''Return string representation.'''
 34 |     values = {'author': self._author,
 35 |               'added': self._added,
 36 |               'removed': self._removed,
 37 |               'delta': self.delta(),
 38 |               'files': str(list(self._files))}
 39 |     return '%(author)s: d%(delta)d, +%(added)i, -%(removed)d %(files)s' % values
 40 | 
 41 |   def author(self):
 42 |     '''Returns author name.'''
 43 |     return self._author
 44 | 
 45 |   def added(self):
 46 |     '''Returns number of lines added in this commit.'''
 47 |     return self._added
 48 | 
 49 |   def removed(self):
 50 |     '''Returns number of lines removed in this commit.'''
 51 |     return self._removed
 52 | 
 53 |   def delta(self):
 54 |     '''Returns delta of lines added and removed in this commit.'''
 55 |     return self._added - self._removed
 56 | 
 57 |   def numFilesChanged(self):
 58 |     '''Returns the number of files changed.'''
 59 |     return len(self._files)
 60 | 
 61 | 
 62 | class CommitStatList(object):
 63 |   '''An aggregated group of CommitStat.'''
 64 |   def __init__(self, name=""):
 65 |     self._name = name
 66 |     self._added = 0
 67 |     self._removed = 0
 68 |     self._commits = list()
 69 |     self._files = set()
 70 | 
 71 |   def name(self):
 72 |     '''Returns the name of this CommitStatList.'''
 73 |     return self._name
 74 | 
 75 |   def added(self):
 76 |     '''Returns number of lines added over all commits.'''
 77 |     return self._added
 78 | 
 79 |   def removed(self):
 80 |     '''Returns number of lines removed over all commits.'''
 81 |     return self._removed
 82 | 
 83 |   def delta(self):
 84 |     '''Returns delta of lines added and removed.'''
 85 |     return self._added - self._removed
 86 | 
 87 |   def numFilesChanged(self):
 88 |     '''Returns the number of files changed'''
 89 |     return len(self._files)
 90 | 
 91 |   def commits(self):
 92 |     '''Returns number of total commits.'''
 93 |     return len(self._commits)
 94 | 
 95 |   def add(self, commitstat):
 96 |     '''Adds a CommitStat, aggregating counts.'''
 97 |     if not isinstance(commitstat, CommitStat):
 98 |       raise TypeError('commitstat is not of type CommitStat')
 99 | 
100 |     self._added += commitstat._added
101 |     self._removed += commitstat._removed
102 |     self._files.update(commitstat._files)
103 |     self._commits.append(commitstat)
104 | 
105 |   def __add__(self, other):
106 |     '''Adds this and another CommitStatList, creating a new CommitStatList.'''
107 |     csl = CommitStatList(self._name)
108 |     csl._added = self._added + other._added
109 |     csl._removed = self._removed + other._removed
110 |     csl._files = self._files.union(other._files)
111 |     csl._commits = self._commits + other._commits
112 |     return csl
113 | 
114 | 
115 | class AuthorMap(object):
116 |   '''The representation of a collection of authors and their CommitStats.'''
117 |   def __init__(self):
118 |     self._stats = {}
119 |     self._orderfield = 'delta'  # default sorting field
120 |     self._ascending = False  # default sorting direction
121 | 
122 |   def __getitem__(self, author):
123 |     '''Returns the CommitStatList for a particular author.'''
124 |     if author not in self._stats:
125 |       self._stats[author] = CommitStatList(author)
126 |     return self._stats[author]
127 | 
128 |   def __iter__(self):
129 |     '''Returns iterator for the sorted keys of the author map.'''
130 |     def statkey(stat):
131 |       '''Returns the value to sort a CommitStatList by.'''
132 |       if self._orderfield == 'author':
133 |         return stat.name().lower()  # named the commit lists by author name
134 |       if self._orderfield == 'files':
135 |         return stat.numFilesChanged()
136 |       return getattr(stat, self._orderfield)()
137 | 
138 |     stats = self._stats.values()
139 |     stats.sort(key=statkey)
140 |     stats = map(lambda statset: statset.name(), stats)  # want keys, not values
141 | 
142 |     if not self._ascending:
143 |       stats.reverse()
144 |     return iter(stats)
145 | 
146 |   def __len__(self):
147 |     return len(self._stats.keys())
148 | 
149 |   def __add__(self, other):
150 |     '''Returns a new AuthorMap representing the sum this and other stats.'''
151 |     am = AuthorMap()
152 |     for author in set(self._stats.keys() + other._stats.keys()):
153 |       am._stats[author] = self[author] + other[author]
154 |     return am
155 | 
156 |   def __str__(self):
157 |     return 'AuthorMap(stats=%s, orderfield=%s, ascending=%s)' % (self._stats,
158 |         self._orderfield, self._ascending)
159 | 
160 |   def order(self, order):
161 |     '''Defines the sort order for this AuthorMap.
162 | 
163 |     Args:
164 |       order: the direction and field to sort by. Accepted orders:
165 |              [+-]?(author|files|commits|added|removed|delta])
166 |     '''
167 |     if order[0] in ('-', '+'):  # direction was specified
168 |       self._orderfield = order[1:]
169 |       self._ascending = order[0] == '+'
170 |     else:
171 |       self._orderfield = order
172 |       self._ascending = order == 'author'  # descending default, except authors
173 | 
174 | 
175 | class GitColor(object):
176 |   '''Gets and uses current git settings to color output.'''
177 |   _colorIsOn = None
178 | 
179 |   @classmethod
180 |   def gitColorIsOn(cls):
181 |     '''Returns whether git config color.ui is on. Only queries once.'''
182 |     if cls._colorIsOn is not None:
183 |       return cls._colorIsOn
184 | 
185 |     cmd = ['git', 'config', 'color.ui']
186 |     proc = subprocess.Popen(cmd, stdout=subprocess.PIPE)
187 |     next = proc.stdout.readline().strip()
188 |     next = 'auto' if next == '' else next
189 |     cls._colorIsOn = next in ['true', 'always', 'auto']
190 |     return cls._colorIsOn
191 | 
192 |   @classmethod
193 |   def red(cls, text):
194 |     '''Returns text, in red if color is on.'''
195 |     if cls.gitColorIsOn():
196 |       return "\033[31m%s\033[0m" % text
197 |     return text
198 | 
199 |   @classmethod
200 |   def green(cls, text):
201 |     '''Returns text, in green if color is on.'''
202 |     if cls.gitColorIsOn():
203 |       return '\033[32m%s\033[0m' % text
204 |     return text
205 | 
206 | 
207 | def parseGitLogData(stream, options, path):
208 |   class ParseCommitData(object):
209 |     '''A utility class to keep track of commit stats currently being parsed.'''
210 |     author = None
211 |     added = 0
212 |     removed = 0
213 |     files = None
214 | 
215 |     @classmethod
216 |     def reset(cls, author):
217 |       cls.author = author
218 |       cls.added = 0
219 |       cls.removed = 0
220 |       cls.files = set()
221 | 
222 |     @classmethod
223 |     def commitStat(cls):
224 |       return CommitStat(cls.author, cls.added, cls.removed, cls.files)
225 | 
226 |   stat_pattern = re.compile(STAT_PATTERN)
227 |   file_pattern = re.compile(FILE_PATTERN)
228 |   exclude_pattern = re.compile(options.exclude) if options.exclude else None
229 |   include_pattern = re.compile(options.include) if options.include else None
230 | 
231 |   authorstats = AuthorMap()
232 | 
233 |   for line in stream:
234 |     line = line.strip()
235 |     if line == '':
236 |       continue  # skip blank lines
237 | 
238 |     # Author line, aggregate last commit (if any) and reset counts
239 |     if line.startswith('+'):
240 |       if ParseCommitData.author != None and len(ParseCommitData.files) > 0:
241 |         authorstats[ParseCommitData.author].add(ParseCommitData.commitStat())
242 |       ParseCommitData.reset(line[1:])  # remove leading '+'
243 | 
244 |     # File line, record per-file changes.
245 |     else:
246 |       match = file_pattern.match(line)
247 |       if not match:
248 |         continue
249 | 
250 |       added, removed, filename = match.group(1, 2, 3)
251 |       exclude_file = options.exclude and exclude_pattern.match(filename)
252 |       include_file = not options.include or include_pattern.match(filename)
253 |       if include_file and not exclude_file:
254 |         ParseCommitData.added += int(added)
255 |         ParseCommitData.removed += int(removed)
256 |         if path:
257 |           filename = os.path.join(path, filename)
258 |         ParseCommitData.files.add(filename)
259 | 
260 |   # Process last commit.
261 |   if ParseCommitData.author != None and len(ParseCommitData.files) > 0:
262 |     authorstats[ParseCommitData.author].add(ParseCommitData.commitStat())
263 | 
264 |   return authorstats
265 | 
266 | 
267 | def gitStats(log_args, options, path=None):
268 |   if path:
269 |     old_path = os.getcwd()
270 |     os.chdir(path)
271 |   cmd = ['git', 'log', '--numstat', '--pretty=format:+%an <%ae>']
272 |   cmd.extend(log_args)
273 |   proc = subprocess.Popen(cmd, stdout=subprocess.PIPE)
274 |   if path:
275 |     os.chdir(old_path)
276 |   return parseGitLogData(proc.stdout, options, path)
277 | 
278 | 
279 | def gitStatsAggregated(log_args, options):
280 |   if options.repos:
281 |     return reduce(operator.add,
282 |         [gitStats(log_args, options, r) for r in options.repos])
283 |   else:
284 |     return gitStats(log_args, options)
285 | 
286 | 
287 | def parseArgs():
288 |   parser = optparse.OptionParser()
289 |   parser.add_option('-e', '--exclude', dest='exclude', metavar='PATTERN',
290 |                     help='exclude files matching PATTERN')
291 |   parser.add_option('-i', '--include', dest='include', metavar='PATTERN',
292 |                     help='include files matching PATTERN')
293 |   parser.add_option('-o', '--order', dest='order', metavar='ORDER',
294 |                     help='sorting order (e.g., +delta)')
295 |   parser.add_option('-r', '--repo', dest='repos', metavar='REPO',
296 |                     action='append',
297 |                     help='repository path, can be specified multiple times')
298 |   options, args = parser.parse_args()
299 | 
300 |   # Validate sorting field.
301 |   valid_fields = ('author', 'files', 'commits', 'delta', 'added', 'removed')
302 |   if options.order:
303 |     has_prefix = options.order.startswith('+') or options.order.startswith('-')
304 |     if (len(options.order) < 2 or
305 |         (has_prefix and options.order[1:] not in valid_fields) or
306 |         (not has_prefix and options.order not in valid_fields)):
307 |       parser.error('invalid sorting order: %s' % options.order)
308 | 
309 |   return options, args
310 | 
311 | 
312 | def main():
313 |   options, log_args = parseArgs()
314 |   stats = gitStatsAggregated(log_args, options)
315 |   if options.order:
316 |     stats.order(options.order)
317 | 
318 |   if len(stats) == 0:
319 |     print 'no commits'
320 |     sys.exit(0)
321 | 
322 |   # Find length of longest author name.
323 |   author_length = max([len(a) for a in stats])
324 |   author_length = min(author_length, MAX_NAME_LENGTH)
325 | 
326 |   print 'author%s\tcommits\tfiles\tdelta\t(+)\t(-)' % (' ' * (author_length-6))
327 |   for author in stats:
328 |     name_field = author[0:author_length]
329 |     name_field = '%s%s' % (name_field, ' ' * (author_length - len(name_field)))
330 |     values = {'name': name_field,
331 |               'commits': stats[author].commits(),
332 |               'files': stats[author].numFilesChanged(),
333 |               'delta': stats[author].delta(),
334 |               'added': GitColor.green(stats[author].added()),
335 |               'removed': GitColor.red(stats[author].removed()) }
336 | 
337 |     print '%(name)s\t%(commits)d\t%(files)d' % values,
338 |     print '\t%(delta)d\t%(added)s\t%(removed)s' % values
339 | 
340 | 
341 | if __name__ == '__main__':
342 |   main()
343 | 


--------------------------------------------------------------------------------