src/git_stats/author.py

   1 #!/usr/bin/env python
   2
   3 import os
   4 import sys
   5
   6 from optparse import OptionParser
   7 from git_python import Git
   8
   9 from git_stats import commit
  10
  11 class Activity:
  12   """Simple storage class containing stats on the activity in one file."""
  13
  14   def __init__(self):
  15     self.count = 1
  16     self.added = 0
  17     self.deleted = 0
  18     self.id = []
  19
  20   def __str__(self):
  21     return str(self.count) + ": +" + str(self.added) + " -" + str(self.deleted)
  22
  23 def activityInArea(log, filterid):
  24   """Parses the specified file containing commit logs.
  25   The output is filtered to contain only the results of the specified author.
  26   The output is expected to be in the format described below:
  27   [
  28     [<id>\n]+
  29     [<lines added>\t<lines deleted>\t<path>]+
  30     \n
  31   ]+
  32
  33   Params:
  34     log: The log formatted as described above.
  35     filterid: The id address to filter on.
  36
  37   Returns:
  38     A dictionary containing activity per path is returned.
  39     Each path contains one Activity object with the aggregated results.
  40   """
  41
  42   # Create a dict to store the activity stats per path
  43   activityByPath = {}
  44
  45   # Create a place to store the result in
  46   activity = Activity()
  47
  48   i = 0
  49
  50   # Parse all the lines in the file
  51   for line in log:
  52     i += 1
  53
  54     # Split the line at the tab and store the data
  55     splitline = line.split('\t')
  56     size = len(splitline)
  57     length = len(line.lstrip())
  58
  59     # There is something on this line, but it contains no separator
  60     if size == 1 and length > 0:
  61       # Get the id address minus the newline
  62       activity.id.append(line[:-1])
  63     elif size == 3:
  64       try:
  65         addpart = splitline[0]
  66         deletepart = splitline[1]
  67
  68         if addpart == '-':
  69           addpart = 0
  70
  71         if deletepart == '-':
  72           deletepart = 0
  73
  74         activity.added = int(addpart)
  75         activity.deleted = int(deletepart)
  76       except ValueError, e:
  77         print("On line '" + str(i) + "', could not convert number: " + str(e))
  78
  79       path = splitline[2][:-1]
  80     elif length == 0:
  81       if not filterid in activity.id:
  82         # Ignore other authors
  83         pass
  84       else:
  85         result = Activity()
  86
  87         # If we already encountered this path, update the count
  88         if activityByPath.has_key(path):
  89           known = activityByPath[path]
  90           result.added = known.added + activity.added
  91           result.deleted = known.deleted + activity.deleted
  92           result.count = known.count + activity.count
  93         else:
  94           result = activity
  95
  96         # Store it under it's path
  97         activityByPath[path] = result
  98
  99         # Create a fresh activity to store the next round in
 100         activity = Activity()
 101     else:
 102       print("Cannot parse line " + str(i) + ".")
 103
 104   # Return the result
 105   return activityByPath
 106
 107 def activity(id, field, startFrom):
 108   """Shows the activity for the specified developer in the current repo.
 109
 110   Params:
 111     id: The id of the developer, currently only e-mail address is accepted.
 112     field: The field to filter on.
 113     startfrom: The commit to start logging from.
 114   """
 115
 116   git = Git(".")
 117   result = git.log("--numstat", "--pretty=format:%" + field, startFrom)
 118
 119   log = result.splitlines(True)
 120   activity = activityInArea(log, id)
 121
 122   result = []
 123
 124   for key, value in activity.iteritems():
 125     result.append(str(key) + " = " + str(value))
 126
 127   return result
 128
 129 def aggregateActivity(idFilter, field, startFrom):
 130   """Aggregates the activity for all developers
 131
 132   Args:
 133     idFilter: The id to filter on, if None all developers will be shown.
 134     field: The field to filter on.
 135   """
 136
 137   result = []
 138
 139   # TODO, implement
 140
 141   return result
 142
 143 def dispatch(*args):
 144   """Dispatches author related commands
 145   """
 146
 147   progname = os.path.basename(sys.argv[0]) + " author"
 148
 149   parser = OptionParser(option_class=commit.CommitOption, prog=progname)
 150
 151   parser.add_option(
 152       "-a", "--aggregate",
 153       action="store_true",
 154       help="aggregate the results")
 155
 156   parser.add_option(
 157       "-d", "--developer",
 158       help="the id to filter on")
 159
 160   parser.add_option(
 161       "-i", "--id",
 162       help="the one/two letter identifier specifying which field to filter on")
 163
 164   parser.add_option(
 165       "-s", "--start-from",
 166       type="commit",
 167       metavar="COMMIT",
 168       help="the commit to start logging from")
 169
 170   parser.set_default("id", "ae")
 171   parser.set_default("start_from", "HEAD")
 172
 173   (options, args) = parser.parse_args(list(args))
 174
 175   if not options.aggregate and not options.developer:
 176     parser.error("When not specifying a developer the -a switch is required")
 177
 178   if options.aggregate:
 179     result = aggregateActivity(options.developer, options.id, options.start_from)
 180   else:
 181     result = activity(options.developer, options.id, options.start_from)
 182
 183   for line in result:
 184     print(line)
 185