src/git_stats/author.py

   1 #!/usr/bin/env python
   2
   3 from git_python import Git
   4
   5 class Activity:
   6   """Simple storage class containing stats on the activity in one file."""
   7
   8   count = 1
   9   added = 0
  10   deleted = 0
  11   id = []
  12
  13   def __str__(self):
  14     return str(self.count) + ": +" + str(self.added) + " -" + str(self.deleted)
  15
  16 def activityInArea(log, filterid):
  17   """Parses the specified file containing commit logs.
  18   The output is filtered to contain only the results of the specified author.
  19   The output is expected to be in the format described below:
  20   [
  21     [<id>\n]+
  22     [<lines added>\t<lines deleted>\t<path>]+
  23     \n
  24   ]+
  25
  26   Params:
  27     log: The log formatted as described above.
  28     filterid: The id address to filter on.
  29
  30   Returns:
  31     A dictionary containing activity per path is returned.
  32     Each path contains one Activity object with the aggregated results.
  33   """
  34
  35   # Create a dict to store the activity stats per path
  36   activityByPath = {}
  37
  38   # Create a place to store the result in
  39   activity = Activity()
  40
  41   i = 0
  42
  43   # Parse all the lines in the file
  44   for line in log:
  45     i += 1
  46
  47     # Split the line at the tab and store the data
  48     splitline = line.split('\t')
  49     size = len(splitline)
  50     length = len(line.lstrip())
  51
  52     # There is something on this line, but it contains no separator
  53     if size == 1 and length > 0:
  54       # Get the id address minus the newline
  55       activity.id.append(line[:-1])
  56     elif size == 3:
  57       try:
  58         addpart = splitline[0]
  59         deletepart = splitline[1]
  60
  61         if addpart == '-':
  62           addpart = 0
  63
  64         if deletepart == '-':
  65           deletepart = 0
  66
  67         activity.added = int(addpart)
  68         activity.deleted = int(deletepart)
  69       except ValueError, e:
  70         print("On line '" + str(i) + "', could not convert number: " + str(e))
  71
  72       path = splitline[2][:-1]
  73     elif length == 0:
  74       if not filterid in activity.id:
  75         # Ignore other authors
  76         pass
  77       else:
  78         result = Activity()
  79
  80         # If we already encountered this path, update the count
  81         if activityByPath.has_key(path):
  82           known = activityByPath[path]
  83           result.added = known.added + activity.added
  84           result.deleted = known.deleted + activity.deleted
  85           result.count = known.count + activity.count
  86         else:
  87           result = activity
  88
  89         # Store it under it's path
  90         activityByPath[path] = result
  91
  92         # Create a fresh activity to store the next round in
  93         activity = Activity()
  94     else:
  95       print("Cannot parse line " + str(i) + ".")
  96
  97   # Return the result
  98   return activityByPath
  99
 100 def activity(id):
 101   """Shows the activity for the specified developer in the current repo.
 102
 103   Params:
 104     id: The id of the developer, currently only e-mail address is accepted.
 105   """
 106
 107   git = Git(".")
 108   result = git.log("--numstat", "--pretty=format:%ae")
 109
 110   log = result.splitlines(True)
 111   activity = activityInArea(log, id)
 112
 113   for key, value in activity.iteritems():
 114     print(str(key) + " = " + str(value))
 115
 116 def dispatch(*args):
 117   """Dispatches author related commands
 118   """
 119
 120   pass
 121
 122 if __name__ == '__main__':
 123   path = "testrepo.log"
 124
 125   # Open the log file
 126   file = open(path)
 127
 128   log = file.readlines()
 129
 130   activity = activityInArea(log, "sverre@rabbelier.nl")
 131
 132   for key, value in activity.iteritems():
 133     print(str(key) + " = " + str(value))
 134