src/git_stats/matcher.py

   1 #!/usr/bin/env python
   2
   3 import difflib
   4 import os
   5 import sys
   6
   7 from optparse import OptionParser
   8 from git_stats import commit
   9 from git_stats import diff
  10 from git_stats import parse
  11
  12 def calculateDiffSize(difference):
  13   """Calculates the true diff size
  14
  15   All lines that start with '+\t', '-\t' are counted.
  16   Lines that are of size 1 are also counted.
  17   """
  18
  19   size = 0
  20
  21   # Take each line and only count if it is part of the diff
  22   for line in difference:
  23     if (len(line) == 1 or line[1] == '\t' and
  24         (line[0] == '+' or line[0] == '-')):
  25       size += 1
  26
  27   return size
  28
  29 def findMatch(left, right):
  30   """Tries to find a match between left and right
  31
  32   If it is plausible that there is a match the difference is returend.
  33   Otherwise False is returned.
  34   """
  35
  36   # Get the diff and convert it to a usable format
  37   res = difflib.unified_diff(left, right, n=0, lineterm="")
  38   res = list(res)
  39
  40   # Get some sizes for easy calculation
  41   ressize = calculateDiffSize(res)
  42   leftsize = len(left)
  43   rightsize = len(right)
  44
  45   # The difference is larger than either side
  46   if ressize > leftsize or ressize > rightsize:
  47     return False
  48
  49   # The difference is larger than the average
  50   if ressize > (leftsize + rightsize)/2:
  51     return False
  52
  53   # This is probably a match, return the difference
  54   return res
  55
  56 def match(target):
  57   """Tries to find a match between added and removed hunks
  58
  59   The diff of the specified commit is retreived and it is
  60   split into hunks. The hunks that were added are compared
  61   with the hunks that were deleted. If they are similar the
  62   pair is deemed a match.
  63   """
  64
  65   # Retrieve the diff
  66   result = commit.getDiff(target)
  67   targetDiff = result.split('\n')
  68
  69   # And have it parsed, but don't add line numbering to the hunks
  70   parsedDiffs = diff.parseCommitDiff(targetDiff, number=False)
  71
  72   # To store the matches in
  73   result = []
  74
  75   # Iterate over all the diffs, e.g., take all pairs
  76   for left in parsedDiffs:
  77     for right in parsedDiffs:
  78       # Don't compare with self, that'd always match
  79       if left == right:
  80         continue
  81
  82       # A removal hunk, not interesting as a left side
  83       # Only interesting when comparing with addition
  84       if not left.linesAdded:
  85         continue
  86
  87       # An add hunk, not interesting as a right side
  88       # We are interested in this as a left side
  89       if not right.linesDeleted:
  90         continue
  91
  92       # Try to find a match for this pair
  93       res = findMatch(left.linesAdded, right.linesDeleted)
  94
  95       # There was no match
  96       if not res:
  97         continue
  98
  99       result.append(left.linesAdded, right.linesAdded, res)
 100
 101   return result
 102
 103 def dispatch(*args):
 104   """Dispatches matching related commands
 105   """
 106
 107   # Make the help show 'progname commit' instead of just 'progname'
 108   progname = os.path.basename(sys.argv[0]) + " matcher"
 109
 110   parser = OptionParser(option_class=parse.GitOption, prog=progname)
 111
 112   parser.add_option(
 113       "-m", "--matcher",
 114       type="commit",
 115       help="match the chunks of a diff to find code moves")
 116
 117   parser.set_default("matcher", "HEAD")
 118
 119   (options, args) = parser.parse_args(list(args))
 120
 121   if args:
 122     parser.error("Please specify a commit to analyse")
 123
 124   # Get the result
 125   result = match(options.matcher)
 126
 127   # And print it
 128   for line in result:
 129     print(line)
 130