src/git_stats/matcher.py

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