tools/intersect-diffs.py

   1 #!/usr/bin/env python
   2 ##  Copyright (c) 2012 The WebM project authors. All Rights Reserved.
   3 ##
   4 ##  Use of this source code is governed by a BSD-style license
   5 ##  that can be found in the LICENSE file in the root of the source
   6 ##  tree. An additional intellectual property rights grant can be found
   7 ##  in the file PATENTS.  All contributing project authors may
   8 ##  be found in the AUTHORS file in the root of the source tree.
   9 ##
  10 """Calculates the "intersection" of two unified diffs.
  11
  12 Given two diffs, A and B, it finds all hunks in B that had non-context lines
  13 in A and prints them to stdout. This is useful to determine the hunks in B that
  14 are relevant to A. The resulting file can be applied with patch(1) on top of A.
  15 """
  16
  17 __author__ = "jkoleszar@google.com"
  18
  19 import re
  20 import sys
  21
  22
  23 class DiffLines(object):
  24     """A container for one half of a diff."""
  25
  26     def __init__(self, filename, offset, length):
  27         self.filename = filename
  28         self.offset = offset
  29         self.length = length
  30         self.lines = []
  31         self.delta_line_nums = []
  32
  33     def Append(self, line):
  34         l = len(self.lines)
  35         if line[0] != " ":
  36             self.delta_line_nums.append(self.offset + l)
  37         self.lines.append(line[1:])
  38         assert l+1 <= self.length
  39
  40     def Complete(self):
  41         return len(self.lines) == self.length
  42
  43     def __contains__(self, item):
  44         return item >= self.offset and item <= self.offset + self.length - 1
  45
  46
  47 class DiffHunk(object):
  48     """A container for one diff hunk, consisting of two DiffLines."""
  49
  50     def __init__(self, header, file_a, file_b, start_a, len_a, start_b, len_b):
  51         self.header = header
  52         self.left = DiffLines(file_a, start_a, len_a)
  53         self.right = DiffLines(file_b, start_b, len_b)
  54         self.lines = []
  55
  56     def Append(self, line):
  57         """Adds a line to the DiffHunk and its DiffLines children."""
  58         if line[0] == "-":
  59             self.left.Append(line)
  60         elif line[0] == "+":
  61             self.right.Append(line)
  62         elif line[0] == " ":
  63             self.left.Append(line)
  64             self.right.Append(line)
  65         else:
  66             assert False, ("Unrecognized character at start of diff line "
  67                            "%r" % line[0])
  68         self.lines.append(line)
  69
  70     def Complete(self):
  71         return self.left.Complete() and self.right.Complete()
  72
  73     def __repr__(self):
  74         return "DiffHunk(%s, %s, len %d)" % (
  75             self.left.filename, self.right.filename,
  76             max(self.left.length, self.right.length))
  77
  78
  79 def ParseDiffHunks(stream):
  80     """Walk a file-like object, yielding DiffHunks as they're parsed."""
  81
  82     file_regex = re.compile(r"(\+\+\+|---) (\S+)")
  83     range_regex = re.compile(r"@@ -(\d+)(,(\d+))? \+(\d+)(,(\d+))?")
  84     hunk = None
  85     while True:
  86         line = stream.readline()
  87         if not line:
  88             break
  89
  90         if hunk is None:
  91             # Parse file names
  92             diff_file = file_regex.match(line)
  93             if diff_file:
  94               if line.startswith("---"):
  95                   a_line = line
  96                   a = diff_file.group(2)
  97                   continue
  98               if line.startswith("+++"):
  99                   b_line = line
 100                   b = diff_file.group(2)
 101                   continue
 102
 103             # Parse offset/lengths
 104             diffrange = range_regex.match(line)
 105             if diffrange:
 106                 if diffrange.group(2):
 107                     start_a = int(diffrange.group(1))
 108                     len_a = int(diffrange.group(3))
 109                 else:
 110                     start_a = 1
 111                     len_a = int(diffrange.group(1))
 112
 113                 if diffrange.group(5):
 114                     start_b = int(diffrange.group(4))
 115                     len_b = int(diffrange.group(6))
 116                 else:
 117                     start_b = 1
 118                     len_b = int(diffrange.group(4))
 119
 120                 header = [a_line, b_line, line]
 121                 hunk = DiffHunk(header, a, b, start_a, len_a, start_b, len_b)
 122         else:
 123             # Add the current line to the hunk
 124             hunk.Append(line)
 125
 126             # See if the whole hunk has been parsed. If so, yield it and prepare
 127             # for the next hunk.
 128             if hunk.Complete():
 129                 yield hunk
 130                 hunk = None
 131
 132     # Partial hunks are a parse error
 133     assert hunk is None
 134
 135
 136 def FormatDiffHunks(hunks):
 137     """Re-serialize a list of DiffHunks."""
 138     r = []
 139     last_header = None
 140     for hunk in hunks:
 141         this_header = hunk.header[0:2]
 142         if last_header != this_header:
 143             r.extend(hunk.header)
 144             last_header = this_header
 145         else:
 146             r.extend(hunk.header[2])
 147         r.extend(hunk.lines)
 148         r.append("\n")
 149     return "".join(r)
 150
 151
 152 def ZipHunks(rhs_hunks, lhs_hunks):
 153     """Join two hunk lists on filename."""
 154     for rhs_hunk in rhs_hunks:
 155         rhs_file = rhs_hunk.right.filename.split("/")[1:]
 156
 157         for lhs_hunk in lhs_hunks:
 158             lhs_file = lhs_hunk.left.filename.split("/")[1:]
 159             if lhs_file != rhs_file:
 160                 continue
 161             yield (rhs_hunk, lhs_hunk)
 162
 163
 164 def main():
 165     old_hunks = [x for x in ParseDiffHunks(open(sys.argv[1], "r"))]
 166     new_hunks = [x for x in ParseDiffHunks(open(sys.argv[2], "r"))]
 167     out_hunks = []
 168
 169     # Join the right hand side of the older diff with the left hand side of the
 170     # newer diff.
 171     for old_hunk, new_hunk in ZipHunks(old_hunks, new_hunks):
 172         if new_hunk in out_hunks:
 173             continue
 174         old_lines = old_hunk.right
 175         new_lines = new_hunk.left
 176
 177         # Determine if this hunk overlaps any non-context line from the other
 178         for i in old_lines.delta_line_nums:
 179             if i in new_lines:
 180                 out_hunks.append(new_hunk)
 181                 break
 182
 183     if out_hunks:
 184         print FormatDiffHunks(out_hunks)
 185         sys.exit(1)
 186
 187 if __name__ == "__main__":
 188     main()