mercurial/mdiff.py
author Dustin Sallings <dustin@spy.net>
Thu, 01 Nov 2007 12:17:59 -0700
changeset 5482 e5eedd74e70f
parent 5367 7530334bf301
child 5863 3d1f9dcecdea
permissions -rw-r--r--
Use both the from and to name in mdiff.unidiff. This fixes a compatibility issue with git diffs. * * *
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
239
75840796e8e2 mdiff.py: kill #! line, add copyright notice
mpm@selenic.com
parents: 184
diff changeset
     1
# mdiff.py - diff and patch routines for mercurial
75840796e8e2 mdiff.py: kill #! line, add copyright notice
mpm@selenic.com
parents: 184
diff changeset
     2
#
2859
345bac2bc4ec update copyrights.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2580
diff changeset
     3
# Copyright 2005, 2006 Matt Mackall <mpm@selenic.com>
239
75840796e8e2 mdiff.py: kill #! line, add copyright notice
mpm@selenic.com
parents: 184
diff changeset
     4
#
75840796e8e2 mdiff.py: kill #! line, add copyright notice
mpm@selenic.com
parents: 184
diff changeset
     5
# This software may be used and distributed according to the terms
75840796e8e2 mdiff.py: kill #! line, add copyright notice
mpm@selenic.com
parents: 184
diff changeset
     6
# of the GNU General Public License, incorporated herein by reference.
75840796e8e2 mdiff.py: kill #! line, add copyright notice
mpm@selenic.com
parents: 184
diff changeset
     7
4108
226df1808f16 merge with crew-stable
Thomas Arendsen Hein <thomas@intevation.de>
parents: 3877 4103
diff changeset
     8
import bdiff, mpatch, re, struct, util, md5
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
     9
2251
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
    10
def splitnewlines(text):
2248
b914f0557832 fix diffs containing embedded "\r".
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2078
diff changeset
    11
    '''like str.splitlines, but only split on newlines.'''
2251
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
    12
    lines = [l + '\n' for l in text.split('\n')]
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
    13
    if lines:
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
    14
        if lines[-1] == '\n':
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
    15
            lines.pop()
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
    16
        else:
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
    17
            lines[-1] = lines[-1][:-1]
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
    18
    return lines
2248
b914f0557832 fix diffs containing embedded "\r".
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2078
diff changeset
    19
2874
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    20
class diffopts(object):
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    21
    '''context is the number of context lines
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    22
    text treats all files as text
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    23
    showfunc enables diff -p output
2907
8b02af865990 Add diff --git option
Brendan Cully <brendan@kublai.com>
parents: 2874
diff changeset
    24
    git enables the git extended patch format
3199
096f1c73cdc3 Add -D/--nodates options to hg diff/export that removes dates from diff headers
Stephen Darnell <stephen@darnell.plus.com>
parents: 3026
diff changeset
    25
    nodates removes dates from diff headers
2874
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    26
    ignorews ignores all whitespace changes in the diff
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    27
    ignorewsamount ignores changes in the amount of whitespace
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    28
    ignoreblanklines ignores changes whose lines are all blank'''
396
8f8bb77d560e Show revisions in diffs like CVS, based on a patch from Goffredo Baroncelli.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 361
diff changeset
    29
2874
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    30
    defaults = {
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    31
        'context': 3,
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    32
        'text': False,
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    33
        'showfunc': True,
2907
8b02af865990 Add diff --git option
Brendan Cully <brendan@kublai.com>
parents: 2874
diff changeset
    34
        'git': False,
3199
096f1c73cdc3 Add -D/--nodates options to hg diff/export that removes dates from diff headers
Stephen Darnell <stephen@darnell.plus.com>
parents: 3026
diff changeset
    35
        'nodates': False,
2874
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    36
        'ignorews': False,
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    37
        'ignorewsamount': False,
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    38
        'ignoreblanklines': False,
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    39
        }
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    40
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    41
    __slots__ = defaults.keys()
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    42
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    43
    def __init__(self, **opts):
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    44
        for k in self.__slots__:
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    45
            v = opts.get(k)
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    46
            if v is None:
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    47
                v = self.defaults[k]
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    48
            setattr(self, k, v)
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    49
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    50
defaultopts = diffopts()
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    51
4878
372d93f03d3a diff: correctly handle combinations of whitespace options
Matt Mackall <mpm@selenic.com>
parents: 4679
diff changeset
    52
def wsclean(opts, text):
372d93f03d3a diff: correctly handle combinations of whitespace options
Matt Mackall <mpm@selenic.com>
parents: 4679
diff changeset
    53
    if opts.ignorews:
372d93f03d3a diff: correctly handle combinations of whitespace options
Matt Mackall <mpm@selenic.com>
parents: 4679
diff changeset
    54
        text = re.sub('[ \t]+', '', text)
372d93f03d3a diff: correctly handle combinations of whitespace options
Matt Mackall <mpm@selenic.com>
parents: 4679
diff changeset
    55
    elif opts.ignorewsamount:
372d93f03d3a diff: correctly handle combinations of whitespace options
Matt Mackall <mpm@selenic.com>
parents: 4679
diff changeset
    56
        text = re.sub('[ \t]+', ' ', text)
372d93f03d3a diff: correctly handle combinations of whitespace options
Matt Mackall <mpm@selenic.com>
parents: 4679
diff changeset
    57
        text = re.sub('[ \t]+\n', '\n', text)
372d93f03d3a diff: correctly handle combinations of whitespace options
Matt Mackall <mpm@selenic.com>
parents: 4679
diff changeset
    58
    if opts.ignoreblanklines:
372d93f03d3a diff: correctly handle combinations of whitespace options
Matt Mackall <mpm@selenic.com>
parents: 4679
diff changeset
    59
        text = re.sub('\n+', '', text)
372d93f03d3a diff: correctly handle combinations of whitespace options
Matt Mackall <mpm@selenic.com>
parents: 4679
diff changeset
    60
    return text
372d93f03d3a diff: correctly handle combinations of whitespace options
Matt Mackall <mpm@selenic.com>
parents: 4679
diff changeset
    61
5482
e5eedd74e70f Use both the from and to name in mdiff.unidiff.
Dustin Sallings <dustin@spy.net>
parents: 5367
diff changeset
    62
def unidiff(a, ad, b, bd, fn1, fn2, r=None, opts=defaultopts):
4679
826659bd8053 git patches: correct handling of filenames with spaces
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4361
diff changeset
    63
    def datetag(date, addtab=True):
826659bd8053 git patches: correct handling of filenames with spaces
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4361
diff changeset
    64
        if not opts.git and not opts.nodates:
826659bd8053 git patches: correct handling of filenames with spaces
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4361
diff changeset
    65
            return '\t%s\n' % date
5482
e5eedd74e70f Use both the from and to name in mdiff.unidiff.
Dustin Sallings <dustin@spy.net>
parents: 5367
diff changeset
    66
        if addtab and ' ' in fn1:
4679
826659bd8053 git patches: correct handling of filenames with spaces
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4361
diff changeset
    67
            return '\t\n'
826659bd8053 git patches: correct handling of filenames with spaces
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4361
diff changeset
    68
        return '\n'
3026
d838bfac668d Remove dates from git export file lines - they confuse git-apply
Brendan Cully <brendan@kublai.com>
parents: 2907
diff changeset
    69
35
9197c26a414b unidiff: punt on comparing empty files
mpm@selenic.com
parents: 0
diff changeset
    70
    if not a and not b: return ""
1379
8ee7ce877be2 Clean up mdiff imports
Matt Mackall <mpm@selenic.com>
parents: 1378
diff changeset
    71
    epoch = util.datestr((0, 0))
264
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
    72
2874
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
    73
    if not opts.text and (util.binary(a) or util.binary(b)):
4103
544838cc1158 Don't lie that "binary file has changed"
tailgunner@smtp.ru
parents: 3199
diff changeset
    74
        def h(v):
544838cc1158 Don't lie that "binary file has changed"
tailgunner@smtp.ru
parents: 3199
diff changeset
    75
            # md5 is used instead of sha1 because md5 is supposedly faster
544838cc1158 Don't lie that "binary file has changed"
tailgunner@smtp.ru
parents: 3199
diff changeset
    76
            return md5.new(v).digest()
544838cc1158 Don't lie that "binary file has changed"
tailgunner@smtp.ru
parents: 3199
diff changeset
    77
        if a and b and len(a) == len(b) and h(a) == h(b):
544838cc1158 Don't lie that "binary file has changed"
tailgunner@smtp.ru
parents: 3199
diff changeset
    78
            return ""
5482
e5eedd74e70f Use both the from and to name in mdiff.unidiff.
Dustin Sallings <dustin@spy.net>
parents: 5367
diff changeset
    79
        l = ['Binary file %s has changed\n' % fn1]
1723
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
    80
    elif not a:
2251
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
    81
        b = splitnewlines(b)
1723
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
    82
        if a is None:
4679
826659bd8053 git patches: correct handling of filenames with spaces
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4361
diff changeset
    83
            l1 = '--- /dev/null%s' % datetag(epoch, False)
1723
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
    84
        else:
5482
e5eedd74e70f Use both the from and to name in mdiff.unidiff.
Dustin Sallings <dustin@spy.net>
parents: 5367
diff changeset
    85
            l1 = "--- %s%s" % ("a/" + fn1, datetag(ad))
e5eedd74e70f Use both the from and to name in mdiff.unidiff.
Dustin Sallings <dustin@spy.net>
parents: 5367
diff changeset
    86
        l2 = "+++ %s%s" % ("b/" + fn2, datetag(bd))
264
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
    87
        l3 = "@@ -0,0 +1,%d @@\n" % len(b)
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
    88
        l = [l1, l2, l3] + ["+" + e for e in b]
1723
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
    89
    elif not b:
2251
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
    90
        a = splitnewlines(a)
5482
e5eedd74e70f Use both the from and to name in mdiff.unidiff.
Dustin Sallings <dustin@spy.net>
parents: 5367
diff changeset
    91
        l1 = "--- %s%s" % ("a/" + fn1, datetag(ad))
1723
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
    92
        if b is None:
4679
826659bd8053 git patches: correct handling of filenames with spaces
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4361
diff changeset
    93
            l2 = '+++ /dev/null%s' % datetag(epoch, False)
1723
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
    94
        else:
5482
e5eedd74e70f Use both the from and to name in mdiff.unidiff.
Dustin Sallings <dustin@spy.net>
parents: 5367
diff changeset
    95
            l2 = "+++ %s%s" % ("b/" + fn2, datetag(bd))
264
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
    96
        l3 = "@@ -1,%d +0,0 @@\n" % len(a)
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
    97
        l = [l1, l2, l3] + ["-" + e for e in a]
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
    98
    else:
2251
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
    99
        al = splitnewlines(a)
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
   100
        bl = splitnewlines(b)
5482
e5eedd74e70f Use both the from and to name in mdiff.unidiff.
Dustin Sallings <dustin@spy.net>
parents: 5367
diff changeset
   101
        l = list(bunidiff(a, b, al, bl, "a/" + fn1, "b/" + fn2, opts=opts))
278
777e388c06d6 unidiff: handle empty diffs more gracefully
mpm@selenic.com
parents: 272
diff changeset
   102
        if not l: return ""
272
467cea2bf2d8 diff: use tab to separate date from filename
mpm@selenic.com
parents: 264
diff changeset
   103
        # difflib uses a space, rather than a tab
3026
d838bfac668d Remove dates from git export file lines - they confuse git-apply
Brendan Cully <brendan@kublai.com>
parents: 2907
diff changeset
   104
        l[0] = "%s%s" % (l[0][:-2], datetag(ad))
d838bfac668d Remove dates from git export file lines - they confuse git-apply
Brendan Cully <brendan@kublai.com>
parents: 2907
diff changeset
   105
        l[1] = "%s%s" % (l[1][:-2], datetag(bd))
170
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
   106
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
   107
    for ln in xrange(len(l)):
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
   108
        if l[ln][-1] != '\n':
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
   109
            l[ln] += "\n\ No newline at end of file\n"
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
   110
396
8f8bb77d560e Show revisions in diffs like CVS, based on a patch from Goffredo Baroncelli.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 361
diff changeset
   111
    if r:
8f8bb77d560e Show revisions in diffs like CVS, based on a patch from Goffredo Baroncelli.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 361
diff changeset
   112
        l.insert(0, "diff %s %s\n" %
5482
e5eedd74e70f Use both the from and to name in mdiff.unidiff.
Dustin Sallings <dustin@spy.net>
parents: 5367
diff changeset
   113
                    (' '.join(["-r %s" % rev for rev in r]), fn1))
396
8f8bb77d560e Show revisions in diffs like CVS, based on a patch from Goffredo Baroncelli.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 361
diff changeset
   114
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
   115
    return "".join(l)
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
   116
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   117
# somewhat self contained replacement for difflib.unified_diff
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   118
# t1 and t2 are the text to be diffed
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   119
# l1 and l2 are the text broken up into lines
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   120
# header1 and header2 are the filenames for the diff output
2874
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
   121
def bunidiff(t1, t2, l1, l2, header1, header2, opts=defaultopts):
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   122
    def contextend(l, len):
2874
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
   123
        ret = l + opts.context
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   124
        if ret > len:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   125
            ret = len
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   126
        return ret
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   127
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   128
    def contextstart(l):
2874
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
   129
        ret = l - opts.context
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   130
        if ret < 0:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   131
            return 0
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   132
        return ret
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   133
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   134
    def yieldhunk(hunk, header):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   135
        if header:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   136
            for x in header:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   137
                yield x
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   138
        (astart, a2, bstart, b2, delta) = hunk
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   139
        aend = contextend(a2, len(l1))
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   140
        alen = aend - astart
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   141
        blen = b2 - bstart + aend - a2
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   142
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   143
        func = ""
2874
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
   144
        if opts.showfunc:
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   145
            # walk backwards from the start of the context
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   146
            # to find a line starting with an alphanumeric char.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   147
            for x in xrange(astart, -1, -1):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   148
                t = l1[x].rstrip()
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   149
                if funcre.match(t):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   150
                    func = ' ' + t[:40]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   151
                    break
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   152
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   153
        yield "@@ -%d,%d +%d,%d @@%s\n" % (astart + 1, alen,
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   154
                                           bstart + 1, blen, func)
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   155
        for x in delta:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   156
            yield x
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   157
        for x in xrange(a2, aend):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   158
            yield ' ' + l1[x]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   159
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   160
    header = [ "--- %s\t\n" % header1, "+++ %s\t\n" % header2 ]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   161
2874
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
   162
    if opts.showfunc:
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   163
        funcre = re.compile('\w')
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   164
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   165
    # bdiff.blocks gives us the matching sequences in the files.  The loop
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   166
    # below finds the spaces between those matching sequences and translates
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   167
    # them into diff output.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   168
    #
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   169
    diff = bdiff.blocks(t1, t2)
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   170
    hunk = None
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   171
    for i in xrange(len(diff)):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   172
        # The first match is special.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   173
        # we've either found a match starting at line 0 or a match later
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   174
        # in the file.  If it starts later, old and new below will both be
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   175
        # empty and we'll continue to the next match.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   176
        if i > 0:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   177
            s = diff[i-1]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   178
        else:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   179
            s = [0, 0, 0, 0]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   180
        delta = []
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   181
        s1 = diff[i]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   182
        a1 = s[1]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   183
        a2 = s1[0]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   184
        b1 = s[3]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   185
        b2 = s1[2]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   186
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   187
        old = l1[a1:a2]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   188
        new = l2[b1:b2]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   189
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   190
        # bdiff sometimes gives huge matches past eof, this check eats them,
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   191
        # and deals with the special first match case described above
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   192
        if not old and not new:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   193
            continue
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   194
4878
372d93f03d3a diff: correctly handle combinations of whitespace options
Matt Mackall <mpm@selenic.com>
parents: 4679
diff changeset
   195
        if opts.ignorews or opts.ignorewsamount or opts.ignoreblanklines:
372d93f03d3a diff: correctly handle combinations of whitespace options
Matt Mackall <mpm@selenic.com>
parents: 4679
diff changeset
   196
            if wsclean(opts, "".join(old)) == wsclean(opts, "".join(new)):
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   197
                continue
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   198
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   199
        astart = contextstart(a1)
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   200
        bstart = contextstart(b1)
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   201
        prev = None
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   202
        if hunk:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   203
            # join with the previous hunk if it falls inside the context
2874
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2859
diff changeset
   204
            if astart < hunk[1] + opts.context + 1:
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   205
                prev = hunk
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   206
                astart = hunk[1]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   207
                bstart = hunk[3]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   208
            else:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   209
                for x in yieldhunk(hunk, header):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   210
                    yield x
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   211
                # we only want to yield the header if the files differ, and
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   212
                # we only want to yield it once.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   213
                header = None
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   214
        if prev:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   215
            # we've joined the previous hunk, record the new ending points.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   216
            hunk[1] = a2
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   217
            hunk[3] = b2
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   218
            delta = hunk[4]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   219
        else:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   220
            # create a new hunk
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   221
            hunk = [ astart, a2, bstart, b2, delta ]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   222
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   223
        delta[len(delta):] = [ ' ' + x for x in l1[astart:a1] ]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   224
        delta[len(delta):] = [ '-' + x for x in old ]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   225
        delta[len(delta):] = [ '+' + x for x in new ]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   226
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   227
    if hunk:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   228
        for x in yieldhunk(hunk, header):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   229
            yield x
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
   230
120
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   231
def patchtext(bin):
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   232
    pos = 0
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   233
    t = []
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   234
    while pos < len(bin):
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   235
        p1, p2, l = struct.unpack(">lll", bin[pos:pos + 12])
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   236
        pos += 12
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   237
        t.append(bin[pos:pos + l])
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   238
        pos += l
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   239
    return "".join(t)
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   240
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
   241
def patch(a, bin):
1379
8ee7ce877be2 Clean up mdiff imports
Matt Mackall <mpm@selenic.com>
parents: 1378
diff changeset
   242
    return mpatch.patches(a, [bin])
432
3b9e3d3d2810 Start using bdiff for generating deltas
mpm@selenic.com
parents: 396
diff changeset
   243
4361
99c853a1408c add mdiff.get_matching_blocks
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4108
diff changeset
   244
# similar to difflib.SequenceMatcher.get_matching_blocks
99c853a1408c add mdiff.get_matching_blocks
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4108
diff changeset
   245
def get_matching_blocks(a, b):
99c853a1408c add mdiff.get_matching_blocks
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4108
diff changeset
   246
    return [(d[0], d[2], d[1] - d[0]) for d in bdiff.blocks(a, b)]
99c853a1408c add mdiff.get_matching_blocks
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4108
diff changeset
   247
5367
7530334bf301 revlog: generate trivial deltas against null revision
Matt Mackall <mpm@selenic.com>
parents: 4878
diff changeset
   248
def trivialdiffheader(length):
7530334bf301 revlog: generate trivial deltas against null revision
Matt Mackall <mpm@selenic.com>
parents: 4878
diff changeset
   249
    return struct.pack(">lll", 0, 0, length)
7530334bf301 revlog: generate trivial deltas against null revision
Matt Mackall <mpm@selenic.com>
parents: 4878
diff changeset
   250
1379
8ee7ce877be2 Clean up mdiff imports
Matt Mackall <mpm@selenic.com>
parents: 1378
diff changeset
   251
patches = mpatch.patches
2078
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 1723
diff changeset
   252
patchedsize = mpatch.patchedsize
432
3b9e3d3d2810 Start using bdiff for generating deltas
mpm@selenic.com
parents: 396
diff changeset
   253
textdiff = bdiff.bdiff