mercurial/mdiff.py
author Matt Mackall <mpm@selenic.com>
Tue, 04 Oct 2005 11:25:48 -0700
changeset 1379 8ee7ce877be2
parent 1378 a0fcfbbf52bb
child 1452 f1755621cb7d
permissions -rw-r--r--
Clean up mdiff imports
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
#
75840796e8e2 mdiff.py: kill #! line, add copyright notice
mpm@selenic.com
parents: 184
diff changeset
     3
# Copyright 2005 Matt Mackall <mpm@selenic.com>
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
1379
8ee7ce877be2 Clean up mdiff imports
Matt Mackall <mpm@selenic.com>
parents: 1378
diff changeset
     8
import difflib, struct, bdiff, util, mpatch
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
     9
1015
22571b8d35d3 Add automatic binary file detection to diff and export
mpm@selenic.com
parents: 582
diff changeset
    10
def unidiff(a, ad, b, bd, fn, r=None, text=False):
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
    11
35
9197c26a414b unidiff: punt on comparing empty files
mpm@selenic.com
parents: 0
diff changeset
    12
    if not a and not b: return ""
1379
8ee7ce877be2 Clean up mdiff imports
Matt Mackall <mpm@selenic.com>
parents: 1378
diff changeset
    13
    epoch = util.datestr((0, 0))
264
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
    14
1379
8ee7ce877be2 Clean up mdiff imports
Matt Mackall <mpm@selenic.com>
parents: 1378
diff changeset
    15
    if not text and (util.binary(a) or util.binary(b)):
1015
22571b8d35d3 Add automatic binary file detection to diff and export
mpm@selenic.com
parents: 582
diff changeset
    16
        l = ['Binary file %s has changed\n' % fn]
22571b8d35d3 Add automatic binary file detection to diff and export
mpm@selenic.com
parents: 582
diff changeset
    17
    elif a == None:
264
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
    18
        b = b.splitlines(1)
1378
a0fcfbbf52bb make diff dates be epoch for add/remove
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1015
diff changeset
    19
        l1 = "--- %s\t%s\n" % ("/dev/null", epoch)
264
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
    20
        l2 = "+++ %s\t%s\n" % ("b/" + fn, bd)
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
    21
        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
    22
        l = [l1, l2, l3] + ["+" + e for e in b]
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
    23
    elif b == None:
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
    24
        a = a.splitlines(1)
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
    25
        l1 = "--- %s\t%s\n" % ("a/" + fn, ad)
1378
a0fcfbbf52bb make diff dates be epoch for add/remove
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 1015
diff changeset
    26
        l2 = "+++ %s\t%s\n" % ("/dev/null", epoch)
264
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
    27
        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
    28
        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
    29
    else:
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
    30
        a = a.splitlines(1)
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
    31
        b = b.splitlines(1)
272
467cea2bf2d8 diff: use tab to separate date from filename
mpm@selenic.com
parents: 264
diff changeset
    32
        l = list(difflib.unified_diff(a, b, "a/" + fn, "b/" + fn))
278
777e388c06d6 unidiff: handle empty diffs more gracefully
mpm@selenic.com
parents: 272
diff changeset
    33
        if not l: return ""
272
467cea2bf2d8 diff: use tab to separate date from filename
mpm@selenic.com
parents: 264
diff changeset
    34
        # difflib uses a space, rather than a tab
467cea2bf2d8 diff: use tab to separate date from filename
mpm@selenic.com
parents: 264
diff changeset
    35
        l[0] = l[0][:-2] + "\t" + ad + "\n"
467cea2bf2d8 diff: use tab to separate date from filename
mpm@selenic.com
parents: 264
diff changeset
    36
        l[1] = l[1][:-2] + "\t" + bd + "\n"
170
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
    37
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
    38
    for ln in xrange(len(l)):
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
    39
        if l[ln][-1] != '\n':
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
    40
            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
    41
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
    42
    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
    43
        l.insert(0, "diff %s %s\n" %
8f8bb77d560e Show revisions in diffs like CVS, based on a patch from Goffredo Baroncelli.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 361
diff changeset
    44
                    (' '.join(["-r %s" % rev for rev in r]), fn))
8f8bb77d560e Show revisions in diffs like CVS, based on a patch from Goffredo Baroncelli.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 361
diff changeset
    45
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
    46
    return "".join(l)
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
    47
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
    48
def sortdiff(a, b):
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
    49
    la = lb = 0
325
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    50
    lena = len(a)
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    51
    lenb = len(b)
515
03f27b1381f9 Whitespace cleanups
mpm@selenic.com
parents: 432
diff changeset
    52
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
    53
    while 1:
325
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    54
        am, bm, = la, lb
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    55
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    56
        # walk over matching lines
326
235443668bea mdiff: fix the fix
mpm@selenic.com
parents: 325
diff changeset
    57
        while lb < lenb and la < lena and a[la] == b[lb] :
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
    58
            la += 1
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
    59
            lb += 1
318
2819f63b16bf mdiff: revert grouping optimization for the time being
mpm@selenic.com
parents: 317
diff changeset
    60
325
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    61
        if la > am:
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    62
            yield (am, bm, la - am) # return a match
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    63
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    64
        # skip mismatched lines from b
361
88268aa2b8d2 Fix another sortdiff cornercase
mpm@selenic.com
parents: 330
diff changeset
    65
        while la < lena and lb < lenb and b[lb] < a[la]:
325
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    66
            lb += 1
318
2819f63b16bf mdiff: revert grouping optimization for the time being
mpm@selenic.com
parents: 317
diff changeset
    67
325
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    68
        if lb >= lenb:
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    69
            break
515
03f27b1381f9 Whitespace cleanups
mpm@selenic.com
parents: 432
diff changeset
    70
325
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    71
        # skip mismatched lines from a
361
88268aa2b8d2 Fix another sortdiff cornercase
mpm@selenic.com
parents: 330
diff changeset
    72
        while la < lena and lb < lenb and b[lb] > a[la]:
325
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    73
            la += 1
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    74
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    75
        if la >= lena:
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    76
            break
515
03f27b1381f9 Whitespace cleanups
mpm@selenic.com
parents: 432
diff changeset
    77
325
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    78
    yield (lena, lenb, 0)
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
    79
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
    80
def diff(a, b, sorted=0):
325
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    81
    if not a:
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    82
        s = "".join(b)
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    83
        return s and (struct.pack(">lll", 0, 0, len(s)) + s)
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    84
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
    85
    bin = []
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
    86
    p = [0]
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
    87
    for i in a: p.append(p[-1] + len(i))
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
    88
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
    89
    if sorted:
317
b18ce742566a hg commit: user and date options
mpm@selenic.com
parents: 278
diff changeset
    90
        try:
b18ce742566a hg commit: user and date options
mpm@selenic.com
parents: 278
diff changeset
    91
            d = sortdiff(a, b)
b18ce742566a hg commit: user and date options
mpm@selenic.com
parents: 278
diff changeset
    92
        except:
b18ce742566a hg commit: user and date options
mpm@selenic.com
parents: 278
diff changeset
    93
            raise
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
    94
    else:
325
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    95
        d = difflib.SequenceMatcher(None, a, b).get_matching_blocks()
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    96
    la = 0
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    97
    lb = 0
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    98
    for am, bm, size in d:
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
    99
        s = "".join(b[lb:bm])
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
   100
        if am > la or s:
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
   101
            bin.append(struct.pack(">lll", p[la], p[am], len(s)) + s)
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
   102
        la = am + size
ad87e19854a6 mdiff: reinstate new algorithm
mpm@selenic.com
parents: 318
diff changeset
   103
        lb = bm + size
515
03f27b1381f9 Whitespace cleanups
mpm@selenic.com
parents: 432
diff changeset
   104
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
   105
    return "".join(bin)
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
   106
120
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   107
def patchtext(bin):
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   108
    pos = 0
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   109
    t = []
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   110
    while pos < len(bin):
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   111
        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
   112
        pos += 12
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   113
        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
   114
        pos += l
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   115
    return "".join(t)
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
   116
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
   117
def patch(a, bin):
1379
8ee7ce877be2 Clean up mdiff imports
Matt Mackall <mpm@selenic.com>
parents: 1378
diff changeset
   118
    return mpatch.patches(a, [bin])
432
3b9e3d3d2810 Start using bdiff for generating deltas
mpm@selenic.com
parents: 396
diff changeset
   119
1379
8ee7ce877be2 Clean up mdiff imports
Matt Mackall <mpm@selenic.com>
parents: 1378
diff changeset
   120
patches = mpatch.patches
432
3b9e3d3d2810 Start using bdiff for generating deltas
mpm@selenic.com
parents: 396
diff changeset
   121
textdiff = bdiff.bdiff
3b9e3d3d2810 Start using bdiff for generating deltas
mpm@selenic.com
parents: 396
diff changeset
   122
3b9e3d3d2810 Start using bdiff for generating deltas
mpm@selenic.com
parents: 396
diff changeset
   123