author | Martin von Zweigbergk <martinvonz@google.com> |
Tue, 05 Apr 2022 12:06:32 -0700 | |
changeset 49123 | 5d205e476057 |
parent 48966 | 6000f5b25c9b |
child 49733 | 594fc56c0af7 |
permissions | -rw-r--r-- |
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
1 |
# bdiff.py - Python implementation of bdiff.c |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
2 |
# |
46819
d4ba4d51f85f
contributor: change mentions of mpm to olivia
Raphaël Gomès <rgomes@octobus.net>
parents:
45878
diff
changeset
|
3 |
# Copyright 2009 Olivia Mackall <olivia@selenic.com> and others |
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
4 |
# |
8225
46293a0c7e9f
updated license to be explicit about GPL version 2
Martin Geisler <mg@lazybytes.net>
parents:
7944
diff
changeset
|
5 |
# This software may be used and distributed according to the terms of the |
10263 | 6 |
# GNU General Public License version 2 or any later version. |
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
7 |
|
27335
c4e3ff497f89
bdiff: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents:
15530
diff
changeset
|
8 |
|
c4e3ff497f89
bdiff: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents:
15530
diff
changeset
|
9 |
import difflib |
c4e3ff497f89
bdiff: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents:
15530
diff
changeset
|
10 |
import re |
c4e3ff497f89
bdiff: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents:
15530
diff
changeset
|
11 |
import struct |
7944
e9b48afd0e78
pure/bdiff: fix circular import
Matt Mackall <mpm@selenic.com>
parents:
7703
diff
changeset
|
12 |
|
43076
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
41269
diff
changeset
|
13 |
|
7944
e9b48afd0e78
pure/bdiff: fix circular import
Matt Mackall <mpm@selenic.com>
parents:
7703
diff
changeset
|
14 |
def splitnewlines(text): |
e9b48afd0e78
pure/bdiff: fix circular import
Matt Mackall <mpm@selenic.com>
parents:
7703
diff
changeset
|
15 |
'''like str.splitlines, but only split on newlines.''' |
43077
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
16 |
lines = [l + b'\n' for l in text.split(b'\n')] |
7944
e9b48afd0e78
pure/bdiff: fix circular import
Matt Mackall <mpm@selenic.com>
parents:
7703
diff
changeset
|
17 |
if lines: |
43077
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
18 |
if lines[-1] == b'\n': |
7944
e9b48afd0e78
pure/bdiff: fix circular import
Matt Mackall <mpm@selenic.com>
parents:
7703
diff
changeset
|
19 |
lines.pop() |
e9b48afd0e78
pure/bdiff: fix circular import
Matt Mackall <mpm@selenic.com>
parents:
7703
diff
changeset
|
20 |
else: |
e9b48afd0e78
pure/bdiff: fix circular import
Matt Mackall <mpm@selenic.com>
parents:
7703
diff
changeset
|
21 |
lines[-1] = lines[-1][:-1] |
e9b48afd0e78
pure/bdiff: fix circular import
Matt Mackall <mpm@selenic.com>
parents:
7703
diff
changeset
|
22 |
return lines |
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
23 |
|
43076
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
41269
diff
changeset
|
24 |
|
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
25 |
def _normalizeblocks(a, b, blocks): |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
26 |
prev = None |
14066
14fac6c0536a
pure bdiff: don't use a generator
Dan Villiom Podlaski Christiansen <danchr@gmail.com>
parents:
10282
diff
changeset
|
27 |
r = [] |
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
28 |
for curr in blocks: |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
29 |
if prev is None: |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
30 |
prev = curr |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
31 |
continue |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
32 |
shift = 0 |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
33 |
|
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
34 |
a1, b1, l1 = prev |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
35 |
a1end = a1 + l1 |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
36 |
b1end = b1 + l1 |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
37 |
|
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
38 |
a2, b2, l2 = curr |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
39 |
a2end = a2 + l2 |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
40 |
b2end = b2 + l2 |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
41 |
if a1end == a2: |
43076
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
41269
diff
changeset
|
42 |
while ( |
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
41269
diff
changeset
|
43 |
a1end + shift < a2end and a[a1end + shift] == b[b1end + shift] |
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
41269
diff
changeset
|
44 |
): |
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
45 |
shift += 1 |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
46 |
elif b1end == b2: |
43076
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
41269
diff
changeset
|
47 |
while ( |
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
41269
diff
changeset
|
48 |
b1end + shift < b2end and a[a1end + shift] == b[b1end + shift] |
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
41269
diff
changeset
|
49 |
): |
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
50 |
shift += 1 |
14066
14fac6c0536a
pure bdiff: don't use a generator
Dan Villiom Podlaski Christiansen <danchr@gmail.com>
parents:
10282
diff
changeset
|
51 |
r.append((a1, b1, l1 + shift)) |
10282
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
10263
diff
changeset
|
52 |
prev = a2 + shift, b2 + shift, l2 - shift |
45878
68aedad4c11c
pure: guard against empty blocks
Gregory Szorc <gregory.szorc@gmail.com>
parents:
43077
diff
changeset
|
53 |
|
68aedad4c11c
pure: guard against empty blocks
Gregory Szorc <gregory.szorc@gmail.com>
parents:
43077
diff
changeset
|
54 |
if prev is not None: |
68aedad4c11c
pure: guard against empty blocks
Gregory Szorc <gregory.szorc@gmail.com>
parents:
43077
diff
changeset
|
55 |
r.append(prev) |
68aedad4c11c
pure: guard against empty blocks
Gregory Szorc <gregory.szorc@gmail.com>
parents:
43077
diff
changeset
|
56 |
|
14066
14fac6c0536a
pure bdiff: don't use a generator
Dan Villiom Podlaski Christiansen <danchr@gmail.com>
parents:
10282
diff
changeset
|
57 |
return r |
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
58 |
|
43076
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
41269
diff
changeset
|
59 |
|
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
60 |
def bdiff(a, b): |
31646
f2b334e6c7e0
py3: use bytes() to cast to immutable bytes in pure.bdiff.bdiff()
Yuya Nishihara <yuya@tcha.org>
parents:
31645
diff
changeset
|
61 |
a = bytes(a).splitlines(True) |
f2b334e6c7e0
py3: use bytes() to cast to immutable bytes in pure.bdiff.bdiff()
Yuya Nishihara <yuya@tcha.org>
parents:
31645
diff
changeset
|
62 |
b = bytes(b).splitlines(True) |
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
63 |
|
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
64 |
if not a: |
43077
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
65 |
s = b"".join(b) |
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
66 |
return s and (struct.pack(b">lll", 0, 0, len(s)) + s) |
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
67 |
|
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
68 |
bin = [] |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
69 |
p = [0] |
34436
5326e4ef1dab
style: never put multiple statements on one line
Alex Gaynor <agaynor@mozilla.com>
parents:
32545
diff
changeset
|
70 |
for i in a: |
5326e4ef1dab
style: never put multiple statements on one line
Alex Gaynor <agaynor@mozilla.com>
parents:
32545
diff
changeset
|
71 |
p.append(p[-1] + len(i)) |
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
72 |
|
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
73 |
d = difflib.SequenceMatcher(None, a, b).get_matching_blocks() |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
74 |
d = _normalizeblocks(a, b, d) |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
75 |
la = 0 |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
76 |
lb = 0 |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
77 |
for am, bm, size in d: |
43077
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
78 |
s = b"".join(b[lb:bm]) |
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
79 |
if am > la or s: |
43077
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
80 |
bin.append(struct.pack(b">lll", p[la], p[am], len(s)) + s) |
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
81 |
la = am + size |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
82 |
lb = bm + size |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
83 |
|
43077
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
84 |
return b"".join(bin) |
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
85 |
|
43076
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
41269
diff
changeset
|
86 |
|
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
87 |
def blocks(a, b): |
7944
e9b48afd0e78
pure/bdiff: fix circular import
Matt Mackall <mpm@selenic.com>
parents:
7703
diff
changeset
|
88 |
an = splitnewlines(a) |
e9b48afd0e78
pure/bdiff: fix circular import
Matt Mackall <mpm@selenic.com>
parents:
7703
diff
changeset
|
89 |
bn = splitnewlines(b) |
7703
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
90 |
d = difflib.SequenceMatcher(None, an, bn).get_matching_blocks() |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
91 |
d = _normalizeblocks(an, bn, d) |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
92 |
return [(i, i + n, j, j + n) for (i, j, n) in d] |
9044d3567f6d
pure Python implementation of bdiff.c
Martin Geisler <mg@daimi.au.dk>
parents:
diff
changeset
|
93 |
|
43076
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
41269
diff
changeset
|
94 |
|
15530
eeac5e179243
mdiff: replace wscleanup() regexps with C loops
Patrick Mezard <pmezard@gmail.com>
parents:
14066
diff
changeset
|
95 |
def fixws(text, allws): |
eeac5e179243
mdiff: replace wscleanup() regexps with C loops
Patrick Mezard <pmezard@gmail.com>
parents:
14066
diff
changeset
|
96 |
if allws: |
43077
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
97 |
text = re.sub(b'[ \t\r]+', b'', text) |
15530
eeac5e179243
mdiff: replace wscleanup() regexps with C loops
Patrick Mezard <pmezard@gmail.com>
parents:
14066
diff
changeset
|
98 |
else: |
43077
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
99 |
text = re.sub(b'[ \t\r]+', b' ', text) |
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
100 |
text = text.replace(b' \n', b'\n') |
15530
eeac5e179243
mdiff: replace wscleanup() regexps with C loops
Patrick Mezard <pmezard@gmail.com>
parents:
14066
diff
changeset
|
101 |
return text |