Show More
@@ -1,252 +1,257 b'' | |||
|
1 | 1 | # mdiff.py - diff and patch routines for mercurial |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2005, 2006 Matt Mackall <mpm@selenic.com> |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms |
|
6 | 6 | # of the GNU General Public License, incorporated herein by reference. |
|
7 | 7 | |
|
8 | 8 | from demandload import demandload |
|
9 | 9 | import bdiff, mpatch |
|
10 | demandload(globals(), "re struct util") | |
|
10 | demandload(globals(), "re struct util md5") | |
|
11 | 11 | |
|
12 | 12 | def splitnewlines(text): |
|
13 | 13 | '''like str.splitlines, but only split on newlines.''' |
|
14 | 14 | lines = [l + '\n' for l in text.split('\n')] |
|
15 | 15 | if lines: |
|
16 | 16 | if lines[-1] == '\n': |
|
17 | 17 | lines.pop() |
|
18 | 18 | else: |
|
19 | 19 | lines[-1] = lines[-1][:-1] |
|
20 | 20 | return lines |
|
21 | 21 | |
|
22 | 22 | class diffopts(object): |
|
23 | 23 | '''context is the number of context lines |
|
24 | 24 | text treats all files as text |
|
25 | 25 | showfunc enables diff -p output |
|
26 | 26 | git enables the git extended patch format |
|
27 | 27 | nodates removes dates from diff headers |
|
28 | 28 | ignorews ignores all whitespace changes in the diff |
|
29 | 29 | ignorewsamount ignores changes in the amount of whitespace |
|
30 | 30 | ignoreblanklines ignores changes whose lines are all blank''' |
|
31 | 31 | |
|
32 | 32 | defaults = { |
|
33 | 33 | 'context': 3, |
|
34 | 34 | 'text': False, |
|
35 | 35 | 'showfunc': True, |
|
36 | 36 | 'git': False, |
|
37 | 37 | 'nodates': False, |
|
38 | 38 | 'ignorews': False, |
|
39 | 39 | 'ignorewsamount': False, |
|
40 | 40 | 'ignoreblanklines': False, |
|
41 | 41 | } |
|
42 | 42 | |
|
43 | 43 | __slots__ = defaults.keys() |
|
44 | 44 | |
|
45 | 45 | def __init__(self, **opts): |
|
46 | 46 | for k in self.__slots__: |
|
47 | 47 | v = opts.get(k) |
|
48 | 48 | if v is None: |
|
49 | 49 | v = self.defaults[k] |
|
50 | 50 | setattr(self, k, v) |
|
51 | 51 | |
|
52 | 52 | defaultopts = diffopts() |
|
53 | 53 | |
|
54 | 54 | def unidiff(a, ad, b, bd, fn, r=None, opts=defaultopts): |
|
55 | 55 | def datetag(date): |
|
56 | 56 | return (opts.git or opts.nodates) and '\n' or '\t%s\n' % date |
|
57 | 57 | |
|
58 | 58 | if not a and not b: return "" |
|
59 | 59 | epoch = util.datestr((0, 0)) |
|
60 | 60 | |
|
61 | 61 | if not opts.text and (util.binary(a) or util.binary(b)): |
|
62 | def h(v): | |
|
63 | # md5 is used instead of sha1 because md5 is supposedly faster | |
|
64 | return md5.new(v).digest() | |
|
65 | if a and b and len(a) == len(b) and h(a) == h(b): | |
|
66 | return "" | |
|
62 | 67 | l = ['Binary file %s has changed\n' % fn] |
|
63 | 68 | elif not a: |
|
64 | 69 | b = splitnewlines(b) |
|
65 | 70 | if a is None: |
|
66 | 71 | l1 = '--- /dev/null%s' % datetag(epoch) |
|
67 | 72 | else: |
|
68 | 73 | l1 = "--- %s%s" % ("a/" + fn, datetag(ad)) |
|
69 | 74 | l2 = "+++ %s%s" % ("b/" + fn, datetag(bd)) |
|
70 | 75 | l3 = "@@ -0,0 +1,%d @@\n" % len(b) |
|
71 | 76 | l = [l1, l2, l3] + ["+" + e for e in b] |
|
72 | 77 | elif not b: |
|
73 | 78 | a = splitnewlines(a) |
|
74 | 79 | l1 = "--- %s%s" % ("a/" + fn, datetag(ad)) |
|
75 | 80 | if b is None: |
|
76 | 81 | l2 = '+++ /dev/null%s' % datetag(epoch) |
|
77 | 82 | else: |
|
78 | 83 | l2 = "+++ %s%s" % ("b/" + fn, datetag(bd)) |
|
79 | 84 | l3 = "@@ -1,%d +0,0 @@\n" % len(a) |
|
80 | 85 | l = [l1, l2, l3] + ["-" + e for e in a] |
|
81 | 86 | else: |
|
82 | 87 | al = splitnewlines(a) |
|
83 | 88 | bl = splitnewlines(b) |
|
84 | 89 | l = list(bunidiff(a, b, al, bl, "a/" + fn, "b/" + fn, opts=opts)) |
|
85 | 90 | if not l: return "" |
|
86 | 91 | # difflib uses a space, rather than a tab |
|
87 | 92 | l[0] = "%s%s" % (l[0][:-2], datetag(ad)) |
|
88 | 93 | l[1] = "%s%s" % (l[1][:-2], datetag(bd)) |
|
89 | 94 | |
|
90 | 95 | for ln in xrange(len(l)): |
|
91 | 96 | if l[ln][-1] != '\n': |
|
92 | 97 | l[ln] += "\n\ No newline at end of file\n" |
|
93 | 98 | |
|
94 | 99 | if r: |
|
95 | 100 | l.insert(0, "diff %s %s\n" % |
|
96 | 101 | (' '.join(["-r %s" % rev for rev in r]), fn)) |
|
97 | 102 | |
|
98 | 103 | return "".join(l) |
|
99 | 104 | |
|
100 | 105 | # somewhat self contained replacement for difflib.unified_diff |
|
101 | 106 | # t1 and t2 are the text to be diffed |
|
102 | 107 | # l1 and l2 are the text broken up into lines |
|
103 | 108 | # header1 and header2 are the filenames for the diff output |
|
104 | 109 | def bunidiff(t1, t2, l1, l2, header1, header2, opts=defaultopts): |
|
105 | 110 | def contextend(l, len): |
|
106 | 111 | ret = l + opts.context |
|
107 | 112 | if ret > len: |
|
108 | 113 | ret = len |
|
109 | 114 | return ret |
|
110 | 115 | |
|
111 | 116 | def contextstart(l): |
|
112 | 117 | ret = l - opts.context |
|
113 | 118 | if ret < 0: |
|
114 | 119 | return 0 |
|
115 | 120 | return ret |
|
116 | 121 | |
|
117 | 122 | def yieldhunk(hunk, header): |
|
118 | 123 | if header: |
|
119 | 124 | for x in header: |
|
120 | 125 | yield x |
|
121 | 126 | (astart, a2, bstart, b2, delta) = hunk |
|
122 | 127 | aend = contextend(a2, len(l1)) |
|
123 | 128 | alen = aend - astart |
|
124 | 129 | blen = b2 - bstart + aend - a2 |
|
125 | 130 | |
|
126 | 131 | func = "" |
|
127 | 132 | if opts.showfunc: |
|
128 | 133 | # walk backwards from the start of the context |
|
129 | 134 | # to find a line starting with an alphanumeric char. |
|
130 | 135 | for x in xrange(astart, -1, -1): |
|
131 | 136 | t = l1[x].rstrip() |
|
132 | 137 | if funcre.match(t): |
|
133 | 138 | func = ' ' + t[:40] |
|
134 | 139 | break |
|
135 | 140 | |
|
136 | 141 | yield "@@ -%d,%d +%d,%d @@%s\n" % (astart + 1, alen, |
|
137 | 142 | bstart + 1, blen, func) |
|
138 | 143 | for x in delta: |
|
139 | 144 | yield x |
|
140 | 145 | for x in xrange(a2, aend): |
|
141 | 146 | yield ' ' + l1[x] |
|
142 | 147 | |
|
143 | 148 | header = [ "--- %s\t\n" % header1, "+++ %s\t\n" % header2 ] |
|
144 | 149 | |
|
145 | 150 | if opts.showfunc: |
|
146 | 151 | funcre = re.compile('\w') |
|
147 | 152 | if opts.ignorewsamount: |
|
148 | 153 | wsamountre = re.compile('[ \t]+') |
|
149 | 154 | wsappendedre = re.compile(' \n') |
|
150 | 155 | if opts.ignoreblanklines: |
|
151 | 156 | wsblanklinesre = re.compile('\n') |
|
152 | 157 | if opts.ignorews: |
|
153 | 158 | wsre = re.compile('[ \t]') |
|
154 | 159 | |
|
155 | 160 | # bdiff.blocks gives us the matching sequences in the files. The loop |
|
156 | 161 | # below finds the spaces between those matching sequences and translates |
|
157 | 162 | # them into diff output. |
|
158 | 163 | # |
|
159 | 164 | diff = bdiff.blocks(t1, t2) |
|
160 | 165 | hunk = None |
|
161 | 166 | for i in xrange(len(diff)): |
|
162 | 167 | # The first match is special. |
|
163 | 168 | # we've either found a match starting at line 0 or a match later |
|
164 | 169 | # in the file. If it starts later, old and new below will both be |
|
165 | 170 | # empty and we'll continue to the next match. |
|
166 | 171 | if i > 0: |
|
167 | 172 | s = diff[i-1] |
|
168 | 173 | else: |
|
169 | 174 | s = [0, 0, 0, 0] |
|
170 | 175 | delta = [] |
|
171 | 176 | s1 = diff[i] |
|
172 | 177 | a1 = s[1] |
|
173 | 178 | a2 = s1[0] |
|
174 | 179 | b1 = s[3] |
|
175 | 180 | b2 = s1[2] |
|
176 | 181 | |
|
177 | 182 | old = l1[a1:a2] |
|
178 | 183 | new = l2[b1:b2] |
|
179 | 184 | |
|
180 | 185 | # bdiff sometimes gives huge matches past eof, this check eats them, |
|
181 | 186 | # and deals with the special first match case described above |
|
182 | 187 | if not old and not new: |
|
183 | 188 | continue |
|
184 | 189 | |
|
185 | 190 | if opts.ignoreblanklines: |
|
186 | 191 | wsold = wsblanklinesre.sub('', "".join(old)) |
|
187 | 192 | wsnew = wsblanklinesre.sub('', "".join(new)) |
|
188 | 193 | if wsold == wsnew: |
|
189 | 194 | continue |
|
190 | 195 | |
|
191 | 196 | if opts.ignorewsamount: |
|
192 | 197 | wsold = wsamountre.sub(' ', "".join(old)) |
|
193 | 198 | wsold = wsappendedre.sub('\n', wsold) |
|
194 | 199 | wsnew = wsamountre.sub(' ', "".join(new)) |
|
195 | 200 | wsnew = wsappendedre.sub('\n', wsnew) |
|
196 | 201 | if wsold == wsnew: |
|
197 | 202 | continue |
|
198 | 203 | |
|
199 | 204 | if opts.ignorews: |
|
200 | 205 | wsold = wsre.sub('', "".join(old)) |
|
201 | 206 | wsnew = wsre.sub('', "".join(new)) |
|
202 | 207 | if wsold == wsnew: |
|
203 | 208 | continue |
|
204 | 209 | |
|
205 | 210 | astart = contextstart(a1) |
|
206 | 211 | bstart = contextstart(b1) |
|
207 | 212 | prev = None |
|
208 | 213 | if hunk: |
|
209 | 214 | # join with the previous hunk if it falls inside the context |
|
210 | 215 | if astart < hunk[1] + opts.context + 1: |
|
211 | 216 | prev = hunk |
|
212 | 217 | astart = hunk[1] |
|
213 | 218 | bstart = hunk[3] |
|
214 | 219 | else: |
|
215 | 220 | for x in yieldhunk(hunk, header): |
|
216 | 221 | yield x |
|
217 | 222 | # we only want to yield the header if the files differ, and |
|
218 | 223 | # we only want to yield it once. |
|
219 | 224 | header = None |
|
220 | 225 | if prev: |
|
221 | 226 | # we've joined the previous hunk, record the new ending points. |
|
222 | 227 | hunk[1] = a2 |
|
223 | 228 | hunk[3] = b2 |
|
224 | 229 | delta = hunk[4] |
|
225 | 230 | else: |
|
226 | 231 | # create a new hunk |
|
227 | 232 | hunk = [ astart, a2, bstart, b2, delta ] |
|
228 | 233 | |
|
229 | 234 | delta[len(delta):] = [ ' ' + x for x in l1[astart:a1] ] |
|
230 | 235 | delta[len(delta):] = [ '-' + x for x in old ] |
|
231 | 236 | delta[len(delta):] = [ '+' + x for x in new ] |
|
232 | 237 | |
|
233 | 238 | if hunk: |
|
234 | 239 | for x in yieldhunk(hunk, header): |
|
235 | 240 | yield x |
|
236 | 241 | |
|
237 | 242 | def patchtext(bin): |
|
238 | 243 | pos = 0 |
|
239 | 244 | t = [] |
|
240 | 245 | while pos < len(bin): |
|
241 | 246 | p1, p2, l = struct.unpack(">lll", bin[pos:pos + 12]) |
|
242 | 247 | pos += 12 |
|
243 | 248 | t.append(bin[pos:pos + l]) |
|
244 | 249 | pos += l |
|
245 | 250 | return "".join(t) |
|
246 | 251 | |
|
247 | 252 | def patch(a, bin): |
|
248 | 253 | return mpatch.patches(a, [bin]) |
|
249 | 254 | |
|
250 | 255 | patches = mpatch.patches |
|
251 | 256 | patchedsize = mpatch.patchedsize |
|
252 | 257 | textdiff = bdiff.bdiff |
General Comments 0
You need to be logged in to leave comments.
Login now