Show More
@@ -1,253 +1,253 | |||
|
1 | 1 | # copies.py - copy detection for Mercurial |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2008 Matt Mackall <mpm@selenic.com> |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | from i18n import _ |
|
9 | 9 | import util |
|
10 | 10 | import heapq |
|
11 | 11 | |
|
12 | 12 | def _nonoverlap(d1, d2, d3): |
|
13 | 13 | "Return list of elements in d1 not in d2 or d3" |
|
14 | 14 | return sorted([d for d in d1 if d not in d3 and d not in d2]) |
|
15 | 15 | |
|
16 | 16 | def _dirname(f): |
|
17 | 17 | s = f.rfind("/") |
|
18 | 18 | if s == -1: |
|
19 | 19 | return "" |
|
20 | 20 | return f[:s] |
|
21 | 21 | |
|
22 | 22 | def _dirs(files): |
|
23 | 23 | d = set() |
|
24 | 24 | for f in files: |
|
25 | 25 | f = _dirname(f) |
|
26 | 26 | while f not in d: |
|
27 | 27 | d.add(f) |
|
28 | 28 | f = _dirname(f) |
|
29 | 29 | return d |
|
30 | 30 | |
|
31 | 31 | def _findlimit(repo, a, b): |
|
32 | 32 | """Find the earliest revision that's an ancestor of a or b but not both, |
|
33 | 33 | None if no such revision exists. |
|
34 | 34 | """ |
|
35 | 35 | # basic idea: |
|
36 | 36 | # - mark a and b with different sides |
|
37 | 37 | # - if a parent's children are all on the same side, the parent is |
|
38 | 38 | # on that side, otherwise it is on no side |
|
39 | 39 | # - walk the graph in topological order with the help of a heap; |
|
40 | 40 | # - add unseen parents to side map |
|
41 | 41 | # - clear side of any parent that has children on different sides |
|
42 | 42 | # - track number of interesting revs that might still be on a side |
|
43 | 43 | # - track the lowest interesting rev seen |
|
44 | 44 | # - quit when interesting revs is zero |
|
45 | 45 | |
|
46 | 46 | cl = repo.changelog |
|
47 | 47 | working = len(cl) # pseudo rev for the working directory |
|
48 | 48 | if a is None: |
|
49 | 49 | a = working |
|
50 | 50 | if b is None: |
|
51 | 51 | b = working |
|
52 | 52 | |
|
53 | 53 | side = {a: -1, b: 1} |
|
54 | 54 | visit = [-a, -b] |
|
55 | 55 | heapq.heapify(visit) |
|
56 | 56 | interesting = len(visit) |
|
57 | 57 | hascommonancestor = False |
|
58 | 58 | limit = working |
|
59 | 59 | |
|
60 | 60 | while interesting: |
|
61 | 61 | r = -heapq.heappop(visit) |
|
62 | 62 | if r == working: |
|
63 | 63 | parents = [cl.rev(p) for p in repo.dirstate.parents()] |
|
64 | 64 | else: |
|
65 | 65 | parents = cl.parentrevs(r) |
|
66 | 66 | for p in parents: |
|
67 | 67 | if p < 0: |
|
68 | 68 | continue |
|
69 | 69 | if p not in side: |
|
70 | 70 | # first time we see p; add it to visit |
|
71 | 71 | side[p] = side[r] |
|
72 | 72 | if side[p]: |
|
73 | 73 | interesting += 1 |
|
74 | 74 | heapq.heappush(visit, -p) |
|
75 | 75 | elif side[p] and side[p] != side[r]: |
|
76 | 76 | # p was interesting but now we know better |
|
77 | 77 | side[p] = 0 |
|
78 | 78 | interesting -= 1 |
|
79 | 79 | hascommonancestor = True |
|
80 | 80 | if side[r]: |
|
81 | 81 | limit = r # lowest rev visited |
|
82 | 82 | interesting -= 1 |
|
83 | 83 | |
|
84 | 84 | if not hascommonancestor: |
|
85 | 85 | return None |
|
86 | 86 | return limit |
|
87 | 87 | |
|
88 | 88 | def copies(repo, c1, c2, ca, checkdirs=False): |
|
89 | 89 | """ |
|
90 | 90 | Find moves and copies between context c1 and c2 |
|
91 | 91 | """ |
|
92 | 92 | # avoid silly behavior for update from empty dir |
|
93 | 93 | if not c1 or not c2 or c1 == c2: |
|
94 | 94 | return {}, {} |
|
95 | 95 | |
|
96 | 96 | # avoid silly behavior for parent -> working dir |
|
97 | 97 | if c2.node() is None and c1.node() == repo.dirstate.parents()[0]: |
|
98 | 98 | return repo.dirstate.copies(), {} |
|
99 | 99 | |
|
100 | 100 | limit = _findlimit(repo, c1.rev(), c2.rev()) |
|
101 | 101 | if limit is None: |
|
102 | 102 | # no common ancestor, no copies |
|
103 | 103 | return {}, {} |
|
104 | 104 | m1 = c1.manifest() |
|
105 | 105 | m2 = c2.manifest() |
|
106 | 106 | ma = ca.manifest() |
|
107 | 107 | |
|
108 | 108 | def makectx(f, n): |
|
109 | 109 | if len(n) != 20: # in a working context? |
|
110 | 110 | if c1.rev() is None: |
|
111 | 111 | return c1.filectx(f) |
|
112 | 112 | return c2.filectx(f) |
|
113 | 113 | return repo.filectx(f, fileid=n) |
|
114 | 114 | |
|
115 | 115 | ctx = util.lrucachefunc(makectx) |
|
116 | 116 | copy = {} |
|
117 | 117 | fullcopy = {} |
|
118 | 118 | diverge = {} |
|
119 | 119 | |
|
120 | 120 | def related(f1, f2, limit): |
|
121 | 121 | g1, g2 = f1.ancestors(), f2.ancestors() |
|
122 | 122 | try: |
|
123 | 123 | while 1: |
|
124 | 124 | f1r, f2r = f1.rev(), f2.rev() |
|
125 | 125 | if f1r > f2r: |
|
126 | 126 | f1 = g1.next() |
|
127 | 127 | elif f2r > f1r: |
|
128 | 128 | f2 = g2.next() |
|
129 | 129 | elif f1 == f2: |
|
130 | 130 | return f1 # a match |
|
131 | 131 | elif f1r == f2r or f1r < limit or f2r < limit: |
|
132 | 132 | return False # copy no longer relevant |
|
133 | 133 | except StopIteration: |
|
134 | 134 | return False |
|
135 | 135 | |
|
136 | 136 | def checkcopies(f, m1, m2): |
|
137 | 137 | '''check possible copies of f from m1 to m2''' |
|
138 | 138 | of = None |
|
139 | 139 | seen = set([f]) |
|
140 | 140 | for oc in ctx(f, m1[f]).ancestors(): |
|
141 | 141 | ocr = oc.rev() |
|
142 | 142 | of = oc.path() |
|
143 | 143 | if of in seen: |
|
144 | 144 | # check limit late - grab last rename before |
|
145 | 145 | if ocr < limit: |
|
146 | 146 | break |
|
147 | 147 | continue |
|
148 | 148 | seen.add(of) |
|
149 | 149 | |
|
150 | 150 | fullcopy[f] = of # remember for dir rename detection |
|
151 | 151 | if of not in m2: |
|
152 | 152 | continue # no match, keep looking |
|
153 | 153 | if m2[of] == ma.get(of): |
|
154 | 154 | break # no merge needed, quit early |
|
155 | 155 | c2 = ctx(of, m2[of]) |
|
156 | 156 | cr = related(oc, c2, ca.rev()) |
|
157 | if of == f or of == c2.path(): # non-divergent | |
|
157 | if cr and (of == f or of == c2.path()): # non-divergent | |
|
158 | 158 | copy[f] = of |
|
159 | 159 | of = None |
|
160 | 160 | break |
|
161 | 161 | |
|
162 | 162 | if of in ma: |
|
163 | 163 | diverge.setdefault(of, []).append(f) |
|
164 | 164 | |
|
165 | 165 | repo.ui.debug(" searching for copies back to rev %d\n" % limit) |
|
166 | 166 | |
|
167 | 167 | u1 = _nonoverlap(m1, m2, ma) |
|
168 | 168 | u2 = _nonoverlap(m2, m1, ma) |
|
169 | 169 | |
|
170 | 170 | if u1: |
|
171 | 171 | repo.ui.debug(" unmatched files in local:\n %s\n" |
|
172 | 172 | % "\n ".join(u1)) |
|
173 | 173 | if u2: |
|
174 | 174 | repo.ui.debug(" unmatched files in other:\n %s\n" |
|
175 | 175 | % "\n ".join(u2)) |
|
176 | 176 | |
|
177 | 177 | for f in u1: |
|
178 | 178 | checkcopies(f, m1, m2) |
|
179 | 179 | for f in u2: |
|
180 | 180 | checkcopies(f, m2, m1) |
|
181 | 181 | |
|
182 | 182 | diverge2 = set() |
|
183 | 183 | for of, fl in diverge.items(): |
|
184 | 184 | if len(fl) == 1: |
|
185 | 185 | del diverge[of] # not actually divergent |
|
186 | 186 | else: |
|
187 | 187 | diverge2.update(fl) # reverse map for below |
|
188 | 188 | |
|
189 | 189 | if fullcopy: |
|
190 | 190 | repo.ui.debug(" all copies found (* = to merge, ! = divergent):\n") |
|
191 | 191 | for f in fullcopy: |
|
192 | 192 | note = "" |
|
193 | 193 | if f in copy: |
|
194 | 194 | note += "*" |
|
195 | 195 | if f in diverge2: |
|
196 | 196 | note += "!" |
|
197 | 197 | repo.ui.debug(" %s -> %s %s\n" % (f, fullcopy[f], note)) |
|
198 | 198 | del diverge2 |
|
199 | 199 | |
|
200 | 200 | if not fullcopy or not checkdirs: |
|
201 | 201 | return copy, diverge |
|
202 | 202 | |
|
203 | 203 | repo.ui.debug(" checking for directory renames\n") |
|
204 | 204 | |
|
205 | 205 | # generate a directory move map |
|
206 | 206 | d1, d2 = _dirs(m1), _dirs(m2) |
|
207 | 207 | invalid = set() |
|
208 | 208 | dirmove = {} |
|
209 | 209 | |
|
210 | 210 | # examine each file copy for a potential directory move, which is |
|
211 | 211 | # when all the files in a directory are moved to a new directory |
|
212 | 212 | for dst, src in fullcopy.iteritems(): |
|
213 | 213 | dsrc, ddst = _dirname(src), _dirname(dst) |
|
214 | 214 | if dsrc in invalid: |
|
215 | 215 | # already seen to be uninteresting |
|
216 | 216 | continue |
|
217 | 217 | elif dsrc in d1 and ddst in d1: |
|
218 | 218 | # directory wasn't entirely moved locally |
|
219 | 219 | invalid.add(dsrc) |
|
220 | 220 | elif dsrc in d2 and ddst in d2: |
|
221 | 221 | # directory wasn't entirely moved remotely |
|
222 | 222 | invalid.add(dsrc) |
|
223 | 223 | elif dsrc in dirmove and dirmove[dsrc] != ddst: |
|
224 | 224 | # files from the same directory moved to two different places |
|
225 | 225 | invalid.add(dsrc) |
|
226 | 226 | else: |
|
227 | 227 | # looks good so far |
|
228 | 228 | dirmove[dsrc + "/"] = ddst + "/" |
|
229 | 229 | |
|
230 | 230 | for i in invalid: |
|
231 | 231 | if i in dirmove: |
|
232 | 232 | del dirmove[i] |
|
233 | 233 | del d1, d2, invalid |
|
234 | 234 | |
|
235 | 235 | if not dirmove: |
|
236 | 236 | return copy, diverge |
|
237 | 237 | |
|
238 | 238 | for d in dirmove: |
|
239 | 239 | repo.ui.debug(" dir %s -> %s\n" % (d, dirmove[d])) |
|
240 | 240 | |
|
241 | 241 | # check unaccounted nonoverlapping files against directory moves |
|
242 | 242 | for f in u1 + u2: |
|
243 | 243 | if f not in fullcopy: |
|
244 | 244 | for d in dirmove: |
|
245 | 245 | if f.startswith(d): |
|
246 | 246 | # new file added in a directory that was moved, move it |
|
247 | 247 | df = dirmove[d] + f[len(d):] |
|
248 | 248 | if df not in copy: |
|
249 | 249 | copy[f] = df |
|
250 | 250 | repo.ui.debug(" file %s -> %s\n" % (f, copy[f])) |
|
251 | 251 | break |
|
252 | 252 | |
|
253 | 253 | return copy, diverge |
General Comments 0
You need to be logged in to leave comments.
Login now