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