2 Replicate (replay) changesets from one SVN repository to another.
5 from .. import base_version
, full_version
7 from .. import svnclient
8 from ..shell
import run_svn
9 from ..errors
import (ExternalCommandFailed
, UnsupportedSVNAction
, InternalError
, VerificationError
)
10 from parse
import HelpFormatter
20 from datetime
import datetime
22 _valid_svn_actions
= "MARD" # The list of known SVN action abbr's, from "svn log"
24 # Module-level variables/parameters
25 source_url
= "" # URL to source path in source SVN repo, e.g. 'http://server/svn/source/trunk'
26 source_repos_url
= "" # URL to root of source SVN repo, e.g. 'http://server/svn/source'
27 source_base
= "" # Relative path of source_url in source SVN repo, e.g. '/trunk'
28 source_repos_uuid
= "" # UUID of source SVN repo
29 target_url
="" # URL to target path in target SVN repo, e.g. 'file:///svn/repo_target/trunk'
30 rev_map
= {} # The running mapping-table dictionary for source_url rev #'s -> target_url rev #'s
31 options
= None # optparser options
33 def parse_svn_commit_rev(output
):
35 Parse the revision number from the output of "svn commit".
37 output_lines
= output
.strip("\n").split("\n")
39 for line
in output_lines
:
40 if line
[0:19] == 'Committed revision ':
41 rev_num
= line
[19:].rstrip('.')
43 assert rev_num
is not None
46 def commit_from_svn_log_entry(log_entry
, commit_paths
=None, target_revprops
=None):
48 Given an SVN log entry and an optional list of changed paths, do an svn commit.
50 # TODO: Run optional external shell hook here, for doing pre-commit filtering
51 # Display the _wc_target "svn status" info if running in -vv (or higher) mode
52 if ui
.get_level() >= ui
.EXTRA
:
53 ui
.status(">> commit_from_svn_log_entry: Pre-commit _wc_target status:", level
=ui
.EXTRA
, color
='CYAN')
54 ui
.status(run_svn(["status"]), level
=ui
.EXTRA
, color
='CYAN')
55 # This will use the local timezone for displaying commit times
56 timestamp
= int(log_entry
['date'])
57 svn_date
= str(datetime
.fromtimestamp(timestamp
))
58 # Uncomment this one one if you prefer UTC commit times
59 #svn_date = "%d 0" % timestamp
60 args
= ["commit", "--force-log"]
61 message
= log_entry
['message']
63 message
+= "\nDate: " + svn_date
64 if options
.log_author
:
65 message
+= "\nAuthor: " + log_entry
['author']
66 if options
.keep_author
:
67 args
+= ["--username", log_entry
['author']]
68 args
+= ["-m", message
]
70 if log_entry
['revprops']:
71 # Carry forward any revprop's from the source revision
72 for v
in log_entry
['revprops']:
73 revprops
[v
['name']] = v
['value']
75 # Add any extra revprop's we want to set for the target repo commits
76 for v
in target_revprops
:
77 revprops
[v
['name']] = v
['value']
80 args
+= ["--with-revprop", "%s=%s" % (key
, str(revprops
[key
]))]
82 if len(commit_paths
)<100:
83 # If we don't have an excessive amount of individual changed paths, pass
84 # those to the "svn commit" command. Else, pass nothing so we commit at
85 # the root of the working-copy.
86 args
+= list(commit_paths
)
88 if not options
.dry_run
:
89 # Run the "svn commit" command, and screen-scrape the target_rev value (if any)
90 output
= run_svn(args
)
91 rev_num
= parse_svn_commit_rev(output
) if output
else None
92 if rev_num
is not None:
93 ui
.status("Committed revision %s.", rev_num
)
95 run_svn(["propset", "--revprop", "-r", rev_num
, "svn:date", log_entry
['date_raw']])
98 def full_svn_revert():
100 Do an "svn revert" and proactively remove any extra files in the working copy.
102 run_svn(["revert", "--recursive", "."])
103 output
= run_svn(["status"])
105 output_lines
= output
.strip("\n").split("\n")
106 for line
in output_lines
:
108 path
= line
[4:].strip(" ")
109 if os
.path
.isfile(path
):
111 if os
.path
.isdir(path
):
114 def gen_tracking_revprops(source_rev
):
116 Build an array of svn2svn-specific source-tracking revprops.
118 revprops
= [{'name':'svn2svn:source_uuid', 'value':source_repos_uuid}
,
119 {'name':'svn2svn:source_url', 'value':source_url}
,
120 {'name':'svn2svn:source_rev', 'value':source_rev}
]
123 def sync_svn_props(source_url
, source_rev
, path_offset
):
125 Carry-forward any unversioned properties from the source repo to the
128 source_props
= svnclient
.get_all_props(source_url
+"/"+path_offset
, source_rev
)
129 target_props
= svnclient
.get_all_props(path_offset
)
130 if 'svn:mergeinfo' in source_props
:
131 # Never carry-forward "svn:mergeinfo"
132 del source_props
['svn:mergeinfo']
133 for prop
in target_props
:
134 if prop
not in source_props
:
135 # Remove any properties which exist in target but not source
136 run_svn(["propdel", prop
, path_offset
])
137 for prop
in source_props
:
138 if prop
not in target_props
or \
139 source_props
[prop
] != target_props
[prop
]:
140 # Set/update any properties which exist in source but not target or
141 # whose value differs between source vs. target.
142 run_svn(["propset", prop
, source_props
[prop
], path_offset
])
144 def in_svn(p
, require_in_repo
=False, prefix
=""):
146 Check if a given file/folder is being tracked by Subversion.
147 Prior to SVN 1.6, we could "cheat" and look for the existence of ".svn" directories.
148 With SVN 1.7 and beyond, WC-NG means only a single top-level ".svn" at the root of the working-copy.
149 Use "svn status" to check the status of the file/folder.
151 entries
= svnclient
.get_svn_status(p
, no_recursive
=True)
155 if require_in_repo
and (d
['status'] == 'added' or d
['revision'] is None):
156 # If caller requires this path to be in the SVN repo, prevent returning True
157 # for paths that are only locally-added.
160 # Don't consider files tracked as deleted in the WC as under source-control.
161 # Consider files which are locally added/copied as under source-control.
162 ret
= True if not (d
['status'] == 'deleted') and (d
['type'] == 'normal' or d
['status'] == 'added' or d
['copied'] == 'true') else False
163 ui
.status(prefix
+ ">> in_svn('%s', require_in_repo=%s) --> %s", p
, str(require_in_repo
), str(ret
), level
=ui
.DEBUG
, color
='GREEN')
166 def is_child_path(path
, p_path
):
167 return True if (path
== p_path
) or (path
.startswith(p_path
+"/")) else False
169 def find_svn_ancestors(svn_repos_url
, base_path
, source_path
, source_rev
, prefix
= ""):
171 Given a source path, walk the SVN history backwards to inspect the ancestory of
172 that path, seeing if it traces back to base_path. Build an array of copyfrom_path
173 and copyfrom_revision pairs for each of the "svn copies". If we find a copyfrom_path
174 which base_path is a substring match of (e.g. we crawled back to the initial branch-
175 copy from trunk), then return the collection of ancestor paths. Otherwise,
176 copyfrom_path has no ancestory compared to base_path.
178 This is useful when comparing "trunk" vs. "branch" paths, to handle cases where a
179 file/folder was renamed in a branch and then that branch was merged back to trunk.
181 'svn_repos_url' is the full URL to the root of the SVN repository,
182 e.g. 'file:///path/to/repo'
183 'base_path' is the path in the SVN repo to the target path we're trying to
184 trace ancestry back to, e.g. '/trunk'.
185 'source_path' is the path in the SVN repo to the source path to start checking
186 ancestry at, e.g. '/branches/fix1/projectA/file1.txt'.
187 (full_path = svn_repos_url+base_path+"/"+path_offset)
188 'source_rev' is the revision to start walking the history of source_path backwards from.
190 ui
.status(prefix
+ ">> find_svn_ancestors: Start: (%s) source_path: %s base_path: %s",
191 svn_repos_url
, source_path
+"@"+str(source_rev
), base_path
, level
=ui
.DEBUG
, color
='YELLOW')
193 working_path
= base_path
+"/"+source_path
194 working_rev
= source_rev
195 first_iter_done
= False
198 # Get the first "svn log" entry for this path (relative to @rev)
199 ui
.status(prefix
+ ">> find_svn_ancestors: %s", svn_repos_url
+ working_path
+"@"+str(working_rev
), level
=ui
.DEBUG
, color
='YELLOW')
200 log_entry
= svnclient
.get_first_svn_log_entry(svn_repos_url
+ working_path
, 1, working_rev
, True)
202 ui
.status(prefix
+ ">> find_svn_ancestors: Done: no log_entry", level
=ui
.DEBUG
, color
='YELLOW')
205 # If we found a copy-from case which matches our base_path, we're done.
206 # ...but only if we've at least tried to search for the first copy-from path.
207 if first_iter_done
and is_child_path(working_path
, base_path
):
208 ui
.status(prefix
+ ">> find_svn_ancestors: Done: Found is_child_path(working_path, base_path) and first_iter_done=True", level
=ui
.DEBUG
, color
='YELLOW')
211 first_iter_done
= True
212 # Search for any actions on our target path (or parent paths).
213 changed_paths_temp
= []
214 for d
in log_entry
['changed_paths']:
216 if is_child_path(working_path
, path
):
217 changed_paths_temp
.append({'path': path, 'data': d}
)
218 if not changed_paths_temp
:
219 # If no matches, then we've hit the end of the chain and this path has no ancestry back to base_path.
220 ui
.status(prefix
+ ">> find_svn_ancestors: Done: No matching changed_paths", level
=ui
.DEBUG
, color
='YELLOW')
223 # Reverse-sort any matches, so that we start with the most-granular (deepest in the tree) path.
224 changed_paths
= sorted(changed_paths_temp
, key
=operator
.itemgetter('path'), reverse
=True)
225 # Find the action for our working_path in this revision. Use a loop to check in reverse order,
226 # so that if the target file/folder is "M" but has a parent folder with an "A" copy-from.
227 for v
in changed_paths
:
230 # Check action-type for this file
232 if action
not in _valid_svn_actions
:
233 raise UnsupportedSVNAction("In SVN rev. %d: action '%s' not supported. Please report a bug!"
234 % (log_entry
['revision'], action
))
235 ui
.status(prefix
+ "> %s %s%s", action
, path
,
236 (" (from %s)" % (d
['copyfrom_path']+"@"+str(d
['copyfrom_revision']))) if d
['copyfrom_path'] else "",
237 level
=ui
.DEBUG
, color
='YELLOW')
239 # If file/folder was deleted, it has no ancestor
241 ui
.status(prefix
+ ">> find_svn_ancestors: Done: deleted", level
=ui
.DEBUG
, color
='YELLOW')
245 # If file/folder was added/replaced but not a copy, it has no ancestor
246 if not d
['copyfrom_path']:
248 ui
.status(prefix
+ ">> find_svn_ancestors: Done: %s with no copyfrom_path",
249 "Added" if action
== "A" else "Replaced",
250 level
=ui
.DEBUG
, color
='YELLOW')
253 # Else, file/folder was added/replaced and is a copy, so add an entry to our ancestors list
254 # and keep checking for ancestors
255 ui
.status(prefix
+ ">> find_svn_ancestors: Found copy-from (action=%s): %s --> %s",
256 action
, path
, d
['copyfrom_path']+"@"+str(d
['copyfrom_revision']),
257 level
=ui
.DEBUG
, color
='YELLOW')
258 ancestors_temp
.append({'path': path
, 'revision': log_entry
['revision'],
259 'copyfrom_path': d
['copyfrom_path'], 'copyfrom_rev': d
['copyfrom_revision']})
260 working_path
= working_path
.replace(d
['path'], d
['copyfrom_path'])
261 working_rev
= d
['copyfrom_revision']
262 # Follow the copy and keep on searching
266 ancestors
.append({'path': base_path+"/"+source_path, 'revision': source_rev}
)
267 working_path
= base_path
+"/"+source_path
268 for idx
in range(len(ancestors_temp
)):
269 d
= ancestors_temp
[idx
]
270 working_path
= working_path
.replace(d
['path'], d
['copyfrom_path'])
271 working_rev
= d
['copyfrom_rev']
272 ancestors
.append({'path': working_path, 'revision': working_rev}
)
273 if ui
.get_level() >= ui
.DEBUG
:
275 for idx
in range(len(ancestors
)):
277 max_len
= max(max_len
, len(d
['path']+"@"+str(d
['revision'])))
278 ui
.status(prefix
+ ">> find_svn_ancestors: Found parent ancestors:", level
=ui
.DEBUG
, color
='YELLOW_B')
279 for idx
in range(len(ancestors
)-1):
281 d_next
= ancestors
[idx
+1]
282 ui
.status(prefix
+ " [%s] %s <-- %s", idx
,
283 str(d
['path']+"@"+str(d
['revision'])).ljust(max_len
),
284 str(d_next
['path']+"@"+str(d_next
['revision'])).ljust(max_len
),
285 level
=ui
.DEBUG
, color
='YELLOW')
287 ui
.status(prefix
+ ">> find_svn_ancestors: No ancestor-chain found: %s",
288 svn_repos_url
+base_path
+"/"+source_path
+"@"+str(source_rev
), level
=ui
.DEBUG
, color
='YELLOW')
291 def get_rev_map(source_rev
, prefix
):
293 Find the equivalent rev # in the target repo for the given rev # from the source repo.
295 ui
.status(prefix
+ ">> get_rev_map(%s)", source_rev
, level
=ui
.DEBUG
, color
='GREEN')
296 # Find the highest entry less-than-or-equal-to source_rev
297 for rev
in range(int(source_rev
), 0, -1):
298 in_rev_map
= True if rev
in rev_map
else False
299 ui
.status(prefix
+ ">> get_rev_map: rev=%s in_rev_map=%s", rev
, str(in_rev_map
), level
=ui
.DEBUG
, color
='BLACK_B')
301 return int(rev_map
[rev
])
302 # Else, we fell off the bottom of the rev_map. Ruh-roh...
305 def set_rev_map(source_rev
, target_rev
):
306 ui
.status(">> set_rev_map: source_rev=%s target_rev=%s", source_rev
, target_rev
, level
=ui
.DEBUG
, color
='GREEN')
308 rev_map
[int(source_rev
)]=int(target_rev
)
310 def build_rev_map(target_url
, target_end_rev
, source_info
):
312 Check for any already-replayed history from source_url (source_info) and
313 build the mapping-table of source_rev -> target_rev.
317 ui
.status("Rebuilding target_rev -> source_rev rev_map...", level
=ui
.VERBOSE
)
319 it_log_entries
= svnclient
.iter_svn_log_entries(target_url
, 1, target_end_rev
, get_changed_paths
=False, get_revprops
=True)
320 for log_entry
in it_log_entries
:
321 if log_entry
['revprops']:
323 for v
in log_entry
['revprops']:
324 if v
['name'].startswith('svn2svn:'):
325 revprops
[v
['name']] = v
['value']
327 revprops
['svn2svn:source_uuid'] == source_info
['repos_uuid'] and \
328 revprops
['svn2svn:source_url'] == source_info
['url']:
329 source_rev
= revprops
['svn2svn:source_rev']
330 target_rev
= log_entry
['revision']
331 set_rev_map(source_rev
, target_rev
)
333 if proc_count
% 500 == 0:
334 ui
.status("...processed %s (%s of %s)..." % (proc_count
, target_rev
, target_end_rev
), level
=ui
.VERBOSE
)
336 def get_svn_dirlist(svn_path
, rev_number
= ""):
338 Get a list of all the child contents (recusive) of the given folder path.
343 args
+= ["-r", rev_number
]
344 path
+= "@"+str(rev_number
)
346 paths
= run_svn(args
, no_fail
=True)
347 paths
= paths
.strip("\n").split("\n") if len(paths
)>1 else []
350 def path_in_list(paths
, path
):
352 if is_child_path(path
, p
):
356 def add_path(paths
, path
):
357 if not path_in_list(paths
, path
):
360 def do_svn_add(path_offset
, source_rev
, parent_copyfrom_path
="", parent_copyfrom_rev
="", \
361 export_paths
={}, is_dir
= False, skip_paths
=[], prefix
= ""):
363 Given the add'd source path, replay the "svn add/copy" commands to correctly
364 track renames across copy-from's.
366 For example, consider a sequence of events like this:
367 1. svn copy /trunk /branches/fix1
368 2. (Make some changes on /branches/fix1)
369 3. svn mv /branches/fix1/Proj1 /branches/fix1/Proj2 " Rename folder
370 4. svn mv /branches/fix1/Proj2/file1.txt /branches/fix1/Proj2/file2.txt " Rename file inside renamed folder
371 5. svn co /trunk && svn merge /branches/fix1
372 After the merge and commit, "svn log -v" with show a delete of /trunk/Proj1
373 and and add of /trunk/Proj2 copy-from /branches/fix1/Proj2. If we were just
374 to do a straight "svn export+add" based on the /branches/fix1/Proj2 folder,
375 we'd lose the logical history that Proj2/file2.txt is really a descendant
378 'path_offset' is the offset from source_base to the file to check ancestry for,
379 e.g. 'projectA/file1.txt'. path = source_repos_url + source_base + path_offset.
380 'source_rev' is the revision ("svn log") that we're processing from the source repo.
381 'parent_copyfrom_path' and 'parent_copyfrom_rev' is the copy-from path of the parent
382 directory, when being called recursively by do_svn_add_dir().
383 'export_paths' is the list of path_offset's that we've deferred running "svn export" on.
384 'is_dir' is whether path_offset is a directory (rather than a file).
386 ui
.status(prefix
+ ">> do_svn_add: %s %s", source_base
+"/"+path_offset
+"@"+str(source_rev
),
387 " (parent-copyfrom: "+parent_copyfrom_path
+"@"+str(parent_copyfrom_rev
)+")" if parent_copyfrom_path
else "",
388 level
=ui
.DEBUG
, color
='GREEN')
389 # Check if the given path has ancestors which chain back to the current source_base
390 found_ancestor
= False
391 ancestors
= find_svn_ancestors(source_repos_url
, source_base
, path_offset
, source_rev
, prefix
+" ")
392 # ancestors[n] is the original (pre-branch-copy) trunk path.
393 # ancestors[n-1] is the first commit on the new branch.
394 copyfrom_path
= ancestors
[len(ancestors
)-1]['path'] if ancestors
else ""
395 copyfrom_rev
= ancestors
[len(ancestors
)-1]['revision'] if ancestors
else ""
397 # The copy-from path has ancestory back to source_url.
398 ui
.status(prefix
+ ">> do_svn_add: Check copy-from: Found parent: %s", copyfrom_path
+"@"+str(copyfrom_rev
),
399 level
=ui
.DEBUG
, color
='GREEN', bold
=True)
400 found_ancestor
= True
401 # Map the copyfrom_rev (source repo) to the equivalent target repo rev #. This can
402 # return None in the case where copyfrom_rev is *before* our source_start_rev.
403 tgt_rev
= get_rev_map(copyfrom_rev
, prefix
+" ")
404 ui
.status(prefix
+ ">> do_svn_add: get_rev_map: %s (source) -> %s (target)", copyfrom_rev
, tgt_rev
, level
=ui
.DEBUG
, color
='GREEN')
406 ui
.status(prefix
+ ">> do_svn_add: Check copy-from: No ancestor chain found.", level
=ui
.DEBUG
, color
='GREEN')
407 found_ancestor
= False
408 if found_ancestor
and tgt_rev
:
409 # Check if this path_offset in the target WC already has this ancestry, in which
410 # case there's no need to run the "svn copy" (again).
411 path_in_svn
= in_svn(path_offset
, prefix
=prefix
+" ")
412 log_entry
= svnclient
.get_last_svn_log_entry(path_offset
, 1, 'HEAD', get_changed_paths
=False) if in_svn(path_offset
, require_in_repo
=True, prefix
=prefix
+" ") else []
413 if (not log_entry
or (log_entry
['revision'] != tgt_rev
)):
414 copyfrom_offset
= copyfrom_path
[len(source_base
):].strip('/')
415 ui
.status(prefix
+ ">> do_svn_add: svn_copy: Copy-from: %s", copyfrom_path
+"@"+str(copyfrom_rev
), level
=ui
.DEBUG
, color
='GREEN')
416 ui
.status(prefix
+ " copyfrom: %s", copyfrom_path
+"@"+str(copyfrom_rev
), level
=ui
.DEBUG
, color
='GREEN')
417 ui
.status(prefix
+ " p_copyfrom: %s", parent_copyfrom_path
+"@"+str(parent_copyfrom_rev
) if parent_copyfrom_path
else "", level
=ui
.DEBUG
, color
='GREEN')
419 ((parent_copyfrom_path
and is_child_path(copyfrom_path
, parent_copyfrom_path
)) and \
420 (parent_copyfrom_rev
and copyfrom_rev
== parent_copyfrom_rev
)):
421 # When being called recursively, if this child entry has the same ancestor as the
422 # the parent, then no need to try to run another "svn copy".
423 ui
.status(prefix
+ ">> do_svn_add: svn_copy: Same ancestry as parent: %s",
424 parent_copyfrom_path
+"@"+str(parent_copyfrom_rev
),level
=ui
.DEBUG
, color
='GREEN')
427 # Copy this path from the equivalent path+rev in the target repo, to create the
428 # equivalent history.
429 if parent_copyfrom_path
:
430 # If we have a parent copy-from path, we mis-match that so display a status
431 # message describing the action we're mimic'ing. If path_in_svn, then this
432 # is logically a "replace" rather than an "add".
433 ui
.status(" %s %s (from %s)", ('R' if path_in_svn
else 'A'), source_base
+"/"+path_offset
, ancestors
[1]['path']+"@"+str(copyfrom_rev
), level
=ui
.VERBOSE
)
435 # If local file is already under version-control, then this is a replace.
436 ui
.status(prefix
+ ">> do_svn_add: pre-copy: local path already exists: %s", path_offset
, level
=ui
.DEBUG
, color
='GREEN')
437 run_svn(["remove", "--force", path_offset
])
438 run_svn(["copy", "-r", tgt_rev
, target_url
+"/"+copyfrom_offset
+"@"+str(tgt_rev
), path_offset
])
440 # Export the final verison of all files in this folder.
441 add_path(export_paths
, path_offset
)
443 # Export the final verison of this file.
444 run_svn(["export", "--force", "-r", source_rev
,
445 source_repos_url
+source_base
+"/"+path_offset
+"@"+str(source_rev
), path_offset
])
446 if options
.keep_prop
:
447 sync_svn_props(source_url
, source_rev
, path_offset
)
449 ui
.status(prefix
+ ">> do_svn_add: Skipped 'svn copy': %s", path_offset
, level
=ui
.DEBUG
, color
='GREEN')
451 # Else, either this copy-from path has no ancestry back to source_url OR copyfrom_rev comes
452 # before our initial source_start_rev (i.e. tgt_rev == None), so can't do a "svn copy".
453 # Create (parent) directory if needed.
454 # TODO: This is (nearly) a duplicate of code in process_svn_log_entry(). Should this be
455 # split-out to a shared tag?
456 p_path
= path_offset
if is_dir
else os
.path
.dirname(path_offset
).strip() or '.'
457 if not os
.path
.exists(p_path
):
458 run_svn(["mkdir", p_path
])
459 if not in_svn(path_offset
, prefix
=prefix
+" "):
461 # Export the final verison of all files in this folder.
462 add_path(export_paths
, path_offset
)
464 # Export the final verison of this file. We *need* to do this before running
465 # the "svn add", even if we end-up re-exporting this file again via export_paths.
466 run_svn(["export", "--force", "-r", source_rev
,
467 source_repos_url
+source_base
+"/"+path_offset
+"@"+str(source_rev
), path_offset
])
468 # If not already under version-control, then "svn add" this file/folder.
469 run_svn(["add", "--parents", path_offset
])
470 if options
.keep_prop
:
471 sync_svn_props(source_url
, source_rev
, path_offset
)
473 # For any folders that we process, process any child contents, so that we correctly
474 # replay copies/replaces/etc.
475 do_svn_add_dir(path_offset
, source_rev
, copyfrom_path
, copyfrom_rev
, export_paths
, skip_paths
, prefix
+" ")
477 def do_svn_add_dir(path_offset
, source_rev
, parent_copyfrom_path
, parent_copyfrom_rev
, \
478 export_paths
, skip_paths
, prefix
=""):
479 # Get the directory contents, to compare between the local WC (target_url) vs. the remote repo (source_url)
480 # TODO: paths_local won't include add'd paths because "svn ls" lists the contents of the
481 # associated remote repo folder. (Is this a problem?)
482 paths_local
= get_svn_dirlist(path_offset
)
483 paths_remote
= get_svn_dirlist(source_url
+"/"+path_offset
, source_rev
)
484 ui
.status(prefix
+ ">> do_svn_add_dir: paths_local: %s", str(paths_local
), level
=ui
.DEBUG
, color
='GREEN')
485 ui
.status(prefix
+ ">> do_svn_add_dir: paths_remote: %s", str(paths_remote
), level
=ui
.DEBUG
, color
='GREEN')
486 # Update files/folders which exist in remote but not local
487 for path
in paths_remote
:
488 path_is_dir
= True if path
[-1] == "/" else False
489 working_path
= path_offset
+"/"+(path
.rstrip('/') if path_is_dir
else path
)
490 if not working_path
in skip_paths
:
491 do_svn_add(working_path
, source_rev
, parent_copyfrom_path
, parent_copyfrom_rev
,
492 export_paths
, path_is_dir
, skip_paths
, prefix
+" ")
493 # Remove files/folders which exist in local but not remote
494 for path
in paths_local
:
495 if not path
in paths_remote
:
496 ui
.status(" %s %s", 'D', source_base
+"/"+path_offset
+"/"+path
, level
=ui
.VERBOSE
)
497 run_svn(["remove", "--force", path_offset
+"/"+path
])
498 # TODO: Does this handle deleted folders too? Wouldn't want to have a case
499 # where we only delete all files from folder but leave orphaned folder around.
501 def process_svn_log_entry(log_entry
, commit_paths
, prefix
= ""):
503 Process SVN changes from the given log entry. Build an array (commit_paths)
504 of the paths in the working-copy that were changed, i.e. the paths which
505 we'll pass to "svn commit".
508 source_rev
= log_entry
['revision']
509 ui
.status(prefix
+ ">> process_svn_log_entry: %s", source_url
+"@"+str(source_rev
), level
=ui
.DEBUG
, color
='GREEN')
510 for d
in log_entry
['changed_paths']:
511 # Get the full path for this changed_path
512 # e.g. '/branches/bug123/projectA/file1.txt'
514 if not is_child_path(path
, source_base
):
515 # Ignore changed files that are not part of this subdir
516 ui
.status(prefix
+ ">> process_svn_log_entry: Unrelated path: %s (base: %s)", path
, source_base
, level
=ui
.DEBUG
, color
='GREEN')
518 if d
['kind'] == "" or d
['kind'] == 'none':
519 # The "kind" value was introduced in SVN 1.6, and "svn log --xml" won't return a "kind"
520 # value for commits made on a pre-1.6 repo, even if the server is now running 1.6.
521 # We need to use other methods to fetch the node-kind for these cases.
522 d
['kind'] = svnclient
.get_kind(source_repos_url
, path
, source_rev
, d
['action'], log_entry
['changed_paths'])
523 assert (d
['kind'] == 'file') or (d
['kind'] == 'dir')
524 path_is_dir
= True if d
['kind'] == 'dir' else False
525 path_is_file
= True if d
['kind'] == 'file' else False
526 # Calculate the offset (based on source_base) for this changed_path
527 # e.g. 'projectA/file1.txt'
528 # (path = source_base + "/" + path_offset)
529 path_offset
= path
[len(source_base
):].strip("/")
530 # Get the action for this path
532 if action
not in _valid_svn_actions
:
533 raise UnsupportedSVNAction("In SVN rev. %d: action '%s' not supported. Please report a bug!"
534 % (source_rev
, action
))
535 ui
.status(" %s %s%s", action
, d
['path'],
536 (" (from %s)" % (d
['copyfrom_path']+"@"+str(d
['copyfrom_revision']))) if d
['copyfrom_path'] else "",
539 # Try to be efficient and keep track of an explicit list of paths in the
540 # working copy that changed. If we commit from the root of the working copy,
541 # then SVN needs to crawl the entire working copy looking for pending changes.
542 commit_paths
.append(path_offset
)
544 # Special-handling for replace's
546 # If file was "replaced" (deleted then re-added, all in same revision),
547 # then we need to run the "svn rm" first, then change action='A'. This
548 # lets the normal code below handle re-"svn add"'ing the files. This
549 # should replicate the "replace".
550 if in_svn(path_offset
):
551 # Target path might not be under version-control yet, e.g. parent "add"
552 # was a copy-from a branch which had no ancestry back to trunk, and each
553 # child folder under that parent folder is a "replace" action on the final
554 # merge to trunk. Since the child folders will be in skip_paths, do_svn_add
555 # wouldn't have created them while processing the parent "add" path.
556 run_svn(["remove", "--force", path_offset
])
559 # Handle all the various action-types
560 # (Handle "add" first, for "svn copy/move" support)
562 # Determine where to export from.
564 # Handle cases where this "add" was a copy from another URL in the source repo
565 if d
['copyfrom_revision']:
566 copyfrom_path
= d
['copyfrom_path']
567 copyfrom_rev
= d
['copyfrom_revision']
569 for tmp_d
in log_entry
['changed_paths']:
570 tmp_path
= tmp_d
['path']
571 if is_child_path(tmp_path
, path
) and tmp_d
['action'] in 'ARD':
572 # Build list of child entries which are also in the changed_paths list,
573 # so that do_svn_add() can skip processing these entries when recursing
574 # since we'll end-up processing them later. Don't include action="M" paths
575 # in this list because it's non-conclusive: it could just mean that the
576 # file was modified *after* the copy-from, so we still want do_svn_add()
577 # to re-create the correct ancestry.
578 tmp_path_offset
= tmp_path
[len(source_base
):].strip("/")
579 skip_paths
.append(tmp_path_offset
)
580 do_svn_add(path_offset
, source_rev
, "", "", export_paths
, path_is_dir
, skip_paths
, prefix
+" ")
581 # Else just "svn export" the files from the source repo and "svn add" them.
583 # Create (parent) directory if needed
584 p_path
= path_offset
if path_is_dir
else os
.path
.dirname(path_offset
).strip() or '.'
585 if not os
.path
.exists(p_path
):
586 run_svn(["mkdir", p_path
])
587 # Export the entire added tree.
589 # For directories, defer the (recurisve) "svn export". Might have a
590 # situation in a branch merge where the entry in the svn-log is a
591 # non-copy-from'd "add" but there are child contents (that we haven't
592 # gotten to yet in log_entry) that are copy-from's. When we try do
593 # the "svn copy" later on in do_svn_add() for those copy-from'd paths,
594 # having pre-existing (svn-add'd) contents creates some trouble.
595 # Instead, just create the stub folders ("svn mkdir" above) and defer
596 # exporting the final file-state until the end.
597 add_path(export_paths
, path_offset
)
599 # Export the final verison of this file. We *need* to do this before running
600 # the "svn add", even if we end-up re-exporting this file again via export_paths.
601 run_svn(["export", "--force", "-r", source_rev
,
602 source_url
+"/"+path_offset
+"@"+str(source_rev
), path_offset
])
603 if not in_svn(path_offset
, prefix
=prefix
+" "):
604 # Need to use in_svn here to handle cases where client committed the parent
605 # folder and each indiv sub-folder.
606 run_svn(["add", "--parents", path_offset
])
607 if options
.keep_prop
:
608 sync_svn_props(source_url
, source_rev
, path_offset
)
611 run_svn(["remove", "--force", path_offset
])
615 run_svn(["export", "--force", "-N" , "-r", source_rev
,
616 source_url
+"/"+path_offset
+"@"+str(source_rev
), path_offset
])
617 if options
.keep_prop
:
618 sync_svn_props(source_url
, source_rev
, path_offset
)
621 raise InternalError("Internal Error: process_svn_log_entry: Unhandled 'action' value: '%s'"
624 # Export the final version of all add'd paths from source_url
626 for path_offset
in export_paths
:
627 run_svn(["export", "--force", "-r", source_rev
,
628 source_url
+"/"+path_offset
+"@"+str(source_rev
), path_offset
])
630 def keep_revnum(source_rev
, target_rev_last
, wc_target_tmp
):
632 Add "padding" target revisions as needed to keep source and target
633 revision #'s identical.
635 if int(source_rev
) <= int(target_rev_last
):
636 raise InternalError("keep-revnum mode is enabled, "
637 "but source revision (r%s) is less-than-or-equal last target revision (r%s)" % \
638 (source_rev
, target_rev_last
))
639 if int(target_rev_last
) < int(source_rev
)-1:
640 # Add "padding" target revisions to keep source and target rev #'s identical
641 if os
.path
.exists(wc_target_tmp
):
642 shutil
.rmtree(wc_target_tmp
)
643 run_svn(["checkout", "-r", "HEAD", "--depth=empty", target_repos_url
, wc_target_tmp
])
644 for rev_num
in range(int(target_rev_last
)+1, int(source_rev
)):
645 run_svn(["propset", "svn2svn:keep-revnum", rev_num
, wc_target_tmp
])
646 output
= run_svn(["commit", "-m", "", wc_target_tmp
])
647 rev_num_tmp
= parse_svn_commit_rev(output
) if output
else None
648 assert rev_num
== rev_num_tmp
649 ui
.status("Committed revision %s (keep-revnum).", rev_num
)
650 target_rev_last
= rev_num
651 shutil
.rmtree(wc_target_tmp
)
652 # Update our target working-copy, to ensure everything says it's at the new HEAD revision
654 return target_rev_last
656 def disp_svn_log_summary(log_entry
):
657 ui
.status("------------------------------------------------------------------------")
658 ui
.status("r%s | %s | %s",
659 log_entry
['revision'],
661 str(datetime
.fromtimestamp(int(log_entry
['date'])).isoformat(' ')))
662 ui
.status(log_entry
['message'])
664 def real_main(args
, parser
):
665 global source_url
, target_url
, rev_map
666 source_url
= args
.pop(0).rstrip("/") # e.g. 'http://server/svn/source/trunk'
667 target_url
= args
.pop(0).rstrip("/") # e.g. 'file:///svn/target/trunk'
668 ui
.status("options: %s", str(options
), level
=ui
.DEBUG
, color
='GREEN')
670 # Make sure that both the source and target URL's are valid
671 source_info
= svnclient
.get_svn_info(source_url
)
672 assert is_child_path(source_url
, source_info
['repos_url'])
673 target_info
= svnclient
.get_svn_info(target_url
)
674 assert is_child_path(target_url
, target_info
['repos_url'])
677 global source_repos_url
,source_base
,source_repos_uuid
678 source_repos_url
= source_info
['repos_url'] # e.g. 'http://server/svn/source'
679 source_base
= source_url
[len(source_repos_url
):] # e.g. '/trunk'
680 source_repos_uuid
= source_info
['repos_uuid']
681 global target_repos_url
682 target_repos_url
= target_info
['repos_url']
684 # Init start and end revision
686 source_start_rev
= svnclient
.get_svn_rev(source_repos_url
, options
.rev_start
if options
.rev_start
else 1)
687 except ExternalCommandFailed
:
688 parser
.error("invalid start source revision value: %s" % (options
.rev_start
))
690 source_end_rev
= svnclient
.get_svn_rev(source_repos_url
, options
.rev_end
if options
.rev_end
else "HEAD")
691 except ExternalCommandFailed
:
692 parser
.error("invalid end source revision value: %s" % (options
.rev_end
))
693 ui
.status("Using source revision range %s:%s", source_start_rev
, source_end_rev
, level
=ui
.VERBOSE
)
695 # TODO: If options.keep_date, should we try doing a "svn propset" on an *existing* revision
696 # as a sanity check, so we check if the pre-revprop-change hook script is correctly setup
697 # before doing first replay-commit?
699 target_rev_last
= target_info
['revision'] # Last revision # in the target repo
700 wc_target
= os
.path
.abspath('_wc_target')
701 wc_target_tmp
= os
.path
.abspath('_tmp_wc_target')
707 # Check out a working copy of target_url if needed
708 wc_exists
= os
.path
.exists(wc_target
)
709 if wc_exists
and not options
.cont_from_break
:
710 shutil
.rmtree(wc_target
)
713 ui
.status("Checking-out _wc_target...", level
=ui
.VERBOSE
)
714 svnclient
.svn_checkout(target_url
, wc_target
)
717 if not options
.cont_from_break
:
718 # TODO: Warn user if trying to start (non-continue) into a non-empty target path?
719 # Get the first log entry at/after source_start_rev, which is where
720 # we'll do the initial import from.
721 it_log_start
= svnclient
.iter_svn_log_entries(source_url
, source_start_rev
, source_end_rev
, get_changed_paths
=False)
722 for source_start_log
in it_log_start
:
724 if not source_start_log
:
725 raise InternalError("Unable to find any matching revisions between %s:%s in source_url: %s" % \
726 (source_start_rev
, source_end_rev
, source_url
))
728 # This is the revision we will start from for source_url
729 source_start_rev
= source_rev
= int(source_start_log
['revision'])
730 ui
.status("Starting at source revision %s.", source_start_rev
, level
=ui
.VERBOSE
)
732 if options
.keep_revnum
and source_rev
> target_rev_last
:
733 target_rev_last
= keep_revnum(source_rev
, target_rev_last
, wc_target_tmp
)
735 # For the initial commit to the target URL, export all the contents from
736 # the source URL at the start-revision.
737 disp_svn_log_summary(svnclient
.get_one_svn_log_entry(source_url
, source_rev
, source_rev
))
738 # Export and add file-contents from source_url@source_start_rev
739 top_paths
= run_svn(["list", "-r", source_rev
, source_url
+"@"+str(source_rev
)])
740 top_paths
= top_paths
.strip("\n").split("\n")
741 for path
in top_paths
:
742 # For each top-level file/folder...
745 # Directories have a trailing slash in the "svn list" output
746 path_is_dir
= True if path
[-1] == "/" else False
747 path_offset
= path
.rstrip('/') if path_is_dir
else path
748 if in_svn(path_offset
, prefix
=" "):
749 raise InternalError("Cannot replay history on top of pre-existing structure: %s" % source_url
+"/"+path_offset
)
750 if path_is_dir
and not os
.path
.exists(path_offset
):
751 os
.makedirs(path_offset
)
752 run_svn(["export", "--force", "-r" , source_rev
, source_url
+"/"+path_offset
+"@"+str(source_rev
), path_offset
])
753 run_svn(["add", path_offset
])
754 # Update any properties on the newly added content
755 paths
= run_svn(["list", "--recursive", "-r", source_rev
, source_url
+"@"+str(source_rev
)])
756 paths
= paths
.strip("\n").split("\n")
757 if options
.keep_prop
:
758 sync_svn_props(source_url
, source_rev
, "")
762 # Directories have a trailing slash in the "svn list" output
763 path_is_dir
= True if path
[-1] == "/" else False
764 path_offset
= path
.rstrip('/') if path_is_dir
else path
765 ui
.status(" A %s", source_base
+"/"+path_offset
, level
=ui
.VERBOSE
)
766 if options
.keep_prop
:
767 sync_svn_props(source_url
, source_rev
, path_offset
)
768 # Commit the initial import
769 num_entries_proc
+= 1
770 target_revprops
= gen_tracking_revprops(source_rev
) # Build source-tracking revprop's
771 target_rev
= commit_from_svn_log_entry(source_start_log
, target_revprops
=target_revprops
)
773 # Update rev_map, mapping table of source-repo rev # -> target-repo rev #
774 set_rev_map(source_rev
, target_rev
)
775 # Update our target working-copy, to ensure everything says it's at the new HEAD revision
778 target_rev_last
= target_rev
780 # Re-build the rev_map based on any already-replayed history in target_url
781 build_rev_map(target_url
, target_rev_last
, source_info
)
783 parser
.error("called with continue-mode, but no already-replayed source history found in target_url")
784 source_start_rev
= int(max(rev_map
, key
=rev_map
.get
))
785 assert source_start_rev
786 ui
.status("Continuing from source revision %s.", source_start_rev
, level
=ui
.VERBOSE
)
789 if options
.keep_revnum
and source_start_rev
< target_rev_last
:
790 parser
.error("last target revision is equal-or-higher than starting source revision; "
791 "cannot use --keep-revnum mode")
793 svn_vers_t
= svnclient
.get_svn_client_version()
794 svn_vers
= float(".".join(map(str, svn_vers_t
[0:2])))
796 # Load SVN log starting from source_start_rev + 1
797 it_log_entries
= svnclient
.iter_svn_log_entries(source_url
, source_start_rev
+1, source_end_rev
, get_revprops
=True) if source_start_rev
< source_end_rev
else []
800 # TODO: Now that commit_from_svn_log_entry() might try to do a "svn propset svn:date",
801 # we might want some better KeyboardInterupt handilng here, to ensure that
802 # commit_from_svn_log_entry() always runs as an atomic unit.
804 for log_entry
in it_log_entries
:
805 if options
.entries_proc_limit
:
806 if num_entries_proc
>= options
.entries_proc_limit
:
808 # Replay this revision from source_url into target_url
809 source_rev
= log_entry
['revision']
810 if options
.keep_revnum
:
811 target_rev_last
= keep_revnum(source_rev
, target_rev_last
, wc_target_tmp
)
812 disp_svn_log_summary(log_entry
)
813 # Process all the changed-paths in this log entry
815 process_svn_log_entry(log_entry
, commit_paths
)
816 num_entries_proc
+= 1
817 # Commit any changes made to _wc_target
818 target_revprops
= gen_tracking_revprops(source_rev
) # Build source-tracking revprop's
819 target_rev
= commit_from_svn_log_entry(log_entry
, commit_paths
, target_revprops
=target_revprops
)
821 # Update rev_map, mapping table of source-repo rev # -> target-repo rev #
822 source_rev
= log_entry
['revision']
823 set_rev_map(source_rev
, target_rev
)
824 target_rev_last
= target_rev
825 # Update our target working-copy, to ensure everything says it's at the new HEAD revision
828 # Run "svn cleanup" every 100 commits if SVN 1.7+, to clean-up orphaned ".svn/pristines/*"
829 if svn_vers
>= 1.7 and (commit_count
% 100 == 0):
832 # If there were no new source_url revisions to process, init source_rev
833 # for the "finally" message below to be the last source revision replayed.
834 source_rev
= source_start_rev
836 except KeyboardInterrupt:
837 print "\nStopped by user."
838 print "\nCleaning-up..."
842 print "\nCommand failed with following error:\n"
843 traceback
.print_exc()
844 print "\nCleaning-up..."
846 print run_svn(["status"])
849 print "\nFinished at source revision %s%s." % (source_rev
, " (dry-run)" if options
.dry_run
else "")
852 # Defined as entry point. Must be callable without arguments.
853 usage
= "svn2svn, version %s\n" % str(full_version
) + \
854 "<http://nynim.org/projects/svn2svn> <https://github.com/tonyduckles/svn2svn>\n\n" + \
855 "Usage: %prog [OPTIONS] source_url target_url\n"
857 Replicate (replay) history from one SVN repository to another. Maintain
858 logical ancestry wherever possible, so that 'svn log' on the replayed repo
859 will correctly follow file/folder renames.
862 Create a copy of only /trunk from source repo, starting at r5000
863 $ svnadmin create /svn/target
864 $ svn mkdir -m 'Add trunk' file:///svn/target/trunk
865 $ svn2svn -av -r 5000 http://server/source/trunk file:///svn/target/trunk
866 1. The target_url will be checked-out to ./_wc_target
867 2. The first commit to http://server/source/trunk at/after r5000 will be
868 exported & added into _wc_target
869 3. All revisions affecting http://server/source/trunk (starting at r5000)
870 will be replayed to _wc_target. Any add/copy/move/replaces that are
871 copy-from'd some path outside of /trunk (e.g. files renamed on a
872 /branch and branch was merged into /trunk) will correctly maintain
873 logical ancestry where possible.
875 Use continue-mode (-c) to pick-up where the last run left-off
876 $ svn2svn -avc http://server/source/trunk file:///svn/target/trunk
877 1. The target_url will be checked-out to ./_wc_target, if not already
879 2. All new revisions affecting http://server/source/trunk starting from
880 the last replayed revision to file:///svn/target/trunk (based on the
881 svn2svn:* revprops) will be replayed to _wc_target, maintaining all
882 logical ancestry where possible."""
883 parser
= optparse
.OptionParser(usage
, description
=description
,
884 formatter
=HelpFormatter(), version
="%prog "+str(full_version
))
885 parser
.add_option("-v", "--verbose", dest
="verbosity", action
="count", default
=1,
886 help="enable additional output (use -vv or -vvv for more)")
887 parser
.add_option("-a", "--archive", action
="store_true", dest
="archive", default
=False,
888 help="archive/mirror mode; same as -UDP (see REQUIRE's below)\n"
889 "maintain same commit author, same commit time, and file/dir properties")
890 parser
.add_option("-U", "--keep-author", action
="store_true", dest
="keep_author", default
=False,
891 help="maintain same commit authors (svn:author) as source\n"
892 "(REQUIRES target_url be non-auth'd, e.g. file://-based, since this uses --username to set author)")
893 parser
.add_option("-D", "--keep-date", action
="store_true", dest
="keep_date", default
=False,
894 help="maintain same commit time (svn:date) as source\n"
895 "(REQUIRES 'pre-revprop-change' hook script to allow 'svn:date' changes)")
896 parser
.add_option("-P", "--keep-prop", action
="store_true", dest
="keep_prop", default
=False,
897 help="maintain same file/dir SVN properties as source")
898 parser
.add_option("-R", "--keep-revnum", action
="store_true", dest
="keep_revnum", default
=False,
899 help="maintain same rev #'s as source. creates placeholder target "
900 "revisions (by modifying a 'svn2svn:keep-revnum' property at the root of the target repo)")
901 parser
.add_option("-c", "--continue", action
="store_true", dest
="cont_from_break",
902 help="continue from last source commit to target (based on svn2svn:* revprops)")
903 parser
.add_option("-r", "--revision", type="string", dest
="revision", metavar
="ARG",
904 help="revision range to replay from source_url\n"
905 "A revision argument can be one of:\n"
906 " START start rev # (end will be 'HEAD')\n"
907 " START:END start and ending rev #'s\n"
908 "Any revision # formats which SVN understands are "
909 "supported, e.g. 'HEAD', '{2010-01-31}', etc.")
910 parser
.add_option("-u", "--log-author", action
="store_true", dest
="log_author", default
=False,
911 help="append source commit author to replayed commit mesages")
912 parser
.add_option("-d", "--log-date", action
="store_true", dest
="log_date", default
=False,
913 help="append source commit time to replayed commit messages")
914 parser
.add_option("-l", "--limit", type="int", dest
="entries_proc_limit", metavar
="NUM",
915 help="maximum number of source revisions to process")
916 parser
.add_option("-n", "--dry-run", action
="store_true", dest
="dry_run", default
=False,
917 help="process next source revision but don't commit changes to "
918 "target working-copy (forces --limit=1)")
919 parser
.add_option("--debug", dest
="verbosity", const
=ui
.DEBUG
, action
="store_const",
920 help="enable debugging output (same as -vvv)")
922 options
, args
= parser
.parse_args()
924 parser
.error("incorrect number of arguments")
925 if options
.verbosity
< 10:
926 # Expand multiple "-v" arguments to a real ui._level value
927 options
.verbosity
*= 10
929 # When in dry-run mode, only try to process the next log_entry
930 options
.entries_proc_limit
= 1
931 options
.rev_start
= None
932 options
.rev_end
= None
934 # Reg-ex for matching a revision arg (http://svnbook.red-bean.com/en/1.5/svn.tour.revs.specifiers.html#svn.tour.revs.dates)
935 rev_patt
= '[0-9A-Z]+|\{[0-9A-Za-z/\\ :-]+\}'
937 match
= re
.match('^('+rev_patt
+'):('+rev_patt
+')$', options
.revision
) # First try start:end match
938 if match
is None: match
= re
.match('^('+rev_patt
+')$', options
.revision
) # Next, try start match
940 parser
.error("unexpected --revision argument format; see 'svn help log' for valid revision formats")
942 options
.rev_start
= rev
[0] if len(rev
)>0 else None
943 options
.rev_end
= rev
[1] if len(rev
)>1 else None
945 options
.keep_author
= True
946 options
.keep_date
= True
947 options
.keep_prop
= True
948 ui
.update_config(options
)
949 return real_main(args
, parser
)
952 if __name__
== "__main__":
953 sys
.exit(main() or 0)