svn2svn/run/svn2svn.py

   1 """
   2 Replicate (replay) changesets from one SVN repository to another.
   3 """
   4
   5 from .. import base_version, full_version
   6 from .. import ui
   7 from .. import svnclient
   8 from ..shell import run_svn,run_shell_command
   9 from ..errors import (ExternalCommandFailed, UnsupportedSVNAction, InternalError, VerificationError)
  10 from parse import HelpFormatter
  11 from breakhandler import BreakHandler
  12
  13 import sys
  14 import os
  15 import time
  16 import traceback
  17 import shutil
  18 import operator
  19 import optparse
  20 import re
  21 import urllib
  22 from datetime import datetime
  23
  24 _valid_svn_actions = "MARD"   # The list of known SVN action abbr's, from "svn log"
  25
  26 # Module-level variables/parameters
  27 source_url = ""          # URL to source path in source SVN repo, e.g. 'http://server/svn/source/trunk'
  28 source_repos_url = ""    # URL to root of source SVN repo,        e.g. 'http://server/svn/source'
  29 source_base = ""         # Relative path of source_url in source SVN repo, e.g. '/trunk'
  30 source_repos_uuid = ""   # UUID of source SVN repo
  31 target_url =""           # URL to target path in target SVN repo, e.g. 'file:///svn/repo_target/trunk'
  32 target_repos_url = ""    # URL to root of target SVN repo,        e.g. 'http://server/svn/target'
  33 target_base = ""         # Relative path of target_url in target SVN repo, e.g. '/trunk'
  34 rev_map = {}             # The running mapping-table dictionary for source_url rev #'s -> target_url rev #'s
  35 options = None           # optparser options
  36
  37 def parse_svn_commit_rev(output):
  38     """
  39     Parse the revision number from the output of "svn commit".
  40     """
  41     output_lines = output.strip("\n").split("\n")
  42     rev_num = None
  43     for line in output_lines:
  44         if line[0:19] == 'Committed revision ':
  45             rev_num = line[19:].rstrip('.')
  46             break
  47     assert rev_num is not None
  48     return int(rev_num)
  49
  50 def commit_from_svn_log_entry(log_entry, commit_paths=None, target_revprops=None):
  51     """
  52     Given an SVN log entry and an optional list of changed paths, do an svn commit.
  53     """
  54     # TODO: Run optional external shell hook here, for doing pre-commit filtering
  55     # Display the _wc_target "svn status" info if running in -vv (or higher) mode
  56     if ui.get_level() >= ui.EXTRA:
  57         ui.status(">> commit_from_svn_log_entry: Pre-commit _wc_target status:", level=ui.EXTRA, color='CYAN')
  58         ui.status(run_svn(["status"]), level=ui.EXTRA, color='CYAN')
  59     # This will use the local timezone for displaying commit times
  60     timestamp = int(log_entry['date'])
  61     svn_date = str(datetime.fromtimestamp(timestamp))
  62     # Uncomment this one one if you prefer UTC commit times
  63     #svn_date = "%d 0" % timestamp
  64     args = ["commit", "--force-log"]
  65     message = log_entry['message']
  66     if options.log_date:
  67         message += "\nDate: " + svn_date
  68     if options.log_author:
  69         message += "\nAuthor: " + log_entry['author']
  70     args += ["-m", message]
  71     revprops = {}
  72     if log_entry['revprops']:
  73         # Carry forward any revprop's from the source revision
  74         for v in log_entry['revprops']:
  75             revprops[v['name']] = v['value']
  76     if target_revprops:
  77         # Add any extra revprop's we want to set for the target repo commits
  78         for v in target_revprops:
  79             revprops[v['name']] = v['value']
  80     if revprops:
  81         for key in revprops:
  82             args += ["--with-revprop", "%s=%s" % (key, str(revprops[key]))]
  83     if commit_paths:
  84         if len(commit_paths)<100:
  85             # If we don't have an excessive amount of individual changed paths, pass
  86             # those to the "svn commit" command. Else, pass nothing so we commit at
  87             # the root of the working-copy.
  88             args += list(commit_paths)
  89     rev_num = None
  90     if not options.dry_run:
  91         # Use BreakHandler class to temporarily redirect SIGINT handler, so that
  92         # "svn commit" + post-commit rev-prop updating is a quasi-atomic unit.
  93         # If user presses Ctrl-C during this, wait until after this full action
  94         # has finished raising the KeyboardInterrupt exception.
  95         bh = BreakHandler()
  96         bh.enable()
  97         # Run the "svn commit" command, and screen-scrape the target_rev value (if any)
  98         output = run_svn(args)
  99         rev_num = parse_svn_commit_rev(output) if output else None
 100         if rev_num is not None:
 101             if options.keep_date:
 102                 run_svn(["propset", "--revprop", "-r", rev_num, "svn:date", log_entry['date_raw']])
 103             if options.keep_author:
 104                 run_svn(["propset", "--revprop", "-r", rev_num, "svn:author",  log_entry['author']])
 105             ui.status("Committed revision %s (source r%s).", rev_num, log_entry['revision'])
 106         bh.disable()
 107         # Check if the user tried to press Ctrl-C
 108         if bh.trapped:
 109             raise KeyboardInterrupt
 110     return rev_num
 111
 112 def verify_commit(source_rev, target_rev, log_entry=None):
 113     """
 114     Compare the ancestry/content/properties between source_url vs target_url
 115     for a given revision.
 116     """
 117     error_cnt = 0
 118     # Gather the offsets in the source repo to check
 119     check_paths = []
 120     remove_paths = []
 121     # TODO: Need to make this ancestry aware
 122     if options.verify == 1 and log_entry is not None:  # Changed only
 123         ui.status("Verifying source revision %s (only-changed)...", source_rev, level=ui.VERBOSE)
 124         for d in log_entry['changed_paths']:
 125             path = d['path']
 126             if not is_child_path(path, source_base):
 127                 continue
 128             if d['kind'] == "":
 129                 d['kind'] = svnclient.get_kind(source_repos_url, path, source_rev, d['action'], log_entry['changed_paths'])
 130             assert (d['kind'] == 'file') or (d['kind'] == 'dir')
 131             path_is_dir =  True if d['kind'] == 'dir'  else False
 132             path_is_file = True if d['kind'] == 'file' else False
 133             path_offset = path[len(source_base):].strip("/")
 134             if d['action'] == 'D':
 135                 remove_paths.append(path_offset)
 136             elif not path_offset in check_paths:
 137                 ui.status("verify_commit: path [mode=changed]: kind=%s: %s", d['kind'], path, level=ui.DEBUG, color='YELLOW')
 138                 if path_is_file:
 139                     ui.status("  "+"verify_commit [mode=changed]: check_paths.append('%s')", path_offset, level=ui.DEBUG, color='GREEN')
 140                     check_paths.append(path_offset)
 141                 if path_is_dir:
 142                     if not d['action'] in 'AR':
 143                         continue
 144                     child_paths = run_svn(["list", "--recursive", "-r", source_rev, source_url.rstrip("/")+"/"+path_offset+"@"+str(source_rev)])
 145                     child_paths = child_paths.strip("\n").split("\n")
 146                     for child_path in child_paths:
 147                         if not child_path:
 148                             continue
 149                         # Directories have a trailing slash in the "svn list" output
 150                         child_path_is_dir = True if child_path[-1] == "/" else False
 151                         child_path_offset = child_path.rstrip('/') if child_path_is_dir else child_path
 152                         if not child_path_is_dir:
 153                             # Only check files
 154                             working_path = (path_offset+"/" if path_offset else "") + child_path_offset
 155                             if not working_path in check_paths:
 156                                 ui.status("    "+"verify_commit [mode=changed]: check_paths.append('%s'+'/'+'%s')", path_offset, child_path_offset, level=ui.DEBUG, color='GREEN')
 157                                 check_paths.append(working_path)
 158     if options.verify == 2:  # All paths
 159         ui.status("Verifying source revision %s (all)...", source_rev, level=ui.VERBOSE)
 160         child_paths = run_svn(["list", "--recursive", "-r", source_rev, source_url+"@"+str(source_rev)])
 161         child_paths = child_paths.strip("\n").split("\n")
 162         for child_path in child_paths:
 163             if not child_path:
 164                 continue
 165             # Directories have a trailing slash in the "svn list" output
 166             child_path_is_dir = True if child_path[-1] == "/" else False
 167             child_path_offset = child_path.rstrip('/') if child_path_is_dir else child_path
 168             if not child_path_is_dir:
 169                 # Only check files
 170                 ui.status("verify_commit [mode=all]: check_paths.append('%s')", child_path_offset, level=ui.DEBUG, color='GREEN')
 171                 check_paths.append(child_path_offset)
 172
 173     # If there were any paths deleted in the last revision (options.verify=1 mode),
 174     # check that they were correctly deleted.
 175     if remove_paths:
 176         count_total = len(remove_paths)
 177         count = 0
 178         for path_offset in remove_paths:
 179             count += 1
 180             if in_svn(path_offset):
 181                 ui.status(" (%s/%s) Verify path: FAIL: %s", str(count).rjust(len(str(count_total))), count_total, path_offset, level=ui.EXTRA, color='RED')
 182                 ui.status("VerificationError: Path removed in source rev r%s, but still exists in target WC: %s", source_rev, path_offset, color='RED')
 183                 error_cnt +=1
 184             else:
 185                 ui.status(" (%s/%s) Verify remove: OK: %s", str(count).rjust(len(str(count_total))), count_total, path_offset, level=ui.EXTRA)
 186
 187     # Compare each of the check_path entries between source vs. target
 188     if check_paths:
 189         source_rev_first = int(min(rev_map, key=rev_map.get)) or 1  # The first source_rev we replayed into target
 190         ui.status("verify_commit: source_rev_first:%s", source_rev_first, level=ui.DEBUG, color='YELLOW')
 191         count_total = len(check_paths)
 192         count = 0
 193         for path_offset in check_paths:
 194             count += 1
 195             if count % 500 == 0:
 196                 ui.status("...processed %s (%s of %s)..." % (count, count, count_total), level=ui.VERBOSE)
 197             ui.status("verify_commit: path_offset:%s", path_offset, level=ui.DEBUG, color='YELLOW')
 198             source_log_entries = svnclient.run_svn_log(source_url.rstrip("/")+"/"+path_offset+"@"+str(source_rev), source_rev, 1, source_rev-source_rev_first+1)
 199             target_log_entries = svnclient.run_svn_log(target_url.rstrip("/")+"/"+path_offset+"@"+str(target_rev), target_rev, 1, target_rev)
 200             # Build a list of commits in source_log_entries which matches our
 201             # target path_offset.
 202             working_path = source_base+"/"+path_offset
 203             source_revs = []
 204             for log_entry in source_log_entries:
 205                 source_rev_tmp = log_entry['revision']
 206                 if source_rev_tmp < source_rev_first:
 207                     # Only process source revisions which have been replayed into target
 208                     break
 209                 #ui.status("  [verify_commit] source_rev_tmp:%s, working_path:%s\n%s", source_rev_tmp, working_path, pp.pformat(log_entry), level=ui.DEBUG, color='MAGENTA')
 210                 changed_paths_temp = []
 211                 for d in log_entry['changed_paths']:
 212                     path = d['path']
 213                     # Match working_path or any parents
 214                     if is_child_path(working_path, path):
 215                         ui.status("  verify_commit: changed_path: %s %s@%s (parent:%s)", d['action'], path, source_rev_tmp, working_path, level=ui.DEBUG, color='YELLOW')
 216                         changed_paths_temp.append({'path': path, 'data': d})
 217                 assert changed_paths_temp
 218                 # Reverse-sort any matches, so that we start with the most-granular (deepest in the tree) path.
 219                 changed_paths = sorted(changed_paths_temp, key=operator.itemgetter('path'), reverse=True)
 220                 # Find the action for our working_path in this revision. Use a loop to check in reverse order,
 221                 # so that if the target file/folder is "M" but has a parent folder with an "A" copy-from.
 222                 working_path_next = working_path
 223                 match_d = {}
 224                 for v in changed_paths:
 225                     d = v['data']
 226                     if not match_d:
 227                         match_d = d
 228                     path = d['path']
 229                     if d['action'] not in _valid_svn_actions:
 230                         raise UnsupportedSVNAction("In SVN rev. %d: action '%s' not supported. Please report a bug!"
 231                             % (log_entry['revision'], d['action']))
 232                     if d['action'] in 'AR' and d['copyfrom_revision']:
 233                         # If we found a copy-from action for a parent path, adjust our
 234                         # working_path to follow the rename/copy-from, just like find_svn_ancestors().
 235                         working_path_next = working_path.replace(d['path'], d['copyfrom_path'])
 236                         match_d = d
 237                         break
 238                 if is_child_path(working_path, source_base):
 239                     # Only add source_rev's where the path changed in this revision was a child
 240                     # of source_base, so that we silently ignore any history that happened on
 241                     # non-source_base paths (e.g. ignore branch history if we're only replaying trunk).
 242                     is_diff = False
 243                     d = match_d
 244                     if d['action'] == 'M':
 245                         # For action="M", we need to throw out cases where the only change was to
 246                         # a property which we ignore, e.g. "svn:mergeinfo".
 247                         if d['kind'] == "":
 248                             d['kind'] = svnclient.get_kind(source_repos_url, working_path, log_entry['revision'], d['action'], log_entry['changed_paths'])
 249                         assert (d['kind'] == 'file') or (d['kind'] == 'dir')
 250                         if d['kind'] == 'file':
 251                             # Check for file-content changes
 252                             # TODO: This should be made ancestor-aware, since the file won't always be at the same path in rev-1
 253                             sum1 = run_shell_command("svn cat -r %s '%s' | md5sum" % (source_rev_tmp, source_repos_url+working_path+"@"+str(source_rev_tmp)))
 254                             sum2 = run_shell_command("svn cat -r %s '%s' | md5sum" % (source_rev_tmp-1, source_repos_url+working_path_next+"@"+str(source_rev_tmp-1)))
 255                             is_diff = True if sum1 <> sum2 else False
 256                         if not is_diff:
 257                             # Check for property changes
 258                             props1 = svnclient.get_all_props(source_repos_url+working_path, source_rev_tmp)
 259                             props2 = svnclient.get_all_props(source_repos_url+working_path_next, source_rev_tmp-1)
 260                             # Ignore changes to "svn:mergeinfo", since we don't copy that
 261                             if 'svn:mergeinfo' in props1: del props1['svn:mergeinfo']
 262                             if 'svn:mergeinfo' in props2: del props2['svn:mergeinfo']
 263                             for prop in props1:
 264                                 if prop not in props2 or \
 265                                         props1[prop] != props2[prop]:
 266                                     is_diff = True
 267                                     break
 268                             for prop in props2:
 269                                 if prop not in props1 or \
 270                                         props1[prop] != props2[prop]:
 271                                     is_diff = True
 272                                     break
 273                         if not is_diff:
 274                             ui.status("  verify_commit: skip %s@%s", working_path, source_rev_tmp, level=ui.DEBUG, color='GREEN_B', bold=True)
 275                     else:
 276                         is_diff = True
 277                     if is_diff:
 278                         ui.status("  verify_commit: source_revs.append(%s), working_path:%s", source_rev_tmp, working_path, level=ui.DEBUG, color='GREEN_B')
 279                         source_revs.append({'path': working_path, 'revision': source_rev_tmp})
 280                 working_path = working_path_next
 281             # Build a list of all the target commits "svn log" returned
 282             target_revs = []
 283             target_revs_rmndr = []
 284             for log_entry in target_log_entries:
 285                 target_rev_tmp = log_entry['revision']
 286                 ui.status("  verify_commit: target_revs.append(%s)", target_rev_tmp, level=ui.DEBUG, color='GREEN_B')
 287                 target_revs.append(target_rev_tmp)
 288                 target_revs_rmndr.append(target_rev_tmp)
 289             # Compare the two lists
 290             for d in source_revs:
 291                 working_path   = d['path']
 292                 source_rev_tmp = d['revision']
 293                 target_rev_tmp = get_rev_map(source_rev_tmp, "  ")
 294                 working_offset = working_path[len(source_base):].strip("/")
 295                 sum1 = run_shell_command("svn cat -r %s '%s' | md5sum" % (source_rev_tmp, source_repos_url+working_path+"@"+str(source_rev_tmp)))
 296                 sum2 = run_shell_command("svn cat -r %s '%s' | md5sum" % (target_rev_tmp, target_url+"/"+working_offset+"@"+str(target_rev_tmp))) if target_rev_tmp is not None else ""
 297                 #print "source@%s: %s" % (str(source_rev_tmp).ljust(6), sum1)
 298                 #print "target@%s: %s" % (str(target_rev_tmp).ljust(6), sum2)
 299                 ui.status("  verify_commit: %s: source=%s target=%s", working_offset, source_rev_tmp, target_rev_tmp, level=ui.DEBUG, color='GREEN')
 300                 if not target_rev_tmp:
 301                     ui.status(" (%s/%s) Verify path: FAIL: %s", str(count).rjust(len(str(count_total))), count_total, path_offset, level=ui.EXTRA, color='RED')
 302                     ui.status("VerificationError: Unable to find corresponding target_rev for source_rev r%s in rev_map (path_offset='%s')", source_rev_tmp, path_offset, color='RED')
 303                     error_cnt +=1
 304                     continue
 305                 if target_rev_tmp not in target_revs:
 306                     # If found a source_rev with no equivalent target_rev in target_revs,
 307                     # check if the only difference in source_rev vs. source_rev-1 is the
 308                     # removal/addition of a trailing newline char, since this seems to get
 309                     # stripped-out sometimes during the replay (via "svn export"?).
 310                     # Strip any trailing \r\n from file-content (http://stackoverflow.com/a/1656218/346778)
 311                     sum1 = run_shell_command("svn cat -r %s '%s' | perl -i -p0777we's/\\r\\n\z//' | md5sum" % (source_rev_tmp,   source_repos_url+working_path+"@"+str(source_rev_tmp)))
 312                     sum2 = run_shell_command("svn cat -r %s '%s' | perl -i -p0777we's/\\r\\n\z//' | md5sum" % (source_rev_tmp-1, source_repos_url+working_path+"@"+str(source_rev_tmp-1)))
 313                     if sum1 <> sum2:
 314                         ui.status(" (%s/%s) Verify path: FAIL: %s", str(count).rjust(len(str(count_total))), count_total, path_offset, level=ui.EXTRA, color='RED')
 315                         ui.status("VerificationError: Found source_rev (r%s) with no corresponding target_rev: path_offset='%s'", source_rev_tmp, path_offset, color='RED')
 316                         error_cnt +=1
 317                     continue
 318                 target_revs_rmndr.remove(target_rev_tmp)
 319             if target_revs_rmndr:
 320                 rmndr_list = ", ".join(map(str, target_revs_rmndr))
 321                 ui.status(" (%s/%s) Verify path: FAIL: %s", str(count).rjust(len(str(count_total))), count_total, path_offset, level=ui.EXTRA, color='RED')
 322                 ui.status("VerificationError: Found one or more *extra* target_revs: path_offset='%s', target_revs='%s'", path_offset, rmndr_list, color='RED')
 323                 error_cnt +=1
 324             ui.status(" (%s/%s) Verify path: OK: %s", str(count).rjust(len(str(count_total))), count_total, path_offset, level=ui.EXTRA)
 325
 326     # Ensure there are no "extra" files in the target side
 327     if options.verify == 2:
 328         target_paths = []
 329         child_paths = run_svn(["list", "--recursive", "-r", target_rev, target_url+"@"+str(target_rev)])
 330         child_paths = child_paths.strip("\n").split("\n")
 331         for child_path in child_paths:
 332             if not child_path:
 333                 continue
 334             # Directories have a trailing slash in the "svn list" output
 335             child_path_is_dir = True if child_path[-1] == "/" else False
 336             child_path_offset = child_path.rstrip('/') if child_path_is_dir else child_path
 337             if not child_path_is_dir:
 338                 target_paths.append(child_path_offset)
 339         # Compare
 340         for path_offset in target_paths:
 341             if not path_offset in check_paths:
 342                 ui.status("VerificationError: Path exists in target (@%s) but not source (@%s): %s", target_rev, source_rev, path_offset, color='RED')
 343                 error_cnt += 1
 344         for path_offset in check_paths:
 345             if not path_offset in target_paths:
 346                 ui.status("VerificationError: Path exists in source (@%s) but not target (@%s): %s", source_rev, target_rev, path_offset, color='RED')
 347                 error_cnt += 1
 348
 349     if error_cnt > 0:
 350         raise VerificationError("Found %s verification errors" % (error_cnt))
 351     ui.status("Verified revision %s (%s).", target_rev, "all" if options.verify == 2 else "only-changed")
 352
 353 def full_svn_revert():
 354     """
 355     Do an "svn revert" and proactively remove any extra files in the working copy.
 356     """
 357     run_svn(["revert", "--recursive", "."])
 358     output = run_svn(["status"])
 359     if output:
 360         output_lines = output.strip("\n").split("\n")
 361         for line in output_lines:
 362             if line[0] == "?":
 363                 path = line[4:].strip(" ")
 364                 if os.path.isfile(path):
 365                     os.remove(path)
 366                 if os.path.isdir(path):
 367                     shutil.rmtree(path)
 368
 369 def gen_tracking_revprops(source_rev):
 370     """
 371     Build an array of svn2svn-specific source-tracking revprops.
 372     """
 373     revprops = [{'name':'svn2svn:source_uuid', 'value':source_repos_uuid},
 374                 {'name':'svn2svn:source_url',  'value':source_url},
 375                 {'name':'svn2svn:source_rev',  'value':source_rev}]
 376     return revprops
 377
 378 def sync_svn_props(source_url, source_rev, path_offset):
 379     """
 380     Carry-forward any unversioned properties from the source repo to the
 381     target WC.
 382     """
 383     source_props = svnclient.get_all_props(join_path(source_url, path_offset),  source_rev)
 384     target_props = svnclient.get_all_props(path_offset)
 385     if 'svn:mergeinfo' in source_props:
 386         # Never carry-forward "svn:mergeinfo"
 387         del source_props['svn:mergeinfo']
 388     for prop in target_props:
 389         if prop not in source_props:
 390             # Remove any properties which exist in target but not source
 391             run_svn(["propdel", prop, path_offset])
 392     for prop in source_props:
 393         if prop not in target_props or \
 394            source_props[prop] != target_props[prop]:
 395             # Set/update any properties which exist in source but not target or
 396             # whose value differs between source vs. target.
 397             run_svn(["propset", prop, source_props[prop], path_offset])
 398
 399 def in_svn(p, require_in_repo=False, prefix=""):
 400     """
 401     Check if a given file/folder is being tracked by Subversion.
 402     Prior to SVN 1.6, we could "cheat" and look for the existence of ".svn" directories.
 403     With SVN 1.7 and beyond, WC-NG means only a single top-level ".svn" at the root of the working-copy.
 404     Use "svn status" to check the status of the file/folder.
 405     """
 406     entries = svnclient.get_svn_status(p, no_recursive=True)
 407     if not entries:
 408         return False
 409     d = entries[0]
 410     if require_in_repo and (d['status'] == 'added' or d['revision'] is None):
 411         # If caller requires this path to be in the SVN repo, prevent returning True
 412         # for paths that are only locally-added.
 413         ret = False
 414     else:
 415         # Don't consider files tracked as deleted in the WC as under source-control.
 416         # Consider files which are locally added/copied as under source-control.
 417         ret = True if not (d['status'] == 'deleted') and (d['type'] == 'normal' or d['status'] == 'added' or d['copied'] == 'true') else False
 418     ui.status(prefix + ">> in_svn('%s', require_in_repo=%s) --> %s", p, str(require_in_repo), str(ret), level=ui.DEBUG, color='GREEN')
 419     return ret
 420
 421 def is_child_path(path, p_path):
 422     return True if (path == p_path) or (path.startswith(p_path+"/")) else False
 423
 424 def join_path(base, child):
 425     base.rstrip('/')
 426     return base+"/"+child if child else base
 427
 428 def find_svn_ancestors(svn_repos_url, start_path, start_rev, stop_base_path=None, prefix=""):
 429     """
 430     Given an initial starting path+rev, walk the SVN history backwards to inspect the
 431     ancestry of that path, optionally seeing if it traces back to stop_base_path.
 432
 433     Build an array of copyfrom_path and copyfrom_revision pairs for each of the "svn copy"'s.
 434     If we find a copyfrom_path which stop_base_path is a substring match of (e.g. we crawled
 435     back to the initial branch-copy from trunk), then return the collection of ancestor
 436     paths.  Otherwise, copyfrom_path has no ancestry compared to stop_base_path.
 437
 438     This is useful when comparing "trunk" vs. "branch" paths, to handle cases where a
 439     file/folder was renamed in a branch and then that branch was merged back to trunk.
 440
 441     'svn_repos_url' is the full URL to the root of the SVN repository,
 442       e.g. 'file:///path/to/repo'
 443     'start_path' is the path in the SVN repo to the source path to start checking
 444       ancestry at, e.g. '/branches/fix1/projectA/file1.txt'.
 445     'start_rev' is the revision to start walking the history of start_path backwards from.
 446     'stop_base_path' is the path in the SVN repo to stop tracing ancestry once we've reached,
 447       i.e. the target path we're trying to trace ancestry back to, e.g. '/trunk'.
 448     """
 449     ui.status(prefix + ">> find_svn_ancestors: Start: (%s) start_path: %s  stop_base_path: %s",
 450         svn_repos_url, start_path+"@"+str(start_rev), stop_base_path, level=ui.DEBUG, color='YELLOW')
 451     done = False
 452     no_ancestry = False
 453     cur_path = start_path
 454     cur_rev  = start_rev
 455     first_iter_done = False
 456     ancestors = []
 457     while not done:
 458         # Get the first "svn log" entry for cur_path (relative to @cur_rev)
 459         ui.status(prefix + ">> find_svn_ancestors: %s", svn_repos_url+cur_path+"@"+str(cur_rev), level=ui.DEBUG, color='YELLOW')
 460         log_entry = svnclient.get_first_svn_log_entry(svn_repos_url+cur_path, 1, cur_rev)
 461         if not log_entry:
 462             ui.status(prefix + ">> find_svn_ancestors: Done: no log_entry", level=ui.DEBUG, color='YELLOW')
 463             done = True
 464             break
 465         # If we found a copy-from case which matches our stop_base_path, we're done.
 466         # ...but only if we've at least tried to search for the first copy-from path.
 467         if stop_base_path is not None and first_iter_done and is_child_path(cur_path, stop_base_path):
 468             ui.status(prefix + ">> find_svn_ancestors: Done: Found is_child_path(cur_path, stop_base_path) and first_iter_done=True", level=ui.DEBUG, color='YELLOW')
 469             done = True
 470             break
 471         first_iter_done = True
 472         # Search for any actions on our target path (or parent paths).
 473         changed_paths_temp = []
 474         for d in log_entry['changed_paths']:
 475             path = d['path']
 476             if is_child_path(cur_path, path):
 477                 changed_paths_temp.append({'path': path, 'data': d})
 478         if not changed_paths_temp:
 479             # If no matches, then we've hit the end of the ancestry-chain.
 480             ui.status(prefix + ">> find_svn_ancestors: Done: No matching changed_paths", level=ui.DEBUG, color='YELLOW')
 481             done = True
 482             continue
 483         # Reverse-sort any matches, so that we start with the most-granular (deepest in the tree) path.
 484         changed_paths = sorted(changed_paths_temp, key=operator.itemgetter('path'), reverse=True)
 485         # Find the action for our cur_path in this revision. Use a loop to check in reverse order,
 486         # so that if the target file/folder is "M" but has a parent folder with an "A" copy-from
 487         # then we still correctly match the deepest copy-from.
 488         for v in changed_paths:
 489             d = v['data']
 490             path = d['path']
 491             # Check action-type for this file
 492             action = d['action']
 493             if action not in _valid_svn_actions:
 494                 raise UnsupportedSVNAction("In SVN rev. %d: action '%s' not supported. Please report a bug!"
 495                     % (log_entry['revision'], action))
 496             ui.status(prefix + "> %s %s%s", action, path,
 497                 (" (from %s)" % (d['copyfrom_path']+"@"+str(d['copyfrom_revision']))) if d['copyfrom_path'] else "",
 498                 level=ui.DEBUG, color='YELLOW')
 499             if action == 'D':
 500                 # If file/folder was deleted, ancestry-chain stops here
 501                 if stop_base_path:
 502                     no_ancestry = True
 503                 ui.status(prefix + ">> find_svn_ancestors: Done: deleted", level=ui.DEBUG, color='YELLOW')
 504                 done = True
 505                 break
 506             if action in 'RA':
 507                 # If file/folder was added/replaced but not a copy, ancestry-chain stops here
 508                 if not d['copyfrom_path']:
 509                     if stop_base_path:
 510                         no_ancestry = True
 511                     ui.status(prefix + ">> find_svn_ancestors: Done: %s with no copyfrom_path",
 512                         "Added" if action == "A" else "Replaced",
 513                         level=ui.DEBUG, color='YELLOW')
 514                     done = True
 515                     break
 516                 # Else, file/folder was added/replaced and is a copy, so add an entry to our ancestors list
 517                 # and keep checking for ancestors
 518                 ui.status(prefix + ">> find_svn_ancestors: Found copy-from (action=%s): %s --> %s",
 519                     action, path, d['copyfrom_path']+"@"+str(d['copyfrom_revision']),
 520                     level=ui.DEBUG, color='YELLOW')
 521                 ancestors.append({'path': cur_path, 'revision': log_entry['revision'],
 522                     'copyfrom_path': cur_path.replace(d['path'], d['copyfrom_path']), 'copyfrom_rev': d['copyfrom_revision']})
 523                 cur_path = cur_path.replace(d['path'], d['copyfrom_path'])
 524                 cur_rev =  d['copyfrom_revision']
 525                 # Follow the copy and keep on searching
 526                 break
 527     if stop_base_path and no_ancestry:
 528         # If we're tracing back ancestry to a specific target stop_base_path and
 529         # the ancestry-chain stopped before we reached stop_base_path, then return
 530         # nothing since there is no ancestry chaining back to that target.
 531         ancestors = []
 532     if ancestors:
 533         if ui.get_level() >= ui.DEBUG:
 534             max_len = 0
 535             for idx in range(len(ancestors)):
 536                 d = ancestors[idx]
 537                 max_len = max(max_len, len(d['path']+"@"+str(d['revision'])))
 538             ui.status(prefix + ">> find_svn_ancestors: Found parent ancestors:", level=ui.DEBUG, color='YELLOW_B')
 539             for idx in range(len(ancestors)):
 540                 d = ancestors[idx]
 541                 ui.status(prefix + " [%s] %s --> %s", idx,
 542                     str(d['path']+"@"+str(d['revision'])).ljust(max_len),
 543                     str(d['copyfrom_path']+"@"+str(d['copyfrom_rev'])),
 544                     level=ui.DEBUG, color='YELLOW')
 545     else:
 546         ui.status(prefix + ">> find_svn_ancestors: No ancestor-chain found: %s",
 547             svn_repos_url+start_path+"@"+str(start_rev), level=ui.DEBUG, color='YELLOW')
 548     return ancestors
 549
 550 def get_rev_map(source_rev, prefix):
 551     """
 552     Find the equivalent rev # in the target repo for the given rev # from the source repo.
 553     """
 554     ui.status(prefix + ">> get_rev_map(%s)", source_rev, level=ui.DEBUG, color='GREEN')
 555     # Find the highest entry less-than-or-equal-to source_rev
 556     for rev in range(int(source_rev), 0, -1):
 557         in_rev_map = True if rev in rev_map else False
 558         ui.status(prefix + ">> get_rev_map: rev=%s  in_rev_map=%s", rev, str(in_rev_map), level=ui.DEBUG, color='BLACK_B')
 559         if in_rev_map:
 560             return int(rev_map[rev])
 561     # Else, we fell off the bottom of the rev_map. Ruh-roh...
 562     return None
 563
 564 def set_rev_map(source_rev, target_rev):
 565     #ui.status(">> set_rev_map: source_rev=%s target_rev=%s", source_rev, target_rev, level=ui.DEBUG, color='GREEN')
 566     global rev_map
 567     rev_map[int(source_rev)]=int(target_rev)
 568
 569 def build_rev_map(target_url, target_end_rev, source_info):
 570     """
 571     Check for any already-replayed history from source_url (source_info) and
 572     build the mapping-table of source_rev -> target_rev.
 573     """
 574     global rev_map
 575     rev_map = {}
 576     ui.status("Rebuilding target_rev -> source_rev rev_map...", level=ui.VERBOSE)
 577     proc_count = 0
 578     it_log_entries = svnclient.iter_svn_log_entries(target_url, 1, target_end_rev, get_changed_paths=False, get_revprops=True)
 579     for log_entry in it_log_entries:
 580         if log_entry['revprops']:
 581             revprops = {}
 582             for v in log_entry['revprops']:
 583                 if v['name'].startswith('svn2svn:'):
 584                     revprops[v['name']] = v['value']
 585             if revprops and \
 586                revprops['svn2svn:source_uuid'] == source_info['repos_uuid'] and \
 587                revprops['svn2svn:source_url'] == source_info['url']:
 588                 source_rev = revprops['svn2svn:source_rev']
 589                 target_rev = log_entry['revision']
 590                 set_rev_map(source_rev, target_rev)
 591                 proc_count += 1
 592                 if proc_count % 500 == 0:
 593                     ui.status("...processed %s (%s of %s)..." % (proc_count, target_rev, target_end_rev), level=ui.VERBOSE)
 594
 595 def get_svn_dirlist(svn_path, rev_number = ""):
 596     """
 597     Get a list of all the child contents (recusive) of the given folder path.
 598     """
 599     args = ["list"]
 600     path = svn_path
 601     if rev_number:
 602         args += ["-r", rev_number]
 603         path += "@"+str(rev_number)
 604     args += [path]
 605     paths = run_svn(args, no_fail=True)
 606     paths = paths.strip("\n").split("\n") if len(paths)>1 else []
 607     return paths
 608
 609 def path_in_list(paths, path):
 610     for p in paths:
 611         if is_child_path(path, p):
 612             return True
 613     return False
 614
 615 def add_path(paths, path):
 616     if not path_in_list(paths, path):
 617         paths.append(path)
 618
 619 def in_ancestors(ancestors, ancestor):
 620     match = True
 621     for idx in range(len(ancestors)-1, 0, -1):
 622         if int(ancestors[idx]['revision']) > ancestor['revision']:
 623             match = is_child_path(ancestor['path'], ancestors[idx]['path'])
 624             break
 625     return match
 626
 627 def do_svn_add(source_url, path_offset, source_rev, source_ancestors, \
 628                parent_copyfrom_path="", parent_copyfrom_rev="", \
 629                export_paths={}, is_dir = False, skip_paths=[], prefix = ""):
 630     """
 631     Given the add'd source path, replay the "svn add/copy" commands to correctly
 632     track renames across copy-from's.
 633
 634     For example, consider a sequence of events like this:
 635     1. svn copy /trunk /branches/fix1
 636     2. (Make some changes on /branches/fix1)
 637     3. svn mv /branches/fix1/Proj1 /branches/fix1/Proj2  " Rename folder
 638     4. svn mv /branches/fix1/Proj2/file1.txt /branches/fix1/Proj2/file2.txt  " Rename file inside renamed folder
 639     5. svn co /trunk && svn merge /branches/fix1
 640     After the merge and commit, "svn log -v" with show a delete of /trunk/Proj1
 641     and and add of /trunk/Proj2 copy-from /branches/fix1/Proj2. If we were just
 642     to do a straight "svn export+add" based on the /branches/fix1/Proj2 folder,
 643     we'd lose the logical history that Proj2/file2.txt is really a descendant
 644     of Proj1/file1.txt.
 645
 646     'path_offset' is the offset from source_base to the file to check ancestry for,
 647       e.g. 'projectA/file1.txt'. path = source_repos_url + source_base + path_offset.
 648     'source_rev' is the revision ("svn log") that we're processing from the source repo.
 649     'parent_copyfrom_path' and 'parent_copyfrom_rev' is the copy-from path of the parent
 650       directory, when being called recursively by do_svn_add_dir().
 651     'export_paths' is the list of path_offset's that we've deferred running "svn export" on.
 652     'is_dir' is whether path_offset is a directory (rather than a file).
 653     """
 654     source_base = source_url[len(source_repos_url):]  # e.g. '/trunk'
 655     ui.status(prefix + ">> do_svn_add: %s  %s", join_path(source_base, path_offset)+"@"+str(source_rev),
 656         "  (parent-copyfrom: "+parent_copyfrom_path+"@"+str(parent_copyfrom_rev)+")" if parent_copyfrom_path else "",
 657         level=ui.DEBUG, color='GREEN')
 658     # Check if the given path has ancestors which chain back to the current source_base
 659     found_ancestor = False
 660     ancestors = find_svn_ancestors(source_repos_url, join_path(source_base, path_offset), source_rev, stop_base_path=source_base, prefix=prefix+"  ")
 661     ancestor = ancestors[len(ancestors)-1] if ancestors else None  # Choose the eldest ancestor, i.e. where we reached stop_base_path=source_base
 662     if ancestor and not in_ancestors(source_ancestors, ancestor):
 663         ancestor = None
 664     copyfrom_path = ancestor['copyfrom_path'] if ancestor else ""
 665     copyfrom_rev  = ancestor['copyfrom_rev']  if ancestor else ""
 666     if ancestor:
 667         # The copy-from path has ancestry back to source_url.
 668         ui.status(prefix + ">> do_svn_add: Check copy-from: Found parent: %s", copyfrom_path+"@"+str(copyfrom_rev),
 669             level=ui.DEBUG, color='GREEN', bold=True)
 670         found_ancestor = True
 671         # Map the copyfrom_rev (source repo) to the equivalent target repo rev #. This can
 672         # return None in the case where copyfrom_rev is *before* our source_start_rev.
 673         tgt_rev = get_rev_map(copyfrom_rev, prefix+"  ")
 674         ui.status(prefix + ">> do_svn_add: get_rev_map: %s (source) -> %s (target)", copyfrom_rev, tgt_rev, level=ui.DEBUG, color='GREEN')
 675     else:
 676         ui.status(prefix + ">> do_svn_add: Check copy-from: No ancestor chain found.", level=ui.DEBUG, color='GREEN')
 677         found_ancestor = False
 678     if found_ancestor and tgt_rev:
 679         # Check if this path_offset in the target WC already has this ancestry, in which
 680         # case there's no need to run the "svn copy" (again).
 681         path_in_svn = in_svn(path_offset, prefix=prefix+"  ")
 682         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 []
 683         if (not log_entry or (log_entry['revision'] != tgt_rev)):
 684             copyfrom_offset = copyfrom_path[len(source_base):].strip('/')
 685             ui.status(prefix + ">> do_svn_add: svn_copy: Copy-from: %s", copyfrom_path+"@"+str(copyfrom_rev), level=ui.DEBUG, color='GREEN')
 686             ui.status(prefix + "   copyfrom: %s", copyfrom_path+"@"+str(copyfrom_rev), level=ui.DEBUG, color='GREEN')
 687             ui.status(prefix + " p_copyfrom: %s", parent_copyfrom_path+"@"+str(parent_copyfrom_rev) if parent_copyfrom_path else "", level=ui.DEBUG, color='GREEN')
 688             if path_in_svn and \
 689                ((parent_copyfrom_path and is_child_path(copyfrom_path, parent_copyfrom_path)) and \
 690                 (parent_copyfrom_rev and copyfrom_rev == parent_copyfrom_rev)):
 691                 # When being called recursively, if this child entry has the same ancestor as the
 692                 # the parent, then no need to try to run another "svn copy".
 693                 ui.status(prefix + ">> do_svn_add: svn_copy: Same ancestry as parent: %s",
 694                     parent_copyfrom_path+"@"+str(parent_copyfrom_rev),level=ui.DEBUG, color='GREEN')
 695                 pass
 696             else:
 697                 # Copy this path from the equivalent path+rev in the target repo, to create the
 698                 # equivalent history.
 699                 if parent_copyfrom_path:
 700                     # If we have a parent copy-from path, we mis-match that so display a status
 701                     # message describing the action we're mimic'ing. If path_in_svn, then this
 702                     # is logically a "replace" rather than an "add".
 703                     ui.status(" %s %s (from %s)", ('R' if path_in_svn else 'A'), join_path(source_base, path_offset), ancestors[0]['copyfrom_path']+"@"+str(copyfrom_rev), level=ui.VERBOSE)
 704                 if path_in_svn:
 705                     # If local file is already under version-control, then this is a replace.
 706                     ui.status(prefix + ">> do_svn_add: pre-copy: local path already exists: %s", path_offset, level=ui.DEBUG, color='GREEN')
 707                     run_svn(["update", path_offset])
 708                     run_svn(["remove", "--force", path_offset])
 709                 run_svn(["copy", "-r", tgt_rev, join_path(target_url, copyfrom_offset)+"@"+str(tgt_rev), path_offset])
 710                 if is_dir:
 711                     # Export the final verison of all files in this folder.
 712                     add_path(export_paths, path_offset)
 713                 else:
 714                     # Export the final verison of this file.
 715                     run_svn(["export", "--force", "-r", source_rev,
 716                              source_repos_url+join_path(source_base, path_offset)+"@"+str(source_rev), path_offset])
 717                 if options.keep_prop:
 718                     sync_svn_props(source_url, source_rev, path_offset)
 719         else:
 720             ui.status(prefix + ">> do_svn_add: Skipped 'svn copy': %s", path_offset, level=ui.DEBUG, color='GREEN')
 721     else:
 722         # Else, either this copy-from path has no ancestry back to source_url OR copyfrom_rev comes
 723         # before our initial source_start_rev (i.e. tgt_rev == None), so can't do a "svn copy".
 724         # Create (parent) directory if needed.
 725         # TODO: This is (nearly) a duplicate of code in process_svn_log_entry(). Should this be
 726         #       split-out to a shared tag?
 727         p_path = path_offset if is_dir else os.path.dirname(path_offset).strip() or None
 728         if p_path and not os.path.exists(p_path):
 729             run_svn(["mkdir", p_path])
 730         if not in_svn(path_offset, prefix=prefix+"  "):
 731             if is_dir:
 732                 # Export the final verison of all files in this folder.
 733                 add_path(export_paths, path_offset)
 734             else:
 735                 # Export the final verison of this file. We *need* to do this before running
 736                 # the "svn add", even if we end-up re-exporting this file again via export_paths.
 737                 run_svn(["export", "--force", "-r", source_rev,
 738                          source_repos_url+join_path(source_base, path_offset)+"@"+str(source_rev), path_offset])
 739             # If not already under version-control, then "svn add" this file/folder.
 740             run_svn(["add", "--parents", path_offset])
 741         if options.keep_prop:
 742             sync_svn_props(source_url, source_rev, path_offset)
 743     if is_dir:
 744         # For any folders that we process, process any child contents, so that we correctly
 745         # replay copies/replaces/etc.
 746         do_svn_add_dir(source_url, path_offset, source_rev, source_ancestors,
 747                        copyfrom_path, copyfrom_rev, export_paths, skip_paths, prefix+"  ")
 748
 749 def do_svn_add_dir(source_url, path_offset, source_rev, source_ancestors, \
 750                    parent_copyfrom_path, parent_copyfrom_rev, \
 751                    export_paths, skip_paths, prefix=""):
 752     source_base = source_url[len(source_repos_url):]  # e.g. '/trunk'
 753     # Get the directory contents, to compare between the local WC (target_url) vs. the remote repo (source_url)
 754     # TODO: paths_local won't include add'd paths because "svn ls" lists the contents of the
 755     #       associated remote repo folder. (Is this a problem?)
 756     paths_local =  get_svn_dirlist(path_offset)
 757     paths_remote = get_svn_dirlist(join_path(source_url, path_offset), source_rev)
 758     ui.status(prefix + ">> do_svn_add_dir: paths_local:  %s", str(paths_local),  level=ui.DEBUG, color='GREEN')
 759     ui.status(prefix + ">> do_svn_add_dir: paths_remote: %s", str(paths_remote), level=ui.DEBUG, color='GREEN')
 760     # Update files/folders which exist in remote but not local
 761     for path in paths_remote:
 762         path_is_dir = True if path[-1] == "/" else False
 763         working_path = join_path(path_offset, (path.rstrip('/') if path_is_dir else path)).lstrip('/')
 764         #print "working_path:%s = path_offset:%s + path:%s" % (working_path, path_offset, path)
 765         if not working_path in skip_paths:
 766             do_svn_add(source_url, working_path, source_rev, source_ancestors,
 767                        parent_copyfrom_path, parent_copyfrom_rev,
 768                        export_paths, path_is_dir, skip_paths, prefix+"  ")
 769     # Remove files/folders which exist in local but not remote
 770     for path in paths_local:
 771         if not path in paths_remote:
 772             path_is_dir = True if path[-1] == "/" else False
 773             working_path = join_path(path_offset, (path.rstrip('/') if path_is_dir else path)).lstrip('/')
 774             ui.status(" %s %s", 'D', join_path(source_base, working_path), level=ui.VERBOSE)
 775             run_svn(["update", working_path])
 776             run_svn(["remove", "--force", working_path])
 777             # TODO: Does this handle deleted folders too? Wouldn't want to have a case
 778             #       where we only delete all files from folder but leave orphaned folder around.
 779
 780 def process_svn_log_entry(log_entry, ancestors, commit_paths, prefix = ""):
 781     """
 782     Process SVN changes from the given log entry. Build an array (commit_paths)
 783     of the paths in the working-copy that were changed, i.e. the paths which
 784     we'll pass to "svn commit".
 785     """
 786     export_paths = []
 787     source_rev = log_entry['revision']
 788     source_url = log_entry['url']
 789     source_base = source_url[len(source_repos_url):]  # e.g. '/trunk'
 790     ui.status(prefix + ">> process_svn_log_entry: %s", source_url+"@"+str(source_rev), level=ui.DEBUG, color='GREEN')
 791     for d in log_entry['changed_paths']:
 792         # Get the full path for this changed_path
 793         # e.g. '/branches/bug123/projectA/file1.txt'
 794         path = d['path']
 795         if not is_child_path(path, source_base):
 796             # Ignore changed files that are not part of this subdir
 797             ui.status(prefix + ">> process_svn_log_entry: Unrelated path: %s  (base: %s)", path, source_base, level=ui.DEBUG, color='GREEN')
 798             continue
 799         if d['kind'] == "" or d['kind'] == 'none':
 800             # The "kind" value was introduced in SVN 1.6, and "svn log --xml" won't return a "kind"
 801             # value for commits made on a pre-1.6 repo, even if the server is now running 1.6.
 802             # We need to use other methods to fetch the node-kind for these cases.
 803             d['kind'] = svnclient.get_kind(source_repos_url, path, source_rev, d['action'], log_entry['changed_paths'])
 804         assert (d['kind'] == 'file') or (d['kind'] == 'dir')
 805         path_is_dir =  True if d['kind'] == 'dir'  else False
 806         path_is_file = True if d['kind'] == 'file' else False
 807         # Calculate the offset (based on source_base) for this changed_path
 808         # e.g. 'projectA/file1.txt'
 809         # (path = source_base + "/" + path_offset)
 810         path_offset = path[len(source_base):].strip("/")
 811         # Get the action for this path
 812         action = d['action']
 813         if action not in _valid_svn_actions:
 814             raise UnsupportedSVNAction("In SVN rev. %d: action '%s' not supported. Please report a bug!"
 815                 % (source_rev, action))
 816         ui.status(" %s %s%s", action, d['path'],
 817             (" (from %s)" % (d['copyfrom_path']+"@"+str(d['copyfrom_revision']))) if d['copyfrom_path'] else "",
 818             level=ui.VERBOSE)
 819
 820         # Try to be efficient and keep track of an explicit list of paths in the
 821         # working copy that changed. If we commit from the root of the working copy,
 822         # then SVN needs to crawl the entire working copy looking for pending changes.
 823         commit_paths.append(path_offset)
 824
 825         # Special-handling for replace's
 826         if action == 'R':
 827             # If file was "replaced" (deleted then re-added, all in same revision),
 828             # then we need to run the "svn rm" first, then change action='A'. This
 829             # lets the normal code below handle re-"svn add"'ing the files. This
 830             # should replicate the "replace".
 831             if path_offset and in_svn(path_offset):
 832                 # Target path might not be under version-control yet, e.g. parent "add"
 833                 # was a copy-from a branch which had no ancestry back to trunk, and each
 834                 # child folder under that parent folder is a "replace" action on the final
 835                 # merge to trunk. Since the child folders will be in skip_paths, do_svn_add
 836                 # wouldn't have created them while processing the parent "add" path.
 837                 if path_is_dir:
 838                     # Need to "svn update" before "svn remove" in case child contents are at
 839                     # a higher rev than the (parent) path_offset.
 840                     run_svn(["update", path_offset])
 841                 run_svn(["remove", "--force", path_offset])
 842             action = 'A'
 843
 844         # Handle all the various action-types
 845         # (Handle "add" first, for "svn copy/move" support)
 846         if action == 'A':
 847             # Determine where to export from.
 848             svn_copy = False
 849             # Handle cases where this "add" was a copy from another URL in the source repo
 850             if d['copyfrom_revision']:
 851                 copyfrom_path = d['copyfrom_path']
 852                 copyfrom_rev =  d['copyfrom_revision']
 853                 skip_paths = []
 854                 for tmp_d in log_entry['changed_paths']:
 855                     tmp_path = tmp_d['path']
 856                     if is_child_path(tmp_path, path) and tmp_d['action'] in 'ARD':
 857                         # Build list of child entries which are also in the changed_paths list,
 858                         # so that do_svn_add() can skip processing these entries when recursing
 859                         # since we'll end-up processing them later. Don't include action="M" paths
 860                         # in this list because it's non-conclusive: it could just mean that the
 861                         # file was modified *after* the copy-from, so we still want do_svn_add()
 862                         # to re-create the correct ancestry.
 863                         tmp_path_offset = tmp_path[len(source_base):].strip("/")
 864                         skip_paths.append(tmp_path_offset)
 865                 do_svn_add(source_url, path_offset, source_rev, ancestors, "", "", export_paths, path_is_dir, skip_paths, prefix+"  ")
 866             # Else just "svn export" the files from the source repo and "svn add" them.
 867             else:
 868                 # Create (parent) directory if needed
 869                 p_path = path_offset if path_is_dir else os.path.dirname(path_offset).strip() or None
 870                 if p_path and not os.path.exists(p_path):
 871                     run_svn(["mkdir", p_path])
 872                 # Export the entire added tree.
 873                 if path_is_dir:
 874                     # For directories, defer the (recurisve) "svn export". Might have a
 875                     # situation in a branch merge where the entry in the svn-log is a
 876                     # non-copy-from'd "add" but there are child contents (that we haven't
 877                     # gotten to yet in log_entry) that are copy-from's.  When we try do
 878                     # the "svn copy" later on in do_svn_add() for those copy-from'd paths,
 879                     # having pre-existing (svn-add'd) contents creates some trouble.
 880                     # Instead, just create the stub folders ("svn mkdir" above) and defer
 881                     # exporting the final file-state until the end.
 882                     add_path(export_paths, path_offset)
 883                 else:
 884                     # Export the final verison of this file. We *need* to do this before running
 885                     # the "svn add", even if we end-up re-exporting this file again via export_paths.
 886                     run_svn(["export", "--force", "-r", source_rev,
 887                              join_path(source_url, path_offset)+"@"+str(source_rev), path_offset])
 888                 if not in_svn(path_offset, prefix=prefix+"  "):
 889                     # Need to use in_svn here to handle cases where client committed the parent
 890                     # folder and each indiv sub-folder.
 891                     run_svn(["add", "--parents", path_offset])
 892                 if options.keep_prop:
 893                     sync_svn_props(source_url, source_rev, path_offset)
 894
 895         elif action == 'D':
 896             if path_is_dir:
 897                 # For dirs, need to "svn update" before "svn remove" because the final
 898                 # "svn commit" will fail if the parent (path_offset) is at a lower rev
 899                 # than any of the child contents. This needs to be a recursive update.
 900                 run_svn(["update", path_offset])
 901             run_svn(["remove", "--force", path_offset])
 902
 903         elif action == 'M':
 904             if path_is_file:
 905                 run_svn(["export", "--force", "-N" , "-r", source_rev,
 906                          join_path(source_url, path_offset)+"@"+str(source_rev), path_offset])
 907             if path_is_dir:
 908                 # For dirs, need to "svn update" before export/prop-sync because the
 909                 # final "svn commit" will fail if the parent is at a lower rev than
 910                 # child contents. Just need to update the rev-state of the dir (d['path']),
 911                 # don't need to recursively update all child contents.
 912                 # (??? is this the right reason?)
 913                 run_svn(["update", "-N", path_offset])
 914             if options.keep_prop:
 915                 sync_svn_props(source_url, source_rev, path_offset)
 916
 917         else:
 918             raise InternalError("Internal Error: process_svn_log_entry: Unhandled 'action' value: '%s'"
 919                 % action)
 920
 921     # Export the final version of all add'd paths from source_url
 922     if export_paths:
 923         for path_offset in export_paths:
 924             run_svn(["export", "--force", "-r", source_rev,
 925                      join_path(source_url, path_offset)+"@"+str(source_rev), path_offset])
 926
 927 def keep_revnum(source_rev, target_rev_last, wc_target_tmp):
 928     """
 929     Add "padding" target revisions as needed to keep source and target
 930     revision #'s identical.
 931     """
 932     bh = BreakHandler()
 933     if int(source_rev) <= int(target_rev_last):
 934         raise InternalError("keep-revnum mode is enabled, "
 935             "but source revision (r%s) is less-than-or-equal last target revision (r%s)" % \
 936             (source_rev, target_rev_last))
 937     if int(target_rev_last) < int(source_rev)-1:
 938         # Add "padding" target revisions to keep source and target rev #'s identical
 939         if os.path.exists(wc_target_tmp):
 940             shutil.rmtree(wc_target_tmp)
 941         run_svn(["checkout", "-r", "HEAD", "--depth=empty", target_repos_url, wc_target_tmp])
 942         for rev_num in range(int(target_rev_last)+1, int(source_rev)):
 943             run_svn(["propset", "svn2svn:keep-revnum", rev_num, wc_target_tmp])
 944             # Prevent Ctrl-C's during this inner part, so we'll always display
 945             # the "Commit revision ..." message if we ran a "svn commit".
 946             bh.enable()
 947             output = run_svn(["commit", "-m", "", wc_target_tmp])
 948             rev_num_tmp = parse_svn_commit_rev(output) if output else None
 949             assert rev_num == rev_num_tmp
 950             ui.status("Committed revision %s (keep-revnum).", rev_num)
 951             bh.disable()
 952             # Check if the user tried to press Ctrl-C
 953             if bh.trapped:
 954                 raise KeyboardInterrupt
 955             target_rev_last = rev_num
 956         shutil.rmtree(wc_target_tmp)
 957     return target_rev_last
 958
 959 def disp_svn_log_summary(log_entry):
 960     ui.status("------------------------------------------------------------------------", level=ui.VERBOSE)
 961     ui.status("r%s | %s | %s",
 962         log_entry['revision'],
 963         log_entry['author'],
 964         str(datetime.fromtimestamp(int(log_entry['date'])).isoformat(' ')), level=ui.VERBOSE)
 965     ui.status(log_entry['message'], level=ui.VERBOSE)
 966
 967 def real_main(args):
 968     global source_url, target_url, rev_map
 969     source_url = urllib.quote(args.pop(0).rstrip("/"),"/:")   # e.g. 'http://server/svn/source/trunk'
 970     target_url = urllib.quote(args.pop(0).rstrip("/"),"/:")   # e.g. 'file:///svn/target/trunk'
 971     ui.status("options: %s", str(options), level=ui.DEBUG, color='GREEN')
 972
 973     # Make sure that both the source and target URL's are valid
 974     source_info = svnclient.get_svn_info(source_url)
 975     assert is_child_path(source_url, source_info['repos_url'])
 976     target_info = svnclient.get_svn_info(target_url)
 977     assert is_child_path(target_url, target_info['repos_url'])
 978
 979     # Init global vars
 980     global source_repos_url,source_base,source_repos_uuid
 981     source_repos_url = source_info['repos_url']       # e.g. 'http://server/svn/source'
 982     source_base = source_url[len(source_repos_url):]  # e.g. '/trunk'
 983     source_repos_uuid = source_info['repos_uuid']
 984     global target_repos_url,target_base
 985     target_repos_url = target_info['repos_url']       # e.g. 'http://server/svn/target'
 986     target_base = target_url[len(target_repos_url):]  # e.g. '/trunk'
 987
 988     # Init start and end revision
 989     try:
 990         source_start_rev = svnclient.get_svn_rev(source_repos_url, options.rev_start if options.rev_start else 1)
 991     except ExternalCommandFailed:
 992         print "Error: Invalid start source revision value: %s" % (options.rev_start)
 993         sys.exit(1)
 994     try:
 995         source_end_rev   = svnclient.get_svn_rev(source_repos_url, options.rev_end   if options.rev_end   else "HEAD")
 996     except ExternalCommandFailed:
 997         print "Error: Invalid end source revision value: %s" % (options.rev_end)
 998         sys.exit(1)
 999     ui.status("Using source revision range %s:%s", source_start_rev, source_end_rev, level=ui.VERBOSE)
1000
1001     # TODO: If options.keep_date, should we try doing a "svn propset" on an *existing* revision
1002     #       as a sanity check, so we check if the pre-revprop-change hook script is correctly setup
1003     #       before doing first replay-commit?
1004
1005     target_rev_last =  target_info['revision']   # Last revision # in the target repo
1006     wc_target = os.path.abspath('_wc_target')
1007     wc_target_tmp = os.path.abspath('_wc_target_tmp')
1008     num_entries_proc = 0
1009     commit_count = 0
1010     source_rev = None
1011     target_rev = None
1012
1013     # Check out a working copy of target_url if needed
1014     wc_exists = os.path.exists(wc_target)
1015     if wc_exists and not options.cont_from_break:
1016         shutil.rmtree(wc_target)
1017         wc_exists = False
1018     if not wc_exists:
1019         ui.status("Checking-out _wc_target...", level=ui.VERBOSE)
1020         svnclient.svn_checkout(target_url, wc_target)
1021     os.chdir(wc_target)
1022     if wc_exists:
1023         # If using an existing WC, make sure it's clean ("svn revert")
1024         ui.status("Cleaning-up _wc_target...", level=ui.VERBOSE)
1025         run_svn(["cleanup"])
1026         full_svn_revert()
1027
1028     if not options.cont_from_break:
1029         # Warn user if trying to start (non-continue) into a non-empty target path
1030         if not options.force_nocont:
1031             top_paths = run_svn(["list", "-r", "HEAD", target_url])
1032             if len(top_paths)>0:
1033                 print "Error: Trying to replay (non-continue-mode) into a non-empty target_url location. " \
1034                       "Use --force if you're sure this is what you want."
1035                 sys.exit(1)
1036         # Get the first log entry at/after source_start_rev, which is where
1037         # we'll do the initial import from.
1038         source_ancestors = find_svn_ancestors(source_repos_url, source_base, source_end_rev, prefix="  ")
1039         it_log_start = svnclient.iter_svn_log_entries(source_url, source_start_rev, source_end_rev, get_changed_paths=False, ancestors=source_ancestors)
1040         source_start_log = None
1041         for log_entry in it_log_start:
1042             # Pick the first entry. Need to use a "for ..." loop since we're using an iterator.
1043             source_start_log = log_entry
1044             break
1045         if not source_start_log:
1046             raise InternalError("Unable to find any matching revisions between %s:%s in source_url: %s" % \
1047                 (source_start_rev, source_end_rev, source_url))
1048
1049         # This is the revision we will start from for source_url
1050         source_start_rev = int(source_start_log['revision'])
1051         ui.status("Starting at source revision %s.", source_start_rev, level=ui.VERBOSE)
1052         ui.status("", level=ui.VERBOSE)
1053         if options.keep_revnum and source_rev > target_rev_last:
1054             target_rev_last = keep_revnum(source_rev, target_rev_last, wc_target_tmp)
1055
1056         # For the initial commit to the target URL, export all the contents from
1057         # the source URL at the start-revision.
1058         disp_svn_log_summary(svnclient.get_one_svn_log_entry(source_repos_url, source_start_rev, source_start_rev))
1059         # Export and add file-contents from source_url@source_start_rev
1060         source_start_url = source_url if not source_ancestors else source_repos_url+source_ancestors[len(source_ancestors)-1]['copyfrom_path']
1061         top_paths = run_svn(["list", "-r", source_start_rev, source_start_url+"@"+str(source_start_rev)])
1062         top_paths = top_paths.strip("\n").split("\n")
1063         for path in top_paths:
1064             # For each top-level file/folder...
1065             if not path:
1066                 continue
1067             # Directories have a trailing slash in the "svn list" output
1068             path_is_dir = True if path[-1] == "/" else False
1069             path_offset = path.rstrip('/') if path_is_dir else path
1070             if in_svn(path_offset, prefix="  "):
1071                 raise InternalError("Cannot replay history on top of pre-existing structure: %s" % join_path(source_start_url, path_offset))
1072             if path_is_dir and not os.path.exists(path_offset):
1073                 os.makedirs(path_offset)
1074             run_svn(["export", "--force", "-r" , source_start_rev, join_path(source_start_url, path_offset)+"@"+str(source_start_rev), path_offset])
1075             run_svn(["add", path_offset])
1076         # Update any properties on the newly added content
1077         paths = run_svn(["list", "--recursive", "-r", source_start_rev, source_start_url+"@"+str(source_start_rev)])
1078         paths = paths.strip("\n").split("\n")
1079         if options.keep_prop:
1080             sync_svn_props(source_start_url, source_start_rev, "")
1081         for path in paths:
1082             if not path:
1083                 continue
1084             # Directories have a trailing slash in the "svn list" output
1085             path_is_dir = True if path[-1] == "/" else False
1086             path_offset = path.rstrip('/') if path_is_dir else path
1087             ui.status(" A %s", join_path(source_base, path_offset), level=ui.VERBOSE)
1088             if options.keep_prop:
1089                 sync_svn_props(source_start_url, source_start_rev, path_offset)
1090         # Commit the initial import
1091         num_entries_proc += 1
1092         target_revprops = gen_tracking_revprops(source_start_rev)   # Build source-tracking revprop's
1093         target_rev = commit_from_svn_log_entry(source_start_log, target_revprops=target_revprops)
1094         if target_rev:
1095             # Update rev_map, mapping table of source-repo rev # -> target-repo rev #
1096             set_rev_map(source_start_rev, target_rev)
1097             commit_count += 1
1098             target_rev_last = target_rev
1099             if options.verify:
1100                 verify_commit(source_rev, target_rev_last)
1101     else:
1102         # Re-build the rev_map based on any already-replayed history in target_url
1103         build_rev_map(target_url, target_rev_last, source_info)
1104         if not rev_map:
1105             print "Error: Called with continue-mode, but no already-replayed source history found in target_url."
1106             sys.exit(1)
1107         source_start_rev = int(max(rev_map, key=rev_map.get))
1108         assert source_start_rev
1109         ui.status("Continuing from source revision %s.", source_start_rev, level=ui.VERBOSE)
1110         ui.status("", level=ui.VERBOSE)
1111
1112     svn_vers_t = svnclient.get_svn_client_version()
1113     svn_vers = float(".".join(map(str, svn_vers_t[0:2])))
1114
1115     # Load SVN log starting from source_start_rev + 1
1116     source_ancestors = find_svn_ancestors(source_repos_url, source_base, source_end_rev, prefix="  ")
1117     it_log_entries = svnclient.iter_svn_log_entries(source_url, source_start_rev+1, source_end_rev, get_revprops=True, ancestors=source_ancestors) if source_start_rev < source_end_rev else []
1118     source_rev_last = source_start_rev
1119
1120     try:
1121         for log_entry in it_log_entries:
1122             if options.entries_proc_limit:
1123                 if num_entries_proc >= options.entries_proc_limit:
1124                     break
1125             # Replay this revision from source_url into target_url
1126             source_rev = log_entry['revision']
1127             log_url =    log_entry['url']
1128             #print "source_url:%s  log_url:%s" % (source_url, log_url)
1129             if options.keep_revnum:
1130                 if source_rev < target_rev_last:
1131                     print "Error: Last target revision (r%s) is equal-or-higher than starting source revision (r%s). " \
1132                         "Cannot use --keep-revnum mode." % (target_rev_last, source_start_rev)
1133                     sys.exit(1)
1134                 target_rev_last = keep_revnum(source_rev, target_rev_last, wc_target_tmp)
1135             disp_svn_log_summary(log_entry)
1136             # Process all the changed-paths in this log entry
1137             commit_paths = []
1138             process_svn_log_entry(log_entry, source_ancestors, commit_paths)
1139             num_entries_proc += 1
1140             # Commit any changes made to _wc_target
1141             target_revprops = gen_tracking_revprops(source_rev)   # Build source-tracking revprop's
1142             target_rev = commit_from_svn_log_entry(log_entry, commit_paths, target_revprops=target_revprops)
1143             source_rev_last = source_rev
1144             if target_rev:
1145                 # Update rev_map, mapping table of source-repo rev # -> target-repo rev #
1146                 source_rev = log_entry['revision']
1147                 set_rev_map(source_rev, target_rev)
1148                 target_rev_last = target_rev
1149                 commit_count += 1
1150                 if options.verify:
1151                     verify_commit(source_rev, target_rev_last, log_entry)
1152                 # Run "svn cleanup" every 100 commits if SVN 1.7+, to clean-up orphaned ".svn/pristines/*"
1153                 if svn_vers >= 1.7 and (commit_count % 100 == 0):
1154                     run_svn(["cleanup"])
1155         if source_rev_last == source_start_rev:
1156             # If there were no new source_url revisions to process, still trigger
1157             # "full-mode" verify check (if enabled).
1158             if options.verify:
1159                 verify_commit(source_rev_last, target_rev_last)
1160
1161     except KeyboardInterrupt:
1162         print "\nStopped by user."
1163         print "\nCleaning-up..."
1164         run_svn(["cleanup"])
1165         full_svn_revert()
1166     except:
1167         print "\nCommand failed with following error:\n"
1168         traceback.print_exc()
1169         print "\nCleaning-up..."
1170         run_svn(["cleanup"])
1171         print run_svn(["status"])
1172         full_svn_revert()
1173     finally:
1174         print "\nFinished at source revision %s%s." % (source_rev_last, " (dry-run)" if options.dry_run else "")
1175
1176 def main():
1177     # Defined as entry point. Must be callable without arguments.
1178     usage = "svn2svn, version %s\n" % str(full_version) + \
1179             "<http://nynim.org/projects/svn2svn> <https://github.com/tonyduckles/svn2svn>\n\n" + \
1180             "Usage: %prog [OPTIONS] source_url target_url\n"
1181     description = """\
1182 Replicate (replay) history from one SVN repository to another. Maintain
1183 logical ancestry wherever possible, so that 'svn log' on the replayed repo
1184 will correctly follow file/folder renames.
1185
1186 Examples:
1187   Create a copy of only /trunk from source repo, starting at r5000
1188   $ svnadmin create /svn/target
1189   $ svn mkdir -m 'Add trunk' file:///svn/target/trunk
1190   $ svn2svn -av -r 5000 http://server/source/trunk file:///svn/target/trunk
1191     1. The target_url will be checked-out to ./_wc_target
1192     2. The first commit to http://server/source/trunk at/after r5000 will be
1193        exported & added into _wc_target
1194     3. All revisions affecting http://server/source/trunk (starting at r5000)
1195        will be replayed to _wc_target. Any add/copy/move/replaces that are
1196        copy-from'd some path outside of /trunk (e.g. files renamed on a
1197        /branch and branch was merged into /trunk) will correctly maintain
1198        logical ancestry where possible.
1199
1200   Use continue-mode (-c) to pick-up where the last run left-off
1201   $ svn2svn -avc http://server/source/trunk file:///svn/target/trunk
1202     1. The target_url will be checked-out to ./_wc_target, if not already
1203        checked-out
1204     2. All new revisions affecting http://server/source/trunk starting from
1205        the last replayed revision to file:///svn/target/trunk (based on the
1206        svn2svn:* revprops) will be replayed to _wc_target, maintaining all
1207        logical ancestry where possible."""
1208     parser = optparse.OptionParser(usage, description=description,
1209                 formatter=HelpFormatter(), version="%prog "+str(full_version))
1210     parser.add_option("-v", "--verbose", dest="verbosity", action="count", default=1,
1211                       help="enable additional output (use -vv or -vvv for more)")
1212     parser.add_option("-a", "--archive", action="store_true", dest="archive", default=False,
1213                       help="archive/mirror mode; same as -UDP (see REQUIRE's below)\n"
1214                            "maintain same commit author, same commit time, and file/dir properties")
1215     parser.add_option("-U", "--keep-author", action="store_true", dest="keep_author", default=False,
1216                       help="maintain same commit authors (svn:author) as source\n"
1217                            "(REQUIRES 'pre-revprop-change' hook script to allow 'svn:author' changes)")
1218     parser.add_option("-D", "--keep-date", action="store_true", dest="keep_date", default=False,
1219                       help="maintain same commit time (svn:date) as source\n"
1220                            "(REQUIRES 'pre-revprop-change' hook script to allow 'svn:date' changes)")
1221     parser.add_option("-P", "--keep-prop", action="store_true", dest="keep_prop", default=False,
1222                       help="maintain same file/dir SVN properties as source")
1223     parser.add_option("-R", "--keep-revnum", action="store_true", dest="keep_revnum", default=False,
1224                       help="maintain same rev #'s as source. creates placeholder target "
1225                             "revisions (by modifying a 'svn2svn:keep-revnum' property at the root of the target repo)")
1226     parser.add_option("-c", "--continue", action="store_true", dest="cont_from_break",
1227                       help="continue from last source commit to target (based on svn2svn:* revprops)")
1228     parser.add_option("-f", "--force", action="store_true", dest="force_nocont",
1229                       help="allow replaying into a non-empty target folder")
1230     parser.add_option("-r", "--revision", type="string", dest="revision", metavar="ARG",
1231                       help="revision range to replay from source_url\n"
1232                            "A revision argument can be one of:\n"
1233                            "   START        start rev # (end will be 'HEAD')\n"
1234                            "   START:END    start and ending rev #'s\n"
1235                            "Any revision # formats which SVN understands are "
1236                            "supported, e.g. 'HEAD', '{2010-01-31}', etc.")
1237     parser.add_option("-u", "--log-author", action="store_true", dest="log_author", default=False,
1238                       help="append source commit author to replayed commit mesages")
1239     parser.add_option("-d", "--log-date", action="store_true", dest="log_date", default=False,
1240                       help="append source commit time to replayed commit messages")
1241     parser.add_option("-l", "--limit", type="int", dest="entries_proc_limit", metavar="NUM",
1242                       help="maximum number of source revisions to process")
1243     parser.add_option("-n", "--dry-run", action="store_true", dest="dry_run", default=False,
1244                       help="process next source revision but don't commit changes to "
1245                            "target working-copy (forces --limit=1)")
1246     parser.add_option("-x", "--verify",     action="store_const", const=1, dest="verify",
1247                       help="verify ancestry and content for changed paths in commit after every target commit or last target commit")
1248     parser.add_option("-X", "--verify-all", action="store_const", const=2, dest="verify",
1249                       help="verify ancestry and content for entire target_url tree after every target commit or last target commit")
1250     parser.add_option("--debug", dest="verbosity", const=ui.DEBUG, action="store_const",
1251                       help="enable debugging output (same as -vvv)")
1252     global options
1253     options, args = parser.parse_args()
1254     if len(args) != 2:
1255         parser.error("incorrect number of arguments")
1256     if options.verbosity < 10:
1257         # Expand multiple "-v" arguments to a real ui._level value
1258         options.verbosity *= 10
1259     if options.dry_run:
1260         # When in dry-run mode, only try to process the next log_entry
1261         options.entries_proc_limit = 1
1262     options.rev_start = None
1263     options.rev_end   = None
1264     if options.revision:
1265         # Reg-ex for matching a revision arg (http://svnbook.red-bean.com/en/1.5/svn.tour.revs.specifiers.html#svn.tour.revs.dates)
1266         rev_patt = '[0-9A-Z]+|\{[0-9A-Za-z/\\ :-]+\}'
1267         rev = None
1268         match = re.match('^('+rev_patt+'):('+rev_patt+')$', options.revision)  # First try start:end match
1269         if match is None: match = re.match('^('+rev_patt+')$', options.revision)   # Next, try start match
1270         if match is None:
1271             parser.error("unexpected --revision argument format; see 'svn help log' for valid revision formats")
1272         rev = match.groups()
1273         options.rev_start = rev[0] if len(rev)>0 else None
1274         options.rev_end   = rev[1] if len(rev)>1 else None
1275     if options.archive:
1276         options.keep_author = True
1277         options.keep_date   = True
1278         options.keep_prop   = True
1279     ui.update_config(options)
1280     return real_main(args)
1281
1282
1283 if __name__ == "__main__":
1284     sys.exit(main() or 0)