p4-fast-export.py

   1 #!/usr/bin/python
   2 #
   3 # p4-fast-export.py
   4 #
   5 # Author: Simon Hausmann <hausmann@kde.org>
   6 # License: MIT <http://www.opensource.org/licenses/mit-license.php>
   7 #
   8 # TODO:
   9 #       - support integrations (at least p4i)
  10 #       - support p4 submit (hah!)
  11 #       - emulate p4's delete behavior: if a directory becomes empty delete it. continue
  12 #         with parent dir until non-empty dir is found.
  13 #
  14 import os, string, sys, time, os.path
  15 import marshal, popen2, getopt, sha
  16 from sets import Set;
  17
  18 cacheDebug = False
  19
  20 silent = False
  21 knownBranches = Set()
  22 createdBranches = Set()
  23 committedChanges = Set()
  24 branch = "refs/heads/master"
  25 globalPrefix = previousDepotPath = os.popen("git-repo-config --get p4.depotpath").read()
  26 detectBranches = False
  27 changesFile = ""
  28 if len(globalPrefix) != 0:
  29     globalPrefix = globalPrefix[:-1]
  30
  31 try:
  32     opts, args = getopt.getopt(sys.argv[1:], "", [ "branch=", "detect-branches", "changesfile=", "silent", "known-branches=",
  33                                                    "cache-debug" ])
  34 except getopt.GetoptError:
  35     print "fixme, syntax error"
  36     sys.exit(1)
  37
  38 for o, a in opts:
  39     if o == "--branch":
  40         branch = "refs/heads/" + a
  41     elif o == "--detect-branches":
  42         detectBranches = True
  43     elif o == "--changesfile":
  44         changesFile = a
  45     elif o == "--silent":
  46         silent= True
  47     elif o == "--known-branches":
  48         for branch in open(a).readlines():
  49             knownBranches.add(branch[:-1])
  50     elif o == "--cache-debug":
  51         cacheDebug = True
  52
  53 if len(args) == 0 and len(globalPrefix) != 0:
  54     if not silent:
  55         print "[using previously specified depot path %s]" % globalPrefix
  56 elif len(args) != 1:
  57     print "usage: %s //depot/path[@revRange]" % sys.argv[0]
  58     print "\n    example:"
  59     print "    %s //depot/my/project/ -- to import the current head"
  60     print "    %s //depot/my/project/@all -- to import everything"
  61     print "    %s //depot/my/project/@1,6 -- to import only from revision 1 to 6"
  62     print ""
  63     print "    (a ... is not needed in the path p4 specification, it's added implicitly)"
  64     print ""
  65     sys.exit(1)
  66 else:
  67     if len(globalPrefix) != 0 and globalPrefix != args[0]:
  68         print "previous import used depot path %s and now %s was specified. this doesn't work!" % (globalPrefix, args[0])
  69         sys.exit(1)
  70     globalPrefix = args[0]
  71
  72 changeRange = ""
  73 revision = ""
  74 users = {}
  75 initialParent = ""
  76 lastChange = 0
  77 initialTag = ""
  78
  79 if globalPrefix.find("@") != -1:
  80     atIdx = globalPrefix.index("@")
  81     changeRange = globalPrefix[atIdx:]
  82     if changeRange == "@all":
  83         changeRange = ""
  84     elif changeRange.find(",") == -1:
  85         revision = changeRange
  86         changeRange = ""
  87     globalPrefix = globalPrefix[0:atIdx]
  88 elif globalPrefix.find("#") != -1:
  89     hashIdx = globalPrefix.index("#")
  90     revision = globalPrefix[hashIdx:]
  91     globalPrefix = globalPrefix[0:hashIdx]
  92 elif len(previousDepotPath) == 0:
  93     revision = "#head"
  94
  95 if globalPrefix.endswith("..."):
  96     globalPrefix = globalPrefix[:-3]
  97
  98 if not globalPrefix.endswith("/"):
  99     globalPrefix += "/"
 100
 101 def p4File(depotPath):
 102     cacheKey = "/tmp/p4cache/data-" + sha.new(depotPath).hexdigest()
 103
 104     data = 0
 105     try:
 106         if not cacheDebug:
 107             raise
 108         data = open(cacheKey, "rb").read()
 109     except:
 110         data = os.popen("p4 print -q \"%s\"" % depotPath, "rb").read()
 111         if cacheDebug:
 112             open(cacheKey, "wb").write(data)
 113
 114     return data
 115
 116 def p4CmdList(cmd):
 117     fullCmd = "p4 -G %s" % cmd;
 118
 119     cacheKey = sha.new(fullCmd).hexdigest()
 120     cacheKey = "/tmp/p4cache/cmd-" + cacheKey
 121
 122     cached = True
 123     pipe = 0
 124     try:
 125         if not cacheDebug:
 126             raise
 127         pipe = open(cacheKey, "rb")
 128     except:
 129         cached = False
 130         pipe = os.popen(fullCmd, "rb")
 131
 132     result = []
 133     try:
 134         while True:
 135             entry = marshal.load(pipe)
 136             result.append(entry)
 137     except EOFError:
 138         pass
 139     pipe.close()
 140
 141     if not cached and cacheDebug:
 142         pipe = open(cacheKey, "wb")
 143         for r in result:
 144             marshal.dump(r, pipe)
 145         pipe.close()
 146
 147     return result
 148
 149 def p4Cmd(cmd):
 150     list = p4CmdList(cmd)
 151     result = {}
 152     for entry in list:
 153         result.update(entry)
 154     return result;
 155
 156 def extractFilesFromCommit(commit):
 157     files = []
 158     fnum = 0
 159     while commit.has_key("depotFile%s" % fnum):
 160         path =  commit["depotFile%s" % fnum]
 161         if not path.startswith(globalPrefix):
 162 #            if not silent:
 163 #                print "\nchanged files: ignoring path %s outside of %s in change %s" % (path, globalPrefix, change)
 164             fnum = fnum + 1
 165             continue
 166
 167         file = {}
 168         file["path"] = path
 169         file["rev"] = commit["rev%s" % fnum]
 170         file["action"] = commit["action%s" % fnum]
 171         file["type"] = commit["type%s" % fnum]
 172         files.append(file)
 173         fnum = fnum + 1
 174     return files
 175
 176 def isSubPathOf(first, second):
 177     if not first.startswith(second):
 178         return False
 179     if first == second:
 180         return True
 181     return first[len(second)] == "/"
 182
 183 def branchesForCommit(files):
 184     global knownBranches
 185     branches = Set()
 186
 187     for file in files:
 188         relativePath = file["path"][len(globalPrefix):]
 189         # strip off the filename
 190         relativePath = relativePath[0:relativePath.rfind("/")]
 191
 192 #        if len(branches) == 0:
 193 #            branches.add(relativePath)
 194 #            knownBranches.add(relativePath)
 195 #            continue
 196
 197         ###### this needs more testing :)
 198         knownBranch = False
 199         for branch in branches:
 200             if relativePath == branch:
 201                 knownBranch = True
 202                 break
 203 #            if relativePath.startswith(branch):
 204             if isSubPathOf(relativePath, branch):
 205                 knownBranch = True
 206                 break
 207 #            if branch.startswith(relativePath):
 208             if isSubPathOf(branch, relativePath):
 209                 branches.remove(branch)
 210                 break
 211
 212         if knownBranch:
 213             continue
 214
 215         for branch in knownBranches:
 216             #if relativePath.startswith(branch):
 217             if isSubPathOf(relativePath, branch):
 218                 if len(branches) == 0:
 219                     relativePath = branch
 220                 else:
 221                     knownBranch = True
 222                 break
 223
 224         if knownBranch:
 225             continue
 226
 227         branches.add(relativePath)
 228         knownBranches.add(relativePath)
 229
 230     return branches
 231
 232 def findBranchParent(branchPrefix, files):
 233     for file in files:
 234         path = file["path"]
 235         if not path.startswith(branchPrefix):
 236             continue
 237         action = file["action"]
 238         if action != "integrate" and action != "branch":
 239             continue
 240         rev = file["rev"]
 241         depotPath = path + "#" + rev
 242
 243         log = p4CmdList("filelog \"%s\"" % depotPath)
 244         if len(log) != 1:
 245             print "eek! I got confused by the filelog of %s" % depotPath
 246             sys.exit(1);
 247
 248         log = log[0]
 249         if log["action0"] != action:
 250             print "eek! wrong action in filelog for %s : found %s, expected %s" % (depotPath, log["action0"], action)
 251             sys.exit(1);
 252
 253         branchAction = log["how0,0"]
 254 #        if branchAction == "branch into" or branchAction == "ignored":
 255 #            continue # ignore for branching
 256
 257         if not branchAction.endswith(" from"):
 258             continue # ignore for branching
 259 #            print "eek! file %s was not branched from but instead: %s" % (depotPath, branchAction)
 260 #            sys.exit(1);
 261
 262         source = log["file0,0"]
 263         if source.startswith(branchPrefix):
 264             continue
 265
 266         lastSourceRev = log["erev0,0"]
 267
 268         sourceLog = p4CmdList("filelog -m 1 \"%s%s\"" % (source, lastSourceRev))
 269         if len(sourceLog) != 1:
 270             print "eek! I got confused by the source filelog of %s%s" % (source, lastSourceRev)
 271             sys.exit(1);
 272         sourceLog = sourceLog[0]
 273
 274         relPath = source[len(globalPrefix):]
 275         # strip off the filename
 276         relPath = relPath[0:relPath.rfind("/")]
 277
 278         for branch in knownBranches:
 279             if isSubPathOf(relPath, branch):
 280 #                print "determined parent branch branch %s due to change in file %s" % (branch, source)
 281                 return branch
 282 #            else:
 283 #                print "%s is not a subpath of branch %s" % (relPath, branch)
 284
 285     return ""
 286
 287 def commit(details, files, branch, branchPrefix, parent):
 288     global users
 289     global lastChange
 290     global committedChanges
 291
 292     epoch = details["time"]
 293     author = details["user"]
 294
 295     gitStream.write("commit %s\n" % branch)
 296     gitStream.write("mark :%s\n" % details["change"])
 297     committedChanges.add(int(details["change"]))
 298     committer = ""
 299     if author in users:
 300         committer = "%s %s %s" % (users[author], epoch, tz)
 301     else:
 302         committer = "%s <a@b> %s %s" % (author, epoch, tz)
 303
 304     gitStream.write("committer %s\n" % committer)
 305
 306     gitStream.write("data <<EOT\n")
 307     gitStream.write(details["desc"])
 308     gitStream.write("\n[ imported from %s; change %s ]\n" % (branchPrefix, details["change"]))
 309     gitStream.write("EOT\n\n")
 310
 311     if len(parent) > 0:
 312         gitStream.write("from %s\n" % parent)
 313
 314     for file in files:
 315         path = file["path"]
 316         if not path.startswith(branchPrefix):
 317 #            if not silent:
 318 #                print "\nchanged files: ignoring path %s outside of branch prefix %s in change %s" % (path, branchPrefix, details["change"])
 319             continue
 320         rev = file["rev"]
 321         depotPath = path + "#" + rev
 322         relPath = path[len(branchPrefix):]
 323         action = file["action"]
 324
 325         if action == "delete":
 326             gitStream.write("D %s\n" % relPath)
 327         else:
 328             mode = 644
 329             if file["type"].startswith("x"):
 330                 mode = 755
 331
 332             data = p4File(depotPath)
 333
 334             gitStream.write("M %s inline %s\n" % (mode, relPath))
 335             gitStream.write("data %s\n" % len(data))
 336             gitStream.write(data)
 337             gitStream.write("\n")
 338
 339     gitStream.write("\n")
 340
 341     lastChange = int(details["change"])
 342
 343 def extractFilesInCommitToBranch(files, branchPrefix):
 344     newFiles = []
 345
 346     for file in files:
 347         path = file["path"]
 348         if path.startswith(branchPrefix):
 349             newFiles.append(file)
 350
 351     return newFiles
 352
 353 def findBranchSourceHeuristic(files, branch, branchPrefix):
 354     for file in files:
 355         action = file["action"]
 356         if action != "integrate" and action != "branch":
 357             continue
 358         path = file["path"]
 359         rev = file["rev"]
 360         depotPath = path + "#" + rev
 361
 362         log = p4CmdList("filelog \"%s\"" % depotPath)
 363         if len(log) != 1:
 364             print "eek! I got confused by the filelog of %s" % depotPath
 365             sys.exit(1);
 366
 367         log = log[0]
 368         if log["action0"] != action:
 369             print "eek! wrong action in filelog for %s : found %s, expected %s" % (depotPath, log["action0"], action)
 370             sys.exit(1);
 371
 372         branchAction = log["how0,0"]
 373
 374         if not branchAction.endswith(" from"):
 375             continue # ignore for branching
 376 #            print "eek! file %s was not branched from but instead: %s" % (depotPath, branchAction)
 377 #            sys.exit(1);
 378
 379         source = log["file0,0"]
 380         if source.startswith(branchPrefix):
 381             continue
 382
 383         lastSourceRev = log["erev0,0"]
 384
 385         sourceLog = p4CmdList("filelog -m 1 \"%s%s\"" % (source, lastSourceRev))
 386         if len(sourceLog) != 1:
 387             print "eek! I got confused by the source filelog of %s%s" % (source, lastSourceRev)
 388             sys.exit(1);
 389         sourceLog = sourceLog[0]
 390
 391         relPath = source[len(globalPrefix):]
 392         # strip off the filename
 393         relPath = relPath[0:relPath.rfind("/")]
 394
 395         for candidate in knownBranches:
 396             if isSubPathOf(relPath, candidate) and candidate != branch:
 397                 return candidate
 398
 399     return ""
 400
 401 def changeIsBranchMerge(sourceBranch, destinationBranch, change):
 402     return False
 403
 404 def getUserMap():
 405     users = {}
 406
 407     for output in p4CmdList("users"):
 408         if not output.has_key("User"):
 409             continue
 410         users[output["User"]] = output["FullName"] + " <" + output["Email"] + ">"
 411     return users
 412
 413 users = getUserMap()
 414
 415 if len(changeRange) == 0:
 416     try:
 417         sout, sin, serr = popen2.popen3("git-name-rev --tags `git-rev-parse %s`" % branch)
 418         output = sout.read()
 419         if output.endswith("\n"):
 420             output = output[:-1]
 421         tagIdx = output.index(" tags/p4/")
 422         caretIdx = output.find("^")
 423         endPos = len(output)
 424         if caretIdx != -1:
 425             endPos = caretIdx
 426         rev = int(output[tagIdx + 9 : endPos]) + 1
 427         changeRange = "@%s,#head" % rev
 428         initialParent = os.popen("git-rev-parse %s" % branch).read()[:-1]
 429         initialTag = "p4/%s" % (int(rev) - 1)
 430     except:
 431         pass
 432
 433 tz = - time.timezone / 36
 434 tzsign = ("%s" % tz)[0]
 435 if tzsign != '+' and tzsign != '-':
 436     tz = "+" + ("%s" % tz)
 437
 438 gitOutput, gitStream, gitError = popen2.popen3("git-fast-import")
 439
 440 if len(revision) > 0:
 441     print "Doing initial import of %s from revision %s" % (globalPrefix, revision)
 442
 443     details = { "user" : "git perforce import user", "time" : int(time.time()) }
 444     details["desc"] = "Initial import of %s from the state at revision %s" % (globalPrefix, revision)
 445     details["change"] = revision
 446     newestRevision = 0
 447
 448     fileCnt = 0
 449     for info in p4CmdList("files %s...%s" % (globalPrefix, revision)):
 450         change = int(info["change"])
 451         if change > newestRevision:
 452             newestRevision = change
 453
 454         if info["action"] == "delete":
 455             continue
 456
 457         for prop in [ "depotFile", "rev", "action", "type" ]:
 458             details["%s%s" % (prop, fileCnt)] = info[prop]
 459
 460         fileCnt = fileCnt + 1
 461
 462     details["change"] = newestRevision
 463
 464     try:
 465         commit(details, extractFilesFromCommit(details), branch, globalPrefix)
 466     except:
 467         print gitError.read()
 468
 469 else:
 470     changes = []
 471
 472     if len(changesFile) > 0:
 473         output = open(changesFile).readlines()
 474         changeSet = Set()
 475         for line in output:
 476             changeSet.add(int(line))
 477
 478         for change in changeSet:
 479             changes.append(change)
 480
 481         changes.sort()
 482     else:
 483         output = os.popen("p4 changes %s...%s" % (globalPrefix, changeRange)).readlines()
 484
 485         for line in output:
 486             changeNum = line.split(" ")[1]
 487             changes.append(changeNum)
 488
 489         changes.reverse()
 490
 491     if len(changes) == 0:
 492         if not silent:
 493             print "no changes to import!"
 494         sys.exit(1)
 495
 496     cnt = 1
 497     for change in changes:
 498         description = p4Cmd("describe %s" % change)
 499
 500         if not silent:
 501             sys.stdout.write("\rimporting revision %s (%s%%)" % (change, cnt * 100 / len(changes)))
 502             sys.stdout.flush()
 503         cnt = cnt + 1
 504
 505 #        try:
 506         files = extractFilesFromCommit(description)
 507         if detectBranches:
 508             for branch in branchesForCommit(files):
 509                 knownBranches.add(branch)
 510                 branchPrefix = globalPrefix + branch + "/"
 511
 512                 filesForCommit = extractFilesInCommitToBranch(files, branchPrefix)
 513
 514                 parent = ""
 515                 ########### remove cnt!!!
 516                 if branch not in createdBranches and cnt > 2:
 517                     createdBranches.add(branch)
 518                     parent = findBranchParent(branchPrefix, files)
 519                     if parent == branch:
 520                         parent = ""
 521 #                    elif len(parent) > 0:
 522 #                        print "%s branched off of %s" % (branch, parent)
 523
 524                 if len(parent) == 0:
 525                     parent = findBranchSourceHeuristic(filesForCommit, branch, branchPrefix)
 526                     if len(parent) > 0:
 527                         print "change %s could be a merge from %s into %s" % (description["change"], parent, branch)
 528                         if not changeIsBranchMerge(parent, branch, description["change"]):
 529                             parent = ""
 530
 531                 branch = "refs/heads/" + branch
 532                 if len(parent) > 0:
 533                     parent = "refs/heads/" + parent
 534                 commit(description, files, branch, branchPrefix, parent)
 535         else:
 536             commit(description, filesForCommit, branch, globalPrefix, initialParent)
 537             initialParent = ""
 538 #        except:
 539 #            print gitError.read()
 540 #            sys.exit(1)
 541
 542 if not silent:
 543     print ""
 544
 545 gitStream.write("reset refs/tags/p4/%s\n" % lastChange)
 546 gitStream.write("from %s\n\n" % branch);
 547
 548
 549 gitStream.close()
 550 gitOutput.close()
 551 gitError.close()
 552
 553 os.popen("git-repo-config p4.depotpath %s" % globalPrefix).read()
 554 if len(initialTag) > 0:
 555     os.popen("git tag -d %s" % initialTag).read()
 556
 557 sys.exit(0)