p4-fast-export.py

   1 #!/usr/bin/env 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 dataCache = False
  19 commandCache = False
  20
  21 silent = False
  22 knownBranches = Set()
  23 createdBranches = Set()
  24 committedChanges = Set()
  25 branch = "refs/heads/master"
  26 globalPrefix = previousDepotPath = os.popen("git-repo-config --get p4.depotpath").read()
  27 detectBranches = False
  28 changesFile = ""
  29 if len(globalPrefix) != 0:
  30     globalPrefix = globalPrefix[:-1]
  31
  32 try:
  33     opts, args = getopt.getopt(sys.argv[1:], "", [ "branch=", "detect-branches", "changesfile=", "silent", "known-branches=",
  34                                                    "cache", "command-cache" ])
  35 except getopt.GetoptError:
  36     print "fixme, syntax error"
  37     sys.exit(1)
  38
  39 for o, a in opts:
  40     if o == "--branch":
  41         branch = "refs/heads/" + a
  42     elif o == "--detect-branches":
  43         detectBranches = True
  44     elif o == "--changesfile":
  45         changesFile = a
  46     elif o == "--silent":
  47         silent= True
  48     elif o == "--known-branches":
  49         for branch in open(a).readlines():
  50             knownBranches.add(branch[:-1])
  51     elif o == "--cache":
  52         dataCache = True
  53         commandCache = True
  54     elif o == "--command-cache":
  55         commandCache = True
  56
  57 if len(args) == 0 and len(globalPrefix) != 0:
  58     if not silent:
  59         print "[using previously specified depot path %s]" % globalPrefix
  60 elif len(args) != 1:
  61     print "usage: %s //depot/path[@revRange]" % sys.argv[0]
  62     print "\n    example:"
  63     print "    %s //depot/my/project/ -- to import the current head"
  64     print "    %s //depot/my/project/@all -- to import everything"
  65     print "    %s //depot/my/project/@1,6 -- to import only from revision 1 to 6"
  66     print ""
  67     print "    (a ... is not needed in the path p4 specification, it's added implicitly)"
  68     print ""
  69     sys.exit(1)
  70 else:
  71     if len(globalPrefix) != 0 and globalPrefix != args[0]:
  72         print "previous import used depot path %s and now %s was specified. this doesn't work!" % (globalPrefix, args[0])
  73         sys.exit(1)
  74     globalPrefix = args[0]
  75
  76 changeRange = ""
  77 revision = ""
  78 users = {}
  79 initialParent = ""
  80 lastChange = 0
  81 initialTag = ""
  82
  83 if globalPrefix.find("@") != -1:
  84     atIdx = globalPrefix.index("@")
  85     changeRange = globalPrefix[atIdx:]
  86     if changeRange == "@all":
  87         changeRange = ""
  88     elif changeRange.find(",") == -1:
  89         revision = changeRange
  90         changeRange = ""
  91     globalPrefix = globalPrefix[0:atIdx]
  92 elif globalPrefix.find("#") != -1:
  93     hashIdx = globalPrefix.index("#")
  94     revision = globalPrefix[hashIdx:]
  95     globalPrefix = globalPrefix[0:hashIdx]
  96 elif len(previousDepotPath) == 0:
  97     revision = "#head"
  98
  99 if globalPrefix.endswith("..."):
 100     globalPrefix = globalPrefix[:-3]
 101
 102 if not globalPrefix.endswith("/"):
 103     globalPrefix += "/"
 104
 105 def p4File(depotPath):
 106     cacheKey = "/tmp/p4cache/data-" + sha.new(depotPath).hexdigest()
 107
 108     data = 0
 109     try:
 110         if not dataCache:
 111             raise
 112         data = open(cacheKey, "rb").read()
 113     except:
 114         data = os.popen("p4 print -q \"%s\"" % depotPath, "rb").read()
 115         if dataCache:
 116             open(cacheKey, "wb").write(data)
 117
 118     return data
 119
 120 def p4CmdList(cmd):
 121     fullCmd = "p4 -G %s" % cmd;
 122
 123     cacheKey = sha.new(fullCmd).hexdigest()
 124     cacheKey = "/tmp/p4cache/cmd-" + cacheKey
 125
 126     cached = True
 127     pipe = 0
 128     try:
 129         if not commandCache:
 130             raise
 131         pipe = open(cacheKey, "rb")
 132     except:
 133         cached = False
 134         pipe = os.popen(fullCmd, "rb")
 135
 136     result = []
 137     try:
 138         while True:
 139             entry = marshal.load(pipe)
 140             result.append(entry)
 141     except EOFError:
 142         pass
 143     pipe.close()
 144
 145     if not cached and commandCache:
 146         pipe = open(cacheKey, "wb")
 147         for r in result:
 148             marshal.dump(r, pipe)
 149         pipe.close()
 150
 151     return result
 152
 153 def p4Cmd(cmd):
 154     list = p4CmdList(cmd)
 155     result = {}
 156     for entry in list:
 157         result.update(entry)
 158     return result;
 159
 160 def extractFilesFromCommit(commit):
 161     files = []
 162     fnum = 0
 163     while commit.has_key("depotFile%s" % fnum):
 164         path =  commit["depotFile%s" % fnum]
 165         if not path.startswith(globalPrefix):
 166 #            if not silent:
 167 #                print "\nchanged files: ignoring path %s outside of %s in change %s" % (path, globalPrefix, change)
 168             fnum = fnum + 1
 169             continue
 170
 171         file = {}
 172         file["path"] = path
 173         file["rev"] = commit["rev%s" % fnum]
 174         file["action"] = commit["action%s" % fnum]
 175         file["type"] = commit["type%s" % fnum]
 176         files.append(file)
 177         fnum = fnum + 1
 178     return files
 179
 180 def isSubPathOf(first, second):
 181     if not first.startswith(second):
 182         return False
 183     if first == second:
 184         return True
 185     return first[len(second)] == "/"
 186
 187 def branchesForCommit(files):
 188     global knownBranches
 189     branches = Set()
 190
 191     for file in files:
 192         relativePath = file["path"][len(globalPrefix):]
 193         # strip off the filename
 194         relativePath = relativePath[0:relativePath.rfind("/")]
 195
 196 #        if len(branches) == 0:
 197 #            branches.add(relativePath)
 198 #            knownBranches.add(relativePath)
 199 #            continue
 200
 201         ###### this needs more testing :)
 202         knownBranch = False
 203         for branch in branches:
 204             if relativePath == branch:
 205                 knownBranch = True
 206                 break
 207 #            if relativePath.startswith(branch):
 208             if isSubPathOf(relativePath, branch):
 209                 knownBranch = True
 210                 break
 211 #            if branch.startswith(relativePath):
 212             if isSubPathOf(branch, relativePath):
 213                 branches.remove(branch)
 214                 break
 215
 216         if knownBranch:
 217             continue
 218
 219         for branch in knownBranches:
 220             #if relativePath.startswith(branch):
 221             if isSubPathOf(relativePath, branch):
 222                 if len(branches) == 0:
 223                     relativePath = branch
 224                 else:
 225                     knownBranch = True
 226                 break
 227
 228         if knownBranch:
 229             continue
 230
 231         branches.add(relativePath)
 232         knownBranches.add(relativePath)
 233
 234     return branches
 235
 236 def findBranchParent(branchPrefix, files):
 237     for file in files:
 238         path = file["path"]
 239         if not path.startswith(branchPrefix):
 240             continue
 241         action = file["action"]
 242         if action != "integrate" and action != "branch":
 243             continue
 244         rev = file["rev"]
 245         depotPath = path + "#" + rev
 246
 247         log = p4CmdList("filelog \"%s\"" % depotPath)
 248         if len(log) != 1:
 249             print "eek! I got confused by the filelog of %s" % depotPath
 250             sys.exit(1);
 251
 252         log = log[0]
 253         if log["action0"] != action:
 254             print "eek! wrong action in filelog for %s : found %s, expected %s" % (depotPath, log["action0"], action)
 255             sys.exit(1);
 256
 257         branchAction = log["how0,0"]
 258 #        if branchAction == "branch into" or branchAction == "ignored":
 259 #            continue # ignore for branching
 260
 261         if not branchAction.endswith(" from"):
 262             continue # ignore for branching
 263 #            print "eek! file %s was not branched from but instead: %s" % (depotPath, branchAction)
 264 #            sys.exit(1);
 265
 266         source = log["file0,0"]
 267         if source.startswith(branchPrefix):
 268             continue
 269
 270         lastSourceRev = log["erev0,0"]
 271
 272         sourceLog = p4CmdList("filelog -m 1 \"%s%s\"" % (source, lastSourceRev))
 273         if len(sourceLog) != 1:
 274             print "eek! I got confused by the source filelog of %s%s" % (source, lastSourceRev)
 275             sys.exit(1);
 276         sourceLog = sourceLog[0]
 277
 278         relPath = source[len(globalPrefix):]
 279         # strip off the filename
 280         relPath = relPath[0:relPath.rfind("/")]
 281
 282         for branch in knownBranches:
 283             if isSubPathOf(relPath, branch):
 284 #                print "determined parent branch branch %s due to change in file %s" % (branch, source)
 285                 return branch
 286 #            else:
 287 #                print "%s is not a subpath of branch %s" % (relPath, branch)
 288
 289     return ""
 290
 291 def commit(details, files, branch, branchPrefix, parent, merged = ""):
 292     global users
 293     global lastChange
 294     global committedChanges
 295
 296     epoch = details["time"]
 297     author = details["user"]
 298
 299     gitStream.write("commit %s\n" % branch)
 300 #    gitStream.write("mark :%s\n" % details["change"])
 301     committedChanges.add(int(details["change"]))
 302     committer = ""
 303     if author in users:
 304         committer = "%s %s %s" % (users[author], epoch, tz)
 305     else:
 306         committer = "%s <a@b> %s %s" % (author, epoch, tz)
 307
 308     gitStream.write("committer %s\n" % committer)
 309
 310     gitStream.write("data <<EOT\n")
 311     gitStream.write(details["desc"])
 312     gitStream.write("\n[ imported from %s; change %s ]\n" % (branchPrefix, details["change"]))
 313     gitStream.write("EOT\n\n")
 314
 315     if len(parent) > 0:
 316         gitStream.write("from %s\n" % parent)
 317
 318     if len(merged) > 0:
 319         gitStream.write("merge %s\n" % merged)
 320
 321     for file in files:
 322         path = file["path"]
 323         if not path.startswith(branchPrefix):
 324 #            if not silent:
 325 #                print "\nchanged files: ignoring path %s outside of branch prefix %s in change %s" % (path, branchPrefix, details["change"])
 326             continue
 327         rev = file["rev"]
 328         depotPath = path + "#" + rev
 329         relPath = path[len(branchPrefix):]
 330         action = file["action"]
 331
 332         if file["type"] == "apple":
 333             print "\nfile %s is a strange apple file that forks. Ignoring!" % path
 334             continue
 335
 336         if action == "delete":
 337             gitStream.write("D %s\n" % relPath)
 338         else:
 339             mode = 644
 340             if file["type"].startswith("x"):
 341                 mode = 755
 342
 343             data = p4File(depotPath)
 344
 345             gitStream.write("M %s inline %s\n" % (mode, relPath))
 346             gitStream.write("data %s\n" % len(data))
 347             gitStream.write(data)
 348             gitStream.write("\n")
 349
 350     gitStream.write("\n")
 351
 352     lastChange = int(details["change"])
 353
 354 def extractFilesInCommitToBranch(files, branchPrefix):
 355     newFiles = []
 356
 357     for file in files:
 358         path = file["path"]
 359         if path.startswith(branchPrefix):
 360             newFiles.append(file)
 361
 362     return newFiles
 363
 364 def findBranchSourceHeuristic(files, branch, branchPrefix):
 365     for file in files:
 366         action = file["action"]
 367         if action != "integrate" and action != "branch":
 368             continue
 369         path = file["path"]
 370         rev = file["rev"]
 371         depotPath = path + "#" + rev
 372
 373         log = p4CmdList("filelog \"%s\"" % depotPath)
 374         if len(log) != 1:
 375             print "eek! I got confused by the filelog of %s" % depotPath
 376             sys.exit(1);
 377
 378         log = log[0]
 379         if log["action0"] != action:
 380             print "eek! wrong action in filelog for %s : found %s, expected %s" % (depotPath, log["action0"], action)
 381             sys.exit(1);
 382
 383         branchAction = log["how0,0"]
 384
 385         if not branchAction.endswith(" from"):
 386             continue # ignore for branching
 387 #            print "eek! file %s was not branched from but instead: %s" % (depotPath, branchAction)
 388 #            sys.exit(1);
 389
 390         source = log["file0,0"]
 391         if source.startswith(branchPrefix):
 392             continue
 393
 394         lastSourceRev = log["erev0,0"]
 395
 396         sourceLog = p4CmdList("filelog -m 1 \"%s%s\"" % (source, lastSourceRev))
 397         if len(sourceLog) != 1:
 398             print "eek! I got confused by the source filelog of %s%s" % (source, lastSourceRev)
 399             sys.exit(1);
 400         sourceLog = sourceLog[0]
 401
 402         relPath = source[len(globalPrefix):]
 403         # strip off the filename
 404         relPath = relPath[0:relPath.rfind("/")]
 405
 406         for candidate in knownBranches:
 407             if isSubPathOf(relPath, candidate) and candidate != branch:
 408                 return candidate
 409
 410     return ""
 411
 412 def changeIsBranchMerge(sourceBranch, destinationBranch, change):
 413     sourceFiles = {}
 414     for file in p4CmdList("files %s...@%s" % (globalPrefix + sourceBranch + "/", change)):
 415         if file["action"] == "delete":
 416             continue
 417         sourceFiles[file["depotFile"]] = file
 418
 419     destinationFiles = {}
 420     for file in p4CmdList("files %s...@%s" % (globalPrefix + destinationBranch + "/", change)):
 421         destinationFiles[file["depotFile"]] = file
 422
 423     for fileName in sourceFiles.keys():
 424         integrations = []
 425         deleted = False
 426         integrationCount = 0
 427         for integration in p4CmdList("integrated \"%s\"" % fileName):
 428             toFile = integration["fromFile"] # yes, it's true, it's fromFile
 429             if not toFile in destinationFiles:
 430                 continue
 431             destFile = destinationFiles[toFile]
 432             if destFile["action"] == "delete":
 433 #                print "file %s has been deleted in %s" % (fileName, toFile)
 434                 deleted = True
 435                 break
 436             integrationCount += 1
 437             if integration["how"] == "branch from":
 438                 continue
 439
 440             if int(integration["change"]) == change:
 441                 integrations.append(integration)
 442                 continue
 443             if int(integration["change"]) > change:
 444                 continue
 445
 446             destRev = int(destFile["rev"])
 447
 448             startRev = integration["startFromRev"][1:]
 449             if startRev == "none":
 450                 startRev = 0
 451             else:
 452                 startRev = int(startRev)
 453
 454             endRev = integration["endFromRev"][1:]
 455             if endRev == "none":
 456                 endRev = 0
 457             else:
 458                 endRev = int(endRev)
 459
 460             initialBranch = (destRev == 1 and integration["how"] != "branch into")
 461             inRange = (destRev >= startRev and destRev <= endRev)
 462             newer = (destRev > startRev and destRev > endRev)
 463
 464             if initialBranch or inRange or newer:
 465                 integrations.append(integration)
 466
 467         if deleted:
 468             continue
 469
 470         if len(integrations) == 0 and integrationCount > 1:
 471             print "file %s was not integrated from %s into %s" % (fileName, sourceBranch, destinationBranch)
 472             return False
 473
 474     return True
 475
 476 def getUserMap():
 477     users = {}
 478
 479     for output in p4CmdList("users"):
 480         if not output.has_key("User"):
 481             continue
 482         users[output["User"]] = output["FullName"] + " <" + output["Email"] + ">"
 483     return users
 484
 485 users = getUserMap()
 486
 487 if len(changeRange) == 0:
 488     try:
 489         sout, sin, serr = popen2.popen3("git-name-rev --tags `git-rev-parse %s`" % branch)
 490         output = sout.read()
 491         if output.endswith("\n"):
 492             output = output[:-1]
 493         tagIdx = output.index(" tags/p4/")
 494         caretIdx = output.find("^")
 495         endPos = len(output)
 496         if caretIdx != -1:
 497             endPos = caretIdx
 498         rev = int(output[tagIdx + 9 : endPos]) + 1
 499         changeRange = "@%s,#head" % rev
 500         initialParent = os.popen("git-rev-parse %s" % branch).read()[:-1]
 501         initialTag = "p4/%s" % (int(rev) - 1)
 502     except:
 503         pass
 504
 505 tz = - time.timezone / 36
 506 tzsign = ("%s" % tz)[0]
 507 if tzsign != '+' and tzsign != '-':
 508     tz = "+" + ("%s" % tz)
 509
 510 gitOutput, gitStream, gitError = popen2.popen3("git-fast-import")
 511
 512 if len(revision) > 0:
 513     print "Doing initial import of %s from revision %s" % (globalPrefix, revision)
 514
 515     details = { "user" : "git perforce import user", "time" : int(time.time()) }
 516     details["desc"] = "Initial import of %s from the state at revision %s" % (globalPrefix, revision)
 517     details["change"] = revision
 518     newestRevision = 0
 519
 520     fileCnt = 0
 521     for info in p4CmdList("files %s...%s" % (globalPrefix, revision)):
 522         change = int(info["change"])
 523         if change > newestRevision:
 524             newestRevision = change
 525
 526         if info["action"] == "delete":
 527             continue
 528
 529         for prop in [ "depotFile", "rev", "action", "type" ]:
 530             details["%s%s" % (prop, fileCnt)] = info[prop]
 531
 532         fileCnt = fileCnt + 1
 533
 534     details["change"] = newestRevision
 535
 536     try:
 537         commit(details, extractFilesFromCommit(details), branch, globalPrefix)
 538     except:
 539         print gitError.read()
 540
 541 else:
 542     changes = []
 543
 544     if len(changesFile) > 0:
 545         output = open(changesFile).readlines()
 546         changeSet = Set()
 547         for line in output:
 548             changeSet.add(int(line))
 549
 550         for change in changeSet:
 551             changes.append(change)
 552
 553         changes.sort()
 554     else:
 555         output = os.popen("p4 changes %s...%s" % (globalPrefix, changeRange)).readlines()
 556
 557         for line in output:
 558             changeNum = line.split(" ")[1]
 559             changes.append(changeNum)
 560
 561         changes.reverse()
 562
 563     if len(changes) == 0:
 564         if not silent:
 565             print "no changes to import!"
 566         sys.exit(1)
 567
 568     cnt = 1
 569     for change in changes:
 570         description = p4Cmd("describe %s" % change)
 571
 572         if not silent:
 573             sys.stdout.write("\rimporting revision %s (%s%%)" % (change, cnt * 100 / len(changes)))
 574             sys.stdout.flush()
 575         cnt = cnt + 1
 576
 577         try:
 578             files = extractFilesFromCommit(description)
 579             if detectBranches:
 580                 for branch in branchesForCommit(files):
 581                     knownBranches.add(branch)
 582                     branchPrefix = globalPrefix + branch + "/"
 583
 584                     filesForCommit = extractFilesInCommitToBranch(files, branchPrefix)
 585
 586                     merged = ""
 587                     parent = ""
 588                     ########### remove cnt!!!
 589                     if branch not in createdBranches and cnt > 2:
 590                         createdBranches.add(branch)
 591                         parent = findBranchParent(branchPrefix, files)
 592                         if parent == branch:
 593                             parent = ""
 594     #                    elif len(parent) > 0:
 595     #                        print "%s branched off of %s" % (branch, parent)
 596
 597                     if len(parent) == 0:
 598                         merged = findBranchSourceHeuristic(filesForCommit, branch, branchPrefix)
 599                         if len(merged) > 0:
 600                             print "change %s could be a merge from %s into %s" % (description["change"], merged, branch)
 601                             if not changeIsBranchMerge(merged, branch, int(description["change"])):
 602                                 merged = ""
 603
 604                     branch = "refs/heads/" + branch
 605                     if len(parent) > 0:
 606                         parent = "refs/heads/" + parent
 607                     if len(merged) > 0:
 608                         merged = "refs/heads/" + merged
 609                     commit(description, files, branch, branchPrefix, parent, merged)
 610             else:
 611                 commit(description, files, branch, globalPrefix, initialParent)
 612                 initialParent = ""
 613         except IOError:
 614             print gitError.read()
 615             sys.exit(1)
 616
 617 if not silent:
 618     print ""
 619
 620 gitStream.write("reset refs/tags/p4/%s\n" % lastChange)
 621 gitStream.write("from %s\n\n" % branch);
 622
 623
 624 gitStream.close()
 625 gitOutput.close()
 626 gitError.close()
 627
 628 os.popen("git-repo-config p4.depotpath %s" % globalPrefix).read()
 629 if len(initialTag) > 0:
 630     os.popen("git tag -d %s" % initialTag).read()
 631
 632 sys.exit(0)