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
  15 import marshal, popen2, getopt
  16 from sets import Set;
  17
  18 knownBranches = Set()
  19 committedChanges = Set()
  20 branch = "refs/heads/master"
  21 globalPrefix = previousDepotPath = os.popen("git-repo-config --get p4.depotpath").read()
  22 detectBranches = False
  23 changesFile = ""
  24 if len(globalPrefix) != 0:
  25     globalPrefix = globalPrefix[:-1]
  26
  27 try:
  28     opts, args = getopt.getopt(sys.argv[1:], "", [ "branch=", "detect-branches", "changesfile=" ])
  29 except getopt.GetoptError:
  30     print "fixme, syntax error"
  31     sys.exit(1)
  32
  33 for o, a in opts:
  34     if o == "--branch":
  35         branch = "refs/heads/" + a
  36     elif o == "--detect-branches":
  37         detectBranches = True
  38     elif o == "--changesfile":
  39         changesFile = a
  40
  41 if len(args) == 0 and len(globalPrefix) != 0:
  42     print "[using previously specified depot path %s]" % globalPrefix
  43 elif len(args) != 1:
  44     print "usage: %s //depot/path[@revRange]" % sys.argv[0]
  45     print "\n    example:"
  46     print "    %s //depot/my/project/ -- to import the current head"
  47     print "    %s //depot/my/project/@all -- to import everything"
  48     print "    %s //depot/my/project/@1,6 -- to import only from revision 1 to 6"
  49     print ""
  50     print "    (a ... is not needed in the path p4 specification, it's added implicitly)"
  51     print ""
  52     sys.exit(1)
  53 else:
  54     if len(globalPrefix) != 0 and globalPrefix != args[0]:
  55         print "previous import used depot path %s and now %s was specified. this doesn't work!" % (globalPrefix, args[0])
  56         sys.exit(1)
  57     globalPrefix = args[0]
  58
  59 changeRange = ""
  60 revision = ""
  61 users = {}
  62 initialParent = ""
  63 lastChange = 0
  64 initialTag = ""
  65
  66 if globalPrefix.find("@") != -1:
  67     atIdx = globalPrefix.index("@")
  68     changeRange = globalPrefix[atIdx:]
  69     if changeRange == "@all":
  70         changeRange = ""
  71     elif changeRange.find(",") == -1:
  72         revision = changeRange
  73         changeRange = ""
  74     globalPrefix = globalPrefix[0:atIdx]
  75 elif globalPrefix.find("#") != -1:
  76     hashIdx = globalPrefix.index("#")
  77     revision = globalPrefix[hashIdx:]
  78     globalPrefix = globalPrefix[0:hashIdx]
  79 elif len(previousDepotPath) == 0:
  80     revision = "#head"
  81
  82 if globalPrefix.endswith("..."):
  83     globalPrefix = globalPrefix[:-3]
  84
  85 if not globalPrefix.endswith("/"):
  86     globalPrefix += "/"
  87
  88 def p4CmdList(cmd):
  89     pipe = os.popen("p4 -G %s" % cmd, "rb")
  90     result = []
  91     try:
  92         while True:
  93             entry = marshal.load(pipe)
  94             result.append(entry)
  95     except EOFError:
  96         pass
  97     pipe.close()
  98     return result
  99
 100 def p4Cmd(cmd):
 101     list = p4CmdList(cmd)
 102     result = {}
 103     for entry in list:
 104         result.update(entry)
 105     return result;
 106
 107 def extractFilesFromCommit(commit):
 108     files = []
 109     fnum = 0
 110     while commit.has_key("depotFile%s" % fnum):
 111         path =  commit["depotFile%s" % fnum]
 112         if not path.startswith(globalPrefix):
 113             print "\nchanged files: ignoring path %s outside of %s in change %s" % (path, globalPrefix, change)
 114             fnum = fnum + 1
 115             continue
 116
 117         file = {}
 118         file["path"] = path
 119         file["rev"] = commit["rev%s" % fnum]
 120         file["action"] = commit["action%s" % fnum]
 121         file["type"] = commit["type%s" % fnum]
 122         files.append(file)
 123         fnum = fnum + 1
 124     return files
 125
 126 def isSubPathOf(first, second):
 127     if not first.startswith(second):
 128         return False
 129     if first == second:
 130         return True
 131     return first[len(second)] == "/"
 132
 133 def branchesForCommit(files):
 134     global knownBranches
 135     branches = Set()
 136
 137     for file in files:
 138         relativePath = file["path"][len(globalPrefix):]
 139         # strip off the filename
 140         relativePath = relativePath[0:relativePath.rfind("/")]
 141
 142 #        if len(branches) == 0:
 143 #            branches.add(relativePath)
 144 #            knownBranches.add(relativePath)
 145 #            continue
 146
 147         ###### this needs more testing :)
 148         knownBranch = False
 149         for branch in branches:
 150             if relativePath == branch:
 151                 knownBranch = True
 152                 break
 153 #            if relativePath.startswith(branch):
 154             if isSubPathOf(relativePath, branch):
 155                 knownBranch = True
 156                 break
 157 #            if branch.startswith(relativePath):
 158             if isSubPathOf(branch, relativePath):
 159                 branches.remove(branch)
 160                 break
 161
 162         if knownBranch:
 163             continue
 164
 165         for branch in knownBranches:
 166             #if relativePath.startswith(branch):
 167             if isSubPathOf(relativePath, branch):
 168                 if len(branches) == 0:
 169                     relativePath = branch
 170                 else:
 171                     knownBranch = True
 172                 break
 173
 174         if knownBranch:
 175             continue
 176
 177         branches.add(relativePath)
 178         knownBranches.add(relativePath)
 179
 180     return branches
 181
 182 def commit(details, files, branch, branchPrefix):
 183     global initialParent
 184     global users
 185     global lastChange
 186     global committedChanges
 187
 188     epoch = details["time"]
 189     author = details["user"]
 190
 191     gitStream.write("commit %s\n" % branch)
 192     gitStream.write("mark :%s\n" % details["change"])
 193     committedChanges.add(int(details["change"]))
 194     committer = ""
 195     if author in users:
 196         committer = "%s %s %s" % (users[author], epoch, tz)
 197     else:
 198         committer = "%s <a@b> %s %s" % (author, epoch, tz)
 199
 200     gitStream.write("committer %s\n" % committer)
 201
 202     gitStream.write("data <<EOT\n")
 203     gitStream.write(details["desc"])
 204     gitStream.write("\n[ imported from %s; change %s ]\n" % (branchPrefix, details["change"]))
 205     gitStream.write("EOT\n\n")
 206
 207     if len(initialParent) > 0:
 208         gitStream.write("from %s\n" % initialParent)
 209         initialParent = ""
 210
 211     #mergedBranches = Set()
 212     merges = Set()
 213
 214     for file in files:
 215         if lastChange == 0 or not detectBranches:
 216             continue
 217         path = file["path"]
 218         if not path.startswith(branchPrefix):
 219             continue
 220         action = file["action"]
 221         if action != "integrate" and action != "branch":
 222             continue
 223         rev = file["rev"]
 224         depotPath = path + "#" + rev
 225
 226         log = p4CmdList("filelog \"%s\"" % depotPath)
 227         if len(log) != 1:
 228             print "eek! I got confused by the filelog of %s" % depotPath
 229             sys.exit(1);
 230
 231         log = log[0]
 232         if log["action0"] != action:
 233             print "eek! wrong action in filelog for %s : found %s, expected %s" % (depotPath, log["action0"], action)
 234             sys.exit(1);
 235
 236         branchAction = log["how0,0"]
 237 #        if branchAction == "branch into" or branchAction == "ignored":
 238 #            continue # ignore for branching
 239
 240         if not branchAction.endswith(" from"):
 241             continue # ignore for branching
 242 #            print "eek! file %s was not branched from but instead: %s" % (depotPath, branchAction)
 243 #            sys.exit(1);
 244
 245         source = log["file0,0"]
 246         if source.startswith(branchPrefix):
 247             continue
 248
 249         lastSourceRev = log["erev0,0"]
 250
 251         sourceLog = p4CmdList("filelog -m 1 \"%s%s\"" % (source, lastSourceRev))
 252         if len(sourceLog) != 1:
 253             print "eek! I got confused by the source filelog of %s%s" % (source, lastSourceRev)
 254             sys.exit(1);
 255         sourceLog = sourceLog[0]
 256
 257         change = int(sourceLog["change0"])
 258         merges.add(change)
 259
 260 #        relPath = source[len(globalPrefix):]
 261 #
 262 #        for branch in knownBranches:
 263 #            if relPath.startswith(branch) and branch not in mergedBranches:
 264 #                gitStream.write("merge refs/heads/%s\n" % branch)
 265 #                mergedBranches.add(branch)
 266 #                break
 267
 268     for merge in merges:
 269         if merge in committedChanges:
 270             gitStream.write("merge :%s\n" % merge)
 271
 272     for file in files:
 273         path = file["path"]
 274         if not path.startswith(branchPrefix):
 275             print "\nchanged files: ignoring path %s outside of branch prefix %s in change %s" % (path, branchPrefix, details["change"])
 276             continue
 277         rev = file["rev"]
 278         depotPath = path + "#" + rev
 279         relPath = path[len(branchPrefix):]
 280         action = file["action"]
 281
 282         if action == "delete":
 283             gitStream.write("D %s\n" % relPath)
 284         else:
 285             mode = 644
 286             if file["type"].startswith("x"):
 287                 mode = 755
 288
 289             data = os.popen("p4 print -q \"%s\"" % depotPath, "rb").read()
 290
 291             gitStream.write("M %s inline %s\n" % (mode, relPath))
 292             gitStream.write("data %s\n" % len(data))
 293             gitStream.write(data)
 294             gitStream.write("\n")
 295
 296     gitStream.write("\n")
 297
 298     lastChange = int(details["change"])
 299
 300 def getUserMap():
 301     users = {}
 302
 303     for output in p4CmdList("users"):
 304         if not output.has_key("User"):
 305             continue
 306         users[output["User"]] = output["FullName"] + " <" + output["Email"] + ">"
 307     return users
 308
 309 users = getUserMap()
 310
 311 if len(changeRange) == 0:
 312     try:
 313         sout, sin, serr = popen2.popen3("git-name-rev --tags `git-rev-parse %s`" % branch)
 314         output = sout.read()
 315         if output.endswith("\n"):
 316             output = output[:-1]
 317         tagIdx = output.index(" tags/p4/")
 318         caretIdx = output.find("^")
 319         endPos = len(output)
 320         if caretIdx != -1:
 321             endPos = caretIdx
 322         rev = int(output[tagIdx + 9 : endPos]) + 1
 323         changeRange = "@%s,#head" % rev
 324         initialParent = os.popen("git-rev-parse %s" % branch).read()[:-1]
 325         initialTag = "p4/%s" % (int(rev) - 1)
 326     except:
 327         pass
 328
 329 sys.stderr.write("\n")
 330
 331 tz = - time.timezone / 36
 332 tzsign = ("%s" % tz)[0]
 333 if tzsign != '+' and tzsign != '-':
 334     tz = "+" + ("%s" % tz)
 335
 336 gitOutput, gitStream, gitError = popen2.popen3("git-fast-import")
 337
 338 if len(revision) > 0:
 339     print "Doing initial import of %s from revision %s" % (globalPrefix, revision)
 340
 341     details = { "user" : "git perforce import user", "time" : int(time.time()) }
 342     details["desc"] = "Initial import of %s from the state at revision %s" % (globalPrefix, revision)
 343     details["change"] = revision
 344     newestRevision = 0
 345
 346     fileCnt = 0
 347     for info in p4CmdList("files %s...%s" % (globalPrefix, revision)):
 348         change = int(info["change"])
 349         if change > newestRevision:
 350             newestRevision = change
 351
 352         if info["action"] == "delete":
 353             continue
 354
 355         for prop in [ "depotFile", "rev", "action", "type" ]:
 356             details["%s%s" % (prop, fileCnt)] = info[prop]
 357
 358         fileCnt = fileCnt + 1
 359
 360     details["change"] = newestRevision
 361
 362     try:
 363         commit(details, extractFilesFromCommit(details), branch, globalPrefix)
 364     except:
 365         print gitError.read()
 366
 367 else:
 368     changes = []
 369
 370     if len(changesFile) > 0:
 371         output = open(changesFile).readlines()
 372         changeSet = Set()
 373         for line in output:
 374             changeSet.add(int(line))
 375
 376         for change in changeSet:
 377             changes.append(change)
 378
 379         changes.sort()
 380     else:
 381         output = os.popen("p4 changes %s...%s" % (globalPrefix, changeRange)).readlines()
 382
 383         for line in output:
 384             changeNum = line.split(" ")[1]
 385             changes.append(changeNum)
 386
 387         changes.reverse()
 388
 389     if len(changes) == 0:
 390         print "no changes to import!"
 391         sys.exit(1)
 392
 393     cnt = 1
 394     for change in changes:
 395         description = p4Cmd("describe %s" % change)
 396
 397         sys.stdout.write("\rimporting revision %s (%s%%)" % (change, cnt * 100 / len(changes)))
 398         sys.stdout.flush()
 399         cnt = cnt + 1
 400
 401 #        try:
 402         files = extractFilesFromCommit(description)
 403         if detectBranches:
 404             for branch in branchesForCommit(files):
 405                 knownBranches.add(branch)
 406                 branchPrefix = globalPrefix + branch + "/"
 407                 branch = "refs/heads/" + branch
 408                 commit(description, files, branch, branchPrefix)
 409         else:
 410             commit(description, files, branch, globalPrefix)
 411 #        except:
 412 #            print gitError.read()
 413 #            sys.exit(1)
 414
 415 print ""
 416
 417 gitStream.write("reset refs/tags/p4/%s\n" % lastChange)
 418 gitStream.write("from %s\n\n" % branch);
 419
 420
 421 gitStream.close()
 422 gitOutput.close()
 423 gitError.close()
 424
 425 os.popen("git-repo-config p4.depotpath %s" % globalPrefix).read()
 426 if len(initialTag) > 0:
 427     os.popen("git tag -d %s" % initialTag).read()
 428
 429 sys.exit(0)