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 #
  12 import os, string, sys, time
  13 import marshal, popen2, getopt
  14
  15 knownBranches = set()
  16 committedChanges = set()
  17 branch = "refs/heads/master"
  18 globalPrefix = previousDepotPath = os.popen("git-repo-config --get p4.depotpath").read()
  19 detectBranches = False
  20 changesFile = ""
  21 if len(globalPrefix) != 0:
  22     globalPrefix = globalPrefix[:-1]
  23
  24 try:
  25     opts, args = getopt.getopt(sys.argv[1:], "", [ "branch=", "detect-branches", "changesfile=" ])
  26 except getopt.GetoptError:
  27     print "fixme, syntax error"
  28     sys.exit(1)
  29
  30 for o, a in opts:
  31     if o == "--branch":
  32         branch = "refs/heads/" + a
  33     elif o == "--detect-branches":
  34         detectBranches = True
  35     elif o == "--changesfile":
  36         changesFile = a
  37
  38 if len(args) == 0 and len(globalPrefix) != 0:
  39     print "[using previously specified depot path %s]" % globalPrefix
  40 elif len(args) != 1:
  41     print "usage: %s //depot/path[@revRange]" % sys.argv[0]
  42     print "\n    example:"
  43     print "    %s //depot/my/project/ -- to import the current head"
  44     print "    %s //depot/my/project/@all -- to import everything"
  45     print "    %s //depot/my/project/@1,6 -- to import only from revision 1 to 6"
  46     print ""
  47     print "    (a ... is not needed in the path p4 specification, it's added implicitly)"
  48     print ""
  49     sys.exit(1)
  50 else:
  51     if len(globalPrefix) != 0 and globalPrefix != args[0]:
  52         print "previous import used depot path %s and now %s was specified. this doesn't work!" % (globalPrefix, args[0])
  53         sys.exit(1)
  54     globalPrefix = args[0]
  55
  56 changeRange = ""
  57 revision = ""
  58 users = {}
  59 initialParent = ""
  60 lastChange = 0
  61 initialTag = ""
  62
  63 if globalPrefix.find("@") != -1:
  64     atIdx = globalPrefix.index("@")
  65     changeRange = globalPrefix[atIdx:]
  66     if changeRange == "@all":
  67         changeRange = ""
  68     elif changeRange.find(",") == -1:
  69         revision = changeRange
  70         changeRange = ""
  71     globalPrefix = globalPrefix[0:atIdx]
  72 elif globalPrefix.find("#") != -1:
  73     hashIdx = globalPrefix.index("#")
  74     revision = globalPrefix[hashIdx:]
  75     globalPrefix = globalPrefix[0:hashIdx]
  76 elif len(previousDepotPath) == 0:
  77     revision = "#head"
  78
  79 if globalPrefix.endswith("..."):
  80     globalPrefix = globalPrefix[:-3]
  81
  82 if not globalPrefix.endswith("/"):
  83     globalPrefix += "/"
  84
  85 def p4CmdList(cmd):
  86     pipe = os.popen("p4 -G %s" % cmd, "rb")
  87     result = []
  88     try:
  89         while True:
  90             entry = marshal.load(pipe)
  91             result.append(entry)
  92     except EOFError:
  93         pass
  94     pipe.close()
  95     return result
  96
  97 def p4Cmd(cmd):
  98     list = p4CmdList(cmd)
  99     result = {}
 100     for entry in list:
 101         result.update(entry)
 102     return result;
 103
 104 def extractFilesFromCommit(commit):
 105     files = []
 106     fnum = 0
 107     while commit.has_key("depotFile%s" % fnum):
 108         path =  commit["depotFile%s" % fnum]
 109         if not path.startswith(globalPrefix):
 110             print "\nchanged files: ignoring path %s outside of %s in change %s" % (path, globalPrefix, change)
 111             fnum = fnum + 1
 112             continue
 113
 114         file = {}
 115         file["path"] = path
 116         file["rev"] = commit["rev%s" % fnum]
 117         file["action"] = commit["action%s" % fnum]
 118         file["type"] = commit["type%s" % fnum]
 119         files.append(file)
 120         fnum = fnum + 1
 121     return files
 122
 123 def isSubPathOf(first, second):
 124     if not first.startswith(second):
 125         return False
 126     if first == second:
 127         return True
 128     return first[len(second)] == "/"
 129
 130 def branchesForCommit(files):
 131     global knownBranches
 132     branches = set()
 133
 134     for file in files:
 135         relativePath = file["path"][len(globalPrefix):]
 136         # strip off the filename
 137         relativePath = relativePath[0:relativePath.rfind("/")]
 138
 139 #        if len(branches) == 0:
 140 #            branches.add(relativePath)
 141 #            knownBranches.add(relativePath)
 142 #            continue
 143
 144         ###### this needs more testing :)
 145         knownBranch = False
 146         for branch in branches:
 147             if relativePath == branch:
 148                 knownBranch = True
 149                 break
 150 #            if relativePath.startswith(branch):
 151             if isSubPathOf(relativePath, branch):
 152                 knownBranch = True
 153                 break
 154 #            if branch.startswith(relativePath):
 155             if isSubPathOf(branch, relativePath):
 156                 branches.remove(branch)
 157                 break
 158
 159         if knownBranch:
 160             continue
 161
 162         for branch in knownBranches:
 163             #if relativePath.startswith(branch):
 164             if isSubPathOf(relativePath, branch):
 165                 if len(branches) == 0:
 166                     relativePath = branch
 167                 else:
 168                     knownBranch = True
 169                 break
 170
 171         if knownBranch:
 172             continue
 173
 174         branches.add(relativePath)
 175         knownBranches.add(relativePath)
 176
 177     return branches
 178
 179 def commit(details, files, branch, branchPrefix):
 180     global initialParent
 181     global users
 182     global lastChange
 183     global committedChanges
 184
 185     epoch = details["time"]
 186     author = details["user"]
 187
 188     gitStream.write("commit %s\n" % branch)
 189     gitStream.write("mark :%s\n" % details["change"])
 190     committedChanges.add(int(details["change"]))
 191     committer = ""
 192     if author in users:
 193         committer = "%s %s %s" % (users[author], epoch, tz)
 194     else:
 195         committer = "%s <a@b> %s %s" % (author, epoch, tz)
 196
 197     gitStream.write("committer %s\n" % committer)
 198
 199     gitStream.write("data <<EOT\n")
 200     gitStream.write(details["desc"])
 201     gitStream.write("\n[ imported from %s; change %s ]\n" % (branchPrefix, details["change"]))
 202     gitStream.write("EOT\n\n")
 203
 204     if len(initialParent) > 0:
 205         gitStream.write("from %s\n" % initialParent)
 206         initialParent = ""
 207
 208     #mergedBranches = set()
 209     merges = set()
 210
 211     for file in files:
 212         if lastChange == 0:
 213             continue
 214         path = file["path"]
 215         if not path.startswith(branchPrefix):
 216             continue
 217         action = file["action"]
 218         if action != "integrate" and action != "branch":
 219             continue
 220         rev = file["rev"]
 221         depotPath = path + "#" + rev
 222
 223         log = p4CmdList("filelog \"%s\"" % depotPath)
 224         if len(log) != 1:
 225             print "eek! I got confused by the filelog of %s" % depotPath
 226             sys.exit(1);
 227
 228         log = log[0]
 229         if log["action0"] != action:
 230             print "eek! wrong action in filelog for %s : found %s, expected %s" % (depotPath, log["action0"], action)
 231             sys.exit(1);
 232
 233         branchAction = log["how0,0"]
 234 #        if branchAction == "branch into" or branchAction == "ignored":
 235 #            continue # ignore for branching
 236
 237         if not branchAction.endswith(" from"):
 238             continue # ignore for branching
 239 #            print "eek! file %s was not branched from but instead: %s" % (depotPath, branchAction)
 240 #            sys.exit(1);
 241
 242         source = log["file0,0"]
 243         if source.startswith(branchPrefix):
 244             continue
 245
 246         lastSourceRev = log["erev0,0"]
 247
 248         sourceLog = p4CmdList("filelog -m 1 \"%s%s\"" % (source, lastSourceRev))
 249         if len(sourceLog) != 1:
 250             print "eek! I got confused by the source filelog of %s%s" % (source, lastSourceRev)
 251             sys.exit(1);
 252         sourceLog = sourceLog[0]
 253
 254         change = int(sourceLog["change0"])
 255         merges.add(change)
 256
 257 #        relPath = source[len(globalPrefix):]
 258 #
 259 #        for branch in knownBranches:
 260 #            if relPath.startswith(branch) and branch not in mergedBranches:
 261 #                gitStream.write("merge refs/heads/%s\n" % branch)
 262 #                mergedBranches.add(branch)
 263 #                break
 264
 265     for merge in merges:
 266         if merge in committedChanges:
 267             gitStream.write("merge :%s\n" % merge)
 268
 269     for file in files:
 270         path = file["path"]
 271         if not path.startswith(branchPrefix):
 272             print "\nchanged files: ignoring path %s outside of branch prefix %s in change %s" % (path, branchPrefix, details["change"])
 273             continue
 274         rev = file["rev"]
 275         depotPath = path + "#" + rev
 276         relPath = path[len(branchPrefix):]
 277         action = file["action"]
 278
 279         if action == "delete":
 280             gitStream.write("D %s\n" % relPath)
 281         else:
 282             mode = 644
 283             if file["type"].startswith("x"):
 284                 mode = 755
 285
 286             data = os.popen("p4 print -q \"%s\"" % depotPath, "rb").read()
 287
 288             gitStream.write("M %s inline %s\n" % (mode, relPath))
 289             gitStream.write("data %s\n" % len(data))
 290             gitStream.write(data)
 291             gitStream.write("\n")
 292
 293     gitStream.write("\n")
 294
 295     lastChange = int(details["change"])
 296
 297 def getUserMap():
 298     users = {}
 299
 300     for output in p4CmdList("users"):
 301         if not output.has_key("User"):
 302             continue
 303         users[output["User"]] = output["FullName"] + " <" + output["Email"] + ">"
 304     return users
 305
 306 users = getUserMap()
 307
 308 if len(changeRange) == 0:
 309     try:
 310         sout, sin, serr = popen2.popen3("git-name-rev --tags `git-rev-parse %s`" % branch)
 311         output = sout.read()
 312         if output.endswith("\n"):
 313             output = output[:-1]
 314         tagIdx = output.index(" tags/p4/")
 315         caretIdx = output.find("^")
 316         endPos = len(output)
 317         if caretIdx != -1:
 318             endPos = caretIdx
 319         rev = int(output[tagIdx + 9 : endPos]) + 1
 320         changeRange = "@%s,#head" % rev
 321         initialParent = os.popen("git-rev-parse %s" % branch).read()[:-1]
 322         initialTag = "p4/%s" % (int(rev) - 1)
 323     except:
 324         pass
 325
 326 sys.stderr.write("\n")
 327
 328 tz = - time.timezone / 36
 329 tzsign = ("%s" % tz)[0]
 330 if tzsign != '+' and tzsign != '-':
 331     tz = "+" + ("%s" % tz)
 332
 333 gitOutput, gitStream, gitError = popen2.popen3("git-fast-import")
 334
 335 if len(revision) > 0:
 336     print "Doing initial import of %s from revision %s" % (globalPrefix, revision)
 337
 338     details = { "user" : "git perforce import user", "time" : int(time.time()) }
 339     details["desc"] = "Initial import of %s from the state at revision %s" % (globalPrefix, revision)
 340     details["change"] = revision
 341     newestRevision = 0
 342
 343     fileCnt = 0
 344     for info in p4CmdList("files %s...%s" % (globalPrefix, revision)):
 345         change = int(info["change"])
 346         if change > newestRevision:
 347             newestRevision = change
 348
 349         if info["action"] == "delete":
 350             continue
 351
 352         for prop in [ "depotFile", "rev", "action", "type" ]:
 353             details["%s%s" % (prop, fileCnt)] = info[prop]
 354
 355         fileCnt = fileCnt + 1
 356
 357     details["change"] = newestRevision
 358
 359     try:
 360         commit(details, extractFilesFromCommit(details), branch, globalPrefix)
 361     except:
 362         print gitError.read()
 363
 364 else:
 365     changes = []
 366
 367     if len(changesFile) > 0:
 368         output = open(changesFile).readlines()
 369         changeSet = set()
 370         for line in output:
 371             changeSet.add(int(line))
 372
 373         for change in changeSet:
 374             changes.append(change)
 375
 376         changes.sort()
 377     else:
 378         output = os.popen("p4 changes %s...%s" % (globalPrefix, changeRange)).readlines()
 379
 380         for line in output:
 381             changeNum = line.split(" ")[1]
 382             changes.append(changeNum)
 383
 384         changes.reverse()
 385
 386     if len(changes) == 0:
 387         print "no changes to import!"
 388         sys.exit(1)
 389
 390     cnt = 1
 391     for change in changes:
 392         description = p4Cmd("describe %s" % change)
 393
 394         sys.stdout.write("\rimporting revision %s (%s%%)" % (change, cnt * 100 / len(changes)))
 395         sys.stdout.flush()
 396         cnt = cnt + 1
 397
 398 #        try:
 399         files = extractFilesFromCommit(description)
 400         if detectBranches:
 401             for branch in branchesForCommit(files):
 402                 knownBranches.add(branch)
 403                 branchPrefix = globalPrefix + branch + "/"
 404                 branch = "refs/heads/" + branch
 405                 commit(description, files, branch, branchPrefix)
 406         else:
 407             commit(description, files, branch, globalPrefix)
 408 #        except:
 409 #            print gitError.read()
 410 #            sys.exit(1)
 411
 412 print ""
 413
 414 gitStream.write("reset refs/tags/p4/%s\n" % lastChange)
 415 gitStream.write("from %s\n\n" % branch);
 416
 417
 418 gitStream.close()
 419 gitOutput.close()
 420 gitError.close()
 421
 422 os.popen("git-repo-config p4.depotpath %s" % globalPrefix).read()
 423 if len(initialTag) > 0:
 424     os.popen("git tag -d %s" % initialTag).read()
 425
 426 sys.exit(0)