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 silent = False
  19 knownBranches = Set()
  20 committedChanges = Set()
  21 branch = "refs/heads/master"
  22 globalPrefix = previousDepotPath = os.popen("git-repo-config --get p4.depotpath").read()
  23 detectBranches = False
  24 changesFile = ""
  25 if len(globalPrefix) != 0:
  26     globalPrefix = globalPrefix[:-1]
  27
  28 try:
  29     opts, args = getopt.getopt(sys.argv[1:], "", [ "branch=", "detect-branches", "changesfile=", "silent" ])
  30 except getopt.GetoptError:
  31     print "fixme, syntax error"
  32     sys.exit(1)
  33
  34 for o, a in opts:
  35     if o == "--branch":
  36         branch = "refs/heads/" + a
  37     elif o == "--detect-branches":
  38         detectBranches = True
  39     elif o == "--changesfile":
  40         changesFile = a
  41     elif o == "--silent":
  42         silent= True
  43
  44 if len(args) == 0 and len(globalPrefix) != 0:
  45     print "[using previously specified depot path %s]" % globalPrefix
  46 elif len(args) != 1:
  47     print "usage: %s //depot/path[@revRange]" % sys.argv[0]
  48     print "\n    example:"
  49     print "    %s //depot/my/project/ -- to import the current head"
  50     print "    %s //depot/my/project/@all -- to import everything"
  51     print "    %s //depot/my/project/@1,6 -- to import only from revision 1 to 6"
  52     print ""
  53     print "    (a ... is not needed in the path p4 specification, it's added implicitly)"
  54     print ""
  55     sys.exit(1)
  56 else:
  57     if len(globalPrefix) != 0 and globalPrefix != args[0]:
  58         print "previous import used depot path %s and now %s was specified. this doesn't work!" % (globalPrefix, args[0])
  59         sys.exit(1)
  60     globalPrefix = args[0]
  61
  62 changeRange = ""
  63 revision = ""
  64 users = {}
  65 initialParent = ""
  66 lastChange = 0
  67 initialTag = ""
  68
  69 if globalPrefix.find("@") != -1:
  70     atIdx = globalPrefix.index("@")
  71     changeRange = globalPrefix[atIdx:]
  72     if changeRange == "@all":
  73         changeRange = ""
  74     elif changeRange.find(",") == -1:
  75         revision = changeRange
  76         changeRange = ""
  77     globalPrefix = globalPrefix[0:atIdx]
  78 elif globalPrefix.find("#") != -1:
  79     hashIdx = globalPrefix.index("#")
  80     revision = globalPrefix[hashIdx:]
  81     globalPrefix = globalPrefix[0:hashIdx]
  82 elif len(previousDepotPath) == 0:
  83     revision = "#head"
  84
  85 if globalPrefix.endswith("..."):
  86     globalPrefix = globalPrefix[:-3]
  87
  88 if not globalPrefix.endswith("/"):
  89     globalPrefix += "/"
  90
  91 def p4CmdList(cmd):
  92     pipe = os.popen("p4 -G %s" % cmd, "rb")
  93     result = []
  94     try:
  95         while True:
  96             entry = marshal.load(pipe)
  97             result.append(entry)
  98     except EOFError:
  99         pass
 100     pipe.close()
 101     return result
 102
 103 def p4Cmd(cmd):
 104     list = p4CmdList(cmd)
 105     result = {}
 106     for entry in list:
 107         result.update(entry)
 108     return result;
 109
 110 def extractFilesFromCommit(commit):
 111     files = []
 112     fnum = 0
 113     while commit.has_key("depotFile%s" % fnum):
 114         path =  commit["depotFile%s" % fnum]
 115         if not path.startswith(globalPrefix):
 116             if not silent:
 117                 print "\nchanged files: ignoring path %s outside of %s in change %s" % (path, globalPrefix, change)
 118             fnum = fnum + 1
 119             continue
 120
 121         file = {}
 122         file["path"] = path
 123         file["rev"] = commit["rev%s" % fnum]
 124         file["action"] = commit["action%s" % fnum]
 125         file["type"] = commit["type%s" % fnum]
 126         files.append(file)
 127         fnum = fnum + 1
 128     return files
 129
 130 def isSubPathOf(first, second):
 131     if not first.startswith(second):
 132         return False
 133     if first == second:
 134         return True
 135     return first[len(second)] == "/"
 136
 137 def branchesForCommit(files):
 138     global knownBranches
 139     branches = Set()
 140
 141     for file in files:
 142         relativePath = file["path"][len(globalPrefix):]
 143         # strip off the filename
 144         relativePath = relativePath[0:relativePath.rfind("/")]
 145
 146 #        if len(branches) == 0:
 147 #            branches.add(relativePath)
 148 #            knownBranches.add(relativePath)
 149 #            continue
 150
 151         ###### this needs more testing :)
 152         knownBranch = False
 153         for branch in branches:
 154             if relativePath == branch:
 155                 knownBranch = True
 156                 break
 157 #            if relativePath.startswith(branch):
 158             if isSubPathOf(relativePath, branch):
 159                 knownBranch = True
 160                 break
 161 #            if branch.startswith(relativePath):
 162             if isSubPathOf(branch, relativePath):
 163                 branches.remove(branch)
 164                 break
 165
 166         if knownBranch:
 167             continue
 168
 169         for branch in knownBranches:
 170             #if relativePath.startswith(branch):
 171             if isSubPathOf(relativePath, branch):
 172                 if len(branches) == 0:
 173                     relativePath = branch
 174                 else:
 175                     knownBranch = True
 176                 break
 177
 178         if knownBranch:
 179             continue
 180
 181         branches.add(relativePath)
 182         knownBranches.add(relativePath)
 183
 184     return branches
 185
 186 def commit(details, files, branch, branchPrefix):
 187     global initialParent
 188     global users
 189     global lastChange
 190     global committedChanges
 191
 192     epoch = details["time"]
 193     author = details["user"]
 194
 195     gitStream.write("commit %s\n" % branch)
 196     gitStream.write("mark :%s\n" % details["change"])
 197     committedChanges.add(int(details["change"]))
 198     committer = ""
 199     if author in users:
 200         committer = "%s %s %s" % (users[author], epoch, tz)
 201     else:
 202         committer = "%s <a@b> %s %s" % (author, epoch, tz)
 203
 204     gitStream.write("committer %s\n" % committer)
 205
 206     gitStream.write("data <<EOT\n")
 207     gitStream.write(details["desc"])
 208     gitStream.write("\n[ imported from %s; change %s ]\n" % (branchPrefix, details["change"]))
 209     gitStream.write("EOT\n\n")
 210
 211     if len(initialParent) > 0:
 212         gitStream.write("from %s\n" % initialParent)
 213         initialParent = ""
 214
 215     #mergedBranches = Set()
 216     merges = Set()
 217
 218     for file in files:
 219         if lastChange == 0 or not detectBranches:
 220             continue
 221         path = file["path"]
 222         if not path.startswith(branchPrefix):
 223             continue
 224         action = file["action"]
 225         if action != "integrate" and action != "branch":
 226             continue
 227         rev = file["rev"]
 228         depotPath = path + "#" + rev
 229
 230         log = p4CmdList("filelog \"%s\"" % depotPath)
 231         if len(log) != 1:
 232             print "eek! I got confused by the filelog of %s" % depotPath
 233             sys.exit(1);
 234
 235         log = log[0]
 236         if log["action0"] != action:
 237             print "eek! wrong action in filelog for %s : found %s, expected %s" % (depotPath, log["action0"], action)
 238             sys.exit(1);
 239
 240         branchAction = log["how0,0"]
 241 #        if branchAction == "branch into" or branchAction == "ignored":
 242 #            continue # ignore for branching
 243
 244         if not branchAction.endswith(" from"):
 245             continue # ignore for branching
 246 #            print "eek! file %s was not branched from but instead: %s" % (depotPath, branchAction)
 247 #            sys.exit(1);
 248
 249         source = log["file0,0"]
 250         if source.startswith(branchPrefix):
 251             continue
 252
 253         lastSourceRev = log["erev0,0"]
 254
 255         sourceLog = p4CmdList("filelog -m 1 \"%s%s\"" % (source, lastSourceRev))
 256         if len(sourceLog) != 1:
 257             print "eek! I got confused by the source filelog of %s%s" % (source, lastSourceRev)
 258             sys.exit(1);
 259         sourceLog = sourceLog[0]
 260
 261         change = int(sourceLog["change0"])
 262         merges.add(change)
 263
 264 #        relPath = source[len(globalPrefix):]
 265 #
 266 #        for branch in knownBranches:
 267 #            if relPath.startswith(branch) and branch not in mergedBranches:
 268 #                gitStream.write("merge refs/heads/%s\n" % branch)
 269 #                mergedBranches.add(branch)
 270 #                break
 271
 272     for merge in merges:
 273         if merge in committedChanges:
 274             gitStream.write("merge :%s\n" % merge)
 275
 276     for file in files:
 277         path = file["path"]
 278         if not path.startswith(branchPrefix):
 279             if not silent:
 280                 print "\nchanged files: ignoring path %s outside of branch prefix %s in change %s" % (path, branchPrefix, details["change"])
 281             continue
 282         rev = file["rev"]
 283         depotPath = path + "#" + rev
 284         relPath = path[len(branchPrefix):]
 285         action = file["action"]
 286
 287         if action == "delete":
 288             gitStream.write("D %s\n" % relPath)
 289         else:
 290             mode = 644
 291             if file["type"].startswith("x"):
 292                 mode = 755
 293
 294             data = os.popen("p4 print -q \"%s\"" % depotPath, "rb").read()
 295
 296             gitStream.write("M %s inline %s\n" % (mode, relPath))
 297             gitStream.write("data %s\n" % len(data))
 298             gitStream.write(data)
 299             gitStream.write("\n")
 300
 301     gitStream.write("\n")
 302
 303     lastChange = int(details["change"])
 304
 305 def getUserMap():
 306     users = {}
 307
 308     for output in p4CmdList("users"):
 309         if not output.has_key("User"):
 310             continue
 311         users[output["User"]] = output["FullName"] + " <" + output["Email"] + ">"
 312     return users
 313
 314 users = getUserMap()
 315
 316 if len(changeRange) == 0:
 317     try:
 318         sout, sin, serr = popen2.popen3("git-name-rev --tags `git-rev-parse %s`" % branch)
 319         output = sout.read()
 320         if output.endswith("\n"):
 321             output = output[:-1]
 322         tagIdx = output.index(" tags/p4/")
 323         caretIdx = output.find("^")
 324         endPos = len(output)
 325         if caretIdx != -1:
 326             endPos = caretIdx
 327         rev = int(output[tagIdx + 9 : endPos]) + 1
 328         changeRange = "@%s,#head" % rev
 329         initialParent = os.popen("git-rev-parse %s" % branch).read()[:-1]
 330         initialTag = "p4/%s" % (int(rev) - 1)
 331     except:
 332         pass
 333
 334 sys.stderr.write("\n")
 335
 336 tz = - time.timezone / 36
 337 tzsign = ("%s" % tz)[0]
 338 if tzsign != '+' and tzsign != '-':
 339     tz = "+" + ("%s" % tz)
 340
 341 gitOutput, gitStream, gitError = popen2.popen3("git-fast-import")
 342
 343 if len(revision) > 0:
 344     print "Doing initial import of %s from revision %s" % (globalPrefix, revision)
 345
 346     details = { "user" : "git perforce import user", "time" : int(time.time()) }
 347     details["desc"] = "Initial import of %s from the state at revision %s" % (globalPrefix, revision)
 348     details["change"] = revision
 349     newestRevision = 0
 350
 351     fileCnt = 0
 352     for info in p4CmdList("files %s...%s" % (globalPrefix, revision)):
 353         change = int(info["change"])
 354         if change > newestRevision:
 355             newestRevision = change
 356
 357         if info["action"] == "delete":
 358             continue
 359
 360         for prop in [ "depotFile", "rev", "action", "type" ]:
 361             details["%s%s" % (prop, fileCnt)] = info[prop]
 362
 363         fileCnt = fileCnt + 1
 364
 365     details["change"] = newestRevision
 366
 367     try:
 368         commit(details, extractFilesFromCommit(details), branch, globalPrefix)
 369     except:
 370         print gitError.read()
 371
 372 else:
 373     changes = []
 374
 375     if len(changesFile) > 0:
 376         output = open(changesFile).readlines()
 377         changeSet = Set()
 378         for line in output:
 379             changeSet.add(int(line))
 380
 381         for change in changeSet:
 382             changes.append(change)
 383
 384         changes.sort()
 385     else:
 386         output = os.popen("p4 changes %s...%s" % (globalPrefix, changeRange)).readlines()
 387
 388         for line in output:
 389             changeNum = line.split(" ")[1]
 390             changes.append(changeNum)
 391
 392         changes.reverse()
 393
 394     if len(changes) == 0:
 395         if not silent:
 396             print "no changes to import!"
 397         sys.exit(1)
 398
 399     cnt = 1
 400     for change in changes:
 401         description = p4Cmd("describe %s" % change)
 402
 403         if not silent:
 404             sys.stdout.write("\rimporting revision %s (%s%%)" % (change, cnt * 100 / len(changes)))
 405             sys.stdout.flush()
 406         cnt = cnt + 1
 407
 408 #        try:
 409         files = extractFilesFromCommit(description)
 410         if detectBranches:
 411             for branch in branchesForCommit(files):
 412                 knownBranches.add(branch)
 413                 branchPrefix = globalPrefix + branch + "/"
 414                 branch = "refs/heads/" + branch
 415                 commit(description, files, branch, branchPrefix)
 416         else:
 417             commit(description, files, branch, globalPrefix)
 418 #        except:
 419 #            print gitError.read()
 420 #            sys.exit(1)
 421
 422 if not silent:
 423     print ""
 424
 425 gitStream.write("reset refs/tags/p4/%s\n" % lastChange)
 426 gitStream.write("from %s\n\n" % branch);
 427
 428
 429 gitStream.close()
 430 gitOutput.close()
 431 gitError.close()
 432
 433 os.popen("git-repo-config p4.depotpath %s" % globalPrefix).read()
 434 if len(initialTag) > 0:
 435     os.popen("git tag -d %s" % initialTag).read()
 436
 437 sys.exit(0)