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