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     mergedBranches = set()
 175
 176     for file in files:
 177         path = file["path"]
 178         if not path.startswith(branchPrefix):
 179             continue
 180         action = file["action"]
 181         if action != "integrate" and action != "branch":
 182             continue
 183         rev = file["rev"]
 184         depotPath = path + "#" + rev
 185
 186         log = p4CmdList("filelog \"%s\"" % depotPath)
 187         if len(log) != 1:
 188             print "eek! I got confused by the filelog of %s" % depotPath
 189             sys.exit(1);
 190
 191         log = log[0]
 192         if log["action0"] != action:
 193             print "eek! wrong action in filelog for %s : found %s, expected %s" % (depotPath, log["action0"], action)
 194             sys.exit(1);
 195
 196         if not log["how0,0"].endswith(" from"):
 197             print "eek! file %s was not branched but instead: %s" % (depotPath, log["how0,0"])
 198             sys.exit(1);
 199
 200         source = log["file0,0"]
 201         if source.startswith(branchPrefix):
 202             continue
 203
 204         relPath = source[len(globalPrefix):]
 205
 206         for branch in knownBranches:
 207             if relPath.startswith(branch) and branch not in mergedBranches:
 208                 gitStream.write("merge refs/heads/%s\n" % branch)
 209                 mergedBranches.add(branch)
 210                 break
 211
 212     for file in files:
 213         path = file["path"]
 214         if not path.startswith(branchPrefix):
 215             print "\nchanged files: ignoring path %s outside of branch prefix %s in change %s" % (path, branchPrefix, change)
 216             continue
 217         rev = file["rev"]
 218         depotPath = path + "#" + rev
 219         relPath = path[len(branchPrefix):]
 220         action = file["action"]
 221
 222         if action == "delete":
 223             gitStream.write("D %s\n" % relPath)
 224         else:
 225             mode = 644
 226             if file["type"].startswith("x"):
 227                 mode = 755
 228
 229             data = os.popen("p4 print -q \"%s\"" % depotPath, "rb").read()
 230
 231             gitStream.write("M %s inline %s\n" % (mode, relPath))
 232             gitStream.write("data %s\n" % len(data))
 233             gitStream.write(data)
 234             gitStream.write("\n")
 235
 236     gitStream.write("\n")
 237
 238     lastChange = details["change"]
 239
 240 def getUserMap():
 241     users = {}
 242
 243     for output in p4CmdList("users"):
 244         if not output.has_key("User"):
 245             continue
 246         users[output["User"]] = output["FullName"] + " <" + output["Email"] + ">"
 247     return users
 248
 249 users = getUserMap()
 250
 251 if len(changeRange) == 0:
 252     try:
 253         sout, sin, serr = popen2.popen3("git-name-rev --tags `git-rev-parse %s`" % branch)
 254         output = sout.read()
 255         if output.endswith("\n"):
 256             output = output[:-1]
 257         tagIdx = output.index(" tags/p4/")
 258         caretIdx = output.find("^")
 259         endPos = len(output)
 260         if caretIdx != -1:
 261             endPos = caretIdx
 262         rev = int(output[tagIdx + 9 : endPos]) + 1
 263         changeRange = "@%s,#head" % rev
 264         initialParent = os.popen("git-rev-parse %s" % branch).read()[:-1]
 265         initialTag = "p4/%s" % (int(rev) - 1)
 266     except:
 267         pass
 268
 269 sys.stderr.write("\n")
 270
 271 tz = - time.timezone / 36
 272 tzsign = ("%s" % tz)[0]
 273 if tzsign != '+' and tzsign != '-':
 274     tz = "+" + ("%s" % tz)
 275
 276 gitOutput, gitStream, gitError = popen2.popen3("git-fast-import")
 277
 278 if len(revision) > 0:
 279     print "Doing initial import of %s from revision %s" % (globalPrefix, revision)
 280
 281     details = { "user" : "git perforce import user", "time" : int(time.time()) }
 282     details["desc"] = "Initial import of %s from the state at revision %s" % (globalPrefix, revision)
 283     details["change"] = revision
 284     newestRevision = 0
 285
 286     fileCnt = 0
 287     for info in p4CmdList("files %s...%s" % (globalPrefix, revision)):
 288         change = int(info["change"])
 289         if change > newestRevision:
 290             newestRevision = change
 291
 292         if info["action"] == "delete":
 293             continue
 294
 295         for prop in [ "depotFile", "rev", "action", "type" ]:
 296             details["%s%s" % (prop, fileCnt)] = info[prop]
 297
 298         fileCnt = fileCnt + 1
 299
 300     details["change"] = newestRevision
 301
 302     try:
 303         commit(details, extractFilesFromCommit(details), branch, globalPrefix)
 304     except:
 305         print gitError.read()
 306
 307 else:
 308     output = os.popen("p4 changes %s...%s" % (globalPrefix, changeRange)).readlines()
 309
 310     changes = []
 311     for line in output:
 312         changeNum = line.split(" ")[1]
 313         changes.append(changeNum)
 314
 315     changes.reverse()
 316
 317     if len(changes) == 0:
 318         print "no changes to import!"
 319         sys.exit(1)
 320
 321     cnt = 1
 322     for change in changes:
 323         description = p4Cmd("describe %s" % change)
 324
 325         sys.stdout.write("\rimporting revision %s (%s%%)" % (change, cnt * 100 / len(changes)))
 326         sys.stdout.flush()
 327         cnt = cnt + 1
 328
 329         try:
 330             files = extractFilesFromCommit(description)
 331             if detectBranches:
 332                 for branch in branchesForCommit(files):
 333                     knownBranches.add(branch)
 334                     branchPrefix = globalPrefix + branch + "/"
 335                     branch = "refs/heads/" + branch
 336                     commit(description, files, branch, branchPrefix)
 337             else:
 338                 commit(description, files, branch, globalPrefix)
 339         except:
 340             print gitError.read()
 341             sys.exit(1)
 342
 343 print ""
 344
 345 gitStream.write("reset refs/tags/p4/%s\n" % lastChange)
 346 gitStream.write("from %s\n\n" % branch);
 347
 348
 349 gitStream.close()
 350 gitOutput.close()
 351 gitError.close()
 352
 353 os.popen("git-repo-config p4.depotpath %s" % globalPrefix).read()
 354 if len(initialTag) > 0:
 355     os.popen("git tag -d %s" % initialTag).read()
 356
 357 sys.exit(0)