# Author: Simon Hausmann <hausmann@kde.org>
# License: MIT <http://www.opensource.org/licenses/mit-license.php>
#
-# TODO: - fix date parsing (how hard can it be?)
+# TODO:
# - support integrations (at least p4i)
-# - support incremental imports
-# - create tags
-# - instead of reading all files into a variable try to pipe from
-# - p4 print directly to stdout. need to figure out file size somehow
-# though.
# - support p4 submit (hah!)
-# - don't hardcode the import to master
+# - emulate p4's delete behavior: if a directory becomes empty delete it. continue
+# with parent dir until non-empty dir is found.
#
-import os, string, sys
-
-if len(sys.argv) != 2:
- sys.stderr.write("usage: %s //depot/path[@revRange]\n" % sys.argv[0]);
- sys.stderr.write("\n example:\n");
- sys.stderr.write(" %s //depot/my/project -- to import everything\n");
- sys.stderr.write(" %s //depot/my/project@1,6 -- to import only from revision 1 to 6\n");
- sys.stderr.write("\n");
+import os, string, sys, time, os.path
+import marshal, popen2, getopt, sha
+from sets import Set;
+
+cacheDebug = False
+
+silent = False
+knownBranches = Set()
+createdBranches = Set()
+committedChanges = Set()
+branch = "refs/heads/master"
+globalPrefix = previousDepotPath = os.popen("git-repo-config --get p4.depotpath").read()
+detectBranches = False
+changesFile = ""
+if len(globalPrefix) != 0:
+ globalPrefix = globalPrefix[:-1]
+
+try:
+ opts, args = getopt.getopt(sys.argv[1:], "", [ "branch=", "detect-branches", "changesfile=", "silent", "known-branches=" ])
+except getopt.GetoptError:
+ print "fixme, syntax error"
+ sys.exit(1)
+
+for o, a in opts:
+ if o == "--branch":
+ branch = "refs/heads/" + a
+ elif o == "--detect-branches":
+ detectBranches = True
+ elif o == "--changesfile":
+ changesFile = a
+ elif o == "--silent":
+ silent= True
+ elif o == "--known-branches":
+ for branch in o.split(","):
+ knownBranches.add(branch)
+
+if len(args) == 0 and len(globalPrefix) != 0:
+ if not silent:
+ print "[using previously specified depot path %s]" % globalPrefix
+elif len(args) != 1:
+ print "usage: %s //depot/path[@revRange]" % sys.argv[0]
+ print "\n example:"
+ print " %s //depot/my/project/ -- to import the current head"
+ print " %s //depot/my/project/@all -- to import everything"
+ print " %s //depot/my/project/@1,6 -- to import only from revision 1 to 6"
+ print ""
+ print " (a ... is not needed in the path p4 specification, it's added implicitly)"
+ print ""
sys.exit(1)
+else:
+ if len(globalPrefix) != 0 and globalPrefix != args[0]:
+ print "previous import used depot path %s and now %s was specified. this doesn't work!" % (globalPrefix, args[0])
+ sys.exit(1)
+ globalPrefix = args[0]
-prefix = sys.argv[1]
changeRange = ""
-try:
- atIdx = prefix.index("@")
- changeRange = prefix[atIdx:]
- prefix = prefix[0:atIdx]
-except ValueError:
- changeRange = ""
+revision = ""
+users = {}
+initialParent = ""
+lastChange = 0
+initialTag = ""
+
+if globalPrefix.find("@") != -1:
+ atIdx = globalPrefix.index("@")
+ changeRange = globalPrefix[atIdx:]
+ if changeRange == "@all":
+ changeRange = ""
+ elif changeRange.find(",") == -1:
+ revision = changeRange
+ changeRange = ""
+ globalPrefix = globalPrefix[0:atIdx]
+elif globalPrefix.find("#") != -1:
+ hashIdx = globalPrefix.index("#")
+ revision = globalPrefix[hashIdx:]
+ globalPrefix = globalPrefix[0:hashIdx]
+elif len(previousDepotPath) == 0:
+ revision = "#head"
+
+if globalPrefix.endswith("..."):
+ globalPrefix = globalPrefix[:-3]
+
+if not globalPrefix.endswith("/"):
+ globalPrefix += "/"
+
+def p4File(depotPath):
+ cacheKey = "/tmp/p4cache/data-" + sha.new(depotPath).hexdigest()
+
+ data = 0
+ try:
+ if not cacheDebug:
+ raise
+ data = open(cacheKey, "rb").read()
+ except:
+ data = os.popen("p4 print -q \"%s\"" % depotPath, "rb").read()
+ if cacheDebug:
+ open(cacheKey, "wb").write(data)
-def describe(change):
- output = os.popen("p4 describe %s" % change).readlines()
+ return data
- firstLine = output[0]
+def p4CmdList(cmd):
+ fullCmd = "p4 -G %s" % cmd;
- author = firstLine.split(" ")[3]
- author = author[:author.find("@")]
+ cacheKey = sha.new(fullCmd).hexdigest()
+ cacheKey = "/tmp/p4cache/cmd-" + cacheKey
- filesSection = 0
+ cached = True
+ pipe = 0
try:
- filesSection = output.index("Affected files ...\n")
- except ValueError:
- sys.stderr.write("Change %s doesn't seem to affect any files. Weird.\n" % change)
- return [], [], [], []
-
- differencesSection = 0
+ if not cacheDebug:
+ raise
+ pipe = open(cacheKey, "rb")
+ except:
+ cached = False
+ pipe = os.popen(fullCmd, "rb")
+
+ result = []
try:
- differencesSection = output.index("Differences ...\n")
- except ValueError:
- sys.stderr.write("Change %s doesn't seem to have a differences section. Weird.\n" % change)
- return [], [], [], []
+ while True:
+ entry = marshal.load(pipe)
+ result.append(entry)
+ except EOFError:
+ pass
+ pipe.close()
+
+ if not cached and cacheDebug:
+ pipe = open(cacheKey, "wb")
+ for r in result:
+ marshal.dump(r, pipe)
+ pipe.close()
+
+ return result
+
+def p4Cmd(cmd):
+ list = p4CmdList(cmd)
+ result = {}
+ for entry in list:
+ result.update(entry)
+ return result;
+
+def extractFilesFromCommit(commit):
+ files = []
+ fnum = 0
+ while commit.has_key("depotFile%s" % fnum):
+ path = commit["depotFile%s" % fnum]
+ if not path.startswith(globalPrefix):
+# if not silent:
+# print "\nchanged files: ignoring path %s outside of %s in change %s" % (path, globalPrefix, change)
+ fnum = fnum + 1
+ continue
- log = output[2:filesSection - 1]
+ file = {}
+ file["path"] = path
+ file["rev"] = commit["rev%s" % fnum]
+ file["action"] = commit["action%s" % fnum]
+ file["type"] = commit["type%s" % fnum]
+ files.append(file)
+ fnum = fnum + 1
+ return files
+
+def isSubPathOf(first, second):
+ if not first.startswith(second):
+ return False
+ if first == second:
+ return True
+ return first[len(second)] == "/"
+
+def branchesForCommit(files):
+ global knownBranches
+ branches = Set()
+
+ for file in files:
+ relativePath = file["path"][len(globalPrefix):]
+ # strip off the filename
+ relativePath = relativePath[0:relativePath.rfind("/")]
+
+# if len(branches) == 0:
+# branches.add(relativePath)
+# knownBranches.add(relativePath)
+# continue
+
+ ###### this needs more testing :)
+ knownBranch = False
+ for branch in branches:
+ if relativePath == branch:
+ knownBranch = True
+ break
+# if relativePath.startswith(branch):
+ if isSubPathOf(relativePath, branch):
+ knownBranch = True
+ break
+# if branch.startswith(relativePath):
+ if isSubPathOf(branch, relativePath):
+ branches.remove(branch)
+ break
+
+ if knownBranch:
+ continue
- lines = output[filesSection + 2:differencesSection - 1]
+ for branch in knownBranches:
+ #if relativePath.startswith(branch):
+ if isSubPathOf(relativePath, branch):
+ if len(branches) == 0:
+ relativePath = branch
+ else:
+ knownBranch = True
+ break
- changed = []
- removed = []
+ if knownBranch:
+ continue
- for line in lines:
- # chop off "... " and trailing newline
- line = line[4:len(line) - 1]
+ branches.add(relativePath)
+ knownBranches.add(relativePath)
- lastSpace = line.rfind(" ")
- if lastSpace == -1:
- sys.stderr.write("trouble parsing line %s, skipping!\n" % line)
+ return branches
+
+def findBranchParent(branchPrefix, files):
+ for file in files:
+ path = file["path"]
+ if not path.startswith(branchPrefix):
+ continue
+ action = file["action"]
+ if action != "integrate" and action != "branch":
+ continue
+ rev = file["rev"]
+ depotPath = path + "#" + rev
+
+ log = p4CmdList("filelog \"%s\"" % depotPath)
+ if len(log) != 1:
+ print "eek! I got confused by the filelog of %s" % depotPath
+ sys.exit(1);
+
+ log = log[0]
+ if log["action0"] != action:
+ print "eek! wrong action in filelog for %s : found %s, expected %s" % (depotPath, log["action0"], action)
+ sys.exit(1);
+
+ branchAction = log["how0,0"]
+# if branchAction == "branch into" or branchAction == "ignored":
+# continue # ignore for branching
+
+ if not branchAction.endswith(" from"):
+ continue # ignore for branching
+# print "eek! file %s was not branched from but instead: %s" % (depotPath, branchAction)
+# sys.exit(1);
+
+ source = log["file0,0"]
+ if source.startswith(branchPrefix):
continue
- operation = line[lastSpace + 1:]
- path = line[:lastSpace]
+ lastSourceRev = log["erev0,0"]
+
+ sourceLog = p4CmdList("filelog -m 1 \"%s%s\"" % (source, lastSourceRev))
+ if len(sourceLog) != 1:
+ print "eek! I got confused by the source filelog of %s%s" % (source, lastSourceRev)
+ sys.exit(1);
+ sourceLog = sourceLog[0]
+
+ relPath = source[len(globalPrefix):]
+ # strip off the filename
+ relPath = relPath[0:relPath.rfind("/")]
+
+ for branch in knownBranches:
+ if isSubPathOf(relPath, branch):
+# print "determined parent branch branch %s due to change in file %s" % (branch, source)
+ return "refs/heads/%s" % branch
+# else:
+# print "%s is not a subpath of branch %s" % (relPath, branch)
- if operation == "delete":
- removed.append(path)
+ return ""
+
+def commit(details, files, branch, branchPrefix, parent):
+ global users
+ global lastChange
+ global committedChanges
+
+ epoch = details["time"]
+ author = details["user"]
+
+ gitStream.write("commit %s\n" % branch)
+ gitStream.write("mark :%s\n" % details["change"])
+ committedChanges.add(int(details["change"]))
+ committer = ""
+ if author in users:
+ committer = "%s %s %s" % (users[author], epoch, tz)
+ else:
+ committer = "%s <a@b> %s %s" % (author, epoch, tz)
+
+ gitStream.write("committer %s\n" % committer)
+
+ gitStream.write("data <<EOT\n")
+ gitStream.write(details["desc"])
+ gitStream.write("\n[ imported from %s; change %s ]\n" % (branchPrefix, details["change"]))
+ gitStream.write("EOT\n\n")
+
+ if len(parent) > 0:
+ gitStream.write("from %s\n" % parent)
+
+ for file in files:
+ path = file["path"]
+ if not path.startswith(branchPrefix):
+# if not silent:
+# print "\nchanged files: ignoring path %s outside of branch prefix %s in change %s" % (path, branchPrefix, details["change"])
+ continue
+ rev = file["rev"]
+ depotPath = path + "#" + rev
+ relPath = path[len(branchPrefix):]
+ action = file["action"]
+
+ if action == "delete":
+ gitStream.write("D %s\n" % relPath)
else:
- changed.append(path)
+ mode = 644
+ if file["type"].startswith("x"):
+ mode = 755
- return author, log, changed, removed
+ data = p4File(depotPath)
-def p4cat(path):
- return os.popen("p4 print -q \"%s\"" % path).read()
+ gitStream.write("M %s inline %s\n" % (mode, relPath))
+ gitStream.write("data %s\n" % len(data))
+ gitStream.write(data)
+ gitStream.write("\n")
-def stripRevision(path):
- hashPos = path.rindex("#")
- return path[:hashPos]
+ gitStream.write("\n")
+
+ lastChange = int(details["change"])
def getUserMap():
users = {}
- output = os.popen("p4 users")
- for line in output:
- firstSpace = line.index(" ")
- secondSpace = line.index(" ", firstSpace + 1)
- key = line[:firstSpace]
- email = line[firstSpace + 1:secondSpace]
- openParenPos = line.index("(", secondSpace)
- closedParenPos = line.index(")", openParenPos)
- name = line[openParenPos + 1:closedParenPos]
-
- users[key] = name + " " + email
+ for output in p4CmdList("users"):
+ if not output.has_key("User"):
+ continue
+ users[output["User"]] = output["FullName"] + " <" + output["Email"] + ">"
return users
-
users = getUserMap()
-output = os.popen("p4 changes %s...%s" % (prefix, changeRange)).readlines()
+if len(changeRange) == 0:
+ try:
+ sout, sin, serr = popen2.popen3("git-name-rev --tags `git-rev-parse %s`" % branch)
+ output = sout.read()
+ if output.endswith("\n"):
+ output = output[:-1]
+ tagIdx = output.index(" tags/p4/")
+ caretIdx = output.find("^")
+ endPos = len(output)
+ if caretIdx != -1:
+ endPos = caretIdx
+ rev = int(output[tagIdx + 9 : endPos]) + 1
+ changeRange = "@%s,#head" % rev
+ initialParent = os.popen("git-rev-parse %s" % branch).read()[:-1]
+ initialTag = "p4/%s" % (int(rev) - 1)
+ except:
+ pass
+
+tz = - time.timezone / 36
+tzsign = ("%s" % tz)[0]
+if tzsign != '+' and tzsign != '-':
+ tz = "+" + ("%s" % tz)
+
+gitOutput, gitStream, gitError = popen2.popen3("git-fast-import")
+
+if len(revision) > 0:
+ print "Doing initial import of %s from revision %s" % (globalPrefix, revision)
+
+ details = { "user" : "git perforce import user", "time" : int(time.time()) }
+ details["desc"] = "Initial import of %s from the state at revision %s" % (globalPrefix, revision)
+ details["change"] = revision
+ newestRevision = 0
+
+ fileCnt = 0
+ for info in p4CmdList("files %s...%s" % (globalPrefix, revision)):
+ change = int(info["change"])
+ if change > newestRevision:
+ newestRevision = change
+
+ if info["action"] == "delete":
+ continue
+
+ for prop in [ "depotFile", "rev", "action", "type" ]:
+ details["%s%s" % (prop, fileCnt)] = info[prop]
-changes = []
-for line in output:
- changeNum = line.split(" ")[1]
- changes.append(changeNum)
+ fileCnt = fileCnt + 1
-changes.reverse()
+ details["change"] = newestRevision
-sys.stderr.write("\n")
+ try:
+ commit(details, extractFilesFromCommit(details), branch, globalPrefix)
+ except:
+ print gitError.read()
-cnt = 0
-for change in changes:
- [ author, log, changedFiles, removedFiles ] = describe(change)
- sys.stderr.write("\rimporting revision %s (%s%%)" % (change, cnt * 100 / len(changes)))
- cnt = cnt + 1
-# sys.stderr.write("%s\n" % log)
-# sys.stderr.write("%s\n" % changedFiles)
-# sys.stderr.write("%s\n" % removedFiles)
+else:
+ changes = []
- print "commit refs/heads/master"
- if author in users:
- print "committer %s 1 2" % users[author]
+ if len(changesFile) > 0:
+ output = open(changesFile).readlines()
+ changeSet = Set()
+ for line in output:
+ changeSet.add(int(line))
+
+ for change in changeSet:
+ changes.append(change)
+
+ changes.sort()
else:
- print "committer %s <a@b> 1 2" % author
- print "data <<EOT"
- for l in log:
- print l[:len(l) - 1]
- print "EOT"
+ output = os.popen("p4 changes %s...%s" % (globalPrefix, changeRange)).readlines()
+
+ for line in output:
+ changeNum = line.split(" ")[1]
+ changes.append(changeNum)
+
+ changes.reverse()
+
+ if len(changes) == 0:
+ if not silent:
+ print "no changes to import!"
+ sys.exit(1)
+
+ cnt = 1
+ for change in changes:
+ description = p4Cmd("describe %s" % change)
+
+ if not silent:
+ sys.stdout.write("\rimporting revision %s (%s%%)" % (change, cnt * 100 / len(changes)))
+ sys.stdout.flush()
+ cnt = cnt + 1
+
+# try:
+ files = extractFilesFromCommit(description)
+ if detectBranches:
+ for branch in branchesForCommit(files):
+ knownBranches.add(branch)
+ branchPrefix = globalPrefix + branch + "/"
+
+ parent = ""
+ ########### remove cnt!!!
+ if branch not in createdBranches and cnt > 2:
+ createdBranches.add(branch)
+ parent = findBranchParent(branchPrefix, files)
+ if parent == branch:
+ parent = ""
+# elif len(parent) > 0:
+# print "%s branched off of %s" % (branch, parent)
+
+ branch = "refs/heads/" + branch
+ commit(description, files, branch, branchPrefix, parent)
+ else:
+ commit(description, files, branch, globalPrefix, initialParent)
+ initialParent = ""
+# except:
+# print gitError.read()
+# sys.exit(1)
+if not silent:
print ""
- for f in changedFiles:
- if not f.startswith(prefix):
- sys.stderr.write("\nchanged files: ignoring path %s outside of %s in change %s\n" % (f, prefix, change))
- continue
- relpath = f[len(prefix):]
- print "M 644 inline %s" % stripRevision(relpath)
- data = p4cat(f)
- print "data %s" % len(data)
- sys.stdout.write(data)
- print ""
-
- for f in removedFiles:
- if not f.startswith(prefix):
- sys.stderr.write("\ndeleted files: ignoring path %s outside of %s in change %s\n" % (f, prefix, change))
- continue
- relpath = f[len(prefix):]
- print "D %s" % stripRevision(relpath)
+gitStream.write("reset refs/tags/p4/%s\n" % lastChange)
+gitStream.write("from %s\n\n" % branch);
- print ""
+gitStream.close()
+gitOutput.close()
+gitError.close()
+
+os.popen("git-repo-config p4.depotpath %s" % globalPrefix).read()
+if len(initialTag) > 0:
+ os.popen("git tag -d %s" % initialTag).read()
+
+sys.exit(0)