-#!/usr/bin/python
+#!/usr/bin/env python
#
# p4-fast-export.py
#
# - 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, time
-import marshal, popen2, getopt
+import os, string, sys, time, os.path
+import marshal, popen2, getopt, sha
from sets import Set;
+dataCache = False
+commandCache = False
+
silent = False
knownBranches = Set()
+createdBranches = Set()
committedChanges = Set()
branch = "refs/heads/master"
globalPrefix = previousDepotPath = os.popen("git-repo-config --get p4.depotpath").read()
globalPrefix = globalPrefix[:-1]
try:
- opts, args = getopt.getopt(sys.argv[1:], "", [ "branch=", "detect-branches", "changesfile=", "silent" ])
+ opts, args = getopt.getopt(sys.argv[1:], "", [ "branch=", "detect-branches", "changesfile=", "silent", "known-branches=",
+ "cache", "command-cache" ])
except getopt.GetoptError:
print "fixme, syntax error"
sys.exit(1)
changesFile = a
elif o == "--silent":
silent= True
+ elif o == "--known-branches":
+ for branch in open(a).readlines():
+ knownBranches.add(branch[:-1])
+ elif o == "--cache":
+ dataCache = True
+ commandCache = True
+ elif o == "--command-cache":
+ commandCache = True
if len(args) == 0 and len(globalPrefix) != 0:
if not silent:
if not globalPrefix.endswith("/"):
globalPrefix += "/"
+def p4File(depotPath):
+ cacheKey = "/tmp/p4cache/data-" + sha.new(depotPath).hexdigest()
+
+ data = 0
+ try:
+ if not dataCache:
+ raise
+ data = open(cacheKey, "rb").read()
+ except:
+ data = os.popen("p4 print -q \"%s\"" % depotPath, "rb").read()
+ if dataCache:
+ open(cacheKey, "wb").write(data)
+
+ return data
+
def p4CmdList(cmd):
- pipe = os.popen("p4 -G %s" % cmd, "rb")
+ fullCmd = "p4 -G %s" % cmd;
+
+ cacheKey = sha.new(fullCmd).hexdigest()
+ cacheKey = "/tmp/p4cache/cmd-" + cacheKey
+
+ cached = True
+ pipe = 0
+ try:
+ if not commandCache:
+ raise
+ pipe = open(cacheKey, "rb")
+ except:
+ cached = False
+ pipe = os.popen(fullCmd, "rb")
+
result = []
try:
while True:
except EOFError:
pass
pipe.close()
+
+ if not cached and commandCache:
+ pipe = open(cacheKey, "wb")
+ for r in result:
+ marshal.dump(r, pipe)
+ pipe.close()
+
return result
def p4Cmd(cmd):
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)
+# if not silent:
+# print "\nchanged files: ignoring path %s outside of %s in change %s" % (path, globalPrefix, change)
fnum = fnum + 1
continue
return branches
-def commit(details, files, branch, branchPrefix):
- global initialParent
- 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(initialParent) > 0:
- gitStream.write("from %s\n" % initialParent)
- initialParent = ""
-
- #mergedBranches = Set()
- merges = Set()
-
+def findBranchParent(branchPrefix, files):
for file in files:
- if lastChange == 0 or not detectBranches:
- continue
path = file["path"]
if not path.startswith(branchPrefix):
continue
sys.exit(1);
sourceLog = sourceLog[0]
- change = int(sourceLog["change0"])
- merges.add(change)
+ relPath = source[len(globalPrefix):]
+ # strip off the filename
+ relPath = relPath[0:relPath.rfind("/")]
-# relPath = source[len(globalPrefix):]
-#
-# for branch in knownBranches:
-# if relPath.startswith(branch) and branch not in mergedBranches:
-# gitStream.write("merge refs/heads/%s\n" % branch)
-# mergedBranches.add(branch)
-# break
+ for branch in knownBranches:
+ if isSubPathOf(relPath, branch):
+# print "determined parent branch branch %s due to change in file %s" % (branch, source)
+ return branch
+# else:
+# print "%s is not a subpath of branch %s" % (relPath, branch)
+
+ return ""
+
+def commit(details, files, branch, branchPrefix, parent, merged = ""):
+ global users
+ global lastChange
+ global committedChanges
- for merge in merges:
- if merge in committedChanges:
- gitStream.write("merge :%s\n" % merge)
+ 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)
+
+ if len(merged) > 0:
+ gitStream.write("merge %s\n" % merged)
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"])
+# 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 file["type"] == "apple":
+ print "\nfile %s is a strange apple file that forks. Ignoring!" % path
+ continue
+
if action == "delete":
gitStream.write("D %s\n" % relPath)
else:
if file["type"].startswith("x"):
mode = 755
- data = os.popen("p4 print -q \"%s\"" % depotPath, "rb").read()
+ data = p4File(depotPath)
gitStream.write("M %s inline %s\n" % (mode, relPath))
gitStream.write("data %s\n" % len(data))
lastChange = int(details["change"])
+def extractFilesInCommitToBranch(files, branchPrefix):
+ newFiles = []
+
+ for file in files:
+ path = file["path"]
+ if path.startswith(branchPrefix):
+ newFiles.append(file)
+
+ return newFiles
+
+def findBranchSourceHeuristic(files, branch, branchPrefix):
+ for file in files:
+ action = file["action"]
+ if action != "integrate" and action != "branch":
+ continue
+ path = file["path"]
+ 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 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
+
+ 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 candidate in knownBranches:
+ if isSubPathOf(relPath, candidate) and candidate != branch:
+ return candidate
+
+ return ""
+
+def changeIsBranchMerge(sourceBranch, destinationBranch, change):
+ sourceFiles = {}
+ for file in p4CmdList("files %s...@%s" % (globalPrefix + sourceBranch + "/", change)):
+ if file["action"] == "delete":
+ continue
+ sourceFiles[file["depotFile"]] = file
+
+ destinationFiles = {}
+ for file in p4CmdList("files %s...@%s" % (globalPrefix + destinationBranch + "/", change)):
+ destinationFiles[file["depotFile"]] = file
+
+ for fileName in sourceFiles.keys():
+ integrations = []
+ deleted = False
+ integrationCount = 0
+ for integration in p4CmdList("integrated \"%s\"" % fileName):
+ toFile = integration["fromFile"] # yes, it's true, it's fromFile
+ if not toFile in destinationFiles:
+ continue
+ destFile = destinationFiles[toFile]
+ if destFile["action"] == "delete":
+# print "file %s has been deleted in %s" % (fileName, toFile)
+ deleted = True
+ break
+ integrationCount += 1
+ if integration["how"] == "branch from":
+ continue
+
+ if int(integration["change"]) == change:
+ integrations.append(integration)
+ continue
+ if int(integration["change"]) > change:
+ continue
+
+ destRev = int(destFile["rev"])
+
+ startRev = integration["startFromRev"][1:]
+ if startRev == "none":
+ startRev = 0
+ else:
+ startRev = int(startRev)
+
+ endRev = integration["endFromRev"][1:]
+ if endRev == "none":
+ endRev = 0
+ else:
+ endRev = int(endRev)
+
+ initialBranch = (destRev == 1 and integration["how"] != "branch into")
+ inRange = (destRev >= startRev and destRev <= endRev)
+ newer = (destRev > startRev and destRev > endRev)
+
+ if initialBranch or inRange or newer:
+ integrations.append(integration)
+
+ if deleted:
+ continue
+
+ if len(integrations) == 0 and integrationCount > 1:
+ print "file %s was not integrated from %s into %s" % (fileName, sourceBranch, destinationBranch)
+ return False
+
+ return True
+
def getUserMap():
users = {}
except:
pass
-sys.stderr.write("\n")
-
tz = - time.timezone / 36
tzsign = ("%s" % tz)[0]
if tzsign != '+' and tzsign != '-':
sys.stdout.flush()
cnt = cnt + 1
-# try:
- files = extractFilesFromCommit(description)
- if detectBranches:
- for branch in branchesForCommit(files):
- knownBranches.add(branch)
- branchPrefix = globalPrefix + branch + "/"
- branch = "refs/heads/" + branch
- commit(description, files, branch, branchPrefix)
- else:
- commit(description, files, branch, globalPrefix)
-# except:
-# print gitError.read()
-# sys.exit(1)
+ try:
+ files = extractFilesFromCommit(description)
+ if detectBranches:
+ for branch in branchesForCommit(files):
+ knownBranches.add(branch)
+ branchPrefix = globalPrefix + branch + "/"
+
+ filesForCommit = extractFilesInCommitToBranch(files, branchPrefix)
+
+ merged = ""
+ 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)
+
+ if len(parent) == 0:
+ merged = findBranchSourceHeuristic(filesForCommit, branch, branchPrefix)
+ if len(merged) > 0:
+ print "change %s could be a merge from %s into %s" % (description["change"], merged, branch)
+ if not changeIsBranchMerge(merged, branch, int(description["change"])):
+ merged = ""
+
+ branch = "refs/heads/" + branch
+ if len(parent) > 0:
+ parent = "refs/heads/" + parent
+ if len(merged) > 0:
+ merged = "refs/heads/" + merged
+ commit(description, files, branch, branchPrefix, parent, merged)
+ else:
+ commit(description, files, branch, globalPrefix, initialParent)
+ initialParent = ""
+ except IOError:
+ print gitError.read()
+ sys.exit(1)
if not silent:
print ""