Code

More cleanups and speedups for labels and branches
[git.git] / contrib / fast-import / git-p4
index 2f3615bd7274f22c1132ec96670ba23f519f2d27..49114d2c6f3e377b2805f0e5e3442d71fb306f41 100755 (executable)
@@ -414,11 +414,9 @@ class P4Sync(Command):
                 optparse.make_option("--detect-branches", dest="detectBranches", action="store_true"),
                 optparse.make_option("--changesfile", dest="changesFile"),
                 optparse.make_option("--silent", dest="silent", action="store_true"),
-                optparse.make_option("--known-branches", dest="knownBranches"),
-                optparse.make_option("--data-cache", dest="dataCache", action="store_true"),
-                optparse.make_option("--command-cache", dest="commandCache", action="store_true"),
                 optparse.make_option("--detect-labels", dest="detectLabels", action="store_true"),
-                optparse.make_option("--with-origin", dest="syncWithOrigin", action="store_true")
+                optparse.make_option("--with-origin", dest="syncWithOrigin", action="store_true"),
+                optparse.make_option("--verbose", dest="verbose", action="store_true")
         ]
         self.description = """Imports from Perforce into a git repository.\n
     example:
@@ -430,10 +428,7 @@ class P4Sync(Command):
 
         self.usage += " //depot/path[@revRange]"
 
-        self.dataCache = False
-        self.commandCache = False
         self.silent = False
-        self.knownBranches = Set()
         self.createdBranches = Set()
         self.committedChanges = Set()
         self.branch = ""
@@ -441,6 +436,7 @@ class P4Sync(Command):
         self.detectLabels = False
         self.changesFile = ""
         self.syncWithOrigin = False
+        self.verbose = False
 
     def p4File(self, depotPath):
         return os.popen("p4 print -q \"%s\"" % depotPath, "rb").read()
@@ -465,120 +461,42 @@ class P4Sync(Command):
             fnum = fnum + 1
         return files
 
-    def isSubPathOf(self, first, second):
-        if not first.startswith(second):
-            return False
-        if first == second:
-            return True
-        return first[len(second)] == "/"
-
-    def branchesForCommit(self, files):
-        branches = Set()
-
-        for file in files:
-            relativePath = file["path"][len(self.depotPath):]
-            # 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 self.isSubPathOf(relativePath, branch):
-                    knownBranch = True
-                    break
-    #            if branch.startswith(relativePath):
-                if self.isSubPathOf(branch, relativePath):
-                    branches.remove(branch)
-                    break
+    def splitFilesIntoBranches(self, commit):
+        branches = {}
 
-            if knownBranch:
+        fnum = 0
+        while commit.has_key("depotFile%s" % fnum):
+            path =  commit["depotFile%s" % fnum]
+            if not path.startswith(self.depotPath):
+    #            if not self.silent:
+    #                print "\nchanged files: ignoring path %s outside of %s in change %s" % (path, self.depotPath, change)
+                fnum = fnum + 1
                 continue
 
-            for branch in self.knownBranches:
-                #if relativePath.startswith(branch):
-                if self.isSubPathOf(relativePath, branch):
-                    if len(branches) == 0:
-                        relativePath = branch
-                    else:
-                        knownBranch = True
-                    break
+            file = {}
+            file["path"] = path
+            file["rev"] = commit["rev%s" % fnum]
+            file["action"] = commit["action%s" % fnum]
+            file["type"] = commit["type%s" % fnum]
+            fnum = fnum + 1
 
-            if knownBranch:
-                continue
+            relPath = path[len(self.depotPath):]
 
-            branches.add(relativePath)
-            self.knownBranches.add(relativePath)
+            for branch in self.knownBranches.keys():
+                if relPath.startswith(branch):
+                    if branch not in branches:
+                        branches[branch] = []
+                    branches[branch].append(file)
 
         return branches
 
-    def findBranchParent(self, 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
-
-            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(self.depotPath):]
-            # strip off the filename
-            relPath = relPath[0:relPath.rfind("/")]
-
-            for branch in self.knownBranches:
-                if self.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(self, details, files, branch, branchPrefix, parent = "", merged = ""):
+    def commit(self, details, files, branch, branchPrefix, parent = ""):
         epoch = details["time"]
         author = details["user"]
 
+        if self.verbose:
+            print "commit into %s" % branch
+
         self.gitStream.write("commit %s\n" % branch)
     #    gitStream.write("mark :%s\n" % details["change"])
         self.committedChanges.add(int(details["change"]))
@@ -596,11 +514,10 @@ class P4Sync(Command):
         self.gitStream.write("EOT\n\n")
 
         if len(parent) > 0:
+            if self.verbose:
+                print "parent %s" % parent
             self.gitStream.write("from %s\n" % parent)
 
-        if len(merged) > 0:
-            self.gitStream.write("merge %s\n" % merged)
-
         for file in files:
             path = file["path"]
             if not path.startswith(branchPrefix):
@@ -640,6 +557,8 @@ class P4Sync(Command):
             label = self.labels[change]
             labelDetails = label[0]
             labelRevisions = label[1]
+            if self.verbose:
+                print "Change %s is labelled %s" % (change, labelDetails)
 
             files = p4CmdList("files %s...@%s" % (branchPrefix, change))
 
@@ -674,128 +593,6 @@ class P4Sync(Command):
                 if not self.silent:
                     print "Tag %s does not match with change %s: file count is different." % (labelDetails["label"], change)
 
-    def extractFilesInCommitToBranch(self, files, branchPrefix):
-        newFiles = []
-
-        for file in files:
-            path = file["path"]
-            if path.startswith(branchPrefix):
-                newFiles.append(file)
-
-        return newFiles
-
-    def findBranchSourceHeuristic(self, 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(self.depotPath):]
-            # strip off the filename
-            relPath = relPath[0:relPath.rfind("/")]
-
-            for candidate in self.knownBranches:
-                if self.isSubPathOf(relPath, candidate) and candidate != branch:
-                    return candidate
-
-        return ""
-
-    def changeIsBranchMerge(self, sourceBranch, destinationBranch, change):
-        sourceFiles = {}
-        for file in p4CmdList("files %s...@%s" % (self.depotPath + sourceBranch + "/", change)):
-            if file["action"] == "delete":
-                continue
-            sourceFiles[file["depotFile"]] = file
-
-        destinationFiles = {}
-        for file in p4CmdList("files %s...@%s" % (self.depotPath + 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(self):
         self.users = {}
 
@@ -815,21 +612,61 @@ class P4Sync(Command):
             label = output["label"]
             revisions = {}
             newestChange = 0
-            for file in p4CmdList("files //...@%s" % label):
+            if self.verbose:
+                print "Querying files for label %s" % label
+            for file in p4CmdList("files %s...@%s" % (self.depotPath, label)):
                 revisions[file["depotFile"]] = file["rev"]
                 change = int(file["change"])
                 if change > newestChange:
                     newestChange = change
 
-            self.labels[newestChange] = [output, revisions]
+            self.labels[int(newestChange)] = [output, revisions]
+
+    def getBranchMapping(self):
+        self.projectName = self.depotPath[self.depotPath[:-1].rfind("/") + 1:]
+
+        for info in p4CmdList("branches"):
+            details = p4Cmd("branch -o %s" % info["branch"])
+            viewIdx = 0
+            while details.has_key("View%s" % viewIdx):
+                paths = details["View%s" % viewIdx].split(" ")
+                viewIdx = viewIdx + 1
+                # require standard //depot/foo/... //depot/bar/... mapping
+                if len(paths) != 2 or not paths[0].endswith("/...") or not paths[1].endswith("/..."):
+                    continue
+                source = paths[0]
+                destination = paths[1]
+                if source.startswith(self.depotPath) and destination.startswith(self.depotPath):
+                    source = source[len(self.depotPath):-4]
+                    destination = destination[len(self.depotPath):-4]
+                    if destination not in self.knownBranches:
+                        self.knownBranches[destination] = source
+                    if source not in self.knownBranches:
+                        self.knownBranches[source] = source
+
+    def listExistingP4GitBranches(self):
+        self.p4BranchesInGit = []
+
+        for line in mypopen("git rev-parse --symbolic --remotes").readlines():
+            if line.startswith("p4/") and line != "p4/HEAD\n":
+                branch = line[3:-1]
+                self.p4BranchesInGit.append(branch)
+                self.initialParents["refs/remotes/p4/" + branch] = parseRevision(line[:-1])
 
     def run(self, args):
         self.depotPath = ""
         self.changeRange = ""
         self.initialParent = ""
         self.previousDepotPath = ""
+        # map from branch depot path to parent branch
+        self.knownBranches = {}
+        self.initialParents = {}
+
+        self.listExistingP4GitBranches()
+
+        if self.syncWithOrigin and gitBranchExists("origin") and gitBranchExists("refs/remotes/p4/master") and not self.detectBranches:
+            ### needs to be ported to multi branch import
 
-        if self.syncWithOrigin and gitBranchExists("origin") and gitBranchExists("refs/remotes/p4/master"):
             print "Syncing with origin first as requested by calling git fetch origin"
             system("git fetch origin")
             [originPreviousDepotPath, originP4Change] = extractDepotPathAndChangeFromGitLog(extractLogMessageFromGitCommit("origin"))
@@ -853,7 +690,8 @@ class P4Sync(Command):
                 system("git symbolic-ref refs/remotes/p4/HEAD %s" % self.branch)
 
         if len(args) == 0:
-            if not gitBranchExists(self.branch) and gitBranchExists("origin"):
+            if not gitBranchExists(self.branch) and gitBranchExists("origin") and not self.detectBranches:
+                ### needs to be ported to multi branch import
                 if not self.silent:
                     print "Creating %s branch in git repository based on origin" % self.branch
                 branch = self.branch
@@ -861,9 +699,30 @@ class P4Sync(Command):
                     branch = "refs/heads/" + branch
                 system("git update-ref %s origin" % branch)
 
-            [self.previousDepotPath, p4Change] = extractDepotPathAndChangeFromGitLog(extractLogMessageFromGitCommit(self.branch))
-            if len(self.previousDepotPath) > 0 and len(p4Change) > 0:
-                p4Change = int(p4Change) + 1
+            if self.verbose:
+                print "branches: %s" % self.p4BranchesInGit
+
+            p4Change = 0
+            for branch in self.p4BranchesInGit:
+                depotPath, change = extractDepotPathAndChangeFromGitLog(extractLogMessageFromGitCommit("refs/remotes/p4/" + branch))
+
+                if self.verbose:
+                    print "path %s change %s" % (depotPath, change)
+
+                if len(depotPath) > 0 and len(change) > 0:
+                    change = int(change) + 1
+                    p4Change = max(p4Change, change)
+
+                    if len(self.previousDepotPath) == 0:
+                        self.previousDepotPath = depotPath
+                    else:
+                        i = 0
+                        l = min(len(self.previousDepotPath), len(depotPath))
+                        while i < l and self.previousDepotPath[i] == depotPath[i]:
+                            i = i + 1
+                        self.previousDepotPath = self.previousDepotPath[:i]
+
+            if p4Change > 0:
                 self.depotPath = self.previousDepotPath
                 self.changeRange = "@%s,#head" % p4Change
                 self.initialParent = parseRevision(self.branch)
@@ -918,6 +777,16 @@ class P4Sync(Command):
         if self.detectLabels:
             self.getLabels();
 
+        if self.detectBranches:
+            self.getBranchMapping();
+            if self.verbose:
+                print "p4-git branches: %s" % self.p4BranchesInGit
+                print "initial parents: %s" % self.initialParents
+            for b in self.p4BranchesInGit:
+                if b != "master":
+                    b = b[len(self.projectName):]
+                self.createdBranches.add(b)
+
         self.tz = "%+03d%02d" % (- time.timezone / 3600, ((- time.timezone % 3600) / 60))
 
         importProcess = subprocess.Popen(["git", "fast-import"], stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE);
@@ -971,6 +840,8 @@ class P4Sync(Command):
 
                 changes.sort()
             else:
+                if self.verbose:
+                    print "Getting p4 changes for %s...%s" % (self.depotPath, self.changeRange)
                 output = mypopen("p4 changes %s...%s" % (self.depotPath, self.changeRange)).readlines()
 
                 for line in output:
@@ -994,39 +865,51 @@ class P4Sync(Command):
                 cnt = cnt + 1
 
                 try:
-                    files = self.extractFilesFromCommit(description)
                     if self.detectBranches:
-                        for branch in self.branchesForCommit(files):
-                            self.knownBranches.add(branch)
+                        branches = self.splitFilesIntoBranches(description)
+                        for branch in branches.keys():
                             branchPrefix = self.depotPath + branch + "/"
 
-                            filesForCommit = self.extractFilesInCommitToBranch(files, branchPrefix)
-
-                            merged = ""
                             parent = ""
-                            ########### remove cnt!!!
-                            if branch not in self.createdBranches and cnt > 2:
+
+                            filesForCommit = branches[branch]
+
+                            if self.verbose:
+                                print "branch is %s" % branch
+
+                            if branch not in self.createdBranches:
                                 self.createdBranches.add(branch)
-                                parent = self.findBranchParent(branchPrefix, files)
+                                parent = self.knownBranches[branch]
                                 if parent == branch:
                                     parent = ""
-            #                    elif len(parent) > 0:
-            #                        print "%s branched off of %s" % (branch, parent)
+                                elif self.verbose:
+                                    print "parent determined through known branches: %s" % parent
+
+                            # main branch? use master
+                            if branch == "main":
+                                branch = "master"
+                            else:
+                                branch = self.projectName + branch
 
-                            if len(parent) == 0:
-                                merged = self.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 self.changeIsBranchMerge(merged, branch, int(description["change"])):
-                                        merged = ""
+                            if parent == "main":
+                                parent = "master"
+                            elif len(parent) > 0:
+                                parent = self.projectName + parent
 
-                            branch = "refs/heads/" + branch
+                            branch = "refs/remotes/p4/" + branch
                             if len(parent) > 0:
-                                parent = "refs/heads/" + parent
-                            if len(merged) > 0:
-                                merged = "refs/heads/" + merged
-                            self.commit(description, files, branch, branchPrefix, parent, merged)
+                                parent = "refs/remotes/p4/" + parent
+
+                            if self.verbose:
+                                print "looking for initial parent for %s; current parent is %s" % (branch, parent)
+
+                            if len(parent) == 0 and branch in self.initialParents:
+                                parent = self.initialParents[branch]
+                                del self.initialParents[branch]
+
+                            self.commit(description, filesForCommit, branch, branchPrefix, parent)
                     else:
+                        files = self.extractFilesFromCommit(description)
                         self.commit(description, files, self.branch, self.depotPath, self.initialParent)
                         self.initialParent = ""
                 except IOError:
@@ -1038,9 +921,10 @@ class P4Sync(Command):
 
 
         self.gitStream.close()
+        if importProcess.wait() != 0:
+            die("fast-import failed: %s" % self.gitError.read())
         self.gitOutput.close()
         self.gitError.close()
-        importProcess.wait()
 
         return True
 
@@ -1107,8 +991,11 @@ class P4Clone(P4Sync):
         if not P4Sync.run(self, [depotPath]):
             return False
         if self.branch != "master":
-            system("git branch master p4")
-            system("git checkout -f")
+            if gitBranchExists("refs/remotes/p4/master"):
+                system("git branch master refs/remotes/p4/master")
+                system("git checkout -f")
+            else:
+                print "Could not detect main branch. No checkout/master branch created."
         return True
 
 class HelpFormatter(optparse.IndentedHelpFormatter):