git-p4: recognize all p4 filetypes
[gitweb.git] / contrib / fast-import / git-p4
index 388151503423aeff0a99ee943f7eebf0c8fec476..0490ca563b377385cafa5e17e52e271894ca5743 100755 (executable)
@@ -118,13 +118,41 @@ def p4_system(cmd):
     real_cmd = p4_build_cmd(cmd)
     return system(real_cmd)
 
-def isP4Exec(kind):
-    """Determine if a Perforce 'kind' should have execute permission
+#
+# Canonicalize the p4 type and return a tuple of the
+# base type, plus any modifiers.  See "p4 help filetypes"
+# for a list and explanation.
+#
+def split_p4_type(p4type):
+
+    p4_filetypes_historical = {
+        "ctempobj": "binary+Sw",
+        "ctext": "text+C",
+        "cxtext": "text+Cx",
+        "ktext": "text+k",
+        "kxtext": "text+kx",
+        "ltext": "text+F",
+        "tempobj": "binary+FSw",
+        "ubinary": "binary+F",
+        "uresource": "resource+F",
+        "uxbinary": "binary+Fx",
+        "xbinary": "binary+x",
+        "xltext": "text+Fx",
+        "xtempobj": "binary+Swx",
+        "xtext": "text+x",
+        "xunicode": "unicode+x",
+        "xutf16": "utf16+x",
+    }
+    if p4type in p4_filetypes_historical:
+        p4type = p4_filetypes_historical[p4type]
+    mods = ""
+    s = p4type.split("+")
+    base = s[0]
+    mods = ""
+    if len(s) > 1:
+        mods = s[1]
+    return (base, mods)
 
-    'p4 help filetypes' gives a list of the types.  If it starts with 'x',
-    or x follows one of a few letters.  Otherwise, if there is an 'x' after
-    a plus sign, it is also executable"""
-    return (re.search(r"(^[cku]?x)|\+.*x", kind) != None)
 
 def setP4ExecBit(file, mode):
     # Reopens an already open file and changes the execute bit to match
@@ -222,10 +250,10 @@ def p4CmdList(cmd, stdin=None, stdin_mode='w+b', cb=None):
     try:
         while True:
             entry = marshal.load(p4.stdout)
-           if cb is not None:
-               cb(entry)
-           else:
-               result.append(entry)
+            if cb is not None:
+                cb(entry)
+            else:
+                result.append(entry)
     except EOFError:
         pass
     exitCode = p4.wait()
@@ -342,6 +370,11 @@ def gitConfig(key, args = None): # set args to "--bool", for instance
         _gitConfig[key] = read_pipe(cmd, ignore_error=True).strip()
     return _gitConfig[key]
 
+def gitConfigList(key):
+    if not _gitConfig.has_key(key):
+        _gitConfig[key] = read_pipe("git config --get-all %s" % key, ignore_error=True).strip().split(os.linesep)
+    return _gitConfig[key]
+
 def p4BranchesInGit(branchesAreInRemotes = True):
     branches = {}
 
@@ -449,8 +482,8 @@ def p4ChangesForPaths(depotPaths, changeRange):
 
     changes = {}
     for line in output:
-       changeNum = int(line.split(" ")[1])
-       changes[changeNum] = True
+        changeNum = int(line.split(" ")[1])
+        changes[changeNum] = True
 
     changelist = changes.keys()
     changelist.sort()
@@ -474,6 +507,47 @@ class Command:
         self.usage = "usage: %prog [options]"
         self.needsGit = True
 
+class P4UserMap:
+    def __init__(self):
+        self.userMapFromPerforceServer = False
+
+    def getUserCacheFilename(self):
+        home = os.environ.get("HOME", os.environ.get("USERPROFILE"))
+        return home + "/.gitp4-usercache.txt"
+
+    def getUserMapFromPerforceServer(self):
+        if self.userMapFromPerforceServer:
+            return
+        self.users = {}
+        self.emails = {}
+
+        for output in p4CmdList("users"):
+            if not output.has_key("User"):
+                continue
+            self.users[output["User"]] = output["FullName"] + " <" + output["Email"] + ">"
+            self.emails[output["Email"]] = output["User"]
+
+
+        s = ''
+        for (key, val) in self.users.items():
+            s += "%s\t%s\n" % (key.expandtabs(1), val.expandtabs(1))
+
+        open(self.getUserCacheFilename(), "wb").write(s)
+        self.userMapFromPerforceServer = True
+
+    def loadUserMapFromCache(self):
+        self.users = {}
+        self.userMapFromPerforceServer = False
+        try:
+            cache = open(self.getUserCacheFilename(), "rb")
+            lines = cache.readlines()
+            cache.close()
+            for line in lines:
+                entry = line.strip().split("\t")
+                self.users[entry[0]] = entry[1]
+        except IOError:
+            self.getUserMapFromPerforceServer()
+
 class P4Debug(Command):
     def __init__(self):
         Command.__init__(self)
@@ -554,13 +628,16 @@ class P4RollBack(Command):
 
         return True
 
-class P4Submit(Command):
+class P4Submit(Command, P4UserMap):
     def __init__(self):
         Command.__init__(self)
+        P4UserMap.__init__(self)
         self.options = [
                 optparse.make_option("--verbose", dest="verbose", action="store_true"),
                 optparse.make_option("--origin", dest="origin"),
                 optparse.make_option("-M", dest="detectRenames", action="store_true"),
+                # preserve the user, requires relevant p4 permissions
+                optparse.make_option("--preserve-user", dest="preserveUser", action="store_true"),
         ]
         self.description = "Submit changes from git to the perforce depot."
         self.usage += " [name of git branch to submit into perforce depot]"
@@ -568,7 +645,9 @@ class P4Submit(Command):
         self.origin = ""
         self.detectRenames = False
         self.verbose = False
+        self.preserveUser = gitConfig("git-p4.preserveUser").lower() == "true"
         self.isWindows = (platform.system() == "Windows")
+        self.myP4UserId = None
 
     def check(self):
         if len(p4CmdList("opened ...")) > 0:
@@ -602,6 +681,99 @@ class P4Submit(Command):
 
         return result
 
+    def p4UserForCommit(self,id):
+        # Return the tuple (perforce user,git email) for a given git commit id
+        self.getUserMapFromPerforceServer()
+        gitEmail = read_pipe("git log --max-count=1 --format='%%ae' %s" % id)
+        gitEmail = gitEmail.strip()
+        if not self.emails.has_key(gitEmail):
+            return (None,gitEmail)
+        else:
+            return (self.emails[gitEmail],gitEmail)
+
+    def checkValidP4Users(self,commits):
+        # check if any git authors cannot be mapped to p4 users
+        for id in commits:
+            (user,email) = self.p4UserForCommit(id)
+            if not user:
+                msg = "Cannot find p4 user for email %s in commit %s." % (email, id)
+                if gitConfig('git-p4.allowMissingP4Users').lower() == "true":
+                    print "%s" % msg
+                else:
+                    die("Error: %s\nSet git-p4.allowMissingP4Users to true to allow this." % msg)
+
+    def lastP4Changelist(self):
+        # Get back the last changelist number submitted in this client spec. This
+        # then gets used to patch up the username in the change. If the same
+        # client spec is being used by multiple processes then this might go
+        # wrong.
+        results = p4CmdList("client -o")        # find the current client
+        client = None
+        for r in results:
+            if r.has_key('Client'):
+                client = r['Client']
+                break
+        if not client:
+            die("could not get client spec")
+        results = p4CmdList("changes -c %s -m 1" % client)
+        for r in results:
+            if r.has_key('change'):
+                return r['change']
+        die("Could not get changelist number for last submit - cannot patch up user details")
+
+    def modifyChangelistUser(self, changelist, newUser):
+        # fixup the user field of a changelist after it has been submitted.
+        changes = p4CmdList("change -o %s" % changelist)
+        if len(changes) != 1:
+            die("Bad output from p4 change modifying %s to user %s" %
+                (changelist, newUser))
+
+        c = changes[0]
+        if c['User'] == newUser: return   # nothing to do
+        c['User'] = newUser
+        input = marshal.dumps(c)
+
+        result = p4CmdList("change -f -i", stdin=input)
+        for r in result:
+            if r.has_key('code'):
+                if r['code'] == 'error':
+                    die("Could not modify user field of changelist %s to %s:%s" % (changelist, newUser, r['data']))
+            if r.has_key('data'):
+                print("Updated user field for changelist %s to %s" % (changelist, newUser))
+                return
+        die("Could not modify user field of changelist %s to %s" % (changelist, newUser))
+
+    def canChangeChangelists(self):
+        # check to see if we have p4 admin or super-user permissions, either of
+        # which are required to modify changelists.
+        results = p4CmdList("protects %s" % self.depotPath)
+        for r in results:
+            if r.has_key('perm'):
+                if r['perm'] == 'admin':
+                    return 1
+                if r['perm'] == 'super':
+                    return 1
+        return 0
+
+    def p4UserId(self):
+        if self.myP4UserId:
+            return self.myP4UserId
+
+        results = p4CmdList("user -o")
+        for r in results:
+            if r.has_key('User'):
+                self.myP4UserId = r['User']
+                return r['User']
+        die("Could not find your p4 user id")
+
+    def p4UserIsMe(self, p4User):
+        # return True if the given p4 user is actually me
+        me = self.p4UserId()
+        if not p4User or p4User != me:
+            return False
+        else:
+            return True
+
     def prepareSubmitTemplate(self):
         # remove lines in the Files section that show changes to files outside the depot path we're committing into
         template = ""
@@ -631,19 +803,26 @@ class P4Submit(Command):
     def applyCommit(self, id):
         print "Applying %s" % (read_pipe("git log --max-count=1 --pretty=oneline %s" % id))
 
+        (p4User, gitEmail) = self.p4UserForCommit(id)
+
         if not self.detectRenames:
             # If not explicitly set check the config variable
-            self.detectRenames = gitConfig("git-p4.detectRenames").lower() == "true"
+            self.detectRenames = gitConfig("git-p4.detectRenames")
 
-        if self.detectRenames:
+        if self.detectRenames.lower() == "false" or self.detectRenames == "":
+            diffOpts = ""
+        elif self.detectRenames.lower() == "true":
             diffOpts = "-M"
         else:
-            diffOpts = ""
+            diffOpts = "-M%s" % self.detectRenames
 
-        if gitConfig("git-p4.detectCopies").lower() == "true":
+        detectCopies = gitConfig("git-p4.detectCopies")
+        if detectCopies.lower() == "true":
             diffOpts += " -C"
+        elif detectCopies != "" and detectCopies.lower() != "false":
+            diffOpts += " -C%s" % detectCopies
 
-        if gitConfig("git-p4.detectCopiesHarder").lower() == "true":
+        if gitConfig("git-p4.detectCopiesHarder", "--bool") == "true":
             diffOpts += " --find-copies-harder"
 
         diff = read_pipe_lines("git diff-tree -r %s \"%s^\" \"%s\"" % (diffOpts, id, id))
@@ -748,6 +927,10 @@ class P4Submit(Command):
 
         if self.interactive:
             submitTemplate = self.prepareLogMessage(template, logMessage)
+
+            if self.preserveUser:
+               submitTemplate = submitTemplate + ("\n######## Actual user %s, modified after commit\n" % p4User)
+
             if os.environ.has_key("P4DIFF"):
                 del(os.environ["P4DIFF"])
             diff = ""
@@ -764,6 +947,11 @@ class P4Submit(Command):
                     newdiff += "+" + line
                 f.close()
 
+            if self.checkAuthorship and not self.p4UserIsMe(p4User):
+                submitTemplate += "######## git author %s does not match your p4 account.\n" % gitEmail
+                submitTemplate += "######## Use git-p4 option --preserve-user to modify authorship\n"
+                submitTemplate += "######## Use git-p4 config git-p4.skipUserNameCheck hides this message.\n"
+
             separatorLine = "######## everything below this line is just the diff #######\n"
 
             [handle, fileName] = tempfile.mkstemp()
@@ -781,8 +969,13 @@ class P4Submit(Command):
                 editor = read_pipe("git var GIT_EDITOR").strip()
             system(editor + " " + fileName)
 
+            if gitConfig("git-p4.skipSubmitEditCheck") == "true":
+                checkModTime = False
+            else:
+                checkModTime = True
+
             response = "y"
-            if os.stat(fileName).st_mtime <= mtime:
+            if checkModTime and (os.stat(fileName).st_mtime <= mtime):
                 response = "x"
                 while response != "y" and response != "n":
                     response = raw_input("Submit template unchanged. Submit anyway? [y]es, [n]o (skip this patch) ")
@@ -795,6 +988,14 @@ class P4Submit(Command):
                 if self.isWindows:
                     submitTemplate = submitTemplate.replace("\r\n", "\n")
                 p4_write_pipe("submit -i", submitTemplate)
+
+                if self.preserveUser:
+                    if p4User:
+                        # Get last changelist number. Cannot easily get it from
+                        # the submit command output as the output is unmarshalled.
+                        changelist = self.lastP4Changelist()
+                        self.modifyChangelistUser(changelist, p4User)
+
             else:
                 for f in editedFiles:
                     p4_system("revert \"%s\"" % f);
@@ -831,6 +1032,10 @@ class P4Submit(Command):
         if len(self.origin) == 0:
             self.origin = upstream
 
+        if self.preserveUser:
+            if not self.canChangeChangelists():
+                die("Cannot preserve user names without p4 super-user or admin permissions")
+
         if self.verbose:
             print "Origin branch is " + self.origin
 
@@ -858,6 +1063,14 @@ class P4Submit(Command):
             commits.append(line.strip())
         commits.reverse()
 
+        if self.preserveUser or (gitConfig("git-p4.skipUserNameCheck") == "true"):
+            self.checkAuthorship = False
+        else:
+            self.checkAuthorship = True
+
+        if self.preserveUser:
+            self.checkValidP4Users(commits)
+
         while len(commits) > 0:
             commit = commits[0]
             commits = commits[1:]
@@ -877,11 +1090,12 @@ class P4Submit(Command):
 
         return True
 
-class P4Sync(Command):
+class P4Sync(Command, P4UserMap):
     delete_actions = ( "delete", "move/delete", "purge" )
 
     def __init__(self):
         Command.__init__(self)
+        P4UserMap.__init__(self)
         self.options = [
                 optparse.make_option("--branch", dest="branch"),
                 optparse.make_option("--detect-branches", dest="detectBranches", action="store_true"),
@@ -1033,38 +1247,54 @@ class P4Sync(Command):
     # - helper for streamP4Files
 
     def streamOneP4File(self, file, contents):
-       if file["type"] == "apple":
-           print "\nfile %s is a strange apple file that forks. Ignoring" % \
-               file['depotFile']
-           return
+        if file["type"] == "apple":
+            print "\nfile %s is a strange apple file that forks. Ignoring" % \
+                file['depotFile']
+            return
 
         relPath = self.stripRepoPath(file['depotFile'], self.branchPrefixes)
         relPath = self.wildcard_decode(relPath)
         if verbose:
             sys.stderr.write("%s\n" % relPath)
 
-        mode = "644"
-        if isP4Exec(file["type"]):
-            mode = "755"
-        elif file["type"] == "symlink":
-            mode = "120000"
-            # p4 print on a symlink contains "target\n", so strip it off
+        (type_base, type_mods) = split_p4_type(file["type"])
+
+        git_mode = "100644"
+        if "x" in type_mods:
+            git_mode = "100755"
+        if type_base == "symlink":
+            git_mode = "120000"
+            # p4 print on a symlink contains "target\n"; remove the newline
             data = ''.join(contents)
             contents = [data[:-1]]
 
-        if self.isWindows and file["type"].endswith("text"):
+        if type_base == "utf16":
+            # p4 delivers different text in the python output to -G
+            # than it does when using "print -o", or normal p4 client
+            # operations.  utf16 is converted to ascii or utf8, perhaps.
+            # But ascii text saved as -t utf16 is completely mangled.
+            # Invoke print -o to get the real contents.
+            text = p4_read_pipe('print -q -o - "%s"' % file['depotFile'])
+            contents = [ text ]
+
+        # Perhaps windows wants unicode, utf16 newlines translated too;
+        # but this is not doing it.
+        if self.isWindows and type_base == "text":
             mangled = []
             for data in contents:
                 data = data.replace("\r\n", "\n")
                 mangled.append(data)
             contents = mangled
 
-        if file['type'] in ('text+ko', 'unicode+ko', 'binary+ko'):
-            contents = map(lambda text: re.sub(r'(?i)\$(Id|Header):[^$]*\$',r'$\1$', text), contents)
-        elif file['type'] in ('text+k', 'ktext', 'kxtext', 'unicode+k', 'binary+k'):
-            contents = map(lambda text: re.sub(r'\$(Id|Header|Author|Date|DateTime|Change|File|Revision):[^$\n]*\$',r'$\1$', text), contents)
+        # Note that we do not try to de-mangle keywords on utf16 files,
+        # even though in theory somebody may want that.
+        if type_base in ("text", "unicode", "binary"):
+            if "ko" in type_mods:
+                contents = map(lambda text: re.sub(r'(?i)\$(Id|Header):[^$]*\$', r'$\1$', text), contents)
+            elif "k" in type_mods:
+                contents = map(lambda text: re.sub(r'\$(Id|Header|Author|Date|DateTime|Change|File|Revision):[^$\n]*\$', r'$\1$', text), contents)
 
-        self.gitStream.write("M %s inline %s\n" % (mode, relPath))
+        self.gitStream.write("M %s inline %s\n" % (git_mode, relPath))
 
         # total length...
         length = 0
@@ -1085,22 +1315,22 @@ class P4Sync(Command):
     # handle another chunk of streaming data
     def streamP4FilesCb(self, marshalled):
 
-       if marshalled.has_key('depotFile') and self.stream_have_file_info:
-           # start of a new file - output the old one first
-           self.streamOneP4File(self.stream_file, self.stream_contents)
-           self.stream_file = {}
-           self.stream_contents = []
-           self.stream_have_file_info = False
+        if marshalled.has_key('depotFile') and self.stream_have_file_info:
+            # start of a new file - output the old one first
+            self.streamOneP4File(self.stream_file, self.stream_contents)
+            self.stream_file = {}
+            self.stream_contents = []
+            self.stream_have_file_info = False
 
-       # pick up the new file information... for the
-       # 'data' field we need to append to our array
-       for k in marshalled.keys():
-           if k == 'data':
-               self.stream_contents.append(marshalled['data'])
-           else:
-               self.stream_file[k] = marshalled[k]
+        # pick up the new file information... for the
+        # 'data' field we need to append to our array
+        for k in marshalled.keys():
+            if k == 'data':
+                self.stream_contents.append(marshalled['data'])
+            else:
+                self.stream_file[k] = marshalled[k]
 
-       self.stream_have_file_info = True
+        self.stream_have_file_info = True
 
     # Stream directly from "p4 files" into "git fast-import"
     def streamP4Files(self, files):
@@ -1132,14 +1362,14 @@ class P4Sync(Command):
             self.stream_contents = []
             self.stream_have_file_info = False
 
-           # curry self argument
-           def streamP4FilesCbSelf(entry):
-               self.streamP4FilesCb(entry)
+            # curry self argument
+            def streamP4FilesCbSelf(entry):
+                self.streamP4FilesCb(entry)
 
-           p4CmdList("-x - print",
-               '\n'.join(['%s#%s' % (f['path'], f['rev'])
+            p4CmdList("-x - print",
+                '\n'.join(['%s#%s' % (f['path'], f['rev'])
                                                   for f in filesToRead]),
-               cb=streamP4FilesCbSelf)
+                cb=streamP4FilesCbSelf)
 
             # do the last chunk
             if self.stream_file.has_key('depotFile'):
@@ -1148,7 +1378,7 @@ class P4Sync(Command):
     def commit(self, details, files, branch, branchPrefixes, parent = ""):
         epoch = details["time"]
         author = details["user"]
-       self.branchPrefixes = branchPrefixes
+        self.branchPrefixes = branchPrefixes
 
         if self.verbose:
             print "commit into %s" % branch
@@ -1236,41 +1466,6 @@ class P4Sync(Command):
                     print ("Tag %s does not match with change %s: file count is different."
                            % (labelDetails["label"], change))
 
-    def getUserCacheFilename(self):
-        home = os.environ.get("HOME", os.environ.get("USERPROFILE"))
-        return home + "/.gitp4-usercache.txt"
-
-    def getUserMapFromPerforceServer(self):
-        if self.userMapFromPerforceServer:
-            return
-        self.users = {}
-
-        for output in p4CmdList("users"):
-            if not output.has_key("User"):
-                continue
-            self.users[output["User"]] = output["FullName"] + " <" + output["Email"] + ">"
-
-
-        s = ''
-        for (key, val) in self.users.items():
-           s += "%s\t%s\n" % (key.expandtabs(1), val.expandtabs(1))
-
-        open(self.getUserCacheFilename(), "wb").write(s)
-        self.userMapFromPerforceServer = True
-
-    def loadUserMapFromCache(self):
-        self.users = {}
-        self.userMapFromPerforceServer = False
-        try:
-            cache = open(self.getUserCacheFilename(), "rb")
-            lines = cache.readlines()
-            cache.close()
-            for line in lines:
-                entry = line.strip().split("\t")
-                self.users[entry[0]] = entry[1]
-        except IOError:
-            self.getUserMapFromPerforceServer()
-
     def getLabels(self):
         self.labels = {}
 
@@ -1309,7 +1504,13 @@ class P4Sync(Command):
     def getBranchMapping(self):
         lostAndFoundBranches = set()
 
-        for info in p4CmdList("branches"):
+        user = gitConfig("git-p4.branchUser")
+        if len(user) > 0:
+            command = "branches -u %s" % user
+        else:
+            command = "branches"
+
+        for info in p4CmdList(command):
             details = p4Cmd("branch -o %s" % info["branch"])
             viewIdx = 0
             while details.has_key("View%s" % viewIdx):
@@ -1338,6 +1539,25 @@ class P4Sync(Command):
                     if source not in self.knownBranches:
                         lostAndFoundBranches.add(source)
 
+        # Perforce does not strictly require branches to be defined, so we also
+        # check git config for a branch list.
+        #
+        # Example of branch definition in git config file:
+        # [git-p4]
+        #   branchList=main:branchA
+        #   branchList=main:branchB
+        #   branchList=branchA:branchC
+        configBranches = gitConfigList("git-p4.branchList")
+        for branch in configBranches:
+            if branch:
+                (source, destination) = branch.split(":")
+                self.knownBranches[destination] = source
+
+                lostAndFoundBranches.discard(destination)
+
+                if source not in self.knownBranches:
+                    lostAndFoundBranches.add(source)
+
 
         for branch in lostAndFoundBranches:
             self.knownBranches[branch] = branch
@@ -1508,7 +1728,8 @@ class P4Sync(Command):
     def importHeadRevision(self, revision):
         print "Doing initial import of %s from revision %s into %s" % (' '.join(self.depotPaths), revision, self.branch)
 
-        details = { "user" : "git perforce import user", "time" : int(time.time()) }
+        details = {}
+        details["user"] = "git perforce import user"
         details["desc"] = ("Initial import of %s from the state at revision %s\n"
                            % (' '.join(self.depotPaths), revision))
         details["change"] = revision
@@ -1548,6 +1769,18 @@ class P4Sync(Command):
             fileCnt = fileCnt + 1
 
         details["change"] = newestRevision
+
+        # Use time from top-most change so that all git-p4 clones of
+        # the same p4 repo have the same commit SHA1s.
+        res = p4CmdList("describe -s %d" % newestRevision)
+        newestTime = None
+        for r in res:
+            if r.has_key('time'):
+                newestTime = int(r['time'])
+        if newestTime is None:
+            die("\"describe -s\" on newest change %d did not give a time")
+        details["time"] = newestTime
+
         self.updateOptionDict(details)
         try:
             self.commit(details, self.extractFilesFromCommit(details), self.branch, self.depotPaths)
@@ -1670,12 +1903,14 @@ class P4Sync(Command):
                     else:
                         paths = []
                         for (prev, cur) in zip(self.previousDepotPaths, depotPaths):
-                            for i in range(0, min(len(cur), len(prev))):
-                                if cur[i] <> prev[i]:
+                            prev_list = prev.split("/")
+                            cur_list = cur.split("/")
+                            for i in range(0, min(len(cur_list), len(prev_list))):
+                                if cur_list[i] <> prev_list[i]:
                                     i = i - 1
                                     break
 
-                            paths.append (cur[:i + 1])
+                            paths.append ("/".join(cur_list[:i + 1]))
 
                         self.previousDepotPaths = paths