git-p4: clean up after p4 submit failure
[git] / git-p4.py
1 #!/usr/bin/env python
2 #
3 # git-p4.py -- A tool for bidirectional operation between a Perforce depot and git.
4 #
5 # Author: Simon Hausmann <simon@lst.de>
6 # Copyright: 2007 Simon Hausmann <simon@lst.de>
7 #            2007 Trolltech ASA
8 # License: MIT <http://www.opensource.org/licenses/mit-license.php>
9 #
10 import sys
11 if sys.hexversion < 0x02040000:
12     # The limiter is the subprocess module
13     sys.stderr.write("git-p4: requires Python 2.4 or later.\n")
14     sys.exit(1)
15 import os
16 import optparse
17 import marshal
18 import subprocess
19 import tempfile
20 import time
21 import platform
22 import re
23 import shutil
24 import stat
25
26 try:
27     from subprocess import CalledProcessError
28 except ImportError:
29     # from python2.7:subprocess.py
30     # Exception classes used by this module.
31     class CalledProcessError(Exception):
32         """This exception is raised when a process run by check_call() returns
33         a non-zero exit status.  The exit status will be stored in the
34         returncode attribute."""
35         def __init__(self, returncode, cmd):
36             self.returncode = returncode
37             self.cmd = cmd
38         def __str__(self):
39             return "Command '%s' returned non-zero exit status %d" % (self.cmd, self.returncode)
40
41 verbose = False
42
43 # Only labels/tags matching this will be imported/exported
44 defaultLabelRegexp = r'[a-zA-Z0-9_\-.]+$'
45
46 # Grab changes in blocks of this many revisions, unless otherwise requested
47 defaultBlockSize = 512
48
49 def p4_build_cmd(cmd):
50     """Build a suitable p4 command line.
51
52     This consolidates building and returning a p4 command line into one
53     location. It means that hooking into the environment, or other configuration
54     can be done more easily.
55     """
56     real_cmd = ["p4"]
57
58     user = gitConfig("git-p4.user")
59     if len(user) > 0:
60         real_cmd += ["-u",user]
61
62     password = gitConfig("git-p4.password")
63     if len(password) > 0:
64         real_cmd += ["-P", password]
65
66     port = gitConfig("git-p4.port")
67     if len(port) > 0:
68         real_cmd += ["-p", port]
69
70     host = gitConfig("git-p4.host")
71     if len(host) > 0:
72         real_cmd += ["-H", host]
73
74     client = gitConfig("git-p4.client")
75     if len(client) > 0:
76         real_cmd += ["-c", client]
77
78
79     if isinstance(cmd,basestring):
80         real_cmd = ' '.join(real_cmd) + ' ' + cmd
81     else:
82         real_cmd += cmd
83     return real_cmd
84
85 def chdir(path, is_client_path=False):
86     """Do chdir to the given path, and set the PWD environment
87        variable for use by P4.  It does not look at getcwd() output.
88        Since we're not using the shell, it is necessary to set the
89        PWD environment variable explicitly.
90
91        Normally, expand the path to force it to be absolute.  This
92        addresses the use of relative path names inside P4 settings,
93        e.g. P4CONFIG=.p4config.  P4 does not simply open the filename
94        as given; it looks for .p4config using PWD.
95
96        If is_client_path, the path was handed to us directly by p4,
97        and may be a symbolic link.  Do not call os.getcwd() in this
98        case, because it will cause p4 to think that PWD is not inside
99        the client path.
100        """
101
102     os.chdir(path)
103     if not is_client_path:
104         path = os.getcwd()
105     os.environ['PWD'] = path
106
107 def die(msg):
108     if verbose:
109         raise Exception(msg)
110     else:
111         sys.stderr.write(msg + "\n")
112         sys.exit(1)
113
114 def write_pipe(c, stdin):
115     if verbose:
116         sys.stderr.write('Writing pipe: %s\n' % str(c))
117
118     expand = isinstance(c,basestring)
119     p = subprocess.Popen(c, stdin=subprocess.PIPE, shell=expand)
120     pipe = p.stdin
121     val = pipe.write(stdin)
122     pipe.close()
123     if p.wait():
124         die('Command failed: %s' % str(c))
125
126     return val
127
128 def p4_write_pipe(c, stdin):
129     real_cmd = p4_build_cmd(c)
130     return write_pipe(real_cmd, stdin)
131
132 def read_pipe(c, ignore_error=False):
133     if verbose:
134         sys.stderr.write('Reading pipe: %s\n' % str(c))
135
136     expand = isinstance(c,basestring)
137     p = subprocess.Popen(c, stdout=subprocess.PIPE, shell=expand)
138     pipe = p.stdout
139     val = pipe.read()
140     if p.wait() and not ignore_error:
141         die('Command failed: %s' % str(c))
142
143     return val
144
145 def p4_read_pipe(c, ignore_error=False):
146     real_cmd = p4_build_cmd(c)
147     return read_pipe(real_cmd, ignore_error)
148
149 def read_pipe_lines(c):
150     if verbose:
151         sys.stderr.write('Reading pipe: %s\n' % str(c))
152
153     expand = isinstance(c, basestring)
154     p = subprocess.Popen(c, stdout=subprocess.PIPE, shell=expand)
155     pipe = p.stdout
156     val = pipe.readlines()
157     if pipe.close() or p.wait():
158         die('Command failed: %s' % str(c))
159
160     return val
161
162 def p4_read_pipe_lines(c):
163     """Specifically invoke p4 on the command supplied. """
164     real_cmd = p4_build_cmd(c)
165     return read_pipe_lines(real_cmd)
166
167 def p4_has_command(cmd):
168     """Ask p4 for help on this command.  If it returns an error, the
169        command does not exist in this version of p4."""
170     real_cmd = p4_build_cmd(["help", cmd])
171     p = subprocess.Popen(real_cmd, stdout=subprocess.PIPE,
172                                    stderr=subprocess.PIPE)
173     p.communicate()
174     return p.returncode == 0
175
176 def p4_has_move_command():
177     """See if the move command exists, that it supports -k, and that
178        it has not been administratively disabled.  The arguments
179        must be correct, but the filenames do not have to exist.  Use
180        ones with wildcards so even if they exist, it will fail."""
181
182     if not p4_has_command("move"):
183         return False
184     cmd = p4_build_cmd(["move", "-k", "@from", "@to"])
185     p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
186     (out, err) = p.communicate()
187     # return code will be 1 in either case
188     if err.find("Invalid option") >= 0:
189         return False
190     if err.find("disabled") >= 0:
191         return False
192     # assume it failed because @... was invalid changelist
193     return True
194
195 def system(cmd):
196     expand = isinstance(cmd,basestring)
197     if verbose:
198         sys.stderr.write("executing %s\n" % str(cmd))
199     retcode = subprocess.call(cmd, shell=expand)
200     if retcode:
201         raise CalledProcessError(retcode, cmd)
202
203 def p4_system(cmd):
204     """Specifically invoke p4 as the system command. """
205     real_cmd = p4_build_cmd(cmd)
206     expand = isinstance(real_cmd, basestring)
207     retcode = subprocess.call(real_cmd, shell=expand)
208     if retcode:
209         raise CalledProcessError(retcode, real_cmd)
210
211 _p4_version_string = None
212 def p4_version_string():
213     """Read the version string, showing just the last line, which
214        hopefully is the interesting version bit.
215
216        $ p4 -V
217        Perforce - The Fast Software Configuration Management System.
218        Copyright 1995-2011 Perforce Software.  All rights reserved.
219        Rev. P4/NTX86/2011.1/393975 (2011/12/16).
220     """
221     global _p4_version_string
222     if not _p4_version_string:
223         a = p4_read_pipe_lines(["-V"])
224         _p4_version_string = a[-1].rstrip()
225     return _p4_version_string
226
227 def p4_integrate(src, dest):
228     p4_system(["integrate", "-Dt", wildcard_encode(src), wildcard_encode(dest)])
229
230 def p4_sync(f, *options):
231     p4_system(["sync"] + list(options) + [wildcard_encode(f)])
232
233 def p4_add(f):
234     # forcibly add file names with wildcards
235     if wildcard_present(f):
236         p4_system(["add", "-f", f])
237     else:
238         p4_system(["add", f])
239
240 def p4_delete(f):
241     p4_system(["delete", wildcard_encode(f)])
242
243 def p4_edit(f):
244     p4_system(["edit", wildcard_encode(f)])
245
246 def p4_revert(f):
247     p4_system(["revert", wildcard_encode(f)])
248
249 def p4_reopen(type, f):
250     p4_system(["reopen", "-t", type, wildcard_encode(f)])
251
252 def p4_move(src, dest):
253     p4_system(["move", "-k", wildcard_encode(src), wildcard_encode(dest)])
254
255 def p4_last_change():
256     results = p4CmdList(["changes", "-m", "1"])
257     return int(results[0]['change'])
258
259 def p4_describe(change):
260     """Make sure it returns a valid result by checking for
261        the presence of field "time".  Return a dict of the
262        results."""
263
264     ds = p4CmdList(["describe", "-s", str(change)])
265     if len(ds) != 1:
266         die("p4 describe -s %d did not return 1 result: %s" % (change, str(ds)))
267
268     d = ds[0]
269
270     if "p4ExitCode" in d:
271         die("p4 describe -s %d exited with %d: %s" % (change, d["p4ExitCode"],
272                                                       str(d)))
273     if "code" in d:
274         if d["code"] == "error":
275             die("p4 describe -s %d returned error code: %s" % (change, str(d)))
276
277     if "time" not in d:
278         die("p4 describe -s %d returned no \"time\": %s" % (change, str(d)))
279
280     return d
281
282 #
283 # Canonicalize the p4 type and return a tuple of the
284 # base type, plus any modifiers.  See "p4 help filetypes"
285 # for a list and explanation.
286 #
287 def split_p4_type(p4type):
288
289     p4_filetypes_historical = {
290         "ctempobj": "binary+Sw",
291         "ctext": "text+C",
292         "cxtext": "text+Cx",
293         "ktext": "text+k",
294         "kxtext": "text+kx",
295         "ltext": "text+F",
296         "tempobj": "binary+FSw",
297         "ubinary": "binary+F",
298         "uresource": "resource+F",
299         "uxbinary": "binary+Fx",
300         "xbinary": "binary+x",
301         "xltext": "text+Fx",
302         "xtempobj": "binary+Swx",
303         "xtext": "text+x",
304         "xunicode": "unicode+x",
305         "xutf16": "utf16+x",
306     }
307     if p4type in p4_filetypes_historical:
308         p4type = p4_filetypes_historical[p4type]
309     mods = ""
310     s = p4type.split("+")
311     base = s[0]
312     mods = ""
313     if len(s) > 1:
314         mods = s[1]
315     return (base, mods)
316
317 #
318 # return the raw p4 type of a file (text, text+ko, etc)
319 #
320 def p4_type(f):
321     results = p4CmdList(["fstat", "-T", "headType", wildcard_encode(f)])
322     return results[0]['headType']
323
324 #
325 # Given a type base and modifier, return a regexp matching
326 # the keywords that can be expanded in the file
327 #
328 def p4_keywords_regexp_for_type(base, type_mods):
329     if base in ("text", "unicode", "binary"):
330         kwords = None
331         if "ko" in type_mods:
332             kwords = 'Id|Header'
333         elif "k" in type_mods:
334             kwords = 'Id|Header|Author|Date|DateTime|Change|File|Revision'
335         else:
336             return None
337         pattern = r"""
338             \$              # Starts with a dollar, followed by...
339             (%s)            # one of the keywords, followed by...
340             (:[^$\n]+)?     # possibly an old expansion, followed by...
341             \$              # another dollar
342             """ % kwords
343         return pattern
344     else:
345         return None
346
347 #
348 # Given a file, return a regexp matching the possible
349 # RCS keywords that will be expanded, or None for files
350 # with kw expansion turned off.
351 #
352 def p4_keywords_regexp_for_file(file):
353     if not os.path.exists(file):
354         return None
355     else:
356         (type_base, type_mods) = split_p4_type(p4_type(file))
357         return p4_keywords_regexp_for_type(type_base, type_mods)
358
359 def setP4ExecBit(file, mode):
360     # Reopens an already open file and changes the execute bit to match
361     # the execute bit setting in the passed in mode.
362
363     p4Type = "+x"
364
365     if not isModeExec(mode):
366         p4Type = getP4OpenedType(file)
367         p4Type = re.sub('^([cku]?)x(.*)', '\\1\\2', p4Type)
368         p4Type = re.sub('(.*?\+.*?)x(.*?)', '\\1\\2', p4Type)
369         if p4Type[-1] == "+":
370             p4Type = p4Type[0:-1]
371
372     p4_reopen(p4Type, file)
373
374 def getP4OpenedType(file):
375     # Returns the perforce file type for the given file.
376
377     result = p4_read_pipe(["opened", wildcard_encode(file)])
378     match = re.match(".*\((.+)\)( \*exclusive\*)?\r?$", result)
379     if match:
380         return match.group(1)
381     else:
382         die("Could not determine file type for %s (result: '%s')" % (file, result))
383
384 # Return the set of all p4 labels
385 def getP4Labels(depotPaths):
386     labels = set()
387     if isinstance(depotPaths,basestring):
388         depotPaths = [depotPaths]
389
390     for l in p4CmdList(["labels"] + ["%s..." % p for p in depotPaths]):
391         label = l['label']
392         labels.add(label)
393
394     return labels
395
396 # Return the set of all git tags
397 def getGitTags():
398     gitTags = set()
399     for line in read_pipe_lines(["git", "tag"]):
400         tag = line.strip()
401         gitTags.add(tag)
402     return gitTags
403
404 def diffTreePattern():
405     # This is a simple generator for the diff tree regex pattern. This could be
406     # a class variable if this and parseDiffTreeEntry were a part of a class.
407     pattern = re.compile(':(\d+) (\d+) (\w+) (\w+) ([A-Z])(\d+)?\t(.*?)((\t(.*))|$)')
408     while True:
409         yield pattern
410
411 def parseDiffTreeEntry(entry):
412     """Parses a single diff tree entry into its component elements.
413
414     See git-diff-tree(1) manpage for details about the format of the diff
415     output. This method returns a dictionary with the following elements:
416
417     src_mode - The mode of the source file
418     dst_mode - The mode of the destination file
419     src_sha1 - The sha1 for the source file
420     dst_sha1 - The sha1 fr the destination file
421     status - The one letter status of the diff (i.e. 'A', 'M', 'D', etc)
422     status_score - The score for the status (applicable for 'C' and 'R'
423                    statuses). This is None if there is no score.
424     src - The path for the source file.
425     dst - The path for the destination file. This is only present for
426           copy or renames. If it is not present, this is None.
427
428     If the pattern is not matched, None is returned."""
429
430     match = diffTreePattern().next().match(entry)
431     if match:
432         return {
433             'src_mode': match.group(1),
434             'dst_mode': match.group(2),
435             'src_sha1': match.group(3),
436             'dst_sha1': match.group(4),
437             'status': match.group(5),
438             'status_score': match.group(6),
439             'src': match.group(7),
440             'dst': match.group(10)
441         }
442     return None
443
444 def isModeExec(mode):
445     # Returns True if the given git mode represents an executable file,
446     # otherwise False.
447     return mode[-3:] == "755"
448
449 def isModeExecChanged(src_mode, dst_mode):
450     return isModeExec(src_mode) != isModeExec(dst_mode)
451
452 def p4CmdList(cmd, stdin=None, stdin_mode='w+b', cb=None):
453
454     if isinstance(cmd,basestring):
455         cmd = "-G " + cmd
456         expand = True
457     else:
458         cmd = ["-G"] + cmd
459         expand = False
460
461     cmd = p4_build_cmd(cmd)
462     if verbose:
463         sys.stderr.write("Opening pipe: %s\n" % str(cmd))
464
465     # Use a temporary file to avoid deadlocks without
466     # subprocess.communicate(), which would put another copy
467     # of stdout into memory.
468     stdin_file = None
469     if stdin is not None:
470         stdin_file = tempfile.TemporaryFile(prefix='p4-stdin', mode=stdin_mode)
471         if isinstance(stdin,basestring):
472             stdin_file.write(stdin)
473         else:
474             for i in stdin:
475                 stdin_file.write(i + '\n')
476         stdin_file.flush()
477         stdin_file.seek(0)
478
479     p4 = subprocess.Popen(cmd,
480                           shell=expand,
481                           stdin=stdin_file,
482                           stdout=subprocess.PIPE)
483
484     result = []
485     try:
486         while True:
487             entry = marshal.load(p4.stdout)
488             if cb is not None:
489                 cb(entry)
490             else:
491                 result.append(entry)
492     except EOFError:
493         pass
494     exitCode = p4.wait()
495     if exitCode != 0:
496         entry = {}
497         entry["p4ExitCode"] = exitCode
498         result.append(entry)
499
500     return result
501
502 def p4Cmd(cmd):
503     list = p4CmdList(cmd)
504     result = {}
505     for entry in list:
506         result.update(entry)
507     return result;
508
509 def p4Where(depotPath):
510     if not depotPath.endswith("/"):
511         depotPath += "/"
512     depotPathLong = depotPath + "..."
513     outputList = p4CmdList(["where", depotPathLong])
514     output = None
515     for entry in outputList:
516         if "depotFile" in entry:
517             # Search for the base client side depot path, as long as it starts with the branch's P4 path.
518             # The base path always ends with "/...".
519             if entry["depotFile"].find(depotPath) == 0 and entry["depotFile"][-4:] == "/...":
520                 output = entry
521                 break
522         elif "data" in entry:
523             data = entry.get("data")
524             space = data.find(" ")
525             if data[:space] == depotPath:
526                 output = entry
527                 break
528     if output == None:
529         return ""
530     if output["code"] == "error":
531         return ""
532     clientPath = ""
533     if "path" in output:
534         clientPath = output.get("path")
535     elif "data" in output:
536         data = output.get("data")
537         lastSpace = data.rfind(" ")
538         clientPath = data[lastSpace + 1:]
539
540     if clientPath.endswith("..."):
541         clientPath = clientPath[:-3]
542     return clientPath
543
544 def currentGitBranch():
545     return read_pipe("git name-rev HEAD").split(" ")[1].strip()
546
547 def isValidGitDir(path):
548     if (os.path.exists(path + "/HEAD")
549         and os.path.exists(path + "/refs") and os.path.exists(path + "/objects")):
550         return True;
551     return False
552
553 def parseRevision(ref):
554     return read_pipe("git rev-parse %s" % ref).strip()
555
556 def branchExists(ref):
557     rev = read_pipe(["git", "rev-parse", "-q", "--verify", ref],
558                      ignore_error=True)
559     return len(rev) > 0
560
561 def extractLogMessageFromGitCommit(commit):
562     logMessage = ""
563
564     ## fixme: title is first line of commit, not 1st paragraph.
565     foundTitle = False
566     for log in read_pipe_lines("git cat-file commit %s" % commit):
567        if not foundTitle:
568            if len(log) == 1:
569                foundTitle = True
570            continue
571
572        logMessage += log
573     return logMessage
574
575 def extractSettingsGitLog(log):
576     values = {}
577     for line in log.split("\n"):
578         line = line.strip()
579         m = re.search (r"^ *\[git-p4: (.*)\]$", line)
580         if not m:
581             continue
582
583         assignments = m.group(1).split (':')
584         for a in assignments:
585             vals = a.split ('=')
586             key = vals[0].strip()
587             val = ('='.join (vals[1:])).strip()
588             if val.endswith ('\"') and val.startswith('"'):
589                 val = val[1:-1]
590
591             values[key] = val
592
593     paths = values.get("depot-paths")
594     if not paths:
595         paths = values.get("depot-path")
596     if paths:
597         values['depot-paths'] = paths.split(',')
598     return values
599
600 def gitBranchExists(branch):
601     proc = subprocess.Popen(["git", "rev-parse", branch],
602                             stderr=subprocess.PIPE, stdout=subprocess.PIPE);
603     return proc.wait() == 0;
604
605 _gitConfig = {}
606
607 def gitConfig(key):
608     if not _gitConfig.has_key(key):
609         cmd = [ "git", "config", key ]
610         s = read_pipe(cmd, ignore_error=True)
611         _gitConfig[key] = s.strip()
612     return _gitConfig[key]
613
614 def gitConfigBool(key):
615     """Return a bool, using git config --bool.  It is True only if the
616        variable is set to true, and False if set to false or not present
617        in the config."""
618
619     if not _gitConfig.has_key(key):
620         cmd = [ "git", "config", "--bool", key ]
621         s = read_pipe(cmd, ignore_error=True)
622         v = s.strip()
623         _gitConfig[key] = v == "true"
624     return _gitConfig[key]
625
626 def gitConfigList(key):
627     if not _gitConfig.has_key(key):
628         s = read_pipe(["git", "config", "--get-all", key], ignore_error=True)
629         _gitConfig[key] = s.strip().split(os.linesep)
630     return _gitConfig[key]
631
632 def p4BranchesInGit(branchesAreInRemotes=True):
633     """Find all the branches whose names start with "p4/", looking
634        in remotes or heads as specified by the argument.  Return
635        a dictionary of { branch: revision } for each one found.
636        The branch names are the short names, without any
637        "p4/" prefix."""
638
639     branches = {}
640
641     cmdline = "git rev-parse --symbolic "
642     if branchesAreInRemotes:
643         cmdline += "--remotes"
644     else:
645         cmdline += "--branches"
646
647     for line in read_pipe_lines(cmdline):
648         line = line.strip()
649
650         # only import to p4/
651         if not line.startswith('p4/'):
652             continue
653         # special symbolic ref to p4/master
654         if line == "p4/HEAD":
655             continue
656
657         # strip off p4/ prefix
658         branch = line[len("p4/"):]
659
660         branches[branch] = parseRevision(line)
661
662     return branches
663
664 def branch_exists(branch):
665     """Make sure that the given ref name really exists."""
666
667     cmd = [ "git", "rev-parse", "--symbolic", "--verify", branch ]
668     p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
669     out, _ = p.communicate()
670     if p.returncode:
671         return False
672     # expect exactly one line of output: the branch name
673     return out.rstrip() == branch
674
675 def findUpstreamBranchPoint(head = "HEAD"):
676     branches = p4BranchesInGit()
677     # map from depot-path to branch name
678     branchByDepotPath = {}
679     for branch in branches.keys():
680         tip = branches[branch]
681         log = extractLogMessageFromGitCommit(tip)
682         settings = extractSettingsGitLog(log)
683         if settings.has_key("depot-paths"):
684             paths = ",".join(settings["depot-paths"])
685             branchByDepotPath[paths] = "remotes/p4/" + branch
686
687     settings = None
688     parent = 0
689     while parent < 65535:
690         commit = head + "~%s" % parent
691         log = extractLogMessageFromGitCommit(commit)
692         settings = extractSettingsGitLog(log)
693         if settings.has_key("depot-paths"):
694             paths = ",".join(settings["depot-paths"])
695             if branchByDepotPath.has_key(paths):
696                 return [branchByDepotPath[paths], settings]
697
698         parent = parent + 1
699
700     return ["", settings]
701
702 def createOrUpdateBranchesFromOrigin(localRefPrefix = "refs/remotes/p4/", silent=True):
703     if not silent:
704         print ("Creating/updating branch(es) in %s based on origin branch(es)"
705                % localRefPrefix)
706
707     originPrefix = "origin/p4/"
708
709     for line in read_pipe_lines("git rev-parse --symbolic --remotes"):
710         line = line.strip()
711         if (not line.startswith(originPrefix)) or line.endswith("HEAD"):
712             continue
713
714         headName = line[len(originPrefix):]
715         remoteHead = localRefPrefix + headName
716         originHead = line
717
718         original = extractSettingsGitLog(extractLogMessageFromGitCommit(originHead))
719         if (not original.has_key('depot-paths')
720             or not original.has_key('change')):
721             continue
722
723         update = False
724         if not gitBranchExists(remoteHead):
725             if verbose:
726                 print "creating %s" % remoteHead
727             update = True
728         else:
729             settings = extractSettingsGitLog(extractLogMessageFromGitCommit(remoteHead))
730             if settings.has_key('change') > 0:
731                 if settings['depot-paths'] == original['depot-paths']:
732                     originP4Change = int(original['change'])
733                     p4Change = int(settings['change'])
734                     if originP4Change > p4Change:
735                         print ("%s (%s) is newer than %s (%s). "
736                                "Updating p4 branch from origin."
737                                % (originHead, originP4Change,
738                                   remoteHead, p4Change))
739                         update = True
740                 else:
741                     print ("Ignoring: %s was imported from %s while "
742                            "%s was imported from %s"
743                            % (originHead, ','.join(original['depot-paths']),
744                               remoteHead, ','.join(settings['depot-paths'])))
745
746         if update:
747             system("git update-ref %s %s" % (remoteHead, originHead))
748
749 def originP4BranchesExist():
750         return gitBranchExists("origin") or gitBranchExists("origin/p4") or gitBranchExists("origin/p4/master")
751
752
753 def p4ParseNumericChangeRange(parts):
754     changeStart = int(parts[0][1:])
755     if parts[1] == '#head':
756         changeEnd = p4_last_change()
757     else:
758         changeEnd = int(parts[1])
759
760     return (changeStart, changeEnd)
761
762 def chooseBlockSize(blockSize):
763     if blockSize:
764         return blockSize
765     else:
766         return defaultBlockSize
767
768 def p4ChangesForPaths(depotPaths, changeRange, requestedBlockSize):
769     assert depotPaths
770
771     # Parse the change range into start and end. Try to find integer
772     # revision ranges as these can be broken up into blocks to avoid
773     # hitting server-side limits (maxrows, maxscanresults). But if
774     # that doesn't work, fall back to using the raw revision specifier
775     # strings, without using block mode.
776
777     if changeRange is None or changeRange == '':
778         changeStart = 1
779         changeEnd = p4_last_change()
780         block_size = chooseBlockSize(requestedBlockSize)
781     else:
782         parts = changeRange.split(',')
783         assert len(parts) == 2
784         try:
785             (changeStart, changeEnd) = p4ParseNumericChangeRange(parts)
786             block_size = chooseBlockSize(requestedBlockSize)
787         except:
788             changeStart = parts[0][1:]
789             changeEnd = parts[1]
790             if requestedBlockSize:
791                 die("cannot use --changes-block-size with non-numeric revisions")
792             block_size = None
793
794     # Accumulate change numbers in a dictionary to avoid duplicates
795     changes = {}
796
797     for p in depotPaths:
798         # Retrieve changes a block at a time, to prevent running
799         # into a MaxResults/MaxScanRows error from the server.
800
801         while True:
802             cmd = ['changes']
803
804             if block_size:
805                 end = min(changeEnd, changeStart + block_size)
806                 revisionRange = "%d,%d" % (changeStart, end)
807             else:
808                 revisionRange = "%s,%s" % (changeStart, changeEnd)
809
810             cmd += ["%s...@%s" % (p, revisionRange)]
811
812             for line in p4_read_pipe_lines(cmd):
813                 changeNum = int(line.split(" ")[1])
814                 changes[changeNum] = True
815
816             if not block_size:
817                 break
818
819             if end >= changeEnd:
820                 break
821
822             changeStart = end + 1
823
824     changelist = changes.keys()
825     changelist.sort()
826     return changelist
827
828 def p4PathStartsWith(path, prefix):
829     # This method tries to remedy a potential mixed-case issue:
830     #
831     # If UserA adds  //depot/DirA/file1
832     # and UserB adds //depot/dira/file2
833     #
834     # we may or may not have a problem. If you have core.ignorecase=true,
835     # we treat DirA and dira as the same directory
836     if gitConfigBool("core.ignorecase"):
837         return path.lower().startswith(prefix.lower())
838     return path.startswith(prefix)
839
840 def getClientSpec():
841     """Look at the p4 client spec, create a View() object that contains
842        all the mappings, and return it."""
843
844     specList = p4CmdList("client -o")
845     if len(specList) != 1:
846         die('Output from "client -o" is %d lines, expecting 1' %
847             len(specList))
848
849     # dictionary of all client parameters
850     entry = specList[0]
851
852     # the //client/ name
853     client_name = entry["Client"]
854
855     # just the keys that start with "View"
856     view_keys = [ k for k in entry.keys() if k.startswith("View") ]
857
858     # hold this new View
859     view = View(client_name)
860
861     # append the lines, in order, to the view
862     for view_num in range(len(view_keys)):
863         k = "View%d" % view_num
864         if k not in view_keys:
865             die("Expected view key %s missing" % k)
866         view.append(entry[k])
867
868     return view
869
870 def getClientRoot():
871     """Grab the client directory."""
872
873     output = p4CmdList("client -o")
874     if len(output) != 1:
875         die('Output from "client -o" is %d lines, expecting 1' % len(output))
876
877     entry = output[0]
878     if "Root" not in entry:
879         die('Client has no "Root"')
880
881     return entry["Root"]
882
883 #
884 # P4 wildcards are not allowed in filenames.  P4 complains
885 # if you simply add them, but you can force it with "-f", in
886 # which case it translates them into %xx encoding internally.
887 #
888 def wildcard_decode(path):
889     # Search for and fix just these four characters.  Do % last so
890     # that fixing it does not inadvertently create new %-escapes.
891     # Cannot have * in a filename in windows; untested as to
892     # what p4 would do in such a case.
893     if not platform.system() == "Windows":
894         path = path.replace("%2A", "*")
895     path = path.replace("%23", "#") \
896                .replace("%40", "@") \
897                .replace("%25", "%")
898     return path
899
900 def wildcard_encode(path):
901     # do % first to avoid double-encoding the %s introduced here
902     path = path.replace("%", "%25") \
903                .replace("*", "%2A") \
904                .replace("#", "%23") \
905                .replace("@", "%40")
906     return path
907
908 def wildcard_present(path):
909     m = re.search("[*#@%]", path)
910     return m is not None
911
912 class Command:
913     def __init__(self):
914         self.usage = "usage: %prog [options]"
915         self.needsGit = True
916         self.verbose = False
917
918 class P4UserMap:
919     def __init__(self):
920         self.userMapFromPerforceServer = False
921         self.myP4UserId = None
922
923     def p4UserId(self):
924         if self.myP4UserId:
925             return self.myP4UserId
926
927         results = p4CmdList("user -o")
928         for r in results:
929             if r.has_key('User'):
930                 self.myP4UserId = r['User']
931                 return r['User']
932         die("Could not find your p4 user id")
933
934     def p4UserIsMe(self, p4User):
935         # return True if the given p4 user is actually me
936         me = self.p4UserId()
937         if not p4User or p4User != me:
938             return False
939         else:
940             return True
941
942     def getUserCacheFilename(self):
943         home = os.environ.get("HOME", os.environ.get("USERPROFILE"))
944         return home + "/.gitp4-usercache.txt"
945
946     def getUserMapFromPerforceServer(self):
947         if self.userMapFromPerforceServer:
948             return
949         self.users = {}
950         self.emails = {}
951
952         for output in p4CmdList("users"):
953             if not output.has_key("User"):
954                 continue
955             self.users[output["User"]] = output["FullName"] + " <" + output["Email"] + ">"
956             self.emails[output["Email"]] = output["User"]
957
958
959         s = ''
960         for (key, val) in self.users.items():
961             s += "%s\t%s\n" % (key.expandtabs(1), val.expandtabs(1))
962
963         open(self.getUserCacheFilename(), "wb").write(s)
964         self.userMapFromPerforceServer = True
965
966     def loadUserMapFromCache(self):
967         self.users = {}
968         self.userMapFromPerforceServer = False
969         try:
970             cache = open(self.getUserCacheFilename(), "rb")
971             lines = cache.readlines()
972             cache.close()
973             for line in lines:
974                 entry = line.strip().split("\t")
975                 self.users[entry[0]] = entry[1]
976         except IOError:
977             self.getUserMapFromPerforceServer()
978
979 class P4Debug(Command):
980     def __init__(self):
981         Command.__init__(self)
982         self.options = []
983         self.description = "A tool to debug the output of p4 -G."
984         self.needsGit = False
985
986     def run(self, args):
987         j = 0
988         for output in p4CmdList(args):
989             print 'Element: %d' % j
990             j += 1
991             print output
992         return True
993
994 class P4RollBack(Command):
995     def __init__(self):
996         Command.__init__(self)
997         self.options = [
998             optparse.make_option("--local", dest="rollbackLocalBranches", action="store_true")
999         ]
1000         self.description = "A tool to debug the multi-branch import. Don't use :)"
1001         self.rollbackLocalBranches = False
1002
1003     def run(self, args):
1004         if len(args) != 1:
1005             return False
1006         maxChange = int(args[0])
1007
1008         if "p4ExitCode" in p4Cmd("changes -m 1"):
1009             die("Problems executing p4");
1010
1011         if self.rollbackLocalBranches:
1012             refPrefix = "refs/heads/"
1013             lines = read_pipe_lines("git rev-parse --symbolic --branches")
1014         else:
1015             refPrefix = "refs/remotes/"
1016             lines = read_pipe_lines("git rev-parse --symbolic --remotes")
1017
1018         for line in lines:
1019             if self.rollbackLocalBranches or (line.startswith("p4/") and line != "p4/HEAD\n"):
1020                 line = line.strip()
1021                 ref = refPrefix + line
1022                 log = extractLogMessageFromGitCommit(ref)
1023                 settings = extractSettingsGitLog(log)
1024
1025                 depotPaths = settings['depot-paths']
1026                 change = settings['change']
1027
1028                 changed = False
1029
1030                 if len(p4Cmd("changes -m 1 "  + ' '.join (['%s...@%s' % (p, maxChange)
1031                                                            for p in depotPaths]))) == 0:
1032                     print "Branch %s did not exist at change %s, deleting." % (ref, maxChange)
1033                     system("git update-ref -d %s `git rev-parse %s`" % (ref, ref))
1034                     continue
1035
1036                 while change and int(change) > maxChange:
1037                     changed = True
1038                     if self.verbose:
1039                         print "%s is at %s ; rewinding towards %s" % (ref, change, maxChange)
1040                     system("git update-ref %s \"%s^\"" % (ref, ref))
1041                     log = extractLogMessageFromGitCommit(ref)
1042                     settings =  extractSettingsGitLog(log)
1043
1044
1045                     depotPaths = settings['depot-paths']
1046                     change = settings['change']
1047
1048                 if changed:
1049                     print "%s rewound to %s" % (ref, change)
1050
1051         return True
1052
1053 class P4Submit(Command, P4UserMap):
1054
1055     conflict_behavior_choices = ("ask", "skip", "quit")
1056
1057     def __init__(self):
1058         Command.__init__(self)
1059         P4UserMap.__init__(self)
1060         self.options = [
1061                 optparse.make_option("--origin", dest="origin"),
1062                 optparse.make_option("-M", dest="detectRenames", action="store_true"),
1063                 # preserve the user, requires relevant p4 permissions
1064                 optparse.make_option("--preserve-user", dest="preserveUser", action="store_true"),
1065                 optparse.make_option("--export-labels", dest="exportLabels", action="store_true"),
1066                 optparse.make_option("--dry-run", "-n", dest="dry_run", action="store_true"),
1067                 optparse.make_option("--prepare-p4-only", dest="prepare_p4_only", action="store_true"),
1068                 optparse.make_option("--conflict", dest="conflict_behavior",
1069                                      choices=self.conflict_behavior_choices),
1070                 optparse.make_option("--branch", dest="branch"),
1071         ]
1072         self.description = "Submit changes from git to the perforce depot."
1073         self.usage += " [name of git branch to submit into perforce depot]"
1074         self.origin = ""
1075         self.detectRenames = False
1076         self.preserveUser = gitConfigBool("git-p4.preserveUser")
1077         self.dry_run = False
1078         self.prepare_p4_only = False
1079         self.conflict_behavior = None
1080         self.isWindows = (platform.system() == "Windows")
1081         self.exportLabels = False
1082         self.p4HasMoveCommand = p4_has_move_command()
1083         self.branch = None
1084
1085     def check(self):
1086         if len(p4CmdList("opened ...")) > 0:
1087             die("You have files opened with perforce! Close them before starting the sync.")
1088
1089     def separate_jobs_from_description(self, message):
1090         """Extract and return a possible Jobs field in the commit
1091            message.  It goes into a separate section in the p4 change
1092            specification.
1093
1094            A jobs line starts with "Jobs:" and looks like a new field
1095            in a form.  Values are white-space separated on the same
1096            line or on following lines that start with a tab.
1097
1098            This does not parse and extract the full git commit message
1099            like a p4 form.  It just sees the Jobs: line as a marker
1100            to pass everything from then on directly into the p4 form,
1101            but outside the description section.
1102
1103            Return a tuple (stripped log message, jobs string)."""
1104
1105         m = re.search(r'^Jobs:', message, re.MULTILINE)
1106         if m is None:
1107             return (message, None)
1108
1109         jobtext = message[m.start():]
1110         stripped_message = message[:m.start()].rstrip()
1111         return (stripped_message, jobtext)
1112
1113     def prepareLogMessage(self, template, message, jobs):
1114         """Edits the template returned from "p4 change -o" to insert
1115            the message in the Description field, and the jobs text in
1116            the Jobs field."""
1117         result = ""
1118
1119         inDescriptionSection = False
1120
1121         for line in template.split("\n"):
1122             if line.startswith("#"):
1123                 result += line + "\n"
1124                 continue
1125
1126             if inDescriptionSection:
1127                 if line.startswith("Files:") or line.startswith("Jobs:"):
1128                     inDescriptionSection = False
1129                     # insert Jobs section
1130                     if jobs:
1131                         result += jobs + "\n"
1132                 else:
1133                     continue
1134             else:
1135                 if line.startswith("Description:"):
1136                     inDescriptionSection = True
1137                     line += "\n"
1138                     for messageLine in message.split("\n"):
1139                         line += "\t" + messageLine + "\n"
1140
1141             result += line + "\n"
1142
1143         return result
1144
1145     def patchRCSKeywords(self, file, pattern):
1146         # Attempt to zap the RCS keywords in a p4 controlled file matching the given pattern
1147         (handle, outFileName) = tempfile.mkstemp(dir='.')
1148         try:
1149             outFile = os.fdopen(handle, "w+")
1150             inFile = open(file, "r")
1151             regexp = re.compile(pattern, re.VERBOSE)
1152             for line in inFile.readlines():
1153                 line = regexp.sub(r'$\1$', line)
1154                 outFile.write(line)
1155             inFile.close()
1156             outFile.close()
1157             # Forcibly overwrite the original file
1158             os.unlink(file)
1159             shutil.move(outFileName, file)
1160         except:
1161             # cleanup our temporary file
1162             os.unlink(outFileName)
1163             print "Failed to strip RCS keywords in %s" % file
1164             raise
1165
1166         print "Patched up RCS keywords in %s" % file
1167
1168     def p4UserForCommit(self,id):
1169         # Return the tuple (perforce user,git email) for a given git commit id
1170         self.getUserMapFromPerforceServer()
1171         gitEmail = read_pipe(["git", "log", "--max-count=1",
1172                               "--format=%ae", id])
1173         gitEmail = gitEmail.strip()
1174         if not self.emails.has_key(gitEmail):
1175             return (None,gitEmail)
1176         else:
1177             return (self.emails[gitEmail],gitEmail)
1178
1179     def checkValidP4Users(self,commits):
1180         # check if any git authors cannot be mapped to p4 users
1181         for id in commits:
1182             (user,email) = self.p4UserForCommit(id)
1183             if not user:
1184                 msg = "Cannot find p4 user for email %s in commit %s." % (email, id)
1185                 if gitConfigBool("git-p4.allowMissingP4Users"):
1186                     print "%s" % msg
1187                 else:
1188                     die("Error: %s\nSet git-p4.allowMissingP4Users to true to allow this." % msg)
1189
1190     def lastP4Changelist(self):
1191         # Get back the last changelist number submitted in this client spec. This
1192         # then gets used to patch up the username in the change. If the same
1193         # client spec is being used by multiple processes then this might go
1194         # wrong.
1195         results = p4CmdList("client -o")        # find the current client
1196         client = None
1197         for r in results:
1198             if r.has_key('Client'):
1199                 client = r['Client']
1200                 break
1201         if not client:
1202             die("could not get client spec")
1203         results = p4CmdList(["changes", "-c", client, "-m", "1"])
1204         for r in results:
1205             if r.has_key('change'):
1206                 return r['change']
1207         die("Could not get changelist number for last submit - cannot patch up user details")
1208
1209     def modifyChangelistUser(self, changelist, newUser):
1210         # fixup the user field of a changelist after it has been submitted.
1211         changes = p4CmdList("change -o %s" % changelist)
1212         if len(changes) != 1:
1213             die("Bad output from p4 change modifying %s to user %s" %
1214                 (changelist, newUser))
1215
1216         c = changes[0]
1217         if c['User'] == newUser: return   # nothing to do
1218         c['User'] = newUser
1219         input = marshal.dumps(c)
1220
1221         result = p4CmdList("change -f -i", stdin=input)
1222         for r in result:
1223             if r.has_key('code'):
1224                 if r['code'] == 'error':
1225                     die("Could not modify user field of changelist %s to %s:%s" % (changelist, newUser, r['data']))
1226             if r.has_key('data'):
1227                 print("Updated user field for changelist %s to %s" % (changelist, newUser))
1228                 return
1229         die("Could not modify user field of changelist %s to %s" % (changelist, newUser))
1230
1231     def canChangeChangelists(self):
1232         # check to see if we have p4 admin or super-user permissions, either of
1233         # which are required to modify changelists.
1234         results = p4CmdList(["protects", self.depotPath])
1235         for r in results:
1236             if r.has_key('perm'):
1237                 if r['perm'] == 'admin':
1238                     return 1
1239                 if r['perm'] == 'super':
1240                     return 1
1241         return 0
1242
1243     def prepareSubmitTemplate(self):
1244         """Run "p4 change -o" to grab a change specification template.
1245            This does not use "p4 -G", as it is nice to keep the submission
1246            template in original order, since a human might edit it.
1247
1248            Remove lines in the Files section that show changes to files
1249            outside the depot path we're committing into."""
1250
1251         template = ""
1252         inFilesSection = False
1253         for line in p4_read_pipe_lines(['change', '-o']):
1254             if line.endswith("\r\n"):
1255                 line = line[:-2] + "\n"
1256             if inFilesSection:
1257                 if line.startswith("\t"):
1258                     # path starts and ends with a tab
1259                     path = line[1:]
1260                     lastTab = path.rfind("\t")
1261                     if lastTab != -1:
1262                         path = path[:lastTab]
1263                         if not p4PathStartsWith(path, self.depotPath):
1264                             continue
1265                 else:
1266                     inFilesSection = False
1267             else:
1268                 if line.startswith("Files:"):
1269                     inFilesSection = True
1270
1271             template += line
1272
1273         return template
1274
1275     def edit_template(self, template_file):
1276         """Invoke the editor to let the user change the submission
1277            message.  Return true if okay to continue with the submit."""
1278
1279         # if configured to skip the editing part, just submit
1280         if gitConfigBool("git-p4.skipSubmitEdit"):
1281             return True
1282
1283         # look at the modification time, to check later if the user saved
1284         # the file
1285         mtime = os.stat(template_file).st_mtime
1286
1287         # invoke the editor
1288         if os.environ.has_key("P4EDITOR") and (os.environ.get("P4EDITOR") != ""):
1289             editor = os.environ.get("P4EDITOR")
1290         else:
1291             editor = read_pipe("git var GIT_EDITOR").strip()
1292         system(["sh", "-c", ('%s "$@"' % editor), editor, template_file])
1293
1294         # If the file was not saved, prompt to see if this patch should
1295         # be skipped.  But skip this verification step if configured so.
1296         if gitConfigBool("git-p4.skipSubmitEditCheck"):
1297             return True
1298
1299         # modification time updated means user saved the file
1300         if os.stat(template_file).st_mtime > mtime:
1301             return True
1302
1303         while True:
1304             response = raw_input("Submit template unchanged. Submit anyway? [y]es, [n]o (skip this patch) ")
1305             if response == 'y':
1306                 return True
1307             if response == 'n':
1308                 return False
1309
1310     def get_diff_description(self, editedFiles, filesToAdd):
1311         # diff
1312         if os.environ.has_key("P4DIFF"):
1313             del(os.environ["P4DIFF"])
1314         diff = ""
1315         for editedFile in editedFiles:
1316             diff += p4_read_pipe(['diff', '-du',
1317                                   wildcard_encode(editedFile)])
1318
1319         # new file diff
1320         newdiff = ""
1321         for newFile in filesToAdd:
1322             newdiff += "==== new file ====\n"
1323             newdiff += "--- /dev/null\n"
1324             newdiff += "+++ %s\n" % newFile
1325             f = open(newFile, "r")
1326             for line in f.readlines():
1327                 newdiff += "+" + line
1328             f.close()
1329
1330         return (diff + newdiff).replace('\r\n', '\n')
1331
1332     def applyCommit(self, id):
1333         """Apply one commit, return True if it succeeded."""
1334
1335         print "Applying", read_pipe(["git", "show", "-s",
1336                                      "--format=format:%h %s", id])
1337
1338         (p4User, gitEmail) = self.p4UserForCommit(id)
1339
1340         diff = read_pipe_lines("git diff-tree -r %s \"%s^\" \"%s\"" % (self.diffOpts, id, id))
1341         filesToAdd = set()
1342         filesToDelete = set()
1343         editedFiles = set()
1344         pureRenameCopy = set()
1345         filesToChangeExecBit = {}
1346
1347         for line in diff:
1348             diff = parseDiffTreeEntry(line)
1349             modifier = diff['status']
1350             path = diff['src']
1351             if modifier == "M":
1352                 p4_edit(path)
1353                 if isModeExecChanged(diff['src_mode'], diff['dst_mode']):
1354                     filesToChangeExecBit[path] = diff['dst_mode']
1355                 editedFiles.add(path)
1356             elif modifier == "A":
1357                 filesToAdd.add(path)
1358                 filesToChangeExecBit[path] = diff['dst_mode']
1359                 if path in filesToDelete:
1360                     filesToDelete.remove(path)
1361             elif modifier == "D":
1362                 filesToDelete.add(path)
1363                 if path in filesToAdd:
1364                     filesToAdd.remove(path)
1365             elif modifier == "C":
1366                 src, dest = diff['src'], diff['dst']
1367                 p4_integrate(src, dest)
1368                 pureRenameCopy.add(dest)
1369                 if diff['src_sha1'] != diff['dst_sha1']:
1370                     p4_edit(dest)
1371                     pureRenameCopy.discard(dest)
1372                 if isModeExecChanged(diff['src_mode'], diff['dst_mode']):
1373                     p4_edit(dest)
1374                     pureRenameCopy.discard(dest)
1375                     filesToChangeExecBit[dest] = diff['dst_mode']
1376                 if self.isWindows:
1377                     # turn off read-only attribute
1378                     os.chmod(dest, stat.S_IWRITE)
1379                 os.unlink(dest)
1380                 editedFiles.add(dest)
1381             elif modifier == "R":
1382                 src, dest = diff['src'], diff['dst']
1383                 if self.p4HasMoveCommand:
1384                     p4_edit(src)        # src must be open before move
1385                     p4_move(src, dest)  # opens for (move/delete, move/add)
1386                 else:
1387                     p4_integrate(src, dest)
1388                     if diff['src_sha1'] != diff['dst_sha1']:
1389                         p4_edit(dest)
1390                     else:
1391                         pureRenameCopy.add(dest)
1392                 if isModeExecChanged(diff['src_mode'], diff['dst_mode']):
1393                     if not self.p4HasMoveCommand:
1394                         p4_edit(dest)   # with move: already open, writable
1395                     filesToChangeExecBit[dest] = diff['dst_mode']
1396                 if not self.p4HasMoveCommand:
1397                     if self.isWindows:
1398                         os.chmod(dest, stat.S_IWRITE)
1399                     os.unlink(dest)
1400                     filesToDelete.add(src)
1401                 editedFiles.add(dest)
1402             else:
1403                 die("unknown modifier %s for %s" % (modifier, path))
1404
1405         diffcmd = "git diff-tree --full-index -p \"%s\"" % (id)
1406         patchcmd = diffcmd + " | git apply "
1407         tryPatchCmd = patchcmd + "--check -"
1408         applyPatchCmd = patchcmd + "--check --apply -"
1409         patch_succeeded = True
1410
1411         if os.system(tryPatchCmd) != 0:
1412             fixed_rcs_keywords = False
1413             patch_succeeded = False
1414             print "Unfortunately applying the change failed!"
1415
1416             # Patch failed, maybe it's just RCS keyword woes. Look through
1417             # the patch to see if that's possible.
1418             if gitConfigBool("git-p4.attemptRCSCleanup"):
1419                 file = None
1420                 pattern = None
1421                 kwfiles = {}
1422                 for file in editedFiles | filesToDelete:
1423                     # did this file's delta contain RCS keywords?
1424                     pattern = p4_keywords_regexp_for_file(file)
1425
1426                     if pattern:
1427                         # this file is a possibility...look for RCS keywords.
1428                         regexp = re.compile(pattern, re.VERBOSE)
1429                         for line in read_pipe_lines(["git", "diff", "%s^..%s" % (id, id), file]):
1430                             if regexp.search(line):
1431                                 if verbose:
1432                                     print "got keyword match on %s in %s in %s" % (pattern, line, file)
1433                                 kwfiles[file] = pattern
1434                                 break
1435
1436                 for file in kwfiles:
1437                     if verbose:
1438                         print "zapping %s with %s" % (line,pattern)
1439                     # File is being deleted, so not open in p4.  Must
1440                     # disable the read-only bit on windows.
1441                     if self.isWindows and file not in editedFiles:
1442                         os.chmod(file, stat.S_IWRITE)
1443                     self.patchRCSKeywords(file, kwfiles[file])
1444                     fixed_rcs_keywords = True
1445
1446             if fixed_rcs_keywords:
1447                 print "Retrying the patch with RCS keywords cleaned up"
1448                 if os.system(tryPatchCmd) == 0:
1449                     patch_succeeded = True
1450
1451         if not patch_succeeded:
1452             for f in editedFiles:
1453                 p4_revert(f)
1454             return False
1455
1456         #
1457         # Apply the patch for real, and do add/delete/+x handling.
1458         #
1459         system(applyPatchCmd)
1460
1461         for f in filesToAdd:
1462             p4_add(f)
1463         for f in filesToDelete:
1464             p4_revert(f)
1465             p4_delete(f)
1466
1467         # Set/clear executable bits
1468         for f in filesToChangeExecBit.keys():
1469             mode = filesToChangeExecBit[f]
1470             setP4ExecBit(f, mode)
1471
1472         #
1473         # Build p4 change description, starting with the contents
1474         # of the git commit message.
1475         #
1476         logMessage = extractLogMessageFromGitCommit(id)
1477         logMessage = logMessage.strip()
1478         (logMessage, jobs) = self.separate_jobs_from_description(logMessage)
1479
1480         template = self.prepareSubmitTemplate()
1481         submitTemplate = self.prepareLogMessage(template, logMessage, jobs)
1482
1483         if self.preserveUser:
1484            submitTemplate += "\n######## Actual user %s, modified after commit\n" % p4User
1485
1486         if self.checkAuthorship and not self.p4UserIsMe(p4User):
1487             submitTemplate += "######## git author %s does not match your p4 account.\n" % gitEmail
1488             submitTemplate += "######## Use option --preserve-user to modify authorship.\n"
1489             submitTemplate += "######## Variable git-p4.skipUserNameCheck hides this message.\n"
1490
1491         separatorLine = "######## everything below this line is just the diff #######\n"
1492         if not self.prepare_p4_only:
1493             submitTemplate += separatorLine
1494             submitTemplate += self.get_diff_description(editedFiles, filesToAdd)
1495
1496         (handle, fileName) = tempfile.mkstemp()
1497         tmpFile = os.fdopen(handle, "w+b")
1498         if self.isWindows:
1499             submitTemplate = submitTemplate.replace("\n", "\r\n")
1500         tmpFile.write(submitTemplate)
1501         tmpFile.close()
1502
1503         if self.prepare_p4_only:
1504             #
1505             # Leave the p4 tree prepared, and the submit template around
1506             # and let the user decide what to do next
1507             #
1508             print
1509             print "P4 workspace prepared for submission."
1510             print "To submit or revert, go to client workspace"
1511             print "  " + self.clientPath
1512             print
1513             print "To submit, use \"p4 submit\" to write a new description,"
1514             print "or \"p4 submit -i <%s\" to use the one prepared by" \
1515                   " \"git p4\"." % fileName
1516             print "You can delete the file \"%s\" when finished." % fileName
1517
1518             if self.preserveUser and p4User and not self.p4UserIsMe(p4User):
1519                 print "To preserve change ownership by user %s, you must\n" \
1520                       "do \"p4 change -f <change>\" after submitting and\n" \
1521                       "edit the User field."
1522             if pureRenameCopy:
1523                 print "After submitting, renamed files must be re-synced."
1524                 print "Invoke \"p4 sync -f\" on each of these files:"
1525                 for f in pureRenameCopy:
1526                     print "  " + f
1527
1528             print
1529             print "To revert the changes, use \"p4 revert ...\", and delete"
1530             print "the submit template file \"%s\"" % fileName
1531             if filesToAdd:
1532                 print "Since the commit adds new files, they must be deleted:"
1533                 for f in filesToAdd:
1534                     print "  " + f
1535             print
1536             return True
1537
1538         #
1539         # Let the user edit the change description, then submit it.
1540         #
1541         submitted = False
1542
1543         try:
1544             if self.edit_template(fileName):
1545                 # read the edited message and submit
1546                 tmpFile = open(fileName, "rb")
1547                 message = tmpFile.read()
1548                 tmpFile.close()
1549                 if self.isWindows:
1550                     message = message.replace("\r\n", "\n")
1551                 submitTemplate = message[:message.index(separatorLine)]
1552                 p4_write_pipe(['submit', '-i'], submitTemplate)
1553
1554                 if self.preserveUser:
1555                     if p4User:
1556                         # Get last changelist number. Cannot easily get it from
1557                         # the submit command output as the output is
1558                         # unmarshalled.
1559                         changelist = self.lastP4Changelist()
1560                         self.modifyChangelistUser(changelist, p4User)
1561
1562                 # The rename/copy happened by applying a patch that created a
1563                 # new file.  This leaves it writable, which confuses p4.
1564                 for f in pureRenameCopy:
1565                     p4_sync(f, "-f")
1566                 submitted = True
1567
1568         finally:
1569             # skip this patch
1570             if not submitted:
1571                 print "Submission cancelled, undoing p4 changes."
1572                 for f in editedFiles:
1573                     p4_revert(f)
1574                 for f in filesToAdd:
1575                     p4_revert(f)
1576                     os.remove(f)
1577                 for f in filesToDelete:
1578                     p4_revert(f)
1579
1580         os.remove(fileName)
1581         return submitted
1582
1583     # Export git tags as p4 labels. Create a p4 label and then tag
1584     # with that.
1585     def exportGitTags(self, gitTags):
1586         validLabelRegexp = gitConfig("git-p4.labelExportRegexp")
1587         if len(validLabelRegexp) == 0:
1588             validLabelRegexp = defaultLabelRegexp
1589         m = re.compile(validLabelRegexp)
1590
1591         for name in gitTags:
1592
1593             if not m.match(name):
1594                 if verbose:
1595                     print "tag %s does not match regexp %s" % (name, validLabelRegexp)
1596                 continue
1597
1598             # Get the p4 commit this corresponds to
1599             logMessage = extractLogMessageFromGitCommit(name)
1600             values = extractSettingsGitLog(logMessage)
1601
1602             if not values.has_key('change'):
1603                 # a tag pointing to something not sent to p4; ignore
1604                 if verbose:
1605                     print "git tag %s does not give a p4 commit" % name
1606                 continue
1607             else:
1608                 changelist = values['change']
1609
1610             # Get the tag details.
1611             inHeader = True
1612             isAnnotated = False
1613             body = []
1614             for l in read_pipe_lines(["git", "cat-file", "-p", name]):
1615                 l = l.strip()
1616                 if inHeader:
1617                     if re.match(r'tag\s+', l):
1618                         isAnnotated = True
1619                     elif re.match(r'\s*$', l):
1620                         inHeader = False
1621                         continue
1622                 else:
1623                     body.append(l)
1624
1625             if not isAnnotated:
1626                 body = ["lightweight tag imported by git p4\n"]
1627
1628             # Create the label - use the same view as the client spec we are using
1629             clientSpec = getClientSpec()
1630
1631             labelTemplate  = "Label: %s\n" % name
1632             labelTemplate += "Description:\n"
1633             for b in body:
1634                 labelTemplate += "\t" + b + "\n"
1635             labelTemplate += "View:\n"
1636             for depot_side in clientSpec.mappings:
1637                 labelTemplate += "\t%s\n" % depot_side
1638
1639             if self.dry_run:
1640                 print "Would create p4 label %s for tag" % name
1641             elif self.prepare_p4_only:
1642                 print "Not creating p4 label %s for tag due to option" \
1643                       " --prepare-p4-only" % name
1644             else:
1645                 p4_write_pipe(["label", "-i"], labelTemplate)
1646
1647                 # Use the label
1648                 p4_system(["tag", "-l", name] +
1649                           ["%s@%s" % (depot_side, changelist) for depot_side in clientSpec.mappings])
1650
1651                 if verbose:
1652                     print "created p4 label for tag %s" % name
1653
1654     def run(self, args):
1655         if len(args) == 0:
1656             self.master = currentGitBranch()
1657             if len(self.master) == 0 or not gitBranchExists("refs/heads/%s" % self.master):
1658                 die("Detecting current git branch failed!")
1659         elif len(args) == 1:
1660             self.master = args[0]
1661             if not branchExists(self.master):
1662                 die("Branch %s does not exist" % self.master)
1663         else:
1664             return False
1665
1666         allowSubmit = gitConfig("git-p4.allowSubmit")
1667         if len(allowSubmit) > 0 and not self.master in allowSubmit.split(","):
1668             die("%s is not in git-p4.allowSubmit" % self.master)
1669
1670         [upstream, settings] = findUpstreamBranchPoint()
1671         self.depotPath = settings['depot-paths'][0]
1672         if len(self.origin) == 0:
1673             self.origin = upstream
1674
1675         if self.preserveUser:
1676             if not self.canChangeChangelists():
1677                 die("Cannot preserve user names without p4 super-user or admin permissions")
1678
1679         # if not set from the command line, try the config file
1680         if self.conflict_behavior is None:
1681             val = gitConfig("git-p4.conflict")
1682             if val:
1683                 if val not in self.conflict_behavior_choices:
1684                     die("Invalid value '%s' for config git-p4.conflict" % val)
1685             else:
1686                 val = "ask"
1687             self.conflict_behavior = val
1688
1689         if self.verbose:
1690             print "Origin branch is " + self.origin
1691
1692         if len(self.depotPath) == 0:
1693             print "Internal error: cannot locate perforce depot path from existing branches"
1694             sys.exit(128)
1695
1696         self.useClientSpec = False
1697         if gitConfigBool("git-p4.useclientspec"):
1698             self.useClientSpec = True
1699         if self.useClientSpec:
1700             self.clientSpecDirs = getClientSpec()
1701
1702         # Check for the existance of P4 branches
1703         branchesDetected = (len(p4BranchesInGit().keys()) > 1)
1704
1705         if self.useClientSpec and not branchesDetected:
1706             # all files are relative to the client spec
1707             self.clientPath = getClientRoot()
1708         else:
1709             self.clientPath = p4Where(self.depotPath)
1710
1711         if self.clientPath == "":
1712             die("Error: Cannot locate perforce checkout of %s in client view" % self.depotPath)
1713
1714         print "Perforce checkout for depot path %s located at %s" % (self.depotPath, self.clientPath)
1715         self.oldWorkingDirectory = os.getcwd()
1716
1717         # ensure the clientPath exists
1718         new_client_dir = False
1719         if not os.path.exists(self.clientPath):
1720             new_client_dir = True
1721             os.makedirs(self.clientPath)
1722
1723         chdir(self.clientPath, is_client_path=True)
1724         if self.dry_run:
1725             print "Would synchronize p4 checkout in %s" % self.clientPath
1726         else:
1727             print "Synchronizing p4 checkout..."
1728             if new_client_dir:
1729                 # old one was destroyed, and maybe nobody told p4
1730                 p4_sync("...", "-f")
1731             else:
1732                 p4_sync("...")
1733         self.check()
1734
1735         commits = []
1736         for line in read_pipe_lines(["git", "rev-list", "--no-merges", "%s..%s" % (self.origin, self.master)]):
1737             commits.append(line.strip())
1738         commits.reverse()
1739
1740         if self.preserveUser or gitConfigBool("git-p4.skipUserNameCheck"):
1741             self.checkAuthorship = False
1742         else:
1743             self.checkAuthorship = True
1744
1745         if self.preserveUser:
1746             self.checkValidP4Users(commits)
1747
1748         #
1749         # Build up a set of options to be passed to diff when
1750         # submitting each commit to p4.
1751         #
1752         if self.detectRenames:
1753             # command-line -M arg
1754             self.diffOpts = "-M"
1755         else:
1756             # If not explicitly set check the config variable
1757             detectRenames = gitConfig("git-p4.detectRenames")
1758
1759             if detectRenames.lower() == "false" or detectRenames == "":
1760                 self.diffOpts = ""
1761             elif detectRenames.lower() == "true":
1762                 self.diffOpts = "-M"
1763             else:
1764                 self.diffOpts = "-M%s" % detectRenames
1765
1766         # no command-line arg for -C or --find-copies-harder, just
1767         # config variables
1768         detectCopies = gitConfig("git-p4.detectCopies")
1769         if detectCopies.lower() == "false" or detectCopies == "":
1770             pass
1771         elif detectCopies.lower() == "true":
1772             self.diffOpts += " -C"
1773         else:
1774             self.diffOpts += " -C%s" % detectCopies
1775
1776         if gitConfigBool("git-p4.detectCopiesHarder"):
1777             self.diffOpts += " --find-copies-harder"
1778
1779         #
1780         # Apply the commits, one at a time.  On failure, ask if should
1781         # continue to try the rest of the patches, or quit.
1782         #
1783         if self.dry_run:
1784             print "Would apply"
1785         applied = []
1786         last = len(commits) - 1
1787         for i, commit in enumerate(commits):
1788             if self.dry_run:
1789                 print " ", read_pipe(["git", "show", "-s",
1790                                       "--format=format:%h %s", commit])
1791                 ok = True
1792             else:
1793                 ok = self.applyCommit(commit)
1794             if ok:
1795                 applied.append(commit)
1796             else:
1797                 if self.prepare_p4_only and i < last:
1798                     print "Processing only the first commit due to option" \
1799                           " --prepare-p4-only"
1800                     break
1801                 if i < last:
1802                     quit = False
1803                     while True:
1804                         # prompt for what to do, or use the option/variable
1805                         if self.conflict_behavior == "ask":
1806                             print "What do you want to do?"
1807                             response = raw_input("[s]kip this commit but apply"
1808                                                  " the rest, or [q]uit? ")
1809                             if not response:
1810                                 continue
1811                         elif self.conflict_behavior == "skip":
1812                             response = "s"
1813                         elif self.conflict_behavior == "quit":
1814                             response = "q"
1815                         else:
1816                             die("Unknown conflict_behavior '%s'" %
1817                                 self.conflict_behavior)
1818
1819                         if response[0] == "s":
1820                             print "Skipping this commit, but applying the rest"
1821                             break
1822                         if response[0] == "q":
1823                             print "Quitting"
1824                             quit = True
1825                             break
1826                     if quit:
1827                         break
1828
1829         chdir(self.oldWorkingDirectory)
1830
1831         if self.dry_run:
1832             pass
1833         elif self.prepare_p4_only:
1834             pass
1835         elif len(commits) == len(applied):
1836             print "All commits applied!"
1837
1838             sync = P4Sync()
1839             if self.branch:
1840                 sync.branch = self.branch
1841             sync.run([])
1842
1843             rebase = P4Rebase()
1844             rebase.rebase()
1845
1846         else:
1847             if len(applied) == 0:
1848                 print "No commits applied."
1849             else:
1850                 print "Applied only the commits marked with '*':"
1851                 for c in commits:
1852                     if c in applied:
1853                         star = "*"
1854                     else:
1855                         star = " "
1856                     print star, read_pipe(["git", "show", "-s",
1857                                            "--format=format:%h %s",  c])
1858                 print "You will have to do 'git p4 sync' and rebase."
1859
1860         if gitConfigBool("git-p4.exportLabels"):
1861             self.exportLabels = True
1862
1863         if self.exportLabels:
1864             p4Labels = getP4Labels(self.depotPath)
1865             gitTags = getGitTags()
1866
1867             missingGitTags = gitTags - p4Labels
1868             self.exportGitTags(missingGitTags)
1869
1870         # exit with error unless everything applied perfectly
1871         if len(commits) != len(applied):
1872                 sys.exit(1)
1873
1874         return True
1875
1876 class View(object):
1877     """Represent a p4 view ("p4 help views"), and map files in a
1878        repo according to the view."""
1879
1880     def __init__(self, client_name):
1881         self.mappings = []
1882         self.client_prefix = "//%s/" % client_name
1883         # cache results of "p4 where" to lookup client file locations
1884         self.client_spec_path_cache = {}
1885
1886     def append(self, view_line):
1887         """Parse a view line, splitting it into depot and client
1888            sides.  Append to self.mappings, preserving order.  This
1889            is only needed for tag creation."""
1890
1891         # Split the view line into exactly two words.  P4 enforces
1892         # structure on these lines that simplifies this quite a bit.
1893         #
1894         # Either or both words may be double-quoted.
1895         # Single quotes do not matter.
1896         # Double-quote marks cannot occur inside the words.
1897         # A + or - prefix is also inside the quotes.
1898         # There are no quotes unless they contain a space.
1899         # The line is already white-space stripped.
1900         # The two words are separated by a single space.
1901         #
1902         if view_line[0] == '"':
1903             # First word is double quoted.  Find its end.
1904             close_quote_index = view_line.find('"', 1)
1905             if close_quote_index <= 0:
1906                 die("No first-word closing quote found: %s" % view_line)
1907             depot_side = view_line[1:close_quote_index]
1908             # skip closing quote and space
1909             rhs_index = close_quote_index + 1 + 1
1910         else:
1911             space_index = view_line.find(" ")
1912             if space_index <= 0:
1913                 die("No word-splitting space found: %s" % view_line)
1914             depot_side = view_line[0:space_index]
1915             rhs_index = space_index + 1
1916
1917         # prefix + means overlay on previous mapping
1918         if depot_side.startswith("+"):
1919             depot_side = depot_side[1:]
1920
1921         # prefix - means exclude this path, leave out of mappings
1922         exclude = False
1923         if depot_side.startswith("-"):
1924             exclude = True
1925             depot_side = depot_side[1:]
1926
1927         if not exclude:
1928             self.mappings.append(depot_side)
1929
1930     def convert_client_path(self, clientFile):
1931         # chop off //client/ part to make it relative
1932         if not clientFile.startswith(self.client_prefix):
1933             die("No prefix '%s' on clientFile '%s'" %
1934                 (self.client_prefix, clientFile))
1935         return clientFile[len(self.client_prefix):]
1936
1937     def update_client_spec_path_cache(self, files):
1938         """ Caching file paths by "p4 where" batch query """
1939
1940         # List depot file paths exclude that already cached
1941         fileArgs = [f['path'] for f in files if f['path'] not in self.client_spec_path_cache]
1942
1943         if len(fileArgs) == 0:
1944             return  # All files in cache
1945
1946         where_result = p4CmdList(["-x", "-", "where"], stdin=fileArgs)
1947         for res in where_result:
1948             if "code" in res and res["code"] == "error":
1949                 # assume error is "... file(s) not in client view"
1950                 continue
1951             if "clientFile" not in res:
1952                 die("No clientFile in 'p4 where' output")
1953             if "unmap" in res:
1954                 # it will list all of them, but only one not unmap-ped
1955                 continue
1956             if gitConfigBool("core.ignorecase"):
1957                 res['depotFile'] = res['depotFile'].lower()
1958             self.client_spec_path_cache[res['depotFile']] = self.convert_client_path(res["clientFile"])
1959
1960         # not found files or unmap files set to ""
1961         for depotFile in fileArgs:
1962             if gitConfigBool("core.ignorecase"):
1963                 depotFile = depotFile.lower()
1964             if depotFile not in self.client_spec_path_cache:
1965                 self.client_spec_path_cache[depotFile] = ""
1966
1967     def map_in_client(self, depot_path):
1968         """Return the relative location in the client where this
1969            depot file should live.  Returns "" if the file should
1970            not be mapped in the client."""
1971
1972         if gitConfigBool("core.ignorecase"):
1973             depot_path = depot_path.lower()
1974
1975         if depot_path in self.client_spec_path_cache:
1976             return self.client_spec_path_cache[depot_path]
1977
1978         die( "Error: %s is not found in client spec path" % depot_path )
1979         return ""
1980
1981 class P4Sync(Command, P4UserMap):
1982     delete_actions = ( "delete", "move/delete", "purge" )
1983
1984     def __init__(self):
1985         Command.__init__(self)
1986         P4UserMap.__init__(self)
1987         self.options = [
1988                 optparse.make_option("--branch", dest="branch"),
1989                 optparse.make_option("--detect-branches", dest="detectBranches", action="store_true"),
1990                 optparse.make_option("--changesfile", dest="changesFile"),
1991                 optparse.make_option("--silent", dest="silent", action="store_true"),
1992                 optparse.make_option("--detect-labels", dest="detectLabels", action="store_true"),
1993                 optparse.make_option("--import-labels", dest="importLabels", action="store_true"),
1994                 optparse.make_option("--import-local", dest="importIntoRemotes", action="store_false",
1995                                      help="Import into refs/heads/ , not refs/remotes"),
1996                 optparse.make_option("--max-changes", dest="maxChanges",
1997                                      help="Maximum number of changes to import"),
1998                 optparse.make_option("--changes-block-size", dest="changes_block_size", type="int",
1999                                      help="Internal block size to use when iteratively calling p4 changes"),
2000                 optparse.make_option("--keep-path", dest="keepRepoPath", action='store_true',
2001                                      help="Keep entire BRANCH/DIR/SUBDIR prefix during import"),
2002                 optparse.make_option("--use-client-spec", dest="useClientSpec", action='store_true',
2003                                      help="Only sync files that are included in the Perforce Client Spec"),
2004                 optparse.make_option("-/", dest="cloneExclude",
2005                                      action="append", type="string",
2006                                      help="exclude depot path"),
2007         ]
2008         self.description = """Imports from Perforce into a git repository.\n
2009     example:
2010     //depot/my/project/ -- to import the current head
2011     //depot/my/project/@all -- to import everything
2012     //depot/my/project/@1,6 -- to import only from revision 1 to 6
2013
2014     (a ... is not needed in the path p4 specification, it's added implicitly)"""
2015
2016         self.usage += " //depot/path[@revRange]"
2017         self.silent = False
2018         self.createdBranches = set()
2019         self.committedChanges = set()
2020         self.branch = ""
2021         self.detectBranches = False
2022         self.detectLabels = False
2023         self.importLabels = False
2024         self.changesFile = ""
2025         self.syncWithOrigin = True
2026         self.importIntoRemotes = True
2027         self.maxChanges = ""
2028         self.changes_block_size = None
2029         self.keepRepoPath = False
2030         self.depotPaths = None
2031         self.p4BranchesInGit = []
2032         self.cloneExclude = []
2033         self.useClientSpec = False
2034         self.useClientSpec_from_options = False
2035         self.clientSpecDirs = None
2036         self.tempBranches = []
2037         self.tempBranchLocation = "git-p4-tmp"
2038
2039         if gitConfig("git-p4.syncFromOrigin") == "false":
2040             self.syncWithOrigin = False
2041
2042     # This is required for the "append" cloneExclude action
2043     def ensure_value(self, attr, value):
2044         if not hasattr(self, attr) or getattr(self, attr) is None:
2045             setattr(self, attr, value)
2046         return getattr(self, attr)
2047
2048     # Force a checkpoint in fast-import and wait for it to finish
2049     def checkpoint(self):
2050         self.gitStream.write("checkpoint\n\n")
2051         self.gitStream.write("progress checkpoint\n\n")
2052         out = self.gitOutput.readline()
2053         if self.verbose:
2054             print "checkpoint finished: " + out
2055
2056     def extractFilesFromCommit(self, commit):
2057         self.cloneExclude = [re.sub(r"\.\.\.$", "", path)
2058                              for path in self.cloneExclude]
2059         files = []
2060         fnum = 0
2061         while commit.has_key("depotFile%s" % fnum):
2062             path =  commit["depotFile%s" % fnum]
2063
2064             if [p for p in self.cloneExclude
2065                 if p4PathStartsWith(path, p)]:
2066                 found = False
2067             else:
2068                 found = [p for p in self.depotPaths
2069                          if p4PathStartsWith(path, p)]
2070             if not found:
2071                 fnum = fnum + 1
2072                 continue
2073
2074             file = {}
2075             file["path"] = path
2076             file["rev"] = commit["rev%s" % fnum]
2077             file["action"] = commit["action%s" % fnum]
2078             file["type"] = commit["type%s" % fnum]
2079             files.append(file)
2080             fnum = fnum + 1
2081         return files
2082
2083     def stripRepoPath(self, path, prefixes):
2084         """When streaming files, this is called to map a p4 depot path
2085            to where it should go in git.  The prefixes are either
2086            self.depotPaths, or self.branchPrefixes in the case of
2087            branch detection."""
2088
2089         if self.useClientSpec:
2090             # branch detection moves files up a level (the branch name)
2091             # from what client spec interpretation gives
2092             path = self.clientSpecDirs.map_in_client(path)
2093             if self.detectBranches:
2094                 for b in self.knownBranches:
2095                     if path.startswith(b + "/"):
2096                         path = path[len(b)+1:]
2097
2098         elif self.keepRepoPath:
2099             # Preserve everything in relative path name except leading
2100             # //depot/; just look at first prefix as they all should
2101             # be in the same depot.
2102             depot = re.sub("^(//[^/]+/).*", r'\1', prefixes[0])
2103             if p4PathStartsWith(path, depot):
2104                 path = path[len(depot):]
2105
2106         else:
2107             for p in prefixes:
2108                 if p4PathStartsWith(path, p):
2109                     path = path[len(p):]
2110                     break
2111
2112         path = wildcard_decode(path)
2113         return path
2114
2115     def splitFilesIntoBranches(self, commit):
2116         """Look at each depotFile in the commit to figure out to what
2117            branch it belongs."""
2118
2119         if self.clientSpecDirs:
2120             files = self.extractFilesFromCommit(commit)
2121             self.clientSpecDirs.update_client_spec_path_cache(files)
2122
2123         branches = {}
2124         fnum = 0
2125         while commit.has_key("depotFile%s" % fnum):
2126             path =  commit["depotFile%s" % fnum]
2127             found = [p for p in self.depotPaths
2128                      if p4PathStartsWith(path, p)]
2129             if not found:
2130                 fnum = fnum + 1
2131                 continue
2132
2133             file = {}
2134             file["path"] = path
2135             file["rev"] = commit["rev%s" % fnum]
2136             file["action"] = commit["action%s" % fnum]
2137             file["type"] = commit["type%s" % fnum]
2138             fnum = fnum + 1
2139
2140             # start with the full relative path where this file would
2141             # go in a p4 client
2142             if self.useClientSpec:
2143                 relPath = self.clientSpecDirs.map_in_client(path)
2144             else:
2145                 relPath = self.stripRepoPath(path, self.depotPaths)
2146
2147             for branch in self.knownBranches.keys():
2148                 # add a trailing slash so that a commit into qt/4.2foo
2149                 # doesn't end up in qt/4.2, e.g.
2150                 if relPath.startswith(branch + "/"):
2151                     if branch not in branches:
2152                         branches[branch] = []
2153                     branches[branch].append(file)
2154                     break
2155
2156         return branches
2157
2158     # output one file from the P4 stream
2159     # - helper for streamP4Files
2160
2161     def streamOneP4File(self, file, contents):
2162         relPath = self.stripRepoPath(file['depotFile'], self.branchPrefixes)
2163         if verbose:
2164             sys.stderr.write("%s\n" % relPath)
2165
2166         (type_base, type_mods) = split_p4_type(file["type"])
2167
2168         git_mode = "100644"
2169         if "x" in type_mods:
2170             git_mode = "100755"
2171         if type_base == "symlink":
2172             git_mode = "120000"
2173             # p4 print on a symlink sometimes contains "target\n";
2174             # if it does, remove the newline
2175             data = ''.join(contents)
2176             if not data:
2177                 # Some version of p4 allowed creating a symlink that pointed
2178                 # to nothing.  This causes p4 errors when checking out such
2179                 # a change, and errors here too.  Work around it by ignoring
2180                 # the bad symlink; hopefully a future change fixes it.
2181                 print "\nIgnoring empty symlink in %s" % file['depotFile']
2182                 return
2183             elif data[-1] == '\n':
2184                 contents = [data[:-1]]
2185             else:
2186                 contents = [data]
2187
2188         if type_base == "utf16":
2189             # p4 delivers different text in the python output to -G
2190             # than it does when using "print -o", or normal p4 client
2191             # operations.  utf16 is converted to ascii or utf8, perhaps.
2192             # But ascii text saved as -t utf16 is completely mangled.
2193             # Invoke print -o to get the real contents.
2194             #
2195             # On windows, the newlines will always be mangled by print, so put
2196             # them back too.  This is not needed to the cygwin windows version,
2197             # just the native "NT" type.
2198             #
2199             text = p4_read_pipe(['print', '-q', '-o', '-', "%s@%s" % (file['depotFile'], file['change']) ])
2200             if p4_version_string().find("/NT") >= 0:
2201                 text = text.replace("\r\n", "\n")
2202             contents = [ text ]
2203
2204         if type_base == "apple":
2205             # Apple filetype files will be streamed as a concatenation of
2206             # its appledouble header and the contents.  This is useless
2207             # on both macs and non-macs.  If using "print -q -o xx", it
2208             # will create "xx" with the data, and "%xx" with the header.
2209             # This is also not very useful.
2210             #
2211             # Ideally, someday, this script can learn how to generate
2212             # appledouble files directly and import those to git, but
2213             # non-mac machines can never find a use for apple filetype.
2214             print "\nIgnoring apple filetype file %s" % file['depotFile']
2215             return
2216
2217         # Note that we do not try to de-mangle keywords on utf16 files,
2218         # even though in theory somebody may want that.
2219         pattern = p4_keywords_regexp_for_type(type_base, type_mods)
2220         if pattern:
2221             regexp = re.compile(pattern, re.VERBOSE)
2222             text = ''.join(contents)
2223             text = regexp.sub(r'$\1$', text)
2224             contents = [ text ]
2225
2226         self.gitStream.write("M %s inline %s\n" % (git_mode, relPath))
2227
2228         # total length...
2229         length = 0
2230         for d in contents:
2231             length = length + len(d)
2232
2233         self.gitStream.write("data %d\n" % length)
2234         for d in contents:
2235             self.gitStream.write(d)
2236         self.gitStream.write("\n")
2237
2238     def streamOneP4Deletion(self, file):
2239         relPath = self.stripRepoPath(file['path'], self.branchPrefixes)
2240         if verbose:
2241             sys.stderr.write("delete %s\n" % relPath)
2242         self.gitStream.write("D %s\n" % relPath)
2243
2244     # handle another chunk of streaming data
2245     def streamP4FilesCb(self, marshalled):
2246
2247         # catch p4 errors and complain
2248         err = None
2249         if "code" in marshalled:
2250             if marshalled["code"] == "error":
2251                 if "data" in marshalled:
2252                     err = marshalled["data"].rstrip()
2253         if err:
2254             f = None
2255             if self.stream_have_file_info:
2256                 if "depotFile" in self.stream_file:
2257                     f = self.stream_file["depotFile"]
2258             # force a failure in fast-import, else an empty
2259             # commit will be made
2260             self.gitStream.write("\n")
2261             self.gitStream.write("die-now\n")
2262             self.gitStream.close()
2263             # ignore errors, but make sure it exits first
2264             self.importProcess.wait()
2265             if f:
2266                 die("Error from p4 print for %s: %s" % (f, err))
2267             else:
2268                 die("Error from p4 print: %s" % err)
2269
2270         if marshalled.has_key('depotFile') and self.stream_have_file_info:
2271             # start of a new file - output the old one first
2272             self.streamOneP4File(self.stream_file, self.stream_contents)
2273             self.stream_file = {}
2274             self.stream_contents = []
2275             self.stream_have_file_info = False
2276
2277         # pick up the new file information... for the
2278         # 'data' field we need to append to our array
2279         for k in marshalled.keys():
2280             if k == 'data':
2281                 self.stream_contents.append(marshalled['data'])
2282             else:
2283                 self.stream_file[k] = marshalled[k]
2284
2285         self.stream_have_file_info = True
2286
2287     # Stream directly from "p4 files" into "git fast-import"
2288     def streamP4Files(self, files):
2289         filesForCommit = []
2290         filesToRead = []
2291         filesToDelete = []
2292
2293         for f in files:
2294             # if using a client spec, only add the files that have
2295             # a path in the client
2296             if self.clientSpecDirs:
2297                 if self.clientSpecDirs.map_in_client(f['path']) == "":
2298                     continue
2299
2300             filesForCommit.append(f)
2301             if f['action'] in self.delete_actions:
2302                 filesToDelete.append(f)
2303             else:
2304                 filesToRead.append(f)
2305
2306         # deleted files...
2307         for f in filesToDelete:
2308             self.streamOneP4Deletion(f)
2309
2310         if len(filesToRead) > 0:
2311             self.stream_file = {}
2312             self.stream_contents = []
2313             self.stream_have_file_info = False
2314
2315             # curry self argument
2316             def streamP4FilesCbSelf(entry):
2317                 self.streamP4FilesCb(entry)
2318
2319             fileArgs = ['%s#%s' % (f['path'], f['rev']) for f in filesToRead]
2320
2321             p4CmdList(["-x", "-", "print"],
2322                       stdin=fileArgs,
2323                       cb=streamP4FilesCbSelf)
2324
2325             # do the last chunk
2326             if self.stream_file.has_key('depotFile'):
2327                 self.streamOneP4File(self.stream_file, self.stream_contents)
2328
2329     def make_email(self, userid):
2330         if userid in self.users:
2331             return self.users[userid]
2332         else:
2333             return "%s <a@b>" % userid
2334
2335     # Stream a p4 tag
2336     def streamTag(self, gitStream, labelName, labelDetails, commit, epoch):
2337         if verbose:
2338             print "writing tag %s for commit %s" % (labelName, commit)
2339         gitStream.write("tag %s\n" % labelName)
2340         gitStream.write("from %s\n" % commit)
2341
2342         if labelDetails.has_key('Owner'):
2343             owner = labelDetails["Owner"]
2344         else:
2345             owner = None
2346
2347         # Try to use the owner of the p4 label, or failing that,
2348         # the current p4 user id.
2349         if owner:
2350             email = self.make_email(owner)
2351         else:
2352             email = self.make_email(self.p4UserId())
2353         tagger = "%s %s %s" % (email, epoch, self.tz)
2354
2355         gitStream.write("tagger %s\n" % tagger)
2356
2357         print "labelDetails=",labelDetails
2358         if labelDetails.has_key('Description'):
2359             description = labelDetails['Description']
2360         else:
2361             description = 'Label from git p4'
2362
2363         gitStream.write("data %d\n" % len(description))
2364         gitStream.write(description)
2365         gitStream.write("\n")
2366
2367     def commit(self, details, files, branch, parent = ""):
2368         epoch = details["time"]
2369         author = details["user"]
2370
2371         if self.verbose:
2372             print "commit into %s" % branch
2373
2374         # start with reading files; if that fails, we should not
2375         # create a commit.
2376         new_files = []
2377         for f in files:
2378             if [p for p in self.branchPrefixes if p4PathStartsWith(f['path'], p)]:
2379                 new_files.append (f)
2380             else:
2381                 sys.stderr.write("Ignoring file outside of prefix: %s\n" % f['path'])
2382
2383         if self.clientSpecDirs:
2384             self.clientSpecDirs.update_client_spec_path_cache(files)
2385
2386         self.gitStream.write("commit %s\n" % branch)
2387 #        gitStream.write("mark :%s\n" % details["change"])
2388         self.committedChanges.add(int(details["change"]))
2389         committer = ""
2390         if author not in self.users:
2391             self.getUserMapFromPerforceServer()
2392         committer = "%s %s %s" % (self.make_email(author), epoch, self.tz)
2393
2394         self.gitStream.write("committer %s\n" % committer)
2395
2396         self.gitStream.write("data <<EOT\n")
2397         self.gitStream.write(details["desc"])
2398         self.gitStream.write("\n[git-p4: depot-paths = \"%s\": change = %s" %
2399                              (','.join(self.branchPrefixes), details["change"]))
2400         if len(details['options']) > 0:
2401             self.gitStream.write(": options = %s" % details['options'])
2402         self.gitStream.write("]\nEOT\n\n")
2403
2404         if len(parent) > 0:
2405             if self.verbose:
2406                 print "parent %s" % parent
2407             self.gitStream.write("from %s\n" % parent)
2408
2409         self.streamP4Files(new_files)
2410         self.gitStream.write("\n")
2411
2412         change = int(details["change"])
2413
2414         if self.labels.has_key(change):
2415             label = self.labels[change]
2416             labelDetails = label[0]
2417             labelRevisions = label[1]
2418             if self.verbose:
2419                 print "Change %s is labelled %s" % (change, labelDetails)
2420
2421             files = p4CmdList(["files"] + ["%s...@%s" % (p, change)
2422                                                 for p in self.branchPrefixes])
2423
2424             if len(files) == len(labelRevisions):
2425
2426                 cleanedFiles = {}
2427                 for info in files:
2428                     if info["action"] in self.delete_actions:
2429                         continue
2430                     cleanedFiles[info["depotFile"]] = info["rev"]
2431
2432                 if cleanedFiles == labelRevisions:
2433                     self.streamTag(self.gitStream, 'tag_%s' % labelDetails['label'], labelDetails, branch, epoch)
2434
2435                 else:
2436                     if not self.silent:
2437                         print ("Tag %s does not match with change %s: files do not match."
2438                                % (labelDetails["label"], change))
2439
2440             else:
2441                 if not self.silent:
2442                     print ("Tag %s does not match with change %s: file count is different."
2443                            % (labelDetails["label"], change))
2444
2445     # Build a dictionary of changelists and labels, for "detect-labels" option.
2446     def getLabels(self):
2447         self.labels = {}
2448
2449         l = p4CmdList(["labels"] + ["%s..." % p for p in self.depotPaths])
2450         if len(l) > 0 and not self.silent:
2451             print "Finding files belonging to labels in %s" % `self.depotPaths`
2452
2453         for output in l:
2454             label = output["label"]
2455             revisions = {}
2456             newestChange = 0
2457             if self.verbose:
2458                 print "Querying files for label %s" % label
2459             for file in p4CmdList(["files"] +
2460                                       ["%s...@%s" % (p, label)
2461                                           for p in self.depotPaths]):
2462                 revisions[file["depotFile"]] = file["rev"]
2463                 change = int(file["change"])
2464                 if change > newestChange:
2465                     newestChange = change
2466
2467             self.labels[newestChange] = [output, revisions]
2468
2469         if self.verbose:
2470             print "Label changes: %s" % self.labels.keys()
2471
2472     # Import p4 labels as git tags. A direct mapping does not
2473     # exist, so assume that if all the files are at the same revision
2474     # then we can use that, or it's something more complicated we should
2475     # just ignore.
2476     def importP4Labels(self, stream, p4Labels):
2477         if verbose:
2478             print "import p4 labels: " + ' '.join(p4Labels)
2479
2480         ignoredP4Labels = gitConfigList("git-p4.ignoredP4Labels")
2481         validLabelRegexp = gitConfig("git-p4.labelImportRegexp")
2482         if len(validLabelRegexp) == 0:
2483             validLabelRegexp = defaultLabelRegexp
2484         m = re.compile(validLabelRegexp)
2485
2486         for name in p4Labels:
2487             commitFound = False
2488
2489             if not m.match(name):
2490                 if verbose:
2491                     print "label %s does not match regexp %s" % (name,validLabelRegexp)
2492                 continue
2493
2494             if name in ignoredP4Labels:
2495                 continue
2496
2497             labelDetails = p4CmdList(['label', "-o", name])[0]
2498
2499             # get the most recent changelist for each file in this label
2500             change = p4Cmd(["changes", "-m", "1"] + ["%s...@%s" % (p, name)
2501                                 for p in self.depotPaths])
2502
2503             if change.has_key('change'):
2504                 # find the corresponding git commit; take the oldest commit
2505                 changelist = int(change['change'])
2506                 gitCommit = read_pipe(["git", "rev-list", "--max-count=1",
2507                      "--reverse", ":/\[git-p4:.*change = %d\]" % changelist])
2508                 if len(gitCommit) == 0:
2509                     print "could not find git commit for changelist %d" % changelist
2510                 else:
2511                     gitCommit = gitCommit.strip()
2512                     commitFound = True
2513                     # Convert from p4 time format
2514                     try:
2515                         tmwhen = time.strptime(labelDetails['Update'], "%Y/%m/%d %H:%M:%S")
2516                     except ValueError:
2517                         print "Could not convert label time %s" % labelDetails['Update']
2518                         tmwhen = 1
2519
2520                     when = int(time.mktime(tmwhen))
2521                     self.streamTag(stream, name, labelDetails, gitCommit, when)
2522                     if verbose:
2523                         print "p4 label %s mapped to git commit %s" % (name, gitCommit)
2524             else:
2525                 if verbose:
2526                     print "Label %s has no changelists - possibly deleted?" % name
2527
2528             if not commitFound:
2529                 # We can't import this label; don't try again as it will get very
2530                 # expensive repeatedly fetching all the files for labels that will
2531                 # never be imported. If the label is moved in the future, the
2532                 # ignore will need to be removed manually.
2533                 system(["git", "config", "--add", "git-p4.ignoredP4Labels", name])
2534
2535     def guessProjectName(self):
2536         for p in self.depotPaths:
2537             if p.endswith("/"):
2538                 p = p[:-1]
2539             p = p[p.strip().rfind("/") + 1:]
2540             if not p.endswith("/"):
2541                p += "/"
2542             return p
2543
2544     def getBranchMapping(self):
2545         lostAndFoundBranches = set()
2546
2547         user = gitConfig("git-p4.branchUser")
2548         if len(user) > 0:
2549             command = "branches -u %s" % user
2550         else:
2551             command = "branches"
2552
2553         for info in p4CmdList(command):
2554             details = p4Cmd(["branch", "-o", info["branch"]])
2555             viewIdx = 0
2556             while details.has_key("View%s" % viewIdx):
2557                 paths = details["View%s" % viewIdx].split(" ")
2558                 viewIdx = viewIdx + 1
2559                 # require standard //depot/foo/... //depot/bar/... mapping
2560                 if len(paths) != 2 or not paths[0].endswith("/...") or not paths[1].endswith("/..."):
2561                     continue
2562                 source = paths[0]
2563                 destination = paths[1]
2564                 ## HACK
2565                 if p4PathStartsWith(source, self.depotPaths[0]) and p4PathStartsWith(destination, self.depotPaths[0]):
2566                     source = source[len(self.depotPaths[0]):-4]
2567                     destination = destination[len(self.depotPaths[0]):-4]
2568
2569                     if destination in self.knownBranches:
2570                         if not self.silent:
2571                             print "p4 branch %s defines a mapping from %s to %s" % (info["branch"], source, destination)
2572                             print "but there exists another mapping from %s to %s already!" % (self.knownBranches[destination], destination)
2573                         continue
2574
2575                     self.knownBranches[destination] = source
2576
2577                     lostAndFoundBranches.discard(destination)
2578
2579                     if source not in self.knownBranches:
2580                         lostAndFoundBranches.add(source)
2581
2582         # Perforce does not strictly require branches to be defined, so we also
2583         # check git config for a branch list.
2584         #
2585         # Example of branch definition in git config file:
2586         # [git-p4]
2587         #   branchList=main:branchA
2588         #   branchList=main:branchB
2589         #   branchList=branchA:branchC
2590         configBranches = gitConfigList("git-p4.branchList")
2591         for branch in configBranches:
2592             if branch:
2593                 (source, destination) = branch.split(":")
2594                 self.knownBranches[destination] = source
2595
2596                 lostAndFoundBranches.discard(destination)
2597
2598                 if source not in self.knownBranches:
2599                     lostAndFoundBranches.add(source)
2600
2601
2602         for branch in lostAndFoundBranches:
2603             self.knownBranches[branch] = branch
2604
2605     def getBranchMappingFromGitBranches(self):
2606         branches = p4BranchesInGit(self.importIntoRemotes)
2607         for branch in branches.keys():
2608             if branch == "master":
2609                 branch = "main"
2610             else:
2611                 branch = branch[len(self.projectName):]
2612             self.knownBranches[branch] = branch
2613
2614     def updateOptionDict(self, d):
2615         option_keys = {}
2616         if self.keepRepoPath:
2617             option_keys['keepRepoPath'] = 1
2618
2619         d["options"] = ' '.join(sorted(option_keys.keys()))
2620
2621     def readOptions(self, d):
2622         self.keepRepoPath = (d.has_key('options')
2623                              and ('keepRepoPath' in d['options']))
2624
2625     def gitRefForBranch(self, branch):
2626         if branch == "main":
2627             return self.refPrefix + "master"
2628
2629         if len(branch) <= 0:
2630             return branch
2631
2632         return self.refPrefix + self.projectName + branch
2633
2634     def gitCommitByP4Change(self, ref, change):
2635         if self.verbose:
2636             print "looking in ref " + ref + " for change %s using bisect..." % change
2637
2638         earliestCommit = ""
2639         latestCommit = parseRevision(ref)
2640
2641         while True:
2642             if self.verbose:
2643                 print "trying: earliest %s latest %s" % (earliestCommit, latestCommit)
2644             next = read_pipe("git rev-list --bisect %s %s" % (latestCommit, earliestCommit)).strip()
2645             if len(next) == 0:
2646                 if self.verbose:
2647                     print "argh"
2648                 return ""
2649             log = extractLogMessageFromGitCommit(next)
2650             settings = extractSettingsGitLog(log)
2651             currentChange = int(settings['change'])
2652             if self.verbose:
2653                 print "current change %s" % currentChange
2654
2655             if currentChange == change:
2656                 if self.verbose:
2657                     print "found %s" % next
2658                 return next
2659
2660             if currentChange < change:
2661                 earliestCommit = "^%s" % next
2662             else:
2663                 latestCommit = "%s" % next
2664
2665         return ""
2666
2667     def importNewBranch(self, branch, maxChange):
2668         # make fast-import flush all changes to disk and update the refs using the checkpoint
2669         # command so that we can try to find the branch parent in the git history
2670         self.gitStream.write("checkpoint\n\n");
2671         self.gitStream.flush();
2672         branchPrefix = self.depotPaths[0] + branch + "/"
2673         range = "@1,%s" % maxChange
2674         #print "prefix" + branchPrefix
2675         changes = p4ChangesForPaths([branchPrefix], range, self.changes_block_size)
2676         if len(changes) <= 0:
2677             return False
2678         firstChange = changes[0]
2679         #print "first change in branch: %s" % firstChange
2680         sourceBranch = self.knownBranches[branch]
2681         sourceDepotPath = self.depotPaths[0] + sourceBranch
2682         sourceRef = self.gitRefForBranch(sourceBranch)
2683         #print "source " + sourceBranch
2684
2685         branchParentChange = int(p4Cmd(["changes", "-m", "1", "%s...@1,%s" % (sourceDepotPath, firstChange)])["change"])
2686         #print "branch parent: %s" % branchParentChange
2687         gitParent = self.gitCommitByP4Change(sourceRef, branchParentChange)
2688         if len(gitParent) > 0:
2689             self.initialParents[self.gitRefForBranch(branch)] = gitParent
2690             #print "parent git commit: %s" % gitParent
2691
2692         self.importChanges(changes)
2693         return True
2694
2695     def searchParent(self, parent, branch, target):
2696         parentFound = False
2697         for blob in read_pipe_lines(["git", "rev-list", "--reverse",
2698                                      "--no-merges", parent]):
2699             blob = blob.strip()
2700             if len(read_pipe(["git", "diff-tree", blob, target])) == 0:
2701                 parentFound = True
2702                 if self.verbose:
2703                     print "Found parent of %s in commit %s" % (branch, blob)
2704                 break
2705         if parentFound:
2706             return blob
2707         else:
2708             return None
2709
2710     def importChanges(self, changes):
2711         cnt = 1
2712         for change in changes:
2713             description = p4_describe(change)
2714             self.updateOptionDict(description)
2715
2716             if not self.silent:
2717                 sys.stdout.write("\rImporting revision %s (%s%%)" % (change, cnt * 100 / len(changes)))
2718                 sys.stdout.flush()
2719             cnt = cnt + 1
2720
2721             try:
2722                 if self.detectBranches:
2723                     branches = self.splitFilesIntoBranches(description)
2724                     for branch in branches.keys():
2725                         ## HACK  --hwn
2726                         branchPrefix = self.depotPaths[0] + branch + "/"
2727                         self.branchPrefixes = [ branchPrefix ]
2728
2729                         parent = ""
2730
2731                         filesForCommit = branches[branch]
2732
2733                         if self.verbose:
2734                             print "branch is %s" % branch
2735
2736                         self.updatedBranches.add(branch)
2737
2738                         if branch not in self.createdBranches:
2739                             self.createdBranches.add(branch)
2740                             parent = self.knownBranches[branch]
2741                             if parent == branch:
2742                                 parent = ""
2743                             else:
2744                                 fullBranch = self.projectName + branch
2745                                 if fullBranch not in self.p4BranchesInGit:
2746                                     if not self.silent:
2747                                         print("\n    Importing new branch %s" % fullBranch);
2748                                     if self.importNewBranch(branch, change - 1):
2749                                         parent = ""
2750                                         self.p4BranchesInGit.append(fullBranch)
2751                                     if not self.silent:
2752                                         print("\n    Resuming with change %s" % change);
2753
2754                                 if self.verbose:
2755                                     print "parent determined through known branches: %s" % parent
2756
2757                         branch = self.gitRefForBranch(branch)
2758                         parent = self.gitRefForBranch(parent)
2759
2760                         if self.verbose:
2761                             print "looking for initial parent for %s; current parent is %s" % (branch, parent)
2762
2763                         if len(parent) == 0 and branch in self.initialParents:
2764                             parent = self.initialParents[branch]
2765                             del self.initialParents[branch]
2766
2767                         blob = None
2768                         if len(parent) > 0:
2769                             tempBranch = "%s/%d" % (self.tempBranchLocation, change)
2770                             if self.verbose:
2771                                 print "Creating temporary branch: " + tempBranch
2772                             self.commit(description, filesForCommit, tempBranch)
2773                             self.tempBranches.append(tempBranch)
2774                             self.checkpoint()
2775                             blob = self.searchParent(parent, branch, tempBranch)
2776                         if blob:
2777                             self.commit(description, filesForCommit, branch, blob)
2778                         else:
2779                             if self.verbose:
2780                                 print "Parent of %s not found. Committing into head of %s" % (branch, parent)
2781                             self.commit(description, filesForCommit, branch, parent)
2782                 else:
2783                     files = self.extractFilesFromCommit(description)
2784                     self.commit(description, files, self.branch,
2785                                 self.initialParent)
2786                     # only needed once, to connect to the previous commit
2787                     self.initialParent = ""
2788             except IOError:
2789                 print self.gitError.read()
2790                 sys.exit(1)
2791
2792     def importHeadRevision(self, revision):
2793         print "Doing initial import of %s from revision %s into %s" % (' '.join(self.depotPaths), revision, self.branch)
2794
2795         details = {}
2796         details["user"] = "git perforce import user"
2797         details["desc"] = ("Initial import of %s from the state at revision %s\n"
2798                            % (' '.join(self.depotPaths), revision))
2799         details["change"] = revision
2800         newestRevision = 0
2801
2802         fileCnt = 0
2803         fileArgs = ["%s...%s" % (p,revision) for p in self.depotPaths]
2804
2805         for info in p4CmdList(["files"] + fileArgs):
2806
2807             if 'code' in info and info['code'] == 'error':
2808                 sys.stderr.write("p4 returned an error: %s\n"
2809                                  % info['data'])
2810                 if info['data'].find("must refer to client") >= 0:
2811                     sys.stderr.write("This particular p4 error is misleading.\n")
2812                     sys.stderr.write("Perhaps the depot path was misspelled.\n");
2813                     sys.stderr.write("Depot path:  %s\n" % " ".join(self.depotPaths))
2814                 sys.exit(1)
2815             if 'p4ExitCode' in info:
2816                 sys.stderr.write("p4 exitcode: %s\n" % info['p4ExitCode'])
2817                 sys.exit(1)
2818
2819
2820             change = int(info["change"])
2821             if change > newestRevision:
2822                 newestRevision = change
2823
2824             if info["action"] in self.delete_actions:
2825                 # don't increase the file cnt, otherwise details["depotFile123"] will have gaps!
2826                 #fileCnt = fileCnt + 1
2827                 continue
2828
2829             for prop in ["depotFile", "rev", "action", "type" ]:
2830                 details["%s%s" % (prop, fileCnt)] = info[prop]
2831
2832             fileCnt = fileCnt + 1
2833
2834         details["change"] = newestRevision
2835
2836         # Use time from top-most change so that all git p4 clones of
2837         # the same p4 repo have the same commit SHA1s.
2838         res = p4_describe(newestRevision)
2839         details["time"] = res["time"]
2840
2841         self.updateOptionDict(details)
2842         try:
2843             self.commit(details, self.extractFilesFromCommit(details), self.branch)
2844         except IOError:
2845             print "IO error with git fast-import. Is your git version recent enough?"
2846             print self.gitError.read()
2847
2848
2849     def run(self, args):
2850         self.depotPaths = []
2851         self.changeRange = ""
2852         self.previousDepotPaths = []
2853         self.hasOrigin = False
2854
2855         # map from branch depot path to parent branch
2856         self.knownBranches = {}
2857         self.initialParents = {}
2858
2859         if self.importIntoRemotes:
2860             self.refPrefix = "refs/remotes/p4/"
2861         else:
2862             self.refPrefix = "refs/heads/p4/"
2863
2864         if self.syncWithOrigin:
2865             self.hasOrigin = originP4BranchesExist()
2866             if self.hasOrigin:
2867                 if not self.silent:
2868                     print 'Syncing with origin first, using "git fetch origin"'
2869                 system("git fetch origin")
2870
2871         branch_arg_given = bool(self.branch)
2872         if len(self.branch) == 0:
2873             self.branch = self.refPrefix + "master"
2874             if gitBranchExists("refs/heads/p4") and self.importIntoRemotes:
2875                 system("git update-ref %s refs/heads/p4" % self.branch)
2876                 system("git branch -D p4")
2877
2878         # accept either the command-line option, or the configuration variable
2879         if self.useClientSpec:
2880             # will use this after clone to set the variable
2881             self.useClientSpec_from_options = True
2882         else:
2883             if gitConfigBool("git-p4.useclientspec"):
2884                 self.useClientSpec = True
2885         if self.useClientSpec:
2886             self.clientSpecDirs = getClientSpec()
2887
2888         # TODO: should always look at previous commits,
2889         # merge with previous imports, if possible.
2890         if args == []:
2891             if self.hasOrigin:
2892                 createOrUpdateBranchesFromOrigin(self.refPrefix, self.silent)
2893
2894             # branches holds mapping from branch name to sha1
2895             branches = p4BranchesInGit(self.importIntoRemotes)
2896
2897             # restrict to just this one, disabling detect-branches
2898             if branch_arg_given:
2899                 short = self.branch.split("/")[-1]
2900                 if short in branches:
2901                     self.p4BranchesInGit = [ short ]
2902             else:
2903                 self.p4BranchesInGit = branches.keys()
2904
2905             if len(self.p4BranchesInGit) > 1:
2906                 if not self.silent:
2907                     print "Importing from/into multiple branches"
2908                 self.detectBranches = True
2909                 for branch in branches.keys():
2910                     self.initialParents[self.refPrefix + branch] = \
2911                         branches[branch]
2912
2913             if self.verbose:
2914                 print "branches: %s" % self.p4BranchesInGit
2915
2916             p4Change = 0
2917             for branch in self.p4BranchesInGit:
2918                 logMsg =  extractLogMessageFromGitCommit(self.refPrefix + branch)
2919
2920                 settings = extractSettingsGitLog(logMsg)
2921
2922                 self.readOptions(settings)
2923                 if (settings.has_key('depot-paths')
2924                     and settings.has_key ('change')):
2925                     change = int(settings['change']) + 1
2926                     p4Change = max(p4Change, change)
2927
2928                     depotPaths = sorted(settings['depot-paths'])
2929                     if self.previousDepotPaths == []:
2930                         self.previousDepotPaths = depotPaths
2931                     else:
2932                         paths = []
2933                         for (prev, cur) in zip(self.previousDepotPaths, depotPaths):
2934                             prev_list = prev.split("/")
2935                             cur_list = cur.split("/")
2936                             for i in range(0, min(len(cur_list), len(prev_list))):
2937                                 if cur_list[i] <> prev_list[i]:
2938                                     i = i - 1
2939                                     break
2940
2941                             paths.append ("/".join(cur_list[:i + 1]))
2942
2943                         self.previousDepotPaths = paths
2944
2945             if p4Change > 0:
2946                 self.depotPaths = sorted(self.previousDepotPaths)
2947                 self.changeRange = "@%s,#head" % p4Change
2948                 if not self.silent and not self.detectBranches:
2949                     print "Performing incremental import into %s git branch" % self.branch
2950
2951         # accept multiple ref name abbreviations:
2952         #    refs/foo/bar/branch -> use it exactly
2953         #    p4/branch -> prepend refs/remotes/ or refs/heads/
2954         #    branch -> prepend refs/remotes/p4/ or refs/heads/p4/
2955         if not self.branch.startswith("refs/"):
2956             if self.importIntoRemotes:
2957                 prepend = "refs/remotes/"
2958             else:
2959                 prepend = "refs/heads/"
2960             if not self.branch.startswith("p4/"):
2961                 prepend += "p4/"
2962             self.branch = prepend + self.branch
2963
2964         if len(args) == 0 and self.depotPaths:
2965             if not self.silent:
2966                 print "Depot paths: %s" % ' '.join(self.depotPaths)
2967         else:
2968             if self.depotPaths and self.depotPaths != args:
2969                 print ("previous import used depot path %s and now %s was specified. "
2970                        "This doesn't work!" % (' '.join (self.depotPaths),
2971                                                ' '.join (args)))
2972                 sys.exit(1)
2973
2974             self.depotPaths = sorted(args)
2975
2976         revision = ""
2977         self.users = {}
2978
2979         # Make sure no revision specifiers are used when --changesfile
2980         # is specified.
2981         bad_changesfile = False
2982         if len(self.changesFile) > 0:
2983             for p in self.depotPaths:
2984                 if p.find("@") >= 0 or p.find("#") >= 0:
2985                     bad_changesfile = True
2986                     break
2987         if bad_changesfile:
2988             die("Option --changesfile is incompatible with revision specifiers")
2989
2990         newPaths = []
2991         for p in self.depotPaths:
2992             if p.find("@") != -1:
2993                 atIdx = p.index("@")
2994                 self.changeRange = p[atIdx:]
2995                 if self.changeRange == "@all":
2996                     self.changeRange = ""
2997                 elif ',' not in self.changeRange:
2998                     revision = self.changeRange
2999                     self.changeRange = ""
3000                 p = p[:atIdx]
3001             elif p.find("#") != -1:
3002                 hashIdx = p.index("#")
3003                 revision = p[hashIdx:]
3004                 p = p[:hashIdx]
3005             elif self.previousDepotPaths == []:
3006                 # pay attention to changesfile, if given, else import
3007                 # the entire p4 tree at the head revision
3008                 if len(self.changesFile) == 0:
3009                     revision = "#head"
3010
3011             p = re.sub ("\.\.\.$", "", p)
3012             if not p.endswith("/"):
3013                 p += "/"
3014
3015             newPaths.append(p)
3016
3017         self.depotPaths = newPaths
3018
3019         # --detect-branches may change this for each branch
3020         self.branchPrefixes = self.depotPaths
3021
3022         self.loadUserMapFromCache()
3023         self.labels = {}
3024         if self.detectLabels:
3025             self.getLabels();
3026
3027         if self.detectBranches:
3028             ## FIXME - what's a P4 projectName ?
3029             self.projectName = self.guessProjectName()
3030
3031             if self.hasOrigin:
3032                 self.getBranchMappingFromGitBranches()
3033             else:
3034                 self.getBranchMapping()
3035             if self.verbose:
3036                 print "p4-git branches: %s" % self.p4BranchesInGit
3037                 print "initial parents: %s" % self.initialParents
3038             for b in self.p4BranchesInGit:
3039                 if b != "master":
3040
3041                     ## FIXME
3042                     b = b[len(self.projectName):]
3043                 self.createdBranches.add(b)
3044
3045         self.tz = "%+03d%02d" % (- time.timezone / 3600, ((- time.timezone % 3600) / 60))
3046
3047         self.importProcess = subprocess.Popen(["git", "fast-import"],
3048                                               stdin=subprocess.PIPE,
3049                                               stdout=subprocess.PIPE,
3050                                               stderr=subprocess.PIPE);
3051         self.gitOutput = self.importProcess.stdout
3052         self.gitStream = self.importProcess.stdin
3053         self.gitError = self.importProcess.stderr
3054
3055         if revision:
3056             self.importHeadRevision(revision)
3057         else:
3058             changes = []
3059
3060             if len(self.changesFile) > 0:
3061                 output = open(self.changesFile).readlines()
3062                 changeSet = set()
3063                 for line in output:
3064                     changeSet.add(int(line))
3065
3066                 for change in changeSet:
3067                     changes.append(change)
3068
3069                 changes.sort()
3070             else:
3071                 # catch "git p4 sync" with no new branches, in a repo that
3072                 # does not have any existing p4 branches
3073                 if len(args) == 0:
3074                     if not self.p4BranchesInGit:
3075                         die("No remote p4 branches.  Perhaps you never did \"git p4 clone\" in here.")
3076
3077                     # The default branch is master, unless --branch is used to
3078                     # specify something else.  Make sure it exists, or complain
3079                     # nicely about how to use --branch.
3080                     if not self.detectBranches:
3081                         if not branch_exists(self.branch):
3082                             if branch_arg_given:
3083                                 die("Error: branch %s does not exist." % self.branch)
3084                             else:
3085                                 die("Error: no branch %s; perhaps specify one with --branch." %
3086                                     self.branch)
3087
3088                 if self.verbose:
3089                     print "Getting p4 changes for %s...%s" % (', '.join(self.depotPaths),
3090                                                               self.changeRange)
3091                 changes = p4ChangesForPaths(self.depotPaths, self.changeRange, self.changes_block_size)
3092
3093                 if len(self.maxChanges) > 0:
3094                     changes = changes[:min(int(self.maxChanges), len(changes))]
3095
3096             if len(changes) == 0:
3097                 if not self.silent:
3098                     print "No changes to import!"
3099             else:
3100                 if not self.silent and not self.detectBranches:
3101                     print "Import destination: %s" % self.branch
3102
3103                 self.updatedBranches = set()
3104
3105                 if not self.detectBranches:
3106                     if args:
3107                         # start a new branch
3108                         self.initialParent = ""
3109                     else:
3110                         # build on a previous revision
3111                         self.initialParent = parseRevision(self.branch)
3112
3113                 self.importChanges(changes)
3114
3115                 if not self.silent:
3116                     print ""
3117                     if len(self.updatedBranches) > 0:
3118                         sys.stdout.write("Updated branches: ")
3119                         for b in self.updatedBranches:
3120                             sys.stdout.write("%s " % b)
3121                         sys.stdout.write("\n")
3122
3123         if gitConfigBool("git-p4.importLabels"):
3124             self.importLabels = True
3125
3126         if self.importLabels:
3127             p4Labels = getP4Labels(self.depotPaths)
3128             gitTags = getGitTags()
3129
3130             missingP4Labels = p4Labels - gitTags
3131             self.importP4Labels(self.gitStream, missingP4Labels)
3132
3133         self.gitStream.close()
3134         if self.importProcess.wait() != 0:
3135             die("fast-import failed: %s" % self.gitError.read())
3136         self.gitOutput.close()
3137         self.gitError.close()
3138
3139         # Cleanup temporary branches created during import
3140         if self.tempBranches != []:
3141             for branch in self.tempBranches:
3142                 read_pipe("git update-ref -d %s" % branch)
3143             os.rmdir(os.path.join(os.environ.get("GIT_DIR", ".git"), self.tempBranchLocation))
3144
3145         # Create a symbolic ref p4/HEAD pointing to p4/<branch> to allow
3146         # a convenient shortcut refname "p4".
3147         if self.importIntoRemotes:
3148             head_ref = self.refPrefix + "HEAD"
3149             if not gitBranchExists(head_ref) and gitBranchExists(self.branch):
3150                 system(["git", "symbolic-ref", head_ref, self.branch])
3151
3152         return True
3153
3154 class P4Rebase(Command):
3155     def __init__(self):
3156         Command.__init__(self)
3157         self.options = [
3158                 optparse.make_option("--import-labels", dest="importLabels", action="store_true"),
3159         ]
3160         self.importLabels = False
3161         self.description = ("Fetches the latest revision from perforce and "
3162                             + "rebases the current work (branch) against it")
3163
3164     def run(self, args):
3165         sync = P4Sync()
3166         sync.importLabels = self.importLabels
3167         sync.run([])
3168
3169         return self.rebase()
3170
3171     def rebase(self):
3172         if os.system("git update-index --refresh") != 0:
3173             die("Some files in your working directory are modified and different than what is in your index. You can use git update-index <filename> to bring the index up-to-date or stash away all your changes with git stash.");
3174         if len(read_pipe("git diff-index HEAD --")) > 0:
3175             die("You have uncommitted changes. Please commit them before rebasing or stash them away with git stash.");
3176
3177         [upstream, settings] = findUpstreamBranchPoint()
3178         if len(upstream) == 0:
3179             die("Cannot find upstream branchpoint for rebase")
3180
3181         # the branchpoint may be p4/foo~3, so strip off the parent
3182         upstream = re.sub("~[0-9]+$", "", upstream)
3183
3184         print "Rebasing the current branch onto %s" % upstream
3185         oldHead = read_pipe("git rev-parse HEAD").strip()
3186         system("git rebase %s" % upstream)
3187         system("git diff-tree --stat --summary -M %s HEAD --" % oldHead)
3188         return True
3189
3190 class P4Clone(P4Sync):
3191     def __init__(self):
3192         P4Sync.__init__(self)
3193         self.description = "Creates a new git repository and imports from Perforce into it"
3194         self.usage = "usage: %prog [options] //depot/path[@revRange]"
3195         self.options += [
3196             optparse.make_option("--destination", dest="cloneDestination",
3197                                  action='store', default=None,
3198                                  help="where to leave result of the clone"),
3199             optparse.make_option("--bare", dest="cloneBare",
3200                                  action="store_true", default=False),
3201         ]
3202         self.cloneDestination = None
3203         self.needsGit = False
3204         self.cloneBare = False
3205
3206     def defaultDestination(self, args):
3207         ## TODO: use common prefix of args?
3208         depotPath = args[0]
3209         depotDir = re.sub("(@[^@]*)$", "", depotPath)
3210         depotDir = re.sub("(#[^#]*)$", "", depotDir)
3211         depotDir = re.sub(r"\.\.\.$", "", depotDir)
3212         depotDir = re.sub(r"/$", "", depotDir)
3213         return os.path.split(depotDir)[1]
3214
3215     def run(self, args):
3216         if len(args) < 1:
3217             return False
3218
3219         if self.keepRepoPath and not self.cloneDestination:
3220             sys.stderr.write("Must specify destination for --keep-path\n")
3221             sys.exit(1)
3222
3223         depotPaths = args
3224
3225         if not self.cloneDestination and len(depotPaths) > 1:
3226             self.cloneDestination = depotPaths[-1]
3227             depotPaths = depotPaths[:-1]
3228
3229         self.cloneExclude = ["/"+p for p in self.cloneExclude]
3230         for p in depotPaths:
3231             if not p.startswith("//"):
3232                 sys.stderr.write('Depot paths must start with "//": %s\n' % p)
3233                 return False
3234
3235         if not self.cloneDestination:
3236             self.cloneDestination = self.defaultDestination(args)
3237
3238         print "Importing from %s into %s" % (', '.join(depotPaths), self.cloneDestination)
3239
3240         if not os.path.exists(self.cloneDestination):
3241             os.makedirs(self.cloneDestination)
3242         chdir(self.cloneDestination)
3243
3244         init_cmd = [ "git", "init" ]
3245         if self.cloneBare:
3246             init_cmd.append("--bare")
3247         retcode = subprocess.call(init_cmd)
3248         if retcode:
3249             raise CalledProcessError(retcode, init_cmd)
3250
3251         if not P4Sync.run(self, depotPaths):
3252             return False
3253
3254         # create a master branch and check out a work tree
3255         if gitBranchExists(self.branch):
3256             system([ "git", "branch", "master", self.branch ])
3257             if not self.cloneBare:
3258                 system([ "git", "checkout", "-f" ])
3259         else:
3260             print 'Not checking out any branch, use ' \
3261                   '"git checkout -q -b master <branch>"'
3262
3263         # auto-set this variable if invoked with --use-client-spec
3264         if self.useClientSpec_from_options:
3265             system("git config --bool git-p4.useclientspec true")
3266
3267         return True
3268
3269 class P4Branches(Command):
3270     def __init__(self):
3271         Command.__init__(self)
3272         self.options = [ ]
3273         self.description = ("Shows the git branches that hold imports and their "
3274                             + "corresponding perforce depot paths")
3275         self.verbose = False
3276
3277     def run(self, args):
3278         if originP4BranchesExist():
3279             createOrUpdateBranchesFromOrigin()
3280
3281         cmdline = "git rev-parse --symbolic "
3282         cmdline += " --remotes"
3283
3284         for line in read_pipe_lines(cmdline):
3285             line = line.strip()
3286
3287             if not line.startswith('p4/') or line == "p4/HEAD":
3288                 continue
3289             branch = line
3290
3291             log = extractLogMessageFromGitCommit("refs/remotes/%s" % branch)
3292             settings = extractSettingsGitLog(log)
3293
3294             print "%s <= %s (%s)" % (branch, ",".join(settings["depot-paths"]), settings["change"])
3295         return True
3296
3297 class HelpFormatter(optparse.IndentedHelpFormatter):
3298     def __init__(self):
3299         optparse.IndentedHelpFormatter.__init__(self)
3300
3301     def format_description(self, description):
3302         if description:
3303             return description + "\n"
3304         else:
3305             return ""
3306
3307 def printUsage(commands):
3308     print "usage: %s <command> [options]" % sys.argv[0]
3309     print ""
3310     print "valid commands: %s" % ", ".join(commands)
3311     print ""
3312     print "Try %s <command> --help for command specific help." % sys.argv[0]
3313     print ""
3314
3315 commands = {
3316     "debug" : P4Debug,
3317     "submit" : P4Submit,
3318     "commit" : P4Submit,
3319     "sync" : P4Sync,
3320     "rebase" : P4Rebase,
3321     "clone" : P4Clone,
3322     "rollback" : P4RollBack,
3323     "branches" : P4Branches
3324 }
3325
3326
3327 def main():
3328     if len(sys.argv[1:]) == 0:
3329         printUsage(commands.keys())
3330         sys.exit(2)
3331
3332     cmdName = sys.argv[1]
3333     try:
3334         klass = commands[cmdName]
3335         cmd = klass()
3336     except KeyError:
3337         print "unknown command %s" % cmdName
3338         print ""
3339         printUsage(commands.keys())
3340         sys.exit(2)
3341
3342     options = cmd.options
3343     cmd.gitdir = os.environ.get("GIT_DIR", None)
3344
3345     args = sys.argv[2:]
3346
3347     options.append(optparse.make_option("--verbose", "-v", dest="verbose", action="store_true"))
3348     if cmd.needsGit:
3349         options.append(optparse.make_option("--git-dir", dest="gitdir"))
3350
3351     parser = optparse.OptionParser(cmd.usage.replace("%prog", "%prog " + cmdName),
3352                                    options,
3353                                    description = cmd.description,
3354                                    formatter = HelpFormatter())
3355
3356     (cmd, args) = parser.parse_args(sys.argv[2:], cmd);
3357     global verbose
3358     verbose = cmd.verbose
3359     if cmd.needsGit:
3360         if cmd.gitdir == None:
3361             cmd.gitdir = os.path.abspath(".git")
3362             if not isValidGitDir(cmd.gitdir):
3363                 cmd.gitdir = read_pipe("git rev-parse --git-dir").strip()
3364                 if os.path.exists(cmd.gitdir):
3365                     cdup = read_pipe("git rev-parse --show-cdup").strip()
3366                     if len(cdup) > 0:
3367                         chdir(cdup);
3368
3369         if not isValidGitDir(cmd.gitdir):
3370             if isValidGitDir(cmd.gitdir + "/.git"):
3371                 cmd.gitdir += "/.git"
3372             else:
3373                 die("fatal: cannot locate git repository at %s" % cmd.gitdir)
3374
3375         os.environ["GIT_DIR"] = cmd.gitdir
3376
3377     if not cmd.run(args):
3378         parser.print_help()
3379         sys.exit(2)
3380
3381
3382 if __name__ == '__main__':
3383     main()