cola/git.py

   1 from __future__ import absolute_import, division, print_function, unicode_literals
   2 from functools import partial
   3 import errno
   4 import os
   5 from os.path import join
   6 import subprocess
   7 import threading
   8 import time
   9
  10 from . import core
  11 from .compat import int_types
  12 from .compat import ustr
  13 from .compat import WIN32
  14 from .decorators import memoize
  15 from .interaction import Interaction
  16
  17
  18 GIT_COLA_TRACE = core.getenv('GIT_COLA_TRACE', '')
  19 GIT = core.getenv('GIT_COLA_GIT', 'git')
  20 STATUS = 0
  21 STDOUT = 1
  22 STDERR = 2
  23
  24 # Object ID / SHA1-related constants
  25 # Git's empty tree is a built-in constant object name.
  26 EMPTY_TREE_OID = '4b825dc642cb6eb9a060e54bf8d69288fbee4904'
  27 # Git's diff machinery returns zeroes for modified files whose content exists
  28 # in the worktree only.
  29 MISSING_BLOB_OID = '0000000000000000000000000000000000000000'
  30 # Git's SHA-1 object IDs are 40 characters  long.
  31 # This will need to change when Git moves away from SHA-1.
  32 # When that happens we'll have to detect and update this at runtime in
  33 # order to support both old and new git.
  34 OID_LENGTH = 40
  35
  36 _index_lock = threading.Lock()
  37
  38
  39 def dashify(s):
  40     return s.replace('_', '-')
  41
  42
  43 def is_git_dir(git_dir):
  44     """From git's setup.c:is_git_directory()."""
  45     result = False
  46     if git_dir:
  47         headref = join(git_dir, 'HEAD')
  48
  49         if (
  50             core.isdir(git_dir)
  51             and (
  52                 core.isdir(join(git_dir, 'objects'))
  53                 and core.isdir(join(git_dir, 'refs'))
  54             )
  55             or (
  56                 core.isfile(join(git_dir, 'gitdir'))
  57                 and core.isfile(join(git_dir, 'commondir'))
  58             )
  59         ):
  60
  61             result = core.isfile(headref) or (
  62                 core.islink(headref) and core.readlink(headref).startswith('refs/')
  63             )
  64         else:
  65             result = is_git_file(git_dir)
  66
  67     return result
  68
  69
  70 def is_git_file(f):
  71     return core.isfile(f) and os.path.basename(f) == '.git'
  72
  73
  74 def is_git_worktree(d):
  75     return is_git_dir(join(d, '.git'))
  76
  77
  78 def is_git_repository(path):
  79     return is_git_worktree(path) or is_git_dir(path)
  80
  81
  82 def read_git_file(path):
  83     """Read the path from a .git-file
  84
  85     `None` is returned when <path> is not a .git-file.
  86
  87     """
  88     result = None
  89     if path and is_git_file(path):
  90         header = 'gitdir: '
  91         data = core.read(path).strip()
  92         if data.startswith(header):
  93             result = data[len(header) :]
  94             if result and not os.path.isabs(result):
  95                 path_folder = os.path.dirname(path)
  96                 repo_relative = join(path_folder, result)
  97                 result = os.path.normpath(repo_relative)
  98     return result
  99
 100
 101 class Paths(object):
 102     """Git repository paths of interest"""
 103
 104     def __init__(self, git_dir=None, git_file=None, worktree=None, common_dir=None):
 105         if git_dir and not is_git_dir(git_dir):
 106             git_dir = None
 107         self.git_dir = git_dir
 108         self.git_file = git_file
 109         self.worktree = worktree
 110         self.common_dir = common_dir
 111
 112     def get(self, path):
 113         ceiling_dirs = set()
 114         ceiling = core.getenv('GIT_CEILING_DIRECTORIES')
 115         if ceiling:
 116             ceiling_dirs.update([x for x in ceiling.split(':') if x])
 117
 118         if path:
 119             path = core.abspath(path)
 120
 121         if not self.git_dir or not self.worktree:
 122             # Search for a .git directory
 123             while path:
 124                 if path in ceiling_dirs:
 125                     break
 126                 if is_git_dir(path):
 127                     if not self.git_dir:
 128                         self.git_dir = path
 129                     basename = os.path.basename(path)
 130                     if not self.worktree and basename == '.git':
 131                         self.worktree = os.path.dirname(path)
 132                 # We are either in a bare repository, or someone set GIT_DIR
 133                 # but did not set GIT_WORK_TREE.
 134                 if self.git_dir:
 135                     if not self.worktree:
 136                         basename = os.path.basename(self.git_dir)
 137                         if basename == '.git':
 138                             self.worktree = os.path.dirname(self.git_dir)
 139                         elif path and not is_git_dir(path):
 140                             self.worktree = path
 141                     break
 142                 gitpath = join(path, '.git')
 143                 if is_git_dir(gitpath):
 144                     if not self.git_dir:
 145                         self.git_dir = gitpath
 146                     if not self.worktree:
 147                         self.worktree = path
 148                     break
 149                 path, dummy = os.path.split(path)
 150                 if not dummy:
 151                     break
 152
 153         if self.git_dir:
 154             git_dir_path = read_git_file(self.git_dir)
 155             if git_dir_path:
 156                 self.git_file = self.git_dir
 157                 self.git_dir = git_dir_path
 158
 159                 commondir_file = join(git_dir_path, 'commondir')
 160                 if core.exists(commondir_file):
 161                     common_path = core.read(commondir_file).strip()
 162                     if common_path:
 163                         if os.path.isabs(common_path):
 164                             common_dir = common_path
 165                         else:
 166                             common_dir = join(git_dir_path, common_path)
 167                             common_dir = os.path.normpath(common_dir)
 168                         self.common_dir = common_dir
 169         # usage: Paths().get()
 170         return self
 171
 172
 173 def find_git_directory(path):
 174     """Perform Git repository discovery"""
 175     return Paths(
 176         git_dir=core.getenv('GIT_DIR'), worktree=core.getenv('GIT_WORK_TREE')
 177     ).get(path)
 178
 179
 180 class Git(object):
 181     """
 182     The Git class manages communication with the Git binary
 183     """
 184
 185     def __init__(self):
 186         self.paths = Paths()
 187
 188         self._valid = {}  #: Store the result of is_git_dir() for performance
 189         self.set_worktree(core.getcwd())
 190
 191     # pylint: disable=no-self-use
 192     def is_git_repository(self, path):
 193         return is_git_repository(path)
 194
 195     def getcwd(self):
 196         """Return the working directory used by git()"""
 197         return self.paths.worktree or self.paths.git_dir
 198
 199     def set_worktree(self, path):
 200         path = core.decode(path)
 201         self.paths = find_git_directory(path)
 202         return self.paths.worktree
 203
 204     def worktree(self):
 205         if not self.paths.worktree:
 206             path = core.abspath(core.getcwd())
 207             self.paths = find_git_directory(path)
 208         return self.paths.worktree
 209
 210     def is_valid(self):
 211         """Is this a valid git repository?
 212
 213         Cache the result to avoid hitting the filesystem.
 214
 215         """
 216         git_dir = self.paths.git_dir
 217         try:
 218             valid = bool(git_dir) and self._valid[git_dir]
 219         except KeyError:
 220             valid = self._valid[git_dir] = is_git_dir(git_dir)
 221
 222         return valid
 223
 224     def git_path(self, *paths):
 225         result = None
 226         if self.paths.git_dir:
 227             result = join(self.paths.git_dir, *paths)
 228         if result and self.paths.common_dir and not core.exists(result):
 229             common_result = join(self.paths.common_dir, *paths)
 230             if core.exists(common_result):
 231                 result = common_result
 232         return result
 233
 234     def git_dir(self):
 235         if not self.paths.git_dir:
 236             path = core.abspath(core.getcwd())
 237             self.paths = find_git_directory(path)
 238         return self.paths.git_dir
 239
 240     def __getattr__(self, name):
 241         git_cmd = partial(self.git, name)
 242         setattr(self, name, git_cmd)
 243         return git_cmd
 244
 245     @staticmethod
 246     def execute(
 247         command,
 248         _cwd=None,
 249         _decode=True,
 250         _encoding=None,
 251         _raw=False,
 252         _stdin=None,
 253         _stderr=subprocess.PIPE,
 254         _stdout=subprocess.PIPE,
 255         _readonly=False,
 256         _no_win32_startupinfo=False,
 257     ):
 258         """
 259         Execute a command and returns its output
 260
 261         :param command: argument list to execute.
 262         :param _cwd: working directory, defaults to the current directory.
 263         :param _decode: whether to decode output, defaults to True.
 264         :param _encoding: default encoding, defaults to None (utf-8).
 265         :param _raw: do not strip trailing whitespace.
 266         :param _stdin: optional stdin filehandle.
 267         :returns (status, out, err): exit status, stdout, stderr
 268
 269         """
 270         # Allow the user to have the command executed in their working dir.
 271         if not _cwd:
 272             _cwd = core.getcwd()
 273
 274         extra = {}
 275
 276         if hasattr(os, 'setsid'):
 277             # SSH uses the SSH_ASKPASS variable only if the process is really
 278             # detached from the TTY (stdin redirection and setting the
 279             # SSH_ASKPASS environment variable is not enough).  To detach a
 280             # process from the console it should fork and call os.setsid().
 281             extra['preexec_fn'] = os.setsid
 282
 283         start_time = time.time()
 284
 285         # Start the process
 286         # Guard against thread-unsafe .git/index.lock files
 287         if not _readonly:
 288             _index_lock.acquire()
 289         try:
 290             status, out, err = core.run_command(
 291                 command,
 292                 cwd=_cwd,
 293                 encoding=_encoding,
 294                 stdin=_stdin,
 295                 stdout=_stdout,
 296                 stderr=_stderr,
 297                 no_win32_startupinfo=_no_win32_startupinfo,
 298                 **extra
 299             )
 300         finally:
 301             # Let the next thread in
 302             if not _readonly:
 303                 _index_lock.release()
 304
 305         end_time = time.time()
 306         elapsed_time = abs(end_time - start_time)
 307
 308         if not _raw and out is not None:
 309             out = core.UStr(out.rstrip('\n'), out.encoding)
 310
 311         cola_trace = GIT_COLA_TRACE
 312         if cola_trace == 'trace':
 313             msg = 'trace: %.3fs: %s' % (elapsed_time, core.list2cmdline(command))
 314             Interaction.log_status(status, msg, '')
 315         elif cola_trace == 'full':
 316             if out or err:
 317                 core.print_stderr(
 318                     "# %.3fs: %s -> %d: '%s' '%s'"
 319                     % (elapsed_time, ' '.join(command), status, out, err)
 320                 )
 321             else:
 322                 core.print_stderr(
 323                     '# %.3fs: %s -> %d' % (elapsed_time, ' '.join(command), status)
 324                 )
 325         elif cola_trace:
 326             core.print_stderr('# %.3fs: %s' % (elapsed_time, ' '.join(command)))
 327
 328         # Allow access to the command's status code
 329         return (status, out, err)
 330
 331     def git(self, cmd, *args, **kwargs):
 332         # Handle optional arguments prior to calling transform_kwargs
 333         # otherwise they'll end up in args, which is bad.
 334         _kwargs = dict(_cwd=self.getcwd())
 335         execute_kwargs = (
 336             '_cwd',
 337             '_decode',
 338             '_encoding',
 339             '_stdin',
 340             '_stdout',
 341             '_stderr',
 342             '_raw',
 343             '_readonly',
 344             '_no_win32_startupinfo',
 345         )
 346
 347         for kwarg in execute_kwargs:
 348             if kwarg in kwargs:
 349                 _kwargs[kwarg] = kwargs.pop(kwarg)
 350
 351         # Prepare the argument list
 352         git_args = [
 353             GIT,
 354             '-c',
 355             'diff.suppressBlankEmpty=false',
 356             '-c',
 357             'log.showSignature=false',
 358             dashify(cmd),
 359         ]
 360         opt_args = transform_kwargs(**kwargs)
 361         call = git_args + opt_args
 362         call.extend(args)
 363         try:
 364             result = self.execute(call, **_kwargs)
 365         except OSError as e:
 366             if WIN32 and e.errno == errno.ENOENT:
 367                 # see if git exists at all. on win32 it can fail with ENOENT in
 368                 # case of argv overflow. we should be safe from that but use
 369                 # defensive coding for the worst-case scenario. On UNIX
 370                 # we have ENAMETOOLONG but that doesn't exist on Windows.
 371                 if _git_is_installed():
 372                     raise e
 373                 _print_win32_git_hint()
 374             result = (1, '', "error: unable to execute '%s'" % GIT)
 375         return result
 376
 377
 378 def _git_is_installed():
 379     """Return True if git is installed"""
 380     # On win32 Git commands can fail with ENOENT in case of argv overflow. we
 381     # should be safe from that but use defensive coding for the worst-case
 382     # scenario. On UNIX we have ENAMETOOLONG but that doesn't exist on
 383     # Windows.
 384     try:
 385         status, _, _ = Git.execute([GIT, '--version'])
 386         result = status == 0
 387     except OSError:
 388         result = False
 389     return result
 390
 391
 392 def transform_kwargs(**kwargs):
 393     """Transform kwargs into git command line options
 394
 395     Callers can assume the following behavior:
 396
 397     Passing foo=None ignores foo, so that callers can
 398     use default values of None that are ignored unless
 399     set explicitly.
 400
 401     Passing foo=False ignore foo, for the same reason.
 402
 403     Passing foo={string-or-number} results in ['--foo=<value>']
 404     in the resulting arguments.
 405
 406     """
 407     args = []
 408     types_to_stringify = (ustr, float, str) + int_types
 409
 410     for k, v in kwargs.items():
 411         if len(k) == 1:
 412             dashes = '-'
 413             eq = ''
 414         else:
 415             dashes = '--'
 416             eq = '='
 417         # isinstance(False, int) is True, so we have to check bool first
 418         if isinstance(v, bool):
 419             if v:
 420                 args.append('%s%s' % (dashes, dashify(k)))
 421             # else: pass  # False is ignored; flag=False inhibits --flag
 422         elif isinstance(v, types_to_stringify):
 423             args.append('%s%s%s%s' % (dashes, dashify(k), eq, v))
 424
 425     return args
 426
 427
 428 def win32_git_error_hint():
 429     return (
 430         '\n'
 431         'NOTE: If you have Git installed in a custom location, e.g.\n'
 432         'C:\\Tools\\Git, then you can create a file at\n'
 433         '~/.config/git-cola/git-bindir with following text\n'
 434         'and git-cola will add the specified location to your $PATH\n'
 435         'automatically when starting cola:\n'
 436         '\n'
 437         r'C:\Tools\Git\bin'
 438     )
 439
 440
 441 @memoize
 442 def _print_win32_git_hint():
 443     hint = '\n' + win32_git_error_hint() + '\n'
 444     core.print_stderr("error: unable to execute 'git'" + hint)
 445
 446
 447 def create():
 448     """Create Git instances
 449
 450     >>> git = create()
 451     >>> status, out, err = git.version()
 452     >>> 'git' == out[:3].lower()
 453     True
 454
 455     """
 456     return Git()