~bzr-pqm/bzr/bzr.dev

# Copyright (C) 2005, 2006 Canonical Ltd

#

# This program is free software; you can redistribute it and/or modify

# it under the terms of the GNU General Public License as published by

# the Free Software Foundation; either version 2 of the License, or

# (at your option) any later version.

#

# This program is distributed in the hope that it will be useful,

# but WITHOUT ANY WARRANTY; without even the implied warranty of

# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the

# GNU General Public License for more details.

#

# You should have received a copy of the GNU General Public License

# along with this program; if not, write to the Free Software

# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA

from cStringIO import StringIO

import os

import re

import stat

from stat import (S_ISREG, S_ISDIR, S_ISLNK, ST_MODE, ST_SIZE,

                  S_ISCHR, S_ISBLK, S_ISFIFO, S_ISSOCK)

import sys

import time

from bzrlib.lazy_import import lazy_import

lazy_import(globals(), """

import codecs

from datetime import datetime

import errno

from ntpath import (abspath as _nt_abspath,

                    join as _nt_join,

                    normpath as _nt_normpath,

                    realpath as _nt_realpath,

                    splitdrive as _nt_splitdrive,

                    )

import posixpath

import sha

import shutil

from shutil import (

    rmtree,

    )

import tempfile

from tempfile import (

    mkdtemp,

    )

import unicodedata

from bzrlib import (

    errors,

    )

""")

import bzrlib

from bzrlib.symbol_versioning import (

    deprecated_function,

    zero_nine,

    )

from bzrlib.trace import mutter

# On win32, O_BINARY is used to indicate the file should

# be opened in binary mode, rather than text mode.

# On other platforms, O_BINARY doesn't exist, because

# they always open in binary mode, so it is okay to

# OR with 0 on those platforms

O_BINARY = getattr(os, 'O_BINARY', 0)

def make_readonly(filename):

    """Make a filename read-only."""

    mod = os.stat(filename).st_mode

    mod = mod & 0777555

    os.chmod(filename, mod)

def make_writable(filename):

    mod = os.stat(filename).st_mode

    mod = mod | 0200

    os.chmod(filename, mod)

_QUOTE_RE = None

def quotefn(f):

    """Return a quoted filename filename

    This previously used backslash quoting, but that works poorly on

    Windows."""

    # TODO: I'm not really sure this is the best format either.x

    global _QUOTE_RE

    if _QUOTE_RE is None:

        _QUOTE_RE = re.compile(r'([^a-zA-Z0-9.,:/\\_~-])')

    if _QUOTE_RE.search(f):

        return '"' + f + '"'

    else:

        return f

_directory_kind = 'directory'

_formats = {

    stat.S_IFDIR:_directory_kind,

    stat.S_IFCHR:'chardev',

    stat.S_IFBLK:'block',

    stat.S_IFREG:'file',

    stat.S_IFIFO:'fifo',

    stat.S_IFLNK:'symlink',

    stat.S_IFSOCK:'socket',

}

def file_kind_from_stat_mode(stat_mode, _formats=_formats, _unknown='unknown'):

    """Generate a file kind from a stat mode. This is used in walkdirs.

    Its performance is critical: Do not mutate without careful benchmarking.

    """

    try:

        return _formats[stat_mode & 0170000]

    except KeyError:

        return _unknown

def file_kind(f, _lstat=os.lstat, _mapper=file_kind_from_stat_mode):

    try:

        return _mapper(_lstat(f).st_mode)

    except OSError, e:

        if getattr(e, 'errno', None) == errno.ENOENT:

            raise errors.NoSuchFile(f)

        raise

def get_umask():

    """Return the current umask"""

    # Assume that people aren't messing with the umask while running

    # XXX: This is not thread safe, but there is no way to get the

    #      umask without setting it

    umask = os.umask(0)

    os.umask(umask)

    return umask

def kind_marker(kind):

    if kind == 'file':

        return ''

    elif kind == _directory_kind:

        return '/'

    elif kind == 'symlink':

        return '@'

    else:

        raise errors.BzrError('invalid file kind %r' % kind)

lexists = getattr(os.path, 'lexists', None)

if lexists is None:

    def lexists(f):

        try:

            if getattr(os, 'lstat') is not None:

                os.lstat(f)

            else:

                os.stat(f)

            return True

        except OSError,e:

            if e.errno == errno.ENOENT:

                return False;

            else:

                raise errors.BzrError("lstat/stat of (%r): %r" % (f, e))

def fancy_rename(old, new, rename_func, unlink_func):

    """A fancy rename, when you don't have atomic rename.

    :param old: The old path, to rename from

    :param new: The new path, to rename to

    :param rename_func: The potentially non-atomic rename function

    :param unlink_func: A way to delete the target file if the full rename succeeds

    """

    # sftp rename doesn't allow overwriting, so play tricks:

    import random

    base = os.path.basename(new)

    dirname = os.path.dirname(new)

    tmp_name = u'tmp.%s.%.9f.%d.%s' % (base, time.time(), os.getpid(), rand_chars(10))

    tmp_name = pathjoin(dirname, tmp_name)

    # Rename the file out of the way, but keep track if it didn't exist

    # We don't want to grab just any exception

    # something like EACCES should prevent us from continuing

    # The downside is that the rename_func has to throw an exception

    # with an errno = ENOENT, or NoSuchFile

    file_existed = False

    try:

        rename_func(new, tmp_name)

    except (errors.NoSuchFile,), e:

        pass

    except IOError, e:

        # RBC 20060103 abstraction leakage: the paramiko SFTP clients rename

        # function raises an IOError with errno is None when a rename fails.

        # This then gets caught here.

        if e.errno not in (None, errno.ENOENT, errno.ENOTDIR):

            raise

    except Exception, e:

        if (getattr(e, 'errno', None) is None

            or e.errno not in (errno.ENOENT, errno.ENOTDIR)):

            raise

    else:

        file_existed = True

    success = False

    try:

        # This may throw an exception, in which case success will

        # not be set.

        rename_func(old, new)

        success = True

    finally:

        if file_existed:

            # If the file used to exist, rename it back into place

            # otherwise just delete it from the tmp location

            if success:

                unlink_func(tmp_name)

            else:

                rename_func(tmp_name, new)

# In Python 2.4.2 and older, os.path.abspath and os.path.realpath

# choke on a Unicode string containing a relative path if

# os.getcwd() returns a non-sys.getdefaultencoding()-encoded

# string.

_fs_enc = sys.getfilesystemencoding() or 'utf-8'

def _posix_abspath(path):

    # jam 20060426 rather than encoding to fsencoding

    # copy posixpath.abspath, but use os.getcwdu instead

    if not posixpath.isabs(path):

        path = posixpath.join(getcwd(), path)

    return posixpath.normpath(path)

def _posix_realpath(path):

    return posixpath.realpath(path.encode(_fs_enc)).decode(_fs_enc)

def _win32_fixdrive(path):

    """Force drive letters to be consistent.

    win32 is inconsistent whether it returns lower or upper case

    and even if it was consistent the user might type the other

    so we force it to uppercase

    running python.exe under cmd.exe return capital C:\\

    running win32 python inside a cygwin shell returns lowercase c:\\

    """

    drive, path = _nt_splitdrive(path)

    return drive.upper() + path

def _win32_abspath(path):

    # Real _nt_abspath doesn't have a problem with a unicode cwd

    return _win32_fixdrive(_nt_abspath(unicode(path)).replace('\\', '/'))

def _win32_realpath(path):

    # Real _nt_realpath doesn't have a problem with a unicode cwd

    return _win32_fixdrive(_nt_realpath(unicode(path)).replace('\\', '/'))

def _win32_pathjoin(*args):

    return _nt_join(*args).replace('\\', '/')

def _win32_normpath(path):

    return _win32_fixdrive(_nt_normpath(unicode(path)).replace('\\', '/'))

def _win32_getcwd():

    return _win32_fixdrive(os.getcwdu().replace('\\', '/'))

def _win32_mkdtemp(*args, **kwargs):

    return _win32_fixdrive(tempfile.mkdtemp(*args, **kwargs).replace('\\', '/'))

def _win32_rename(old, new):

    """We expect to be able to atomically replace 'new' with old.

    On win32, if new exists, it must be moved out of the way first,

    and then deleted. 

    """

    try:

        fancy_rename(old, new, rename_func=os.rename, unlink_func=os.unlink)

    except OSError, e:

        if e.errno in (errno.EPERM, errno.EACCES, errno.EBUSY, errno.EINVAL):

            # If we try to rename a non-existant file onto cwd, we get 

            # EPERM or EACCES instead of ENOENT, this will raise ENOENT 

            # if the old path doesn't exist, sometimes we get EACCES

            # On Linux, we seem to get EBUSY, on Mac we get EINVAL

            os.lstat(old)

        raise

def _mac_getcwd():

    return unicodedata.normalize('NFKC', os.getcwdu())

# Default is to just use the python builtins, but these can be rebound on

# particular platforms.

abspath = _posix_abspath

realpath = _posix_realpath

pathjoin = os.path.join

normpath = os.path.normpath

getcwd = os.getcwdu

rename = os.rename

dirname = os.path.dirname

basename = os.path.basename

split = os.path.split

splitext = os.path.splitext

# These were already imported into local scope

# mkdtemp = tempfile.mkdtemp

# rmtree = shutil.rmtree

MIN_ABS_PATHLENGTH = 1

if sys.platform == 'win32':

    abspath = _win32_abspath

    realpath = _win32_realpath

    pathjoin = _win32_pathjoin

    normpath = _win32_normpath

    getcwd = _win32_getcwd

    mkdtemp = _win32_mkdtemp

    rename = _win32_rename

    MIN_ABS_PATHLENGTH = 3

    def _win32_delete_readonly(function, path, excinfo):

        """Error handler for shutil.rmtree function [for win32]

        Helps to remove files and dirs marked as read-only.

        """

        exception = excinfo[1]

        if function in (os.remove, os.rmdir) \

            and isinstance(exception, OSError) \

            and exception.errno == errno.EACCES:

            make_writable(path)

            function(path)

        else:

            raise

    def rmtree(path, ignore_errors=False, onerror=_win32_delete_readonly):

        """Replacer for shutil.rmtree: could remove readonly dirs/files"""

        return shutil.rmtree(path, ignore_errors, onerror)

elif sys.platform == 'darwin':

    getcwd = _mac_getcwd

def get_terminal_encoding():

    """Find the best encoding for printing to the screen.

    This attempts to check both sys.stdout and sys.stdin to see

    what encoding they are in, and if that fails it falls back to

    bzrlib.user_encoding.

    The problem is that on Windows, locale.getpreferredencoding()

    is not the same encoding as that used by the console:

    http://mail.python.org/pipermail/python-list/2003-May/162357.html

    On my standard US Windows XP, the preferred encoding is

    cp1252, but the console is cp437

    """

    output_encoding = getattr(sys.stdout, 'encoding', None)

    if not output_encoding:

        input_encoding = getattr(sys.stdin, 'encoding', None)

        if not input_encoding:

            output_encoding = bzrlib.user_encoding

            mutter('encoding stdout as bzrlib.user_encoding %r', output_encoding)

        else:

            output_encoding = input_encoding

            mutter('encoding stdout as sys.stdin encoding %r', output_encoding)

    else:

        mutter('encoding stdout as sys.stdout encoding %r', output_encoding)

    if output_encoding == 'cp0':

        # invalid encoding (cp0 means 'no codepage' on Windows)

        output_encoding = bzrlib.user_encoding

        mutter('cp0 is invalid encoding.'

               ' encoding stdout as bzrlib.user_encoding %r', output_encoding)

    # check encoding

    try:

        codecs.lookup(output_encoding)

    except LookupError:

        sys.stderr.write('bzr: warning:'

                         ' unknown terminal encoding %s.\n'

                         '  Using encoding %s instead.\n'

                         % (output_encoding, bzrlib.user_encoding)

                        )

        output_encoding = bzrlib.user_encoding

    return output_encoding

def normalizepath(f):

    if getattr(os.path, 'realpath', None) is not None:

        F = realpath

    else:

        F = abspath

    [p,e] = os.path.split(f)

    if e == "" or e == "." or e == "..":

        return F(f)

    else:

        return pathjoin(F(p), e)

def backup_file(fn):

    """Copy a file to a backup.

    Backups are named in GNU-style, with a ~ suffix.

    If the file is already a backup, it's not copied.

    """

    if fn[-1] == '~':

        return

    bfn = fn + '~'

    if has_symlinks() and os.path.islink(fn):

        target = os.readlink(fn)

        os.symlink(target, bfn)

        return

    inf = file(fn, 'rb')

    try:

        content = inf.read()

    finally:

        inf.close()

    outf = file(bfn, 'wb')

    try:

        outf.write(content)

    finally:

        outf.close()

def isdir(f):

    """True if f is an accessible directory."""

    try:

        return S_ISDIR(os.lstat(f)[ST_MODE])

    except OSError:

        return False

def isfile(f):

    """True if f is a regular file."""

    try:

        return S_ISREG(os.lstat(f)[ST_MODE])

    except OSError:

        return False

def islink(f):

    """True if f is a symlink."""

    try:

        return S_ISLNK(os.lstat(f)[ST_MODE])

    except OSError:

        return False

def is_inside(dir, fname):

    """True if fname is inside dir.

    The parameters should typically be passed to osutils.normpath first, so

    that . and .. and repeated slashes are eliminated, and the separators

    are canonical for the platform.

    The empty string as a dir name is taken as top-of-tree and matches 

    everything.

    """

    # XXX: Most callers of this can actually do something smarter by 

    # looking at the inventory

    if dir == fname:

        return True

    if dir == '':

        return True

    if dir[-1] != '/':

        dir += '/'

    return fname.startswith(dir)

def is_inside_any(dir_list, fname):

    """True if fname is inside any of given dirs."""

    for dirname in dir_list:

        if is_inside(dirname, fname):

            return True

    else:

        return False

def is_inside_or_parent_of_any(dir_list, fname):

    """True if fname is a child or a parent of any of the given files."""

    for dirname in dir_list:

        if is_inside(dirname, fname) or is_inside(fname, dirname):

            return True

    else:

        return False

def pumpfile(fromfile, tofile):

    """Copy contents of one file to another."""

    BUFSIZE = 32768

    while True:

        b = fromfile.read(BUFSIZE)

        if not b:

            break

        tofile.write(b)

def file_iterator(input_file, readsize=32768):

    while True:

        b = input_file.read(readsize)

        if len(b) == 0:

            break

        yield b

def sha_file(f):

    if getattr(f, 'tell', None) is not None:

        assert f.tell() == 0

    s = sha.new()

    BUFSIZE = 128<<10

    while True:

        b = f.read(BUFSIZE)

        if not b:

            break

        s.update(b)

    return s.hexdigest()

def sha_strings(strings):

    """Return the sha-1 of concatenation of strings"""

    s = sha.new()

    map(s.update, strings)

    return s.hexdigest()

def sha_string(f):

    s = sha.new()

    s.update(f)

    return s.hexdigest()

def fingerprint_file(f):

    s = sha.new()

    b = f.read()

    s.update(b)

    size = len(b)

    return {'size': size,

            'sha1': s.hexdigest()}

def compare_files(a, b):

    """Returns true if equal in contents"""

    BUFSIZE = 4096

    while True:

        ai = a.read(BUFSIZE)

        bi = b.read(BUFSIZE)

        if ai != bi:

            return False

        if ai == '':

            return True

def local_time_offset(t=None):

    """Return offset of local zone from GMT, either at present or at time t."""

    if t is None:

        t = time.time()

    offset = datetime.fromtimestamp(t) - datetime.utcfromtimestamp(t)

    return offset.days * 86400 + offset.seconds

def format_date(t, offset=0, timezone='original', date_fmt=None, 

                show_offset=True):

    ## TODO: Perhaps a global option to use either universal or local time?

    ## Or perhaps just let people set $TZ?

    assert isinstance(t, float)

    if timezone == 'utc':

        tt = time.gmtime(t)

        offset = 0

    elif timezone == 'original':

        if offset is None:

            offset = 0

        tt = time.gmtime(t + offset)

    elif timezone == 'local':

        tt = time.localtime(t)

        offset = local_time_offset(t)

    else:

        raise errors.BzrError("unsupported timezone format %r" % timezone,

                              ['options are "utc", "original", "local"'])

    if date_fmt is None:

        date_fmt = "%a %Y-%m-%d %H:%M:%S"

    if show_offset:

        offset_str = ' %+03d%02d' % (offset / 3600, (offset / 60) % 60)

    else:

        offset_str = ''

    return (time.strftime(date_fmt, tt) +  offset_str)

def compact_date(when):

    return time.strftime('%Y%m%d%H%M%S', time.gmtime(when))

def format_delta(delta):

    """Get a nice looking string for a time delta.

    :param delta: The time difference in seconds, can be positive or negative.

        positive indicates time in the past, negative indicates time in the

        future. (usually time.time() - stored_time)

    :return: String formatted to show approximate resolution

    """

    delta = int(delta)

    if delta >= 0:

        direction = 'ago'

    else:

        direction = 'in the future'

        delta = -delta

    seconds = delta

    if seconds < 90: # print seconds up to 90 seconds

        if seconds == 1:

            return '%d second %s' % (seconds, direction,)

        else:

            return '%d seconds %s' % (seconds, direction)

    minutes = int(seconds / 60)

    seconds -= 60 * minutes

    if seconds == 1:

        plural_seconds = ''

    else:

        plural_seconds = 's'

    if minutes < 90: # print minutes, seconds up to 90 minutes

        if minutes == 1:

            return '%d minute, %d second%s %s' % (

                    minutes, seconds, plural_seconds, direction)

        else:

            return '%d minutes, %d second%s %s' % (

                    minutes, seconds, plural_seconds, direction)

    hours = int(minutes / 60)

    minutes -= 60 * hours

    if minutes == 1:

        plural_minutes = ''

    else:

        plural_minutes = 's'

    if hours == 1:

        return '%d hour, %d minute%s %s' % (hours, minutes,

                                            plural_minutes, direction)

    return '%d hours, %d minute%s %s' % (hours, minutes,

                                         plural_minutes, direction)

def filesize(f):

    """Return size of given open file."""

    return os.fstat(f.fileno())[ST_SIZE]

# Define rand_bytes based on platform.

try:

    # Python 2.4 and later have os.urandom,

    # but it doesn't work on some arches

    os.urandom(1)

    rand_bytes = os.urandom

except (NotImplementedError, AttributeError):

    # If python doesn't have os.urandom, or it doesn't work,

    # then try to first pull random data from /dev/urandom

    try:

        rand_bytes = file('/dev/urandom', 'rb').read

    # Otherwise, use this hack as a last resort

    except (IOError, OSError):

        # not well seeded, but better than nothing

        def rand_bytes(n):

            import random

            s = ''

            while n:

                s += chr(random.randint(0, 255))

                n -= 1

            return s

ALNUM = '0123456789abcdefghijklmnopqrstuvwxyz'

def rand_chars(num):

    """Return a random string of num alphanumeric characters

    The result only contains lowercase chars because it may be used on 

    case-insensitive filesystems.

    """

    s = ''

    for raw_byte in rand_bytes(num):

        s += ALNUM[ord(raw_byte) % 36]

    return s

## TODO: We could later have path objects that remember their list

## decomposition (might be too tricksy though.)

def splitpath(p):

    """Turn string into list of parts."""

    assert isinstance(p, basestring)

    # split on either delimiter because people might use either on

    # Windows

    ps = re.split(r'[\\/]', p)

    rps = []

    for f in ps:

        if f == '..':

            raise errors.BzrError("sorry, %r not allowed in path" % f)