~bzr-pqm/bzr/bzr.dev

# Copyright (C) 2005-2010 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., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA

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

import codecs

import warnings

from bzrlib.lazy_import import lazy_import

lazy_import(globals(), """

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 shutil

from shutil import (

    rmtree,

    )

import socket

import subprocess

import tempfile

from tempfile import (

    mkdtemp,

    )

import unicodedata

from bzrlib import (

    cache_utf8,

    errors,

    trace,

    win32utils,

    )

""")

from bzrlib.symbol_versioning import (

    deprecated_function,

    deprecated_in,

    )

# sha and md5 modules are deprecated in python2.6 but hashlib is available as

# of 2.5

if sys.version_info < (2, 5):

    import md5 as _mod_md5

    md5 = _mod_md5.new

    import sha as _mod_sha

    sha = _mod_sha.new

else:

    from hashlib import (

        md5,

        sha1 as sha,

        )

import bzrlib

from bzrlib import symbol_versioning

# Cross platform wall-clock time functionality with decent resolution.

# On Linux ``time.clock`` returns only CPU time. On Windows, ``time.time()``

# only has a resolution of ~15ms. Note that ``time.clock()`` is not

# synchronized with ``time.time()``, this is only meant to be used to find

# delta times by subtracting from another call to this function.

timer_func = time.time

if sys.platform == 'win32':

    timer_func = time.clock

# 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_NOINHERIT and O_TEXT exists only on win32 too.

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

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

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

def get_unicode_argv():

    try:

        user_encoding = get_user_encoding()

        return [a.decode(user_encoding) for a in sys.argv[1:]]

    except UnicodeDecodeError:

        raise errors.BzrError(("Parameter '%r' is unsupported by the current "

                                                            "encoding." % a))

def make_readonly(filename):

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

    mod = os.lstat(filename).st_mode

    if not stat.S_ISLNK(mod):

        mod = mod & 0777555

        os.chmod(filename, mod)

def make_writable(filename):

    mod = os.lstat(filename).st_mode

    if not stat.S_ISLNK(mod):

        mod = mod | 0200

        os.chmod(filename, mod)

def minimum_path_selection(paths):

    """Return the smallset subset of paths which are outside paths.

    :param paths: A container (and hence not None) of paths.

    :return: A set of paths sufficient to include everything in paths via

        is_inside, drawn from the paths parameter.

    """

    if len(paths) < 2:

        return set(paths)

    def sort_key(path):

        return path.split('/')

    sorted_paths = sorted(list(paths), key=sort_key)

    search_paths = [sorted_paths[0]]

    for path in sorted_paths[1:]:

        if not is_inside(search_paths[-1], path):

            # This path is unique, add it

            search_paths.append(path)

    return set(search_paths)

_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'

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

_kind_marker_map = {

    "file": "",

    _directory_kind: "/",

    "symlink": "@",

    'tree-reference': '+',

}

def kind_marker(kind):

    try:

        return _kind_marker_map[kind]

    except KeyError:

        # Slightly faster than using .get(, '') when the common case is that

        # kind will be found

        return ''

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

if lexists is None:

    def lexists(f):

        try:

            stat = getattr(os, 'lstat', os.stat)

            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:

    base = os.path.basename(new)

    dirname = os.path.dirname(new)

    # callers use different encodings for the paths so the following MUST

    # respect that. We rely on python upcasting to unicode if new is unicode

    # and keeping a str if not.

    tmp_name = '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

    failure_exc = None

    success = False

    try:

        try:

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

            # not be set.

            rename_func(old, new)

            success = True

        except (IOError, OSError), e:

            # source and target may be aliases of each other (e.g. on a

            # case-insensitive filesystem), so we may have accidentally renamed

            # source by when we tried to rename target

            failure_exc = sys.exc_info()

            if (file_existed and e.errno in (None, errno.ENOENT)

                and old.lower() == new.lower()):

                # source and target are the same file on a case-insensitive

                # filesystem, so we don't generate an exception

                failure_exc = None

    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)

    if failure_exc is not None:

        raise failure_exc[0], failure_exc[1], failure_exc[2]

# 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 _win98_abspath(path):

    """Return the absolute version of a path.

    Windows 98 safe implementation (python reimplementation

    of Win32 API function GetFullPathNameW)

    """

    # Corner cases:

    #   C:\path     => C:/path

    #   C:/path     => C:/path

    #   \\HOST\path => //HOST/path

    #   //HOST/path => //HOST/path

    #   path        => C:/cwd/path

    #   /path       => C:/path

    path = unicode(path)

    # check for absolute path

    drive = _nt_splitdrive(path)[0]

    if drive == '' and path[:2] not in('//','\\\\'):

        cwd = os.getcwdu()

        # we cannot simply os.path.join cwd and path

        # because os.path.join('C:','/path') produce '/path'

        # and this is incorrect

        if path[:1] in ('/','\\'):

            cwd = _nt_splitdrive(cwd)[0]

            path = path[1:]

        path = cwd + '\\' + path

    return _win32_fixdrive(_nt_normpath(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('NFC', 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':

    if win32utils.winver == 'Windows 98':

        abspath = _win98_abspath

    else:

        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)

    f = win32utils.get_unicode_argv     # special function or None

    if f is not None:

        get_unicode_argv = f

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

    osutils.get_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

    """

    from bzrlib.trace import mutter

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

    if not output_encoding:

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

        if not input_encoding:

            output_encoding = get_user_encoding()

            mutter('encoding stdout as osutils.get_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 = get_user_encoding()

        mutter('cp0 is invalid encoding.'

               ' encoding stdout as osutils.get_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, get_user_encoding())

                        )

        output_encoding = get_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 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

    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

    return False

def pumpfile(from_file, to_file, read_length=-1, buff_size=32768,

             report_activity=None, direction='read'):

    """Copy contents of one file to another.

    The read_length can either be -1 to read to end-of-file (EOF) or

    it can specify the maximum number of bytes to read.

    The buff_size represents the maximum size for each read operation

    performed on from_file.

    :param report_activity: Call this as bytes are read, see

        Transport._report_activity

    :param direction: Will be passed to report_activity

    :return: The number of bytes copied.

    """

    length = 0

    if read_length >= 0:

        # read specified number of bytes

        while read_length > 0:

            num_bytes_to_read = min(read_length, buff_size)

            block = from_file.read(num_bytes_to_read)

            if not block:

                # EOF reached

                break

            if report_activity is not None:

                report_activity(len(block), direction)

            to_file.write(block)

            actual_bytes_read = len(block)

            read_length -= actual_bytes_read

            length += actual_bytes_read

    else:

        # read to EOF

        while True:

            block = from_file.read(buff_size)

            if not block:

                # EOF reached

                break

            if report_activity is not None:

                report_activity(len(block), direction)

            to_file.write(block)

            length += len(block)

    return length

def pump_string_file(bytes, file_handle, segment_size=None):

    """Write bytes to file_handle in many smaller writes.

    :param bytes: The string to write.

    :param file_handle: The file to write to.

    """

    # Write data in chunks rather than all at once, because very large

    # writes fail on some platforms (e.g. Windows with SMB  mounted

    # drives).

    if not segment_size:

        segment_size = 5242880 # 5MB

    segments = range(len(bytes) / segment_size + 1)

    write = file_handle.write

    for segment_index in segments:

        segment = buffer(bytes, segment_index * segment_size, segment_size)

        write(segment)

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):

    """Calculate the hexdigest of an open file.

    The file cursor should be already at the start.

    """

    s = sha()

    BUFSIZE = 128<<10

    while True:

        b = f.read(BUFSIZE)

        if not b:

            break

        s.update(b)

    return s.hexdigest()

def size_sha_file(f):

    """Calculate the size and hexdigest of an open file.

    The file cursor should be already at the start and

    the caller is responsible for closing the file afterwards.

    """

    size = 0

    s = sha()

    BUFSIZE = 128<<10

    while True:

        b = f.read(BUFSIZE)

        if not b:

            break

        size += len(b)

        s.update(b)

    return size, s.hexdigest()

def sha_file_by_name(fname):

    """Calculate the SHA1 of a file by reading the full text"""

    s = sha()

    f = os.open(fname, os.O_RDONLY | O_BINARY | O_NOINHERIT)

    try:

        while True:

            b = os.read(f, 1<<16)

            if not b:

                return s.hexdigest()

            s.update(b)

    finally:

        os.close(f)

def sha_strings(strings, _factory=sha):

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

    s = _factory()

    map(s.update, strings)

    return s.hexdigest()

def sha_string(f, _factory=sha):

    return _factory(f).hexdigest()

def fingerprint_file(f):

    b = f.read()

    return {'size': len(b),

            'sha1': sha(b).hexdigest()}