# encoding.py - character transcoding support for Mercurial

#

#  Copyright 2005-2009 Olivia Mackall <olivia@selenic.com> and others

#

# This software may be used and distributed according to the terms of the

# GNU General Public License version 2 or any later version.

import locale

import os

import re

import typing

import unicodedata

from typing import (

    Any,

    Callable,

    List,

    Text,

    Type,

    TypeVar,

    Union,

)

from . import (

    error,

    policy,

    pycompat,

)

from .pure import charencode as charencodepure

# keep pyflakes happy

for t in (Any, Callable, List, Text, Type, Union):

    assert t

_Tlocalstr = TypeVar('_Tlocalstr', bound='localstr')

charencode = policy.importmod('charencode')

isasciistr = charencode.isasciistr

asciilower = charencode.asciilower

asciiupper = charencode.asciiupper

_jsonescapeu8fast = charencode.jsonescapeu8fast

_sysstr = pycompat.sysstr

unichr = chr

# These unicode characters are ignored by HFS+ (Apple Technote 1150,

# "Unicode Subtleties"), so we need to ignore them in some places for

# sanity.

_ignore = [

    unichr(int(x, 16)).encode("utf-8")

    for x in b"200c 200d 200e 200f 202a 202b 202c 202d 202e "

    b"206a 206b 206c 206d 206e 206f feff".split()

]

# verify the next function will work

assert all(i.startswith((b"\xe2", b"\xef")) for i in _ignore)

    """Remove codepoints ignored by HFS+ from s.

    >>> hfsignoreclean(u'.h\u200cg'.encode('utf-8'))

    '.hg'

    >>> hfsignoreclean(u'.h\ufeffg'.encode('utf-8'))

    '.hg'

    """

    if b"\xe2" in s or b"\xef" in s:

        for c in _ignore:

            s = s.replace(c, b'')

    return s

# encoding.environ is provided read-only, which may not be used to modify

# the process environment

_nativeenviron = os.supports_bytes_environ

if _nativeenviron:

    environ = os.environb  # re-exports

    if pycompat.sysplatform == b'OpenVMS':

        # workaround for a bug in VSI 3.10 port

        # os.environb is only populated with a few Predefined symbols

        def newget(self, key, default=None):

            # pytype on linux does not understand OpenVMS special modules

            import _decc  # pytype: disable=import-error

            v = _decc.getenv(key, None)

            if isinstance(key, bytes):

                return default if v is None else v.encode('latin-1')

            else:

                return default if v is None else v

        environ.__class__.get = newget

else:

    # preferred encoding isn't known yet; use utf-8 to avoid unicode error

    # and recreate it once encoding is settled

    environ = {

        k.encode('utf-8'): v.encode('utf-8')

        for k, v in os.environ.items()  # re-exports

    }

_encodingrewrites = {

    b'646': b'ascii',

    b'ANSI_X3.4-1968': b'ascii',

}

# cp65001 is a Windows variant of utf-8, which isn't supported on Python 2.

# No idea if it should be rewritten to the canonical name 'utf-8' on Python 3.

# https://bugs.python.org/issue13216

if pycompat.iswindows:

    _encodingrewrites[b'cp65001'] = b'utf-8'

try:

    encoding = environ.get(b"HGENCODING")

    if not encoding:

        encoding = locale.getpreferredencoding().encode('ascii') or b'ascii'

        encoding = _encodingrewrites.get(encoding, encoding)

except locale.Error:

    encoding = b'ascii'

encodingmode = environ.get(b"HGENCODINGMODE", b"strict")

fallbackencoding = b'ISO-8859-1'

class localstr(bytes):

    """This class allows strings that are unmodified to be

    round-tripped to the local encoding and back"""

    def __new__(cls, u, l):

        s = bytes.__new__(cls, l)

        s._utf8 = u

        return s

    if typing.TYPE_CHECKING:

        # pseudo implementation to help pytype see localstr() constructor

            super(localstr, self).__init__(l)

            self._utf8 = u

    def __hash__(self):

        return hash(self._utf8)  # avoid collisions in local string space

class safelocalstr(bytes):

    """Tagged string denoting it was previously an internal UTF-8 string,

    and can be converted back to UTF-8 losslessly

    >>> assert safelocalstr(b'\\xc3') == b'\\xc3'

    >>> assert b'\\xc3' == safelocalstr(b'\\xc3')

    >>> assert b'\\xc3' in {safelocalstr(b'\\xc3'): 0}

    >>> assert safelocalstr(b'\\xc3') in {b'\\xc3': 0}

    """

    """

    Convert a string from internal UTF-8 to local encoding

    All internal strings should be UTF-8 but some repos before the

    implementation of locale support may contain latin1 or possibly

    other character sets. We attempt to decode everything strictly

    using UTF-8, then Latin-1, and failing that, we use UTF-8 and

    replace unknown characters.

    The localstr class is used to cache the known UTF-8 encoding of

    strings next to their local representation to allow lossless

    round-trip conversion back to UTF-8.

    >>> u = b'foo: \\xc3\\xa4' # utf-8

    >>> l = tolocal(u)

    >>> l

    'foo: ?'

    >>> fromlocal(l)

    'foo: \\xc3\\xa4'

    >>> u2 = b'foo: \\xc3\\xa1'

    >>> d = { l: 1, tolocal(u2): 2 }

    >>> len(d) # no collision

    2

    >>> b'foo: ?' in d

    False

    >>> l1 = b'foo: \\xe4' # historical latin1 fallback

    >>> l = tolocal(l1)

    >>> l

    'foo: ?'

    >>> fromlocal(l) # magically in utf-8

    'foo: \\xc3\\xa4'

    """

    if isasciistr(s):

        return s

    try:

        try:

            # make sure string is actually stored in UTF-8

            u = s.decode('UTF-8')

            if encoding == b'UTF-8':

                # fast path

                return s

            r = u.encode(_sysstr(encoding), "replace")

            if u == r.decode(_sysstr(encoding)):

                # r is a safe, non-lossy encoding of s

                return safelocalstr(r)

            return localstr(s, r)

        except UnicodeDecodeError:

            # we should only get here if we're looking at an ancient changeset

            try:

                u = s.decode(_sysstr(fallbackencoding))

                r = u.encode(_sysstr(encoding), "replace")

                if u == r.decode(_sysstr(encoding)):

                    # r is a safe, non-lossy encoding of s

                    return safelocalstr(r)

                return localstr(u.encode('UTF-8'), r)

            except UnicodeDecodeError:

                u = s.decode("utf-8", "replace")  # last ditch

                # can't round-trip

                return u.encode(_sysstr(encoding), "replace")

    except LookupError as k:

        raise error.Abort(

            pycompat.bytestr(k), hint=b"please check your locale settings"

        )

    """

    Convert a string from the local character encoding to UTF-8

    We attempt to decode strings using the encoding mode set by

    HGENCODINGMODE, which defaults to 'strict'. In this mode, unknown

    characters will cause an error message. Other modes include

    'replace', which replaces unknown characters with a special

    Unicode character, and 'ignore', which drops the character.

    """

    # can we do a lossless round-trip?

    if isinstance(s, localstr):

        return s._utf8

    if isasciistr(s):

        return s

    try:

        u = s.decode(_sysstr(encoding), _sysstr(encodingmode))

        return u.encode("utf-8")

    except UnicodeDecodeError as inst:

        sub = s[max(0, inst.start - 10) : inst.start + 10]

        raise error.Abort(

            b"decoding near '%s': %s!" % (sub, pycompat.bytestr(inst))

        )

    except LookupError as k:

        raise error.Abort(

            pycompat.bytestr(k), hint=b"please check your locale settings"

        )

    """Convert a unicode string to a byte string of local encoding"""

    return tolocal(u.encode('utf-8'))

    """Convert a byte string of local encoding to a unicode string"""

    return fromlocal(s).decode('utf-8')

    """Create a proxy method that forwards __unicode__() and __str__() of

    Python 3 to __bytes__()"""

    def unifunc(obj):

        return unifromlocal(bytesfunc(obj))

    return unifunc

# converter functions between native str and byte string. use these if the

# character encoding is not aware (e.g. exception message) or is known to

# be locale dependent (e.g. date formatting.)

strtolocal = unitolocal

strfromlocal = unifromlocal

strmethod = unimethod

    """best-effort encoding-aware case-folding of local string s"""

    try:

        return asciilower(s)

    except UnicodeDecodeError:

        pass

    try:

        if isinstance(s, localstr):

            u = s._utf8.decode("utf-8")

        else:

            u = s.decode(_sysstr(encoding), _sysstr(encodingmode))

        lu = u.lower()

        if u == lu:

            return s  # preserve localstring

        return lu.encode(_sysstr(encoding))

    except UnicodeError:

        return s.lower()  # we don't know how to fold this except in ASCII

    except LookupError as k:

        raise error.Abort(

            pycompat.bytestr(k), hint=b"please check your locale settings"

        )

    """best-effort encoding-aware case-folding of local string s"""

    try:

        return asciiupper(s)

    except UnicodeDecodeError:

        return upperfallback(s)

    try:

        if isinstance(s, localstr):

            u = s._utf8.decode("utf-8")

        else:

            u = s.decode(_sysstr(encoding), _sysstr(encodingmode))

        uu = u.upper()

        if u == uu:

            return s  # preserve localstring

        return uu.encode(_sysstr(encoding))

    except UnicodeError:

        return s.upper()  # we don't know how to fold this except in ASCII

    except LookupError as k:

        raise error.Abort(

            pycompat.bytestr(k), hint=b"please check your locale settings"

        )

if not _nativeenviron:

    # now encoding and helper functions are available, recreate the environ

    # dict to be exported to other modules

    if pycompat.iswindows:

        class WindowsEnviron(dict):

            """`os.environ` normalizes environment variables to uppercase on windows"""

            def get(self, key, default=None):

                return super().get(upper(key), default)

        environ = WindowsEnviron()

    for k, v in os.environ.items():  # re-exports

        environ[tolocal(k.encode('utf-8'))] = tolocal(v.encode('utf-8'))

DRIVE_RE = re.compile(b'^[a-z]:')

# os.getcwd() on Python 3 returns string, but it has os.getcwdb() which

# returns bytes.

if pycompat.iswindows:

    # Python 3 on Windows issues a DeprecationWarning about using the bytes

    # API when os.getcwdb() is called.

    #

    # Additionally, py3.8+ uppercases the drive letter when calling

    # os.path.realpath(), which is used on ``repo.root``.  Since those

    # strings are compared in various places as simple strings, also call

    # realpath here.  See https://bugs.python.org/issue40368

    #

    # However this is not reliable, so lets explicitly make this drive

    # letter upper case.

    #

    # note: we should consider dropping realpath here since it seems to

    # change the semantic of `getcwd`.

    def getcwd():

        cwd = os.getcwd()  # re-exports

        cwd = os.path.realpath(cwd)

        cwd = strtolocal(cwd)

        if DRIVE_RE.match(cwd):

            cwd = cwd[0:1].upper() + cwd[1:]

        return cwd

else:

    getcwd = os.getcwdb  # re-exports

# How to treat ambiguous-width characters. Set to 'wide' to treat as wide.

_wide = _sysstr(

    environ.get(b"HGENCODINGAMBIGUOUS", b"narrow") == b"wide"

    and b"WFA"

    or b"WF"

)

    """Find the column width of a string for display in the local encoding"""

    return ucolwidth(s.decode(_sysstr(encoding), 'replace'))

    """Find the column width of a Unicode string for display"""

    eaw = getattr(unicodedata, 'east_asian_width', None)

    if eaw is not None:

        return sum([eaw(c) in _wide and 2 or 1 for c in d])

    return len(d)

    """Use colwidth to find a c-column substring of s starting at byte

    index start"""

    for x in range(start + c, len(s)):

        t = s[start:x]

        if colwidth(t) == c:

            return t

    raise ValueError('substring not found')

    """Trim string 's' to at most 'width' columns (including 'ellipsis').

    If 'leftside' is True, left side of string 's' is trimmed.

    'ellipsis' is always placed at trimmed side.

    >>> from .node import bin

    >>> def bprint(s):

    ...     print(pycompat.sysstr(s))

    >>> ellipsis = b'+++'

    >>> from . import encoding

    >>> encoding.encoding = b'utf-8'

    >>> t = b'1234567890'

    >>> bprint(trim(t, 12, ellipsis=ellipsis))

    1234567890

    >>> bprint(trim(t, 10, ellipsis=ellipsis))

    1234567890

    >>> bprint(trim(t, 8, ellipsis=ellipsis))

    12345+++

    >>> bprint(trim(t, 8, ellipsis=ellipsis, leftside=True))

    +++67890

    >>> bprint(trim(t, 8))

    12345678

    >>> bprint(trim(t, 8, leftside=True))

    34567890

    >>> bprint(trim(t, 3, ellipsis=ellipsis))

    +++

    >>> bprint(trim(t, 1, ellipsis=ellipsis))

    +

    >>> u = u'\u3042\u3044\u3046\u3048\u304a' # 2 x 5 = 10 columns

    >>> t = u.encode(pycompat.sysstr(encoding.encoding))

    >>> bprint(trim(t, 12, ellipsis=ellipsis))

    \xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\xe3\x81\x88\xe3\x81\x8a

    >>> bprint(trim(t, 10, ellipsis=ellipsis))

    \xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\xe3\x81\x88\xe3\x81\x8a

    >>> bprint(trim(t, 8, ellipsis=ellipsis))

    \xe3\x81\x82\xe3\x81\x84+++

    >>> bprint(trim(t, 8, ellipsis=ellipsis, leftside=True))

    +++\xe3\x81\x88\xe3\x81\x8a

    >>> bprint(trim(t, 5))

    \xe3\x81\x82\xe3\x81\x84

    >>> bprint(trim(t, 5, leftside=True))

    \xe3\x81\x88\xe3\x81\x8a

    >>> bprint(trim(t, 4, ellipsis=ellipsis))

    +++

    >>> bprint(trim(t, 4, ellipsis=ellipsis, leftside=True))

    +++

    >>> t = bin(b'112233445566778899aa') # invalid byte sequence

    >>> bprint(trim(t, 12, ellipsis=ellipsis))

    \x11\x22\x33\x44\x55\x66\x77\x88\x99\xaa

    >>> bprint(trim(t, 10, ellipsis=ellipsis))

    \x11\x22\x33\x44\x55\x66\x77\x88\x99\xaa

    >>> bprint(trim(t, 8, ellipsis=ellipsis))

    \x11\x22\x33\x44\x55+++

    >>> bprint(trim(t, 8, ellipsis=ellipsis, leftside=True))

    +++\x66\x77\x88\x99\xaa

    >>> bprint(trim(t, 8))

    \x11\x22\x33\x44\x55\x66\x77\x88

    >>> bprint(trim(t, 8, leftside=True))

    \x33\x44\x55\x66\x77\x88\x99\xaa

    >>> bprint(trim(t, 3, ellipsis=ellipsis))

    +++

    >>> bprint(trim(t, 1, ellipsis=ellipsis))

    +

    """

    try: