summaryrefslogtreecommitdiffhomepage
path: root/libs/click/parser.py
blob: 5fa7adfac842bfa5689fd1a41ae4017be1ebff6f (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
"""
This module started out as largely a copy paste from the stdlib's
optparse module with the features removed that we do not need from
optparse because we implement them in Click on a higher level (for
instance type handling, help formatting and a lot more).

The plan is to remove more and more from here over time.

The reason this is a different module and not optparse from the stdlib
is that there are differences in 2.x and 3.x about the error messages
generated and optparse in the stdlib uses gettext for no good reason
and might cause us issues.

Click uses parts of optparse written by Gregory P. Ward and maintained
by the Python Software Foundation. This is limited to code in parser.py.

Copyright 2001-2006 Gregory P. Ward. All rights reserved.
Copyright 2002-2006 Python Software Foundation. All rights reserved.
"""
# This code uses parts of optparse written by Gregory P. Ward and
# maintained by the Python Software Foundation.
# Copyright 2001-2006 Gregory P. Ward
# Copyright 2002-2006 Python Software Foundation
import typing as t
from collections import deque
from gettext import gettext as _
from gettext import ngettext

from .exceptions import BadArgumentUsage
from .exceptions import BadOptionUsage
from .exceptions import NoSuchOption
from .exceptions import UsageError

if t.TYPE_CHECKING:
    import typing_extensions as te
    from .core import Argument as CoreArgument
    from .core import Context
    from .core import Option as CoreOption
    from .core import Parameter as CoreParameter

V = t.TypeVar("V")

# Sentinel value that indicates an option was passed as a flag without a
# value but is not a flag option. Option.consume_value uses this to
# prompt or use the flag_value.
_flag_needs_value = object()


def _unpack_args(
    args: t.Sequence[str], nargs_spec: t.Sequence[int]
) -> t.Tuple[t.Sequence[t.Union[str, t.Sequence[t.Optional[str]], None]], t.List[str]]:
    """Given an iterable of arguments and an iterable of nargs specifications,
    it returns a tuple with all the unpacked arguments at the first index
    and all remaining arguments as the second.

    The nargs specification is the number of arguments that should be consumed
    or `-1` to indicate that this position should eat up all the remainders.

    Missing items are filled with `None`.
    """
    args = deque(args)
    nargs_spec = deque(nargs_spec)
    rv: t.List[t.Union[str, t.Tuple[t.Optional[str], ...], None]] = []
    spos: t.Optional[int] = None

    def _fetch(c: "te.Deque[V]") -> t.Optional[V]:
        try:
            if spos is None:
                return c.popleft()
            else:
                return c.pop()
        except IndexError:
            return None

    while nargs_spec:
        nargs = _fetch(nargs_spec)

        if nargs is None:
            continue

        if nargs == 1:
            rv.append(_fetch(args))
        elif nargs > 1:
            x = [_fetch(args) for _ in range(nargs)]

            # If we're reversed, we're pulling in the arguments in reverse,
            # so we need to turn them around.
            if spos is not None:
                x.reverse()

            rv.append(tuple(x))
        elif nargs < 0:
            if spos is not None:
                raise TypeError("Cannot have two nargs < 0")

            spos = len(rv)
            rv.append(None)

    # spos is the position of the wildcard (star).  If it's not `None`,
    # we fill it with the remainder.
    if spos is not None:
        rv[spos] = tuple(args)
        args = []
        rv[spos + 1 :] = reversed(rv[spos + 1 :])

    return tuple(rv), list(args)


def split_opt(opt: str) -> t.Tuple[str, str]:
    first = opt[:1]
    if first.isalnum():
        return "", opt
    if opt[1:2] == first:
        return opt[:2], opt[2:]
    return first, opt[1:]


def normalize_opt(opt: str, ctx: t.Optional["Context"]) -> str:
    if ctx is None or ctx.token_normalize_func is None:
        return opt
    prefix, opt = split_opt(opt)
    return f"{prefix}{ctx.token_normalize_func(opt)}"


def split_arg_string(string: str) -> t.List[str]:
    """Split an argument string as with :func:`shlex.split`, but don't
    fail if the string is incomplete. Ignores a missing closing quote or
    incomplete escape sequence and uses the partial token as-is.

    .. code-block:: python

        split_arg_string("example 'my file")
        ["example", "my file"]

        split_arg_string("example my\\")
        ["example", "my"]

    :param string: String to split.
    """
    import shlex

    lex = shlex.shlex(string, posix=True)
    lex.whitespace_split = True
    lex.commenters = ""
    out = []

    try:
        for token in lex:
            out.append(token)
    except ValueError:
        # Raised when end-of-string is reached in an invalid state. Use
        # the partial token as-is. The quote or escape character is in
        # lex.state, not lex.token.
        out.append(lex.token)

    return out


class Option:
    def __init__(
        self,
        obj: "CoreOption",
        opts: t.Sequence[str],
        dest: t.Optional[str],
        action: t.Optional[str] = None,
        nargs: int = 1,
        const: t.Optional[t.Any] = None,
    ):
        self._short_opts = []
        self._long_opts = []
        self.prefixes: t.Set[str] = set()

        for opt in opts:
            prefix, value = split_opt(opt)
            if not prefix:
                raise ValueError(f"Invalid start character for option ({opt})")
            self.prefixes.add(prefix[0])
            if len(prefix) == 1 and len(value) == 1:
                self._short_opts.append(opt)
            else:
                self._long_opts.append(opt)
                self.prefixes.add(prefix)

        if action is None:
            action = "store"

        self.dest = dest
        self.action = action
        self.nargs = nargs
        self.const = const
        self.obj = obj

    @property
    def takes_value(self) -> bool:
        return self.action in ("store", "append")

    def process(self, value: t.Any, state: "ParsingState") -> None:
        if self.action == "store":
            state.opts[self.dest] = value  # type: ignore
        elif self.action == "store_const":
            state.opts[self.dest] = self.const  # type: ignore
        elif self.action == "append":
            state.opts.setdefault(self.dest, []).append(value)  # type: ignore
        elif self.action == "append_const":
            state.opts.setdefault(self.dest, []).append(self.const)  # type: ignore
        elif self.action == "count":
            state.opts[self.dest] = state.opts.get(self.dest, 0) + 1  # type: ignore
        else:
            raise ValueError(f"unknown action '{self.action}'")
        state.order.append(self.obj)


class Argument:
    def __init__(self, obj: "CoreArgument", dest: t.Optional[str], nargs: int = 1):
        self.dest = dest
        self.nargs = nargs
        self.obj = obj

    def process(
        self,
        value: t.Union[t.Optional[str], t.Sequence[t.Optional[str]]],
        state: "ParsingState",
    ) -> None:
        if self.nargs > 1:
            assert value is not None
            holes = sum(1 for x in value if x is None)
            if holes == len(value):
                value = None
            elif holes != 0:
                raise BadArgumentUsage(
                    _("Argument {name!r} takes {nargs} values.").format(
                        name=self.dest, nargs=self.nargs
                    )
                )

        if self.nargs == -1 and self.obj.envvar is not None and value == ():
            # Replace empty tuple with None so that a value from the
            # environment may be tried.
            value = None

        state.opts[self.dest] = value  # type: ignore
        state.order.append(self.obj)


class ParsingState:
    def __init__(self, rargs: t.List[str]) -> None:
        self.opts: t.Dict[str, t.Any] = {}
        self.largs: t.List[str] = []
        self.rargs = rargs
        self.order: t.List["CoreParameter"] = []


class OptionParser:
    """The option parser is an internal class that is ultimately used to
    parse options and arguments.  It's modelled after optparse and brings
    a similar but vastly simplified API.  It should generally not be used
    directly as the high level Click classes wrap it for you.

    It's not nearly as extensible as optparse or argparse as it does not
    implement features that are implemented on a higher level (such as
    types or defaults).

    :param ctx: optionally the :class:`~click.Context` where this parser
                should go with.
    """

    def __init__(self, ctx: t.Optional["Context"] = None) -> None:
        #: The :class:`~click.Context` for this parser.  This might be
        #: `None` for some advanced use cases.
        self.ctx = ctx
        #: This controls how the parser deals with interspersed arguments.
        #: If this is set to `False`, the parser will stop on the first
        #: non-option.  Click uses this to implement nested subcommands
        #: safely.
        self.allow_interspersed_args: bool = True
        #: This tells the parser how to deal with unknown options.  By
        #: default it will error out (which is sensible), but there is a
        #: second mode where it will ignore it and continue processing
        #: after shifting all the unknown options into the resulting args.
        self.ignore_unknown_options: bool = False

        if ctx is not None:
            self.allow_interspersed_args = ctx.allow_interspersed_args
            self.ignore_unknown_options = ctx.ignore_unknown_options

        self._short_opt: t.Dict[str, Option] = {}
        self._long_opt: t.Dict[str, Option] = {}
        self._opt_prefixes = {"-", "--"}
        self._args: t.List[Argument] = []

    def add_option(
        self,
        obj: "CoreOption",
        opts: t.Sequence[str],
        dest: t.Optional[str],
        action: t.Optional[str] = None,
        nargs: int = 1,
        const: t.Optional[t.Any] = None,
    ) -> None:
        """Adds a new option named `dest` to the parser.  The destination
        is not inferred (unlike with optparse) and needs to be explicitly
        provided.  Action can be any of ``store``, ``store_const``,
        ``append``, ``append_const`` or ``count``.

        The `obj` can be used to identify the option in the order list
        that is returned from the parser.
        """
        opts = [normalize_opt(opt, self.ctx) for opt in opts]
        option = Option(obj, opts, dest, action=action, nargs=nargs, const=const)
        self._opt_prefixes.update(option.prefixes)
        for opt in option._short_opts:
            self._short_opt[opt] = option
        for opt in option._long_opts:
            self._long_opt[opt] = option

    def add_argument(
        self, obj: "CoreArgument", dest: t.Optional[str], nargs: int = 1
    ) -> None:
        """Adds a positional argument named `dest` to the parser.

        The `obj` can be used to identify the option in the order list
        that is returned from the parser.
        """
        self._args.append(Argument(obj, dest=dest, nargs=nargs))

    def parse_args(
        self, args: t.List[str]
    ) -> t.Tuple[t.Dict[str, t.Any], t.List[str], t.List["CoreParameter"]]:
        """Parses positional arguments and returns ``(values, args, order)``
        for the parsed options and arguments as well as the leftover
        arguments if there are any.  The order is a list of objects as they
        appear on the command line.  If arguments appear multiple times they
        will be memorized multiple times as well.
        """
        state = ParsingState(args)
        try:
            self._process_args_for_options(state)
            self._process_args_for_args(state)
        except UsageError:
            if self.ctx is None or not self.ctx.resilient_parsing:
                raise
        return state.opts, state.largs, state.order

    def _process_args_for_args(self, state: ParsingState) -> None:
        pargs, args = _unpack_args(
            state.largs + state.rargs, [x.nargs for x in self._args]
        )

        for idx, arg in enumerate(self._args):
            arg.process(pargs[idx], state)

        state.largs = args
        state.rargs = []

    def _process_args_for_options(self, state: ParsingState) -> None:
        while state.rargs:
            arg = state.rargs.pop(0)
            arglen = len(arg)
            # Double dashes always handled explicitly regardless of what
            # prefixes are valid.
            if arg == "--":
                return
            elif arg[:1] in self._opt_prefixes and arglen > 1:
                self._process_opts(arg, state)
            elif self.allow_interspersed_args:
                state.largs.append(arg)
            else:
                state.rargs.insert(0, arg)
                return

        # Say this is the original argument list:
        # [arg0, arg1, ..., arg(i-1), arg(i), arg(i+1), ..., arg(N-1)]
        #                            ^
        # (we are about to process arg(i)).
        #
        # Then rargs is [arg(i), ..., arg(N-1)] and largs is a *subset* of
        # [arg0, ..., arg(i-1)] (any options and their arguments will have
        # been removed from largs).
        #
        # The while loop will usually consume 1 or more arguments per pass.
        # If it consumes 1 (eg. arg is an option that takes no arguments),
        # then after _process_arg() is done the situation is:
        #
        #   largs = subset of [arg0, ..., arg(i)]
        #   rargs = [arg(i+1), ..., arg(N-1)]
        #
        # If allow_interspersed_args is false, largs will always be
        # *empty* -- still a subset of [arg0, ..., arg(i-1)], but
        # not a very interesting subset!

    def _match_long_opt(
        self, opt: str, explicit_value: t.Optional[str], state: ParsingState
    ) -> None:
        if opt not in self._long_opt:
            from difflib import get_close_matches

            possibilities = get_close_matches(opt, self._long_opt)
            raise NoSuchOption(opt, possibilities=possibilities, ctx=self.ctx)

        option = self._long_opt[opt]
        if option.takes_value:
            # At this point it's safe to modify rargs by injecting the
            # explicit value, because no exception is raised in this
            # branch.  This means that the inserted value will be fully
            # consumed.
            if explicit_value is not None:
                state.rargs.insert(0, explicit_value)

            value = self._get_value_from_state(opt, option, state)

        elif explicit_value is not None:
            raise BadOptionUsage(
                opt, _("Option {name!r} does not take a value.").format(name=opt)
            )

        else:
            value = None

        option.process(value, state)

    def _match_short_opt(self, arg: str, state: ParsingState) -> None:
        stop = False
        i = 1
        prefix = arg[0]
        unknown_options = []

        for ch in arg[1:]:
            opt = normalize_opt(f"{prefix}{ch}", self.ctx)
            option = self._short_opt.get(opt)
            i += 1

            if not option:
                if self.ignore_unknown_options:
                    unknown_options.append(ch)
                    continue
                raise NoSuchOption(opt, ctx=self.ctx)
            if option.takes_value:
                # Any characters left in arg?  Pretend they're the
                # next arg, and stop consuming characters of arg.
                if i < len(arg):
                    state.rargs.insert(0, arg[i:])
                    stop = True

                value = self._get_value_from_state(opt, option, state)

            else:
                value = None

            option.process(value, state)

            if stop:
                break

        # If we got any unknown options we recombine the string of the
        # remaining options and re-attach the prefix, then report that
        # to the state as new larg.  This way there is basic combinatorics
        # that can be achieved while still ignoring unknown arguments.
        if self.ignore_unknown_options and unknown_options:
            state.largs.append(f"{prefix}{''.join(unknown_options)}")

    def _get_value_from_state(
        self, option_name: str, option: Option, state: ParsingState
    ) -> t.Any:
        nargs = option.nargs

        if len(state.rargs) < nargs:
            if option.obj._flag_needs_value:
                # Option allows omitting the value.
                value = _flag_needs_value
            else:
                raise BadOptionUsage(
                    option_name,
                    ngettext(
                        "Option {name!r} requires an argument.",
                        "Option {name!r} requires {nargs} arguments.",
                        nargs,
                    ).format(name=option_name, nargs=nargs),
                )
        elif nargs == 1:
            next_rarg = state.rargs[0]

            if (
                option.obj._flag_needs_value
                and isinstance(next_rarg, str)
                and next_rarg[:1] in self._opt_prefixes
                and len(next_rarg) > 1
            ):
                # The next arg looks like the start of an option, don't
                # use it as the value if omitting the value is allowed.
                value = _flag_needs_value
            else:
                value = state.rargs.pop(0)
        else:
            value = tuple(state.rargs[:nargs])
            del state.rargs[:nargs]

        return value

    def _process_opts(self, arg: str, state: ParsingState) -> None:
        explicit_value = None
        # Long option handling happens in two parts.  The first part is
        # supporting explicitly attached values.  In any case, we will try
        # to long match the option first.
        if "=" in arg:
            long_opt, explicit_value = arg.split("=", 1)
        else:
            long_opt = arg
        norm_long_opt = normalize_opt(long_opt, self.ctx)

        # At this point we will match the (assumed) long option through
        # the long option matching code.  Note that this allows options
        # like "-foo" to be matched as long options.
        try:
            self._match_long_opt(norm_long_opt, explicit_value, state)
        except NoSuchOption:
            # At this point the long option matching failed, and we need
            # to try with short options.  However there is a special rule
            # which says, that if we have a two character options prefix
            # (applies to "--foo" for instance), we do not dispatch to the
            # short option code and will instead raise the no option
            # error.
            if arg[:2] not in self._opt_prefixes:
                self._match_short_opt(arg, state)
                return

            if not self.ignore_unknown_options:
                raise

            state.largs.append(arg)