usage_argv/lib.rs
1//! A zero-allocation argv parser for [usage](https://usage.jdx.dev) specs.
2//!
3//! This crate implements the binding rules of [the argv grammar]: which token
4//! becomes which flag or argument, when a word selects a subcommand, and what
5//! is an error. It does so without building a command tree, without allocating,
6//! and in one pass.
7//!
8//! It is the runtime half of a compiled parser. The tables it reads are meant to
9//! be emitted by a derive macro as `static` data, so that starting a parse costs
10//! nothing at all: there is no construction step to pay for, only the walk over
11//! `argv`.
12//!
13//! # Shape of the API
14//!
15//! Parsing yields [`Event`]s rather than a map. A map would have to allocate,
16//! and would then have to be read back out again — whereas generated code can
17//! assign an event straight into a struct field. This is the same reason serde
18//! deserializes into your type instead of into a `Value`.
19//!
20//! ```
21//! use usage_argv::{Arg, Command, Event, Flag, Parser};
22//!
23//! static FORCE: Flag = Flag { key: 0, longs: &["force"], shorts: b"f", ..Flag::BOOL };
24//! static FILE: Arg = Arg { key: 1, ..Arg::REQUIRED };
25//! static ROOT: Command = Command {
26//! name: "ex",
27//! flags: &[&FORCE],
28//! args: &[&FILE],
29//! ..Command::EMPTY
30//! };
31//!
32//! let argv = ["--force", "a.txt"].map(std::ffi::OsStr::new);
33//! let mut parser = Parser::new(&ROOT, &argv);
34//!
35//! let mut force = false;
36//! let mut file = None;
37//! while let Some(event) = parser.next_event() {
38//! match event.expect("valid command line") {
39//! Event::Flag { flag, .. } if flag.key == 0 => force = true,
40//! Event::Arg { value, .. } => file = Some(value),
41//! _ => {}
42//! }
43//! }
44//! assert!(force);
45//! assert_eq!(file, Some(&b"a.txt"[..]));
46//! ```
47//!
48//! # Values are bytes
49//!
50//! An [`Event`] carries `&[u8]`, borrowed from `argv`. Converting to `&str` is
51//! the caller's step ([`as_str`]), and it is the right place for the only
52//! failure a value can have: a command line that is not valid UTF-8 still
53//! *parses* — flags match, subcommands route — and only the values that are
54//! actually looked at can fail to convert.
55//!
56//! Slicing an `OsStr` into `&str` pieces safely is not possible without
57//! allocating or `unsafe`. Bytes are what is left, and they turn out to be the
58//! honest interface anyway.
59//!
60//! The reverse conversion is [`os_string_from_bytes`], which lets a `PathBuf`
61//! field hold a filename that is not UTF-8 rather than a mangled copy of one. On
62//! Unix that is lossless and safe; on Windows, where WTF-8 makes it partial, a
63//! value that will not convert is reported. Either way this crate contains no
64//! `unsafe`, which a conversion that guessed would have cost.
65//!
66//! # What this crate does not do
67//!
68//! Only binding. Required-ness, `choices`, `env` fallback, defaults, `var_min`
69//! and `var_max` are all decided *after* the last token is read, and they need to
70//! know a value's type, so they belong to the layer that owns the target struct.
71//! Keeping them out is what makes this loop small.
72//!
73//! # Features
74//!
75//! - `spec` — a parallel tree of cold metadata (help text, choices, defaults,
76//! effects) and a writer that emits it as a usage spec. Off by default: a
77//! successful parse never reads any of it, so a CLI that only wants a parser
78//! should not compile it.
79//! - `complete` — answering a partial command line ([`complete`]), the shell
80//! scripts that ask ([`script`]), and putting one of those scripts where its
81//! shell will look for it ([`install`]). Installing ships with the scripts
82//! rather than behind a gate of its own: a script a CLI still has to tell its
83//! users to redirect by hand is the unfinished half of shipping one.
84//!
85//! [the argv grammar]: https://usage.jdx.dev/spec/argv
86
87#![forbid(unsafe_code)]
88
89/// Terminate at the compiled CLI entry-point boundary.
90///
91/// Kept in the runtime rather than expanded into an adopter crate so a project that
92/// forbids direct `std::process::exit` calls does not attribute the derive's process
93/// boundary to application code. `Cli::parse_from*` continues to return errors.
94#[doc(hidden)]
95#[allow(clippy::disallowed_methods)]
96pub fn __usage_process_exit(status: i32) -> ! {
97 std::process::exit(status)
98}
99
100use std::ffi::{OsStr, OsString};
101
102/// A value's shell-native completion class for `#[usage(value_hint = ...)]`.
103///
104/// This lives in the runtime crate so a declaration never needs clap merely to describe what
105/// kind of path a shell should offer. It is metadata only and adds no work to a successful
106/// parse.
107#[derive(Clone, Copy, Debug, Eq, PartialEq)]
108pub enum ValueHint {
109 /// Let the shell use its normal fallback behavior.
110 Unknown,
111 /// No structured hint applies; suppress the shell's path fallback.
112 Other,
113 /// A path to a file.
114 FilePath,
115 /// A path to either a file or a directory.
116 AnyPath,
117 /// A path to a directory.
118 DirPath,
119 /// A path to an executable file.
120 ExecutablePath,
121 /// A command name, resolved through the shell's command table and `PATH`.
122 CommandName,
123 /// One string containing a command and any arguments.
124 CommandString,
125 /// A trailing argv vector: complete the first value as a command, then its arguments.
126 CommandWithArguments,
127 /// A local operating-system user name.
128 Username,
129 /// A host name known to the shell or operating system.
130 Hostname,
131 /// A web address. This suppresses path fallback but offers no finite candidate set.
132 Url,
133 /// An email address. This suppresses path fallback but offers no finite candidate set.
134 EmailAddress,
135}
136
137#[cfg(feature = "complete")]
138pub mod complete;
139#[cfg(feature = "diagnostics")]
140pub mod diagnostic;
141#[cfg(feature = "spec")]
142pub mod embedded;
143#[cfg(feature = "complete")]
144pub mod install;
145#[cfg(feature = "complete")]
146pub mod script;
147
148/// Checks that the `complete` feature is on, with an explanation when it is not.
149///
150/// `#[usage(completion)]` generates code that reaches into [`complete`], which is behind a
151/// feature the *depending* crate enables — a derive cannot turn on a feature of another crate.
152/// Without this, the failure was `unresolved module complete`, which says nothing about the
153/// attribute that caused it.
154#[cfg(feature = "complete")]
155#[macro_export]
156macro_rules! __usage_needs_complete_feature {
157 () => {};
158}
159
160/// See [`__usage_needs_complete_feature`].
161#[cfg(not(feature = "complete"))]
162#[macro_export]
163macro_rules! __usage_needs_complete_feature {
164 () => {
165 ::core::compile_error!(
166 "`#[usage(completion)]` needs usage-argv's `complete` feature. Add it where \
167 usage-argv is depended on: usage-argv = { version = \"…\", features = \
168 [\"spec\", \"complete\"] }"
169 );
170 };
171}
172#[cfg(feature = "spec")]
173pub mod help;
174// Behind no feature: two traits and no code, so there is nothing here for a binary that
175// does not dispatch to pay for, and a hand-written CLI on the bare runtime can use them.
176pub mod run;
177#[cfg(feature = "spec")]
178pub mod spec;
179#[cfg(feature = "spec")]
180pub use spec::{parse_args_from, parse_args_from_argv};
181#[cfg(feature = "spec")]
182pub mod warn;
183
184pub use run::{Run, RunAsync, RunAsyncWith, RunWith};
185
186/// How deep a command tree this parser will descend.
187///
188/// The ancestor chain is kept in a fixed-size array so that a parse allocates
189/// nothing; this is that array's size. mise, the largest usage CLI, is four
190/// levels deep.
191pub const MAX_DEPTH: usize = 16;
192
193/// A command: its flags, its positional arguments, and its subcommands.
194///
195/// Every field is a borrowed slice so that a derive can emit the whole tree as
196/// `static` data. Use `..Command::EMPTY` to fill in the parts you do not need.
197#[derive(Debug, Clone, Copy, PartialEq, Eq)]
198pub struct Command<'a> {
199 /// The canonical name, used to select this command.
200 pub name: &'a str,
201 /// Alternative names that also select it.
202 pub aliases: &'a [&'a str],
203 pub flags: &'a [&'a Flag<'a>],
204 /// Positional arguments, in the order they are filled.
205 pub args: &'a [&'a Arg<'a>],
206 /// A repeatable group of scoped flags and positional arguments, if this command has one.
207 pub clause: ::core::option::Option<Clause<'a>>,
208 pub subcommands: &'a [&'a Command<'a>],
209 /// Where a word goes when it names no subcommand of this one.
210 ///
211 /// The spec's `default_subcommand`. `mise build` means `mise run build`: the word names
212 /// no command, so the parser descends into `run` and lets *`run`* have it — even where
213 /// this command declares an argument of its own, which is what makes the property worth
214 /// having rather than a synonym for a positional.
215 ///
216 /// Applied at most once per parse, so a CLI cannot loop through it, and only where a
217 /// subcommand could still be selected.
218 ///
219 /// Resolve it with [`find_subcommand`], which turns a name that no subcommand answers to
220 /// into a compile error.
221 pub default_subcommand: ::core::option::Option<&'a Command<'a>>,
222 /// Look ahead past parent/default flags before implicitly selecting the default.
223 /// Explicit siblings win; parent-only flags retain their ordinary meaning.
224 pub default_subcommand_flags: bool,
225 /// Whether an unmatched word is forwarded as an external command plus the rest of argv.
226 ///
227 /// clap's `allow_external_subcommands`. Known subcommands still win; a
228 /// [`default_subcommand`](Self::default_subcommand) still catches first. Once the
229 /// unmatched word is taken, remaining tokens — including `--help` — are not parsed
230 /// as this command's flags.
231 pub external_subcommand: bool,
232 /// Show this command's help when no argv token follows its name.
233 ///
234 /// This is clap's `arg_required_else_help`. It deliberately observes argv rather than
235 /// bound values: an environment variable or default may fill a field, but neither means
236 /// the user supplied an argument to this invocation.
237 pub arg_required_else_help: bool,
238 /// Selecting a subcommand suppresses this command's required arguments.
239 pub subcommand_negates_reqs: bool,
240 /// Once this command binds a flag or positional, selecting one of its
241 /// subcommands is an error.
242 pub args_conflicts_with_subcommands: bool,
243 /// Let a known subcommand interrupt a variadic argument that would otherwise consume it.
244 pub subcommand_precedence_over_arg: bool,
245 /// Let a later required positional take a word while an earlier optional positional
246 /// remains empty.
247 pub allow_missing_positional: bool,
248 /// Disable delimiter splitting for positional values after `--` or on an
249 /// automatic trailing argument. Inherited by subcommands.
250 pub dont_delimit_trailing_values: bool,
251 /// What an unrecognized flag-like token means here, or `None` to keep whatever the
252 /// enclosing command said. See [`UnknownFlags`].
253 ///
254 /// Inherited rather than resolved per command, which is what usage-lib does — its
255 /// `effective_unknown_flags` walks outward from the command that ran and falls back to
256 /// the spec's. Resolving it in the tables instead was possible only for a builder that
257 /// can see the whole tree: a derive expands one struct at a time and cannot see its
258 /// parent, so `#[usage(unknown_flags = "error")]` on the root reached the root alone and
259 /// a subcommand had no way to say it at all.
260 ///
261 /// The parser carries the effective value down as it descends, so a command that states
262 /// nothing costs nothing.
263 pub unknown_flags: ::core::option::Option<UnknownFlags>,
264 /// Whether this command answers to `--version` and `-V`.
265 ///
266 /// Set on the root, and only when the CLI declares a version: clap adds the flag exactly
267 /// then, and a `--version` that answers with nothing is worse than one that is not there.
268 /// A field rather than a rule about depth, so a CLI that wants it on a subcommand — clap's
269 /// `propagate_version` — has somewhere to say so.
270 pub version: bool,
271 /// Do not synthesize `--help` and `-h` for this command.
272 pub disable_help_flag: bool,
273 /// Do not synthesize the `help` subcommand route for this command.
274 pub disable_help_subcommand: bool,
275 /// Do not synthesize `--version` and `-V` for this command.
276 pub disable_version_flag: bool,
277 /// Caller-assigned identifier, echoed back in [`Event::Command`].
278 ///
279 /// Wide enough for a derive to make these unique without coordination: two
280 /// macro expansions cannot see each other, so the generated keys carry a hash
281 /// of the type they came from in the high half and a per-type index in the low
282 /// half. A parse dispatches on this, so a collision would bind the wrong field
283 /// — [`Spec::to_kdl`](crate::spec::Spec::to_kdl) checks the tree for duplicates
284 /// in debug builds.
285 pub key: u64,
286}
287
288impl Command<'_> {
289 /// A command with nothing declared, for use with struct update syntax.
290 pub const EMPTY: Command<'static> = Command {
291 name: "",
292 aliases: &[],
293 flags: &[],
294 args: &[],
295 clause: ::core::option::Option::None,
296 subcommands: &[],
297 default_subcommand: ::core::option::Option::None,
298 default_subcommand_flags: false,
299 external_subcommand: false,
300 arg_required_else_help: false,
301 subcommand_negates_reqs: false,
302 args_conflicts_with_subcommands: false,
303 subcommand_precedence_over_arg: false,
304 allow_missing_positional: false,
305 dont_delimit_trailing_values: false,
306 unknown_flags: ::core::option::Option::None,
307 version: false,
308 disable_help_flag: false,
309 disable_help_subcommand: false,
310 disable_version_flag: false,
311 key: 0,
312 };
313}
314
315/// A repeatable group of scoped flags and positional arguments.
316#[derive(Debug, Clone, Copy, PartialEq, Eq)]
317pub struct Clause<'a> {
318 pub key: u64,
319 pub name: &'a str,
320 /// An explicit token that starts the next instance. When absent, consuming the
321 /// clause's sole required positional completes the current instance.
322 pub separator: ::core::option::Option<&'a [u8]>,
323 pub flags: &'a [&'a Flag<'a>],
324 pub args: &'a [&'a Arg<'a>],
325}
326
327/// Basename of argv[0] for a multicall CLI: last path component, with a trailing
328/// `.exe` stripped so Windows and Unix agree.
329pub fn multicall_basename(argv0: &str) -> &str {
330 let name = argv0.rsplit(['/', '\\']).next().unwrap_or(argv0);
331 match name.get(name.len().saturating_sub(4)..) {
332 Some(ext) if ext.eq_ignore_ascii_case(".exe") => &name[..name.len() - 4],
333 _ => name,
334 }
335}
336
337/// The applet name to parse as the first word, when argv[0] is not the dispatcher.
338///
339/// `None` means a dispatcher invocation (`busybox ls`): skip argv[0] and parse the
340/// rest. `Some` is a symlink invocation (`ls -l`): inject the basename.
341pub fn multicall_applet<'a>(argv0: &'a str, name: &str, bin: Option<&str>) -> Option<&'a str> {
342 let base = multicall_basename(argv0);
343 if !name.is_empty() && base == multicall_basename(name) {
344 return None;
345 }
346 if let Some(bin) = bin {
347 if !bin.is_empty() && base == multicall_basename(bin) {
348 return None;
349 }
350 }
351 Some(base)
352}
353
354/// Resolved identity of a derive-generated binding type.
355#[derive(Clone, Copy)]
356pub struct BindingType(pub fn() -> &'static str);
357
358impl BindingType {
359 pub fn name(self) -> &'static str {
360 (self.0)()
361 }
362}
363
364impl ::core::fmt::Debug for BindingType {
365 fn fmt(&self, f: &mut ::core::fmt::Formatter<'_>) -> ::core::fmt::Result {
366 f.debug_tuple("BindingType").field(&self.name()).finish()
367 }
368}
369
370impl PartialEq for BindingType {
371 fn eq(&self, other: &Self) -> bool {
372 self.name() == other.name()
373 }
374}
375
376impl Eq for BindingType {}
377
378/// A flag, addressed by any of its long or short forms.
379#[derive(Debug, Clone, Copy, PartialEq, Eq)]
380pub struct Flag<'a> {
381 /// Caller-assigned identifier, echoed back in [`Event::Flag`]. This is how
382 /// generated code knows which field to assign without any string comparison.
383 /// See [`Command::key`] on why it is this wide.
384 pub key: u64,
385 /// Compatibility key for mirroring a redeclared child global into an ancestor field.
386 ///
387 /// Zero means no typed binding contract is declared. Derive-generated tables hash the
388 /// binding shape and portable metadata so only equivalent bindings receive the same event.
389 pub binding_key: u64,
390 /// Resolved Rust value type for a derive-generated binding.
391 ///
392 /// This is separate from [`Self::binding_key`] because token spellings are not type
393 /// identities: an imported alias and a fully qualified path can name the same type.
394 pub binding_type: Option<BindingType>,
395 /// Unused by binding, kept so a table entry can carry its own name for
396 /// diagnostics.
397 pub name: &'a str,
398 /// Long forms, written without the leading `--`.
399 pub longs: &'a [&'a str],
400 /// Short forms, as single bytes.
401 ///
402 /// **Should be ASCII.** A cluster like `-xyz` is walked one byte at a time, so a
403 /// non-ASCII short can never be matched, and the remainder after a value-taking one —
404 /// which becomes its value — would begin in the middle of a character.
405 /// `#[derive(Cli)]` rejects a non-ASCII `short`; a table written by hand should keep to
406 /// it. Nothing is unsound if it does not: the value would simply be cut in a place that
407 /// makes no sense, and on Windows would then fail to convert.
408 pub shorts: &'a [u8],
409 /// A long form that sets the flag to false, written without the `--`.
410 pub negate: Option<&'a str>,
411 /// Whether the flag takes a value.
412 pub takes_value: bool,
413 /// Whether one occurrence of this flag keeps taking values, until a flag-like
414 /// token or the end of the command line.
415 ///
416 /// This is the spec's variadic flag *argument* (`--include <pattern>...`). It
417 /// is not the spec's flag-level `var=#true`, which means the flag may be
418 /// repeated and takes one value each time — repetition needs nothing from the
419 /// parser, since it already reports every occurrence separately. Conflating
420 /// the two makes a merely repeatable flag greedy enough to eat a positional.
421 pub variadic: bool,
422 /// How many values one variadic occurrence may take, after which the next word
423 /// belongs to whatever comes next.
424 ///
425 /// Only for [`variadic`](Self::variadic). A merely repeatable flag — the spec's
426 /// `var=#true` — is bounded on how many times it was *given*, which no single token
427 /// can decide, so that bound stays with the metadata and is checked after the parse.
428 pub var_max: ::core::option::Option<u32>,
429 /// The byte that makes one word several values, if the flag declares one.
430 ///
431 /// Here rather than with the metadata for the same reason [`var_max`](Self::var_max)
432 /// is: it decides *where* a word lands. A bound counts values, and a delimiter is what
433 /// makes a word stop being one of them — `--include a,b,c` is three, so a `var_max` of
434 /// two is already past its bound on the single word it was entitled to take. Binding
435 /// cannot count without it.
436 pub delimiter: ::core::option::Option<u8>,
437 /// Whether a detached value may itself look like a flag.
438 ///
439 /// The default is to refuse: `--jobs --force` is far more likely a forgotten
440 /// value than a jobs of `"--force"`. Declared, the next token is taken
441 /// whatever it looks like — including `--` — which is clap's
442 /// `allow_hyphen_values` and the spec's property of the same name. A variadic
443 /// occurrence still stops collecting at a later flag-like token, so a second
444 /// occurrence of the flag is not eaten as a value.
445 pub allow_hyphen_values: bool,
446 /// Whether a detached value may be a negative number while other flag-like
447 /// tokens still stop collection or report as flags.
448 pub allow_negative_numbers: bool,
449 /// A token that ends one variadic occurrence without becoming a value.
450 pub value_terminator: ::core::option::Option<&'a [u8]>,
451 /// Whether the value must be attached with `=`.
452 ///
453 /// `--flag=value` is accepted and `--flag value` is not, which is clap's
454 /// `require_equals` and the spec's property of the same name. A short's
455 /// attached form (`-i9229`, `-i=9229`) still binds: only the following word
456 /// is refused.
457 pub require_equals: bool,
458 /// Whether this value-taking flag may be present without a value.
459 ///
460 /// A missing value emits the flag event with `value: None`; bindings such as
461 /// `Option<Option<T>>` can therefore distinguish an absent flag from a bare
462 /// flag and from a flag with an explicit value.
463 pub value_optional: bool,
464 /// Whether a boolean long flag accepts an attached `true` or `false` value.
465 ///
466 /// This does not make the flag value-taking in the ordinary sense: a detached
467 /// word is never consumed, and help keeps rendering a switch. Only
468 /// `--flag=true` and `--flag=false` opt into an explicit boolean value.
469 pub bool_value: bool,
470 /// Value used when the flag is present but no value is given.
471 ///
472 /// clap's `default_missing_value` and the spec's `default_missing`. `--color`
473 /// binds this, `--color=never` binds `never`, and an absent flag is not bound.
474 /// Combined with [`Self::require_equals`], a following word is still refused.
475 pub default_missing: ::core::option::Option<&'a [u8]>,
476 /// Whether the flag is recognized by every command beneath the one that
477 /// declares it.
478 pub global: bool,
479 /// Whether this declared flag binds a field or requests a built-in response.
480 pub action: ArgAction,
481}
482
483impl Flag<'_> {
484 /// A value-less flag, for use with struct update syntax.
485 pub const BOOL: Flag<'static> = Flag {
486 key: 0,
487 binding_key: 0,
488 binding_type: None,
489 name: "",
490 longs: &[],
491 shorts: &[],
492 negate: None,
493 takes_value: false,
494 variadic: false,
495 var_max: ::core::option::Option::None,
496 delimiter: ::core::option::Option::None,
497 allow_hyphen_values: false,
498 allow_negative_numbers: false,
499 value_terminator: ::core::option::Option::None,
500 require_equals: false,
501 value_optional: false,
502 bool_value: false,
503 default_missing: ::core::option::Option::None,
504 global: false,
505 action: ArgAction::Set,
506 };
507
508 /// A flag that takes a value, for use with struct update syntax.
509 pub const VALUE: Flag<'static> = Flag {
510 takes_value: true,
511 ..Flag::BOOL
512 };
513}
514
515/// What supplying a declared flag does.
516#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
517pub enum ArgAction {
518 /// Bind the flag to its declared field.
519 #[default]
520 Set,
521 /// Show help, choosing the long form for a long spelling and the short form otherwise.
522 Help,
523 /// Always show short help.
524 HelpShort,
525 /// Always show long help.
526 HelpLong,
527 /// Show long help for this command and every visible descendant.
528 HelpAll,
529 /// Show version information.
530 Version,
531}
532
533/// A positional argument.
534#[derive(Debug, Clone, Copy, PartialEq, Eq)]
535pub struct Arg<'a> {
536 /// Caller-assigned identifier, echoed back in [`Event::Arg`]. See
537 /// [`Command::key`] on why it is this wide.
538 pub key: u64,
539 /// Prefix that classifies this positional independently of declaration order.
540 /// The prefix is removed from the value emitted in [`Event::Arg`].
541 pub sigil: ::core::option::Option<&'a [u8]>,
542 /// Whether post-binding requires this positional to have a value. Kept in the hot
543 /// table because `allow_missing_positional` must reserve words for later required args.
544 pub required: bool,
545 /// Whether this argument keeps taking values once it has one.
546 pub var: bool,
547 /// How many words a variadic may take before the next argument gets the rest.
548 ///
549 /// A bound belongs here, in the table binding reads, rather than with the metadata:
550 /// it decides *where* a word lands, not whether what landed is acceptable. clap's
551 /// `num_args` works the same way, and every spec in the wild is generated from a clap
552 /// command. `u32` rather than `usize` because a CLI that bounds a variadic above four
553 /// billion has other problems, and this table is read on the hot path.
554 pub var_max: ::core::option::Option<u32>,
555 /// The byte that makes one word several values, if the argument declares one.
556 ///
557 /// See [`Flag::delimiter`]: a bound counts values, and only this says how many values a
558 /// word carries.
559 pub delimiter: ::core::option::Option<u8>,
560 /// Whether a negative-number token is accepted as this positional even in
561 /// strict flag mode.
562 pub allow_negative_numbers: bool,
563 /// A token that ends this variadic positional without becoming a value.
564 pub value_terminator: ::core::option::Option<&'a [u8]>,
565 /// This argument's relationship to the `--` separator.
566 pub double_dash: DoubleDash,
567 /// Unused by binding, kept so a table entry can carry its own name for
568 /// diagnostics.
569 pub name: &'a str,
570}
571
572impl Arg<'_> {
573 /// A single-value argument, for use with struct update syntax.
574 pub const REQUIRED: Arg<'static> = Arg {
575 key: 0,
576 sigil: ::core::option::Option::None,
577 required: true,
578 var: false,
579 var_max: ::core::option::Option::None,
580 delimiter: ::core::option::Option::None,
581 allow_negative_numbers: false,
582 value_terminator: ::core::option::Option::None,
583 double_dash: DoubleDash::Optional,
584 name: "",
585 };
586
587 /// A variadic argument, for use with struct update syntax.
588 pub const VAR: Arg<'static> = Arg {
589 var: true,
590 ..Arg::REQUIRED
591 };
592}
593
594/// What to do with a flag-like token that names no flag in scope.
595///
596/// The default is [`UnknownFlags::Value`]: the token carries on to the positional
597/// arguments, because a spec is often parsing a command line whose flags belong to
598/// something else — a wrapped tool, a task script. A CLI that owns all of its
599/// flags declares [`UnknownFlags::Error`] and gets typo detection instead.
600///
601/// Stored per command and already resolved: inheritance is a question for whoever
602/// builds the tables, and answering it at compile time keeps it out of the parse.
603#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
604pub enum UnknownFlags {
605 /// Offer the token to the positionals. If none can take it, it is an
606 /// unexpected argument.
607 #[default]
608 Value,
609 /// Reject the token.
610 Error,
611}
612
613/// How an argument relates to the `--` separator.
614#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
615pub enum DoubleDash {
616 /// Values may appear on either side of a `--`.
617 #[default]
618 Optional,
619 /// Values are accepted only after a `--`.
620 Required,
621 /// A `--` is kept as a value rather than consumed as a separator.
622 Preserve,
623 /// Once the argument takes a value, behave as if a `--` had been given, so
624 /// the rest of the command line is values. A wrapper can then forward flags
625 /// without its caller typing the separator.
626 Automatic,
627}
628
629/// Something the parser bound.
630#[derive(Debug, Clone, Copy, PartialEq, Eq)]
631pub enum Event<'t, 'a, 'v> {
632 /// A subcommand was selected; parsing continues inside it.
633 Command(&'t Command<'t>),
634 /// A flag was given. `value` is `Some` for a flag that takes one, and
635 /// `negated` is true when the flag was set through its `negate` form.
636 Flag {
637 flag: &'t Flag<'t>,
638 value: Option<&'v [u8]>,
639 negated: bool,
640 },
641 /// A word was bound to a positional argument. A variadic argument produces
642 /// one event per value.
643 Arg {
644 arg: &'t Arg<'t>,
645 value: &'v [u8],
646 /// Whether this value should be split by the argument's declared delimiter.
647 delimit: bool,
648 },
649 /// Ended one clause instance and began the next.
650 ClauseSeparator { clause: Clause<'t> },
651 /// An unmatched word was forwarded as an external command: the name, then
652 /// every remaining token, including flags.
653 External { values: &'a [&'v OsStr] },
654}
655
656/// A binding failure.
657///
658/// Carries the offending token so a caller can render a good message, but no
659/// message of its own: rendering belongs to a cold path, and building a string
660/// here would allocate on the way to reporting that nothing was allocated.
661///
662/// `non_exhaustive`, because an error enum grows: a caller matching on it needs a
663/// fallback arm so that recognizing a new failure is never a breaking change.
664// No `Copy`: one variant owns its message. `Clone` stays, and the enum is still 40 bytes
665// because that variant is boxed, so nothing on the hot path grew.
666#[derive(Debug, Clone, PartialEq, Eq)]
667#[non_exhaustive]
668pub enum Error<'t, 'v> {
669 /// A flag-like token matched no flag in scope. `token` is the whole token as
670 /// typed, so a bundle containing an unrecognized letter reports `-fz` rather
671 /// than the letter alone — which is also the unit in which it is rejected.
672 UnknownFlag { token: &'v [u8] },
673 /// A flag that needs a value did not get one, either because the command
674 /// line ended or because the next token was flag-like.
675 MissingFlagValue { flag: &'t Flag<'t> },
676 /// A word arrived with no argument left to hold it.
677 UnexpectedArg { token: &'v [u8] },
678 /// A word was offered to a `double_dash = "required"` argument before any
679 /// `--` had been seen.
680 ArgRequiresDoubleDash { arg: &'t Arg<'t> },
681 /// A subcommand was selected after this command had already bound an argument.
682 SubcommandConflict { subcommand: &'t Command<'t> },
683 /// The command tree is deeper than [`MAX_DEPTH`].
684 TooDeep,
685
686 // The rest are raised *after* the parse, by whoever owns the target type: they
687 // need to know a value's declared type, which the parser deliberately does not.
688 // They share this enum so that a caller has one error to handle rather than two.
689 /// Something the command requires was never given.
690 MissingRequired {
691 /// The flag or argument's name, as the spec calls it.
692 name: &'t str,
693 },
694 /// A flag that is not repeatable was given more than once.
695 DuplicateFlag {
696 /// The flag's name, as the spec calls it.
697 name: &'t str,
698 },
699 /// A value was given that is not among the declared choices.
700 ///
701 /// Carries the choices rather than the offending value: rendering the value means
702 /// owning it, and an error that allocates on a path this crate promises not to
703 /// allocate on would be a poor trade for a better message. Diagnostics are a
704 /// separate layer.
705 InvalidChoice {
706 name: &'t str,
707 choices: &'t [&'t str],
708 },
709 /// Fewer values than `var_min`.
710 VarTooFew {
711 name: &'t str,
712 min: usize,
713 got: usize,
714 },
715 /// More values than `var_max`.
716 VarTooMany {
717 name: &'t str,
718 max: usize,
719 got: usize,
720 },
721 /// Two flags declared to conflict were both given.
722 ///
723 /// Carries both names because either one alone reads as a puzzle: which flag is
724 /// unwelcome depends entirely on what else is on the command line.
725 ConflictingFlags {
726 /// The flag whose declaration names the conflict.
727 name: &'t str,
728 /// The flag it cannot be given with, as the declaration spells it.
729 other: &'t str,
730 },
731 /// A value was given that the field's type could not be built from.
732 ///
733 /// Boxed, and the only error here that owns anything. Everything else borrows the
734 /// tables or argv, which is what keeps a *successful* parse allocation-free — and the
735 /// box keeps `Error` the size it was, so the `Result` this rides in on the hot path
736 /// does not grow. A value that will not convert has already failed, and a message
737 /// worth reading is worth one allocation.
738 InvalidValue(::std::boxed::Box<InvalidValue<'t>>),
739 /// A required group had none of its members given.
740 ///
741 /// Carries the members as members rather than as a rendered sentence: the caller
742 /// decides how to say it, and a completion asking what would satisfy this needs the
743 /// list rather than the prose.
744 MissingGroup {
745 /// The group's declared name, which appears in the message so a command with
746 /// several groups does not report the same sentence twice.
747 group: &'t str,
748 /// The flags that would satisfy it, as the declaration spells them.
749 members: &'t [&'t str],
750 },
751 /// A subcommand was required, and none was given.
752 MissingSubcommand,
753 /// `--help` or `-h` was given, and `cmd` is what it was asked about.
754 ///
755 /// Not a failure, and returned as one anyway: a parse that stops to print help has not
756 /// produced a value, and every caller already handles the "no value" shape. clap does the
757 /// same thing for the same reason.
758 ///
759 /// `long` distinguishes the two: `-h` prints the short form and `--help` the long one, as
760 /// clap has them. The caller renders — this crate does not print, because a library that
761 /// writes to stdout on its own is one an adopter cannot embed.
762 Help { cmd: &'t Command<'t>, long: bool },
763 /// `arg_required_else_help` found no command-line arguments for `cmd`.
764 ///
765 /// Unlike an explicit help request, this is a usage failure: clap prints the short help to
766 /// stderr and exits with status 2. Keeping the shape separate lets embedders preserve that
767 /// terminal contract without guessing why [`Error::Help`] was returned.
768 MissingArgsHelp { cmd: &'t Command<'t> },
769 /// Recursive long help was requested for `cmd` and every visible descendant.
770 HelpAll { cmd: &'t Command<'t> },
771 /// `--version` or `-V` was asked for. Not a failure either — the caller prints and leaves.
772 ///
773 /// The version string lives in the spec rather than the parse tables. `long` lets the
774 /// caller choose `long_version` for `--version` while `-V` retains the concise value.
775 Version { long: bool },
776}
777
778/// The high half of every key one declaration's items get.
779///
780/// A derive cannot see other expansions, so it cannot hand out keys from a shared
781/// counter: it hashes the declaration it was given instead. It cannot see a module path
782/// either, which is why the module is mixed in *here* — `module_path!()` is available to
783/// the generated code as a compile-time string, so two byte-identical declarations in
784/// different modules end up with different keys rather than colliding.
785///
786/// `declaration` is a hash the derive computed over the item's own tokens.
787pub const fn key_base(module: &str, declaration: u32) -> u64 {
788 // FNV-1a, continuing from the declaration's hash rather than starting over, so both
789 // halves contribute. Spelled out rather than taken from a `Hasher`, which is not
790 // guaranteed to be stable between compilations — and these are baked into a binary.
791 let mut hash: u32 = declaration;
792 let bytes = module.as_bytes();
793 let mut i = 0;
794 while i < bytes.len() {
795 hash ^= bytes[i] as u32;
796 hash = hash.wrapping_mul(0x0100_0193);
797 i += 1;
798 }
799 (hash as u64) << 32
800}
801
802/// Why a value would not convert into the type its field holds.
803///
804/// Separate from [`Error`] so that the enum stays small: this is reached through a `Box`.
805#[derive(Debug, Clone, PartialEq, Eq)]
806pub struct InvalidValue<'t> {
807 /// The flag or argument's name, as the spec calls it.
808 pub name: &'t str,
809 /// The text that would not convert.
810 pub value: ::std::string::String,
811 /// What the type's own conversion complained about.
812 pub reason: ::std::string::String,
813}
814
815/// A command-wide validation or finalization failure.
816///
817/// Return this from a `#[usage(validate_with = ...)]` hook or from the
818/// `TryFrom` implementation named by `#[usage(try_into = ...)]`. The derive
819/// turns it into the same [`Error::InvalidValue`] diagnostic used by field
820/// conversion, so callers keep one parse error type.
821#[derive(Debug, Clone, PartialEq, Eq)]
822pub struct ValidationError {
823 name: &'static str,
824 value: String,
825 reason: String,
826}
827
828impl ValidationError {
829 /// Start an error attributed to a flag, positional, or command name.
830 pub fn field(name: &'static str) -> Self {
831 Self {
832 name,
833 value: String::new(),
834 reason: String::new(),
835 }
836 }
837
838 /// Record the value that failed the command-wide invariant.
839 pub fn value(mut self, value: impl Into<String>) -> Self {
840 self.value = value.into();
841 self
842 }
843
844 /// Explain the invariant that the value did not satisfy.
845 pub fn reason(mut self, reason: impl Into<String>) -> Self {
846 self.reason = reason.into();
847 self
848 }
849
850 /// Convert this application-level failure into the parser's diagnostic.
851 pub fn into_parse_error<'v>(self) -> Error<'static, 'v> {
852 Error::InvalidValue(Box::new(InvalidValue {
853 name: self.name,
854 value: self.value,
855 reason: self.reason,
856 }))
857 }
858}
859
860/// Interpret a value as UTF-8.
861///
862/// The parser hands back bytes borrowed from `argv`; this is the conversion most
863/// callers want, and the point at which a non-UTF-8 command line is rejected —
864/// but only for the values actually inspected.
865pub fn as_str(value: &[u8]) -> Result<&str, std::str::Utf8Error> {
866 std::str::from_utf8(value)
867}
868
869/// How many entries a group of tables holds in total.
870///
871/// The length for [`concat_flags`] and [`concat_args`], which need it as a const generic — so
872/// it has to be computable separately from the concatenation itself.
873///
874/// ```
875/// use usage_argv::{table_len, Flag};
876///
877/// static A: Flag = Flag { name: "a", ..Flag::BOOL };
878/// static B: Flag = Flag { name: "b", ..Flag::BOOL };
879/// const GROUPS: &[&[&Flag]] = &[&[&A], &[], &[&B]];
880/// const N: usize = table_len(GROUPS);
881/// assert_eq!(N, 2);
882/// ```
883pub const fn table_len<T>(groups: &[&[T]]) -> usize {
884 let mut total = 0;
885 let mut i = 0;
886 while i < groups.len() {
887 total += groups[i].len();
888 i += 1;
889 }
890 total
891}
892
893/// Join groups of flag tables into one, at compile time.
894///
895/// This is how `#[usage(flatten)]` stays free. A flattened struct's flags have to appear in
896/// the parent's own table, and the parent's macro expansion cannot see them — it has only a
897/// type. But it can name that type's [`CommandArgs::COMMAND`](crate::spec::CommandArgs::COMMAND),
898/// and a `const fn` can read through it, so the two lists become one `static` array before the
899/// program runs. The parser then walks a single flat slice, exactly as it does for a command
900/// that declared everything itself: flatten costs nothing at run time.
901///
902/// Groups are laid out in the order given, which is what lets a flattened group sit *between*
903/// two of the parent's own declarations — necessary for positional arguments, where order is
904/// the meaning.
905///
906/// `N` must be [`table_len`] of the same groups. It cannot be inferred, and a wrong one fails
907/// to compile rather than leaving the difference filled with padding.
908///
909/// ```
910/// use usage_argv::{concat_flags, table_len, Flag};
911///
912/// static FORCE: Flag = Flag { name: "force", longs: &["force"], ..Flag::BOOL };
913/// static QUIET: Flag = Flag { name: "quiet", longs: &["quiet"], ..Flag::BOOL };
914/// static SHARED: &[&Flag] = &[&QUIET];
915///
916/// const GROUPS: &[&[&Flag]] = &[&[&FORCE], SHARED];
917/// static FLAGS: [&Flag; table_len(GROUPS)] = concat_flags(GROUPS);
918///
919/// assert_eq!(FLAGS.iter().map(|f| f.name).collect::<Vec<_>>(), ["force", "quiet"]);
920/// ```
921pub const fn concat_flags<const N: usize>(
922 groups: &[&[&'static Flag<'static>]],
923) -> [&'static Flag<'static>; N] {
924 // Every slot is written below, but an array has to start somewhere and `MaybeUninit`
925 // would mean `unsafe`. A `Flag` nobody can reach is cheaper than that.
926 static PLACEHOLDER: Flag<'static> = Flag::BOOL;
927 let mut out = [&PLACEHOLDER; N];
928 let mut at = 0;
929 let mut g = 0;
930 while g < groups.len() {
931 let group = groups[g];
932 let mut i = 0;
933 while i < group.len() {
934 out[at] = group[i];
935 at += 1;
936 i += 1;
937 }
938 g += 1;
939 }
940 assert!(
941 at == N,
942 "`N` must be `table_len` of the same groups, or the table would keep a placeholder \
943 that answers to nothing"
944 );
945 out
946}
947
948/// Join groups of argument tables into one, at compile time.
949///
950/// The positional counterpart of [`concat_flags`] — see there for why this exists. Order
951/// matters more here: an argument's position *is* its identity, so a flattened group has to
952/// land exactly where the field was written.
953///
954/// Two functions rather than one generic: each needs a value to fill an array with before
955/// overwriting it, and there is no way to ask a type parameter for one in a `const fn`.
956pub const fn concat_args<const N: usize>(
957 groups: &[&[&'static Arg<'static>]],
958) -> [&'static Arg<'static>; N] {
959 static PLACEHOLDER: Arg<'static> = Arg::REQUIRED;
960 let mut out = [&PLACEHOLDER; N];
961 let mut at = 0;
962 let mut g = 0;
963 while g < groups.len() {
964 let group = groups[g];
965 let mut i = 0;
966 while i < group.len() {
967 out[at] = group[i];
968 at += 1;
969 i += 1;
970 }
971 g += 1;
972 }
973 assert!(
974 at == N,
975 "`N` must be `table_len` of the same groups, or the table would keep a placeholder \
976 that answers to nothing"
977 );
978 out
979}
980
981/// The key `--help` answers to, and the one `-h` does.
982///
983/// Reserved rather than generated: a derive builds keys from a hash of the type they came from
984/// in the high half and an index in the low half, so the top of the range belongs to nobody.
985/// Generated code compares against these to tell a help request from a flag of its own.
986pub const HELP_LONG_KEY: u64 = u64::MAX;
987/// See [`HELP_LONG_KEY`].
988pub const HELP_SHORT_KEY: u64 = u64::MAX - 1;
989
990/// `--help`, which every command answers to.
991///
992/// In the parse table and *not* in the metadata, which is the whole trick: the parser has to
993/// recognise the flag, and help output must not list it — a spec does not declare `--help`, so
994/// showing one would make the rendered page disagree with the spec it came from.
995pub static HELP_LONG: Flag<'static> = Flag {
996 key: HELP_LONG_KEY,
997 name: "help",
998 longs: &["help"],
999 action: ArgAction::HelpLong,
1000 ..Flag::BOOL
1001};
1002
1003/// See [`HELP_LONG_KEY`].
1004pub const VERSION_LONG_KEY: u64 = u64::MAX - 2;
1005/// See [`HELP_LONG_KEY`].
1006pub const VERSION_SHORT_KEY: u64 = u64::MAX - 3;
1007
1008/// `--version`, where the CLI declared one.
1009///
1010/// In the parse table and not in the metadata, exactly as `--help` is: a spec does not declare
1011/// `--version`, so listing one would make the rendered page disagree with the spec it came from.
1012pub static VERSION_LONG: Flag<'static> = Flag {
1013 key: VERSION_LONG_KEY,
1014 name: "version",
1015 longs: &["version"],
1016 action: ArgAction::Version,
1017 ..Flag::BOOL
1018};
1019
1020/// `-V`, which clap also supplies.
1021pub static VERSION_SHORT: Flag<'static> = Flag {
1022 key: VERSION_SHORT_KEY,
1023 name: "version",
1024 shorts: b"V",
1025 action: ArgAction::Version,
1026 ..Flag::BOOL
1027};
1028
1029/// `-h`, which prints the shorter form.
1030pub static HELP_SHORT: Flag<'static> = Flag {
1031 key: HELP_SHORT_KEY,
1032 name: "help",
1033 shorts: b"h",
1034 action: ArgAction::HelpShort,
1035 ..Flag::BOOL
1036};
1037
1038/// A named subcommand of a given command, by name or alias.
1039///
1040/// Free rather than a method because `help` resolves a path *without* descending: the words
1041/// after it are a question about a command rather than a walk into one.
1042///
1043/// Names across every subcommand before any alias, the precedence the grammar states — and
1044/// the reason this is the only implementation of it on argv's side. `ex run` and `ex help run`
1045/// selecting different commands would be exactly the divergence this rule was written to end.
1046pub(crate) fn find_named<'t>(cmd: &'t Command<'t>, name: &[u8]) -> Option<&'t Command<'t>> {
1047 let subcommands = || cmd.subcommands.iter().copied();
1048 subcommands()
1049 .find(|c| c.name.as_bytes() == name)
1050 .or_else(|| subcommands().find(|c| c.aliases.iter().any(|a| a.as_bytes() == name)))
1051}
1052
1053/// What a caller should print for a parse failure, and what to exit with.
1054///
1055/// The one entry point a generated `parse()` reaches for, and the reason it exists here rather
1056/// than in the derive: whether the good rendering is available is a *feature of this crate* in
1057/// the adopter's dependency graph, and a `#[cfg]` written into generated code is evaluated in
1058/// the adopter's crate, where the feature is not theirs to see. That is how a metadata field
1059/// once got silently dropped; the answer is that the cfg lives beside the thing it gates.
1060///
1061/// With `diagnostics` on, this is the clap-shaped message. Without it, the error's `Debug`
1062/// form — which is still better than nothing and is what a parser-only build asked for.
1063///
1064/// [`Error::Help`] and [`Error::Version`] are not failures and must be handled before this.
1065#[cfg(feature = "diagnostics")]
1066pub fn render_failure(spec: &spec::Spec<'_>, argv: &[&OsStr], error: &Error<'_, '_>) -> String {
1067 diagnostic::render(spec, argv, error, diagnostic::Style::auto())
1068}
1069
1070/// A parse failure, never coloured.
1071///
1072/// [`render_failure`] asks the environment whether to colour, which is right for a process and
1073/// wrong for anything that keeps the string: a test that asserts on a message, or a snapshot of
1074/// one, would pass or fail by whether stderr happened to be a terminal. The renderer is the
1075/// same; only the answer to that question is fixed.
1076#[cfg(feature = "diagnostics")]
1077pub fn render_failure_plain(
1078 spec: &spec::Spec<'_>,
1079 argv: &[&OsStr],
1080 error: &Error<'_, '_>,
1081) -> String {
1082 diagnostic::render(spec, argv, error, diagnostic::Style::PLAIN)
1083}
1084
1085/// Render a failure through a spec-declared executable view.
1086///
1087/// `argv` is the original full argv, including the view executable as argv0.
1088#[cfg(feature = "diagnostics")]
1089pub fn render_failure_view<'a>(
1090 spec: &'a spec::Spec<'a>,
1091 argv: &[&OsStr],
1092 error: &Error<'_, '_>,
1093 view: &'a spec::ViewMeta<'a>,
1094) -> String {
1095 diagnostic::render_view(spec, argv, error, diagnostic::Style::auto(), view)
1096}
1097
1098/// What a caller should print for a parse failure, without the renderer that makes it readable.
1099///
1100/// See the other half. A caller that wants the clap-shaped message turns on `diagnostics`;
1101/// this is what a parser-only build asked for, and it still says which error it was.
1102#[cfg(all(feature = "spec", not(feature = "diagnostics")))]
1103pub fn render_failure(spec: &spec::Spec<'_>, argv: &[&OsStr], error: &Error<'_, '_>) -> String {
1104 let _ = (spec, argv);
1105 ::std::format!("error: {error:?}\n")
1106}
1107
1108/// A parse failure without the renderer, which is plain either way.
1109#[cfg(all(feature = "spec", not(feature = "diagnostics")))]
1110pub fn render_failure_plain(
1111 spec: &spec::Spec<'_>,
1112 argv: &[&OsStr],
1113 error: &Error<'_, '_>,
1114) -> String {
1115 render_failure(spec, argv, error)
1116}
1117
1118/// Render a failure through a declared view without the optional diagnostics renderer.
1119#[cfg(all(feature = "spec", not(feature = "diagnostics")))]
1120pub fn render_failure_view(
1121 spec: &spec::Spec<'_>,
1122 argv: &[&OsStr],
1123 error: &Error<'_, '_>,
1124 view: &spec::ViewMeta<'_>,
1125) -> String {
1126 let _ = (spec, argv, view);
1127 ::std::format!("error: {error:?}\n")
1128}
1129
1130/// What a caller should print for the deprecations a command line used.
1131///
1132/// The same arrangement as [`render_failure`], and for the same reason: whether the coloured
1133/// rendering is available is a feature of *this* crate in the adopter's dependency graph, so the
1134/// `#[cfg]` lives beside the thing it gates rather than in generated code.
1135///
1136/// Warnings are not failures. A caller prints these to stderr and carries on.
1137#[cfg(feature = "diagnostics")]
1138pub fn render_warnings(warnings: &[warn::Warning<'_>]) -> String {
1139 diagnostic::render_warnings(warnings, diagnostic::Style::auto())
1140}
1141
1142/// The same wording without the renderer that colours it. See the other half.
1143#[cfg(all(feature = "spec", not(feature = "diagnostics")))]
1144pub fn render_warnings(warnings: &[warn::Warning<'_>]) -> String {
1145 warn::render_warnings(warnings)
1146}
1147
1148/// The word a tool sends to ask a binary for its own spec.
1149///
1150/// Not a flag and not a command: a spec request is not something this CLI *does*, so it is
1151/// answered before the parse and stays out of the tables — the same reason
1152/// `__complete_word__` is a word rather than a subcommand. It also keeps the endpoint from
1153/// perturbing the document it prints, which a declared flag would not.
1154pub const SPEC_REQUEST: &str = "__usage_spec__";
1155
1156/// Whether this argv asks for the spec rather than for the CLI to run.
1157///
1158/// Only the first word counts: `mycli build __usage_spec__` passes the word through as an
1159/// ordinary value, because a request is the whole invocation or it is nothing.
1160///
1161/// A root that declares a command of that name keeps it, which is the precedence the `help`
1162/// subcommand already has. The check is here rather than in the derive because a `Cli` derive
1163/// expands one struct and cannot see the variant names of a separate `Subcommands` enum — the
1164/// static tables can.
1165pub fn is_spec_request(root: &Command<'_>, argv: &[&OsStr]) -> bool {
1166 let [first, ..] = argv else { return false };
1167 first.as_encoded_bytes() == SPEC_REQUEST.as_bytes()
1168 && find_named(root, SPEC_REQUEST.as_bytes()).is_none()
1169}
1170
1171/// Whether a flag is one of the two the parser supplies rather than the CLI declaring it.
1172pub fn is_help_flag(flag: &Flag<'_>) -> bool {
1173 matches!(
1174 flag.action,
1175 ArgAction::Help | ArgAction::HelpShort | ArgAction::HelpLong | ArgAction::HelpAll
1176 )
1177}
1178
1179/// Whether a flag is one of the two the parser supplies for `--version`.
1180pub fn is_version_flag(flag: &Flag<'_>) -> bool {
1181 flag.action == ArgAction::Version
1182}
1183
1184/// Whether one exact root argument selects a declared or synthesized version action.
1185///
1186/// Declared flags are checked first because they shadow the built-in `--version` and `-V`
1187/// spellings. Executable views use this before projection so custom version spellings keep
1188/// reporting the package that owns the view.
1189pub fn is_version_arg(cmd: &Command<'_>, word: &OsStr) -> bool {
1190 let token = word.as_encoded_bytes();
1191 if let Some(long) = token.strip_prefix(b"--") {
1192 if let Some(flag) = cmd.flags.iter().find(|flag| {
1193 flag.longs
1194 .iter()
1195 .any(|spelling| spelling.as_bytes() == long)
1196 }) {
1197 return is_version_flag(flag);
1198 }
1199 if cmd.flags.iter().any(|flag| {
1200 flag.negate
1201 .is_some_and(|spelling| spelling.as_bytes() == long)
1202 }) {
1203 return false;
1204 }
1205 return long == b"version" && cmd.version && !cmd.disable_version_flag;
1206 }
1207 if let [b'-', short] = token {
1208 if let Some(flag) = cmd.flags.iter().find(|flag| flag.shorts.contains(short)) {
1209 return is_version_flag(flag);
1210 }
1211 return *short == b'V' && cmd.version && !cmd.disable_version_flag;
1212 }
1213 false
1214}
1215
1216/// Resolve a subcommand by name or alias, at compile time.
1217///
1218/// For [`Command::default_subcommand`], which names a command that a derive cannot see: the
1219/// variants of a subcommand enum are a different macro expansion, so the name is all the
1220/// parent has. Searching the list in a `const fn` closes that gap — the answer is the same
1221/// `&'static` the table already holds, found before the program runs.
1222///
1223/// A name no subcommand answers to is a **compile error**, since this panics during const
1224/// evaluation. That is the whole point of doing it here rather than at startup.
1225///
1226/// ```
1227/// use usage_argv::{find_subcommand, Command};
1228///
1229/// static RUN: Command = Command { name: "run", ..Command::EMPTY };
1230/// static SUBS: &[&Command] = &[&RUN];
1231/// static ROOT: Command = Command {
1232/// name: "ex",
1233/// subcommands: SUBS,
1234/// default_subcommand: Some(find_subcommand(SUBS, "run")),
1235/// ..Command::EMPTY
1236/// };
1237/// assert_eq!(ROOT.default_subcommand.unwrap().name, "run");
1238/// ```
1239pub const fn find_subcommand<'a>(
1240 subcommands: &'a [&'a Command<'a>],
1241 name: &str,
1242) -> &'a Command<'a> {
1243 // Names first, then aliases: a command's own name outranks another command's alias, so
1244 // the answer does not depend on the order the table happens to list them in. Checking
1245 // each candidate's name *and* aliases in one pass instead let whichever command came
1246 // first win, and usage-lib resolved the same spec to the last one.
1247 let mut i = 0;
1248 while i < subcommands.len() {
1249 if str_eq(subcommands[i].name, name) {
1250 return subcommands[i];
1251 }
1252 i += 1;
1253 }
1254 // Aliases answer too, because usage-lib resolves the name against names, aliases and
1255 // hidden aliases alike — so a spec may point `default_subcommand` at any of them.
1256 let mut i = 0;
1257 while i < subcommands.len() {
1258 let candidate = subcommands[i];
1259 let mut a = 0;
1260 while a < candidate.aliases.len() {
1261 if str_eq(candidate.aliases[a], name) {
1262 return candidate;
1263 }
1264 a += 1;
1265 }
1266 i += 1;
1267 }
1268 panic!("`default_subcommand` names a command that this one does not have")
1269}
1270
1271/// Refuse two subcommands that answer to the same name, aliases included.
1272///
1273/// A derive expansion can validate aliases written on one enum, but aliases may also live on
1274/// the independently expanded `Args` structs its variants wrap. This final, joined-table check
1275/// is where both declarations are visible.
1276pub const fn assert_unique_subcommand_names(subcommands: &[&Command<'_>]) {
1277 const fn form<'a>(cmd: &'a Command<'a>, at: usize) -> Option<&'a str> {
1278 if at == 0 {
1279 Some(cmd.name)
1280 } else if at <= cmd.aliases.len() {
1281 Some(cmd.aliases[at - 1])
1282 } else {
1283 None
1284 }
1285 }
1286
1287 let mut command = 0;
1288 while command < subcommands.len() {
1289 let mut at = 0;
1290 while let Some(name) = form(subcommands[command], at) {
1291 let mut other_command = command;
1292 while other_command < subcommands.len() {
1293 let mut other_at = if other_command == command { at + 1 } else { 0 };
1294 while let Some(other) = form(subcommands[other_command], other_at) {
1295 assert!(
1296 !str_eq(name, other),
1297 "two subcommands answer to the same name, counting aliases"
1298 );
1299 other_at += 1;
1300 }
1301 other_command += 1;
1302 }
1303 at += 1;
1304 }
1305 command += 1;
1306 }
1307}
1308
1309/// Assert that flags scoped to a clause do not reuse a command-level spelling.
1310///
1311/// Derive-generated command tables include scoped flags so the parser can find them. Entries
1312/// that share a key are therefore the same declaration and are ignored here; every other pair
1313/// must have disjoint long, short, alias, and negated spellings.
1314pub const fn assert_clause_flag_spellings(command: &Command<'_>) {
1315 let Some(clause) = command.clause else {
1316 return;
1317 };
1318 let mut scoped_at = 0;
1319 while scoped_at < clause.flags.len() {
1320 let scoped = clause.flags[scoped_at];
1321 let mut command_at = 0;
1322 while command_at < command.flags.len() {
1323 let flag = command.flags[command_at];
1324 if flag.key != scoped.key && const_flag_forms_overlap(flag, scoped) {
1325 panic!("a clause flag spelling conflicts with a command-level flag");
1326 }
1327 command_at += 1;
1328 }
1329 scoped_at += 1;
1330 }
1331}
1332
1333const fn const_flag_forms_overlap(a: &Flag<'_>, b: &Flag<'_>) -> bool {
1334 let mut i = 0;
1335 while i < a.longs.len() {
1336 let mut j = 0;
1337 while j < b.longs.len() {
1338 if const_bytes_eq(a.longs[i].as_bytes(), b.longs[j].as_bytes()) {
1339 return true;
1340 }
1341 j += 1;
1342 }
1343 if let Some(negate) = b.negate {
1344 if const_bytes_eq(a.longs[i].as_bytes(), negate.as_bytes()) {
1345 return true;
1346 }
1347 }
1348 i += 1;
1349 }
1350 i = 0;
1351 while i < a.shorts.len() {
1352 let mut j = 0;
1353 while j < b.shorts.len() {
1354 if a.shorts[i] == b.shorts[j] {
1355 return true;
1356 }
1357 j += 1;
1358 }
1359 i += 1;
1360 }
1361 if let Some(negate) = a.negate {
1362 let mut j = 0;
1363 while j < b.longs.len() {
1364 if const_bytes_eq(negate.as_bytes(), b.longs[j].as_bytes()) {
1365 return true;
1366 }
1367 j += 1;
1368 }
1369 if let Some(other) = b.negate {
1370 if const_bytes_eq(negate.as_bytes(), other.as_bytes()) {
1371 return true;
1372 }
1373 }
1374 }
1375 false
1376}
1377
1378const fn const_bytes_eq(a: &[u8], b: &[u8]) -> bool {
1379 if a.len() != b.len() {
1380 return false;
1381 }
1382 let mut i = 0;
1383 while i < a.len() {
1384 if a[i] != b[i] {
1385 return false;
1386 }
1387 i += 1;
1388 }
1389 true
1390}
1391
1392/// `==` on strings, in a `const fn`.
1393const fn str_eq(a: &str, b: &str) -> bool {
1394 let (a, b) = (a.as_bytes(), b.as_bytes());
1395 if a.len() != b.len() {
1396 return false;
1397 }
1398 let mut i = 0;
1399 while i < a.len() {
1400 if a[i] != b[i] {
1401 return false;
1402 }
1403 i += 1;
1404 }
1405 true
1406}
1407
1408/// Rebuild an [`OsString`] from bytes the parser handed back.
1409///
1410/// This is the reverse of [`OsStr::as_encoded_bytes`], and it is how a `PathBuf` field
1411/// receives a filename the operating system accepts but UTF-8 does not — `/tmp/\xff` stays
1412/// `/tmp/\xff` rather than becoming a *different* filename with `U+FFFD` in it.
1413///
1414/// Where the platform cannot hold those bytes, they are handed back in the `Err` — as
1415/// `String::from_utf8` does — so the caller can name the value in its error without this
1416/// having to copy it for a case that is nearly never taken.
1417///
1418/// # Why this is not `unsafe`, and why it is not lossless everywhere
1419///
1420/// On **Unix** an `OsString` is an arbitrary byte sequence, so the conversion is total and
1421/// uses the safe [`OsStringExt::from_vec`]. Every byte survives, which is the case that
1422/// matters: non-UTF-8 filenames are ordinary there.
1423///
1424/// [`OsStringExt::from_vec`]: std::os::unix::ffi::OsStringExt::from_vec
1425///
1426/// On **Windows** the encoding is WTF-8, where not every byte sequence is valid, and the only
1427/// constructor that accepts one is `OsString::from_encoded_bytes_unchecked` — whose
1428/// precondition this function cannot enforce. It takes a `Vec<u8>` from a safe caller, so
1429/// there is no way to know the bytes came from `as_encoded_bytes` rather than from anywhere
1430/// else, and a safe function with a precondition that can be violated is unsound however
1431/// carefully its callers behave today.
1432///
1433/// So on Windows the bytes go through UTF-8, and one that is not valid UTF-8 is refused
1434/// rather than assumed. What that gives up is a Windows argument containing an unpaired
1435/// surrogate, which is reported instead of accepted; what it buys is that this crate needs no
1436/// `unsafe` at all.
1437pub fn os_string_from_bytes(value: Vec<u8>) -> Result<OsString, Vec<u8>> {
1438 #[cfg(unix)]
1439 {
1440 Ok(std::os::unix::ffi::OsStringExt::from_vec(value))
1441 }
1442 #[cfg(not(unix))]
1443 {
1444 match String::from_utf8(value) {
1445 Ok(text) => Ok(OsString::from(text)),
1446 Err(bad) => Err(bad.into_bytes()),
1447 }
1448 }
1449}
1450
1451/// One [`Error::InvalidValue`], built out of line.
1452///
1453/// Cold and never inlined on purpose: this is the failure path of every value
1454/// conversion in every generated `build`, and inlining it there is what made
1455/// those functions large.
1456#[cold]
1457#[inline(never)]
1458pub(crate) fn invalid_value_error<'t, 'v>(
1459 name: &'t str,
1460 value: String,
1461 reason: String,
1462) -> Error<'t, 'v> {
1463 Error::InvalidValue(Box::new(InvalidValue {
1464 name,
1465 value,
1466 reason,
1467 }))
1468}
1469
1470/// One [`Error::InvalidValue`] for a word that was not UTF-8.
1471///
1472/// The error half of what a generated `build` does per text field: the check stays
1473/// inline at the field, and this — the lossy rendering and the allocations — lives
1474/// here once instead of once per field.
1475#[cold]
1476#[inline(never)]
1477pub fn invalid_utf8_value<'t, 'v>(name: &'t str, bad: std::string::FromUtf8Error) -> Error<'t, 'v> {
1478 invalid_value_error(
1479 name,
1480 String::from_utf8_lossy(bad.as_bytes()).into_owned(),
1481 bad.utf8_error().to_string(),
1482 )
1483}
1484
1485/// One [`Error::InvalidValue`] for a value whose type would not build from it.
1486///
1487/// Takes the reason as `&dyn Display` so one copy serves every `FromStr` error type.
1488#[cold]
1489#[inline(never)]
1490pub fn invalid_parsed_value<'t, 'v>(
1491 name: &'t str,
1492 value: String,
1493 reason: &dyn std::fmt::Display,
1494) -> Error<'t, 'v> {
1495 invalid_value_error(name, value, reason.to_string())
1496}
1497
1498/// One [`Error::InvalidValue`] for a word that is not one of a value enum's choices.
1499#[cold]
1500#[inline(never)]
1501pub fn invalid_choice_value<'t, 'v>(name: &'t str, value: String) -> Error<'t, 'v> {
1502 invalid_value_error(name, value, String::from("not one of the declared values"))
1503}
1504
1505/// One [`Error::InvalidValue`] for bytes the platform cannot hold in a path.
1506#[cold]
1507#[inline(never)]
1508pub fn invalid_os_value<'t, 'v>(name: &'t str, bytes: Vec<u8>) -> Error<'t, 'v> {
1509 invalid_value_error(
1510 name,
1511 String::from_utf8_lossy(&bytes).into_owned(),
1512 "this platform cannot hold these bytes in a path".to_string(),
1513 )
1514}
1515
1516/// Convert every repeated value of one text field, reporting `name` for the
1517/// first that is not UTF-8.
1518///
1519/// The shared body of what a generated `build` does per collecting text field:
1520/// one loop in the binary rather than one per field. Converts element by
1521/// element rather than with `collect` so the error can carry the value that
1522/// failed rather than only that one did.
1523///
1524/// The empty case is answered here rather than in the shared loop, and this is
1525/// what keeps sharing the loop free: a command at mise's scale declares dozens
1526/// of collecting fields and a command line names one or two of them, so most of
1527/// these calls have nothing to convert. Testing that at the field costs a branch;
1528/// reaching the loop to learn it costs the call.
1529#[inline]
1530pub fn utf8_values<'t, 'v>(
1531 values: Vec<Vec<u8>>,
1532 name: &'t str,
1533) -> Result<Vec<String>, Error<'t, 'v>> {
1534 if values.is_empty() {
1535 return Ok(Vec::new());
1536 }
1537 utf8_values_given(values, name)
1538}
1539
1540#[inline(never)]
1541fn utf8_values_given<'t, 'v>(
1542 values: Vec<Vec<u8>>,
1543 name: &'t str,
1544) -> Result<Vec<String>, Error<'t, 'v>> {
1545 let mut out = Vec::with_capacity(values.len());
1546 for value in values {
1547 match String::from_utf8(value) {
1548 Ok(text) => out.push(text),
1549 Err(bad) => return Err(invalid_utf8_value(name, bad)),
1550 }
1551 }
1552 Ok(out)
1553}
1554
1555/// Convert every repeated value of one field through
1556/// [`FromStr`](std::str::FromStr), reporting `name` for the first that fails.
1557///
1558/// Monomorphized once per target type rather than expanded once per field, and
1559/// the empty case is answered at the field for the reason [`utf8_values`] gives.
1560#[inline]
1561pub fn parsed_values<'t, 'v, T>(
1562 values: Vec<Vec<u8>>,
1563 name: &'t str,
1564) -> Result<Vec<T>, Error<'t, 'v>>
1565where
1566 T: std::str::FromStr,
1567 T::Err: std::fmt::Display,
1568{
1569 if values.is_empty() {
1570 return Ok(Vec::new());
1571 }
1572 parsed_values_given(values, name)
1573}
1574
1575#[inline(never)]
1576fn parsed_values_given<'t, 'v, T>(
1577 values: Vec<Vec<u8>>,
1578 name: &'t str,
1579) -> Result<Vec<T>, Error<'t, 'v>>
1580where
1581 T: std::str::FromStr,
1582 T::Err: std::fmt::Display,
1583{
1584 let mut out = Vec::with_capacity(values.len());
1585 for value in values {
1586 let text = match String::from_utf8(value) {
1587 Ok(text) => text,
1588 Err(bad) => return Err(invalid_utf8_value(name, bad)),
1589 };
1590 match text.parse() {
1591 Ok(parsed) => out.push(parsed),
1592 Err(reason) => return Err(invalid_parsed_value(name, text, &reason)),
1593 }
1594 }
1595 Ok(out)
1596}
1597
1598/// Convert every repeated value of one path-like field, reporting `name` for
1599/// the first the platform cannot hold.
1600///
1601/// `T` is what the field collects — [`PathBuf`](std::path::PathBuf) or
1602/// [`OsString`] — so one body serves both. The same platform note as
1603/// [`os_string_from_bytes`] applies: lossless on Unix, partial on Windows. The
1604/// empty case is answered at the field for the reason [`utf8_values`] gives.
1605#[inline]
1606pub fn os_values<'t, 'v, T: From<OsString>>(
1607 values: Vec<Vec<u8>>,
1608 name: &'t str,
1609) -> Result<Vec<T>, Error<'t, 'v>> {
1610 if values.is_empty() {
1611 return Ok(Vec::new());
1612 }
1613 os_values_given(values, name)
1614}
1615
1616#[inline(never)]
1617fn os_values_given<'t, 'v, T: From<OsString>>(
1618 values: Vec<Vec<u8>>,
1619 name: &'t str,
1620) -> Result<Vec<T>, Error<'t, 'v>> {
1621 let mut out = Vec::with_capacity(values.len());
1622 for value in values {
1623 match os_string_from_bytes(value) {
1624 Ok(os) => out.push(T::from(os)),
1625 Err(bytes) => return Err(invalid_os_value(name, bytes)),
1626 }
1627 }
1628 Ok(out)
1629}
1630
1631/// A single binding pass over `argv`.
1632///
1633/// [`Command::default_subcommand_flags`] adds a read-only lookahead before binding
1634/// to choose an implicit command boundary.
1635///
1636/// Created with [`Parser::new`] and driven with [`Parser::next_event`].
1637pub struct Parser<'t, 'a, 'v> {
1638 argv: &'a [&'v OsStr],
1639 /// Index of the next token to read.
1640 pos: usize,
1641 /// The command currently in scope.
1642 cmd: &'t Command<'t>,
1643 /// The canonical root, used to hide root globals omitted by an executable view.
1644 #[cfg(feature = "spec")]
1645 root: &'t Command<'t>,
1646 /// The executable projection being parsed, if argv0 selected one.
1647 #[cfg(feature = "spec")]
1648 view: Option<&'t spec::ViewMeta<'t>>,
1649 /// What an unrecognized flag-like token means in the command currently in scope.
1650 ///
1651 /// Carried rather than looked up, because it is inherited: a command that states
1652 /// nothing keeps what the enclosing one said, and walking back up the ancestors on
1653 /// every unrecognized token would pay for the inheritance at the wrong moment.
1654 unknown_flags: UnknownFlags,
1655 /// Effective inherited trailing-delimiter policy.
1656 dont_delimit_trailing_values: bool,
1657 /// The chain above `cmd`, used to find inherited global flags. Fixed size so
1658 /// that nothing is allocated.
1659 ancestors: [Option<&'t Command<'t>>; MAX_DEPTH],
1660 depth: usize,
1661 /// Bytes left in a short-flag bundle, if one is partly read.
1662 bundle: &'v [u8],
1663 /// The whole token the current bundle came from, so an error raised part way
1664 /// through it can still name what the user typed.
1665 bundle_token: &'v [u8],
1666 /// A variadic flag that is still collecting values.
1667 collecting: Option<&'t Flag<'t>>,
1668 /// Where the command in scope began, as an index into `argv`.
1669 cmd_start: usize,
1670 /// Where each ancestor's own words began, in step with `ancestors`.
1671 starts: [usize; MAX_DEPTH],
1672 /// How many values it has taken, so a bound can stop it.
1673 collected: u32,
1674 /// Which of `cmd.args` is next to fill.
1675 arg_pos: usize,
1676 /// How many words the variadic at `arg_pos` has taken, for the same reason.
1677 arg_taken: u32,
1678 /// Whether any word has been bound to a positional of `cmd`. Once one has,
1679 /// no further word can select a subcommand.
1680 arg_filled: bool,
1681 /// Whether this command has bound any flag or positional. Unlike
1682 /// `arg_filled`, flags count because clap's command policy treats both as
1683 /// arguments that exclude a later subcommand.
1684 command_arg_found: bool,
1685 /// Whether flag interpretation has stopped. A `--` does this, and so does an
1686 /// `automatic` argument taking a value.
1687 flags_stopped: bool,
1688 /// Whether a `--` was actually consumed as a separator.
1689 ///
1690 /// Tracked apart from `flags_stopped` because the two can differ: an
1691 /// `automatic` argument stops flag interpretation without any separator being
1692 /// typed, and a `preserve` argument keeps one as a value rather than
1693 /// consuming it. Callers asking this question want to know what the user
1694 /// wrote, not what state the parser reached.
1695 separator_seen: bool,
1696 /// Whether the default subcommand has already been taken.
1697 ///
1698 /// Once, per parse: a default subcommand that itself declares one would otherwise
1699 /// descend on every word until the tree ran out.
1700 default_taken: bool,
1701 /// First default-only flag, when lookahead found no explicit sibling.
1702 default_flag_at: Option<usize>,
1703 /// Cursor just after the implicit boundary bundle, whose shorts keep parent ownership.
1704 default_bundle_end: usize,
1705 /// Set once a fatal error has been reported, so iteration stops.
1706 done: bool,
1707 /// Whether declared built-in actions stop parsing with their action error.
1708 ///
1709 /// Invocation parsing does; completion walking only needs the grammar position after the
1710 /// flag, and must not execute an action while inspecting a partial command line.
1711 action_errors: bool,
1712 /// The `argv` range the `help` *word* resolved as a command path, if one was typed.
1713 ///
1714 /// Empty for `--help`, which asks about wherever the parse had got to. For the word, the
1715 /// question is about a command deeper than the parse reached, and only this walk knows
1716 /// which tokens named it: a caller re-scanning `argv` would count a flag's detached value
1717 /// that happens to spell a sibling's name. Two indices rather than the commands
1718 /// themselves, so the parser keeps allocating nothing.
1719 help_span: (usize, usize),
1720 /// An implicit clause boundary follows the positional event that completed it.
1721 pending_clause_boundary: ::core::option::Option<Clause<'t>>,
1722}
1723
1724impl<'t: 'v, 'a, 'v> Parser<'t, 'a, 'v> {
1725 /// Begin parsing `argv` against `root`.
1726 ///
1727 /// `argv` excludes the program name.
1728 pub fn new(root: &'t Command<'t>, argv: &'a [&'v OsStr]) -> Self {
1729 Self::with_action_errors(root, argv, true)
1730 }
1731
1732 /// Begin a non-executing parse for completion walking.
1733 #[cfg(feature = "complete")]
1734 pub(crate) fn for_completion(root: &'t Command<'t>, argv: &'a [&'v OsStr]) -> Self {
1735 Self::with_action_errors(root, argv, false)
1736 }
1737
1738 fn with_action_errors(
1739 root: &'t Command<'t>,
1740 argv: &'a [&'v OsStr],
1741 action_errors: bool,
1742 ) -> Self {
1743 let mut parser = Parser {
1744 argv,
1745 pos: 0,
1746 cmd: root,
1747 #[cfg(feature = "spec")]
1748 root,
1749 #[cfg(feature = "spec")]
1750 view: None,
1751 unknown_flags: match root.unknown_flags {
1752 ::core::option::Option::Some(mode) => mode,
1753 // Nothing above the root to inherit from, so the default stands.
1754 ::core::option::Option::None => UnknownFlags::Value,
1755 },
1756 dont_delimit_trailing_values: root.dont_delimit_trailing_values,
1757 ancestors: [None; MAX_DEPTH],
1758 depth: 0,
1759 bundle: &[],
1760 bundle_token: &[],
1761 collecting: None,
1762 cmd_start: 0,
1763 starts: [0; MAX_DEPTH],
1764 collected: 0,
1765 arg_pos: 0,
1766 arg_taken: 0,
1767 arg_filled: false,
1768 command_arg_found: false,
1769 flags_stopped: false,
1770 separator_seen: false,
1771 default_taken: false,
1772 default_flag_at: None,
1773 default_bundle_end: 0,
1774 done: false,
1775 action_errors,
1776 help_span: (0, 0),
1777 pending_clause_boundary: ::core::option::Option::None,
1778 };
1779 if root.default_subcommand_flags {
1780 parser.default_flag_at = parser.default_flag_route();
1781 }
1782 parser
1783 }
1784
1785 /// Find the implicit boundary without binding anything. Parent spellings win while
1786 /// scanning the prefix; after the boundary the ordinary child/global scope applies.
1787 /// Unknown flags stop lookahead because their value arity cannot be guessed.
1788 fn default_flag_route(&self) -> Option<usize> {
1789 let default = self.cmd.default_subcommand?;
1790 let mut at = None;
1791 let mut i = 0;
1792 while let Some(token) = self.argv.get(i).map(bytes) {
1793 let scope = if at.is_some() { default } else { self.cmd };
1794 if token == b"--"
1795 || token == b"-"
1796 || scope.clause.is_some_and(|c| c.separator == Some(token))
1797 {
1798 return at;
1799 }
1800 if !is_flag_like(token) {
1801 return if self.find_subcommand(token).is_some()
1802 || (token == b"help" && !self.cmd.disable_help_subcommand)
1803 {
1804 None
1805 } else {
1806 at
1807 };
1808 }
1809 let mut value_flag = None;
1810 let mut attached = None;
1811 if let Some(body) = token.strip_prefix(b"--") {
1812 let end = body.iter().position(|b| *b == b'=').unwrap_or(body.len());
1813 let name = &body[..end];
1814 let parent = self.find_long(name).or_else(|| self.find_negation(name));
1815 if parent.is_none()
1816 && ((name == b"help" && !self.cmd.disable_help_flag)
1817 || (name == b"version"
1818 && self.cmd.version
1819 && !self.cmd.disable_version_flag))
1820 {
1821 i += 1;
1822 continue;
1823 }
1824 let flag = parent.or_else(|| {
1825 default.flags.iter().copied().find(|f| {
1826 f.longs.iter().any(|l| l.as_bytes() == name)
1827 || f.negate.is_some_and(|n| n.as_bytes() == name)
1828 })
1829 });
1830 let Some(flag) = flag else {
1831 return at;
1832 };
1833 if parent.is_none() {
1834 at.get_or_insert(i);
1835 }
1836 if flag.takes_value && flag.negate.is_none_or(|n| n.as_bytes() != name) {
1837 value_flag = Some(flag);
1838 attached = (end < body.len()).then(|| &body[end + 1..]);
1839 }
1840 } else {
1841 for (offset, byte) in token[1..].iter().enumerate() {
1842 let parent = self.find_short(*byte);
1843 let Some(flag) = parent.or_else(|| {
1844 default
1845 .flags
1846 .iter()
1847 .copied()
1848 .find(|f| f.shorts.contains(byte))
1849 }) else {
1850 return at;
1851 };
1852 if parent.is_none() {
1853 at.get_or_insert(i);
1854 }
1855 if flag.takes_value {
1856 value_flag = Some(flag);
1857 let rest = &token[offset + 2..];
1858 attached =
1859 (!rest.is_empty()).then_some(rest.strip_prefix(b"=").unwrap_or(rest));
1860 break;
1861 }
1862 }
1863 }
1864 i += 1;
1865 if let Some(flag) = value_flag {
1866 let scope = if at.is_some() { default } else { self.cmd };
1867 let is_separator =
1868 |next: &[u8]| scope.clause.is_some_and(|c| c.separator == Some(next));
1869 let first = if let Some(value) = attached {
1870 Some(value)
1871 } else if !flag.require_equals {
1872 self.argv
1873 .get(i)
1874 .map(bytes)
1875 .filter(|next| {
1876 !is_separator(next)
1877 && (flag.allow_hyphen_values
1878 || !is_flag_like(next)
1879 || (flag.allow_negative_numbers && is_negative_number(next)))
1880 })
1881 .inspect(|_| i += 1)
1882 } else {
1883 None
1884 };
1885 if first.is_none() && !flag.value_optional && flag.default_missing.is_none() {
1886 return at;
1887 }
1888 if flag.variadic {
1889 let mut count = first.map_or(0, |v| values_in(v, flag.delimiter));
1890 while flag.var_max.is_none_or(|max| count < max) {
1891 let Some(next) = self.argv.get(i).map(bytes) else {
1892 break;
1893 };
1894 if is_separator(next) {
1895 break;
1896 }
1897 if next == b"--" || flag.value_terminator.is_some_and(|end| end == next) {
1898 if next != b"--" {
1899 i += 1;
1900 }
1901 break;
1902 }
1903 if is_flag_like(next)
1904 && !(flag.allow_negative_numbers && is_negative_number(next))
1905 {
1906 break;
1907 }
1908 if self.cmd.subcommand_precedence_over_arg
1909 && self.find_subcommand(next).is_some()
1910 {
1911 return None;
1912 }
1913 count += values_in(next, flag.delimiter);
1914 i += 1;
1915 }
1916 }
1917 }
1918 }
1919 at
1920 }
1921
1922 /// Restrict inherited root globals to those carried by an executable view.
1923 #[cfg(feature = "spec")]
1924 pub fn with_view(mut self, view: &'t spec::ViewMeta<'t>) -> Self {
1925 self.view = Some(view);
1926 self
1927 }
1928
1929 /// The command in scope: the root, or the deepest subcommand selected so far.
1930 pub fn command(&self) -> &'t Command<'t> {
1931 self.cmd
1932 }
1933
1934 /// Whether a `--` was consumed as a separator.
1935 ///
1936 /// False when flag interpretation stopped for another reason, such as an
1937 /// `automatic` argument taking a value, and false for a `--` that a
1938 /// `preserve` argument kept as a value.
1939 pub fn double_dash_seen(&self) -> bool {
1940 self.separator_seen
1941 }
1942
1943 /// Every command entered so far, and where each one's own words begin.
1944 ///
1945 /// The ancestors are already kept for flag scoping; this is the same chain with the offsets,
1946 /// which is what lets a completion hand a callback the words of *its* command rather than of
1947 /// the deepest one — a global flag is declared on an ancestor.
1948 pub fn command_path(&self) -> Vec<(&'t Command<'t>, usize)> {
1949 let mut out = Vec::with_capacity(self.depth + 1);
1950 for (i, ancestor) in self.ancestors[..self.depth].iter().enumerate() {
1951 if let Some(cmd) = ancestor {
1952 // An ancestor's own words start where the one before it descended, and the
1953 // root's start at the beginning.
1954 out.push((*cmd, self.starts[i]));
1955 }
1956 }
1957 out.push((self.cmd, self.cmd_start));
1958 out
1959 }
1960
1961 /// The `argv` range the `help` word resolved as a command path.
1962 ///
1963 /// Empty unless the word was typed. Every token in it named a subcommand of the one before
1964 /// it — the parser resolved them itself, so nothing here is a flag or a flag's value.
1965 pub fn help_span(&self) -> (usize, usize) {
1966 self.help_span
1967 }
1968
1969 /// Where the command in scope began: the index in `argv` just after its name, or at the
1970 /// unmatched word routed into a default subcommand.
1971 ///
1972 /// `argv[command_start()..]` is what that command was given, which is what a completion
1973 /// callback needs to be handed its own command's half-parsed struct rather than the root's.
1974 pub fn command_start(&self) -> usize {
1975 self.cmd_start
1976 }
1977
1978 /// Whether flag interpretation has stopped, for any reason.
1979 ///
1980 /// Wider than [`double_dash_seen`](Self::double_dash_seen), and the question completion
1981 /// asks: past a separator *or* past the first value of an `automatic` argument, a
1982 /// dash-prefixed word is a value, so there is no flag there to offer.
1983 pub fn flags_stopped(&self) -> bool {
1984 self.flags_stopped
1985 }
1986
1987 /// A variadic flag that is still claiming words.
1988 ///
1989 /// Asked *between* events, because the answer is gone by the end: the call that finds argv
1990 /// exhausted is the one that clears it. A completion needs it — the next word after
1991 /// `--tools a ⌶` is another tool, not the positional that follows.
1992 pub fn collecting(&self) -> Option<&'t Flag<'t>> {
1993 self.collecting
1994 }
1995
1996 /// The positional the next word would fill, if there is one left.
1997 ///
1998 /// A variadic stays here until it reaches its bound, which is what makes it the answer to
1999 /// "what could go where the cursor is" as many times as it can be filled.
2000 pub fn pending_arg(&self) -> Option<&'t Arg<'t>> {
2001 self.next_arg()
2002 }
2003
2004 /// Flags a word here could name: this command's own, then any ancestor's globals.
2005 ///
2006 /// The same set the parser itself would look in, so what is offered and what is accepted
2007 /// cannot disagree — including the shadowing rule, where a subcommand redeclaring an
2008 /// inherited name hides it.
2009 pub fn flags_in_scope(&self) -> impl Iterator<Item = &'t Flag<'t>> + '_ {
2010 self.in_scope()
2011 }
2012
2013 /// Read the next event.
2014 ///
2015 /// Returns `None` when `argv` is exhausted. An `Err` is terminal: the parse
2016 /// stops there, since continuing past a token that could not be understood
2017 /// would only produce bindings derived from a guess. Events already yielded
2018 /// before an error are therefore not a partial result to be used — a caller
2019 /// that assigned them into fields should discard the whole attempt.
2020 ///
2021 /// One case is stronger than that, because the grammar demands it: a short
2022 /// bundle containing an unrecognized letter yields the error *instead of*, not
2023 /// after, the letters that did match.
2024 #[allow(clippy::should_implement_trait)] // not an Iterator: items borrow from self's tables
2025 pub fn next_event(&mut self) -> Option<Result<Event<'t, 'a, 'v>, Error<'t, 'v>>> {
2026 if self.done {
2027 return None;
2028 }
2029 let event = self.step();
2030 if matches!(event, Some(Ok(Event::Flag { .. } | Event::Arg { .. }))) {
2031 self.command_arg_found = true;
2032 }
2033 if let Some(Err(_)) = event {
2034 self.done = true;
2035 }
2036 event
2037 }
2038
2039 /// Count parent flags in the shared boundary token before entering the child.
2040 fn default_bundle_has_parent_flag(&self, default: &Command<'_>) -> bool {
2041 let Some(token) = self.argv.get(self.pos).map(bytes) else {
2042 return false;
2043 };
2044 if !token.starts_with(b"-") || token.starts_with(b"--") {
2045 return false;
2046 }
2047 for byte in &token[1..] {
2048 if self.find_short(*byte).is_some() {
2049 return true;
2050 }
2051 match default.flags.iter().find(|f| f.shorts.contains(byte)) {
2052 Some(flag) if !flag.takes_value => {}
2053 _ => break,
2054 }
2055 }
2056 false
2057 }
2058
2059 fn step(&mut self) -> Option<Result<Event<'t, 'a, 'v>, Error<'t, 'v>>> {
2060 if self.default_flag_at == Some(self.pos) && self.bundle.is_empty() {
2061 self.default_flag_at = None;
2062 if let Some(default) = self.cmd.default_subcommand {
2063 if self.cmd.args_conflicts_with_subcommands
2064 && (self.command_arg_found || self.default_bundle_has_parent_flag(default))
2065 {
2066 return Some(Err(Error::SubcommandConflict {
2067 subcommand: default,
2068 }));
2069 }
2070 if self.argv.get(self.pos).is_some_and(|word| {
2071 let token = bytes(word);
2072 token.starts_with(b"-") && !token.starts_with(b"--")
2073 }) {
2074 self.default_bundle_end = self.pos + 1;
2075 }
2076 self.default_taken = true;
2077 return Some(self.descend(default).map(|()| Event::Command(default)));
2078 }
2079 }
2080 if let Some(clause) = self.pending_clause_boundary.take() {
2081 self.arg_pos = 0;
2082 self.arg_taken = 0;
2083 self.arg_filled = false;
2084 self.collecting = None;
2085 self.flags_stopped = false;
2086 return Some(Ok(Event::ClauseSeparator { clause }));
2087 }
2088 // A partly-read short bundle takes priority: its remaining bytes are
2089 // still part of the token being processed.
2090 if !self.bundle.is_empty() {
2091 return Some(self.short_flag());
2092 }
2093
2094 if self.cmd.subcommand_precedence_over_arg && !self.flags_stopped {
2095 if let Some(token) = self.argv.get(self.pos).map(bytes) {
2096 if let Some(sub) = self.find_subcommand(token) {
2097 if self.cmd.args_conflicts_with_subcommands && self.command_arg_found {
2098 return Some(Err(Error::SubcommandConflict { subcommand: sub }));
2099 }
2100 self.pos += 1;
2101 return Some(self.descend(sub).map(|()| Event::Command(sub)));
2102 }
2103 }
2104 }
2105
2106 // A variadic flag keeps claiming tokens until one of them could be
2107 // something else.
2108 if let Some(flag) = self.collecting {
2109 match self.argv.get(self.pos) {
2110 Some(next)
2111 if flag
2112 .value_terminator
2113 .is_some_and(|terminator| bytes(next) == terminator) =>
2114 {
2115 self.pos += 1;
2116 self.collecting = None;
2117 return self.step();
2118 }
2119 Some(next)
2120 if (!is_flag_like(bytes(next))
2121 || (flag.allow_negative_numbers && is_negative_number(bytes(next))))
2122 && self.clause_separator(bytes(next)).is_none()
2123 && bytes(next) != b"--" =>
2124 {
2125 self.pos += 1;
2126 self.collected += values_in(bytes(next), flag.delimiter);
2127 // Same rule as a positional: a bounded occurrence takes that many and
2128 // leaves the rest to whatever follows.
2129 if flag.var_max.is_some_and(|max| self.collected >= max) {
2130 self.collecting = None;
2131 }
2132 // Stopping is only the same as staying within the bound while one word
2133 // is one value. A delimited word can carry the occurrence past it in a
2134 // single step, and that word cannot be split between two owners, so the
2135 // overshoot is an error rather than a place to stop.
2136 if let Some(max) = flag.var_max.filter(|max| self.collected > *max) {
2137 return Some(Err(Error::VarTooMany {
2138 name: flag.name,
2139 max: max as usize,
2140 got: self.collected as usize,
2141 }));
2142 }
2143 return Some(Ok(Event::Flag {
2144 flag,
2145 value: Some(bytes(next)),
2146 negated: false,
2147 }));
2148 }
2149 // A token that could be something else ends the run — but the *end of argv*
2150 // decides nothing. Clearing there threw away the answer to "would the next
2151 // word be claimed?", which is the question a completion asks and no parse
2152 // ever does: once argv is exhausted there are no more events either way.
2153 Some(_) => self.collecting = None,
2154 None => {}
2155 }
2156 }
2157
2158 let token = bytes(self.argv.get(self.pos)?);
2159 self.pos += 1;
2160
2161 // A clause separator remains syntax after an automatic positional stopped flags.
2162 // Only an explicit `--` protects a literal separator.
2163 if let Some(clause) = self.clause_separator(token) {
2164 self.arg_pos = 0;
2165 self.arg_taken = 0;
2166 self.arg_filled = false;
2167 self.collecting = None;
2168 self.flags_stopped = false;
2169 return Some(Ok(Event::ClauseSeparator { clause }));
2170 }
2171
2172 // An automatic trailing argument stops flag interpretation without consuming an
2173 // explicit separator. A later `--` must still unlock a required trailing argument
2174 // (clap's `last`), while a separator already consumed makes every later `--` data.
2175 if self.flags_stopped && (token != b"--" || self.separator_seen) {
2176 return Some(self.word(token));
2177 }
2178
2179 if token == b"--" {
2180 // `preserve` wants the separator itself as a value, so ask the
2181 // argument that would receive it before treating it as syntax.
2182 if self
2183 .next_arg()
2184 .is_some_and(|a| a.double_dash == DoubleDash::Preserve)
2185 {
2186 return Some(self.word(token));
2187 }
2188 self.flags_stopped = true;
2189 self.separator_seen = true;
2190 // An explicit separator unlocks any argument that required one, even
2191 // if earlier arguments are still unfilled.
2192 if let Some(idx) = self.current_args()[self.arg_pos..]
2193 .iter()
2194 .position(|a| a.double_dash == DoubleDash::Required)
2195 {
2196 // The count belongs to the argument at `arg_pos`, so jumping past it has
2197 // to leave the count behind: a bounded variadic before the separator would
2198 // otherwise lend its total to the argument after it, which then stops
2199 // early or at once.
2200 self.arg_pos += idx;
2201 self.arg_taken = 0;
2202 }
2203 return self.step();
2204 }
2205
2206 if self.arg_taken > 0
2207 && self.next_arg().is_some_and(|arg| {
2208 arg.value_terminator
2209 .is_some_and(|terminator| token == terminator)
2210 })
2211 {
2212 self.advance_arg();
2213 return self.step();
2214 }
2215
2216 // An exact declared short outranks the numeric shape. This keeps ordinary negative
2217 // numbers available as values while allowing clap-compatible spellings such as fd's
2218 // `-0` / `--print0` switch.
2219 let declared_numeric_short = matches!(token, [b'-', short]
2220 if short.is_ascii_digit() && self.find_short(*short).is_some());
2221
2222 if !declared_numeric_short
2223 && is_negative_number(token)
2224 && self
2225 .next_arg()
2226 .is_some_and(|arg| arg.allow_negative_numbers)
2227 {
2228 return Some(self.word(token));
2229 }
2230
2231 if !declared_numeric_short
2232 && is_negative_number(token)
2233 && self.cmd.external_subcommand
2234 && !self.arg_filled
2235 {
2236 return Some(self.word(token));
2237 }
2238
2239 if is_flag_like(token) {
2240 if token.starts_with(b"--") {
2241 return Some(self.long_flag(token));
2242 }
2243 // Check the whole bundle before emitting anything from it. Events go
2244 // out one at a time, so discovering an unknown letter half way
2245 // through would mean the earlier letters had already been applied —
2246 // and the grammar rejects the entire token, not the tail of it.
2247 match self.check_bundle(token) {
2248 Ok(()) => {}
2249 // Unrecognized, so it is a word unless this command wants it refused.
2250 Err(e) if self.unknown_flags == UnknownFlags::Error => {
2251 return Some(Err(e));
2252 }
2253 Err(_) => return Some(self.word(token)),
2254 }
2255 self.bundle = &token[1..];
2256 self.bundle_token = token;
2257 return Some(self.short_flag());
2258 }
2259
2260 Some(self.word(token))
2261 }
2262
2263 fn long_flag(&mut self, token: &'v [u8]) -> Result<Event<'t, 'a, 'v>, Error<'t, 'v>> {
2264 let body = &token[2..];
2265 let (name, attached) = match body.iter().position(|&b| b == b'=') {
2266 Some(i) => (&body[..i], Some(&body[i + 1..])),
2267 None => (body, None),
2268 };
2269
2270 if let Some(flag) = self.find_long(name) {
2271 let value = if flag.takes_value {
2272 match attached {
2273 Some(v) => Some(v),
2274 None => self.take_detached_value(flag)?,
2275 }
2276 } else if flag.bool_value {
2277 validate_bool_value(flag, attached)?
2278 } else {
2279 None
2280 };
2281 if flag.variadic {
2282 if let Some(value) = value {
2283 self.start_collecting(flag, value)?;
2284 }
2285 }
2286 if let Some(error) = self.flag_action(flag, true) {
2287 return Err(error);
2288 }
2289 return Ok(Event::Flag {
2290 flag,
2291 value,
2292 negated: false,
2293 });
2294 }
2295
2296 if let Some(flag) = self.find_negation(name) {
2297 return Ok(Event::Flag {
2298 flag,
2299 value: if flag.bool_value {
2300 validate_bool_value(flag, attached)?
2301 } else {
2302 None
2303 },
2304 negated: true,
2305 });
2306 }
2307
2308 // Where the CLI declared a version, `--version` answers with it — asked after the
2309 // command's own flags, so a CLI declaring its own keeps it.
2310 let version_command = self.version_command();
2311 if name == b"version" && version_command.version && !version_command.disable_version_flag {
2312 return Ok(Event::Flag {
2313 flag: &VERSION_LONG,
2314 value: None,
2315 negated: false,
2316 });
2317 }
2318
2319 // Every CLI answers to `--help`, and none of them declares it. Asked *after* the
2320 // command's own flags, so a CLI that declares its own `--help` keeps it.
2321 if name == b"help" && !self.cmd.disable_help_flag {
2322 return Ok(Event::Flag {
2323 flag: &HELP_LONG,
2324 value: None,
2325 negated: false,
2326 });
2327 }
2328
2329 if self.unknown_flags == UnknownFlags::Error {
2330 return Err(Error::UnknownFlag { token });
2331 }
2332 // Not a flag here, so it is a word like any other.
2333 self.word(token)
2334 }
2335
2336 /// Walk a short-flag token without binding anything, to find out whether all
2337 /// of it is recognized.
2338 ///
2339 /// Scanning stops at the first letter whose flag takes a value, because
2340 /// everything after it is that value rather than more letters.
2341 fn check_bundle(&self, token: &'v [u8]) -> Result<(), Error<'t, 'v>> {
2342 let mut rest = &token[1..];
2343 while let Some((&byte, tail)) = rest.split_first() {
2344 match self.find_short(byte) {
2345 None => return Err(Error::UnknownFlag { token }),
2346 Some(flag) if flag.takes_value => return Ok(()),
2347 Some(_) => rest = tail,
2348 }
2349 }
2350 Ok(())
2351 }
2352
2353 fn short_flag(&mut self) -> Result<Event<'t, 'a, 'v>, Error<'t, 'v>> {
2354 let byte = self.bundle[0];
2355 let rest = &self.bundle[1..];
2356
2357 let Some(flag) = self.find_short(byte) else {
2358 // check_bundle already rejected any token containing an unrecognized
2359 // letter, so this is unreachable — but a parser should report rather
2360 // than panic if that ever stops being true.
2361 self.bundle = &[];
2362 return Err(Error::UnknownFlag {
2363 token: self.bundle_token,
2364 });
2365 };
2366
2367 if !flag.takes_value {
2368 self.bundle = rest;
2369 if let Some(error) = self.flag_action(flag, false) {
2370 self.bundle = &[];
2371 return Err(error);
2372 }
2373 return Ok(Event::Flag {
2374 flag,
2375 value: None,
2376 negated: false,
2377 });
2378 }
2379
2380 // A value-taking short ends the token: everything after it is the value,
2381 // less one separating `=`.
2382 self.bundle = &[];
2383 let value = if rest.is_empty() {
2384 self.take_detached_value(flag)?
2385 } else if rest[0] == b'=' {
2386 Some(&rest[1..])
2387 } else {
2388 Some(rest)
2389 };
2390 if flag.variadic {
2391 if let Some(value) = value {
2392 self.start_collecting(flag, value)?;
2393 }
2394 }
2395 if let Some(error) = self.flag_action(flag, false) {
2396 return Err(error);
2397 }
2398 Ok(Event::Flag {
2399 flag,
2400 value,
2401 negated: false,
2402 })
2403 }
2404
2405 fn flag_action(&self, flag: &'t Flag<'t>, long_spelling: bool) -> Option<Error<'t, 'v>> {
2406 if matches!(
2407 flag.key,
2408 HELP_LONG_KEY | HELP_SHORT_KEY | VERSION_LONG_KEY | VERSION_SHORT_KEY
2409 ) || !self.action_errors
2410 {
2411 return None;
2412 }
2413 match flag.action {
2414 ArgAction::Set => None,
2415 ArgAction::Help => Some(Error::Help {
2416 cmd: self.cmd,
2417 long: long_spelling,
2418 }),
2419 ArgAction::HelpShort => Some(Error::Help {
2420 cmd: self.cmd,
2421 long: false,
2422 }),
2423 ArgAction::HelpLong => Some(Error::Help {
2424 cmd: self.cmd,
2425 long: true,
2426 }),
2427 ArgAction::HelpAll => Some(Error::HelpAll { cmd: self.cmd }),
2428 ArgAction::Version => Some(Error::Version {
2429 long: long_spelling,
2430 }),
2431 }
2432 }
2433
2434 /// Take the following token as a flag's value.
2435 ///
2436 /// Refuses a flag-like token unless [`Flag::allow_hyphen_values`] is set:
2437 /// `--jobs --force` is far more likely a forgotten value than a deliberate
2438 /// one, and the attached form is available for the deliberate case. Declared,
2439 /// the next token is taken whatever it looks like, including `--`.
2440 fn take_detached_value(
2441 &mut self,
2442 flag: &'t Flag<'t>,
2443 ) -> Result<Option<&'v [u8]>, Error<'t, 'v>> {
2444 if flag.require_equals {
2445 return self.missing_or_default(flag);
2446 }
2447 match self.argv.get(self.pos) {
2448 Some(next)
2449 if self.clause_separator(bytes(next)).is_none()
2450 && (flag.allow_hyphen_values
2451 || !is_flag_like(bytes(next))
2452 || (flag.allow_negative_numbers && is_negative_number(bytes(next)))) =>
2453 {
2454 self.pos += 1;
2455 Ok(Some(bytes(next)))
2456 }
2457 _ => self.missing_or_default(flag),
2458 }
2459 }
2460
2461 fn missing_or_default(&self, flag: &'t Flag<'t>) -> Result<Option<&'v [u8]>, Error<'t, 'v>> {
2462 match flag.default_missing {
2463 Some(value) => Ok(Some(value)),
2464 None if flag.value_optional => Ok(None),
2465 None => Err(Error::MissingFlagValue { flag }),
2466 }
2467 }
2468
2469 fn clause_separator(&self, token: &[u8]) -> Option<Clause<'t>> {
2470 (!self.separator_seen)
2471 .then_some(self.cmd.clause)
2472 .flatten()
2473 .filter(|clause| clause.separator.is_some_and(|separator| token == separator))
2474 }
2475
2476 fn word(&mut self, token: &'v [u8]) -> Result<Event<'t, 'a, 'v>, Error<'t, 'v>> {
2477 // Subcommands are only matched where descent is still possible: once a
2478 // positional of this command has taken a word, a later word that happens
2479 // to equal a subcommand name is just a value.
2480 if !self.arg_filled && !self.flags_stopped {
2481 if let Some(sub) = self.find_subcommand(token) {
2482 if self.cmd.args_conflicts_with_subcommands && self.command_arg_found {
2483 return Err(Error::SubcommandConflict { subcommand: sub });
2484 }
2485 self.descend(sub)?;
2486 return Ok(Event::Command(sub));
2487 }
2488
2489 // `ex help config ls` — the line every page with a Commands section has printed
2490 // all along ("help Print this message or the help of the given subcommand(s)"),
2491 // and which until now did nothing. The page is what decides the condition here:
2492 // it prints that line where there are subcommands, so that is where the word is
2493 // answered, and to a leaf `help` is a word like any other.
2494 //
2495 // Asked *after* the subcommand lookup, so a CLI that declares a `help` of its own
2496 // keeps it — the same rule the two help flags follow.
2497 //
2498 // The words after it name a command, resolved here rather than descended into:
2499 // descending would bind them, and they are a question rather than an invocation.
2500 if token == b"help"
2501 && !self.cmd.disable_help_subcommand
2502 && !self.cmd.subcommands.is_empty()
2503 {
2504 let mut cmd = self.cmd;
2505 let from = self.pos;
2506 while let Some(next) = self.argv.get(self.pos) {
2507 let Some(sub) = find_named(cmd, bytes(next)) else {
2508 break;
2509 };
2510 cmd = sub;
2511 self.pos += 1;
2512 }
2513 // Kept for `help::route_to`: which mount was asked about is not recoverable
2514 // from `cmd`, since two mounts of one `Subcommands` type are one address.
2515 self.help_span = (from, self.pos);
2516 // The long form, as `ex config --help` gives: someone who typed a whole word to
2517 // ask for help wants the fuller answer.
2518 return Err(Error::Help { cmd, long: true });
2519 }
2520
2521 // A word that names no subcommand goes to the default one, if there is one.
2522 //
2523 // Only a word, though. A dash-prefixed token that named no flag arrives here as a
2524 // value — that is what `unknown_flags = value` means — and it was never a
2525 // candidate to *select* anything, so it binds where it was typed. usage-lib stops
2526 // looking for subcommands at an unrecognised flag for the same reason. `--` is
2527 // excluded on the same grounds: it reaches this function only when a `preserve`
2528 // argument wants it as a value.
2529 //
2530 // The token is *not* consumed: the cursor steps back so the next event reads it
2531 // again, now against the command just descended into. That is what lets it be a
2532 // subcommand of the default (`mise build` where `build` is a task the mount
2533 // added) as easily as an argument of it, without this function having to decide
2534 // which — and without yielding two events for one word.
2535 if let Some(default) = self.cmd.default_subcommand {
2536 // `-` joins `--` in being excluded, and for the reason already written above:
2537 // a value was never a candidate to *select* anything. `is_flag_like` calls a
2538 // lone `-` a value — conventionally stdin — so it passed this guard and
2539 // descended, where mise's `run` has no positional and the parse failed.
2540 // usage-lib and clap both bind it to the root's own `[TASK]` instead.
2541 let default_accepts_negative = is_negative_number(token)
2542 && default
2543 .args
2544 .first()
2545 .is_some_and(|arg| arg.allow_negative_numbers);
2546 if !self.default_taken
2547 && (!is_flag_like(token) || default_accepts_negative)
2548 && token != b"--"
2549 && token != b"-"
2550 {
2551 self.default_taken = true;
2552 self.descend(default)?;
2553 self.pos -= 1;
2554 // Unlike an explicitly named command, the default command receives the
2555 // word that caused descent. Keep its argv boundary at that word so
2556 // command-level policies and completion callbacks see the same input the
2557 // command parser is about to re-read.
2558 self.cmd_start = self.pos;
2559 return Ok(Event::Command(default));
2560 }
2561 }
2562
2563 // Known subcommands and the default route keep precedence. A sigil positional
2564 // then claims its classified word before an external-subcommand catch-all can.
2565 if let Some((arg, sigil)) = self.match_sigil_arg(token) {
2566 if token.len() == sigil.len() {
2567 return Err(invalid_value_error(
2568 arg.name,
2569 as_str(token).unwrap_or_default().to_string(),
2570 format!(
2571 "expected a value after sigil {:?}",
2572 as_str(sigil).unwrap_or_default()
2573 ),
2574 ));
2575 }
2576 return Ok(Event::Arg {
2577 arg,
2578 value: &token[sigil.len()..],
2579 delimit: true,
2580 });
2581 }
2582
2583 // An unmatched word that names no subcommand is forwarded as an external
2584 // command: this word, then every token after it, including flags. Known
2585 // subcommands already won above, and a default_subcommand already caught.
2586 if self.cmd.external_subcommand
2587 && (!is_flag_like(token) || is_negative_number(token))
2588 && token != b"--"
2589 && token != b"-"
2590 {
2591 let from = self.pos - 1;
2592 self.pos = self.argv.len();
2593 return Ok(Event::External {
2594 values: &self.argv[from..],
2595 });
2596 }
2597 }
2598
2599 if self.arg_filled && !self.flags_stopped {
2600 if let Some((arg, sigil)) = self.match_sigil_arg(token) {
2601 if token.len() == sigil.len() {
2602 return Err(invalid_value_error(
2603 arg.name,
2604 as_str(token).unwrap_or_default().to_string(),
2605 format!(
2606 "expected a value after sigil {:?}",
2607 as_str(sigil).unwrap_or_default()
2608 ),
2609 ));
2610 }
2611 return Ok(Event::Arg {
2612 arg,
2613 value: &token[sigil.len()..],
2614 delimit: true,
2615 });
2616 }
2617 }
2618
2619 self.skip_sigil_args();
2620 self.reserve_for_required_positionals();
2621 let Some(arg) = self.next_arg() else {
2622 return Err(Error::UnexpectedArg { token });
2623 };
2624
2625 if arg.double_dash == DoubleDash::Required && !self.separator_seen {
2626 return Err(Error::ArgRequiresDoubleDash { arg });
2627 }
2628
2629 self.arg_filled = true;
2630 // An `automatic` argument stops flag interpretation from here on, as
2631 // though the caller had typed the separator themselves.
2632 let trailing_value = self.separator_seen || arg.double_dash == DoubleDash::Automatic;
2633 let delimit = !(self.dont_delimit_trailing_values && trailing_value);
2634 if arg.double_dash == DoubleDash::Automatic {
2635 self.flags_stopped = true;
2636 }
2637 // A variadic keeps taking values, so the cursor stays put — until it reaches its
2638 // bound, at which point the words after it belong to whatever comes next. That is
2639 // what makes `[a]… [b]` expressible at all.
2640 if arg.var {
2641 self.arg_taken += values_in(token, delimit.then_some(arg.delimiter).flatten());
2642 // Before advancing, which resets the count: as with a variadic flag, reaching
2643 // the bound and passing it are the same event once a word can carry several
2644 // values, and only the second is a mistake.
2645 if let Some(max) = arg.var_max.filter(|max| self.arg_taken > *max) {
2646 return Err(Error::VarTooMany {
2647 name: arg.name,
2648 max: max as usize,
2649 got: self.arg_taken as usize,
2650 });
2651 }
2652 if arg.var_max.is_some_and(|max| self.arg_taken >= max) {
2653 self.advance_arg();
2654 }
2655 } else {
2656 self.advance_arg();
2657 }
2658 if self
2659 .cmd
2660 .clause
2661 .is_some_and(|clause| clause.separator.is_none() && self.next_arg().is_none())
2662 {
2663 self.pending_clause_boundary = self.cmd.clause;
2664 }
2665 Ok(Event::Arg {
2666 arg,
2667 value: token,
2668 delimit,
2669 })
2670 }
2671
2672 fn descend(&mut self, sub: &'t Command<'t>) -> Result<(), Error<'t, 'v>> {
2673 if self.depth >= MAX_DEPTH {
2674 return Err(Error::TooDeep);
2675 }
2676 self.ancestors[self.depth] = Some(self.cmd);
2677 self.starts[self.depth] = self.cmd_start;
2678 self.depth += 1;
2679 self.cmd = sub;
2680 // Only a command that says something changes it, which is what inheriting means.
2681 if let ::core::option::Option::Some(mode) = sub.unknown_flags {
2682 self.unknown_flags = mode;
2683 }
2684 self.dont_delimit_trailing_values |= sub.dont_delimit_trailing_values;
2685 // Where this command's own words start, which is what lets a completion hand a callback
2686 // the half-parsed struct of the command it was declared on rather than of the root.
2687 self.cmd_start = self.pos;
2688 self.arg_pos = 0;
2689 self.arg_taken = 0;
2690 self.arg_filled = false;
2691 self.command_arg_found = false;
2692 Ok(())
2693 }
2694
2695 /// Move to the next positional, forgetting what the last one took.
2696 fn advance_arg(&mut self) {
2697 self.skip_sigil_args();
2698 self.arg_pos += 1;
2699 self.arg_taken = 0;
2700 self.skip_sigil_args();
2701 }
2702
2703 /// A variadic flag occurrence begins, counting from zero.
2704 ///
2705 /// The value it was given on the same token counts, which is why this starts at what
2706 /// that value holds: `--include a b` with `var_max=2` takes `a` and `b`, not three
2707 /// words — and `--include a,b` has already taken both on the one token.
2708 fn start_collecting(&mut self, flag: &'t Flag<'t>, first: &[u8]) -> Result<(), Error<'t, 'v>> {
2709 self.collected = values_in(first, flag.delimiter);
2710 if let Some(max) = flag.var_max.filter(|max| self.collected > *max) {
2711 return Err(Error::VarTooMany {
2712 name: flag.name,
2713 max: max as usize,
2714 got: self.collected as usize,
2715 });
2716 }
2717 self.collecting = if flag.var_max.is_some_and(|max| self.collected >= max) {
2718 None
2719 } else {
2720 Some(flag)
2721 };
2722 Ok(())
2723 }
2724
2725 fn next_arg(&self) -> Option<&'t Arg<'t>> {
2726 self.current_args()[self.arg_pos..]
2727 .iter()
2728 .find(|arg| arg.sigil.is_none())
2729 .copied()
2730 }
2731
2732 fn skip_sigil_args(&mut self) {
2733 while self
2734 .current_args()
2735 .get(self.arg_pos)
2736 .is_some_and(|arg| arg.sigil.is_some())
2737 {
2738 self.arg_pos += 1;
2739 }
2740 }
2741
2742 fn match_sigil_arg(&self, token: &[u8]) -> Option<(&'t Arg<'t>, &'t [u8])> {
2743 if self.flags_stopped {
2744 return None;
2745 }
2746 let own = self.current_args().iter().copied();
2747 let inherited = self.ancestors[..self.depth]
2748 .iter()
2749 .rev()
2750 .filter_map(|cmd| *cmd)
2751 .flat_map(|cmd| cmd.args.iter().copied());
2752 own.chain(inherited)
2753 .filter_map(|arg| {
2754 let sigil = arg.sigil?;
2755 (token.len() >= sigil.len() && token.starts_with(sigil)).then_some((arg, sigil))
2756 })
2757 .max_by_key(|(_, sigil)| sigil.len())
2758 }
2759
2760 fn current_args(&self) -> &'t [&'t Arg<'t>] {
2761 self.cmd
2762 .clause
2763 .map(|clause| clause.args)
2764 .unwrap_or(self.cmd.args)
2765 }
2766
2767 /// Skip empty optional positionals when every remaining value is needed by a later
2768 /// required positional. This is clap's opt-in `allow_missing_positional` policy.
2769 fn reserve_for_required_positionals(&mut self) {
2770 if !self.cmd.allow_missing_positional || self.arg_taken != 0 {
2771 return;
2772 }
2773 loop {
2774 let Some(current) = self.next_arg() else {
2775 return;
2776 };
2777 if current.required {
2778 return;
2779 }
2780 let required_after = self.current_args()[self.arg_pos + 1..]
2781 .iter()
2782 .filter(|arg| arg.required && arg.sigil.is_none())
2783 .count();
2784 if required_after == 0 {
2785 return;
2786 }
2787 let remaining_values = 1 + self.argv[self.pos..]
2788 .iter()
2789 .filter(|word| {
2790 (self.flags_stopped || !is_flag_like(bytes(word)))
2791 && self.match_sigil_arg(bytes(word)).is_none()
2792 })
2793 .count();
2794 if remaining_values > required_after {
2795 return;
2796 }
2797 self.advance_arg();
2798 }
2799 }
2800
2801 #[cfg(feature = "spec")]
2802 fn view_allows_own_flag(&self, flag: &Flag<'_>) -> bool {
2803 match self.view {
2804 None => true,
2805 // The promoted command keeps its own surface. While the injected path is
2806 // still at the host root, however, only explicitly carried globals belong
2807 // to the view; root-local flags are not part of the projected executable.
2808 Some(view) => {
2809 !core::ptr::eq(self.cmd, self.root) || is_version_flag(flag) || view.carries(flag)
2810 }
2811 }
2812 }
2813
2814 #[cfg(not(feature = "spec"))]
2815 fn view_allows_own_flag(&self, _flag: &Flag<'_>) -> bool {
2816 true
2817 }
2818
2819 #[cfg(feature = "spec")]
2820 fn view_allows_inherited_flag(&self, flag: &Flag<'_>) -> bool {
2821 match self.view {
2822 None => true,
2823 // A portable view carries selected host globals, not globals declared
2824 // by intermediate commands on a multi-segment promoted path.
2825 Some(view) => {
2826 self.root
2827 .flags
2828 .iter()
2829 .any(|root| core::ptr::eq(*root, flag))
2830 && (is_version_flag(flag) || view.carries(flag))
2831 }
2832 }
2833 }
2834
2835 #[cfg(not(feature = "spec"))]
2836 fn view_allows_inherited_flag(&self, _flag: &Flag<'_>) -> bool {
2837 true
2838 }
2839
2840 #[cfg(feature = "spec")]
2841 fn inherited_flag_is_in_scope(&self, flag: &Flag<'_>) -> bool {
2842 flag.global || (self.view.is_some() && is_version_flag(flag))
2843 }
2844
2845 #[cfg(not(feature = "spec"))]
2846 fn inherited_flag_is_in_scope(&self, flag: &Flag<'_>) -> bool {
2847 flag.global
2848 }
2849
2850 /// Flags in scope: this command's own, then any ancestor's globals.
2851 ///
2852 /// Own flags come first so that a subcommand redeclaring an inherited name
2853 /// shadows it, which is what mise relies on when it redeclares root globals
2854 /// on `run` with different shorts.
2855 fn in_scope(&self) -> impl Iterator<Item = &'t Flag<'t>> + '_ {
2856 let own = self
2857 .cmd
2858 .flags
2859 .iter()
2860 .copied()
2861 .filter(|flag| self.view_allows_own_flag(flag));
2862 let inherited = self.ancestors[..self.depth]
2863 .iter()
2864 .rev()
2865 .filter_map(|c| *c)
2866 .flat_map(|c| c.flags.iter().copied())
2867 .filter(|flag| self.inherited_flag_is_in_scope(flag))
2868 .filter(|flag| self.view_allows_inherited_flag(flag));
2869 own.chain(inherited)
2870 }
2871
2872 fn find_long(&self, name: &[u8]) -> Option<&'t Flag<'t>> {
2873 self.in_scope()
2874 .find(|f| f.longs.iter().any(|l| l.as_bytes() == name))
2875 }
2876
2877 fn find_negation(&self, name: &[u8]) -> Option<&'t Flag<'t>> {
2878 self.in_scope()
2879 .find(|f| f.negate.is_some_and(|n| n.as_bytes() == name))
2880 }
2881
2882 fn find_short(&self, byte: u8) -> Option<&'t Flag<'t>> {
2883 // Only the boundary token is shared. Later tokens use ordinary child/global scope.
2884 if self.default_taken && self.pos == self.default_bundle_end {
2885 if let Some(flag) = self.ancestors[0].and_then(|parent| {
2886 parent
2887 .flags
2888 .iter()
2889 .copied()
2890 .find(|f| f.shorts.contains(&byte))
2891 }) {
2892 return Some(flag);
2893 }
2894 }
2895 self.in_scope()
2896 .find(|f| f.shorts.contains(&byte))
2897 // As for `--help`: supplied by the parser, and only where the command has not
2898 // declared a `-h` of its own.
2899 .or(if byte == b'h' && !self.cmd.disable_help_flag {
2900 Some(&HELP_SHORT)
2901 } else if byte == b'V'
2902 && self.version_command().version
2903 && !self.version_command().disable_version_flag
2904 {
2905 Some(&VERSION_SHORT)
2906 } else {
2907 None
2908 })
2909 }
2910
2911 #[cfg(feature = "spec")]
2912 fn version_command(&self) -> &'t Command<'t> {
2913 if self.view.is_some() {
2914 self.root
2915 } else {
2916 self.cmd
2917 }
2918 }
2919
2920 #[cfg(not(feature = "spec"))]
2921 fn version_command(&self) -> &'t Command<'t> {
2922 self.cmd
2923 }
2924
2925 fn find_subcommand(&self, name: &[u8]) -> Option<&'t Command<'t>> {
2926 // Shared with `help` rather than spelled out again, so descending into a command and
2927 // asking about one cannot drift apart.
2928 find_named(self.cmd, name)
2929 }
2930}
2931
2932/// View a token as bytes.
2933///
2934/// `as_encoded_bytes` is a plain accessor with no conversion and no allocation.
2935/// The reverse direction is the one with a cost — see [`os_string_from_bytes`] —
2936/// which is why values come back as bytes.
2937fn bytes<'v>(s: &&'v OsStr) -> &'v [u8] {
2938 s.as_encoded_bytes()
2939}
2940
2941/// How many values one word carries.
2942///
2943/// One, until a delimiter is declared — and then one per separator, counting the same way
2944/// splitting on it does: `a,b` is two, `a,` is two with an empty second, and `` is one.
2945/// Counted rather than split because binding only needs the number, and the split itself
2946/// belongs to the layer that owns the values.
2947fn values_in(word: &[u8], delimiter: ::core::option::Option<u8>) -> u32 {
2948 match delimiter {
2949 Some(d) => 1 + word.iter().filter(|b| **b == d).count() as u32,
2950 None => 1,
2951 }
2952}
2953
2954/// Whether a token should be read as a flag.
2955///
2956/// `-` alone is a value, conventionally stdin. Other dash-prefixed tokens are
2957/// flag-like; a field may make the narrower negative-number exception.
2958fn is_flag_like(token: &[u8]) -> bool {
2959 matches!(token, [b'-', rest @ ..] if !rest.is_empty())
2960}
2961
2962fn is_negative_number(token: &[u8]) -> bool {
2963 token.strip_prefix(b"-").is_some_and(is_number)
2964}
2965
2966/// Whether the text after a `-` is a number, so `-1`, `-2.5`, and `-1e5` are values
2967/// while `-1x` is a flag-shaped token that names nothing.
2968///
2969/// Digits, at most one `.`, and an optional exponent. Deliberately narrower than
2970/// `f64::from_str`, which also accepts `inf` and `NaN` — `-inf` is far likelier to be
2971/// a misspelled flag than a number somebody meant to pass.
2972///
2973/// usage-lib applies the same rule, and the corpus pins the edges so the two cannot
2974/// drift apart: they disagreed about `-1e5` when this was a hand-rolled scanner on
2975/// one side and a float parse on the other.
2976///
2977/// Written out rather than deferred to `f64::from_str` because this runs on the hot
2978/// path, and a parse would mean a UTF-8 check on a slice already decided by its
2979/// bytes.
2980fn is_number(rest: &[u8]) -> bool {
2981 let (mantissa, exponent) = match rest.iter().position(|b| matches!(b, b'e' | b'E')) {
2982 Some(at) => (&rest[..at], Some(&rest[at + 1..])),
2983 None => (rest, None),
2984 };
2985
2986 let mut seen_digit = false;
2987 let mut seen_dot = false;
2988 for &b in mantissa {
2989 match b {
2990 b'0'..=b'9' => seen_digit = true,
2991 b'.' if !seen_dot => seen_dot = true,
2992 _ => return false,
2993 }
2994 }
2995 if !seen_digit {
2996 return false;
2997 }
2998
2999 match exponent {
3000 None => true,
3001 // An exponent needs digits of its own, and may carry a sign.
3002 Some(exp) => {
3003 let digits = exp
3004 .strip_prefix(b"+")
3005 .or_else(|| exp.strip_prefix(b"-"))
3006 .unwrap_or(exp);
3007 !digits.is_empty() && digits.iter().all(|b| b.is_ascii_digit())
3008 }
3009 }
3010}
3011
3012fn validate_bool_value<'t, 'v>(
3013 flag: &'t Flag<'t>,
3014 value: Option<&'v [u8]>,
3015) -> Result<Option<&'v [u8]>, Error<'t, 'v>> {
3016 match value {
3017 None | Some(b"true" | b"false") => Ok(value),
3018 Some(_) => Err(Error::InvalidChoice {
3019 name: flag.name,
3020 choices: &["true", "false"],
3021 }),
3022 }
3023}
3024
3025#[cfg(test)]
3026mod tests {
3027 use super::*;
3028
3029 #[test]
3030 fn clause_separator_resets_inner_args_and_survives_automatic_mode() {
3031 static TASK: Arg = Arg {
3032 key: 91,
3033 name: "task",
3034 ..Arg::REQUIRED
3035 };
3036 static REST: Arg = Arg {
3037 key: 92,
3038 name: "args",
3039 double_dash: DoubleDash::Automatic,
3040 ..Arg::VAR
3041 };
3042 static ROOT: Command = Command {
3043 name: "ex",
3044 clause: Some(Clause {
3045 key: 90,
3046 name: "tasks",
3047 separator: Some(b":::"),
3048 flags: &[],
3049 args: &[&TASK, &REST],
3050 }),
3051 ..Command::EMPTY
3052 };
3053 let argv = [
3054 OsStr::new("lint"),
3055 OsStr::new("--fix"),
3056 OsStr::new(":::"),
3057 OsStr::new("test"),
3058 ];
3059 let mut parser = Parser::new(&ROOT, &argv);
3060 let mut seen = Vec::new();
3061 while let Some(event) = parser.next_event() {
3062 match event.expect("valid clause") {
3063 Event::Arg { arg, value, .. } => {
3064 seen.push((arg.name, String::from_utf8_lossy(value).into_owned()))
3065 }
3066 Event::ClauseSeparator { clause } => seen.push((clause.name, ":::".into())),
3067 _ => {}
3068 }
3069 }
3070 assert_eq!(
3071 seen,
3072 [
3073 ("task", "lint".into()),
3074 ("args", "--fix".into()),
3075 ("tasks", ":::".into()),
3076 ("task", "test".into())
3077 ]
3078 );
3079 }
3080
3081 #[test]
3082 fn clause_separator_is_not_a_scoped_flag_value() {
3083 static POSTINSTALL: Flag = Flag {
3084 key: 93,
3085 name: "postinstall",
3086 longs: &["postinstall"],
3087 takes_value: true,
3088 variadic: true,
3089 ..Flag::BOOL
3090 };
3091 static TOOL: Arg = Arg {
3092 key: 94,
3093 name: "tool",
3094 ..Arg::REQUIRED
3095 };
3096 static ROOT: Command = Command {
3097 name: "ex",
3098 flags: &[&POSTINSTALL],
3099 clause: Some(Clause {
3100 key: 95,
3101 name: "tools",
3102 separator: Some(b":::"),
3103 flags: &[&POSTINSTALL],
3104 args: &[&TOOL],
3105 }),
3106 ..Command::EMPTY
3107 };
3108 let argv = [OsStr::new("--postinstall"), OsStr::new(":::")];
3109 let mut parser = Parser::new(&ROOT, &argv);
3110 assert!(matches!(
3111 parser.next_event(),
3112 Some(Err(Error::MissingFlagValue { flag })) if flag.key == POSTINSTALL.key
3113 ));
3114
3115 let argv = [
3116 OsStr::new("--postinstall"),
3117 OsStr::new("setup"),
3118 OsStr::new(":::"),
3119 OsStr::new("tool"),
3120 ];
3121 let mut parser = Parser::new(&ROOT, &argv);
3122 assert!(matches!(
3123 parser.next_event(),
3124 Some(Ok(Event::Flag {
3125 value: Some(b"setup"),
3126 ..
3127 }))
3128 ));
3129 assert!(matches!(
3130 parser.next_event(),
3131 Some(Ok(Event::ClauseSeparator { clause })) if clause.key == 95
3132 ));
3133 assert!(matches!(
3134 parser.next_event(),
3135 Some(Ok(Event::Arg { value: b"tool", .. }))
3136 ));
3137 }
3138
3139 static FORCE: Flag = Flag {
3140 key: 1,
3141 longs: &["force"],
3142 shorts: b"f",
3143 ..Flag::BOOL
3144 };
3145 static EXPLICIT_BOOL: Flag = Flag {
3146 key: 20,
3147 name: "color",
3148 longs: &["color"],
3149 negate: Some("no-color"),
3150 bool_value: true,
3151 ..Flag::BOOL
3152 };
3153 static EXPLICIT_BOOL_ROOT: Command = Command {
3154 name: "ex",
3155 flags: &[&EXPLICIT_BOOL],
3156 ..Command::EMPTY
3157 };
3158 static JOBS: Flag = Flag {
3159 key: 2,
3160 longs: &["jobs"],
3161 shorts: b"j",
3162 allow_negative_numbers: true,
3163 ..Flag::VALUE
3164 };
3165 static COLOR: Flag = Flag {
3166 key: 3,
3167 longs: &["color"],
3168 negate: Some("no-color"),
3169 ..Flag::BOOL
3170 };
3171 static VERBOSE: Flag = Flag {
3172 key: 4,
3173 longs: &["verbose"],
3174 shorts: b"v",
3175 global: true,
3176 ..Flag::BOOL
3177 };
3178 static FILE: Arg = Arg {
3179 key: 10,
3180 name: "file",
3181 allow_negative_numbers: true,
3182 ..Arg::REQUIRED
3183 };
3184 static REST: Arg = Arg {
3185 key: 11,
3186 name: "rest",
3187 ..Arg::VAR
3188 };
3189 static INSTALL: Command = Command {
3190 name: "install",
3191 aliases: &["i"],
3192 flags: &[&FORCE],
3193 key: 100,
3194 ..Command::EMPTY
3195 };
3196 /// Same shape as ROOT, but a CLI that owns all of its flags. The subcommand says
3197 /// nothing and inherits it, which is the point: only the root declares the mode.
3198 static STRICT_INSTALL: Command = Command {
3199 name: "install",
3200 aliases: &["i"],
3201 flags: &[&FORCE],
3202 key: 100,
3203 ..Command::EMPTY
3204 };
3205 static STRICT: Command = Command {
3206 name: "ex",
3207 flags: &[&FORCE, &JOBS, &COLOR, &VERBOSE],
3208 args: &[&FILE, &REST],
3209 subcommands: &[&STRICT_INSTALL],
3210 unknown_flags: Some(UnknownFlags::Error),
3211 ..Command::EMPTY
3212 };
3213 static ROOT: Command = Command {
3214 name: "ex",
3215 flags: &[&FORCE, &JOBS, &COLOR, &VERBOSE],
3216 args: &[&FILE, &REST],
3217 subcommands: &[&INSTALL],
3218 ..Command::EMPTY
3219 };
3220 static ARGUMENT_CONFLICT: Command = Command {
3221 name: "ex",
3222 flags: &[&FORCE],
3223 subcommands: &[&INSTALL],
3224 args_conflicts_with_subcommands: true,
3225 ..Command::EMPTY
3226 };
3227
3228 // A CLI shaped exactly like mise's root: a default subcommand, a positional of its own,
3229 // and a subcommand under the default — which is the arrangement that tells routing from
3230 // a plain positional.
3231 static TASK: Arg = Arg {
3232 key: 20,
3233 name: "task",
3234 ..Arg::REQUIRED
3235 };
3236 static RUN_TASK: Arg = Arg {
3237 key: 21,
3238 name: "run_task",
3239 ..Arg::REQUIRED
3240 };
3241 static DEEP: Command = Command {
3242 name: "deep",
3243 args: &[&RUN_TASK],
3244 key: 203,
3245 ..Command::EMPTY
3246 };
3247 static LINT: Command = Command {
3248 name: "lint",
3249 subcommands: &[&DEEP],
3250 // A default of its own, so that a parse which forgot it had already taken one would
3251 // have somewhere to go. Nothing else in these fixtures can show the latch working.
3252 default_subcommand: Some(&DEEP),
3253 key: 202,
3254 ..Command::EMPTY
3255 };
3256 static RUN: Command = Command {
3257 name: "run",
3258 args: &[&RUN_TASK],
3259 subcommands: &[&LINT],
3260 key: 200,
3261 ..Command::EMPTY
3262 };
3263 static DEFAULTING: Command = Command {
3264 name: "mise",
3265 flags: &[&VERBOSE],
3266 args: &[&TASK],
3267 subcommands: &[&RUN, &INSTALL],
3268 default_subcommand: Some(find_subcommand(&[&RUN, &INSTALL], "run")),
3269 ..Command::EMPTY
3270 };
3271
3272 /// Collect every event, or the first error.
3273 fn parse<'t: 'v, 'v>(
3274 root: &'t Command<'t>,
3275 argv: &'v [&'v OsStr],
3276 ) -> Result<Vec<Event<'t, 'v, 'v>>, Error<'t, 'v>> {
3277 let mut parser = Parser::new(root, argv);
3278 let mut events = Vec::new();
3279 while let Some(event) = parser.next_event() {
3280 events.push(event?);
3281 }
3282 Ok(events)
3283 }
3284
3285 fn argv<const N: usize>(tokens: [&str; N]) -> [&OsStr; N] {
3286 tokens.map(OsStr::new)
3287 }
3288
3289 #[test]
3290 fn long_boolean() {
3291 let a = argv(["--force"]);
3292 assert_eq!(
3293 parse(&ROOT, &a).unwrap(),
3294 vec![Event::Flag {
3295 flag: &FORCE,
3296 value: None,
3297 negated: false
3298 }]
3299 );
3300 }
3301
3302 #[test]
3303 fn long_boolean_accepts_only_opted_in_attached_values() {
3304 for (token, negated, value) in [
3305 ("--color=false", false, b"false".as_slice()),
3306 ("--color=true", false, b"true".as_slice()),
3307 ("--no-color=false", true, b"false".as_slice()),
3308 ] {
3309 let a = argv([token]);
3310 assert_eq!(
3311 parse(&EXPLICIT_BOOL_ROOT, &a).unwrap(),
3312 vec![Event::Flag {
3313 flag: &EXPLICIT_BOOL,
3314 value: Some(value),
3315 negated,
3316 }]
3317 );
3318 }
3319
3320 let a = argv(["--color=maybe"]);
3321 assert!(matches!(
3322 parse(&EXPLICIT_BOOL_ROOT, &a),
3323 Err(Error::InvalidChoice { name: "color", .. })
3324 ));
3325
3326 let a = argv(["--force=false"]);
3327 assert_eq!(
3328 parse(&ROOT, &a).unwrap(),
3329 vec![Event::Flag {
3330 flag: &FORCE,
3331 value: None,
3332 negated: false,
3333 }]
3334 );
3335 }
3336
3337 #[test]
3338 fn long_value_forms() {
3339 for tokens in [vec!["--jobs=8"], vec!["--jobs", "8"]] {
3340 let a: Vec<&OsStr> = tokens.iter().map(|t| OsStr::new(*t)).collect();
3341 assert_eq!(
3342 parse(&ROOT, &a).unwrap(),
3343 vec![Event::Flag {
3344 flag: &JOBS,
3345 value: Some(b"8"),
3346 negated: false
3347 }],
3348 "{tokens:?}"
3349 );
3350 }
3351 }
3352
3353 #[test]
3354 fn long_value_keeps_later_equals() {
3355 let a = argv(["--jobs=a=b"]);
3356 let Event::Flag { value, .. } = parse(&ROOT, &a).unwrap()[0] else {
3357 panic!("expected a flag");
3358 };
3359 assert_eq!(value, Some(&b"a=b"[..]));
3360 }
3361
3362 #[test]
3363 fn long_value_attached_empty_is_empty_not_absent() {
3364 let a = argv(["--jobs="]);
3365 let Event::Flag { value, .. } = parse(&ROOT, &a).unwrap()[0] else {
3366 panic!("expected a flag");
3367 };
3368 assert_eq!(value, Some(&b""[..]));
3369 }
3370
3371 #[test]
3372 fn long_value_refuses_flaglike_next_word() {
3373 let a = argv(["--jobs", "--force"]);
3374 assert_eq!(
3375 parse(&ROOT, &a),
3376 Err(Error::MissingFlagValue { flag: &JOBS })
3377 );
3378 }
3379
3380 #[test]
3381 fn long_value_accepts_negative_number() {
3382 let a = argv(["--jobs", "-1"]);
3383 let Event::Flag { value, .. } = parse(&ROOT, &a).unwrap()[0] else {
3384 panic!("expected a flag");
3385 };
3386 assert_eq!(value, Some(&b"-1"[..]));
3387 }
3388
3389 #[test]
3390 fn missing_optional_positional_reserves_the_last_word() {
3391 static OPTIONAL: Arg = Arg {
3392 key: 90,
3393 name: "optional",
3394 required: false,
3395 ..Arg::REQUIRED
3396 };
3397 static REQUIRED: Arg = Arg {
3398 key: 91,
3399 name: "required",
3400 ..Arg::REQUIRED
3401 };
3402 static CMD: Command = Command {
3403 name: "ex",
3404 args: &[&OPTIONAL, &REQUIRED],
3405 allow_missing_positional: true,
3406 ..Command::EMPTY
3407 };
3408
3409 let one = argv(["value"]);
3410 assert_eq!(
3411 parse(&CMD, &one).unwrap(),
3412 vec![Event::Arg {
3413 arg: &REQUIRED,
3414 value: b"value",
3415 delimit: true
3416 }]
3417 );
3418 let two = argv(["optional", "required"]);
3419 assert_eq!(
3420 parse(&CMD, &two).unwrap(),
3421 vec![
3422 Event::Arg {
3423 arg: &OPTIONAL,
3424 value: b"optional",
3425 delimit: true
3426 },
3427 Event::Arg {
3428 arg: &REQUIRED,
3429 value: b"required",
3430 delimit: true
3431 },
3432 ]
3433 );
3434 }
3435
3436 #[test]
3437 fn negative_numbers_are_narrowly_opted_in() {
3438 static PLAIN: Flag = Flag {
3439 key: 90,
3440 name: "plain",
3441 longs: &["plain"],
3442 ..Flag::VALUE
3443 };
3444 static VALUE: Arg = Arg {
3445 key: 91,
3446 name: "value",
3447 ..Arg::REQUIRED
3448 };
3449 static CMD: Command = Command {
3450 name: "ex",
3451 flags: &[&PLAIN],
3452 args: &[&VALUE],
3453 unknown_flags: Some(UnknownFlags::Error),
3454 ..Command::EMPTY
3455 };
3456
3457 let flag = argv(["--plain", "-1"]);
3458 assert_eq!(
3459 parse(&CMD, &flag),
3460 Err(Error::MissingFlagValue { flag: &PLAIN })
3461 );
3462 let positional = argv(["-1"]);
3463 assert_eq!(
3464 parse(&CMD, &positional),
3465 Err(Error::UnknownFlag { token: b"-1" })
3466 );
3467 }
3468
3469 #[test]
3470 fn an_exact_declared_digit_short_outranks_a_negative_number() {
3471 static PRINT0: Flag = Flag {
3472 key: 92,
3473 name: "print0",
3474 shorts: b"0",
3475 ..Flag::BOOL
3476 };
3477 static VALUE: Arg = Arg {
3478 key: 93,
3479 name: "value",
3480 required: false,
3481 allow_negative_numbers: true,
3482 ..Arg::REQUIRED
3483 };
3484 static CMD: Command = Command {
3485 name: "fd",
3486 flags: &[&PRINT0],
3487 args: &[&VALUE],
3488 unknown_flags: Some(UnknownFlags::Error),
3489 ..Command::EMPTY
3490 };
3491
3492 assert_eq!(
3493 parse(&CMD, &argv(["-0"])),
3494 Ok(vec![Event::Flag {
3495 flag: &PRINT0,
3496 value: None,
3497 negated: false,
3498 }])
3499 );
3500 assert!(matches!(
3501 parse(&CMD, &argv(["-1"])),
3502 Ok(events) if matches!(events.as_slice(), [Event::Arg { value: b"-1", .. }])
3503 ));
3504 }
3505
3506 #[test]
3507 fn negation_of_value_flag_does_not_consume_a_value() {
3508 static MODE: Flag = Flag {
3509 key: 9,
3510 name: "mode",
3511 longs: &["mode"],
3512 negate: Some("no-mode"),
3513 ..Flag::VALUE
3514 };
3515 static NEGATED_VALUE: Command = Command {
3516 name: "ex",
3517 flags: &[&MODE],
3518 args: &[&FILE],
3519 ..Command::EMPTY
3520 };
3521
3522 let a = argv(["--no-mode", "input"]);
3523 assert_eq!(
3524 parse(&NEGATED_VALUE, &a).unwrap(),
3525 vec![
3526 Event::Flag {
3527 flag: &MODE,
3528 value: None,
3529 negated: true
3530 },
3531 Event::Arg {
3532 arg: &FILE,
3533 value: b"input",
3534 delimit: true,
3535 }
3536 ]
3537 );
3538 }
3539
3540 #[test]
3541 fn no_abbreviation() {
3542 // A prefix names no flag, so by default it is a value like any other word.
3543 let a = argv(["--forc"]);
3544 assert_eq!(
3545 parse(&ROOT, &a).unwrap(),
3546 vec![Event::Arg {
3547 arg: &FILE,
3548 value: b"--forc",
3549 delimit: true,
3550 }]
3551 );
3552
3553 // And a CLI that owns its flags hears about it, which is the whole reason
3554 // the strict mode exists.
3555 assert!(matches!(
3556 parse(&STRICT, &a),
3557 Err(Error::UnknownFlag { token: b"--forc" })
3558 ));
3559 }
3560
3561 #[test]
3562 fn an_unknown_flag_is_a_value_by_default() {
3563 // The default, and the case it is for: a command line being forwarded to
3564 // something whose flags this spec does not know.
3565 let a = argv(["--wat", "keep"]);
3566 assert_eq!(
3567 parse(&ROOT, &a).unwrap(),
3568 vec![
3569 Event::Arg {
3570 arg: &FILE,
3571 value: b"--wat",
3572 delimit: true,
3573 },
3574 Event::Arg {
3575 arg: &REST,
3576 value: b"keep",
3577 delimit: true,
3578 },
3579 ]
3580 );
3581
3582 // With nowhere to put it, it is an unexpected argument — the same error an
3583 // extra word gets, rather than a special one about flags.
3584 static ONE: Command = Command {
3585 name: "ex",
3586 args: &[&FILE],
3587 ..Command::EMPTY
3588 };
3589 let a = argv(["a", "--wat"]);
3590 assert_eq!(
3591 parse(&ONE, &a),
3592 Err(Error::UnexpectedArg { token: b"--wat" })
3593 );
3594 }
3595
3596 #[test]
3597 fn negation() {
3598 let a = argv(["--no-color"]);
3599 assert_eq!(
3600 parse(&ROOT, &a).unwrap(),
3601 vec![Event::Flag {
3602 flag: &COLOR,
3603 value: None,
3604 negated: true
3605 }]
3606 );
3607 }
3608
3609 #[test]
3610 fn short_bundle_and_attached_value() {
3611 let a = argv(["-fj8"]);
3612 assert_eq!(
3613 parse(&ROOT, &a).unwrap(),
3614 vec![
3615 Event::Flag {
3616 flag: &FORCE,
3617 value: None,
3618 negated: false
3619 },
3620 Event::Flag {
3621 flag: &JOBS,
3622 value: Some(b"8"),
3623 negated: false
3624 },
3625 ]
3626 );
3627 }
3628
3629 #[test]
3630 fn short_value_strips_one_equals() {
3631 for (tokens, want) in [(["-j=8"], &b"8"[..]), (["-j==8"], &b"=8"[..])] {
3632 let a = argv(tokens);
3633 let Event::Flag { value, .. } = parse(&ROOT, &a).unwrap()[0] else {
3634 panic!("expected a flag");
3635 };
3636 assert_eq!(value, Some(want), "{tokens:?}");
3637 }
3638 }
3639
3640 #[test]
3641 fn bare_dash_is_a_value() {
3642 let a = argv(["-"]);
3643 assert_eq!(
3644 parse(&ROOT, &a).unwrap(),
3645 vec![Event::Arg {
3646 arg: &FILE,
3647 value: b"-",
3648 delimit: true,
3649 }]
3650 );
3651 }
3652
3653 #[test]
3654 fn positionals_then_variadic() {
3655 let a = argv(["one", "two", "three"]);
3656 assert_eq!(
3657 parse(&ROOT, &a).unwrap(),
3658 vec![
3659 Event::Arg {
3660 arg: &FILE,
3661 value: b"one",
3662 delimit: true,
3663 },
3664 Event::Arg {
3665 arg: &REST,
3666 value: b"two",
3667 delimit: true,
3668 },
3669 Event::Arg {
3670 arg: &REST,
3671 value: b"three",
3672 delimit: true,
3673 },
3674 ]
3675 );
3676 }
3677
3678 #[test]
3679 fn subcommand_and_alias_route_the_same() {
3680 for token in ["install", "i"] {
3681 let a = argv([token]);
3682 assert_eq!(
3683 parse(&ROOT, &a).unwrap(),
3684 vec![Event::Command(&INSTALL)],
3685 "{token}"
3686 );
3687 }
3688 }
3689
3690 #[test]
3691 fn a_parent_argument_can_exclude_a_later_subcommand() {
3692 let a = argv(["--force", "install"]);
3693 assert!(matches!(
3694 parse(&ARGUMENT_CONFLICT, &a),
3695 Err(Error::SubcommandConflict { subcommand }) if subcommand.name == "install"
3696 ));
3697 }
3698
3699 #[test]
3700 fn subcommand_only_routes_before_a_positional_is_filled() {
3701 let a = argv(["other", "install"]);
3702 assert_eq!(
3703 parse(&ROOT, &a).unwrap(),
3704 vec![
3705 Event::Arg {
3706 arg: &FILE,
3707 value: b"other",
3708 delimit: true,
3709 },
3710 Event::Arg {
3711 arg: &REST,
3712 value: b"install",
3713 delimit: true,
3714 },
3715 ]
3716 );
3717 }
3718
3719 #[test]
3720 fn a_word_naming_no_subcommand_goes_to_the_default_one() {
3721 // usage-lib's answer, which this reproduces: `mise build` comes back as commands
3722 // `["mise", "run"]` with the word bound to *run's* argument — not to `mise`'s own
3723 // `[TASK]`, which is what makes this more than a synonym for a positional.
3724 let a = argv(["build"]);
3725 assert_eq!(
3726 parse(&DEFAULTING, &a).unwrap(),
3727 vec![
3728 Event::Command(&RUN),
3729 Event::Arg {
3730 arg: &RUN_TASK,
3731 value: b"build",
3732 delimit: true,
3733 },
3734 ]
3735 );
3736 }
3737
3738 // A shared table, as a flattened struct's would be. Declared outside the tests so both
3739 // can splice it, which is the arrangement it exists to model.
3740 static SHARED_QUIET: Flag = Flag {
3741 key: 300,
3742 name: "quiet",
3743 longs: &["quiet"],
3744 ..Flag::BOOL
3745 };
3746 static SHARED_FLAGS: &[&Flag] = &[&SHARED_QUIET];
3747 static SHARED_WHAT: Arg = Arg {
3748 key: 301,
3749 name: "what",
3750 ..Arg::REQUIRED
3751 };
3752 static SHARED_ARGS: &[&Arg] = &[&SHARED_WHAT];
3753
3754 #[test]
3755 fn concatenating_tables_keeps_the_order_they_were_given_in() {
3756 // The property positional arguments depend on: a flattened group lands where the
3757 // field was written, not at the end. `[&FILE], SHARED, [&REST]` has to stay in that
3758 // order or `ex a b c` binds the wrong words.
3759 const ARGS: &[&[&Arg]] = &[&[&FILE], SHARED_ARGS, &[&REST]];
3760 static TABLE: [&Arg; table_len(ARGS)] = concat_args(ARGS);
3761 assert_eq!(
3762 TABLE.iter().map(|a| a.name).collect::<Vec<_>>(),
3763 ["file", "what", "rest"]
3764 );
3765
3766 // Empty groups contribute nothing and disturb nothing, which is what lets the derive
3767 // emit a group per field without checking whether it is empty first.
3768 const WITH_GAPS: &[&[&Flag]] = &[&[], &[&FORCE], &[], SHARED_FLAGS, &[]];
3769 static FLAGS: [&Flag; table_len(WITH_GAPS)] = concat_flags(WITH_GAPS);
3770 // By long form: these fixtures do not all set `name`, and the placeholder's is also
3771 // empty — so comparing names could not tell a real entry from a leftover slot.
3772 assert_eq!(
3773 FLAGS.iter().map(|f| f.longs).collect::<Vec<_>>(),
3774 [&["force"], &["quiet"]]
3775 );
3776 }
3777
3778 #[test]
3779 fn a_concatenated_table_parses_like_a_declared_one() {
3780 // The point of doing this at compile time: what the parser walks is one flat slice,
3781 // indistinguishable from a command that declared everything itself.
3782 const FLAG_GROUPS: &[&[&Flag]] = &[&[&FORCE], SHARED_FLAGS];
3783 const ARG_GROUPS: &[&[&Arg]] = &[SHARED_ARGS, &[&REST]];
3784 static FLAGS: [&Flag; table_len(FLAG_GROUPS)] = concat_flags(FLAG_GROUPS);
3785 static ARGS: [&Arg; table_len(ARG_GROUPS)] = concat_args(ARG_GROUPS);
3786 static JOINED: Command = Command {
3787 name: "joined",
3788 flags: &FLAGS,
3789 args: &ARGS,
3790 ..Command::EMPTY
3791 };
3792
3793 let a = argv(["--quiet", "one", "two", "--force"]);
3794 assert_eq!(
3795 parse(&JOINED, &a).unwrap(),
3796 vec![
3797 Event::Flag {
3798 flag: &SHARED_QUIET,
3799 value: None,
3800 negated: false
3801 },
3802 Event::Arg {
3803 arg: &SHARED_WHAT,
3804 value: b"one",
3805 delimit: true,
3806 },
3807 Event::Arg {
3808 arg: &REST,
3809 value: b"two",
3810 delimit: true,
3811 },
3812 Event::Flag {
3813 flag: &FORCE,
3814 value: None,
3815 negated: false
3816 },
3817 ]
3818 );
3819 }
3820
3821 #[test]
3822 fn an_unknown_flag_is_not_routed() {
3823 // A dash-prefixed token that names no flag becomes a value here (the default for
3824 // `unknown_flags`), and it must not thereby become a *subcommand* word: usage-lib
3825 // stops looking for subcommands at an unrecognised flag, and binds it to the command
3826 // still in scope. Verified against usage-lib, where `ex --wat` comes back as commands
3827 // `["ex"]` with `ROOT_TASK = "--wat"`.
3828 for token in ["--wat", "-x"] {
3829 let a = argv([token]);
3830 assert_eq!(
3831 parse(&DEFAULTING, &a).unwrap(),
3832 vec![Event::Arg {
3833 arg: &TASK,
3834 value: token.as_bytes(),
3835 delimit: true,
3836 }],
3837 "{token} should bind where it was typed, not in the default subcommand"
3838 );
3839 }
3840 }
3841
3842 #[test]
3843 fn a_named_subcommand_is_not_routed() {
3844 // The default is for words that name nothing. A word that names a sibling still
3845 // selects it, and the root's own argument is still reachable behind one.
3846 let a = argv(["install"]);
3847 assert_eq!(
3848 parse(&DEFAULTING, &a).unwrap(),
3849 vec![Event::Command(&INSTALL)]
3850 );
3851 }
3852
3853 #[test]
3854 fn an_unmatched_word_is_forwarded_when_external_subcommand_is_set() {
3855 static CATCH: Command = Command {
3856 name: "ex",
3857 flags: &[&VERBOSE],
3858 subcommands: &[&INSTALL],
3859 external_subcommand: true,
3860 unknown_flags: Some(UnknownFlags::Error),
3861 ..Command::EMPTY
3862 };
3863 let a = argv(["foo", "--help", "bar"]);
3864 assert_eq!(
3865 parse(&CATCH, &a).unwrap(),
3866 vec![Event::External { values: &a[..] }]
3867 );
3868
3869 let a = argv(["install"]);
3870 assert_eq!(parse(&CATCH, &a).unwrap(), vec![Event::Command(&INSTALL)]);
3871
3872 let a = argv(["--verbose", "foo", "--verbose"]);
3873 assert_eq!(
3874 parse(&CATCH, &a).unwrap(),
3875 vec![
3876 Event::Flag {
3877 flag: &VERBOSE,
3878 value: None,
3879 negated: false
3880 },
3881 Event::External { values: &a[1..] }
3882 ]
3883 );
3884
3885 let a = argv(["--wat"]);
3886 assert_eq!(
3887 parse(&CATCH, &a),
3888 Err(Error::UnknownFlag { token: b"--wat" })
3889 );
3890
3891 // A negative number is a value, not a flag, so it can be the unmatched word.
3892 let a = argv(["-1", "rest"]);
3893 assert_eq!(
3894 parse(&CATCH, &a).unwrap(),
3895 vec![Event::External { values: &a[..] }]
3896 );
3897 }
3898
3899 #[test]
3900 fn a_default_subcommand_outranks_an_external_one() {
3901 static CATCH_DEFAULT: Command = Command {
3902 name: "ex",
3903 subcommands: &[&RUN],
3904 default_subcommand: Some(&RUN),
3905 external_subcommand: true,
3906 ..Command::EMPTY
3907 };
3908 let a = argv(["build"]);
3909 assert_eq!(
3910 parse(&CATCH_DEFAULT, &a).unwrap(),
3911 vec![
3912 Event::Command(&RUN),
3913 Event::Arg {
3914 arg: &RUN_TASK,
3915 value: b"build",
3916 delimit: true,
3917 }
3918 ]
3919 );
3920 }
3921
3922 #[test]
3923 fn a_default_subcommand_starts_at_the_word_it_receives() {
3924 let a = argv(["build"]);
3925 let mut parser = Parser::new(&DEFAULTING, &a);
3926 assert_eq!(parser.next_event(), Some(Ok(Event::Command(&RUN))));
3927 assert_eq!(parser.command_start(), 0);
3928 assert_eq!(
3929 parser.next_event(),
3930 Some(Ok(Event::Arg {
3931 arg: &RUN_TASK,
3932 value: b"build",
3933 delimit: true,
3934 }))
3935 );
3936 }
3937
3938 #[test]
3939 fn the_default_can_be_named_by_an_alias() {
3940 // usage-lib resolves the name against subcommand names, aliases and hidden aliases
3941 // alike, so a spec may point `default_subcommand` at any of them.
3942 static BY_ALIAS: Command = Command {
3943 name: "mise",
3944 args: &[&TASK],
3945 subcommands: &[&INSTALL],
3946 // `INSTALL` answers to "i" as well as to its name.
3947 default_subcommand: Some(find_subcommand(&[&INSTALL], "i")),
3948 ..Command::EMPTY
3949 };
3950 assert!(::core::ptr::eq(
3951 BY_ALIAS.default_subcommand.expect("declared"),
3952 &INSTALL
3953 ));
3954 }
3955
3956 #[test]
3957 fn a_name_outranks_another_commands_alias() {
3958 // A spec `assert_unique_subcommand_names` would reject, resolved anyway: a parser
3959 // handed a table nothing validated still has to answer, and the answer is the
3960 // command whose own name it is. Both orders, because taking the first candidate
3961 // that matched on either name or alias made this depend on which was listed first
3962 // — and usage-lib, building a map, took the last.
3963 static ALPHA: Command = Command {
3964 name: "alpha",
3965 aliases: &["run"],
3966 key: 300,
3967 ..Command::EMPTY
3968 };
3969 static PLAIN_RUN: Command = Command {
3970 name: "run",
3971 key: 301,
3972 ..Command::EMPTY
3973 };
3974 for subcommands in [&[&ALPHA, &PLAIN_RUN] as &[&Command], &[&PLAIN_RUN, &ALPHA]] {
3975 assert!(::core::ptr::eq(
3976 find_subcommand(subcommands, "run"),
3977 &PLAIN_RUN
3978 ));
3979 let root: Command = Command {
3980 name: "ex",
3981 subcommands,
3982 ..Command::EMPTY
3983 };
3984 let a = argv(["run"]);
3985 assert_eq!(parse(&root, &a).unwrap(), vec![Event::Command(&PLAIN_RUN)]);
3986 // The alias still reaches its own command by every name it does not share.
3987 let a = argv(["alpha"]);
3988 assert_eq!(parse(&root, &a).unwrap(), vec![Event::Command(&ALPHA)]);
3989 // `ex help run` asks about the command `ex run` selects. These are separate
3990 // lookups — help resolves a path without descending — and answering differently
3991 // for a colliding word is the divergence this rule exists to end.
3992 let a = argv(["help", "run"]);
3993 match parse(&root, &a) {
3994 Err(Error::Help { cmd, .. }) => {
3995 assert!(
3996 ::core::ptr::eq(cmd, &PLAIN_RUN),
3997 "got help for {}",
3998 cmd.name
3999 )
4000 }
4001 other => panic!("expected a help request, got {other:?}"),
4002 }
4003 }
4004 }
4005
4006 #[test]
4007 #[should_panic(expected = "two subcommands answer to the same name")]
4008 fn an_alias_cannot_shadow_a_sibling_command() {
4009 static ADD: Command = Command {
4010 name: "add",
4011 aliases: &["install"],
4012 ..Command::EMPTY
4013 };
4014 assert_unique_subcommand_names(&[&INSTALL, &ADD]);
4015 }
4016
4017 #[test]
4018 fn the_word_is_re_examined_against_the_command_it_reached() {
4019 // The reason the cursor steps back rather than the token being consumed: `lint` names
4020 // nothing at the root, and once inside `run` it names a subcommand. mise's mounted
4021 // task names arrive exactly this way.
4022 let a = argv(["lint"]);
4023 assert_eq!(
4024 parse(&DEFAULTING, &a).unwrap(),
4025 vec![Event::Command(&RUN), Event::Command(&LINT)]
4026 );
4027 }
4028
4029 #[test]
4030 fn the_default_is_taken_at_most_once_per_parse() {
4031 // usage-lib latches this for the whole parse rather than per command, and the shape
4032 // that shows the difference needs two of them: `lint` routes through `run`, and `lint`
4033 // declares a default too. A second word there would descend again — walking a CLI
4034 // deeper than anything the user typed — so the answer is that it does not.
4035 let a = argv(["lint", "zzz"]);
4036 assert_eq!(
4037 parse(&DEFAULTING, &a),
4038 Err(Error::UnexpectedArg { token: b"zzz" }),
4039 "the second word must not reach `deep`"
4040 );
4041
4042 // Reached explicitly, the same command still takes it: the latch bounds routing, not
4043 // the tree.
4044 let a = argv(["lint", "deep", "zzz"]);
4045 assert_eq!(
4046 parse(&DEFAULTING, &a).unwrap(),
4047 vec![
4048 Event::Command(&RUN),
4049 Event::Command(&LINT),
4050 Event::Command(&DEEP),
4051 Event::Arg {
4052 arg: &RUN_TASK,
4053 value: b"zzz",
4054 delimit: true,
4055 },
4056 ]
4057 );
4058 }
4059
4060 #[test]
4061 fn a_flag_before_the_word_still_belongs_to_the_root() {
4062 // Routing happens at the word, so anything typed before it was addressed to the
4063 // command the user was actually at.
4064 let a = argv(["--verbose", "build"]);
4065 assert_eq!(
4066 parse(&DEFAULTING, &a).unwrap(),
4067 vec![
4068 Event::Flag {
4069 flag: &VERBOSE,
4070 value: None,
4071 negated: false
4072 },
4073 Event::Command(&RUN),
4074 Event::Arg {
4075 arg: &RUN_TASK,
4076 value: b"build",
4077 delimit: true,
4078 },
4079 ]
4080 );
4081 }
4082
4083 #[test]
4084 fn nothing_routes_after_the_separator() {
4085 // Past `--` there are no subcommands left to select, so there is no default to reach
4086 // either: the words are values of whatever the command declares.
4087 let a = argv(["--", "build"]);
4088 assert_eq!(
4089 parse(&DEFAULTING, &a).unwrap(),
4090 vec![Event::Arg {
4091 arg: &TASK,
4092 value: b"build",
4093 delimit: true,
4094 }]
4095 );
4096 }
4097
4098 #[test]
4099 fn globals_are_inherited_but_plain_flags_are_not() {
4100 let a = argv(["install", "--verbose"]);
4101 assert_eq!(
4102 parse(&ROOT, &a).unwrap(),
4103 vec![
4104 Event::Command(&INSTALL),
4105 Event::Flag {
4106 flag: &VERBOSE,
4107 value: None,
4108 negated: false
4109 }
4110 ]
4111 );
4112
4113 // `--jobs` belongs to the root and is not global, so it is not a flag here.
4114 // Strictly that is an unknown flag; leniently it is a word, and `install`
4115 // declares no argument to hold one — either way it is never read as the
4116 // root's flag, which is what this test is about.
4117 let a = argv(["install", "--jobs", "8"]);
4118 assert!(matches!(parse(&STRICT, &a), Err(Error::UnknownFlag { .. })));
4119 assert!(matches!(
4120 parse(&ROOT, &a),
4121 Err(Error::UnexpectedArg { token: b"--jobs" })
4122 ));
4123 }
4124
4125 #[test]
4126 fn double_dash_protects_flaglike_values() {
4127 let a = argv(["--", "--force", "-x"]);
4128 assert_eq!(
4129 parse(&ROOT, &a).unwrap(),
4130 vec![
4131 Event::Arg {
4132 arg: &FILE,
4133 value: b"--force",
4134 delimit: true,
4135 },
4136 Event::Arg {
4137 arg: &REST,
4138 value: b"-x",
4139 delimit: true,
4140 },
4141 ]
4142 );
4143 }
4144
4145 #[test]
4146 fn second_double_dash_is_a_value() {
4147 let a = argv(["--", "a", "--", "b"]);
4148 let values: Vec<&[u8]> = parse(&ROOT, &a)
4149 .unwrap()
4150 .iter()
4151 .filter_map(|e| match e {
4152 Event::Arg { value, .. } => Some(*value),
4153 _ => None,
4154 })
4155 .collect();
4156 assert_eq!(values, vec![&b"a"[..], &b"--"[..], &b"b"[..]]);
4157 }
4158
4159 #[test]
4160 fn allow_hyphen_values_takes_a_flaglike_detached_value() {
4161 static ARGS: Flag = Flag {
4162 key: 6,
4163 name: "args",
4164 longs: &["args"],
4165 shorts: b"a",
4166 takes_value: true,
4167 allow_hyphen_values: true,
4168 ..Flag::BOOL
4169 };
4170 static DIR: Flag = Flag {
4171 key: 7,
4172 name: "working-dir",
4173 longs: &["working-dir"],
4174 shorts: b"d",
4175 ..Flag::VALUE
4176 };
4177 static HYPHEN: Command = Command {
4178 name: "ex",
4179 flags: &[&ARGS, &DIR],
4180 args: &[&REST],
4181 ..Command::EMPTY
4182 };
4183
4184 let a = argv(["-a", "-destroy"]);
4185 assert_eq!(
4186 parse(&HYPHEN, &a).unwrap(),
4187 vec![Event::Flag {
4188 flag: &ARGS,
4189 value: Some(b"-destroy"),
4190 negated: false
4191 }]
4192 );
4193
4194 let a = argv(["--args", "--", "-x"]);
4195 assert_eq!(
4196 parse(&HYPHEN, &a).unwrap(),
4197 vec![
4198 Event::Flag {
4199 flag: &ARGS,
4200 value: Some(b"--"),
4201 negated: false
4202 },
4203 Event::Arg {
4204 arg: &REST,
4205 value: b"-x",
4206 delimit: true,
4207 },
4208 ]
4209 );
4210 }
4211
4212 #[test]
4213 fn require_equals_refuses_a_detached_value() {
4214 static INSPECT: Flag = Flag {
4215 key: 8,
4216 name: "inspect",
4217 longs: &["inspect"],
4218 shorts: b"i",
4219 takes_value: true,
4220 require_equals: true,
4221 ..Flag::BOOL
4222 };
4223 static EQ: Command = Command {
4224 name: "ex",
4225 flags: &[&INSPECT],
4226 ..Command::EMPTY
4227 };
4228
4229 let a = argv(["--inspect=9229"]);
4230 assert_eq!(
4231 parse(&EQ, &a).unwrap(),
4232 vec![Event::Flag {
4233 flag: &INSPECT,
4234 value: Some(b"9229"),
4235 negated: false
4236 }]
4237 );
4238
4239 let a = argv(["--inspect", "9229"]);
4240 assert!(matches!(
4241 parse(&EQ, &a),
4242 Err(Error::MissingFlagValue { .. })
4243 ));
4244
4245 let a = argv(["-i9229"]);
4246 assert_eq!(
4247 parse(&EQ, &a).unwrap(),
4248 vec![Event::Flag {
4249 flag: &INSPECT,
4250 value: Some(b"9229"),
4251 negated: false
4252 }]
4253 );
4254
4255 static ALL: Flag = Flag {
4256 key: 9,
4257 name: "all",
4258 longs: &["all"],
4259 shorts: b"a",
4260 ..Flag::BOOL
4261 };
4262 static BUNDLE: Command = Command {
4263 name: "ex",
4264 flags: &[&ALL, &INSPECT],
4265 ..Command::EMPTY
4266 };
4267 let a = argv(["-ai", "9229"]);
4268 assert!(
4269 matches!(parse(&BUNDLE, &a), Err(Error::MissingFlagValue { .. })),
4270 "a require_equals short reached through a bundle still refuses the following word"
4271 );
4272 }
4273
4274 #[test]
4275 fn default_missing_binds_when_the_value_is_left_off() {
4276 static COLOR: Flag = Flag {
4277 key: 9,
4278 name: "color",
4279 longs: &["color"],
4280 takes_value: true,
4281 default_missing: Some(b"always"),
4282 ..Flag::BOOL
4283 };
4284 static VERBOSE: Flag = Flag {
4285 key: 10,
4286 name: "verbose",
4287 longs: &["verbose"],
4288 ..Flag::BOOL
4289 };
4290 static MISSING: Command = Command {
4291 name: "ex",
4292 flags: &[&COLOR, &VERBOSE],
4293 ..Command::EMPTY
4294 };
4295
4296 let a = argv(["--color"]);
4297 assert_eq!(
4298 parse(&MISSING, &a).unwrap(),
4299 vec![Event::Flag {
4300 flag: &COLOR,
4301 value: Some(b"always"),
4302 negated: false
4303 }]
4304 );
4305
4306 let a = argv(["--color=never"]);
4307 assert_eq!(
4308 parse(&MISSING, &a).unwrap(),
4309 vec![Event::Flag {
4310 flag: &COLOR,
4311 value: Some(b"never"),
4312 negated: false
4313 }]
4314 );
4315
4316 let a = argv(["--color", "--verbose"]);
4317 assert_eq!(
4318 parse(&MISSING, &a).unwrap(),
4319 vec![
4320 Event::Flag {
4321 flag: &COLOR,
4322 value: Some(b"always"),
4323 negated: false
4324 },
4325 Event::Flag {
4326 flag: &VERBOSE,
4327 value: None,
4328 negated: false
4329 },
4330 ]
4331 );
4332
4333 let a = argv(["--color="]);
4334 assert_eq!(
4335 parse(&MISSING, &a).unwrap(),
4336 vec![Event::Flag {
4337 flag: &COLOR,
4338 value: Some(b""),
4339 negated: false
4340 }]
4341 );
4342 }
4343
4344 #[test]
4345 fn optional_flag_value_distinguishes_bare_and_explicit_forms() {
4346 static BUMP: Flag = Flag {
4347 key: 11,
4348 name: "bump",
4349 longs: &["bump"],
4350 takes_value: true,
4351 value_optional: true,
4352 ..Flag::BOOL
4353 };
4354 static OPTIONAL: Command = Command {
4355 name: "ex",
4356 flags: &[&BUMP],
4357 ..Command::EMPTY
4358 };
4359
4360 assert_eq!(parse(&OPTIONAL, &argv([])).unwrap(), vec![]);
4361 assert_eq!(
4362 parse(&OPTIONAL, &argv(["--bump"])).unwrap(),
4363 vec![Event::Flag {
4364 flag: &BUMP,
4365 value: None,
4366 negated: false,
4367 }]
4368 );
4369 assert_eq!(
4370 parse(&OPTIONAL, &argv(["--bump=5"])).unwrap(),
4371 vec![Event::Flag {
4372 flag: &BUMP,
4373 value: Some(b"5"),
4374 negated: false,
4375 }]
4376 );
4377
4378 static INCLUDE: Flag = Flag {
4379 key: 12,
4380 name: "include",
4381 longs: &["include"],
4382 takes_value: true,
4383 variadic: true,
4384 value_optional: true,
4385 ..Flag::BOOL
4386 };
4387 static VERBOSE: Flag = Flag {
4388 key: 13,
4389 name: "verbose",
4390 longs: &["verbose"],
4391 ..Flag::BOOL
4392 };
4393 static VARIADIC: Command = Command {
4394 name: "ex",
4395 flags: &[&INCLUDE, &VERBOSE],
4396 args: &[&REST],
4397 ..Command::EMPTY
4398 };
4399 assert_eq!(
4400 parse(&VARIADIC, &argv(["--include", "--verbose", "file"])).unwrap(),
4401 vec![
4402 Event::Flag {
4403 flag: &INCLUDE,
4404 value: None,
4405 negated: false,
4406 },
4407 Event::Flag {
4408 flag: &VERBOSE,
4409 value: None,
4410 negated: false,
4411 },
4412 Event::Arg {
4413 arg: &REST,
4414 value: b"file",
4415 delimit: true,
4416 },
4417 ]
4418 );
4419 }
4420
4421 #[test]
4422 fn default_missing_with_require_equals_leaves_the_following_word() {
4423 static INSPECT: Flag = Flag {
4424 key: 11,
4425 name: "inspect",
4426 longs: &["inspect"],
4427 takes_value: true,
4428 require_equals: true,
4429 default_missing: Some(b"9229"),
4430 ..Flag::BOOL
4431 };
4432 static BOTH: Command = Command {
4433 name: "ex",
4434 flags: &[&INSPECT],
4435 args: &[&REST],
4436 ..Command::EMPTY
4437 };
4438
4439 let a = argv(["--inspect"]);
4440 assert_eq!(
4441 parse(&BOTH, &a).unwrap(),
4442 vec![Event::Flag {
4443 flag: &INSPECT,
4444 value: Some(b"9229"),
4445 negated: false
4446 }]
4447 );
4448
4449 let a = argv(["--inspect", "80"]);
4450 assert_eq!(
4451 parse(&BOTH, &a).unwrap(),
4452 vec![
4453 Event::Flag {
4454 flag: &INSPECT,
4455 value: Some(b"9229"),
4456 negated: false
4457 },
4458 Event::Arg {
4459 arg: &REST,
4460 value: b"80",
4461 delimit: true,
4462 },
4463 ]
4464 );
4465
4466 let a = argv(["--inspect="]);
4467 assert_eq!(
4468 parse(&BOTH, &a).unwrap(),
4469 vec![Event::Flag {
4470 flag: &INSPECT,
4471 value: Some(b""),
4472 negated: false
4473 }]
4474 );
4475 }
4476
4477 #[test]
4478 fn variadic_flag_collects_until_a_flaglike_token() {
4479 static INCLUDE: Flag = Flag {
4480 key: 5,
4481 name: "include",
4482 longs: &["include"],
4483 shorts: b"i",
4484 takes_value: true,
4485 variadic: true,
4486 ..Flag::BOOL
4487 };
4488 static GREEDY: Command = Command {
4489 name: "ex",
4490 flags: &[&INCLUDE, &FORCE],
4491 args: &[&FILE],
4492 ..Command::EMPTY
4493 };
4494
4495 let a = argv(["--include", "x", "y", "--force"]);
4496 assert_eq!(
4497 parse(&GREEDY, &a).unwrap(),
4498 vec![
4499 Event::Flag {
4500 flag: &INCLUDE,
4501 value: Some(b"x"),
4502 negated: false
4503 },
4504 Event::Flag {
4505 flag: &INCLUDE,
4506 value: Some(b"y"),
4507 negated: false
4508 },
4509 Event::Flag {
4510 flag: &FORCE,
4511 value: None,
4512 negated: false
4513 },
4514 ]
4515 );
4516 }
4517
4518 #[test]
4519 fn value_terminators_end_variadic_owners_without_binding() {
4520 static INCLUDE: Flag = Flag {
4521 key: 92,
4522 name: "include",
4523 longs: &["include"],
4524 takes_value: true,
4525 variadic: true,
4526 value_terminator: Some(b";"),
4527 ..Flag::BOOL
4528 };
4529 static ITEMS: Arg = Arg {
4530 key: 93,
4531 name: "items",
4532 var: true,
4533 value_terminator: Some(b";"),
4534 ..Arg::REQUIRED
4535 };
4536 static AFTER: Arg = Arg {
4537 key: 94,
4538 name: "after",
4539 ..Arg::REQUIRED
4540 };
4541 static FLAG_CMD: Command = Command {
4542 name: "ex",
4543 flags: &[&INCLUDE],
4544 args: &[&AFTER],
4545 ..Command::EMPTY
4546 };
4547 static ARG_CMD: Command = Command {
4548 name: "ex",
4549 args: &[&ITEMS, &AFTER],
4550 ..Command::EMPTY
4551 };
4552
4553 let flag = argv(["--include", "a", ";", "tail"]);
4554 assert_eq!(
4555 parse(&FLAG_CMD, &flag).unwrap(),
4556 vec![
4557 Event::Flag {
4558 flag: &INCLUDE,
4559 value: Some(b"a"),
4560 negated: false,
4561 },
4562 Event::Arg {
4563 arg: &AFTER,
4564 value: b"tail",
4565 delimit: true,
4566 },
4567 ]
4568 );
4569
4570 let positional = argv(["a", ";", "tail"]);
4571 assert_eq!(
4572 parse(&ARG_CMD, &positional).unwrap(),
4573 vec![
4574 Event::Arg {
4575 arg: &ITEMS,
4576 value: b"a",
4577 delimit: true,
4578 },
4579 Event::Arg {
4580 arg: &AFTER,
4581 value: b"tail",
4582 delimit: true,
4583 },
4584 ]
4585 );
4586 }
4587
4588 #[test]
4589 fn a_non_variadic_flag_leaves_the_next_word_alone() {
4590 // The counterpart to the test above: a flag that takes one value must not
4591 // swallow the word after it, which would silently steal a positional.
4592 let a = argv(["--jobs", "8", "keep-me"]);
4593 assert_eq!(
4594 parse(&ROOT, &a).unwrap(),
4595 vec![
4596 Event::Flag {
4597 flag: &JOBS,
4598 value: Some(b"8"),
4599 negated: false
4600 },
4601 Event::Arg {
4602 arg: &FILE,
4603 value: b"keep-me",
4604 delimit: true,
4605 },
4606 ]
4607 );
4608 }
4609
4610 #[test]
4611 fn double_dash_seen_means_a_separator_was_typed() {
4612 static FILES: Arg = Arg {
4613 key: 23,
4614 name: "files",
4615 double_dash: DoubleDash::Automatic,
4616 ..Arg::VAR
4617 };
4618 static AUTO: Command = Command {
4619 name: "ex",
4620 flags: &[&FORCE],
4621 args: &[&FILES],
4622 ..Command::EMPTY
4623 };
4624
4625 let a = argv(["--", "x"]);
4626 let mut parser = Parser::new(&ROOT, &a);
4627 while parser.next_event().is_some() {}
4628 assert!(parser.double_dash_seen(), "a real separator was consumed");
4629
4630 // `automatic` stops flag interpretation without a separator being typed,
4631 // and reporting one would be a lie to any caller that forwards argv.
4632 let a = argv(["x", "--force"]);
4633 let mut parser = Parser::new(&AUTO, &a);
4634 while parser.next_event().is_some() {}
4635 assert!(
4636 !parser.double_dash_seen(),
4637 "automatic mode must not claim a separator was given"
4638 );
4639 }
4640
4641 #[test]
4642 fn a_wrapper_still_forwards_a_help_flag() {
4643 // Supplying `--help` must not take the two forwarding mechanisms away from a wrapper,
4644 // which is the one place a CLI means to hand the token on rather than answer it.
4645 static ARGS: Arg = Arg {
4646 key: 24,
4647 name: "args",
4648 ..Arg::VAR
4649 };
4650 static WRAP: Command = Command {
4651 name: "wrap",
4652 args: &[&ARGS],
4653 ..Command::EMPTY
4654 };
4655
4656 // A typed separator: everything after it is a value, `--help` included.
4657 let a = argv(["--", "--help", "-h"]);
4658 assert_eq!(
4659 parse(&WRAP, &a).unwrap(),
4660 vec![
4661 Event::Arg {
4662 arg: &ARGS,
4663 value: b"--help",
4664 delimit: true,
4665 },
4666 Event::Arg {
4667 arg: &ARGS,
4668 value: b"-h",
4669 delimit: true,
4670 },
4671 ]
4672 );
4673
4674 // And `automatic`, for the wrapper whose caller should not have to type one: the
4675 // first value stops flag interpretation, so the flags after it forward.
4676 static AUTO_ARGS: Arg = Arg {
4677 key: 25,
4678 name: "args",
4679 double_dash: DoubleDash::Automatic,
4680 ..Arg::VAR
4681 };
4682 static AUTO_WRAP: Command = Command {
4683 name: "wrap",
4684 args: &[&AUTO_ARGS],
4685 ..Command::EMPTY
4686 };
4687
4688 let a = argv(["node", "--help"]);
4689 assert_eq!(
4690 parse(&AUTO_WRAP, &a).unwrap(),
4691 vec![
4692 Event::Arg {
4693 arg: &AUTO_ARGS,
4694 value: b"node",
4695 delimit: true,
4696 },
4697 Event::Arg {
4698 arg: &AUTO_ARGS,
4699 value: b"--help",
4700 delimit: true,
4701 },
4702 ]
4703 );
4704
4705 // Before either takes effect, though, the wrapper's own help is what `--help` asks
4706 // for — `mise run --help` is a question about `run`, not a value for it.
4707 let a = argv(["--help"]);
4708 assert_eq!(
4709 parse(&AUTO_WRAP, &a).unwrap(),
4710 vec![Event::Flag {
4711 flag: &HELP_LONG,
4712 value: None,
4713 negated: false
4714 }]
4715 );
4716 }
4717
4718 #[test]
4719 fn double_dash_required_arg() {
4720 static CMD: Arg = Arg {
4721 key: 20,
4722 name: "cmd",
4723 double_dash: DoubleDash::Required,
4724 ..Arg::REQUIRED
4725 };
4726 static EXEC: Command = Command {
4727 name: "ex",
4728 args: &[&CMD],
4729 ..Command::EMPTY
4730 };
4731
4732 let a = argv(["--", "ls"]);
4733 assert_eq!(
4734 parse(&EXEC, &a).unwrap(),
4735 vec![Event::Arg {
4736 arg: &CMD,
4737 value: b"ls",
4738 delimit: true,
4739 }]
4740 );
4741
4742 let a = argv(["ls"]);
4743 assert_eq!(
4744 parse(&EXEC, &a),
4745 Err(Error::ArgRequiresDoubleDash { arg: &CMD })
4746 );
4747 }
4748
4749 #[test]
4750 fn double_dash_preserve_keeps_the_separator() {
4751 static ARGS: Arg = Arg {
4752 key: 21,
4753 name: "args",
4754 double_dash: DoubleDash::Preserve,
4755 ..Arg::VAR
4756 };
4757 static WRAP: Command = Command {
4758 name: "ex",
4759 args: &[&ARGS],
4760 ..Command::EMPTY
4761 };
4762
4763 let a = argv(["a", "--", "b"]);
4764 let values: Vec<&[u8]> = parse(&WRAP, &a)
4765 .unwrap()
4766 .iter()
4767 .filter_map(|e| match e {
4768 Event::Arg { value, .. } => Some(*value),
4769 _ => None,
4770 })
4771 .collect();
4772 assert_eq!(values, vec![&b"a"[..], &b"--"[..], &b"b"[..]]);
4773 }
4774
4775 #[test]
4776 fn double_dash_automatic_stops_flag_interpretation() {
4777 static FILES: Arg = Arg {
4778 key: 22,
4779 name: "files",
4780 double_dash: DoubleDash::Automatic,
4781 ..Arg::VAR
4782 };
4783 static AUTO: Command = Command {
4784 name: "ex",
4785 flags: &[&FORCE],
4786 args: &[&FILES],
4787 ..Command::EMPTY
4788 };
4789
4790 // The flag before the first value is still a flag; the one after it is a
4791 // value.
4792 let a = argv(["-f", "one", "--force"]);
4793 assert_eq!(
4794 parse(&AUTO, &a).unwrap(),
4795 vec![
4796 Event::Flag {
4797 flag: &FORCE,
4798 value: None,
4799 negated: false
4800 },
4801 Event::Arg {
4802 arg: &FILES,
4803 value: b"one",
4804 delimit: true,
4805 },
4806 Event::Arg {
4807 arg: &FILES,
4808 value: b"--force",
4809 delimit: true,
4810 },
4811 ]
4812 );
4813 }
4814
4815 #[test]
4816 fn too_many_words() {
4817 static ONE: Command = Command {
4818 name: "ex",
4819 args: &[&FILE],
4820 ..Command::EMPTY
4821 };
4822 let a = argv(["a", "b"]);
4823 assert_eq!(parse(&ONE, &a), Err(Error::UnexpectedArg { token: b"b" }));
4824 }
4825
4826 #[test]
4827 fn unknown_letter_rejects_the_whole_bundle() {
4828 // `-f` is real and `-z` is not. The first event must be the error: if the
4829 // flag event came out first, a caller would have applied `-f` from a
4830 // command line that was rejected.
4831 let a = argv(["-fz"]);
4832 let mut parser = Parser::new(&STRICT, &a);
4833 assert_eq!(
4834 parser.next_event(),
4835 Some(Err(Error::UnknownFlag { token: b"-fz" })),
4836 "an unknown letter must reject the token before any of it is applied"
4837 );
4838 assert!(parser.next_event().is_none());
4839
4840 // Leniently, the same token is a value — and `-f` is *not* applied, since
4841 // the token was never a bundle at all.
4842 let a = argv(["-fz"]);
4843 assert_eq!(
4844 parse(&ROOT, &a).unwrap(),
4845 vec![Event::Arg {
4846 arg: &FILE,
4847 value: b"-fz",
4848 delimit: true,
4849 }]
4850 );
4851 }
4852
4853 #[test]
4854 fn unknown_short_error_names_the_whole_token() {
4855 for (tokens, want) in [(["-z"], &b"-z"[..]), (["-fz"], &b"-fz"[..])] {
4856 let a = argv(tokens);
4857 assert_eq!(
4858 parse(&STRICT, &a),
4859 Err(Error::UnknownFlag { token: want }),
4860 "{tokens:?}"
4861 );
4862 }
4863 }
4864
4865 #[test]
4866 fn errors_are_terminal() {
4867 let a = argv(["--wat", "--force"]);
4868 let mut parser = Parser::new(&STRICT, &a);
4869 assert!(parser.next_event().unwrap().is_err());
4870 assert!(parser.next_event().is_none());
4871 }
4872
4873 #[test]
4874 fn non_utf8_values_still_parse() {
4875 // A value that is not valid UTF-8 binds; only converting it fails, and
4876 // only if a caller asks.
4877 let raw = OsStr::new("--force");
4878 let a = [raw];
4879 assert!(parse(&ROOT, &a).is_ok());
4880
4881 assert!(as_str(b"ok").is_ok());
4882 assert!(as_str(&[0xff, 0xfe]).is_err());
4883 }
4884
4885 #[test]
4886 fn a_multicall_applet_is_the_basename_unless_it_is_the_dispatcher() {
4887 assert_eq!(multicall_basename("/usr/bin/ls"), "ls");
4888 assert_eq!(multicall_basename(r"C:\busybox\ls.exe"), "ls");
4889 assert_eq!(
4890 multicall_applet("/usr/bin/ls", "busybox", Some("busybox")),
4891 Some("ls")
4892 );
4893 assert_eq!(
4894 multicall_applet("/usr/bin/busybox", "busybox", Some("busybox")),
4895 None
4896 );
4897 assert_eq!(
4898 multicall_applet("ls.exe", "busybox", Some("busybox")),
4899 Some("ls")
4900 );
4901 assert_eq!(
4902 multicall_applet("/usr/bin/busybox", "BusyBox", Some("/opt/bin/busybox")),
4903 None
4904 );
4905 assert_eq!(
4906 multicall_applet("busybox.exe", "BusyBox", Some("busybox.exe")),
4907 None
4908 );
4909 }
4910
4911 #[test]
4912 fn a_spec_request_is_the_first_word_and_nothing_else() {
4913 let request = [OsStr::new(SPEC_REQUEST)];
4914 assert!(is_spec_request(&ROOT, &request));
4915
4916 // Anywhere but the front it is an ordinary value, which is what makes the endpoint
4917 // safe for a CLI whose arguments are arbitrary text.
4918 let later = ["install", SPEC_REQUEST].map(OsStr::new);
4919 assert!(!is_spec_request(&ROOT, &later));
4920 assert!(!is_spec_request(&ROOT, &[]));
4921 assert!(!is_spec_request(&ROOT, &[OsStr::new("--help")]));
4922 }
4923
4924 #[test]
4925 fn a_declared_command_of_that_name_keeps_it() {
4926 static DECLARED: Command = Command {
4927 name: SPEC_REQUEST,
4928 key: 200,
4929 ..Command::EMPTY
4930 };
4931 static ALIASED: Command = Command {
4932 name: "describe",
4933 aliases: &[SPEC_REQUEST],
4934 key: 201,
4935 ..Command::EMPTY
4936 };
4937 static DECLARES_IT: Command = Command {
4938 name: "ex",
4939 subcommands: &[&DECLARED],
4940 ..Command::EMPTY
4941 };
4942 static ALIASES_IT: Command = Command {
4943 name: "ex",
4944 subcommands: &[&ALIASED],
4945 ..Command::EMPTY
4946 };
4947
4948 let request = [OsStr::new(SPEC_REQUEST)];
4949 assert!(!is_spec_request(&DECLARES_IT, &request));
4950 // An alias selects a command just as its name does, so it wins here too.
4951 assert!(!is_spec_request(&ALIASES_IT, &request));
4952 }
4953}