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