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