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// One module opts out: `tty`, which asks the kernel how wide the terminal is. `forbid` would
88// make that opt-out impossible, and the alternative to the opt-out is a dependency — this
89// crate has none, and a CLI that wraps its help at 80 columns on a wide terminal is worse
90// than two dozen lines of reviewed FFI. Everything else in the crate is still refused.
91#![deny(unsafe_code)]
92
93/// Terminate at the compiled CLI entry-point boundary.
94///
95/// Kept in the runtime rather than expanded into an adopter crate so a project that
96/// forbids direct `std::process::exit` calls does not attribute the derive's process
97/// boundary to application code. `Cli::parse_from*` continues to return errors.
98#[doc(hidden)]
99#[allow(clippy::disallowed_methods)]
100pub fn __usage_process_exit(status: i32) -> ! {
101 std::process::exit(status)
102}
103
104use std::ffi::{OsStr, OsString};
105
106/// A value's shell-native completion class for `#[usage(value_hint = ...)]`.
107///
108/// This lives in the runtime crate so a declaration never needs clap merely to describe what
109/// kind of path a shell should offer. It is metadata only and adds no work to a successful
110/// parse.
111#[derive(Clone, Copy, Debug, Eq, PartialEq)]
112pub enum ValueHint {
113 /// Let the shell use its normal fallback behavior.
114 Unknown,
115 /// No structured hint applies; suppress the shell's path fallback.
116 Other,
117 /// A path to a file.
118 FilePath,
119 /// A path to either a file or a directory.
120 AnyPath,
121 /// A path to a directory.
122 DirPath,
123 /// A path to an executable file.
124 ExecutablePath,
125 /// A command name, resolved through the shell's command table and `PATH`.
126 CommandName,
127 /// One string containing a command and any arguments.
128 CommandString,
129 /// A trailing argv vector: complete the first value as a command, then its arguments.
130 CommandWithArguments,
131 /// A local operating-system user name.
132 Username,
133 /// A host name known to the shell or operating system.
134 Hostname,
135 /// A web address. This suppresses path fallback but offers no finite candidate set.
136 Url,
137 /// An email address. This suppresses path fallback but offers no finite candidate set.
138 EmailAddress,
139}
140
141#[cfg(feature = "complete")]
142pub mod complete;
143#[cfg(feature = "diagnostics")]
144pub mod diagnostic;
145#[cfg(feature = "spec")]
146pub mod embedded;
147#[cfg(feature = "complete")]
148pub mod install;
149#[cfg(feature = "complete")]
150pub mod script;
151
152/// Checks that the `complete` feature is on, with an explanation when it is not.
153///
154/// `#[usage(completion)]` generates code that reaches into [`complete`], which is behind a
155/// feature the *depending* crate enables — a derive cannot turn on a feature of another crate.
156/// Without this, the failure was `unresolved module complete`, which says nothing about the
157/// attribute that caused it.
158#[cfg(feature = "complete")]
159#[macro_export]
160macro_rules! __usage_needs_complete_feature {
161 () => {};
162}
163
164/// See [`__usage_needs_complete_feature`].
165#[cfg(not(feature = "complete"))]
166#[macro_export]
167macro_rules! __usage_needs_complete_feature {
168 () => {
169 ::core::compile_error!(
170 "`#[usage(completion)]` needs usage-argv's `complete` feature. Add it where \
171 usage-argv is depended on: usage-argv = { version = \"…\", features = \
172 [\"spec\", \"complete\"] }"
173 );
174 };
175}
176#[cfg(feature = "spec")]
177pub mod help;
178#[cfg(feature = "spec")]
179mod order;
180// Behind no feature: two traits and no code, so there is nothing here for a binary that
181// does not dispatch to pay for, and a hand-written CLI on the bare runtime can use them.
182pub mod run;
183// Behind no feature either, and public: an application that lays out its own output beside a
184// help page — a table, a progress line — should not have to work the width out a second way
185// and disagree with the page it sits next to.
186#[cfg(feature = "spec")]
187pub mod spec;
188pub mod tty;
189#[cfg(feature = "spec")]
190pub use spec::{parse_args_from, parse_args_from_argv};
191#[cfg(feature = "spec")]
192pub mod warn;
193
194pub use run::{Run, RunAsync, RunAsyncWith, RunWith};
195
196/// How deep a command tree this parser will descend.
197///
198/// The ancestor chain is kept in a fixed-size array so that a parse allocates
199/// nothing; this is that array's size. mise, the largest usage CLI, is four
200/// levels deep.
201pub const MAX_DEPTH: usize = 16;
202
203/// A command: its flags, its positional arguments, and its subcommands.
204///
205/// Every field is a borrowed slice so that a derive can emit the whole tree as
206/// `static` data. Use `..Command::EMPTY` to fill in the parts you do not need.
207#[derive(Debug, Clone, Copy, PartialEq, Eq)]
208pub struct Command<'a> {
209 /// The canonical name, used to select this command.
210 pub name: &'a str,
211 /// Alternative names that also select it.
212 pub aliases: &'a [&'a str],
213 pub flags: &'a [&'a Flag<'a>],
214 /// Positional arguments, in the order they are filled.
215 pub args: &'a [&'a Arg<'a>],
216 /// A repeatable group of scoped flags and positional arguments, if this command has one.
217 ///
218 /// Behind a reference because hardly any command has one: held inline, the clause's five
219 /// slices were two thirds of every command's table whether it declared one or not.
220 pub clause: ::core::option::Option<&'a Clause<'a>>,
221 pub subcommands: &'a [&'a Command<'a>],
222 /// Where a word goes when it names no subcommand of this one.
223 ///
224 /// The spec's `default_subcommand`. `mise build` means `mise run build`: the word names
225 /// no command, so the parser descends into `run` and lets *`run`* have it — even where
226 /// this command declares an argument of its own, which is what makes the property worth
227 /// having rather than a synonym for a positional.
228 ///
229 /// Applied at most once per parse, so a CLI cannot loop through it, and only where a
230 /// subcommand could still be selected.
231 ///
232 /// Resolve it with [`find_subcommand`], which turns a name that no subcommand answers to
233 /// into a compile error.
234 pub default_subcommand: ::core::option::Option<&'a Command<'a>>,
235 /// Look ahead past parent/default flags before implicitly selecting the default.
236 /// A sibling name before any default-only flag stays on the parent; after one,
237 /// later words are the default command's args.
238 pub default_subcommand_flags: bool,
239 /// Select the default subcommand for an otherwise empty successful invocation.
240 pub default_subcommand_on_empty: bool,
241 /// Whether an unmatched word is forwarded as an external command plus the rest of argv.
242 ///
243 /// clap's `allow_external_subcommands`. Known subcommands still win; a
244 /// [`default_subcommand`](Self::default_subcommand) still catches first. Once the
245 /// unmatched word is taken, remaining tokens — including `--help` — are not parsed
246 /// as this command's flags.
247 pub external_subcommand: bool,
248 /// Show this command's help when no argv token follows its name.
249 ///
250 /// This is clap's `arg_required_else_help`. It deliberately observes argv rather than
251 /// bound values: an environment variable or default may fill a field, but neither means
252 /// the user supplied an argument to this invocation.
253 pub arg_required_else_help: bool,
254 /// Selecting a subcommand suppresses this command's required arguments.
255 pub subcommand_negates_reqs: bool,
256 /// Once this command binds a flag or positional, selecting one of its
257 /// subcommands is an error.
258 pub args_conflicts_with_subcommands: bool,
259 /// Let a known subcommand interrupt a variadic argument that would otherwise consume it.
260 pub subcommand_precedence_over_arg: bool,
261 /// Let a later required positional take a word while an earlier optional positional
262 /// remains empty.
263 pub allow_missing_positional: bool,
264 /// Disable delimiter splitting for positional values after `--` or on an
265 /// automatic trailing argument. Inherited by subcommands.
266 pub dont_delimit_trailing_values: bool,
267 /// What an unrecognized flag-like token means here, or `None` to keep whatever the
268 /// enclosing command said. See [`UnknownFlags`].
269 ///
270 /// Inherited rather than resolved per command, which is what usage-lib does — its
271 /// `effective_unknown_flags` walks outward from the command that ran and falls back to
272 /// the spec's. Resolving it in the tables instead was possible only for a builder that
273 /// can see the whole tree: a derive expands one struct at a time and cannot see its
274 /// parent, so `#[usage(unknown_flags = "error")]` on the root reached the root alone and
275 /// a subcommand had no way to say it at all.
276 ///
277 /// The parser carries the effective value down as it descends, so a command that states
278 /// nothing costs nothing.
279 pub unknown_flags: ::core::option::Option<UnknownFlags>,
280 /// Whether this command answers to `--version` and `-V`.
281 ///
282 /// Set on the root, and only when the CLI declares a version: clap adds the flag exactly
283 /// then, and a `--version` that answers with nothing is worse than one that is not there.
284 /// A field rather than a rule about depth, so a CLI that wants it on a subcommand — clap's
285 /// `propagate_version` — has somewhere to say so.
286 pub version: bool,
287 /// Do not synthesize `--help` and `-h` for this command.
288 pub disable_help_flag: bool,
289 /// Do not synthesize the `help` subcommand route for this command.
290 pub disable_help_subcommand: bool,
291 /// Do not synthesize `--version` and `-V` for this command.
292 pub disable_version_flag: bool,
293 /// Caller-assigned identifier, echoed back in [`Event::Command`].
294 ///
295 /// Wide enough for a derive to make these unique without coordination: two
296 /// macro expansions cannot see each other, so the generated keys carry a hash
297 /// of the type they came from in the high half and a per-type index in the low
298 /// half. A parse dispatches on this, so a collision would bind the wrong field
299 /// — [`Spec::to_kdl`](crate::spec::Spec::to_kdl) checks the tree for duplicates
300 /// in debug builds.
301 pub key: u64,
302}
303
304impl Command<'_> {
305 /// A command with nothing declared, for use with struct update syntax.
306 pub const EMPTY: Command<'static> = Command {
307 name: "",
308 aliases: &[],
309 flags: &[],
310 args: &[],
311 clause: ::core::option::Option::None,
312 subcommands: &[],
313 default_subcommand: ::core::option::Option::None,
314 default_subcommand_flags: false,
315 default_subcommand_on_empty: false,
316 external_subcommand: false,
317 arg_required_else_help: false,
318 subcommand_negates_reqs: false,
319 args_conflicts_with_subcommands: false,
320 subcommand_precedence_over_arg: false,
321 allow_missing_positional: false,
322 dont_delimit_trailing_values: false,
323 unknown_flags: ::core::option::Option::None,
324 version: false,
325 disable_help_flag: false,
326 disable_help_subcommand: false,
327 disable_version_flag: false,
328 key: 0,
329 };
330}
331
332/// A repeatable group of scoped flags and positional arguments.
333#[derive(Debug, Clone, Copy, PartialEq, Eq)]
334pub struct Clause<'a> {
335 pub key: u64,
336 pub name: &'a str,
337 /// An explicit token that starts the next instance. When absent, consuming the
338 /// clause's sole required positional completes the current instance.
339 pub separator: ::core::option::Option<&'a [u8]>,
340 pub flags: &'a [&'a Flag<'a>],
341 pub args: &'a [&'a Arg<'a>],
342}
343
344/// Basename of argv[0] for a multicall CLI: last path component, with a trailing
345/// `.exe` stripped so Windows and Unix agree.
346pub fn multicall_basename(argv0: &str) -> &str {
347 let name = argv0.rsplit(['/', '\\']).next().unwrap_or(argv0);
348 match name.get(name.len().saturating_sub(4)..) {
349 Some(ext) if ext.eq_ignore_ascii_case(".exe") => &name[..name.len() - 4],
350 _ => name,
351 }
352}
353
354/// The applet name to parse as the first word, when argv[0] is not the dispatcher.
355///
356/// `None` means a dispatcher invocation (`busybox ls`): skip argv[0] and parse the
357/// rest. `Some` is a symlink invocation (`ls -l`): inject the basename.
358pub fn multicall_applet<'a>(argv0: &'a str, name: &str, bin: Option<&str>) -> Option<&'a str> {
359 let base = multicall_basename(argv0);
360 if !name.is_empty() && base == multicall_basename(name) {
361 return None;
362 }
363 if let Some(bin) = bin {
364 if !bin.is_empty() && base == multicall_basename(bin) {
365 return None;
366 }
367 }
368 Some(base)
369}
370
371/// Resolved identity of a derive-generated binding type.
372#[derive(Clone, Copy)]
373pub struct BindingType(pub fn() -> &'static str);
374
375impl BindingType {
376 pub fn name(self) -> &'static str {
377 (self.0)()
378 }
379}
380
381impl ::core::fmt::Debug for BindingType {
382 fn fmt(&self, f: &mut ::core::fmt::Formatter<'_>) -> ::core::fmt::Result {
383 f.debug_tuple("BindingType").field(&self.name()).finish()
384 }
385}
386
387impl PartialEq for BindingType {
388 fn eq(&self, other: &Self) -> bool {
389 self.name() == other.name()
390 }
391}
392
393impl Eq for BindingType {}
394
395/// A flag, addressed by any of its long or short forms.
396#[derive(Debug, Clone, Copy, PartialEq, Eq)]
397pub struct Flag<'a> {
398 /// Caller-assigned identifier, echoed back in [`Event::Flag`]. This is how
399 /// generated code knows which field to assign without any string comparison.
400 /// See [`Command::key`] on why it is this wide.
401 pub key: u64,
402 /// Compatibility key for mirroring a redeclared child global into an ancestor field.
403 ///
404 /// Zero means no typed binding contract is declared. Derive-generated tables hash the
405 /// binding shape and portable metadata so only equivalent bindings receive the same event.
406 pub binding_key: u64,
407 /// Resolved Rust value type for a derive-generated binding.
408 ///
409 /// This is separate from [`Self::binding_key`] because token spellings are not type
410 /// identities: an imported alias and a fully qualified path can name the same type.
411 pub binding_type: Option<BindingType>,
412 /// Unused by binding, kept so a table entry can carry its own name for
413 /// diagnostics.
414 pub name: &'a str,
415 /// Long forms, written without the leading `--`.
416 pub longs: &'a [&'a str],
417 /// Short forms, as single bytes.
418 ///
419 /// **Should be ASCII.** A cluster like `-xyz` is walked one byte at a time, so a
420 /// non-ASCII short can never be matched, and the remainder after a value-taking one —
421 /// which becomes its value — would begin in the middle of a character.
422 /// `#[derive(Cli)]` rejects a non-ASCII `short`; a table written by hand should keep to
423 /// it. Nothing is unsound if it does not: the value would simply be cut in a place that
424 /// makes no sense, and on Windows would then fail to convert.
425 pub shorts: &'a [u8],
426 /// A long form that sets the flag to false, written without the `--`.
427 pub negate: Option<&'a str>,
428 /// Whether the flag takes a value.
429 pub takes_value: bool,
430 /// Whether one occurrence of this flag keeps taking values, until a flag-like
431 /// token or the end of the command line.
432 ///
433 /// This is the spec's variadic flag *argument* (`--include <pattern>...`). It
434 /// is not the spec's flag-level `var=#true`, which means the flag may be
435 /// repeated and takes one value each time — repetition needs nothing from the
436 /// parser, since it already reports every occurrence separately. Conflating
437 /// the two makes a merely repeatable flag greedy enough to eat a positional.
438 pub variadic: bool,
439 /// How many values one variadic occurrence may take, after which the next word
440 /// belongs to whatever comes next.
441 ///
442 /// Only for [`variadic`](Self::variadic). A merely repeatable flag — the spec's
443 /// `var=#true` — is bounded on how many times it was *given*, which no single token
444 /// can decide, so that bound stays with the metadata and is checked after the parse.
445 pub var_max: ::core::option::Option<u32>,
446 /// The byte that makes one word several values, if the flag declares one.
447 ///
448 /// Here rather than with the metadata for the same reason [`var_max`](Self::var_max)
449 /// is: it decides *where* a word lands. A bound counts values, and a delimiter is what
450 /// makes a word stop being one of them — `--include a,b,c` is three, so a `var_max` of
451 /// two is already past its bound on the single word it was entitled to take. Binding
452 /// cannot count without it.
453 pub delimiter: ::core::option::Option<u8>,
454 /// Whether a detached value may itself look like a flag.
455 ///
456 /// The default is to refuse: `--jobs --force` is far more likely a forgotten
457 /// value than a jobs of `"--force"`. Declared, the next token is taken
458 /// whatever it looks like — including `--` — which is clap's
459 /// `allow_hyphen_values` and the spec's property of the same name. A variadic
460 /// occurrence still stops collecting at a later flag-like token, so a second
461 /// occurrence of the flag is not eaten as a value.
462 pub allow_hyphen_values: bool,
463 /// Whether a detached value may be a negative number while other flag-like
464 /// tokens still stop collection or report as flags.
465 pub allow_negative_numbers: bool,
466 /// A token that ends one variadic occurrence without becoming a value.
467 pub value_terminator: ::core::option::Option<&'a [u8]>,
468 /// Whether the value must be attached with `=`.
469 ///
470 /// `--flag=value` is accepted and `--flag value` is not, which is clap's
471 /// `require_equals` and the spec's property of the same name. A short's
472 /// attached form (`-i9229`, `-i=9229`) still binds: only the following word
473 /// is refused.
474 pub require_equals: bool,
475 /// Whether this value-taking flag may be present without a value.
476 ///
477 /// A missing value emits the flag event with `value: None`; bindings such as
478 /// `Option<Option<T>>` can therefore distinguish an absent flag from a bare
479 /// flag and from a flag with an explicit value.
480 pub value_optional: bool,
481 /// Whether a boolean long flag accepts an attached `true` or `false` value.
482 ///
483 /// This does not make the flag value-taking in the ordinary sense: a detached
484 /// word is never consumed, and help keeps rendering a switch. Only
485 /// `--flag=true` and `--flag=false` opt into an explicit boolean value.
486 pub bool_value: bool,
487 /// Value used when the flag is present but no value is given.
488 ///
489 /// clap's `default_missing_value` and the spec's `default_missing`. `--color`
490 /// binds this, `--color=never` binds `never`, and an absent flag is not bound.
491 /// Combined with [`Self::require_equals`], a following word is still refused.
492 pub default_missing: ::core::option::Option<&'a [u8]>,
493 /// Whether the flag is recognized by every command beneath the one that
494 /// declares it.
495 pub global: bool,
496 /// Whether this declared flag binds a field or requests a built-in response.
497 pub action: ArgAction,
498}
499
500impl Flag<'_> {
501 /// A value-less flag, for use with struct update syntax.
502 pub const BOOL: Flag<'static> = Flag {
503 key: 0,
504 binding_key: 0,
505 binding_type: None,
506 name: "",
507 longs: &[],
508 shorts: &[],
509 negate: None,
510 takes_value: false,
511 variadic: false,
512 var_max: ::core::option::Option::None,
513 delimiter: ::core::option::Option::None,
514 allow_hyphen_values: false,
515 allow_negative_numbers: false,
516 value_terminator: ::core::option::Option::None,
517 require_equals: false,
518 value_optional: false,
519 bool_value: false,
520 default_missing: ::core::option::Option::None,
521 global: false,
522 action: ArgAction::Set,
523 };
524
525 /// A flag that takes a value, for use with struct update syntax.
526 pub const VALUE: Flag<'static> = Flag {
527 takes_value: true,
528 ..Flag::BOOL
529 };
530}
531
532/// What supplying a declared flag does.
533#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
534pub enum ArgAction {
535 /// Bind the flag to its declared field.
536 #[default]
537 Set,
538 /// Show help, choosing the long form for a long spelling and the short form otherwise.
539 Help,
540 /// Always show short help.
541 HelpShort,
542 /// Always show long help.
543 HelpLong,
544 /// Show long help for this command and every visible descendant.
545 HelpAll,
546 /// Show version information.
547 Version,
548}
549
550/// A positional argument.
551#[derive(Debug, Clone, Copy, PartialEq, Eq)]
552pub struct Arg<'a> {
553 /// Caller-assigned identifier, echoed back in [`Event::Arg`]. See
554 /// [`Command::key`] on why it is this wide.
555 pub key: u64,
556 /// Prefix that classifies this positional independently of declaration order.
557 /// The prefix is removed from the value emitted in [`Event::Arg`].
558 pub sigil: ::core::option::Option<&'a [u8]>,
559 /// Whether post-binding requires this positional to have a value. Kept in the hot
560 /// table because `allow_missing_positional` must reserve words for later required args.
561 pub required: bool,
562 /// Whether this argument keeps taking values once it has one.
563 pub var: bool,
564 /// How many words a variadic may take before the next argument gets the rest.
565 ///
566 /// A bound belongs here, in the table binding reads, rather than with the metadata:
567 /// it decides *where* a word lands, not whether what landed is acceptable. clap's
568 /// `num_args` works the same way, and every spec in the wild is generated from a clap
569 /// command. `u32` rather than `usize` because a CLI that bounds a variadic above four
570 /// billion has other problems, and this table is read on the hot path.
571 pub var_max: ::core::option::Option<u32>,
572 /// The byte that makes one word several values, if the argument declares one.
573 ///
574 /// See [`Flag::delimiter`]: a bound counts values, and only this says how many values a
575 /// word carries.
576 pub delimiter: ::core::option::Option<u8>,
577 /// Whether a negative-number token is accepted as this positional even in
578 /// strict flag mode.
579 pub allow_negative_numbers: bool,
580 /// A token that ends this variadic positional without becoming a value.
581 pub value_terminator: ::core::option::Option<&'a [u8]>,
582 /// This argument's relationship to the `--` separator.
583 pub double_dash: DoubleDash,
584 /// Unused by binding, kept so a table entry can carry its own name for
585 /// diagnostics.
586 pub name: &'a str,
587}
588
589impl Arg<'_> {
590 /// A single-value argument, for use with struct update syntax.
591 pub const REQUIRED: Arg<'static> = Arg {
592 key: 0,
593 sigil: ::core::option::Option::None,
594 required: true,
595 var: false,
596 var_max: ::core::option::Option::None,
597 delimiter: ::core::option::Option::None,
598 allow_negative_numbers: false,
599 value_terminator: ::core::option::Option::None,
600 double_dash: DoubleDash::Optional,
601 name: "",
602 };
603
604 /// A variadic argument, for use with struct update syntax.
605 pub const VAR: Arg<'static> = Arg {
606 var: true,
607 ..Arg::REQUIRED
608 };
609}
610
611/// What to do with a flag-like token that names no flag in scope.
612///
613/// The default is [`UnknownFlags::Value`]: the token carries on to the positional
614/// arguments, because a spec is often parsing a command line whose flags belong to
615/// something else — a wrapped tool, a task script. A CLI that owns all of its
616/// flags declares [`UnknownFlags::Error`] and gets typo detection instead.
617///
618/// Stored per command and already resolved: inheritance is a question for whoever
619/// builds the tables, and answering it at compile time keeps it out of the parse.
620#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
621pub enum UnknownFlags {
622 /// Offer the token to the positionals. If none can take it, it is an
623 /// unexpected argument.
624 #[default]
625 Value,
626 /// Reject the token.
627 Error,
628}
629
630/// How an argument relates to the `--` separator.
631#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
632pub enum DoubleDash {
633 /// Values may appear on either side of a `--`.
634 #[default]
635 Optional,
636 /// Values are accepted only after a `--`.
637 Required,
638 /// A `--` is kept as a value rather than consumed as a separator.
639 Preserve,
640 /// Once the argument takes a value, behave as if a `--` had been given, so
641 /// the rest of the command line is values. A wrapper can then forward flags
642 /// without its caller typing the separator.
643 Automatic,
644}
645
646/// Something the parser bound.
647#[derive(Debug, Clone, Copy, PartialEq, Eq)]
648pub enum Event<'t, 'a, 'v> {
649 /// A subcommand was selected; parsing continues inside it.
650 Command(&'t Command<'t>),
651 /// A flag was given. `value` is `Some` for a flag that takes one, and
652 /// `negated` is true when the flag was set through its `negate` form.
653 Flag {
654 flag: &'t Flag<'t>,
655 value: Option<&'v [u8]>,
656 negated: bool,
657 },
658 /// A word was bound to a positional argument. A variadic argument produces
659 /// one event per value.
660 Arg {
661 arg: &'t Arg<'t>,
662 value: &'v [u8],
663 /// Whether this value should be split by the argument's declared delimiter.
664 delimit: bool,
665 },
666 /// Ended one clause instance and began the next.
667 ClauseSeparator { clause: Clause<'t> },
668 /// An unmatched word was forwarded as an external command: the name, then
669 /// every remaining token, including flags.
670 External { values: &'a [&'v OsStr] },
671}
672
673/// A binding failure.
674///
675/// Carries the offending token so a caller can render a good message, but no
676/// message of its own: rendering belongs to a cold path, and building a string
677/// here would allocate on the way to reporting that nothing was allocated.
678///
679/// `non_exhaustive`, because an error enum grows: a caller matching on it needs a
680/// fallback arm so that recognizing a new failure is never a breaking change.
681// No `Copy`: one variant owns its message. `Clone` stays, and the enum is still 40 bytes
682// because that variant is boxed, so nothing on the hot path grew.
683#[derive(Debug, Clone, PartialEq, Eq)]
684#[non_exhaustive]
685pub enum Error<'t, 'v> {
686 /// A flag-like token matched no flag in scope. `token` is the whole token as
687 /// typed, so a bundle containing an unrecognized letter reports `-fz` rather
688 /// than the letter alone — which is also the unit in which it is rejected.
689 UnknownFlag { token: &'v [u8] },
690 /// A flag that needs a value did not get one, either because the command
691 /// line ended or because the next token was flag-like.
692 MissingFlagValue { flag: &'t Flag<'t> },
693 /// A word arrived with no argument left to hold it.
694 UnexpectedArg { token: &'v [u8] },
695 /// A word was offered to a `double_dash = "required"` argument before any
696 /// `--` had been seen.
697 ArgRequiresDoubleDash { arg: &'t Arg<'t> },
698 /// A subcommand was selected after this command had already bound an argument.
699 SubcommandConflict { subcommand: &'t Command<'t> },
700 /// The command tree is deeper than [`MAX_DEPTH`].
701 TooDeep,
702
703 // The rest are raised *after* the parse, by whoever owns the target type: they
704 // need to know a value's declared type, which the parser deliberately does not.
705 // They share this enum so that a caller has one error to handle rather than two.
706 /// Something the command requires was never given.
707 MissingRequired {
708 /// The flag or argument's name, as the spec calls it.
709 name: &'t str,
710 },
711 /// A flag that is not repeatable was given more than once.
712 DuplicateFlag {
713 /// The flag's name, as the spec calls it.
714 name: &'t str,
715 },
716 /// A value was given that is not among the declared choices.
717 ///
718 /// Carries the choices rather than the offending value: rendering the value means
719 /// owning it, and an error that allocates on a path this crate promises not to
720 /// allocate on would be a poor trade for a better message. Diagnostics are a
721 /// separate layer.
722 InvalidChoice {
723 name: &'t str,
724 choices: &'t [&'t str],
725 },
726 /// Fewer values than `var_min`.
727 VarTooFew {
728 name: &'t str,
729 min: usize,
730 got: usize,
731 },
732 /// More values than `var_max`.
733 VarTooMany {
734 name: &'t str,
735 max: usize,
736 got: usize,
737 },
738 /// Two flags declared to conflict were both given.
739 ///
740 /// Carries both names because either one alone reads as a puzzle: which flag is
741 /// unwelcome depends entirely on what else is on the command line.
742 ConflictingFlags {
743 /// The flag whose declaration names the conflict.
744 name: &'t str,
745 /// The flag it cannot be given with, as the declaration spells it.
746 other: &'t str,
747 },
748 /// A value was given that the field's type could not be built from.
749 ///
750 /// Boxed, and the only error here that owns anything. Everything else borrows the
751 /// tables or argv, which is what keeps a *successful* parse allocation-free — and the
752 /// box keeps `Error` the size it was, so the `Result` this rides in on the hot path
753 /// does not grow. A value that will not convert has already failed, and a message
754 /// worth reading is worth one allocation.
755 InvalidValue(::std::boxed::Box<InvalidValue<'t>>),
756 /// A required group had none of its members given.
757 ///
758 /// Carries the members as members rather than as a rendered sentence: the caller
759 /// decides how to say it, and a completion asking what would satisfy this needs the
760 /// list rather than the prose.
761 MissingGroup {
762 /// The group's declared name, which appears in the message so a command with
763 /// several groups does not report the same sentence twice.
764 group: &'t str,
765 /// The flags that would satisfy it, as the declaration spells them.
766 members: &'t [&'t str],
767 },
768 /// A subcommand was required, and none was given.
769 MissingSubcommand,
770 /// `--help` or `-h` was given, and `cmd` is what it was asked about.
771 ///
772 /// Not a failure, and returned as one anyway: a parse that stops to print help has not
773 /// produced a value, and every caller already handles the "no value" shape. clap does the
774 /// same thing for the same reason.
775 ///
776 /// `long` distinguishes the two: `-h` prints the short form and `--help` the long one, as
777 /// clap has them. The caller renders — this crate does not print, because a library that
778 /// writes to stdout on its own is one an adopter cannot embed.
779 Help { cmd: &'t Command<'t>, long: bool },
780 /// `arg_required_else_help` found no command-line arguments for `cmd`.
781 ///
782 /// Unlike an explicit help request, this is a usage failure: clap prints the short help to
783 /// stderr and exits with status 2. Keeping the shape separate lets embedders preserve that
784 /// terminal contract without guessing why [`Error::Help`] was returned.
785 MissingArgsHelp { cmd: &'t Command<'t> },
786 /// Recursive long help was requested for `cmd` and every visible descendant.
787 HelpAll { cmd: &'t Command<'t> },
788 /// `--version` or `-V` was asked for. Not a failure either — the caller prints and leaves.
789 ///
790 /// The version string lives in the spec rather than the parse tables. `long` lets the
791 /// caller choose `long_version` for `--version` while `-V` retains the concise value.
792 Version { long: bool },
793}
794
795/// The high half of every key one declaration's items get.
796///
797/// A derive cannot see other expansions, so it cannot hand out keys from a shared
798/// counter: it hashes the declaration it was given instead. It cannot see a module path
799/// either, which is why the module is mixed in *here* — `module_path!()` is available to
800/// the generated code as a compile-time string, so two byte-identical declarations in
801/// different modules end up with different keys rather than colliding.
802///
803/// `declaration` is a hash the derive computed over the item's own tokens.
804pub const fn key_base(module: &str, declaration: u32) -> u64 {
805 // FNV-1a, continuing from the declaration's hash rather than starting over, so both
806 // halves contribute. Spelled out rather than taken from a `Hasher`, which is not
807 // guaranteed to be stable between compilations — and these are baked into a binary.
808 let mut hash: u32 = declaration;
809 let bytes = module.as_bytes();
810 let mut i = 0;
811 while i < bytes.len() {
812 hash ^= bytes[i] as u32;
813 hash = hash.wrapping_mul(0x0100_0193);
814 i += 1;
815 }
816 (hash as u64) << 32
817}
818
819/// Why a value would not convert into the type its field holds.
820///
821/// Separate from [`Error`] so that the enum stays small: this is reached through a `Box`.
822#[derive(Debug, Clone, PartialEq, Eq)]
823pub struct InvalidValue<'t> {
824 /// The flag or argument's name, as the spec calls it.
825 pub name: &'t str,
826 /// The text that would not convert.
827 pub value: ::std::string::String,
828 /// What the type's own conversion complained about.
829 pub reason: ::std::string::String,
830}
831
832/// A command-wide validation or finalization failure.
833///
834/// Return this from a `#[usage(validate_with = ...)]` hook or from the
835/// `TryFrom` implementation named by `#[usage(try_into = ...)]`. The derive
836/// turns it into the same [`Error::InvalidValue`] diagnostic used by field
837/// conversion, so callers keep one parse error type.
838#[derive(Debug, Clone, PartialEq, Eq)]
839pub struct ValidationError {
840 name: &'static str,
841 value: String,
842 reason: String,
843}
844
845impl ValidationError {
846 /// Start an error attributed to a flag, positional, or command name.
847 pub fn field(name: &'static str) -> Self {
848 Self {
849 name,
850 value: String::new(),
851 reason: String::new(),
852 }
853 }
854
855 /// Record the value that failed the command-wide invariant.
856 pub fn value(mut self, value: impl Into<String>) -> Self {
857 self.value = value.into();
858 self
859 }
860
861 /// Explain the invariant that the value did not satisfy.
862 pub fn reason(mut self, reason: impl Into<String>) -> Self {
863 self.reason = reason.into();
864 self
865 }
866
867 /// Convert this application-level failure into the parser's diagnostic.
868 pub fn into_parse_error<'v>(self) -> Error<'static, 'v> {
869 Error::InvalidValue(Box::new(InvalidValue {
870 name: self.name,
871 value: self.value,
872 reason: self.reason,
873 }))
874 }
875}
876
877/// Interpret a value as UTF-8.
878///
879/// The parser hands back bytes borrowed from `argv`; this is the conversion most
880/// callers want, and the point at which a non-UTF-8 command line is rejected —
881/// but only for the values actually inspected.
882pub fn as_str(value: &[u8]) -> Result<&str, std::str::Utf8Error> {
883 std::str::from_utf8(value)
884}
885
886/// How many entries a group of tables holds in total.
887///
888/// The length for [`concat_flags`] and [`concat_args`], which need it as a const generic — so
889/// it has to be computable separately from the concatenation itself.
890///
891/// ```
892/// use usage_argv::{table_len, Flag};
893///
894/// static A: Flag = Flag { name: "a", ..Flag::BOOL };
895/// static B: Flag = Flag { name: "b", ..Flag::BOOL };
896/// const GROUPS: &[&[&Flag]] = &[&[&A], &[], &[&B]];
897/// const N: usize = table_len(GROUPS);
898/// assert_eq!(N, 2);
899/// ```
900pub const fn table_len<T>(groups: &[&[T]]) -> usize {
901 let mut total = 0;
902 let mut i = 0;
903 while i < groups.len() {
904 total += groups[i].len();
905 i += 1;
906 }
907 total
908}
909
910/// Join groups of flag tables into one, at compile time.
911///
912/// This is how `#[usage(flatten)]` stays free. A flattened struct's flags have to appear in
913/// the parent's own table, and the parent's macro expansion cannot see them — it has only a
914/// type. But it can name that type's [`CommandArgs::COMMAND`](crate::spec::CommandArgs::COMMAND),
915/// and a `const fn` can read through it, so the two lists become one `static` array before the
916/// program runs. The parser then walks a single flat slice, exactly as it does for a command
917/// that declared everything itself: flatten costs nothing at run time.
918///
919/// Groups are laid out in the order given, which is what lets a flattened group sit *between*
920/// two of the parent's own declarations — necessary for positional arguments, where order is
921/// the meaning.
922///
923/// `N` must be [`table_len`] of the same groups. It cannot be inferred, and a wrong one fails
924/// to compile rather than leaving the difference filled with padding.
925///
926/// ```
927/// use usage_argv::{concat_flags, table_len, Flag};
928///
929/// static FORCE: Flag = Flag { name: "force", longs: &["force"], ..Flag::BOOL };
930/// static QUIET: Flag = Flag { name: "quiet", longs: &["quiet"], ..Flag::BOOL };
931/// static SHARED: &[&Flag] = &[&QUIET];
932///
933/// const GROUPS: &[&[&Flag]] = &[&[&FORCE], SHARED];
934/// static FLAGS: [&Flag; table_len(GROUPS)] = concat_flags(GROUPS);
935///
936/// assert_eq!(FLAGS.iter().map(|f| f.name).collect::<Vec<_>>(), ["force", "quiet"]);
937/// ```
938pub const fn concat_flags<const N: usize>(
939 groups: &[&[&'static Flag<'static>]],
940) -> [&'static Flag<'static>; N] {
941 // Every slot is written below, but an array has to start somewhere and `MaybeUninit`
942 // would mean `unsafe`. A `Flag` nobody can reach is cheaper than that.
943 static PLACEHOLDER: Flag<'static> = Flag::BOOL;
944 let mut out = [&PLACEHOLDER; N];
945 let mut at = 0;
946 let mut g = 0;
947 while g < groups.len() {
948 let group = groups[g];
949 let mut i = 0;
950 while i < group.len() {
951 out[at] = group[i];
952 at += 1;
953 i += 1;
954 }
955 g += 1;
956 }
957 assert!(
958 at == N,
959 "`N` must be `table_len` of the same groups, or the table would keep a placeholder \
960 that answers to nothing"
961 );
962 out
963}
964
965/// Join groups of argument tables into one, at compile time.
966///
967/// The positional counterpart of [`concat_flags`] — see there for why this exists. Order
968/// matters more here: an argument's position *is* its identity, so a flattened group has to
969/// land exactly where the field was written.
970///
971/// Two functions rather than one generic: each needs a value to fill an array with before
972/// overwriting it, and there is no way to ask a type parameter for one in a `const fn`.
973pub const fn concat_args<const N: usize>(
974 groups: &[&[&'static Arg<'static>]],
975) -> [&'static Arg<'static>; N] {
976 static PLACEHOLDER: Arg<'static> = Arg::REQUIRED;
977 let mut out = [&PLACEHOLDER; N];
978 let mut at = 0;
979 let mut g = 0;
980 while g < groups.len() {
981 let group = groups[g];
982 let mut i = 0;
983 while i < group.len() {
984 out[at] = group[i];
985 at += 1;
986 i += 1;
987 }
988 g += 1;
989 }
990 assert!(
991 at == N,
992 "`N` must be `table_len` of the same groups, or the table would keep a placeholder \
993 that answers to nothing"
994 );
995 out
996}
997
998/// The key `--help` answers to, and the one `-h` does.
999///
1000/// Reserved rather than generated: a derive builds keys from a hash of the type they came from
1001/// in the high half and an index in the low half, so the top of the range belongs to nobody.
1002/// Generated code compares against these to tell a help request from a flag of its own.
1003pub const HELP_LONG_KEY: u64 = u64::MAX;
1004/// See [`HELP_LONG_KEY`].
1005pub const HELP_SHORT_KEY: u64 = u64::MAX - 1;
1006
1007/// `--help`, which every command answers to.
1008///
1009/// In the parse table and *not* in the metadata, which is the whole trick: the parser has to
1010/// recognise the flag, and help output must not list it — a spec does not declare `--help`, so
1011/// showing one would make the rendered page disagree with the spec it came from.
1012pub static HELP_LONG: Flag<'static> = Flag {
1013 key: HELP_LONG_KEY,
1014 name: "help",
1015 longs: &["help"],
1016 action: ArgAction::HelpLong,
1017 ..Flag::BOOL
1018};
1019
1020/// See [`HELP_LONG_KEY`].
1021pub const VERSION_LONG_KEY: u64 = u64::MAX - 2;
1022/// See [`HELP_LONG_KEY`].
1023pub const VERSION_SHORT_KEY: u64 = u64::MAX - 3;
1024
1025/// `--version`, where the CLI declared one.
1026///
1027/// In the parse table and not in the metadata, exactly as `--help` is: a spec does not declare
1028/// `--version`, so listing one would make the rendered page disagree with the spec it came from.
1029pub static VERSION_LONG: Flag<'static> = Flag {
1030 key: VERSION_LONG_KEY,
1031 name: "version",
1032 longs: &["version"],
1033 action: ArgAction::Version,
1034 ..Flag::BOOL
1035};
1036
1037/// `-V`, which clap also supplies.
1038pub static VERSION_SHORT: Flag<'static> = Flag {
1039 key: VERSION_SHORT_KEY,
1040 name: "version",
1041 shorts: b"V",
1042 action: ArgAction::Version,
1043 ..Flag::BOOL
1044};
1045
1046/// `-h`, which prints the shorter form.
1047pub static HELP_SHORT: Flag<'static> = Flag {
1048 key: HELP_SHORT_KEY,
1049 name: "help",
1050 shorts: b"h",
1051 action: ArgAction::HelpShort,
1052 ..Flag::BOOL
1053};
1054
1055/// A named subcommand of a given command, by name or alias.
1056///
1057/// Free rather than a method because `help` resolves a path *without* descending: the words
1058/// after it are a question about a command rather than a walk into one.
1059///
1060/// Names across every subcommand before any alias, the precedence the grammar states — and
1061/// the reason this is the only implementation of it on argv's side. `ex run` and `ex help run`
1062/// selecting different commands would be exactly the divergence this rule was written to end.
1063pub(crate) fn find_named<'t>(cmd: &'t Command<'t>, name: &[u8]) -> Option<&'t Command<'t>> {
1064 let subcommands = || cmd.subcommands.iter().copied();
1065 subcommands()
1066 .find(|c| c.name.as_bytes() == name)
1067 .or_else(|| subcommands().find(|c| c.aliases.iter().any(|a| a.as_bytes() == name)))
1068}
1069
1070/// What a caller should print for a parse failure, and what to exit with.
1071///
1072/// The one entry point a generated `parse()` reaches for, and the reason it exists here rather
1073/// than in the derive: whether the good rendering is available is a *feature of this crate* in
1074/// the adopter's dependency graph, and a `#[cfg]` written into generated code is evaluated in
1075/// the adopter's crate, where the feature is not theirs to see. That is how a metadata field
1076/// once got silently dropped; the answer is that the cfg lives beside the thing it gates.
1077///
1078/// With `diagnostics` on, this is the clap-shaped message. Without it, the error's `Debug`
1079/// form — which is still better than nothing and is what a parser-only build asked for.
1080///
1081/// [`Error::Help`] and [`Error::Version`] are not failures and must be handled before this.
1082#[cfg(feature = "diagnostics")]
1083pub fn render_failure(spec: &spec::Spec<'_>, argv: &[&OsStr], error: &Error<'_, '_>) -> String {
1084 diagnostic::render(spec, argv, error, diagnostic::Style::auto())
1085}
1086
1087/// A parse failure, never coloured.
1088///
1089/// [`render_failure`] asks the environment whether to colour, which is right for a process and
1090/// wrong for anything that keeps the string: a test that asserts on a message, or a snapshot of
1091/// one, would pass or fail by whether stderr happened to be a terminal. The renderer is the
1092/// same; only the answer to that question is fixed.
1093#[cfg(feature = "diagnostics")]
1094pub fn render_failure_plain(
1095 spec: &spec::Spec<'_>,
1096 argv: &[&OsStr],
1097 error: &Error<'_, '_>,
1098) -> String {
1099 diagnostic::render(spec, argv, error, diagnostic::Style::PLAIN)
1100}
1101
1102/// Render a failure through a spec-declared executable view.
1103///
1104/// `argv` is the original full argv, including the view executable as argv0.
1105#[cfg(feature = "diagnostics")]
1106pub fn render_failure_view<'a>(
1107 spec: &'a spec::Spec<'a>,
1108 argv: &[&OsStr],
1109 error: &Error<'_, '_>,
1110 view: &'a spec::ViewMeta<'a>,
1111) -> String {
1112 diagnostic::render_view(spec, argv, error, diagnostic::Style::auto(), view)
1113}
1114
1115/// What a caller should print for a parse failure, without the renderer that makes it readable.
1116///
1117/// See the other half. A caller that wants the clap-shaped message turns on `diagnostics`;
1118/// this is what a parser-only build asked for, and it still says which error it was.
1119#[cfg(all(feature = "spec", not(feature = "diagnostics")))]
1120pub fn render_failure(spec: &spec::Spec<'_>, argv: &[&OsStr], error: &Error<'_, '_>) -> String {
1121 let _ = (spec, argv);
1122 ::std::format!("error: {error:?}\n")
1123}
1124
1125/// A parse failure without the renderer, which is plain either way.
1126#[cfg(all(feature = "spec", not(feature = "diagnostics")))]
1127pub fn render_failure_plain(
1128 spec: &spec::Spec<'_>,
1129 argv: &[&OsStr],
1130 error: &Error<'_, '_>,
1131) -> String {
1132 render_failure(spec, argv, error)
1133}
1134
1135/// Render a failure through a declared view without the optional diagnostics renderer.
1136#[cfg(all(feature = "spec", not(feature = "diagnostics")))]
1137pub fn render_failure_view(
1138 spec: &spec::Spec<'_>,
1139 argv: &[&OsStr],
1140 error: &Error<'_, '_>,
1141 view: &spec::ViewMeta<'_>,
1142) -> String {
1143 let _ = (spec, argv, view);
1144 ::std::format!("error: {error:?}\n")
1145}
1146
1147/// Answer an error from a generated `parse()` the way a command-line program does, and leave.
1148///
1149/// A help request goes to stdout and exits 0; a request for help because nothing was given goes
1150/// to stderr and exits 2; any other failure is rendered to stderr and exits 2, which is clap's
1151/// status, so a script that checks for it keeps working. [`Error::Version`] is the caller's to
1152/// answer before this, since only the generated code knows the version it prints.
1153///
1154/// Here rather than expanded into every CLI: the body is the same for all of them, and only the
1155/// spec — which carries any computed identity the generated code has already evaluated — and
1156/// the tables differ. `all_argv` includes argv0, which a view's route needs; `argv` does not.
1157#[cfg(feature = "spec")]
1158#[doc(hidden)]
1159pub fn __usage_exit_on_error(
1160 error: Error<'_, '_>,
1161 spec: &spec::Spec<'_>,
1162 root: &Command<'_>,
1163 all_argv: &[&OsStr],
1164 argv: &[&OsStr],
1165 view: Option<&spec::ViewMeta<'_>>,
1166) -> ! {
1167 let (want, to_stderr, cmd) = match error {
1168 Error::Help { cmd, long } => {
1169 let want = if long {
1170 help::Page::Long
1171 } else {
1172 help::Page::Short
1173 };
1174 (want, false, cmd)
1175 }
1176 Error::MissingArgsHelp { cmd } => (help::Page::Short, true, cmd),
1177 Error::HelpAll { cmd } => (help::Page::All, false, cmd),
1178 error => {
1179 let failure = match view {
1180 Some(view) => render_failure_view(spec, all_argv, &error, view),
1181 None => render_failure(spec, argv, &error),
1182 };
1183 ::std::eprint!("{failure}");
1184 __usage_process_exit(2);
1185 }
1186 };
1187 let style = if to_stderr {
1188 help::Style::auto_stderr()
1189 } else {
1190 help::Style::auto()
1191 };
1192 let page = match view {
1193 Some(view) => help::page_view(spec, root, all_argv, cmd, view, want, style),
1194 None => help::page(spec, root, argv, cmd, want, style),
1195 };
1196 // `None` is only reachable if the command came from another CLI's tables.
1197 if let Some(page) = page {
1198 if to_stderr {
1199 ::std::eprint!("{page}");
1200 } else {
1201 ::std::print!("{page}");
1202 }
1203 }
1204 __usage_process_exit(if to_stderr { 2 } else { 0 })
1205}
1206
1207/// Print `bin version` for a generated `parse()`, and leave.
1208///
1209/// The half of [`__usage_exit_on_error`] the generated code answers itself, because the
1210/// version is a declaration of that CLI; the printing and the exit are the same everywhere.
1211#[cfg(feature = "spec")]
1212#[doc(hidden)]
1213pub fn __usage_exit_version(
1214 bin: &str,
1215 view: Option<&spec::ViewMeta<'_>>,
1216 version: &dyn std::fmt::Display,
1217) -> ! {
1218 let bin = view.map_or(bin, |view| view.bin);
1219 ::std::println!("{bin} {version}");
1220 __usage_process_exit(0)
1221}
1222
1223/// What a caller should print for the deprecations a command line used.
1224///
1225/// The same arrangement as [`render_failure`], and for the same reason: whether the coloured
1226/// rendering is available is a feature of *this* crate in the adopter's dependency graph, so the
1227/// `#[cfg]` lives beside the thing it gates rather than in generated code.
1228///
1229/// Warnings are not failures. A caller prints these to stderr and carries on.
1230#[cfg(feature = "diagnostics")]
1231pub fn render_warnings(warnings: &[warn::Warning<'_>]) -> String {
1232 diagnostic::render_warnings(warnings, diagnostic::Style::auto())
1233}
1234
1235/// The same wording without the renderer that colours it. See the other half.
1236#[cfg(all(feature = "spec", not(feature = "diagnostics")))]
1237pub fn render_warnings(warnings: &[warn::Warning<'_>]) -> String {
1238 warn::render_warnings(warnings)
1239}
1240
1241/// The word a tool sends to ask a binary for its own spec.
1242///
1243/// Not a flag and not a command: a spec request is not something this CLI *does*, so it is
1244/// answered before the parse and stays out of the tables — the same reason
1245/// `__complete_word__` is a word rather than a subcommand. It also keeps the endpoint from
1246/// perturbing the document it prints, which a declared flag would not.
1247pub const SPEC_REQUEST: &str = "__usage_spec__";
1248
1249/// Whether this argv asks for the spec rather than for the CLI to run.
1250///
1251/// Only the first word counts: `mycli build __usage_spec__` passes the word through as an
1252/// ordinary value, because a request is the whole invocation or it is nothing.
1253///
1254/// A root that declares a command of that name keeps it, which is the precedence the `help`
1255/// subcommand already has. The check is here rather than in the derive because a `Cli` derive
1256/// expands one struct and cannot see the variant names of a separate `Subcommands` enum — the
1257/// static tables can.
1258pub fn is_spec_request(root: &Command<'_>, argv: &[&OsStr]) -> bool {
1259 let [first, ..] = argv else { return false };
1260 first.as_encoded_bytes() == SPEC_REQUEST.as_bytes()
1261 && find_named(root, SPEC_REQUEST.as_bytes()).is_none()
1262}
1263
1264/// Whether a flag is one of the two the parser supplies rather than the CLI declaring it.
1265pub fn is_help_flag(flag: &Flag<'_>) -> bool {
1266 matches!(
1267 flag.action,
1268 ArgAction::Help | ArgAction::HelpShort | ArgAction::HelpLong | ArgAction::HelpAll
1269 )
1270}
1271
1272/// Whether a flag is one of the two the parser supplies for `--version`.
1273pub fn is_version_flag(flag: &Flag<'_>) -> bool {
1274 flag.action == ArgAction::Version
1275}
1276
1277/// Whether one exact root argument selects a declared or synthesized version action.
1278///
1279/// Declared flags are checked first because they shadow the built-in `--version` and `-V`
1280/// spellings. Executable views use this before projection so custom version spellings keep
1281/// reporting the package that owns the view.
1282pub fn is_version_arg(cmd: &Command<'_>, word: &OsStr) -> bool {
1283 let token = word.as_encoded_bytes();
1284 if let Some(long) = token.strip_prefix(b"--") {
1285 if let Some(flag) = cmd.flags.iter().find(|flag| {
1286 flag.longs
1287 .iter()
1288 .any(|spelling| spelling.as_bytes() == long)
1289 }) {
1290 return is_version_flag(flag);
1291 }
1292 if cmd.flags.iter().any(|flag| {
1293 flag.negate
1294 .is_some_and(|spelling| spelling.as_bytes() == long)
1295 }) {
1296 return false;
1297 }
1298 return long == b"version" && cmd.version && !cmd.disable_version_flag;
1299 }
1300 if let [b'-', short] = token {
1301 if let Some(flag) = cmd.flags.iter().find(|flag| flag.shorts.contains(short)) {
1302 return is_version_flag(flag);
1303 }
1304 return *short == b'V' && cmd.version && !cmd.disable_version_flag;
1305 }
1306 false
1307}
1308
1309/// Resolve a subcommand by name or alias, at compile time.
1310///
1311/// For [`Command::default_subcommand`], which names a command that a derive cannot see: the
1312/// variants of a subcommand enum are a different macro expansion, so the name is all the
1313/// parent has. Searching the list in a `const fn` closes that gap — the answer is the same
1314/// `&'static` the table already holds, found before the program runs.
1315///
1316/// A name no subcommand answers to is a **compile error**, since this panics during const
1317/// evaluation. That is the whole point of doing it here rather than at startup.
1318///
1319/// ```
1320/// use usage_argv::{find_subcommand, Command};
1321///
1322/// static RUN: Command = Command { name: "run", ..Command::EMPTY };
1323/// static SUBS: &[&Command] = &[&RUN];
1324/// static ROOT: Command = Command {
1325/// name: "ex",
1326/// subcommands: SUBS,
1327/// default_subcommand: Some(find_subcommand(SUBS, "run")),
1328/// ..Command::EMPTY
1329/// };
1330/// assert_eq!(ROOT.default_subcommand.unwrap().name, "run");
1331/// ```
1332pub const fn find_subcommand<'a>(
1333 subcommands: &'a [&'a Command<'a>],
1334 name: &str,
1335) -> &'a Command<'a> {
1336 // Names first, then aliases: a command's own name outranks another command's alias, so
1337 // the answer does not depend on the order the table happens to list them in. Checking
1338 // each candidate's name *and* aliases in one pass instead let whichever command came
1339 // first win, and usage-lib resolved the same spec to the last one.
1340 let mut i = 0;
1341 while i < subcommands.len() {
1342 if str_eq(subcommands[i].name, name) {
1343 return subcommands[i];
1344 }
1345 i += 1;
1346 }
1347 // Aliases answer too, because usage-lib resolves the name against names, aliases and
1348 // hidden aliases alike — so a spec may point `default_subcommand` at any of them.
1349 let mut i = 0;
1350 while i < subcommands.len() {
1351 let candidate = subcommands[i];
1352 let mut a = 0;
1353 while a < candidate.aliases.len() {
1354 if str_eq(candidate.aliases[a], name) {
1355 return candidate;
1356 }
1357 a += 1;
1358 }
1359 i += 1;
1360 }
1361 panic!("`default_subcommand` names a command that this one does not have")
1362}
1363
1364/// Refuse two subcommands that answer to the same name, aliases included.
1365///
1366/// A derive expansion can validate aliases written on one enum, but aliases may also live on
1367/// the independently expanded `Args` structs its variants wrap. This final, joined-table check
1368/// is where both declarations are visible.
1369pub const fn assert_unique_subcommand_names(subcommands: &[&Command<'_>]) {
1370 const fn form<'a>(cmd: &'a Command<'a>, at: usize) -> Option<&'a str> {
1371 if at == 0 {
1372 Some(cmd.name)
1373 } else if at <= cmd.aliases.len() {
1374 Some(cmd.aliases[at - 1])
1375 } else {
1376 None
1377 }
1378 }
1379
1380 let mut command = 0;
1381 while command < subcommands.len() {
1382 let mut at = 0;
1383 while let Some(name) = form(subcommands[command], at) {
1384 let mut other_command = command;
1385 while other_command < subcommands.len() {
1386 let mut other_at = if other_command == command { at + 1 } else { 0 };
1387 while let Some(other) = form(subcommands[other_command], other_at) {
1388 assert!(
1389 !str_eq(name, other),
1390 "two subcommands answer to the same name, counting aliases"
1391 );
1392 other_at += 1;
1393 }
1394 other_command += 1;
1395 }
1396 at += 1;
1397 }
1398 command += 1;
1399 }
1400}
1401
1402/// Assert that flags scoped to a clause do not reuse a command-level spelling.
1403///
1404/// Derive-generated command tables include scoped flags so the parser can find them. Entries
1405/// that share a key are therefore the same declaration and are ignored here; every other pair
1406/// must have disjoint long, short, alias, and negated spellings.
1407pub const fn assert_clause_flag_spellings(command: &Command<'_>) {
1408 let Some(clause) = command.clause else {
1409 return;
1410 };
1411 let mut scoped_at = 0;
1412 while scoped_at < clause.flags.len() {
1413 let scoped = clause.flags[scoped_at];
1414 let mut command_at = 0;
1415 while command_at < command.flags.len() {
1416 let flag = command.flags[command_at];
1417 if flag.key != scoped.key && const_flag_forms_overlap(flag, scoped) {
1418 panic!("a clause flag spelling conflicts with a command-level flag");
1419 }
1420 command_at += 1;
1421 }
1422 scoped_at += 1;
1423 }
1424}
1425
1426const fn const_flag_forms_overlap(a: &Flag<'_>, b: &Flag<'_>) -> bool {
1427 let mut i = 0;
1428 while i < a.longs.len() {
1429 let mut j = 0;
1430 while j < b.longs.len() {
1431 if const_bytes_eq(a.longs[i].as_bytes(), b.longs[j].as_bytes()) {
1432 return true;
1433 }
1434 j += 1;
1435 }
1436 if let Some(negate) = b.negate {
1437 if const_bytes_eq(a.longs[i].as_bytes(), negate.as_bytes()) {
1438 return true;
1439 }
1440 }
1441 i += 1;
1442 }
1443 i = 0;
1444 while i < a.shorts.len() {
1445 let mut j = 0;
1446 while j < b.shorts.len() {
1447 if a.shorts[i] == b.shorts[j] {
1448 return true;
1449 }
1450 j += 1;
1451 }
1452 i += 1;
1453 }
1454 if let Some(negate) = a.negate {
1455 let mut j = 0;
1456 while j < b.longs.len() {
1457 if const_bytes_eq(negate.as_bytes(), b.longs[j].as_bytes()) {
1458 return true;
1459 }
1460 j += 1;
1461 }
1462 if let Some(other) = b.negate {
1463 if const_bytes_eq(negate.as_bytes(), other.as_bytes()) {
1464 return true;
1465 }
1466 }
1467 }
1468 false
1469}
1470
1471const fn const_bytes_eq(a: &[u8], b: &[u8]) -> bool {
1472 if a.len() != b.len() {
1473 return false;
1474 }
1475 let mut i = 0;
1476 while i < a.len() {
1477 if a[i] != b[i] {
1478 return false;
1479 }
1480 i += 1;
1481 }
1482 true
1483}
1484
1485/// `==` on strings, in a `const fn`.
1486const fn str_eq(a: &str, b: &str) -> bool {
1487 let (a, b) = (a.as_bytes(), b.as_bytes());
1488 if a.len() != b.len() {
1489 return false;
1490 }
1491 let mut i = 0;
1492 while i < a.len() {
1493 if a[i] != b[i] {
1494 return false;
1495 }
1496 i += 1;
1497 }
1498 true
1499}
1500
1501/// Rebuild an [`OsString`] from bytes the parser handed back.
1502///
1503/// This is the reverse of [`OsStr::as_encoded_bytes`], and it is how a `PathBuf` field
1504/// receives a filename the operating system accepts but UTF-8 does not — `/tmp/\xff` stays
1505/// `/tmp/\xff` rather than becoming a *different* filename with `U+FFFD` in it.
1506///
1507/// Where the platform cannot hold those bytes, they are handed back in the `Err` — as
1508/// `String::from_utf8` does — so the caller can name the value in its error without this
1509/// having to copy it for a case that is nearly never taken.
1510///
1511/// # Why this is not `unsafe`, and why it is not lossless everywhere
1512///
1513/// On **Unix** an `OsString` is an arbitrary byte sequence, so the conversion is total and
1514/// uses the safe [`OsStringExt::from_vec`]. Every byte survives, which is the case that
1515/// matters: non-UTF-8 filenames are ordinary there.
1516///
1517/// [`OsStringExt::from_vec`]: std::os::unix::ffi::OsStringExt::from_vec
1518///
1519/// On **Windows** the encoding is WTF-8, where not every byte sequence is valid, and the only
1520/// constructor that accepts one is `OsString::from_encoded_bytes_unchecked` — whose
1521/// precondition this function cannot enforce. It takes a `Vec<u8>` from a safe caller, so
1522/// there is no way to know the bytes came from `as_encoded_bytes` rather than from anywhere
1523/// else, and a safe function with a precondition that can be violated is unsound however
1524/// carefully its callers behave today.
1525///
1526/// So on Windows the bytes go through UTF-8, and one that is not valid UTF-8 is refused
1527/// rather than assumed. What that gives up is a Windows argument containing an unpaired
1528/// surrogate, which is reported instead of accepted; what it buys is that this crate needs no
1529/// `unsafe` at all.
1530pub fn os_string_from_bytes(value: Vec<u8>) -> Result<OsString, Vec<u8>> {
1531 #[cfg(unix)]
1532 {
1533 Ok(std::os::unix::ffi::OsStringExt::from_vec(value))
1534 }
1535 #[cfg(not(unix))]
1536 {
1537 match String::from_utf8(value) {
1538 Ok(text) => Ok(OsString::from(text)),
1539 Err(bad) => Err(bad.into_bytes()),
1540 }
1541 }
1542}
1543
1544/// One [`Error::InvalidValue`], built out of line.
1545///
1546/// Cold and never inlined on purpose: this is the failure path of every value
1547/// conversion in every generated `build`, and inlining it there is what made
1548/// those functions large.
1549#[cold]
1550#[inline(never)]
1551pub(crate) fn invalid_value_error<'t, 'v>(
1552 name: &'t str,
1553 value: String,
1554 reason: String,
1555) -> Error<'t, 'v> {
1556 Error::InvalidValue(Box::new(InvalidValue {
1557 name,
1558 value,
1559 reason,
1560 }))
1561}
1562
1563/// One [`Error::InvalidValue`] for a word that was not UTF-8.
1564///
1565/// The error half of what a generated `build` does per text field: the check stays
1566/// inline at the field, and this — the lossy rendering and the allocations — lives
1567/// here once instead of once per field.
1568#[cold]
1569#[inline(never)]
1570pub fn invalid_utf8_value<'t, 'v>(name: &'t str, bad: std::string::FromUtf8Error) -> Error<'t, 'v> {
1571 invalid_value_error(
1572 name,
1573 String::from_utf8_lossy(bad.as_bytes()).into_owned(),
1574 bad.utf8_error().to_string(),
1575 )
1576}
1577
1578/// One [`Error::InvalidValue`] for a value whose type would not build from it.
1579///
1580/// Takes the reason as `&dyn Display` so one copy serves every `FromStr` error type.
1581#[cold]
1582#[inline(never)]
1583pub fn invalid_parsed_value<'t, 'v>(
1584 name: &'t str,
1585 value: String,
1586 reason: &dyn std::fmt::Display,
1587) -> Error<'t, 'v> {
1588 invalid_value_error(name, value, reason.to_string())
1589}
1590
1591/// One [`Error::InvalidValue`] for a word that is not one of a value enum's choices.
1592#[cold]
1593#[inline(never)]
1594pub fn invalid_choice_value<'t, 'v>(name: &'t str, value: String) -> Error<'t, 'v> {
1595 invalid_value_error(name, value, String::from("not one of the declared values"))
1596}
1597
1598/// One [`Error::InvalidValue`] for bytes the platform cannot hold in a path.
1599#[cold]
1600#[inline(never)]
1601pub fn invalid_os_value<'t, 'v>(name: &'t str, bytes: Vec<u8>) -> Error<'t, 'v> {
1602 invalid_value_error(
1603 name,
1604 String::from_utf8_lossy(&bytes).into_owned(),
1605 "this platform cannot hold these bytes in a path".to_string(),
1606 )
1607}
1608
1609/// Convert every repeated value of one text field, reporting `name` for the
1610/// first that is not UTF-8.
1611///
1612/// The shared body of what a generated `build` does per collecting text field:
1613/// one loop in the binary rather than one per field. Converts element by
1614/// element rather than with `collect` so the error can carry the value that
1615/// failed rather than only that one did.
1616///
1617/// The empty case is answered here rather than in the shared loop, and this is
1618/// what keeps sharing the loop free: a command at mise's scale declares dozens
1619/// of collecting fields and a command line names one or two of them, so most of
1620/// these calls have nothing to convert. Testing that at the field costs a branch;
1621/// reaching the loop to learn it costs the call.
1622#[inline]
1623pub fn utf8_values<'t, 'v>(
1624 values: Vec<Vec<u8>>,
1625 name: &'t str,
1626) -> Result<Vec<String>, Error<'t, 'v>> {
1627 if values.is_empty() {
1628 return Ok(Vec::new());
1629 }
1630 utf8_values_given(values, name)
1631}
1632
1633#[inline(never)]
1634fn utf8_values_given<'t, 'v>(
1635 values: Vec<Vec<u8>>,
1636 name: &'t str,
1637) -> Result<Vec<String>, Error<'t, 'v>> {
1638 let mut out = Vec::with_capacity(values.len());
1639 for value in values {
1640 match String::from_utf8(value) {
1641 Ok(text) => out.push(text),
1642 Err(bad) => return Err(invalid_utf8_value(name, bad)),
1643 }
1644 }
1645 Ok(out)
1646}
1647
1648/// Convert every repeated value of one field through
1649/// [`FromStr`](std::str::FromStr), reporting `name` for the first that fails.
1650///
1651/// Monomorphized once per target type rather than expanded once per field, and
1652/// the empty case is answered at the field for the reason [`utf8_values`] gives.
1653#[inline]
1654pub fn parsed_values<'t, 'v, T>(
1655 values: Vec<Vec<u8>>,
1656 name: &'t str,
1657) -> Result<Vec<T>, Error<'t, 'v>>
1658where
1659 T: std::str::FromStr,
1660 T::Err: std::fmt::Display,
1661{
1662 if values.is_empty() {
1663 return Ok(Vec::new());
1664 }
1665 parsed_values_given(values, name)
1666}
1667
1668#[inline(never)]
1669fn parsed_values_given<'t, 'v, T>(
1670 values: Vec<Vec<u8>>,
1671 name: &'t str,
1672) -> Result<Vec<T>, Error<'t, 'v>>
1673where
1674 T: std::str::FromStr,
1675 T::Err: std::fmt::Display,
1676{
1677 let mut out = Vec::with_capacity(values.len());
1678 for value in values {
1679 let text = match String::from_utf8(value) {
1680 Ok(text) => text,
1681 Err(bad) => return Err(invalid_utf8_value(name, bad)),
1682 };
1683 match text.parse() {
1684 Ok(parsed) => out.push(parsed),
1685 Err(reason) => return Err(invalid_parsed_value(name, text, &reason)),
1686 }
1687 }
1688 Ok(out)
1689}
1690
1691/// Convert every repeated value of one path-like field, reporting `name` for
1692/// the first the platform cannot hold.
1693///
1694/// `T` is what the field collects — [`PathBuf`](std::path::PathBuf) or
1695/// [`OsString`] — so one body serves both. The same platform note as
1696/// [`os_string_from_bytes`] applies: lossless on Unix, partial on Windows. The
1697/// empty case is answered at the field for the reason [`utf8_values`] gives.
1698#[inline]
1699pub fn os_values<'t, 'v, T: From<OsString>>(
1700 values: Vec<Vec<u8>>,
1701 name: &'t str,
1702) -> Result<Vec<T>, Error<'t, 'v>> {
1703 if values.is_empty() {
1704 return Ok(Vec::new());
1705 }
1706 os_values_given(values, name)
1707}
1708
1709#[inline(never)]
1710fn os_values_given<'t, 'v, T: From<OsString>>(
1711 values: Vec<Vec<u8>>,
1712 name: &'t str,
1713) -> Result<Vec<T>, Error<'t, 'v>> {
1714 let mut out = Vec::with_capacity(values.len());
1715 for value in values {
1716 match os_string_from_bytes(value) {
1717 Ok(os) => out.push(T::from(os)),
1718 Err(bytes) => return Err(invalid_os_value(name, bytes)),
1719 }
1720 }
1721 Ok(out)
1722}
1723
1724/// What a generated `apply` stores for a flag that takes one value: the word, or nothing
1725/// when the flag arrived without one.
1726///
1727/// Out of line because the copy is an allocation, a `memcpy`, and the drop of whatever the
1728/// field held before — a few dozen instructions that would otherwise be expanded at every
1729/// value-taking field of every command.
1730#[doc(hidden)]
1731#[inline(never)]
1732pub fn bind_text(slot: &mut Vec<u8>, value: Option<&[u8]>) {
1733 *slot = value.map(<[u8]>::to_vec).unwrap_or_default();
1734}
1735
1736/// [`bind_text`] for an `Option<T>` field, which a flag without a value still fills.
1737#[doc(hidden)]
1738#[inline(never)]
1739pub fn bind_some_text(slot: &mut Option<Vec<u8>>, value: Option<&[u8]>) {
1740 *slot = Some(value.map(<[u8]>::to_vec).unwrap_or_default());
1741}
1742
1743/// [`bind_text`] for an `Option<Option<T>>` field, where a flag given without a value is
1744/// the inner `None`.
1745#[doc(hidden)]
1746#[inline(never)]
1747pub fn bind_optional_text(slot: &mut Option<Vec<u8>>, value: Option<&[u8]>) {
1748 *slot = value.map(<[u8]>::to_vec);
1749}
1750
1751/// [`bind_text`] for a collecting field: one more occurrence, appended.
1752#[doc(hidden)]
1753#[inline(never)]
1754pub fn bind_push_text(slot: &mut Vec<Vec<u8>>, value: Option<&[u8]>) {
1755 slot.push(value.map(<[u8]>::to_vec).unwrap_or_default());
1756}
1757
1758/// [`bind_push_text`] for a field with a value delimiter: every part, appended in order.
1759#[doc(hidden)]
1760#[inline(never)]
1761pub fn bind_push_split(slot: &mut Vec<Vec<u8>>, value: Option<&[u8]>, delimiter: u8) {
1762 for part in value.unwrap_or_default().split(|b| *b == delimiter) {
1763 slot.push(part.to_vec());
1764 }
1765}
1766
1767/// A single binding pass over `argv`.
1768///
1769/// [`Command::default_subcommand_flags`] adds a read-only lookahead before binding
1770/// to choose an implicit command boundary.
1771///
1772/// Created with [`Parser::new`] and driven with [`Parser::next_event`].
1773pub struct Parser<'t, 'a, 'v> {
1774 argv: &'a [&'v OsStr],
1775 /// Index of the next token to read.
1776 pos: usize,
1777 /// The command currently in scope.
1778 cmd: &'t Command<'t>,
1779 /// The canonical root, used to hide root globals omitted by an executable view.
1780 #[cfg(feature = "spec")]
1781 root: &'t Command<'t>,
1782 /// The executable projection being parsed, if argv0 selected one.
1783 #[cfg(feature = "spec")]
1784 view: Option<&'t spec::ViewMeta<'t>>,
1785 /// What an unrecognized flag-like token means in the command currently in scope.
1786 ///
1787 /// Carried rather than looked up, because it is inherited: a command that states
1788 /// nothing keeps what the enclosing one said, and walking back up the ancestors on
1789 /// every unrecognized token would pay for the inheritance at the wrong moment.
1790 unknown_flags: UnknownFlags,
1791 /// Effective inherited trailing-delimiter policy.
1792 dont_delimit_trailing_values: bool,
1793 /// The chain above `cmd`, used to find inherited global flags. Fixed size so
1794 /// that nothing is allocated.
1795 ancestors: [Option<&'t Command<'t>>; MAX_DEPTH],
1796 depth: usize,
1797 /// Bytes left in a short-flag bundle, if one is partly read.
1798 bundle: &'v [u8],
1799 /// The whole token the current bundle came from, so an error raised part way
1800 /// through it can still name what the user typed.
1801 bundle_token: &'v [u8],
1802 /// A variadic flag that is still collecting values.
1803 collecting: Option<&'t Flag<'t>>,
1804 /// Where the command in scope began, as an index into `argv`.
1805 cmd_start: usize,
1806 /// Where each ancestor's own words began, in step with `ancestors`.
1807 starts: [usize; MAX_DEPTH],
1808 /// How many values it has taken, so a bound can stop it.
1809 collected: u32,
1810 /// Which of `cmd.args` is next to fill.
1811 arg_pos: usize,
1812 /// How many words the variadic at `arg_pos` has taken, for the same reason.
1813 arg_taken: u32,
1814 /// Whether any word has been bound to a positional of `cmd`. Once one has,
1815 /// no further word can select a subcommand.
1816 arg_filled: bool,
1817 /// Whether this command has bound any flag or positional. Unlike
1818 /// `arg_filled`, flags count because clap's command policy treats both as
1819 /// arguments that exclude a later subcommand.
1820 command_arg_found: bool,
1821 positional_arg_found: bool,
1822 /// Whether flag interpretation has stopped. A `--` does this, and so does an
1823 /// `automatic` argument taking a value.
1824 flags_stopped: bool,
1825 /// Whether a `--` was actually consumed as a separator.
1826 ///
1827 /// Tracked apart from `flags_stopped` because the two can differ: an
1828 /// `automatic` argument stops flag interpretation without any separator being
1829 /// typed, and a `preserve` argument keeps one as a value rather than
1830 /// consuming it. Callers asking this question want to know what the user
1831 /// wrote, not what state the parser reached.
1832 separator_seen: bool,
1833 /// Whether a root clause separator was consumed, so empty-default routing cannot fire.
1834 clause_separator_seen: bool,
1835 /// Whether the default subcommand has already been taken.
1836 ///
1837 /// Once, per parse: a default subcommand that itself declares one would otherwise
1838 /// descend on every word until the tree ran out.
1839 default_taken: bool,
1840 /// First default-only flag, when lookahead found no explicit sibling.
1841 default_flag_at: Option<usize>,
1842 /// Cursor just after the implicit boundary bundle, whose shorts keep parent ownership.
1843 default_bundle_end: usize,
1844 /// Set once a fatal error has been reported, so iteration stops.
1845 done: bool,
1846 /// Whether declared built-in actions stop parsing with their action error.
1847 ///
1848 /// Invocation parsing does; completion walking only needs the grammar position after the
1849 /// flag, and must not execute an action while inspecting a partial command line.
1850 action_errors: bool,
1851 /// The `argv` range the `help` *word* resolved as a command path, if one was typed.
1852 ///
1853 /// Empty for `--help`, which asks about wherever the parse had got to. For the word, the
1854 /// question is about a command deeper than the parse reached, and only this walk knows
1855 /// which tokens named it: a caller re-scanning `argv` would count a flag's detached value
1856 /// that happens to spell a sibling's name. Two indices rather than the commands
1857 /// themselves, so the parser keeps allocating nothing.
1858 help_span: (usize, usize),
1859 /// An implicit clause boundary follows the positional event that completed it.
1860 pending_clause_boundary: ::core::option::Option<&'t Clause<'t>>,
1861}
1862
1863impl<'t: 'v, 'a, 'v> Parser<'t, 'a, 'v> {
1864 /// Begin parsing `argv` against `root`.
1865 ///
1866 /// `argv` excludes the program name.
1867 pub fn new(root: &'t Command<'t>, argv: &'a [&'v OsStr]) -> Self {
1868 Self::with_action_errors(root, argv, true)
1869 }
1870
1871 /// Begin a non-executing parse for completion walking.
1872 #[cfg(feature = "complete")]
1873 pub(crate) fn for_completion(root: &'t Command<'t>, argv: &'a [&'v OsStr]) -> Self {
1874 Self::with_action_errors(root, argv, false)
1875 }
1876
1877 fn with_action_errors(
1878 root: &'t Command<'t>,
1879 argv: &'a [&'v OsStr],
1880 action_errors: bool,
1881 ) -> Self {
1882 let mut parser = Parser {
1883 argv,
1884 pos: 0,
1885 cmd: root,
1886 #[cfg(feature = "spec")]
1887 root,
1888 #[cfg(feature = "spec")]
1889 view: None,
1890 unknown_flags: match root.unknown_flags {
1891 ::core::option::Option::Some(mode) => mode,
1892 // Nothing above the root to inherit from, so the default stands.
1893 ::core::option::Option::None => UnknownFlags::Value,
1894 },
1895 dont_delimit_trailing_values: root.dont_delimit_trailing_values,
1896 ancestors: [None; MAX_DEPTH],
1897 depth: 0,
1898 bundle: &[],
1899 bundle_token: &[],
1900 collecting: None,
1901 cmd_start: 0,
1902 starts: [0; MAX_DEPTH],
1903 collected: 0,
1904 arg_pos: 0,
1905 arg_taken: 0,
1906 arg_filled: false,
1907 command_arg_found: false,
1908 positional_arg_found: false,
1909 flags_stopped: false,
1910 separator_seen: false,
1911 clause_separator_seen: false,
1912 default_taken: false,
1913 default_flag_at: None,
1914 default_bundle_end: 0,
1915 done: false,
1916 action_errors,
1917 help_span: (0, 0),
1918 pending_clause_boundary: ::core::option::Option::None,
1919 };
1920 if root.default_subcommand_flags {
1921 parser.default_flag_at = parser.default_flag_route();
1922 }
1923 parser
1924 }
1925
1926 /// Find the implicit boundary without binding anything. Parent spellings win while
1927 /// scanning the prefix; after the boundary the ordinary child/global scope applies.
1928 /// Unknown flags stop lookahead because their value arity cannot be guessed.
1929 fn default_flag_route(&self) -> Option<usize> {
1930 let default = self.cmd.default_subcommand?;
1931 let mut at = None;
1932 let mut i = 0;
1933 while let Some(token) = self.argv.get(i).map(bytes) {
1934 let scope = if at.is_some() { default } else { self.cmd };
1935 if token == b"--"
1936 || token == b"-"
1937 || scope.clause.is_some_and(|c| c.separator == Some(token))
1938 {
1939 return at;
1940 }
1941 if !is_flag_like(token) {
1942 // A sibling name before any default-only flag stays on the parent.
1943 // After one, later words are the default command's args (`-u pkg`
1944 // when `pkg` is also a command).
1945 if at.is_none()
1946 && (self.find_subcommand(token).is_some()
1947 || (token == b"help" && !self.cmd.disable_help_subcommand))
1948 {
1949 return None;
1950 }
1951 return at;
1952 }
1953 let mut value_flag = None;
1954 let mut attached = None;
1955 if let Some(body) = token.strip_prefix(b"--") {
1956 let end = body.iter().position(|b| *b == b'=').unwrap_or(body.len());
1957 let name = &body[..end];
1958 let parent = self.find_long(name).or_else(|| self.find_negation(name));
1959 if parent.is_none()
1960 && ((name == b"help" && !self.cmd.disable_help_flag)
1961 || (name == b"version"
1962 && self.cmd.version
1963 && !self.cmd.disable_version_flag))
1964 {
1965 i += 1;
1966 continue;
1967 }
1968 let flag = parent.or_else(|| {
1969 default.flags.iter().copied().find(|f| {
1970 f.longs.iter().any(|l| l.as_bytes() == name)
1971 || f.negate.is_some_and(|n| n.as_bytes() == name)
1972 })
1973 });
1974 let Some(flag) = flag else {
1975 return at;
1976 };
1977 if parent.is_none() {
1978 at.get_or_insert(i);
1979 }
1980 if flag.takes_value && flag.negate.is_none_or(|n| n.as_bytes() != name) {
1981 value_flag = Some(flag);
1982 attached = (end < body.len()).then(|| &body[end + 1..]);
1983 }
1984 } else {
1985 for (offset, byte) in token[1..].iter().enumerate() {
1986 let parent = self.find_short(*byte);
1987 let Some(flag) = parent.or_else(|| {
1988 default
1989 .flags
1990 .iter()
1991 .copied()
1992 .find(|f| f.shorts.contains(byte))
1993 }) else {
1994 return at;
1995 };
1996 if parent.is_none() {
1997 at.get_or_insert(i);
1998 }
1999 if flag.takes_value {
2000 value_flag = Some(flag);
2001 let rest = &token[offset + 2..];
2002 attached =
2003 (!rest.is_empty()).then_some(rest.strip_prefix(b"=").unwrap_or(rest));
2004 break;
2005 }
2006 }
2007 }
2008 i += 1;
2009 if let Some(flag) = value_flag {
2010 let scope = if at.is_some() { default } else { self.cmd };
2011 let is_separator =
2012 |next: &[u8]| scope.clause.is_some_and(|c| c.separator == Some(next));
2013 let first = if let Some(value) = attached {
2014 Some(value)
2015 } else if !flag.require_equals {
2016 self.argv
2017 .get(i)
2018 .map(bytes)
2019 .filter(|next| {
2020 !is_separator(next)
2021 && (flag.allow_hyphen_values
2022 || !is_flag_like(next)
2023 || (flag.allow_negative_numbers && is_negative_number(next)))
2024 })
2025 .inspect(|_| i += 1)
2026 } else {
2027 None
2028 };
2029 if first.is_none() && !flag.value_optional && flag.default_missing.is_none() {
2030 return at;
2031 }
2032 if flag.variadic {
2033 let mut count = first.map_or(0, |v| values_in(v, flag.delimiter));
2034 while flag.var_max.is_none_or(|max| count < max) {
2035 let Some(next) = self.argv.get(i).map(bytes) else {
2036 break;
2037 };
2038 if is_separator(next) {
2039 break;
2040 }
2041 if next == b"--" || flag.value_terminator.is_some_and(|end| end == next) {
2042 if next != b"--" {
2043 i += 1;
2044 }
2045 break;
2046 }
2047 if is_flag_like(next)
2048 && !(flag.allow_negative_numbers && is_negative_number(next))
2049 {
2050 break;
2051 }
2052 if self.cmd.subcommand_precedence_over_arg
2053 && self.find_subcommand(next).is_some()
2054 {
2055 break;
2056 }
2057 count += values_in(next, flag.delimiter);
2058 i += 1;
2059 }
2060 }
2061 }
2062 }
2063 at
2064 }
2065
2066 /// Restrict inherited root globals to those carried by an executable view.
2067 #[cfg(feature = "spec")]
2068 pub fn with_view(mut self, view: &'t spec::ViewMeta<'t>) -> Self {
2069 self.view = Some(view);
2070 self
2071 }
2072
2073 /// The command in scope: the root, or the deepest subcommand selected so far.
2074 pub fn command(&self) -> &'t Command<'t> {
2075 self.cmd
2076 }
2077
2078 /// Whether a `--` was consumed as a separator.
2079 ///
2080 /// False when flag interpretation stopped for another reason, such as an
2081 /// `automatic` argument taking a value, and false for a `--` that a
2082 /// `preserve` argument kept as a value.
2083 pub fn double_dash_seen(&self) -> bool {
2084 self.separator_seen
2085 }
2086
2087 /// Every command entered so far, and where each one's own words begin.
2088 ///
2089 /// The ancestors are already kept for flag scoping; this is the same chain with the offsets,
2090 /// which is what lets a completion hand a callback the words of *its* command rather than of
2091 /// the deepest one — a global flag is declared on an ancestor.
2092 pub fn command_path(&self) -> Vec<(&'t Command<'t>, usize)> {
2093 let mut out = Vec::with_capacity(self.depth + 1);
2094 for (i, ancestor) in self.ancestors[..self.depth].iter().enumerate() {
2095 if let Some(cmd) = ancestor {
2096 // An ancestor's own words start where the one before it descended, and the
2097 // root's start at the beginning.
2098 out.push((*cmd, self.starts[i]));
2099 }
2100 }
2101 out.push((self.cmd, self.cmd_start));
2102 out
2103 }
2104
2105 /// The `argv` range the `help` word resolved as a command path.
2106 ///
2107 /// Empty unless the word was typed. Every token in it named a subcommand of the one before
2108 /// it — the parser resolved them itself, so nothing here is a flag or a flag's value.
2109 pub fn help_span(&self) -> (usize, usize) {
2110 self.help_span
2111 }
2112
2113 /// Where the command in scope began: the index in `argv` just after its name, or at the
2114 /// unmatched word routed into a default subcommand.
2115 ///
2116 /// `argv[command_start()..]` is what that command was given, which is what a completion
2117 /// callback needs to be handed its own command's half-parsed struct rather than the root's.
2118 pub fn command_start(&self) -> usize {
2119 self.cmd_start
2120 }
2121
2122 /// Whether flag interpretation has stopped, for any reason.
2123 ///
2124 /// Wider than [`double_dash_seen`](Self::double_dash_seen), and the question completion
2125 /// asks: past a separator *or* past the first value of an `automatic` argument, a
2126 /// dash-prefixed word is a value, so there is no flag there to offer.
2127 pub fn flags_stopped(&self) -> bool {
2128 self.flags_stopped
2129 }
2130
2131 /// A variadic flag that is still claiming words.
2132 ///
2133 /// Asked *between* events, because the answer is gone by the end: the call that finds argv
2134 /// exhausted is the one that clears it. A completion needs it — the next word after
2135 /// `--tools a ⌶` is another tool, not the positional that follows.
2136 pub fn collecting(&self) -> Option<&'t Flag<'t>> {
2137 self.collecting
2138 }
2139
2140 /// The positional the next word would fill, if there is one left.
2141 ///
2142 /// A variadic stays here until it reaches its bound, which is what makes it the answer to
2143 /// "what could go where the cursor is" as many times as it can be filled.
2144 pub fn pending_arg(&self) -> Option<&'t Arg<'t>> {
2145 self.next_arg()
2146 }
2147
2148 /// Flags a word here could name: this command's own, then any ancestor's globals.
2149 ///
2150 /// The same set the parser itself would look in, so what is offered and what is accepted
2151 /// cannot disagree — including the shadowing rule, where a subcommand redeclaring an
2152 /// inherited name hides it.
2153 pub fn flags_in_scope(&self) -> impl Iterator<Item = &'t Flag<'t>> + '_ {
2154 self.in_scope()
2155 }
2156
2157 /// Read the next event.
2158 ///
2159 /// Returns `None` when `argv` is exhausted. An `Err` is terminal: the parse
2160 /// stops there, since continuing past a token that could not be understood
2161 /// would only produce bindings derived from a guess. Events already yielded
2162 /// before an error are therefore not a partial result to be used — a caller
2163 /// that assigned them into fields should discard the whole attempt.
2164 ///
2165 /// One case is stronger than that, because the grammar demands it: a short
2166 /// bundle containing an unrecognized letter yields the error *instead of*, not
2167 /// after, the letters that did match.
2168 #[allow(clippy::should_implement_trait)] // not an Iterator: items borrow from self's tables
2169 pub fn next_event(&mut self) -> Option<Result<Event<'t, 'a, 'v>, Error<'t, 'v>>> {
2170 if self.done {
2171 return None;
2172 }
2173 let event = self.step();
2174 if matches!(event, Some(Ok(Event::Flag { .. } | Event::Arg { .. }))) {
2175 self.command_arg_found = true;
2176 }
2177 if matches!(event, Some(Ok(Event::Arg { .. }))) {
2178 self.positional_arg_found = true;
2179 }
2180 if event.is_none()
2181 && self.depth == 0
2182 && self.cmd.default_subcommand_on_empty
2183 && !self.default_taken
2184 && !self.positional_arg_found
2185 && !self.separator_seen
2186 && !self.clause_separator_seen
2187 {
2188 if let Some(default) = self.cmd.default_subcommand {
2189 if self.cmd.args_conflicts_with_subcommands && self.command_arg_found {
2190 self.done = true;
2191 return Some(Err(Error::SubcommandConflict {
2192 subcommand: default,
2193 }));
2194 }
2195 self.default_taken = true;
2196 match self.descend(default) {
2197 Ok(()) => return Some(Ok(Event::Command(default))),
2198 Err(error) => {
2199 self.done = true;
2200 return Some(Err(error));
2201 }
2202 }
2203 }
2204 }
2205 if let Some(Err(_)) = event {
2206 self.done = true;
2207 }
2208 event
2209 }
2210
2211 /// Count parent flags in the shared boundary token before entering the child.
2212 fn default_bundle_has_parent_flag(&self, default: &Command<'_>) -> bool {
2213 let Some(token) = self.argv.get(self.pos).map(bytes) else {
2214 return false;
2215 };
2216 if !token.starts_with(b"-") || token.starts_with(b"--") {
2217 return false;
2218 }
2219 for byte in &token[1..] {
2220 if self.find_short(*byte).is_some() {
2221 return true;
2222 }
2223 match default.flags.iter().find(|f| f.shorts.contains(byte)) {
2224 Some(flag) if !flag.takes_value => {}
2225 _ => break,
2226 }
2227 }
2228 false
2229 }
2230
2231 fn step(&mut self) -> Option<Result<Event<'t, 'a, 'v>, Error<'t, 'v>>> {
2232 if self.default_flag_at == Some(self.pos) && self.bundle.is_empty() {
2233 self.default_flag_at = None;
2234 if let Some(default) = self.cmd.default_subcommand {
2235 if self.cmd.args_conflicts_with_subcommands
2236 && (self.command_arg_found || self.default_bundle_has_parent_flag(default))
2237 {
2238 return Some(Err(Error::SubcommandConflict {
2239 subcommand: default,
2240 }));
2241 }
2242 if self.argv.get(self.pos).is_some_and(|word| {
2243 let token = bytes(word);
2244 token.starts_with(b"-") && !token.starts_with(b"--")
2245 }) {
2246 self.default_bundle_end = self.pos + 1;
2247 }
2248 self.default_taken = true;
2249 return Some(self.descend(default).map(|()| Event::Command(default)));
2250 }
2251 }
2252 if let Some(clause) = self.pending_clause_boundary.take() {
2253 self.arg_pos = 0;
2254 self.arg_taken = 0;
2255 self.arg_filled = false;
2256 self.collecting = None;
2257 self.flags_stopped = false;
2258 return Some(Ok(Event::ClauseSeparator { clause: *clause }));
2259 }
2260 // A partly-read short bundle takes priority: its remaining bytes are
2261 // still part of the token being processed.
2262 if !self.bundle.is_empty() {
2263 return Some(self.short_flag());
2264 }
2265
2266 if self.cmd.subcommand_precedence_over_arg && !self.flags_stopped {
2267 if let Some(token) = self.argv.get(self.pos).map(bytes) {
2268 if let Some(sub) = self.find_subcommand(token) {
2269 if self.cmd.args_conflicts_with_subcommands && self.command_arg_found {
2270 return Some(Err(Error::SubcommandConflict { subcommand: sub }));
2271 }
2272 self.pos += 1;
2273 return Some(self.descend(sub).map(|()| Event::Command(sub)));
2274 }
2275 }
2276 }
2277
2278 // A variadic flag keeps claiming tokens until one of them could be
2279 // something else.
2280 if let Some(flag) = self.collecting {
2281 match self.argv.get(self.pos) {
2282 Some(next)
2283 if flag
2284 .value_terminator
2285 .is_some_and(|terminator| bytes(next) == terminator) =>
2286 {
2287 self.pos += 1;
2288 self.collecting = None;
2289 return self.step();
2290 }
2291 Some(next)
2292 if (!is_flag_like(bytes(next))
2293 || (flag.allow_negative_numbers && is_negative_number(bytes(next))))
2294 && self.clause_separator(bytes(next)).is_none()
2295 && bytes(next) != b"--" =>
2296 {
2297 self.pos += 1;
2298 self.collected += values_in(bytes(next), flag.delimiter);
2299 // Same rule as a positional: a bounded occurrence takes that many and
2300 // leaves the rest to whatever follows.
2301 if flag.var_max.is_some_and(|max| self.collected >= max) {
2302 self.collecting = None;
2303 }
2304 // Stopping is only the same as staying within the bound while one word
2305 // is one value. A delimited word can carry the occurrence past it in a
2306 // single step, and that word cannot be split between two owners, so the
2307 // overshoot is an error rather than a place to stop.
2308 if let Some(max) = flag.var_max.filter(|max| self.collected > *max) {
2309 return Some(Err(Error::VarTooMany {
2310 name: flag.name,
2311 max: max as usize,
2312 got: self.collected as usize,
2313 }));
2314 }
2315 return Some(Ok(Event::Flag {
2316 flag,
2317 value: Some(bytes(next)),
2318 negated: false,
2319 }));
2320 }
2321 // A token that could be something else ends the run — but the *end of argv*
2322 // decides nothing. Clearing there threw away the answer to "would the next
2323 // word be claimed?", which is the question a completion asks and no parse
2324 // ever does: once argv is exhausted there are no more events either way.
2325 Some(_) => self.collecting = None,
2326 None => {}
2327 }
2328 }
2329
2330 let token = bytes(self.argv.get(self.pos)?);
2331 self.pos += 1;
2332
2333 // A clause separator remains syntax after an automatic positional stopped flags.
2334 // Only an explicit `--` protects a literal separator.
2335 if let Some(clause) = self.clause_separator(token) {
2336 self.arg_pos = 0;
2337 self.arg_taken = 0;
2338 self.arg_filled = false;
2339 self.collecting = None;
2340 self.flags_stopped = false;
2341 if self.depth == 0 {
2342 self.clause_separator_seen = true;
2343 }
2344 return Some(Ok(Event::ClauseSeparator { clause: *clause }));
2345 }
2346
2347 // An automatic trailing argument stops flag interpretation without consuming an
2348 // explicit separator. A later `--` must still unlock a required trailing argument
2349 // (clap's `last`), while a separator already consumed makes every later `--` data.
2350 if self.flags_stopped && (token != b"--" || self.separator_seen) {
2351 return Some(self.word(token));
2352 }
2353
2354 if token == b"--" {
2355 // `preserve` wants the separator itself as a value, so ask the
2356 // argument that would receive it before treating it as syntax.
2357 if self
2358 .next_arg()
2359 .is_some_and(|a| a.double_dash == DoubleDash::Preserve)
2360 {
2361 return Some(self.word(token));
2362 }
2363 self.flags_stopped = true;
2364 self.separator_seen = true;
2365 // An explicit separator unlocks any argument that required one, even
2366 // if earlier arguments are still unfilled.
2367 if let Some(idx) = self.current_args()[self.arg_pos..]
2368 .iter()
2369 .position(|a| a.double_dash == DoubleDash::Required)
2370 {
2371 // The count belongs to the argument at `arg_pos`, so jumping past it has
2372 // to leave the count behind: a bounded variadic before the separator would
2373 // otherwise lend its total to the argument after it, which then stops
2374 // early or at once.
2375 self.arg_pos += idx;
2376 self.arg_taken = 0;
2377 }
2378 return self.step();
2379 }
2380
2381 if self.arg_taken > 0
2382 && self.next_arg().is_some_and(|arg| {
2383 arg.value_terminator
2384 .is_some_and(|terminator| token == terminator)
2385 })
2386 {
2387 self.advance_arg();
2388 return self.step();
2389 }
2390
2391 // An exact declared short outranks the numeric shape. This keeps ordinary negative
2392 // numbers available as values while allowing clap-compatible spellings such as fd's
2393 // `-0` / `--print0` switch.
2394 let declared_numeric_short = matches!(token, [b'-', short]
2395 if short.is_ascii_digit() && self.find_short(*short).is_some());
2396
2397 if !declared_numeric_short
2398 && is_negative_number(token)
2399 && self
2400 .next_arg()
2401 .is_some_and(|arg| arg.allow_negative_numbers)
2402 {
2403 return Some(self.word(token));
2404 }
2405
2406 if !declared_numeric_short
2407 && is_negative_number(token)
2408 && self.cmd.external_subcommand
2409 && !self.arg_filled
2410 {
2411 return Some(self.word(token));
2412 }
2413
2414 if is_flag_like(token) {
2415 if token.starts_with(b"--") {
2416 return Some(self.long_flag(token));
2417 }
2418 // Check the whole bundle before emitting anything from it. Events go
2419 // out one at a time, so discovering an unknown letter half way
2420 // through would mean the earlier letters had already been applied —
2421 // and the grammar rejects the entire token, not the tail of it.
2422 match self.check_bundle(token) {
2423 Ok(()) => {}
2424 // Unrecognized, so it is a word unless this command wants it refused.
2425 Err(e) if self.unknown_flags == UnknownFlags::Error => {
2426 return Some(Err(e));
2427 }
2428 Err(_) => return Some(self.word(token)),
2429 }
2430 self.bundle = &token[1..];
2431 self.bundle_token = token;
2432 return Some(self.short_flag());
2433 }
2434
2435 Some(self.word(token))
2436 }
2437
2438 fn long_flag(&mut self, token: &'v [u8]) -> Result<Event<'t, 'a, 'v>, Error<'t, 'v>> {
2439 let body = &token[2..];
2440 let (name, attached) = match body.iter().position(|&b| b == b'=') {
2441 Some(i) => (&body[..i], Some(&body[i + 1..])),
2442 None => (body, None),
2443 };
2444
2445 if let Some(flag) = self.find_long(name) {
2446 let value = if flag.takes_value {
2447 match attached {
2448 Some(v) => Some(v),
2449 None => self.take_detached_value(flag)?,
2450 }
2451 } else if flag.bool_value {
2452 validate_bool_value(flag, attached)?
2453 } else {
2454 None
2455 };
2456 if flag.variadic {
2457 if let Some(value) = value {
2458 self.start_collecting(flag, value)?;
2459 }
2460 }
2461 if let Some(error) = self.flag_action(flag, true) {
2462 return Err(error);
2463 }
2464 return Ok(Event::Flag {
2465 flag,
2466 value,
2467 negated: false,
2468 });
2469 }
2470
2471 if let Some(flag) = self.find_negation(name) {
2472 return Ok(Event::Flag {
2473 flag,
2474 value: if flag.bool_value {
2475 validate_bool_value(flag, attached)?
2476 } else {
2477 None
2478 },
2479 negated: true,
2480 });
2481 }
2482
2483 // Where the CLI declared a version, `--version` answers with it — asked after the
2484 // command's own flags, so a CLI declaring its own keeps it.
2485 let version_command = self.version_command();
2486 if name == b"version" && version_command.version && !version_command.disable_version_flag {
2487 return Ok(Event::Flag {
2488 flag: &VERSION_LONG,
2489 value: None,
2490 negated: false,
2491 });
2492 }
2493
2494 // Every CLI answers to `--help`, and none of them declares it. Asked *after* the
2495 // command's own flags, so a CLI that declares its own `--help` keeps it.
2496 if name == b"help" && !self.cmd.disable_help_flag {
2497 return Ok(Event::Flag {
2498 flag: &HELP_LONG,
2499 value: None,
2500 negated: false,
2501 });
2502 }
2503
2504 if self.unknown_flags == UnknownFlags::Error {
2505 return Err(Error::UnknownFlag { token });
2506 }
2507 // Not a flag here, so it is a word like any other.
2508 self.word(token)
2509 }
2510
2511 /// Walk a short-flag token without binding anything, to find out whether all
2512 /// of it is recognized.
2513 ///
2514 /// Scanning stops at the first letter whose flag takes a value, because
2515 /// everything after it is that value rather than more letters.
2516 fn check_bundle(&self, token: &'v [u8]) -> Result<(), Error<'t, 'v>> {
2517 let mut rest = &token[1..];
2518 while let Some((&byte, tail)) = rest.split_first() {
2519 match self.find_short(byte) {
2520 None => return Err(Error::UnknownFlag { token }),
2521 Some(flag) if flag.takes_value => return Ok(()),
2522 Some(_) => rest = tail,
2523 }
2524 }
2525 Ok(())
2526 }
2527
2528 fn short_flag(&mut self) -> Result<Event<'t, 'a, 'v>, Error<'t, 'v>> {
2529 let byte = self.bundle[0];
2530 let rest = &self.bundle[1..];
2531
2532 let Some(flag) = self.find_short(byte) else {
2533 // check_bundle already rejected any token containing an unrecognized
2534 // letter, so this is unreachable — but a parser should report rather
2535 // than panic if that ever stops being true.
2536 self.bundle = &[];
2537 return Err(Error::UnknownFlag {
2538 token: self.bundle_token,
2539 });
2540 };
2541
2542 if !flag.takes_value {
2543 self.bundle = rest;
2544 if let Some(error) = self.flag_action(flag, false) {
2545 self.bundle = &[];
2546 return Err(error);
2547 }
2548 return Ok(Event::Flag {
2549 flag,
2550 value: None,
2551 negated: false,
2552 });
2553 }
2554
2555 // A value-taking short ends the token: everything after it is the value,
2556 // less one separating `=`.
2557 self.bundle = &[];
2558 let value = if rest.is_empty() {
2559 self.take_detached_value(flag)?
2560 } else if rest[0] == b'=' {
2561 Some(&rest[1..])
2562 } else {
2563 Some(rest)
2564 };
2565 if flag.variadic {
2566 if let Some(value) = value {
2567 self.start_collecting(flag, value)?;
2568 }
2569 }
2570 if let Some(error) = self.flag_action(flag, false) {
2571 return Err(error);
2572 }
2573 Ok(Event::Flag {
2574 flag,
2575 value,
2576 negated: false,
2577 })
2578 }
2579
2580 fn flag_action(&self, flag: &'t Flag<'t>, long_spelling: bool) -> Option<Error<'t, 'v>> {
2581 if matches!(
2582 flag.key,
2583 HELP_LONG_KEY | HELP_SHORT_KEY | VERSION_LONG_KEY | VERSION_SHORT_KEY
2584 ) || !self.action_errors
2585 {
2586 return None;
2587 }
2588 match flag.action {
2589 ArgAction::Set => None,
2590 ArgAction::Help => Some(Error::Help {
2591 cmd: self.cmd,
2592 long: long_spelling,
2593 }),
2594 ArgAction::HelpShort => Some(Error::Help {
2595 cmd: self.cmd,
2596 long: false,
2597 }),
2598 ArgAction::HelpLong => Some(Error::Help {
2599 cmd: self.cmd,
2600 long: true,
2601 }),
2602 ArgAction::HelpAll => Some(Error::HelpAll { cmd: self.cmd }),
2603 ArgAction::Version => Some(Error::Version {
2604 long: long_spelling,
2605 }),
2606 }
2607 }
2608
2609 /// Take the following token as a flag's value.
2610 ///
2611 /// Refuses a flag-like token unless [`Flag::allow_hyphen_values`] is set:
2612 /// `--jobs --force` is far more likely a forgotten value than a deliberate
2613 /// one, and the attached form is available for the deliberate case. Declared,
2614 /// the next token is taken whatever it looks like, including `--`.
2615 fn take_detached_value(
2616 &mut self,
2617 flag: &'t Flag<'t>,
2618 ) -> Result<Option<&'v [u8]>, Error<'t, 'v>> {
2619 if flag.require_equals {
2620 return self.missing_or_default(flag);
2621 }
2622 match self.argv.get(self.pos) {
2623 Some(next)
2624 if self.clause_separator(bytes(next)).is_none()
2625 && (flag.allow_hyphen_values
2626 || !is_flag_like(bytes(next))
2627 || (flag.allow_negative_numbers && is_negative_number(bytes(next)))) =>
2628 {
2629 self.pos += 1;
2630 Ok(Some(bytes(next)))
2631 }
2632 _ => self.missing_or_default(flag),
2633 }
2634 }
2635
2636 fn missing_or_default(&self, flag: &'t Flag<'t>) -> Result<Option<&'v [u8]>, Error<'t, 'v>> {
2637 match flag.default_missing {
2638 Some(value) => Ok(Some(value)),
2639 None if flag.value_optional => Ok(None),
2640 None => Err(Error::MissingFlagValue { flag }),
2641 }
2642 }
2643
2644 fn clause_separator(&self, token: &[u8]) -> Option<&'t Clause<'t>> {
2645 (!self.separator_seen)
2646 .then_some(self.cmd.clause)
2647 .flatten()
2648 .filter(|clause| clause.separator.is_some_and(|separator| token == separator))
2649 }
2650
2651 fn word(&mut self, token: &'v [u8]) -> Result<Event<'t, 'a, 'v>, Error<'t, 'v>> {
2652 // Subcommands are only matched where descent is still possible: once a
2653 // positional of this command has taken a word, a later word that happens
2654 // to equal a subcommand name is just a value.
2655 if !self.arg_filled && !self.flags_stopped {
2656 if let Some(sub) = self.find_subcommand(token) {
2657 if self.cmd.args_conflicts_with_subcommands && self.command_arg_found {
2658 return Err(Error::SubcommandConflict { subcommand: sub });
2659 }
2660 self.descend(sub)?;
2661 return Ok(Event::Command(sub));
2662 }
2663
2664 // `ex help config ls` — the line every page with a Commands section has printed
2665 // all along ("help Print this message or the help of the given subcommand(s)"),
2666 // and which until now did nothing. The page is what decides the condition here:
2667 // it prints that line where there are subcommands, so that is where the word is
2668 // answered, and to a leaf `help` is a word like any other.
2669 //
2670 // Asked *after* the subcommand lookup, so a CLI that declares a `help` of its own
2671 // keeps it — the same rule the two help flags follow.
2672 //
2673 // The words after it name a command, resolved here rather than descended into:
2674 // descending would bind them, and they are a question rather than an invocation.
2675 if token == b"help"
2676 && !self.cmd.disable_help_subcommand
2677 && !self.cmd.subcommands.is_empty()
2678 {
2679 let mut cmd = self.cmd;
2680 let from = self.pos;
2681 while let Some(next) = self.argv.get(self.pos) {
2682 let Some(sub) = find_named(cmd, bytes(next)) else {
2683 break;
2684 };
2685 cmd = sub;
2686 self.pos += 1;
2687 }
2688 // Kept for `help::route_to`: which mount was asked about is not recoverable
2689 // from `cmd`, since two mounts of one `Subcommands` type are one address.
2690 self.help_span = (from, self.pos);
2691 // The long form, as `ex config --help` gives: someone who typed a whole word to
2692 // ask for help wants the fuller answer.
2693 return Err(Error::Help { cmd, long: true });
2694 }
2695
2696 // A word that names no subcommand goes to the default one, if there is one.
2697 //
2698 // Only a word, though. A dash-prefixed token that named no flag arrives here as a
2699 // value — that is what `unknown_flags = value` means — and it was never a
2700 // candidate to *select* anything, so it binds where it was typed. usage-lib stops
2701 // looking for subcommands at an unrecognised flag for the same reason. `--` is
2702 // excluded on the same grounds: it reaches this function only when a `preserve`
2703 // argument wants it as a value.
2704 //
2705 // The token is *not* consumed: the cursor steps back so the next event reads it
2706 // again, now against the command just descended into. That is what lets it be a
2707 // subcommand of the default (`mise build` where `build` is a task the mount
2708 // added) as easily as an argument of it, without this function having to decide
2709 // which — and without yielding two events for one word.
2710 if let Some(default) = self.cmd.default_subcommand {
2711 // `-` joins `--` in being excluded, and for the reason already written above:
2712 // a value was never a candidate to *select* anything. `is_flag_like` calls a
2713 // lone `-` a value — conventionally stdin — so it passed this guard and
2714 // descended, where mise's `run` has no positional and the parse failed.
2715 // usage-lib and clap both bind it to the root's own `[TASK]` instead.
2716 let default_accepts_negative = is_negative_number(token)
2717 && default
2718 .args
2719 .first()
2720 .is_some_and(|arg| arg.allow_negative_numbers);
2721 if !self.default_taken
2722 && (!is_flag_like(token) || default_accepts_negative)
2723 && token != b"--"
2724 && token != b"-"
2725 {
2726 self.default_taken = true;
2727 self.descend(default)?;
2728 self.pos -= 1;
2729 // Unlike an explicitly named command, the default command receives the
2730 // word that caused descent. Keep its argv boundary at that word so
2731 // command-level policies and completion callbacks see the same input the
2732 // command parser is about to re-read.
2733 self.cmd_start = self.pos;
2734 return Ok(Event::Command(default));
2735 }
2736 }
2737
2738 // Known subcommands and the default route keep precedence. A sigil positional
2739 // then claims its classified word before an external-subcommand catch-all can.
2740 if let Some((arg, sigil)) = self.match_sigil_arg(token) {
2741 if token.len() == sigil.len() {
2742 return Err(invalid_value_error(
2743 arg.name,
2744 as_str(token).unwrap_or_default().to_string(),
2745 format!(
2746 "expected a value after sigil {:?}",
2747 as_str(sigil).unwrap_or_default()
2748 ),
2749 ));
2750 }
2751 return Ok(Event::Arg {
2752 arg,
2753 value: &token[sigil.len()..],
2754 delimit: true,
2755 });
2756 }
2757
2758 // An unmatched word that names no subcommand is forwarded as an external
2759 // command: this word, then every token after it, including flags. Known
2760 // subcommands already won above, and a default_subcommand already caught.
2761 if self.cmd.external_subcommand
2762 && (!is_flag_like(token) || is_negative_number(token))
2763 && token != b"--"
2764 && token != b"-"
2765 {
2766 let from = self.pos - 1;
2767 self.pos = self.argv.len();
2768 return Ok(Event::External {
2769 values: &self.argv[from..],
2770 });
2771 }
2772 }
2773
2774 if self.arg_filled && !self.flags_stopped {
2775 if let Some((arg, sigil)) = self.match_sigil_arg(token) {
2776 if token.len() == sigil.len() {
2777 return Err(invalid_value_error(
2778 arg.name,
2779 as_str(token).unwrap_or_default().to_string(),
2780 format!(
2781 "expected a value after sigil {:?}",
2782 as_str(sigil).unwrap_or_default()
2783 ),
2784 ));
2785 }
2786 return Ok(Event::Arg {
2787 arg,
2788 value: &token[sigil.len()..],
2789 delimit: true,
2790 });
2791 }
2792 }
2793
2794 self.skip_sigil_args();
2795 self.reserve_for_required_positionals();
2796 let Some(arg) = self.next_arg() else {
2797 return Err(Error::UnexpectedArg { token });
2798 };
2799
2800 if arg.double_dash == DoubleDash::Required && !self.separator_seen {
2801 return Err(Error::ArgRequiresDoubleDash { arg });
2802 }
2803
2804 self.arg_filled = true;
2805 // An `automatic` argument stops flag interpretation from here on, as
2806 // though the caller had typed the separator themselves.
2807 let trailing_value = self.separator_seen || arg.double_dash == DoubleDash::Automatic;
2808 let delimit = !(self.dont_delimit_trailing_values && trailing_value);
2809 if arg.double_dash == DoubleDash::Automatic {
2810 self.flags_stopped = true;
2811 }
2812 // A variadic keeps taking values, so the cursor stays put — until it reaches its
2813 // bound, at which point the words after it belong to whatever comes next. That is
2814 // what makes `[a]… [b]` expressible at all.
2815 if arg.var {
2816 self.arg_taken += values_in(token, delimit.then_some(arg.delimiter).flatten());
2817 // Before advancing, which resets the count: as with a variadic flag, reaching
2818 // the bound and passing it are the same event once a word can carry several
2819 // values, and only the second is a mistake.
2820 if let Some(max) = arg.var_max.filter(|max| self.arg_taken > *max) {
2821 return Err(Error::VarTooMany {
2822 name: arg.name,
2823 max: max as usize,
2824 got: self.arg_taken as usize,
2825 });
2826 }
2827 if arg.var_max.is_some_and(|max| self.arg_taken >= max) {
2828 self.advance_arg();
2829 }
2830 } else {
2831 self.advance_arg();
2832 }
2833 if self
2834 .cmd
2835 .clause
2836 .is_some_and(|clause| clause.separator.is_none() && self.next_arg().is_none())
2837 {
2838 self.pending_clause_boundary = self.cmd.clause;
2839 }
2840 Ok(Event::Arg {
2841 arg,
2842 value: token,
2843 delimit,
2844 })
2845 }
2846
2847 fn descend(&mut self, sub: &'t Command<'t>) -> Result<(), Error<'t, 'v>> {
2848 if self.depth >= MAX_DEPTH {
2849 return Err(Error::TooDeep);
2850 }
2851 self.ancestors[self.depth] = Some(self.cmd);
2852 self.starts[self.depth] = self.cmd_start;
2853 self.depth += 1;
2854 self.cmd = sub;
2855 // Only a command that says something changes it, which is what inheriting means.
2856 if let ::core::option::Option::Some(mode) = sub.unknown_flags {
2857 self.unknown_flags = mode;
2858 }
2859 self.dont_delimit_trailing_values |= sub.dont_delimit_trailing_values;
2860 // Where this command's own words start, which is what lets a completion hand a callback
2861 // the half-parsed struct of the command it was declared on rather than of the root.
2862 self.cmd_start = self.pos;
2863 self.arg_pos = 0;
2864 self.arg_taken = 0;
2865 self.arg_filled = false;
2866 self.command_arg_found = false;
2867 Ok(())
2868 }
2869
2870 /// Move to the next positional, forgetting what the last one took.
2871 fn advance_arg(&mut self) {
2872 self.skip_sigil_args();
2873 self.arg_pos += 1;
2874 self.arg_taken = 0;
2875 self.skip_sigil_args();
2876 }
2877
2878 /// A variadic flag occurrence begins, counting from zero.
2879 ///
2880 /// The value it was given on the same token counts, which is why this starts at what
2881 /// that value holds: `--include a b` with `var_max=2` takes `a` and `b`, not three
2882 /// words — and `--include a,b` has already taken both on the one token.
2883 fn start_collecting(&mut self, flag: &'t Flag<'t>, first: &[u8]) -> Result<(), Error<'t, 'v>> {
2884 self.collected = values_in(first, flag.delimiter);
2885 if let Some(max) = flag.var_max.filter(|max| self.collected > *max) {
2886 return Err(Error::VarTooMany {
2887 name: flag.name,
2888 max: max as usize,
2889 got: self.collected as usize,
2890 });
2891 }
2892 self.collecting = if flag.var_max.is_some_and(|max| self.collected >= max) {
2893 None
2894 } else {
2895 Some(flag)
2896 };
2897 Ok(())
2898 }
2899
2900 fn next_arg(&self) -> Option<&'t Arg<'t>> {
2901 self.current_args()[self.arg_pos..]
2902 .iter()
2903 .find(|arg| arg.sigil.is_none())
2904 .copied()
2905 }
2906
2907 fn skip_sigil_args(&mut self) {
2908 while self
2909 .current_args()
2910 .get(self.arg_pos)
2911 .is_some_and(|arg| arg.sigil.is_some())
2912 {
2913 self.arg_pos += 1;
2914 }
2915 }
2916
2917 fn match_sigil_arg(&self, token: &[u8]) -> Option<(&'t Arg<'t>, &'t [u8])> {
2918 if self.flags_stopped {
2919 return None;
2920 }
2921 let own = self.current_args().iter().copied();
2922 let inherited = self.ancestors[..self.depth]
2923 .iter()
2924 .rev()
2925 .filter_map(|cmd| *cmd)
2926 .flat_map(|cmd| cmd.args.iter().copied());
2927 own.chain(inherited)
2928 .filter_map(|arg| {
2929 let sigil = arg.sigil?;
2930 (token.len() >= sigil.len() && token.starts_with(sigil)).then_some((arg, sigil))
2931 })
2932 .max_by_key(|(_, sigil)| sigil.len())
2933 }
2934
2935 fn current_args(&self) -> &'t [&'t Arg<'t>] {
2936 self.cmd
2937 .clause
2938 .map(|clause| clause.args)
2939 .unwrap_or(self.cmd.args)
2940 }
2941
2942 /// Skip empty optional positionals when every remaining value is needed by a later
2943 /// required positional. This is clap's opt-in `allow_missing_positional` policy.
2944 fn reserve_for_required_positionals(&mut self) {
2945 if !self.cmd.allow_missing_positional || self.arg_taken != 0 {
2946 return;
2947 }
2948 loop {
2949 let Some(current) = self.next_arg() else {
2950 return;
2951 };
2952 if current.required {
2953 return;
2954 }
2955 let required_after = self.current_args()[self.arg_pos + 1..]
2956 .iter()
2957 .filter(|arg| arg.required && arg.sigil.is_none())
2958 .count();
2959 if required_after == 0 {
2960 return;
2961 }
2962 let remaining_values = 1 + self.argv[self.pos..]
2963 .iter()
2964 .filter(|word| {
2965 (self.flags_stopped || !is_flag_like(bytes(word)))
2966 && self.match_sigil_arg(bytes(word)).is_none()
2967 })
2968 .count();
2969 if remaining_values > required_after {
2970 return;
2971 }
2972 self.advance_arg();
2973 }
2974 }
2975
2976 #[cfg(feature = "spec")]
2977 fn view_allows_own_flag(&self, flag: &Flag<'_>) -> bool {
2978 match self.view {
2979 None => true,
2980 // The promoted command keeps its own surface. While the injected path is
2981 // still at the host root, however, only explicitly carried globals belong
2982 // to the view; root-local flags are not part of the projected executable.
2983 Some(view) => {
2984 !core::ptr::eq(self.cmd, self.root) || is_version_flag(flag) || view.carries(flag)
2985 }
2986 }
2987 }
2988
2989 #[cfg(not(feature = "spec"))]
2990 fn view_allows_own_flag(&self, _flag: &Flag<'_>) -> bool {
2991 true
2992 }
2993
2994 #[cfg(feature = "spec")]
2995 fn view_allows_inherited_flag(&self, flag: &Flag<'_>) -> bool {
2996 match self.view {
2997 None => true,
2998 // A portable view carries selected host globals, not globals declared
2999 // by intermediate commands on a multi-segment promoted path.
3000 Some(view) => {
3001 self.root
3002 .flags
3003 .iter()
3004 .any(|root| core::ptr::eq(*root, flag))
3005 && (is_version_flag(flag) || view.carries(flag))
3006 }
3007 }
3008 }
3009
3010 #[cfg(not(feature = "spec"))]
3011 fn view_allows_inherited_flag(&self, _flag: &Flag<'_>) -> bool {
3012 true
3013 }
3014
3015 #[cfg(feature = "spec")]
3016 fn inherited_flag_is_in_scope(&self, flag: &Flag<'_>) -> bool {
3017 flag.global || (self.view.is_some() && is_version_flag(flag))
3018 }
3019
3020 #[cfg(not(feature = "spec"))]
3021 fn inherited_flag_is_in_scope(&self, flag: &Flag<'_>) -> bool {
3022 flag.global
3023 }
3024
3025 /// Flags in scope: this command's own, then any ancestor's globals.
3026 ///
3027 /// Own flags come first so that a subcommand redeclaring an inherited name
3028 /// shadows it, which is what mise relies on when it redeclares root globals
3029 /// on `run` with different shorts.
3030 fn in_scope(&self) -> impl Iterator<Item = &'t Flag<'t>> + '_ {
3031 let own = self
3032 .cmd
3033 .flags
3034 .iter()
3035 .copied()
3036 .filter(|flag| self.view_allows_own_flag(flag));
3037 let inherited = self.ancestors[..self.depth]
3038 .iter()
3039 .rev()
3040 .filter_map(|c| *c)
3041 .flat_map(|c| c.flags.iter().copied())
3042 .filter(|flag| self.inherited_flag_is_in_scope(flag))
3043 .filter(|flag| self.view_allows_inherited_flag(flag));
3044 own.chain(inherited)
3045 }
3046
3047 fn find_long(&self, name: &[u8]) -> Option<&'t Flag<'t>> {
3048 self.in_scope()
3049 .find(|f| f.longs.iter().any(|l| l.as_bytes() == name))
3050 }
3051
3052 fn find_negation(&self, name: &[u8]) -> Option<&'t Flag<'t>> {
3053 self.in_scope()
3054 .find(|f| f.negate.is_some_and(|n| n.as_bytes() == name))
3055 }
3056
3057 fn find_short(&self, byte: u8) -> Option<&'t Flag<'t>> {
3058 // Only the boundary token is shared. Later tokens use ordinary child/global scope.
3059 if self.default_taken && self.pos == self.default_bundle_end {
3060 if let Some(flag) = self.ancestors[0].and_then(|parent| {
3061 parent
3062 .flags
3063 .iter()
3064 .copied()
3065 .find(|f| f.shorts.contains(&byte))
3066 }) {
3067 return Some(flag);
3068 }
3069 }
3070 self.in_scope()
3071 .find(|f| f.shorts.contains(&byte))
3072 // As for `--help`: supplied by the parser, and only where the command has not
3073 // declared a `-h` of its own.
3074 .or(if byte == b'h' && !self.cmd.disable_help_flag {
3075 Some(&HELP_SHORT)
3076 } else if byte == b'V'
3077 && self.version_command().version
3078 && !self.version_command().disable_version_flag
3079 {
3080 Some(&VERSION_SHORT)
3081 } else {
3082 None
3083 })
3084 }
3085
3086 #[cfg(feature = "spec")]
3087 fn version_command(&self) -> &'t Command<'t> {
3088 if self.view.is_some() {
3089 self.root
3090 } else {
3091 self.cmd
3092 }
3093 }
3094
3095 #[cfg(not(feature = "spec"))]
3096 fn version_command(&self) -> &'t Command<'t> {
3097 self.cmd
3098 }
3099
3100 fn find_subcommand(&self, name: &[u8]) -> Option<&'t Command<'t>> {
3101 // Shared with `help` rather than spelled out again, so descending into a command and
3102 // asking about one cannot drift apart.
3103 find_named(self.cmd, name)
3104 }
3105}
3106
3107/// View a token as bytes.
3108///
3109/// `as_encoded_bytes` is a plain accessor with no conversion and no allocation.
3110/// The reverse direction is the one with a cost — see [`os_string_from_bytes`] —
3111/// which is why values come back as bytes.
3112fn bytes<'v>(s: &&'v OsStr) -> &'v [u8] {
3113 s.as_encoded_bytes()
3114}
3115
3116/// How many values one word carries.
3117///
3118/// One, until a delimiter is declared — and then one per separator, counting the same way
3119/// splitting on it does: `a,b` is two, `a,` is two with an empty second, and `` is one.
3120/// Counted rather than split because binding only needs the number, and the split itself
3121/// belongs to the layer that owns the values.
3122fn values_in(word: &[u8], delimiter: ::core::option::Option<u8>) -> u32 {
3123 match delimiter {
3124 Some(d) => 1 + word.iter().filter(|b| **b == d).count() as u32,
3125 None => 1,
3126 }
3127}
3128
3129/// Whether a token should be read as a flag.
3130///
3131/// `-` alone is a value, conventionally stdin. Other dash-prefixed tokens are
3132/// flag-like; a field may make the narrower negative-number exception.
3133fn is_flag_like(token: &[u8]) -> bool {
3134 matches!(token, [b'-', rest @ ..] if !rest.is_empty())
3135}
3136
3137fn is_negative_number(token: &[u8]) -> bool {
3138 token.strip_prefix(b"-").is_some_and(is_number)
3139}
3140
3141/// Whether the text after a `-` is a number, so `-1`, `-2.5`, and `-1e5` are values
3142/// while `-1x` is a flag-shaped token that names nothing.
3143///
3144/// Digits, at most one `.`, and an optional exponent. Deliberately narrower than
3145/// `f64::from_str`, which also accepts `inf` and `NaN` — `-inf` is far likelier to be
3146/// a misspelled flag than a number somebody meant to pass.
3147///
3148/// usage-lib applies the same rule, and the corpus pins the edges so the two cannot
3149/// drift apart: they disagreed about `-1e5` when this was a hand-rolled scanner on
3150/// one side and a float parse on the other.
3151///
3152/// Written out rather than deferred to `f64::from_str` because this runs on the hot
3153/// path, and a parse would mean a UTF-8 check on a slice already decided by its
3154/// bytes.
3155fn is_number(rest: &[u8]) -> bool {
3156 let (mantissa, exponent) = match rest.iter().position(|b| matches!(b, b'e' | b'E')) {
3157 Some(at) => (&rest[..at], Some(&rest[at + 1..])),
3158 None => (rest, None),
3159 };
3160
3161 let mut seen_digit = false;
3162 let mut seen_dot = false;
3163 for &b in mantissa {
3164 match b {
3165 b'0'..=b'9' => seen_digit = true,
3166 b'.' if !seen_dot => seen_dot = true,
3167 _ => return false,
3168 }
3169 }
3170 if !seen_digit {
3171 return false;
3172 }
3173
3174 match exponent {
3175 None => true,
3176 // An exponent needs digits of its own, and may carry a sign.
3177 Some(exp) => {
3178 let digits = exp
3179 .strip_prefix(b"+")
3180 .or_else(|| exp.strip_prefix(b"-"))
3181 .unwrap_or(exp);
3182 !digits.is_empty() && digits.iter().all(|b| b.is_ascii_digit())
3183 }
3184 }
3185}
3186
3187fn validate_bool_value<'t, 'v>(
3188 flag: &'t Flag<'t>,
3189 value: Option<&'v [u8]>,
3190) -> Result<Option<&'v [u8]>, Error<'t, 'v>> {
3191 match value {
3192 None | Some(b"true" | b"false") => Ok(value),
3193 Some(_) => Err(Error::InvalidChoice {
3194 name: flag.name,
3195 choices: &["true", "false"],
3196 }),
3197 }
3198}
3199
3200#[cfg(test)]
3201mod tests {
3202 use super::*;
3203
3204 #[test]
3205 fn clause_separator_resets_inner_args_and_survives_automatic_mode() {
3206 static TASK: Arg = Arg {
3207 key: 91,
3208 name: "task",
3209 ..Arg::REQUIRED
3210 };
3211 static REST: Arg = Arg {
3212 key: 92,
3213 name: "args",
3214 double_dash: DoubleDash::Automatic,
3215 ..Arg::VAR
3216 };
3217 static ROOT: Command = Command {
3218 name: "ex",
3219 clause: Some(&Clause {
3220 key: 90,
3221 name: "tasks",
3222 separator: Some(b":::"),
3223 flags: &[],
3224 args: &[&TASK, &REST],
3225 }),
3226 ..Command::EMPTY
3227 };
3228 let argv = [
3229 OsStr::new("lint"),
3230 OsStr::new("--fix"),
3231 OsStr::new(":::"),
3232 OsStr::new("test"),
3233 ];
3234 let mut parser = Parser::new(&ROOT, &argv);
3235 let mut seen = Vec::new();
3236 while let Some(event) = parser.next_event() {
3237 match event.expect("valid clause") {
3238 Event::Arg { arg, value, .. } => {
3239 seen.push((arg.name, String::from_utf8_lossy(value).into_owned()))
3240 }
3241 Event::ClauseSeparator { clause } => seen.push((clause.name, ":::".into())),
3242 _ => {}
3243 }
3244 }
3245 assert_eq!(
3246 seen,
3247 [
3248 ("task", "lint".into()),
3249 ("args", "--fix".into()),
3250 ("tasks", ":::".into()),
3251 ("task", "test".into())
3252 ]
3253 );
3254 }
3255
3256 #[test]
3257 fn clause_separator_is_not_a_scoped_flag_value() {
3258 static POSTINSTALL: Flag = Flag {
3259 key: 93,
3260 name: "postinstall",
3261 longs: &["postinstall"],
3262 takes_value: true,
3263 variadic: true,
3264 ..Flag::BOOL
3265 };
3266 static TOOL: Arg = Arg {
3267 key: 94,
3268 name: "tool",
3269 ..Arg::REQUIRED
3270 };
3271 static ROOT: Command = Command {
3272 name: "ex",
3273 flags: &[&POSTINSTALL],
3274 clause: Some(&Clause {
3275 key: 95,
3276 name: "tools",
3277 separator: Some(b":::"),
3278 flags: &[&POSTINSTALL],
3279 args: &[&TOOL],
3280 }),
3281 ..Command::EMPTY
3282 };
3283 let argv = [OsStr::new("--postinstall"), OsStr::new(":::")];
3284 let mut parser = Parser::new(&ROOT, &argv);
3285 assert!(matches!(
3286 parser.next_event(),
3287 Some(Err(Error::MissingFlagValue { flag })) if flag.key == POSTINSTALL.key
3288 ));
3289
3290 let argv = [
3291 OsStr::new("--postinstall"),
3292 OsStr::new("setup"),
3293 OsStr::new(":::"),
3294 OsStr::new("tool"),
3295 ];
3296 let mut parser = Parser::new(&ROOT, &argv);
3297 assert!(matches!(
3298 parser.next_event(),
3299 Some(Ok(Event::Flag {
3300 value: Some(b"setup"),
3301 ..
3302 }))
3303 ));
3304 assert!(matches!(
3305 parser.next_event(),
3306 Some(Ok(Event::ClauseSeparator { clause })) if clause.key == 95
3307 ));
3308 assert!(matches!(
3309 parser.next_event(),
3310 Some(Ok(Event::Arg { value: b"tool", .. }))
3311 ));
3312 }
3313
3314 static FORCE: Flag = Flag {
3315 key: 1,
3316 longs: &["force"],
3317 shorts: b"f",
3318 ..Flag::BOOL
3319 };
3320 static EXPLICIT_BOOL: Flag = Flag {
3321 key: 20,
3322 name: "color",
3323 longs: &["color"],
3324 negate: Some("no-color"),
3325 bool_value: true,
3326 ..Flag::BOOL
3327 };
3328 static EXPLICIT_BOOL_ROOT: Command = Command {
3329 name: "ex",
3330 flags: &[&EXPLICIT_BOOL],
3331 ..Command::EMPTY
3332 };
3333 static JOBS: Flag = Flag {
3334 key: 2,
3335 longs: &["jobs"],
3336 shorts: b"j",
3337 allow_negative_numbers: true,
3338 ..Flag::VALUE
3339 };
3340 static COLOR: Flag = Flag {
3341 key: 3,
3342 longs: &["color"],
3343 negate: Some("no-color"),
3344 ..Flag::BOOL
3345 };
3346 static VERBOSE: Flag = Flag {
3347 key: 4,
3348 longs: &["verbose"],
3349 shorts: b"v",
3350 global: true,
3351 ..Flag::BOOL
3352 };
3353 static FILE: Arg = Arg {
3354 key: 10,
3355 name: "file",
3356 allow_negative_numbers: true,
3357 ..Arg::REQUIRED
3358 };
3359 static REST: Arg = Arg {
3360 key: 11,
3361 name: "rest",
3362 ..Arg::VAR
3363 };
3364 static INSTALL: Command = Command {
3365 name: "install",
3366 aliases: &["i"],
3367 flags: &[&FORCE],
3368 key: 100,
3369 ..Command::EMPTY
3370 };
3371 /// Same shape as ROOT, but a CLI that owns all of its flags. The subcommand says
3372 /// nothing and inherits it, which is the point: only the root declares the mode.
3373 static STRICT_INSTALL: Command = Command {
3374 name: "install",
3375 aliases: &["i"],
3376 flags: &[&FORCE],
3377 key: 100,
3378 ..Command::EMPTY
3379 };
3380 static STRICT: Command = Command {
3381 name: "ex",
3382 flags: &[&FORCE, &JOBS, &COLOR, &VERBOSE],
3383 args: &[&FILE, &REST],
3384 subcommands: &[&STRICT_INSTALL],
3385 unknown_flags: Some(UnknownFlags::Error),
3386 ..Command::EMPTY
3387 };
3388 static ROOT: Command = Command {
3389 name: "ex",
3390 flags: &[&FORCE, &JOBS, &COLOR, &VERBOSE],
3391 args: &[&FILE, &REST],
3392 subcommands: &[&INSTALL],
3393 ..Command::EMPTY
3394 };
3395 static ARGUMENT_CONFLICT: Command = Command {
3396 name: "ex",
3397 flags: &[&FORCE],
3398 subcommands: &[&INSTALL],
3399 args_conflicts_with_subcommands: true,
3400 ..Command::EMPTY
3401 };
3402
3403 // A CLI shaped exactly like mise's root: a default subcommand, a positional of its own,
3404 // and a subcommand under the default — which is the arrangement that tells routing from
3405 // a plain positional.
3406 static TASK: Arg = Arg {
3407 key: 20,
3408 name: "task",
3409 ..Arg::REQUIRED
3410 };
3411 static RUN_TASK: Arg = Arg {
3412 key: 21,
3413 name: "run_task",
3414 ..Arg::REQUIRED
3415 };
3416 static DEEP: Command = Command {
3417 name: "deep",
3418 args: &[&RUN_TASK],
3419 key: 203,
3420 ..Command::EMPTY
3421 };
3422 static LINT: Command = Command {
3423 name: "lint",
3424 subcommands: &[&DEEP],
3425 // A default of its own, so that a parse which forgot it had already taken one would
3426 // have somewhere to go. Nothing else in these fixtures can show the latch working.
3427 default_subcommand: Some(&DEEP),
3428 key: 202,
3429 ..Command::EMPTY
3430 };
3431 static RUN: Command = Command {
3432 name: "run",
3433 args: &[&RUN_TASK],
3434 subcommands: &[&LINT],
3435 key: 200,
3436 ..Command::EMPTY
3437 };
3438 static DEFAULTING: Command = Command {
3439 name: "mise",
3440 flags: &[&VERBOSE],
3441 args: &[&TASK],
3442 subcommands: &[&RUN, &INSTALL],
3443 default_subcommand: Some(find_subcommand(&[&RUN, &INSTALL], "run")),
3444 ..Command::EMPTY
3445 };
3446
3447 /// Collect every event, or the first error.
3448 fn parse<'t: 'v, 'v>(
3449 root: &'t Command<'t>,
3450 argv: &'v [&'v OsStr],
3451 ) -> Result<Vec<Event<'t, 'v, 'v>>, Error<'t, 'v>> {
3452 let mut parser = Parser::new(root, argv);
3453 let mut events = Vec::new();
3454 while let Some(event) = parser.next_event() {
3455 events.push(event?);
3456 }
3457 Ok(events)
3458 }
3459
3460 fn argv<const N: usize>(tokens: [&str; N]) -> [&OsStr; N] {
3461 tokens.map(OsStr::new)
3462 }
3463
3464 #[test]
3465 fn long_boolean() {
3466 let a = argv(["--force"]);
3467 assert_eq!(
3468 parse(&ROOT, &a).unwrap(),
3469 vec![Event::Flag {
3470 flag: &FORCE,
3471 value: None,
3472 negated: false
3473 }]
3474 );
3475 }
3476
3477 #[test]
3478 fn long_boolean_accepts_only_opted_in_attached_values() {
3479 for (token, negated, value) in [
3480 ("--color=false", false, b"false".as_slice()),
3481 ("--color=true", false, b"true".as_slice()),
3482 ("--no-color=false", true, b"false".as_slice()),
3483 ] {
3484 let a = argv([token]);
3485 assert_eq!(
3486 parse(&EXPLICIT_BOOL_ROOT, &a).unwrap(),
3487 vec![Event::Flag {
3488 flag: &EXPLICIT_BOOL,
3489 value: Some(value),
3490 negated,
3491 }]
3492 );
3493 }
3494
3495 let a = argv(["--color=maybe"]);
3496 assert!(matches!(
3497 parse(&EXPLICIT_BOOL_ROOT, &a),
3498 Err(Error::InvalidChoice { name: "color", .. })
3499 ));
3500
3501 let a = argv(["--force=false"]);
3502 assert_eq!(
3503 parse(&ROOT, &a).unwrap(),
3504 vec![Event::Flag {
3505 flag: &FORCE,
3506 value: None,
3507 negated: false,
3508 }]
3509 );
3510 }
3511
3512 #[test]
3513 fn long_value_forms() {
3514 for tokens in [vec!["--jobs=8"], vec!["--jobs", "8"]] {
3515 let a: Vec<&OsStr> = tokens.iter().map(|t| OsStr::new(*t)).collect();
3516 assert_eq!(
3517 parse(&ROOT, &a).unwrap(),
3518 vec![Event::Flag {
3519 flag: &JOBS,
3520 value: Some(b"8"),
3521 negated: false
3522 }],
3523 "{tokens:?}"
3524 );
3525 }
3526 }
3527
3528 #[test]
3529 fn long_value_keeps_later_equals() {
3530 let a = argv(["--jobs=a=b"]);
3531 let Event::Flag { value, .. } = parse(&ROOT, &a).unwrap()[0] else {
3532 panic!("expected a flag");
3533 };
3534 assert_eq!(value, Some(&b"a=b"[..]));
3535 }
3536
3537 #[test]
3538 fn long_value_attached_empty_is_empty_not_absent() {
3539 let a = argv(["--jobs="]);
3540 let Event::Flag { value, .. } = parse(&ROOT, &a).unwrap()[0] else {
3541 panic!("expected a flag");
3542 };
3543 assert_eq!(value, Some(&b""[..]));
3544 }
3545
3546 #[test]
3547 fn long_value_refuses_flaglike_next_word() {
3548 let a = argv(["--jobs", "--force"]);
3549 assert_eq!(
3550 parse(&ROOT, &a),
3551 Err(Error::MissingFlagValue { flag: &JOBS })
3552 );
3553 }
3554
3555 #[test]
3556 fn long_value_accepts_negative_number() {
3557 let a = argv(["--jobs", "-1"]);
3558 let Event::Flag { value, .. } = parse(&ROOT, &a).unwrap()[0] else {
3559 panic!("expected a flag");
3560 };
3561 assert_eq!(value, Some(&b"-1"[..]));
3562 }
3563
3564 #[test]
3565 fn missing_optional_positional_reserves_the_last_word() {
3566 static OPTIONAL: Arg = Arg {
3567 key: 90,
3568 name: "optional",
3569 required: false,
3570 ..Arg::REQUIRED
3571 };
3572 static REQUIRED: Arg = Arg {
3573 key: 91,
3574 name: "required",
3575 ..Arg::REQUIRED
3576 };
3577 static CMD: Command = Command {
3578 name: "ex",
3579 args: &[&OPTIONAL, &REQUIRED],
3580 allow_missing_positional: true,
3581 ..Command::EMPTY
3582 };
3583
3584 let one = argv(["value"]);
3585 assert_eq!(
3586 parse(&CMD, &one).unwrap(),
3587 vec![Event::Arg {
3588 arg: &REQUIRED,
3589 value: b"value",
3590 delimit: true
3591 }]
3592 );
3593 let two = argv(["optional", "required"]);
3594 assert_eq!(
3595 parse(&CMD, &two).unwrap(),
3596 vec![
3597 Event::Arg {
3598 arg: &OPTIONAL,
3599 value: b"optional",
3600 delimit: true
3601 },
3602 Event::Arg {
3603 arg: &REQUIRED,
3604 value: b"required",
3605 delimit: true
3606 },
3607 ]
3608 );
3609 }
3610
3611 #[test]
3612 fn negative_numbers_are_narrowly_opted_in() {
3613 static PLAIN: Flag = Flag {
3614 key: 90,
3615 name: "plain",
3616 longs: &["plain"],
3617 ..Flag::VALUE
3618 };
3619 static VALUE: Arg = Arg {
3620 key: 91,
3621 name: "value",
3622 ..Arg::REQUIRED
3623 };
3624 static CMD: Command = Command {
3625 name: "ex",
3626 flags: &[&PLAIN],
3627 args: &[&VALUE],
3628 unknown_flags: Some(UnknownFlags::Error),
3629 ..Command::EMPTY
3630 };
3631
3632 let flag = argv(["--plain", "-1"]);
3633 assert_eq!(
3634 parse(&CMD, &flag),
3635 Err(Error::MissingFlagValue { flag: &PLAIN })
3636 );
3637 let positional = argv(["-1"]);
3638 assert_eq!(
3639 parse(&CMD, &positional),
3640 Err(Error::UnknownFlag { token: b"-1" })
3641 );
3642 }
3643
3644 #[test]
3645 fn an_exact_declared_digit_short_outranks_a_negative_number() {
3646 static PRINT0: Flag = Flag {
3647 key: 92,
3648 name: "print0",
3649 shorts: b"0",
3650 ..Flag::BOOL
3651 };
3652 static VALUE: Arg = Arg {
3653 key: 93,
3654 name: "value",
3655 required: false,
3656 allow_negative_numbers: true,
3657 ..Arg::REQUIRED
3658 };
3659 static CMD: Command = Command {
3660 name: "fd",
3661 flags: &[&PRINT0],
3662 args: &[&VALUE],
3663 unknown_flags: Some(UnknownFlags::Error),
3664 ..Command::EMPTY
3665 };
3666
3667 assert_eq!(
3668 parse(&CMD, &argv(["-0"])),
3669 Ok(vec![Event::Flag {
3670 flag: &PRINT0,
3671 value: None,
3672 negated: false,
3673 }])
3674 );
3675 assert!(matches!(
3676 parse(&CMD, &argv(["-1"])),
3677 Ok(events) if matches!(events.as_slice(), [Event::Arg { value: b"-1", .. }])
3678 ));
3679 }
3680
3681 #[test]
3682 fn negation_of_value_flag_does_not_consume_a_value() {
3683 static MODE: Flag = Flag {
3684 key: 9,
3685 name: "mode",
3686 longs: &["mode"],
3687 negate: Some("no-mode"),
3688 ..Flag::VALUE
3689 };
3690 static NEGATED_VALUE: Command = Command {
3691 name: "ex",
3692 flags: &[&MODE],
3693 args: &[&FILE],
3694 ..Command::EMPTY
3695 };
3696
3697 let a = argv(["--no-mode", "input"]);
3698 assert_eq!(
3699 parse(&NEGATED_VALUE, &a).unwrap(),
3700 vec![
3701 Event::Flag {
3702 flag: &MODE,
3703 value: None,
3704 negated: true
3705 },
3706 Event::Arg {
3707 arg: &FILE,
3708 value: b"input",
3709 delimit: true,
3710 }
3711 ]
3712 );
3713 }
3714
3715 #[test]
3716 fn no_abbreviation() {
3717 // A prefix names no flag, so by default it is a value like any other word.
3718 let a = argv(["--forc"]);
3719 assert_eq!(
3720 parse(&ROOT, &a).unwrap(),
3721 vec![Event::Arg {
3722 arg: &FILE,
3723 value: b"--forc",
3724 delimit: true,
3725 }]
3726 );
3727
3728 // And a CLI that owns its flags hears about it, which is the whole reason
3729 // the strict mode exists.
3730 assert!(matches!(
3731 parse(&STRICT, &a),
3732 Err(Error::UnknownFlag { token: b"--forc" })
3733 ));
3734 }
3735
3736 #[test]
3737 fn an_unknown_flag_is_a_value_by_default() {
3738 // The default, and the case it is for: a command line being forwarded to
3739 // something whose flags this spec does not know.
3740 let a = argv(["--wat", "keep"]);
3741 assert_eq!(
3742 parse(&ROOT, &a).unwrap(),
3743 vec![
3744 Event::Arg {
3745 arg: &FILE,
3746 value: b"--wat",
3747 delimit: true,
3748 },
3749 Event::Arg {
3750 arg: &REST,
3751 value: b"keep",
3752 delimit: true,
3753 },
3754 ]
3755 );
3756
3757 // With nowhere to put it, it is an unexpected argument — the same error an
3758 // extra word gets, rather than a special one about flags.
3759 static ONE: Command = Command {
3760 name: "ex",
3761 args: &[&FILE],
3762 ..Command::EMPTY
3763 };
3764 let a = argv(["a", "--wat"]);
3765 assert_eq!(
3766 parse(&ONE, &a),
3767 Err(Error::UnexpectedArg { token: b"--wat" })
3768 );
3769 }
3770
3771 #[test]
3772 fn negation() {
3773 let a = argv(["--no-color"]);
3774 assert_eq!(
3775 parse(&ROOT, &a).unwrap(),
3776 vec![Event::Flag {
3777 flag: &COLOR,
3778 value: None,
3779 negated: true
3780 }]
3781 );
3782 }
3783
3784 #[test]
3785 fn short_bundle_and_attached_value() {
3786 let a = argv(["-fj8"]);
3787 assert_eq!(
3788 parse(&ROOT, &a).unwrap(),
3789 vec![
3790 Event::Flag {
3791 flag: &FORCE,
3792 value: None,
3793 negated: false
3794 },
3795 Event::Flag {
3796 flag: &JOBS,
3797 value: Some(b"8"),
3798 negated: false
3799 },
3800 ]
3801 );
3802 }
3803
3804 #[test]
3805 fn short_value_strips_one_equals() {
3806 for (tokens, want) in [(["-j=8"], &b"8"[..]), (["-j==8"], &b"=8"[..])] {
3807 let a = argv(tokens);
3808 let Event::Flag { value, .. } = parse(&ROOT, &a).unwrap()[0] else {
3809 panic!("expected a flag");
3810 };
3811 assert_eq!(value, Some(want), "{tokens:?}");
3812 }
3813 }
3814
3815 #[test]
3816 fn bare_dash_is_a_value() {
3817 let a = argv(["-"]);
3818 assert_eq!(
3819 parse(&ROOT, &a).unwrap(),
3820 vec![Event::Arg {
3821 arg: &FILE,
3822 value: b"-",
3823 delimit: true,
3824 }]
3825 );
3826 }
3827
3828 #[test]
3829 fn positionals_then_variadic() {
3830 let a = argv(["one", "two", "three"]);
3831 assert_eq!(
3832 parse(&ROOT, &a).unwrap(),
3833 vec![
3834 Event::Arg {
3835 arg: &FILE,
3836 value: b"one",
3837 delimit: true,
3838 },
3839 Event::Arg {
3840 arg: &REST,
3841 value: b"two",
3842 delimit: true,
3843 },
3844 Event::Arg {
3845 arg: &REST,
3846 value: b"three",
3847 delimit: true,
3848 },
3849 ]
3850 );
3851 }
3852
3853 #[test]
3854 fn subcommand_and_alias_route_the_same() {
3855 for token in ["install", "i"] {
3856 let a = argv([token]);
3857 assert_eq!(
3858 parse(&ROOT, &a).unwrap(),
3859 vec![Event::Command(&INSTALL)],
3860 "{token}"
3861 );
3862 }
3863 }
3864
3865 #[test]
3866 fn a_parent_argument_can_exclude_a_later_subcommand() {
3867 let a = argv(["--force", "install"]);
3868 assert!(matches!(
3869 parse(&ARGUMENT_CONFLICT, &a),
3870 Err(Error::SubcommandConflict { subcommand }) if subcommand.name == "install"
3871 ));
3872 }
3873
3874 #[test]
3875 fn subcommand_only_routes_before_a_positional_is_filled() {
3876 let a = argv(["other", "install"]);
3877 assert_eq!(
3878 parse(&ROOT, &a).unwrap(),
3879 vec![
3880 Event::Arg {
3881 arg: &FILE,
3882 value: b"other",
3883 delimit: true,
3884 },
3885 Event::Arg {
3886 arg: &REST,
3887 value: b"install",
3888 delimit: true,
3889 },
3890 ]
3891 );
3892 }
3893
3894 #[test]
3895 fn a_word_naming_no_subcommand_goes_to_the_default_one() {
3896 // usage-lib's answer, which this reproduces: `mise build` comes back as commands
3897 // `["mise", "run"]` with the word bound to *run's* argument — not to `mise`'s own
3898 // `[TASK]`, which is what makes this more than a synonym for a positional.
3899 let a = argv(["build"]);
3900 assert_eq!(
3901 parse(&DEFAULTING, &a).unwrap(),
3902 vec![
3903 Event::Command(&RUN),
3904 Event::Arg {
3905 arg: &RUN_TASK,
3906 value: b"build",
3907 delimit: true,
3908 },
3909 ]
3910 );
3911 }
3912
3913 // A shared table, as a flattened struct's would be. Declared outside the tests so both
3914 // can splice it, which is the arrangement it exists to model.
3915 static SHARED_QUIET: Flag = Flag {
3916 key: 300,
3917 name: "quiet",
3918 longs: &["quiet"],
3919 ..Flag::BOOL
3920 };
3921 static SHARED_FLAGS: &[&Flag] = &[&SHARED_QUIET];
3922 static SHARED_WHAT: Arg = Arg {
3923 key: 301,
3924 name: "what",
3925 ..Arg::REQUIRED
3926 };
3927 static SHARED_ARGS: &[&Arg] = &[&SHARED_WHAT];
3928
3929 #[test]
3930 fn concatenating_tables_keeps_the_order_they_were_given_in() {
3931 // The property positional arguments depend on: a flattened group lands where the
3932 // field was written, not at the end. `[&FILE], SHARED, [&REST]` has to stay in that
3933 // order or `ex a b c` binds the wrong words.
3934 const ARGS: &[&[&Arg]] = &[&[&FILE], SHARED_ARGS, &[&REST]];
3935 static TABLE: [&Arg; table_len(ARGS)] = concat_args(ARGS);
3936 assert_eq!(
3937 TABLE.iter().map(|a| a.name).collect::<Vec<_>>(),
3938 ["file", "what", "rest"]
3939 );
3940
3941 // Empty groups contribute nothing and disturb nothing, which is what lets the derive
3942 // emit a group per field without checking whether it is empty first.
3943 const WITH_GAPS: &[&[&Flag]] = &[&[], &[&FORCE], &[], SHARED_FLAGS, &[]];
3944 static FLAGS: [&Flag; table_len(WITH_GAPS)] = concat_flags(WITH_GAPS);
3945 // By long form: these fixtures do not all set `name`, and the placeholder's is also
3946 // empty — so comparing names could not tell a real entry from a leftover slot.
3947 assert_eq!(
3948 FLAGS.iter().map(|f| f.longs).collect::<Vec<_>>(),
3949 [&["force"], &["quiet"]]
3950 );
3951 }
3952
3953 #[test]
3954 fn a_concatenated_table_parses_like_a_declared_one() {
3955 // The point of doing this at compile time: what the parser walks is one flat slice,
3956 // indistinguishable from a command that declared everything itself.
3957 const FLAG_GROUPS: &[&[&Flag]] = &[&[&FORCE], SHARED_FLAGS];
3958 const ARG_GROUPS: &[&[&Arg]] = &[SHARED_ARGS, &[&REST]];
3959 static FLAGS: [&Flag; table_len(FLAG_GROUPS)] = concat_flags(FLAG_GROUPS);
3960 static ARGS: [&Arg; table_len(ARG_GROUPS)] = concat_args(ARG_GROUPS);
3961 static JOINED: Command = Command {
3962 name: "joined",
3963 flags: &FLAGS,
3964 args: &ARGS,
3965 ..Command::EMPTY
3966 };
3967
3968 let a = argv(["--quiet", "one", "two", "--force"]);
3969 assert_eq!(
3970 parse(&JOINED, &a).unwrap(),
3971 vec![
3972 Event::Flag {
3973 flag: &SHARED_QUIET,
3974 value: None,
3975 negated: false
3976 },
3977 Event::Arg {
3978 arg: &SHARED_WHAT,
3979 value: b"one",
3980 delimit: true,
3981 },
3982 Event::Arg {
3983 arg: &REST,
3984 value: b"two",
3985 delimit: true,
3986 },
3987 Event::Flag {
3988 flag: &FORCE,
3989 value: None,
3990 negated: false
3991 },
3992 ]
3993 );
3994 }
3995
3996 #[test]
3997 fn an_unknown_flag_is_not_routed() {
3998 // A dash-prefixed token that names no flag becomes a value here (the default for
3999 // `unknown_flags`), and it must not thereby become a *subcommand* word: usage-lib
4000 // stops looking for subcommands at an unrecognised flag, and binds it to the command
4001 // still in scope. Verified against usage-lib, where `ex --wat` comes back as commands
4002 // `["ex"]` with `ROOT_TASK = "--wat"`.
4003 for token in ["--wat", "-x"] {
4004 let a = argv([token]);
4005 assert_eq!(
4006 parse(&DEFAULTING, &a).unwrap(),
4007 vec![Event::Arg {
4008 arg: &TASK,
4009 value: token.as_bytes(),
4010 delimit: true,
4011 }],
4012 "{token} should bind where it was typed, not in the default subcommand"
4013 );
4014 }
4015 }
4016
4017 #[test]
4018 fn a_named_subcommand_is_not_routed() {
4019 // The default is for words that name nothing. A word that names a sibling still
4020 // selects it, and the root's own argument is still reachable behind one.
4021 let a = argv(["install"]);
4022 assert_eq!(
4023 parse(&DEFAULTING, &a).unwrap(),
4024 vec![Event::Command(&INSTALL)]
4025 );
4026 }
4027
4028 #[test]
4029 fn an_unmatched_word_is_forwarded_when_external_subcommand_is_set() {
4030 static CATCH: Command = Command {
4031 name: "ex",
4032 flags: &[&VERBOSE],
4033 subcommands: &[&INSTALL],
4034 external_subcommand: true,
4035 unknown_flags: Some(UnknownFlags::Error),
4036 ..Command::EMPTY
4037 };
4038 let a = argv(["foo", "--help", "bar"]);
4039 assert_eq!(
4040 parse(&CATCH, &a).unwrap(),
4041 vec![Event::External { values: &a[..] }]
4042 );
4043
4044 let a = argv(["install"]);
4045 assert_eq!(parse(&CATCH, &a).unwrap(), vec![Event::Command(&INSTALL)]);
4046
4047 let a = argv(["--verbose", "foo", "--verbose"]);
4048 assert_eq!(
4049 parse(&CATCH, &a).unwrap(),
4050 vec![
4051 Event::Flag {
4052 flag: &VERBOSE,
4053 value: None,
4054 negated: false
4055 },
4056 Event::External { values: &a[1..] }
4057 ]
4058 );
4059
4060 let a = argv(["--wat"]);
4061 assert_eq!(
4062 parse(&CATCH, &a),
4063 Err(Error::UnknownFlag { token: b"--wat" })
4064 );
4065
4066 // A negative number is a value, not a flag, so it can be the unmatched word.
4067 let a = argv(["-1", "rest"]);
4068 assert_eq!(
4069 parse(&CATCH, &a).unwrap(),
4070 vec![Event::External { values: &a[..] }]
4071 );
4072 }
4073
4074 #[test]
4075 fn a_default_subcommand_outranks_an_external_one() {
4076 static CATCH_DEFAULT: Command = Command {
4077 name: "ex",
4078 subcommands: &[&RUN],
4079 default_subcommand: Some(&RUN),
4080 external_subcommand: true,
4081 ..Command::EMPTY
4082 };
4083 let a = argv(["build"]);
4084 assert_eq!(
4085 parse(&CATCH_DEFAULT, &a).unwrap(),
4086 vec![
4087 Event::Command(&RUN),
4088 Event::Arg {
4089 arg: &RUN_TASK,
4090 value: b"build",
4091 delimit: true,
4092 }
4093 ]
4094 );
4095 }
4096
4097 #[test]
4098 fn a_default_subcommand_starts_at_the_word_it_receives() {
4099 let a = argv(["build"]);
4100 let mut parser = Parser::new(&DEFAULTING, &a);
4101 assert_eq!(parser.next_event(), Some(Ok(Event::Command(&RUN))));
4102 assert_eq!(parser.command_start(), 0);
4103 assert_eq!(
4104 parser.next_event(),
4105 Some(Ok(Event::Arg {
4106 arg: &RUN_TASK,
4107 value: b"build",
4108 delimit: true,
4109 }))
4110 );
4111 }
4112
4113 #[test]
4114 fn a_root_clause_separator_does_not_select_empty_default() {
4115 static CHILD: Command = Command {
4116 name: "run",
4117 ..Command::EMPTY
4118 };
4119 static ROOT: Command = Command {
4120 name: "ex",
4121 clause: Some(&Clause {
4122 key: 400,
4123 name: "items",
4124 separator: Some(b":::"),
4125 flags: &[],
4126 args: &[],
4127 }),
4128 subcommands: &[&CHILD],
4129 default_subcommand: Some(&CHILD),
4130 default_subcommand_on_empty: true,
4131 ..Command::EMPTY
4132 };
4133 let a = argv([":::"]);
4134 assert_eq!(
4135 parse(&ROOT, &a).unwrap(),
4136 vec![Event::ClauseSeparator {
4137 clause: *ROOT.clause.unwrap(),
4138 }]
4139 );
4140 }
4141
4142 #[test]
4143 fn the_default_can_be_named_by_an_alias() {
4144 // usage-lib resolves the name against subcommand names, aliases and hidden aliases
4145 // alike, so a spec may point `default_subcommand` at any of them.
4146 static BY_ALIAS: Command = Command {
4147 name: "mise",
4148 args: &[&TASK],
4149 subcommands: &[&INSTALL],
4150 // `INSTALL` answers to "i" as well as to its name.
4151 default_subcommand: Some(find_subcommand(&[&INSTALL], "i")),
4152 ..Command::EMPTY
4153 };
4154 assert!(::core::ptr::eq(
4155 BY_ALIAS.default_subcommand.expect("declared"),
4156 &INSTALL
4157 ));
4158 }
4159
4160 #[test]
4161 fn a_name_outranks_another_commands_alias() {
4162 // A spec `assert_unique_subcommand_names` would reject, resolved anyway: a parser
4163 // handed a table nothing validated still has to answer, and the answer is the
4164 // command whose own name it is. Both orders, because taking the first candidate
4165 // that matched on either name or alias made this depend on which was listed first
4166 // — and usage-lib, building a map, took the last.
4167 static ALPHA: Command = Command {
4168 name: "alpha",
4169 aliases: &["run"],
4170 key: 300,
4171 ..Command::EMPTY
4172 };
4173 static PLAIN_RUN: Command = Command {
4174 name: "run",
4175 key: 301,
4176 ..Command::EMPTY
4177 };
4178 for subcommands in [&[&ALPHA, &PLAIN_RUN] as &[&Command], &[&PLAIN_RUN, &ALPHA]] {
4179 assert!(::core::ptr::eq(
4180 find_subcommand(subcommands, "run"),
4181 &PLAIN_RUN
4182 ));
4183 let root: Command = Command {
4184 name: "ex",
4185 subcommands,
4186 ..Command::EMPTY
4187 };
4188 let a = argv(["run"]);
4189 assert_eq!(parse(&root, &a).unwrap(), vec![Event::Command(&PLAIN_RUN)]);
4190 // The alias still reaches its own command by every name it does not share.
4191 let a = argv(["alpha"]);
4192 assert_eq!(parse(&root, &a).unwrap(), vec![Event::Command(&ALPHA)]);
4193 // `ex help run` asks about the command `ex run` selects. These are separate
4194 // lookups — help resolves a path without descending — and answering differently
4195 // for a colliding word is the divergence this rule exists to end.
4196 let a = argv(["help", "run"]);
4197 match parse(&root, &a) {
4198 Err(Error::Help { cmd, .. }) => {
4199 assert!(
4200 ::core::ptr::eq(cmd, &PLAIN_RUN),
4201 "got help for {}",
4202 cmd.name
4203 )
4204 }
4205 other => panic!("expected a help request, got {other:?}"),
4206 }
4207 }
4208 }
4209
4210 #[test]
4211 #[should_panic(expected = "two subcommands answer to the same name")]
4212 fn an_alias_cannot_shadow_a_sibling_command() {
4213 static ADD: Command = Command {
4214 name: "add",
4215 aliases: &["install"],
4216 ..Command::EMPTY
4217 };
4218 assert_unique_subcommand_names(&[&INSTALL, &ADD]);
4219 }
4220
4221 #[test]
4222 fn the_word_is_re_examined_against_the_command_it_reached() {
4223 // The reason the cursor steps back rather than the token being consumed: `lint` names
4224 // nothing at the root, and once inside `run` it names a subcommand. mise's mounted
4225 // task names arrive exactly this way.
4226 let a = argv(["lint"]);
4227 assert_eq!(
4228 parse(&DEFAULTING, &a).unwrap(),
4229 vec![Event::Command(&RUN), Event::Command(&LINT)]
4230 );
4231 }
4232
4233 #[test]
4234 fn the_default_is_taken_at_most_once_per_parse() {
4235 // usage-lib latches this for the whole parse rather than per command, and the shape
4236 // that shows the difference needs two of them: `lint` routes through `run`, and `lint`
4237 // declares a default too. A second word there would descend again — walking a CLI
4238 // deeper than anything the user typed — so the answer is that it does not.
4239 let a = argv(["lint", "zzz"]);
4240 assert_eq!(
4241 parse(&DEFAULTING, &a),
4242 Err(Error::UnexpectedArg { token: b"zzz" }),
4243 "the second word must not reach `deep`"
4244 );
4245
4246 // Reached explicitly, the same command still takes it: the latch bounds routing, not
4247 // the tree.
4248 let a = argv(["lint", "deep", "zzz"]);
4249 assert_eq!(
4250 parse(&DEFAULTING, &a).unwrap(),
4251 vec![
4252 Event::Command(&RUN),
4253 Event::Command(&LINT),
4254 Event::Command(&DEEP),
4255 Event::Arg {
4256 arg: &RUN_TASK,
4257 value: b"zzz",
4258 delimit: true,
4259 },
4260 ]
4261 );
4262 }
4263
4264 #[test]
4265 fn a_flag_before_the_word_still_belongs_to_the_root() {
4266 // Routing happens at the word, so anything typed before it was addressed to the
4267 // command the user was actually at.
4268 let a = argv(["--verbose", "build"]);
4269 assert_eq!(
4270 parse(&DEFAULTING, &a).unwrap(),
4271 vec![
4272 Event::Flag {
4273 flag: &VERBOSE,
4274 value: None,
4275 negated: false
4276 },
4277 Event::Command(&RUN),
4278 Event::Arg {
4279 arg: &RUN_TASK,
4280 value: b"build",
4281 delimit: true,
4282 },
4283 ]
4284 );
4285 }
4286
4287 #[test]
4288 fn nothing_routes_after_the_separator() {
4289 // Past `--` there are no subcommands left to select, so there is no default to reach
4290 // either: the words are values of whatever the command declares.
4291 let a = argv(["--", "build"]);
4292 assert_eq!(
4293 parse(&DEFAULTING, &a).unwrap(),
4294 vec![Event::Arg {
4295 arg: &TASK,
4296 value: b"build",
4297 delimit: true,
4298 }]
4299 );
4300 }
4301
4302 #[test]
4303 fn globals_are_inherited_but_plain_flags_are_not() {
4304 let a = argv(["install", "--verbose"]);
4305 assert_eq!(
4306 parse(&ROOT, &a).unwrap(),
4307 vec![
4308 Event::Command(&INSTALL),
4309 Event::Flag {
4310 flag: &VERBOSE,
4311 value: None,
4312 negated: false
4313 }
4314 ]
4315 );
4316
4317 // `--jobs` belongs to the root and is not global, so it is not a flag here.
4318 // Strictly that is an unknown flag; leniently it is a word, and `install`
4319 // declares no argument to hold one — either way it is never read as the
4320 // root's flag, which is what this test is about.
4321 let a = argv(["install", "--jobs", "8"]);
4322 assert!(matches!(parse(&STRICT, &a), Err(Error::UnknownFlag { .. })));
4323 assert!(matches!(
4324 parse(&ROOT, &a),
4325 Err(Error::UnexpectedArg { token: b"--jobs" })
4326 ));
4327 }
4328
4329 #[test]
4330 fn double_dash_protects_flaglike_values() {
4331 let a = argv(["--", "--force", "-x"]);
4332 assert_eq!(
4333 parse(&ROOT, &a).unwrap(),
4334 vec![
4335 Event::Arg {
4336 arg: &FILE,
4337 value: b"--force",
4338 delimit: true,
4339 },
4340 Event::Arg {
4341 arg: &REST,
4342 value: b"-x",
4343 delimit: true,
4344 },
4345 ]
4346 );
4347 }
4348
4349 #[test]
4350 fn second_double_dash_is_a_value() {
4351 let a = argv(["--", "a", "--", "b"]);
4352 let values: Vec<&[u8]> = parse(&ROOT, &a)
4353 .unwrap()
4354 .iter()
4355 .filter_map(|e| match e {
4356 Event::Arg { value, .. } => Some(*value),
4357 _ => None,
4358 })
4359 .collect();
4360 assert_eq!(values, vec![&b"a"[..], &b"--"[..], &b"b"[..]]);
4361 }
4362
4363 #[test]
4364 fn allow_hyphen_values_takes_a_flaglike_detached_value() {
4365 static ARGS: Flag = Flag {
4366 key: 6,
4367 name: "args",
4368 longs: &["args"],
4369 shorts: b"a",
4370 takes_value: true,
4371 allow_hyphen_values: true,
4372 ..Flag::BOOL
4373 };
4374 static DIR: Flag = Flag {
4375 key: 7,
4376 name: "working-dir",
4377 longs: &["working-dir"],
4378 shorts: b"d",
4379 ..Flag::VALUE
4380 };
4381 static HYPHEN: Command = Command {
4382 name: "ex",
4383 flags: &[&ARGS, &DIR],
4384 args: &[&REST],
4385 ..Command::EMPTY
4386 };
4387
4388 let a = argv(["-a", "-destroy"]);
4389 assert_eq!(
4390 parse(&HYPHEN, &a).unwrap(),
4391 vec![Event::Flag {
4392 flag: &ARGS,
4393 value: Some(b"-destroy"),
4394 negated: false
4395 }]
4396 );
4397
4398 let a = argv(["--args", "--", "-x"]);
4399 assert_eq!(
4400 parse(&HYPHEN, &a).unwrap(),
4401 vec![
4402 Event::Flag {
4403 flag: &ARGS,
4404 value: Some(b"--"),
4405 negated: false
4406 },
4407 Event::Arg {
4408 arg: &REST,
4409 value: b"-x",
4410 delimit: true,
4411 },
4412 ]
4413 );
4414 }
4415
4416 #[test]
4417 fn require_equals_refuses_a_detached_value() {
4418 static INSPECT: Flag = Flag {
4419 key: 8,
4420 name: "inspect",
4421 longs: &["inspect"],
4422 shorts: b"i",
4423 takes_value: true,
4424 require_equals: true,
4425 ..Flag::BOOL
4426 };
4427 static EQ: Command = Command {
4428 name: "ex",
4429 flags: &[&INSPECT],
4430 ..Command::EMPTY
4431 };
4432
4433 let a = argv(["--inspect=9229"]);
4434 assert_eq!(
4435 parse(&EQ, &a).unwrap(),
4436 vec![Event::Flag {
4437 flag: &INSPECT,
4438 value: Some(b"9229"),
4439 negated: false
4440 }]
4441 );
4442
4443 let a = argv(["--inspect", "9229"]);
4444 assert!(matches!(
4445 parse(&EQ, &a),
4446 Err(Error::MissingFlagValue { .. })
4447 ));
4448
4449 let a = argv(["-i9229"]);
4450 assert_eq!(
4451 parse(&EQ, &a).unwrap(),
4452 vec![Event::Flag {
4453 flag: &INSPECT,
4454 value: Some(b"9229"),
4455 negated: false
4456 }]
4457 );
4458
4459 static ALL: Flag = Flag {
4460 key: 9,
4461 name: "all",
4462 longs: &["all"],
4463 shorts: b"a",
4464 ..Flag::BOOL
4465 };
4466 static BUNDLE: Command = Command {
4467 name: "ex",
4468 flags: &[&ALL, &INSPECT],
4469 ..Command::EMPTY
4470 };
4471 let a = argv(["-ai", "9229"]);
4472 assert!(
4473 matches!(parse(&BUNDLE, &a), Err(Error::MissingFlagValue { .. })),
4474 "a require_equals short reached through a bundle still refuses the following word"
4475 );
4476 }
4477
4478 #[test]
4479 fn default_missing_binds_when_the_value_is_left_off() {
4480 static COLOR: Flag = Flag {
4481 key: 9,
4482 name: "color",
4483 longs: &["color"],
4484 takes_value: true,
4485 default_missing: Some(b"always"),
4486 ..Flag::BOOL
4487 };
4488 static VERBOSE: Flag = Flag {
4489 key: 10,
4490 name: "verbose",
4491 longs: &["verbose"],
4492 ..Flag::BOOL
4493 };
4494 static MISSING: Command = Command {
4495 name: "ex",
4496 flags: &[&COLOR, &VERBOSE],
4497 ..Command::EMPTY
4498 };
4499
4500 let a = argv(["--color"]);
4501 assert_eq!(
4502 parse(&MISSING, &a).unwrap(),
4503 vec![Event::Flag {
4504 flag: &COLOR,
4505 value: Some(b"always"),
4506 negated: false
4507 }]
4508 );
4509
4510 let a = argv(["--color=never"]);
4511 assert_eq!(
4512 parse(&MISSING, &a).unwrap(),
4513 vec![Event::Flag {
4514 flag: &COLOR,
4515 value: Some(b"never"),
4516 negated: false
4517 }]
4518 );
4519
4520 let a = argv(["--color", "--verbose"]);
4521 assert_eq!(
4522 parse(&MISSING, &a).unwrap(),
4523 vec![
4524 Event::Flag {
4525 flag: &COLOR,
4526 value: Some(b"always"),
4527 negated: false
4528 },
4529 Event::Flag {
4530 flag: &VERBOSE,
4531 value: None,
4532 negated: false
4533 },
4534 ]
4535 );
4536
4537 let a = argv(["--color="]);
4538 assert_eq!(
4539 parse(&MISSING, &a).unwrap(),
4540 vec![Event::Flag {
4541 flag: &COLOR,
4542 value: Some(b""),
4543 negated: false
4544 }]
4545 );
4546 }
4547
4548 #[test]
4549 fn optional_flag_value_distinguishes_bare_and_explicit_forms() {
4550 static BUMP: Flag = Flag {
4551 key: 11,
4552 name: "bump",
4553 longs: &["bump"],
4554 takes_value: true,
4555 value_optional: true,
4556 ..Flag::BOOL
4557 };
4558 static OPTIONAL: Command = Command {
4559 name: "ex",
4560 flags: &[&BUMP],
4561 ..Command::EMPTY
4562 };
4563
4564 assert_eq!(parse(&OPTIONAL, &argv([])).unwrap(), vec![]);
4565 assert_eq!(
4566 parse(&OPTIONAL, &argv(["--bump"])).unwrap(),
4567 vec![Event::Flag {
4568 flag: &BUMP,
4569 value: None,
4570 negated: false,
4571 }]
4572 );
4573 assert_eq!(
4574 parse(&OPTIONAL, &argv(["--bump=5"])).unwrap(),
4575 vec![Event::Flag {
4576 flag: &BUMP,
4577 value: Some(b"5"),
4578 negated: false,
4579 }]
4580 );
4581
4582 static INCLUDE: Flag = Flag {
4583 key: 12,
4584 name: "include",
4585 longs: &["include"],
4586 takes_value: true,
4587 variadic: true,
4588 value_optional: true,
4589 ..Flag::BOOL
4590 };
4591 static VERBOSE: Flag = Flag {
4592 key: 13,
4593 name: "verbose",
4594 longs: &["verbose"],
4595 ..Flag::BOOL
4596 };
4597 static VARIADIC: Command = Command {
4598 name: "ex",
4599 flags: &[&INCLUDE, &VERBOSE],
4600 args: &[&REST],
4601 ..Command::EMPTY
4602 };
4603 assert_eq!(
4604 parse(&VARIADIC, &argv(["--include", "--verbose", "file"])).unwrap(),
4605 vec![
4606 Event::Flag {
4607 flag: &INCLUDE,
4608 value: None,
4609 negated: false,
4610 },
4611 Event::Flag {
4612 flag: &VERBOSE,
4613 value: None,
4614 negated: false,
4615 },
4616 Event::Arg {
4617 arg: &REST,
4618 value: b"file",
4619 delimit: true,
4620 },
4621 ]
4622 );
4623 }
4624
4625 #[test]
4626 fn default_missing_with_require_equals_leaves_the_following_word() {
4627 static INSPECT: Flag = Flag {
4628 key: 11,
4629 name: "inspect",
4630 longs: &["inspect"],
4631 takes_value: true,
4632 require_equals: true,
4633 default_missing: Some(b"9229"),
4634 ..Flag::BOOL
4635 };
4636 static BOTH: Command = Command {
4637 name: "ex",
4638 flags: &[&INSPECT],
4639 args: &[&REST],
4640 ..Command::EMPTY
4641 };
4642
4643 let a = argv(["--inspect"]);
4644 assert_eq!(
4645 parse(&BOTH, &a).unwrap(),
4646 vec![Event::Flag {
4647 flag: &INSPECT,
4648 value: Some(b"9229"),
4649 negated: false
4650 }]
4651 );
4652
4653 let a = argv(["--inspect", "80"]);
4654 assert_eq!(
4655 parse(&BOTH, &a).unwrap(),
4656 vec![
4657 Event::Flag {
4658 flag: &INSPECT,
4659 value: Some(b"9229"),
4660 negated: false
4661 },
4662 Event::Arg {
4663 arg: &REST,
4664 value: b"80",
4665 delimit: true,
4666 },
4667 ]
4668 );
4669
4670 let a = argv(["--inspect="]);
4671 assert_eq!(
4672 parse(&BOTH, &a).unwrap(),
4673 vec![Event::Flag {
4674 flag: &INSPECT,
4675 value: Some(b""),
4676 negated: false
4677 }]
4678 );
4679 }
4680
4681 #[test]
4682 fn variadic_flag_collects_until_a_flaglike_token() {
4683 static INCLUDE: Flag = Flag {
4684 key: 5,
4685 name: "include",
4686 longs: &["include"],
4687 shorts: b"i",
4688 takes_value: true,
4689 variadic: true,
4690 ..Flag::BOOL
4691 };
4692 static GREEDY: Command = Command {
4693 name: "ex",
4694 flags: &[&INCLUDE, &FORCE],
4695 args: &[&FILE],
4696 ..Command::EMPTY
4697 };
4698
4699 let a = argv(["--include", "x", "y", "--force"]);
4700 assert_eq!(
4701 parse(&GREEDY, &a).unwrap(),
4702 vec![
4703 Event::Flag {
4704 flag: &INCLUDE,
4705 value: Some(b"x"),
4706 negated: false
4707 },
4708 Event::Flag {
4709 flag: &INCLUDE,
4710 value: Some(b"y"),
4711 negated: false
4712 },
4713 Event::Flag {
4714 flag: &FORCE,
4715 value: None,
4716 negated: false
4717 },
4718 ]
4719 );
4720 }
4721
4722 #[test]
4723 fn value_terminators_end_variadic_owners_without_binding() {
4724 static INCLUDE: Flag = Flag {
4725 key: 92,
4726 name: "include",
4727 longs: &["include"],
4728 takes_value: true,
4729 variadic: true,
4730 value_terminator: Some(b";"),
4731 ..Flag::BOOL
4732 };
4733 static ITEMS: Arg = Arg {
4734 key: 93,
4735 name: "items",
4736 var: true,
4737 value_terminator: Some(b";"),
4738 ..Arg::REQUIRED
4739 };
4740 static AFTER: Arg = Arg {
4741 key: 94,
4742 name: "after",
4743 ..Arg::REQUIRED
4744 };
4745 static FLAG_CMD: Command = Command {
4746 name: "ex",
4747 flags: &[&INCLUDE],
4748 args: &[&AFTER],
4749 ..Command::EMPTY
4750 };
4751 static ARG_CMD: Command = Command {
4752 name: "ex",
4753 args: &[&ITEMS, &AFTER],
4754 ..Command::EMPTY
4755 };
4756
4757 let flag = argv(["--include", "a", ";", "tail"]);
4758 assert_eq!(
4759 parse(&FLAG_CMD, &flag).unwrap(),
4760 vec![
4761 Event::Flag {
4762 flag: &INCLUDE,
4763 value: Some(b"a"),
4764 negated: false,
4765 },
4766 Event::Arg {
4767 arg: &AFTER,
4768 value: b"tail",
4769 delimit: true,
4770 },
4771 ]
4772 );
4773
4774 let positional = argv(["a", ";", "tail"]);
4775 assert_eq!(
4776 parse(&ARG_CMD, &positional).unwrap(),
4777 vec![
4778 Event::Arg {
4779 arg: &ITEMS,
4780 value: b"a",
4781 delimit: true,
4782 },
4783 Event::Arg {
4784 arg: &AFTER,
4785 value: b"tail",
4786 delimit: true,
4787 },
4788 ]
4789 );
4790 }
4791
4792 #[test]
4793 fn a_non_variadic_flag_leaves_the_next_word_alone() {
4794 // The counterpart to the test above: a flag that takes one value must not
4795 // swallow the word after it, which would silently steal a positional.
4796 let a = argv(["--jobs", "8", "keep-me"]);
4797 assert_eq!(
4798 parse(&ROOT, &a).unwrap(),
4799 vec![
4800 Event::Flag {
4801 flag: &JOBS,
4802 value: Some(b"8"),
4803 negated: false
4804 },
4805 Event::Arg {
4806 arg: &FILE,
4807 value: b"keep-me",
4808 delimit: true,
4809 },
4810 ]
4811 );
4812 }
4813
4814 #[test]
4815 fn double_dash_seen_means_a_separator_was_typed() {
4816 static FILES: Arg = Arg {
4817 key: 23,
4818 name: "files",
4819 double_dash: DoubleDash::Automatic,
4820 ..Arg::VAR
4821 };
4822 static AUTO: Command = Command {
4823 name: "ex",
4824 flags: &[&FORCE],
4825 args: &[&FILES],
4826 ..Command::EMPTY
4827 };
4828
4829 let a = argv(["--", "x"]);
4830 let mut parser = Parser::new(&ROOT, &a);
4831 while parser.next_event().is_some() {}
4832 assert!(parser.double_dash_seen(), "a real separator was consumed");
4833
4834 // `automatic` stops flag interpretation without a separator being typed,
4835 // and reporting one would be a lie to any caller that forwards argv.
4836 let a = argv(["x", "--force"]);
4837 let mut parser = Parser::new(&AUTO, &a);
4838 while parser.next_event().is_some() {}
4839 assert!(
4840 !parser.double_dash_seen(),
4841 "automatic mode must not claim a separator was given"
4842 );
4843 }
4844
4845 #[test]
4846 fn a_wrapper_still_forwards_a_help_flag() {
4847 // Supplying `--help` must not take the two forwarding mechanisms away from a wrapper,
4848 // which is the one place a CLI means to hand the token on rather than answer it.
4849 static ARGS: Arg = Arg {
4850 key: 24,
4851 name: "args",
4852 ..Arg::VAR
4853 };
4854 static WRAP: Command = Command {
4855 name: "wrap",
4856 args: &[&ARGS],
4857 ..Command::EMPTY
4858 };
4859
4860 // A typed separator: everything after it is a value, `--help` included.
4861 let a = argv(["--", "--help", "-h"]);
4862 assert_eq!(
4863 parse(&WRAP, &a).unwrap(),
4864 vec![
4865 Event::Arg {
4866 arg: &ARGS,
4867 value: b"--help",
4868 delimit: true,
4869 },
4870 Event::Arg {
4871 arg: &ARGS,
4872 value: b"-h",
4873 delimit: true,
4874 },
4875 ]
4876 );
4877
4878 // And `automatic`, for the wrapper whose caller should not have to type one: the
4879 // first value stops flag interpretation, so the flags after it forward.
4880 static AUTO_ARGS: Arg = Arg {
4881 key: 25,
4882 name: "args",
4883 double_dash: DoubleDash::Automatic,
4884 ..Arg::VAR
4885 };
4886 static AUTO_WRAP: Command = Command {
4887 name: "wrap",
4888 args: &[&AUTO_ARGS],
4889 ..Command::EMPTY
4890 };
4891
4892 let a = argv(["node", "--help"]);
4893 assert_eq!(
4894 parse(&AUTO_WRAP, &a).unwrap(),
4895 vec![
4896 Event::Arg {
4897 arg: &AUTO_ARGS,
4898 value: b"node",
4899 delimit: true,
4900 },
4901 Event::Arg {
4902 arg: &AUTO_ARGS,
4903 value: b"--help",
4904 delimit: true,
4905 },
4906 ]
4907 );
4908
4909 // Before either takes effect, though, the wrapper's own help is what `--help` asks
4910 // for — `mise run --help` is a question about `run`, not a value for it.
4911 let a = argv(["--help"]);
4912 assert_eq!(
4913 parse(&AUTO_WRAP, &a).unwrap(),
4914 vec![Event::Flag {
4915 flag: &HELP_LONG,
4916 value: None,
4917 negated: false
4918 }]
4919 );
4920 }
4921
4922 #[test]
4923 fn double_dash_required_arg() {
4924 static CMD: Arg = Arg {
4925 key: 20,
4926 name: "cmd",
4927 double_dash: DoubleDash::Required,
4928 ..Arg::REQUIRED
4929 };
4930 static EXEC: Command = Command {
4931 name: "ex",
4932 args: &[&CMD],
4933 ..Command::EMPTY
4934 };
4935
4936 let a = argv(["--", "ls"]);
4937 assert_eq!(
4938 parse(&EXEC, &a).unwrap(),
4939 vec![Event::Arg {
4940 arg: &CMD,
4941 value: b"ls",
4942 delimit: true,
4943 }]
4944 );
4945
4946 let a = argv(["ls"]);
4947 assert_eq!(
4948 parse(&EXEC, &a),
4949 Err(Error::ArgRequiresDoubleDash { arg: &CMD })
4950 );
4951 }
4952
4953 #[test]
4954 fn double_dash_preserve_keeps_the_separator() {
4955 static ARGS: Arg = Arg {
4956 key: 21,
4957 name: "args",
4958 double_dash: DoubleDash::Preserve,
4959 ..Arg::VAR
4960 };
4961 static WRAP: Command = Command {
4962 name: "ex",
4963 args: &[&ARGS],
4964 ..Command::EMPTY
4965 };
4966
4967 let a = argv(["a", "--", "b"]);
4968 let values: Vec<&[u8]> = parse(&WRAP, &a)
4969 .unwrap()
4970 .iter()
4971 .filter_map(|e| match e {
4972 Event::Arg { value, .. } => Some(*value),
4973 _ => None,
4974 })
4975 .collect();
4976 assert_eq!(values, vec![&b"a"[..], &b"--"[..], &b"b"[..]]);
4977 }
4978
4979 #[test]
4980 fn double_dash_automatic_stops_flag_interpretation() {
4981 static FILES: Arg = Arg {
4982 key: 22,
4983 name: "files",
4984 double_dash: DoubleDash::Automatic,
4985 ..Arg::VAR
4986 };
4987 static AUTO: Command = Command {
4988 name: "ex",
4989 flags: &[&FORCE],
4990 args: &[&FILES],
4991 ..Command::EMPTY
4992 };
4993
4994 // The flag before the first value is still a flag; the one after it is a
4995 // value.
4996 let a = argv(["-f", "one", "--force"]);
4997 assert_eq!(
4998 parse(&AUTO, &a).unwrap(),
4999 vec![
5000 Event::Flag {
5001 flag: &FORCE,
5002 value: None,
5003 negated: false
5004 },
5005 Event::Arg {
5006 arg: &FILES,
5007 value: b"one",
5008 delimit: true,
5009 },
5010 Event::Arg {
5011 arg: &FILES,
5012 value: b"--force",
5013 delimit: true,
5014 },
5015 ]
5016 );
5017 }
5018
5019 #[test]
5020 fn too_many_words() {
5021 static ONE: Command = Command {
5022 name: "ex",
5023 args: &[&FILE],
5024 ..Command::EMPTY
5025 };
5026 let a = argv(["a", "b"]);
5027 assert_eq!(parse(&ONE, &a), Err(Error::UnexpectedArg { token: b"b" }));
5028 }
5029
5030 #[test]
5031 fn unknown_letter_rejects_the_whole_bundle() {
5032 // `-f` is real and `-z` is not. The first event must be the error: if the
5033 // flag event came out first, a caller would have applied `-f` from a
5034 // command line that was rejected.
5035 let a = argv(["-fz"]);
5036 let mut parser = Parser::new(&STRICT, &a);
5037 assert_eq!(
5038 parser.next_event(),
5039 Some(Err(Error::UnknownFlag { token: b"-fz" })),
5040 "an unknown letter must reject the token before any of it is applied"
5041 );
5042 assert!(parser.next_event().is_none());
5043
5044 // Leniently, the same token is a value — and `-f` is *not* applied, since
5045 // the token was never a bundle at all.
5046 let a = argv(["-fz"]);
5047 assert_eq!(
5048 parse(&ROOT, &a).unwrap(),
5049 vec![Event::Arg {
5050 arg: &FILE,
5051 value: b"-fz",
5052 delimit: true,
5053 }]
5054 );
5055 }
5056
5057 #[test]
5058 fn unknown_short_error_names_the_whole_token() {
5059 for (tokens, want) in [(["-z"], &b"-z"[..]), (["-fz"], &b"-fz"[..])] {
5060 let a = argv(tokens);
5061 assert_eq!(
5062 parse(&STRICT, &a),
5063 Err(Error::UnknownFlag { token: want }),
5064 "{tokens:?}"
5065 );
5066 }
5067 }
5068
5069 #[test]
5070 fn errors_are_terminal() {
5071 let a = argv(["--wat", "--force"]);
5072 let mut parser = Parser::new(&STRICT, &a);
5073 assert!(parser.next_event().unwrap().is_err());
5074 assert!(parser.next_event().is_none());
5075 }
5076
5077 #[test]
5078 fn non_utf8_values_still_parse() {
5079 // A value that is not valid UTF-8 binds; only converting it fails, and
5080 // only if a caller asks.
5081 let raw = OsStr::new("--force");
5082 let a = [raw];
5083 assert!(parse(&ROOT, &a).is_ok());
5084
5085 assert!(as_str(b"ok").is_ok());
5086 assert!(as_str(&[0xff, 0xfe]).is_err());
5087 }
5088
5089 #[test]
5090 fn a_multicall_applet_is_the_basename_unless_it_is_the_dispatcher() {
5091 assert_eq!(multicall_basename("/usr/bin/ls"), "ls");
5092 assert_eq!(multicall_basename(r"C:\busybox\ls.exe"), "ls");
5093 assert_eq!(
5094 multicall_applet("/usr/bin/ls", "busybox", Some("busybox")),
5095 Some("ls")
5096 );
5097 assert_eq!(
5098 multicall_applet("/usr/bin/busybox", "busybox", Some("busybox")),
5099 None
5100 );
5101 assert_eq!(
5102 multicall_applet("ls.exe", "busybox", Some("busybox")),
5103 Some("ls")
5104 );
5105 assert_eq!(
5106 multicall_applet("/usr/bin/busybox", "BusyBox", Some("/opt/bin/busybox")),
5107 None
5108 );
5109 assert_eq!(
5110 multicall_applet("busybox.exe", "BusyBox", Some("busybox.exe")),
5111 None
5112 );
5113 }
5114
5115 #[test]
5116 fn a_spec_request_is_the_first_word_and_nothing_else() {
5117 let request = [OsStr::new(SPEC_REQUEST)];
5118 assert!(is_spec_request(&ROOT, &request));
5119
5120 // Anywhere but the front it is an ordinary value, which is what makes the endpoint
5121 // safe for a CLI whose arguments are arbitrary text.
5122 let later = ["install", SPEC_REQUEST].map(OsStr::new);
5123 assert!(!is_spec_request(&ROOT, &later));
5124 assert!(!is_spec_request(&ROOT, &[]));
5125 assert!(!is_spec_request(&ROOT, &[OsStr::new("--help")]));
5126 }
5127
5128 #[test]
5129 fn a_declared_command_of_that_name_keeps_it() {
5130 static DECLARED: Command = Command {
5131 name: SPEC_REQUEST,
5132 key: 200,
5133 ..Command::EMPTY
5134 };
5135 static ALIASED: Command = Command {
5136 name: "describe",
5137 aliases: &[SPEC_REQUEST],
5138 key: 201,
5139 ..Command::EMPTY
5140 };
5141 static DECLARES_IT: Command = Command {
5142 name: "ex",
5143 subcommands: &[&DECLARED],
5144 ..Command::EMPTY
5145 };
5146 static ALIASES_IT: Command = Command {
5147 name: "ex",
5148 subcommands: &[&ALIASED],
5149 ..Command::EMPTY
5150 };
5151
5152 let request = [OsStr::new(SPEC_REQUEST)];
5153 assert!(!is_spec_request(&DECLARES_IT, &request));
5154 // An alias selects a command just as its name does, so it wins here too.
5155 assert!(!is_spec_request(&ALIASES_IT, &request));
5156 }
5157}