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
104/// Print a generated entry point's output to stdout, ignoring a failed write.
105///
106/// `print!` panics when the write fails, and the usual failure is a reader that has gone
107/// away: `cli --help | head -1`, a pager quit early, or a completion the shell cancelled while
108/// typing. Every caller is about to leave (or has finished a successful parse), so there is
109/// nobody left to tell, and a panic only turns that into a crash — an abort with a core dump
110/// in a `panic = "abort"` build. The exit status stays the one the caller chose.
111#[doc(hidden)]
112pub fn __usage_print(output: std::fmt::Arguments<'_>) {
113 use std::io::Write;
114 let mut stdout = std::io::stdout().lock();
115 let _ = stdout.write_fmt(output).and_then(|()| stdout.flush());
116}
117
118/// Build a command's future on the heap, for a generated async dispatch to await.
119///
120/// The future is created inside this function, not by the caller, and that is the point:
121/// without optimizations every temporary in an `async fn` gets its own stack slot, held for
122/// as long as the future runs, so a dispatch that wrote `run_async(inner).await` in each arm
123/// kept a slot the size of every command's future on the stack while running any one of them.
124/// Created here, the future only passes through this function's frame, and the dispatch keeps
125/// a `Box` per arm instead. `Send` still follows the command's future, since a box of it is
126/// `Send` exactly when it is.
127#[doc(hidden)]
128pub fn __usage_box_future<F, Fut>(make: F) -> std::pin::Pin<Box<Fut>>
129where
130 F: FnOnce() -> Fut,
131 Fut: std::future::Future,
132{
133 Box::pin(make())
134}
135
136/// [`__usage_print`] for stderr: help and failures that go there, and deprecation warnings.
137#[doc(hidden)]
138pub fn __usage_eprint(output: std::fmt::Arguments<'_>) {
139 use std::io::Write;
140 let _ = std::io::stderr().lock().write_fmt(output);
141}
142
143use std::ffi::{OsStr, OsString};
144
145/// A value's shell-native completion class for `#[usage(value_hint = ...)]`.
146///
147/// This lives in the runtime crate so a declaration never needs clap merely to describe what
148/// kind of path a shell should offer. It is metadata only and adds no work to a successful
149/// parse.
150#[derive(Clone, Copy, Debug, Eq, PartialEq)]
151pub enum ValueHint {
152 /// Let the shell use its normal fallback behavior.
153 Unknown,
154 /// No structured hint applies; suppress the shell's path fallback.
155 Other,
156 /// A path to a file.
157 FilePath,
158 /// A path to either a file or a directory.
159 AnyPath,
160 /// A path to a directory.
161 DirPath,
162 /// A path to an executable file.
163 ExecutablePath,
164 /// A command name, resolved through the shell's command table and `PATH`.
165 CommandName,
166 /// One string containing a command and any arguments.
167 CommandString,
168 /// A trailing argv vector: complete the first value as a command, then its arguments.
169 CommandWithArguments,
170 /// A local operating-system user name.
171 Username,
172 /// A host name known to the shell or operating system.
173 Hostname,
174 /// A web address. This suppresses path fallback but offers no finite candidate set.
175 Url,
176 /// An email address. This suppresses path fallback but offers no finite candidate set.
177 EmailAddress,
178}
179
180#[cfg(feature = "complete")]
181pub mod complete;
182#[cfg(feature = "diagnostics")]
183pub mod diagnostic;
184#[cfg(feature = "spec")]
185pub mod embedded;
186#[cfg(feature = "complete")]
187pub mod install;
188#[cfg(feature = "complete")]
189pub mod script;
190
191/// Checks that the `complete` feature is on, with an explanation when it is not.
192///
193/// `#[usage(completion)]` generates code that reaches into [`complete`], which is behind a
194/// feature the *depending* crate enables — a derive cannot turn on a feature of another crate.
195/// Without this, the failure was `unresolved module complete`, which says nothing about the
196/// attribute that caused it.
197#[cfg(feature = "complete")]
198#[macro_export]
199macro_rules! __usage_needs_complete_feature {
200 () => {};
201}
202
203/// See [`__usage_needs_complete_feature`].
204#[cfg(not(feature = "complete"))]
205#[macro_export]
206macro_rules! __usage_needs_complete_feature {
207 () => {
208 ::core::compile_error!(
209 "`#[usage(completion)]` needs usage-argv's `complete` feature. Add it where \
210 usage-argv is depended on: usage-argv = { version = \"…\", features = \
211 [\"spec\", \"complete\"] }"
212 );
213 };
214}
215#[cfg(feature = "spec")]
216pub mod help;
217#[cfg(feature = "spec")]
218mod order;
219// Behind no feature: two traits and no code, so there is nothing here for a binary that
220// does not dispatch to pay for, and a hand-written CLI on the bare runtime can use them.
221pub mod run;
222// Behind no feature either, and public: an application that lays out its own output beside a
223// help page — a table, a progress line — should not have to work the width out a second way
224// and disagree with the page it sits next to.
225#[cfg(feature = "spec")]
226pub mod spec;
227pub mod tty;
228#[cfg(feature = "spec")]
229pub use spec::{parse_args_from, parse_args_from_argv};
230#[cfg(feature = "spec")]
231pub mod warn;
232
233pub use run::{Run, RunAsync, RunAsyncWith, RunWith};
234
235/// How deep a command tree this parser will descend.
236///
237/// The ancestor chain is kept in a fixed-size array so that a parse allocates
238/// nothing; this is that array's size. mise, the largest usage CLI, is four
239/// levels deep.
240pub const MAX_DEPTH: usize = 16;
241
242/// A command: its flags, its positional arguments, and its subcommands.
243///
244/// Every field is a borrowed slice so that a derive can emit the whole tree as
245/// `static` data. Use `..Command::EMPTY` to fill in the parts you do not need.
246#[derive(Debug, Clone, Copy, PartialEq, Eq)]
247pub struct Command<'a> {
248 /// The canonical name, used to select this command.
249 pub name: &'a str,
250 /// Alternative names that also select it.
251 pub aliases: &'a [&'a str],
252 pub flags: &'a [&'a Flag<'a>],
253 /// Positional arguments, in the order they are filled.
254 pub args: &'a [&'a Arg<'a>],
255 /// A repeatable group of scoped flags and positional arguments, if this command has one.
256 ///
257 /// Behind a reference because hardly any command has one: held inline, the clause's five
258 /// slices were two thirds of every command's table whether it declared one or not.
259 pub clause: ::core::option::Option<&'a Clause<'a>>,
260 pub subcommands: &'a [&'a Command<'a>],
261 /// Where a word goes when it names no subcommand of this one.
262 ///
263 /// The spec's `default_subcommand`. `mise build` means `mise run build`: the word names
264 /// no command, so the parser descends into `run` and lets *`run`* have it — even where
265 /// this command declares an argument of its own, which is what makes the property worth
266 /// having rather than a synonym for a positional.
267 ///
268 /// Applied at most once per parse, so a CLI cannot loop through it, and only where a
269 /// subcommand could still be selected.
270 ///
271 /// Resolve it with [`find_subcommand`], which turns a name that no subcommand answers to
272 /// into a compile error.
273 pub default_subcommand: ::core::option::Option<&'a Command<'a>>,
274 /// Look ahead past parent/default flags before implicitly selecting the default.
275 /// A sibling name before any default-only flag stays on the parent; after one,
276 /// later words are the default command's args.
277 pub default_subcommand_flags: bool,
278 /// Select the default subcommand for an otherwise empty successful invocation.
279 pub default_subcommand_on_empty: bool,
280 /// Whether an unmatched word is forwarded as an external command plus the rest of argv.
281 ///
282 /// clap's `allow_external_subcommands`. Known subcommands still win; a
283 /// [`default_subcommand`](Self::default_subcommand) still catches first. Once the
284 /// unmatched word is taken, remaining tokens — including `--help` — are not parsed
285 /// as this command's flags.
286 pub external_subcommand: bool,
287 /// Show this command's help when no argv token follows its name.
288 ///
289 /// This is clap's `arg_required_else_help`. It deliberately observes argv rather than
290 /// bound values: an environment variable or default may fill a field, but neither means
291 /// the user supplied an argument to this invocation.
292 pub arg_required_else_help: bool,
293 /// Selecting a subcommand suppresses this command's required arguments.
294 pub subcommand_negates_reqs: bool,
295 /// Once this command binds a flag or positional, selecting one of its
296 /// subcommands is an error.
297 pub args_conflicts_with_subcommands: bool,
298 /// Let a known subcommand interrupt a variadic argument that would otherwise consume it.
299 pub subcommand_precedence_over_arg: bool,
300 /// Let a later required positional take a word while an earlier optional positional
301 /// remains empty.
302 pub allow_missing_positional: bool,
303 /// Disable delimiter splitting for positional values after `--` or on an
304 /// automatic trailing argument. Inherited by subcommands.
305 pub dont_delimit_trailing_values: bool,
306 /// What an unrecognized flag-like token means here, or `None` to keep whatever the
307 /// enclosing command said. See [`UnknownFlags`].
308 ///
309 /// Inherited rather than resolved per command, which is what usage-lib does — its
310 /// `effective_unknown_flags` walks outward from the command that ran and falls back to
311 /// the spec's. Resolving it in the tables instead was possible only for a builder that
312 /// can see the whole tree: a derive expands one struct at a time and cannot see its
313 /// parent, so `#[usage(unknown_flags = "error")]` on the root reached the root alone and
314 /// a subcommand had no way to say it at all.
315 ///
316 /// The parser carries the effective value down as it descends, so a command that states
317 /// nothing costs nothing.
318 pub unknown_flags: ::core::option::Option<UnknownFlags>,
319 /// Whether this command answers to `--version` and `-V`.
320 ///
321 /// Set on the root, and only when the CLI declares a version: clap adds the flag exactly
322 /// then, and a `--version` that answers with nothing is worse than one that is not there.
323 /// A field rather than a rule about depth, so a CLI that wants it on a subcommand — clap's
324 /// `propagate_version` — has somewhere to say so.
325 pub version: bool,
326 /// Do not synthesize `--help` and `-h` for this command.
327 pub disable_help_flag: bool,
328 /// Do not synthesize the `help` subcommand route for this command.
329 pub disable_help_subcommand: bool,
330 /// Do not synthesize `--version` and `-V` for this command.
331 pub disable_version_flag: bool,
332 /// Caller-assigned identifier, echoed back in [`Event::Command`].
333 ///
334 /// Wide enough for a derive to make these unique without coordination: two
335 /// macro expansions cannot see each other, so the generated keys carry a hash
336 /// of the type they came from in the high half and a per-type index in the low
337 /// half. A parse dispatches on this, so a collision would bind the wrong field
338 /// — [`Spec::to_kdl`](crate::spec::Spec::to_kdl) checks the tree for duplicates
339 /// in debug builds.
340 pub key: u64,
341}
342
343impl Command<'_> {
344 /// A command with nothing declared, for use with struct update syntax.
345 pub const EMPTY: Command<'static> = Command {
346 name: "",
347 aliases: &[],
348 flags: &[],
349 args: &[],
350 clause: ::core::option::Option::None,
351 subcommands: &[],
352 default_subcommand: ::core::option::Option::None,
353 default_subcommand_flags: false,
354 default_subcommand_on_empty: false,
355 external_subcommand: false,
356 arg_required_else_help: false,
357 subcommand_negates_reqs: false,
358 args_conflicts_with_subcommands: false,
359 subcommand_precedence_over_arg: false,
360 allow_missing_positional: false,
361 dont_delimit_trailing_values: false,
362 unknown_flags: ::core::option::Option::None,
363 version: false,
364 disable_help_flag: false,
365 disable_help_subcommand: false,
366 disable_version_flag: false,
367 key: 0,
368 };
369}
370
371/// A repeatable group of scoped flags and positional arguments.
372#[derive(Debug, Clone, Copy, PartialEq, Eq)]
373pub struct Clause<'a> {
374 pub key: u64,
375 pub name: &'a str,
376 /// An explicit token that starts the next instance. When absent, consuming the
377 /// clause's sole required positional completes the current instance.
378 pub separator: ::core::option::Option<&'a [u8]>,
379 pub flags: &'a [&'a Flag<'a>],
380 pub args: &'a [&'a Arg<'a>],
381}
382
383/// Basename of argv[0] for a multicall CLI: last path component, with a trailing
384/// `.exe` stripped so Windows and Unix agree.
385pub fn multicall_basename(argv0: &str) -> &str {
386 let name = argv0.rsplit(['/', '\\']).next().unwrap_or(argv0);
387 match name.get(name.len().saturating_sub(4)..) {
388 Some(ext) if ext.eq_ignore_ascii_case(".exe") => &name[..name.len() - 4],
389 _ => name,
390 }
391}
392
393/// The applet name to parse as the first word, when argv[0] is not the dispatcher.
394///
395/// `None` means a dispatcher invocation (`busybox ls`): skip argv[0] and parse the
396/// rest. `Some` is a symlink invocation (`ls -l`): inject the basename.
397pub fn multicall_applet<'a>(argv0: &'a str, name: &str, bin: Option<&str>) -> Option<&'a str> {
398 let base = multicall_basename(argv0);
399 if !name.is_empty() && base == multicall_basename(name) {
400 return None;
401 }
402 if let Some(bin) = bin
403 && !bin.is_empty()
404 && base == multicall_basename(bin)
405 {
406 return None;
407 }
408 Some(base)
409}
410
411/// Resolved identity of a derive-generated binding type.
412#[derive(Clone, Copy)]
413pub struct BindingType(pub fn() -> &'static str);
414
415impl BindingType {
416 pub fn name(self) -> &'static str {
417 (self.0)()
418 }
419}
420
421impl ::core::fmt::Debug for BindingType {
422 fn fmt(&self, f: &mut ::core::fmt::Formatter<'_>) -> ::core::fmt::Result {
423 f.debug_tuple("BindingType").field(&self.name()).finish()
424 }
425}
426
427impl PartialEq for BindingType {
428 fn eq(&self, other: &Self) -> bool {
429 self.name() == other.name()
430 }
431}
432
433impl Eq for BindingType {}
434
435/// A flag, addressed by any of its long or short forms.
436#[derive(Debug, Clone, Copy, PartialEq, Eq)]
437pub struct Flag<'a> {
438 /// Caller-assigned identifier, echoed back in [`Event::Flag`]. This is how
439 /// generated code knows which field to assign without any string comparison.
440 /// See [`Command::key`] on why it is this wide.
441 pub key: u64,
442 /// Compatibility key for mirroring a redeclared child global into an ancestor field.
443 ///
444 /// Zero means no typed binding contract is declared. Derive-generated tables hash the
445 /// binding shape and portable metadata so only equivalent bindings receive the same event.
446 pub binding_key: u64,
447 /// Resolved Rust value type for a derive-generated binding.
448 ///
449 /// This is separate from [`Self::binding_key`] because token spellings are not type
450 /// identities: an imported alias and a fully qualified path can name the same type.
451 pub binding_type: Option<BindingType>,
452 /// Unused by binding, kept so a table entry can carry its own name for
453 /// diagnostics.
454 pub name: &'a str,
455 /// Long forms, written without the leading `--`.
456 pub longs: &'a [&'a str],
457 /// Short forms, as single bytes.
458 ///
459 /// **Should be ASCII.** A cluster like `-xyz` is walked one byte at a time, so a
460 /// non-ASCII short can never be matched, and the remainder after a value-taking one —
461 /// which becomes its value — would begin in the middle of a character.
462 /// `#[derive(Cli)]` rejects a non-ASCII `short`; a table written by hand should keep to
463 /// it. Nothing is unsound if it does not: the value would simply be cut in a place that
464 /// makes no sense, and on Windows would then fail to convert.
465 pub shorts: &'a [u8],
466 /// A long form that sets the flag to false, written without the `--`.
467 pub negate: Option<&'a str>,
468 /// Whether the flag takes a value.
469 pub takes_value: bool,
470 /// Whether one occurrence of this flag keeps taking values, until a flag-like
471 /// token or the end of the command line.
472 ///
473 /// This is the spec's variadic flag *argument* (`--include <pattern>...`). It
474 /// is not the spec's flag-level `var=#true`, which means the flag may be
475 /// repeated and takes one value each time — repetition needs nothing from the
476 /// parser, since it already reports every occurrence separately. Conflating
477 /// the two makes a merely repeatable flag greedy enough to eat a positional.
478 pub variadic: bool,
479 /// How many values one variadic occurrence may take, after which the next word
480 /// belongs to whatever comes next.
481 ///
482 /// Only for [`variadic`](Self::variadic). A merely repeatable flag — the spec's
483 /// `var=#true` — is bounded on how many times it was *given*, which no single token
484 /// can decide, so that bound stays with the metadata and is checked after the parse.
485 pub var_max: ::core::option::Option<u32>,
486 /// The byte that makes one word several values, if the flag declares one.
487 ///
488 /// Here rather than with the metadata for the same reason [`var_max`](Self::var_max)
489 /// is: it decides *where* a word lands. A bound counts values, and a delimiter is what
490 /// makes a word stop being one of them — `--include a,b,c` is three, so a `var_max` of
491 /// two is already past its bound on the single word it was entitled to take. Binding
492 /// cannot count without it.
493 pub delimiter: ::core::option::Option<u8>,
494 /// Whether a detached value may itself look like a flag.
495 ///
496 /// The default is to refuse: `--jobs --force` is far more likely a forgotten
497 /// value than a jobs of `"--force"`. Declared, the next token is taken
498 /// whatever it looks like — including `--` — which is clap's
499 /// `allow_hyphen_values` and the spec's property of the same name. A variadic
500 /// occurrence still stops collecting at a later flag-like token, so a second
501 /// occurrence of the flag is not eaten as a value.
502 pub allow_hyphen_values: bool,
503 /// Whether a detached value may be a negative number while other flag-like
504 /// tokens still stop collection or report as flags.
505 pub allow_negative_numbers: bool,
506 /// A token that ends one variadic occurrence without becoming a value.
507 pub value_terminator: ::core::option::Option<&'a [u8]>,
508 /// Whether the value must be attached with `=`.
509 ///
510 /// `--flag=value` is accepted and `--flag value` is not, which is clap's
511 /// `require_equals` and the spec's property of the same name. A short's
512 /// attached form (`-i9229`, `-i=9229`) still binds: only the following word
513 /// is refused.
514 pub require_equals: bool,
515 /// Whether this value-taking flag may be present without a value.
516 ///
517 /// A missing value emits the flag event with `value: None`; bindings such as
518 /// `Option<Option<T>>` can therefore distinguish an absent flag from a bare
519 /// flag and from a flag with an explicit value.
520 pub value_optional: bool,
521 /// Whether a boolean long flag accepts an attached `true` or `false` value.
522 ///
523 /// This does not make the flag value-taking in the ordinary sense: a detached
524 /// word is never consumed, and help keeps rendering a switch. Only
525 /// `--flag=true` and `--flag=false` opt into an explicit boolean value.
526 pub bool_value: bool,
527 /// Value used when the flag is present but no value is given.
528 ///
529 /// clap's `default_missing_value` and the spec's `default_missing`. `--color`
530 /// binds this, `--color=never` binds `never`, and an absent flag is not bound.
531 /// Combined with [`Self::require_equals`], a following word is still refused.
532 pub default_missing: ::core::option::Option<&'a [u8]>,
533 /// Whether the flag is recognized by every command beneath the one that
534 /// declares it.
535 pub global: bool,
536 /// Whether this declared flag binds a field or requests a built-in response.
537 pub action: ArgAction,
538}
539
540impl Flag<'_> {
541 /// A value-less flag, for use with struct update syntax.
542 pub const BOOL: Flag<'static> = Flag {
543 key: 0,
544 binding_key: 0,
545 binding_type: None,
546 name: "",
547 longs: &[],
548 shorts: &[],
549 negate: None,
550 takes_value: false,
551 variadic: false,
552 var_max: ::core::option::Option::None,
553 delimiter: ::core::option::Option::None,
554 allow_hyphen_values: false,
555 allow_negative_numbers: false,
556 value_terminator: ::core::option::Option::None,
557 require_equals: false,
558 value_optional: false,
559 bool_value: false,
560 default_missing: ::core::option::Option::None,
561 global: false,
562 action: ArgAction::Set,
563 };
564
565 /// A flag that takes a value, for use with struct update syntax.
566 pub const VALUE: Flag<'static> = Flag {
567 takes_value: true,
568 ..Flag::BOOL
569 };
570}
571
572/// What supplying a declared flag does.
573#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
574pub enum ArgAction {
575 /// Bind the flag to its declared field.
576 #[default]
577 Set,
578 /// Show help, choosing the long form for a long spelling and the short form otherwise.
579 Help,
580 /// Always show short help.
581 HelpShort,
582 /// Always show long help.
583 HelpLong,
584 /// Show long help for this command and every visible descendant.
585 HelpAll,
586 /// Show version information.
587 Version,
588}
589
590/// A positional argument.
591#[derive(Debug, Clone, Copy, PartialEq, Eq)]
592pub struct Arg<'a> {
593 /// Caller-assigned identifier, echoed back in [`Event::Arg`]. See
594 /// [`Command::key`] on why it is this wide.
595 pub key: u64,
596 /// Prefix that classifies this positional independently of declaration order.
597 /// The prefix is removed from the value emitted in [`Event::Arg`].
598 pub sigil: ::core::option::Option<&'a [u8]>,
599 /// Whether post-binding requires this positional to have a value. Kept in the hot
600 /// table because `allow_missing_positional` must reserve words for later required args.
601 pub required: bool,
602 /// Whether this argument keeps taking values once it has one.
603 pub var: bool,
604 /// How many words a variadic may take before the next argument gets the rest.
605 ///
606 /// A bound belongs here, in the table binding reads, rather than with the metadata:
607 /// it decides *where* a word lands, not whether what landed is acceptable. clap's
608 /// `num_args` works the same way, and every spec in the wild is generated from a clap
609 /// command. `u32` rather than `usize` because a CLI that bounds a variadic above four
610 /// billion has other problems, and this table is read on the hot path.
611 pub var_max: ::core::option::Option<u32>,
612 /// The byte that makes one word several values, if the argument declares one.
613 ///
614 /// See [`Flag::delimiter`]: a bound counts values, and only this says how many values a
615 /// word carries.
616 pub delimiter: ::core::option::Option<u8>,
617 /// Whether a negative-number token is accepted as this positional even in
618 /// strict flag mode.
619 pub allow_negative_numbers: bool,
620 /// A token that ends this variadic positional without becoming a value.
621 pub value_terminator: ::core::option::Option<&'a [u8]>,
622 /// This argument's relationship to the `--` separator.
623 pub double_dash: DoubleDash,
624 /// Unused by binding, kept so a table entry can carry its own name for
625 /// diagnostics.
626 pub name: &'a str,
627}
628
629impl Arg<'_> {
630 /// A single-value argument, for use with struct update syntax.
631 pub const REQUIRED: Arg<'static> = Arg {
632 key: 0,
633 sigil: ::core::option::Option::None,
634 required: true,
635 var: false,
636 var_max: ::core::option::Option::None,
637 delimiter: ::core::option::Option::None,
638 allow_negative_numbers: false,
639 value_terminator: ::core::option::Option::None,
640 double_dash: DoubleDash::Optional,
641 name: "",
642 };
643
644 /// A variadic argument, for use with struct update syntax.
645 pub const VAR: Arg<'static> = Arg {
646 var: true,
647 ..Arg::REQUIRED
648 };
649}
650
651/// What to do with a flag-like token that names no flag in scope.
652///
653/// The default is [`UnknownFlags::Value`]: the token carries on to the positional
654/// arguments, because a spec is often parsing a command line whose flags belong to
655/// something else — a wrapped tool, a task script. A CLI that owns all of its
656/// flags declares [`UnknownFlags::Error`] and gets typo detection instead.
657///
658/// Stored per command and already resolved: inheritance is a question for whoever
659/// builds the tables, and answering it at compile time keeps it out of the parse.
660#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
661pub enum UnknownFlags {
662 /// Offer the token to the positionals. If none can take it, it is an
663 /// unexpected argument.
664 #[default]
665 Value,
666 /// Reject the token.
667 Error,
668}
669
670/// How an argument relates to the `--` separator.
671#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
672pub enum DoubleDash {
673 /// Values may appear on either side of a `--`.
674 #[default]
675 Optional,
676 /// Values are accepted only after a `--`.
677 Required,
678 /// A `--` is kept as a value rather than consumed as a separator.
679 Preserve,
680 /// Once the argument takes a value, behave as if a `--` had been given, so
681 /// the rest of the command line is values. A wrapper can then forward flags
682 /// without its caller typing the separator.
683 Automatic,
684}
685
686/// Something the parser bound.
687#[derive(Debug, Clone, Copy, PartialEq, Eq)]
688pub enum Event<'t, 'a, 'v> {
689 /// A subcommand was selected; parsing continues inside it.
690 Command(&'t Command<'t>),
691 /// A flag was given. `value` is `Some` for a flag that takes one, and
692 /// `negated` is true when the flag was set through its `negate` form.
693 Flag {
694 flag: &'t Flag<'t>,
695 value: Option<&'v [u8]>,
696 negated: bool,
697 },
698 /// A word was bound to a positional argument. A variadic argument produces
699 /// one event per value.
700 Arg {
701 arg: &'t Arg<'t>,
702 value: &'v [u8],
703 /// Whether this value should be split by the argument's declared delimiter.
704 delimit: bool,
705 },
706 /// Ended one clause instance and began the next.
707 ClauseSeparator { clause: Clause<'t> },
708 /// An unmatched word was forwarded as an external command: the name, then
709 /// every remaining token, including flags.
710 External { values: &'a [&'v OsStr] },
711}
712
713/// A binding failure.
714///
715/// Carries the offending token so a caller can render a good message, but no
716/// message of its own: rendering belongs to a cold path, and building a string
717/// here would allocate on the way to reporting that nothing was allocated.
718///
719/// `non_exhaustive`, because an error enum grows: a caller matching on it needs a
720/// fallback arm so that recognizing a new failure is never a breaking change.
721// No `Copy`: one variant owns its message. `Clone` stays, and the enum is still 40 bytes
722// because that variant is boxed, so nothing on the hot path grew.
723#[derive(Debug, Clone, PartialEq, Eq)]
724#[non_exhaustive]
725pub enum Error<'t, 'v> {
726 /// A flag-like token matched no flag in scope. `token` is the whole token as
727 /// typed, so a bundle containing an unrecognized letter reports `-fz` rather
728 /// than the letter alone — which is also the unit in which it is rejected.
729 UnknownFlag { token: &'v [u8] },
730 /// A flag that needs a value did not get one, either because the command
731 /// line ended or because the next token was flag-like.
732 MissingFlagValue { flag: &'t Flag<'t> },
733 /// A word arrived with no argument left to hold it.
734 UnexpectedArg { token: &'v [u8] },
735 /// A word was offered to a `double_dash = "required"` argument before any
736 /// `--` had been seen.
737 ArgRequiresDoubleDash { arg: &'t Arg<'t> },
738 /// A subcommand was selected after this command had already bound an argument.
739 SubcommandConflict { subcommand: &'t Command<'t> },
740 /// The command tree is deeper than [`MAX_DEPTH`].
741 TooDeep,
742
743 // The rest are raised *after* the parse, by whoever owns the target type: they
744 // need to know a value's declared type, which the parser deliberately does not.
745 // They share this enum so that a caller has one error to handle rather than two.
746 /// Something the command requires was never given.
747 MissingRequired {
748 /// The flag or argument's name, as the spec calls it.
749 name: &'t str,
750 },
751 /// A flag that is not repeatable was given more than once.
752 DuplicateFlag {
753 /// The flag's name, as the spec calls it.
754 name: &'t str,
755 },
756 /// A value was given that is not among the declared choices.
757 ///
758 /// Carries the choices rather than the offending value: rendering the value means
759 /// owning it, and an error that allocates on a path this crate promises not to
760 /// allocate on would be a poor trade for a better message. Diagnostics are a
761 /// separate layer.
762 InvalidChoice {
763 name: &'t str,
764 choices: &'t [&'t str],
765 },
766 /// Fewer values than `var_min`.
767 VarTooFew {
768 name: &'t str,
769 min: usize,
770 got: usize,
771 },
772 /// More values than `var_max`.
773 VarTooMany {
774 name: &'t str,
775 max: usize,
776 got: usize,
777 },
778 /// Two flags declared to conflict were both given.
779 ///
780 /// Carries both names because either one alone reads as a puzzle: which flag is
781 /// unwelcome depends entirely on what else is on the command line.
782 ConflictingFlags {
783 /// The flag whose declaration names the conflict.
784 name: &'t str,
785 /// The flag it cannot be given with, as the declaration spells it.
786 other: &'t str,
787 },
788 /// A value was given that the field's type could not be built from.
789 ///
790 /// Boxed, and the only error here that owns anything. Everything else borrows the
791 /// tables or argv, which is what keeps a *successful* parse allocation-free — and the
792 /// box keeps `Error` the size it was, so the `Result` this rides in on the hot path
793 /// does not grow. A value that will not convert has already failed, and a message
794 /// worth reading is worth one allocation.
795 InvalidValue(::std::boxed::Box<InvalidValue<'t>>),
796 /// A required group had none of its members given.
797 ///
798 /// Carries the members as members rather than as a rendered sentence: the caller
799 /// decides how to say it, and a completion asking what would satisfy this needs the
800 /// list rather than the prose.
801 MissingGroup {
802 /// The group's declared name, which appears in the message so a command with
803 /// several groups does not report the same sentence twice.
804 group: &'t str,
805 /// The flags that would satisfy it, as the declaration spells them.
806 members: &'t [&'t str],
807 },
808 /// A subcommand was required, and none was given.
809 MissingSubcommand,
810 /// `--help` or `-h` was given, and `cmd` is what it was asked about.
811 ///
812 /// Not a failure, and returned as one anyway: a parse that stops to print help has not
813 /// produced a value, and every caller already handles the "no value" shape. clap does the
814 /// same thing for the same reason.
815 ///
816 /// `long` distinguishes the two: `-h` prints the short form and `--help` the long one, as
817 /// clap has them. The caller renders — this crate does not print, because a library that
818 /// writes to stdout on its own is one an adopter cannot embed.
819 Help { cmd: &'t Command<'t>, long: bool },
820 /// `arg_required_else_help` found no command-line arguments for `cmd`.
821 ///
822 /// Unlike an explicit help request, this is a usage failure: clap prints the short help to
823 /// stderr and exits with status 2. Keeping the shape separate lets embedders preserve that
824 /// terminal contract without guessing why [`Error::Help`] was returned.
825 MissingArgsHelp { cmd: &'t Command<'t> },
826 /// Recursive long help was requested for `cmd` and every visible descendant.
827 HelpAll { cmd: &'t Command<'t> },
828 /// `--version` or `-V` was asked for. Not a failure either — the caller prints and leaves.
829 ///
830 /// The version string lives in the spec rather than the parse tables. `long` lets the
831 /// caller choose `long_version` for `--version` while `-V` retains the concise value.
832 Version { long: bool },
833}
834
835/// The high half of every key one declaration's items get.
836///
837/// A derive cannot see other expansions, so it cannot hand out keys from a shared
838/// counter: it hashes the declaration it was given instead. It cannot see a module path
839/// either, which is why the module is mixed in *here* — `module_path!()` is available to
840/// the generated code as a compile-time string, so two byte-identical declarations in
841/// different modules end up with different keys rather than colliding.
842///
843/// `declaration` is a hash the derive computed over the item's own tokens.
844pub const fn key_base(module: &str, declaration: u32) -> u64 {
845 // FNV-1a, continuing from the declaration's hash rather than starting over, so both
846 // halves contribute. Spelled out rather than taken from a `Hasher`, which is not
847 // guaranteed to be stable between compilations — and these are baked into a binary.
848 let mut hash: u32 = declaration;
849 let bytes = module.as_bytes();
850 let mut i = 0;
851 while i < bytes.len() {
852 hash ^= bytes[i] as u32;
853 hash = hash.wrapping_mul(0x0100_0193);
854 i += 1;
855 }
856 (hash as u64) << 32
857}
858
859/// Why a value would not convert into the type its field holds.
860///
861/// Separate from [`Error`] so that the enum stays small: this is reached through a `Box`.
862#[derive(Debug, Clone, PartialEq, Eq)]
863pub struct InvalidValue<'t> {
864 /// The flag or argument's name, as the spec calls it.
865 pub name: &'t str,
866 /// The text that would not convert.
867 pub value: ::std::string::String,
868 /// What the type's own conversion complained about.
869 pub reason: ::std::string::String,
870}
871
872/// A command-wide validation or finalization failure.
873///
874/// Return this from a `#[usage(validate_with = ...)]` hook or from the
875/// `TryFrom` implementation named by `#[usage(try_into = ...)]`. The derive
876/// turns it into the same [`Error::InvalidValue`] diagnostic used by field
877/// conversion, so callers keep one parse error type.
878#[derive(Debug, Clone, PartialEq, Eq)]
879pub struct ValidationError {
880 name: &'static str,
881 value: String,
882 reason: String,
883}
884
885impl ValidationError {
886 /// Start an error attributed to a flag, positional, or command name.
887 pub fn field(name: &'static str) -> Self {
888 Self {
889 name,
890 value: String::new(),
891 reason: String::new(),
892 }
893 }
894
895 /// Record the value that failed the command-wide invariant.
896 pub fn value(mut self, value: impl Into<String>) -> Self {
897 self.value = value.into();
898 self
899 }
900
901 /// Explain the invariant that the value did not satisfy.
902 pub fn reason(mut self, reason: impl Into<String>) -> Self {
903 self.reason = reason.into();
904 self
905 }
906
907 /// Convert this application-level failure into the parser's diagnostic.
908 pub fn into_parse_error<'v>(self) -> Error<'static, 'v> {
909 Error::InvalidValue(Box::new(InvalidValue {
910 name: self.name,
911 value: self.value,
912 reason: self.reason,
913 }))
914 }
915}
916
917/// Interpret a value as UTF-8.
918///
919/// The parser hands back bytes borrowed from `argv`; this is the conversion most
920/// callers want, and the point at which a non-UTF-8 command line is rejected —
921/// but only for the values actually inspected.
922pub fn as_str(value: &[u8]) -> Result<&str, std::str::Utf8Error> {
923 std::str::from_utf8(value)
924}
925
926/// How many entries a group of tables holds in total.
927///
928/// The length for [`concat_flags`] and [`concat_args`], which need it as a const generic — so
929/// it has to be computable separately from the concatenation itself.
930///
931/// ```
932/// use usage_argv::{table_len, Flag};
933///
934/// static A: Flag = Flag { name: "a", ..Flag::BOOL };
935/// static B: Flag = Flag { name: "b", ..Flag::BOOL };
936/// const GROUPS: &[&[&Flag]] = &[&[&A], &[], &[&B]];
937/// const N: usize = table_len(GROUPS);
938/// assert_eq!(N, 2);
939/// ```
940pub const fn table_len<T>(groups: &[&[T]]) -> usize {
941 let mut total = 0;
942 let mut i = 0;
943 while i < groups.len() {
944 total += groups[i].len();
945 i += 1;
946 }
947 total
948}
949
950/// Join groups of flag tables into one, at compile time.
951///
952/// This is how `#[usage(flatten)]` stays free. A flattened struct's flags have to appear in
953/// the parent's own table, and the parent's macro expansion cannot see them — it has only a
954/// type. But it can name that type's [`CommandArgs::COMMAND`](crate::spec::CommandArgs::COMMAND),
955/// and a `const fn` can read through it, so the two lists become one `static` array before the
956/// program runs. The parser then walks a single flat slice, exactly as it does for a command
957/// that declared everything itself: flatten costs nothing at run time.
958///
959/// Groups are laid out in the order given, which is what lets a flattened group sit *between*
960/// two of the parent's own declarations — necessary for positional arguments, where order is
961/// the meaning.
962///
963/// `N` must be [`table_len`] of the same groups. It cannot be inferred, and a wrong one fails
964/// to compile rather than leaving the difference filled with padding.
965///
966/// ```
967/// use usage_argv::{concat_flags, table_len, Flag};
968///
969/// static FORCE: Flag = Flag { name: "force", longs: &["force"], ..Flag::BOOL };
970/// static QUIET: Flag = Flag { name: "quiet", longs: &["quiet"], ..Flag::BOOL };
971/// static SHARED: &[&Flag] = &[&QUIET];
972///
973/// const GROUPS: &[&[&Flag]] = &[&[&FORCE], SHARED];
974/// static FLAGS: [&Flag; table_len(GROUPS)] = concat_flags(GROUPS);
975///
976/// assert_eq!(FLAGS.iter().map(|f| f.name).collect::<Vec<_>>(), ["force", "quiet"]);
977/// ```
978pub const fn concat_flags<const N: usize>(
979 groups: &[&[&'static Flag<'static>]],
980) -> [&'static Flag<'static>; N] {
981 // Every slot is written below, but an array has to start somewhere and `MaybeUninit`
982 // would mean `unsafe`. A `Flag` nobody can reach is cheaper than that.
983 static PLACEHOLDER: Flag<'static> = Flag::BOOL;
984 let mut out = [&PLACEHOLDER; N];
985 let mut at = 0;
986 let mut g = 0;
987 while g < groups.len() {
988 let group = groups[g];
989 let mut i = 0;
990 while i < group.len() {
991 out[at] = group[i];
992 at += 1;
993 i += 1;
994 }
995 g += 1;
996 }
997 assert!(
998 at == N,
999 "`N` must be `table_len` of the same groups, or the table would keep a placeholder \
1000 that answers to nothing"
1001 );
1002 out
1003}
1004
1005/// Join groups of argument tables into one, at compile time.
1006///
1007/// The positional counterpart of [`concat_flags`] — see there for why this exists. Order
1008/// matters more here: an argument's position *is* its identity, so a flattened group has to
1009/// land exactly where the field was written.
1010///
1011/// Two functions rather than one generic: each needs a value to fill an array with before
1012/// overwriting it, and there is no way to ask a type parameter for one in a `const fn`.
1013pub const fn concat_args<const N: usize>(
1014 groups: &[&[&'static Arg<'static>]],
1015) -> [&'static Arg<'static>; N] {
1016 static PLACEHOLDER: Arg<'static> = Arg::REQUIRED;
1017 let mut out = [&PLACEHOLDER; N];
1018 let mut at = 0;
1019 let mut g = 0;
1020 while g < groups.len() {
1021 let group = groups[g];
1022 let mut i = 0;
1023 while i < group.len() {
1024 out[at] = group[i];
1025 at += 1;
1026 i += 1;
1027 }
1028 g += 1;
1029 }
1030 assert!(
1031 at == N,
1032 "`N` must be `table_len` of the same groups, or the table would keep a placeholder \
1033 that answers to nothing"
1034 );
1035 out
1036}
1037
1038/// The key `--help` answers to, and the one `-h` does.
1039///
1040/// Reserved rather than generated: a derive builds keys from a hash of the type they came from
1041/// in the high half and an index in the low half, so the top of the range belongs to nobody.
1042/// Generated code compares against these to tell a help request from a flag of its own.
1043pub const HELP_LONG_KEY: u64 = u64::MAX;
1044/// See [`HELP_LONG_KEY`].
1045pub const HELP_SHORT_KEY: u64 = u64::MAX - 1;
1046
1047/// `--help`, which every command answers to.
1048///
1049/// In the parse table and *not* in the metadata, which is the whole trick: the parser has to
1050/// recognise the flag, and help output must not list it — a spec does not declare `--help`, so
1051/// showing one would make the rendered page disagree with the spec it came from.
1052pub static HELP_LONG: Flag<'static> = Flag {
1053 key: HELP_LONG_KEY,
1054 name: "help",
1055 longs: &["help"],
1056 action: ArgAction::HelpLong,
1057 ..Flag::BOOL
1058};
1059
1060/// See [`HELP_LONG_KEY`].
1061pub const VERSION_LONG_KEY: u64 = u64::MAX - 2;
1062/// See [`HELP_LONG_KEY`].
1063pub const VERSION_SHORT_KEY: u64 = u64::MAX - 3;
1064
1065/// `--version`, where the CLI declared one.
1066///
1067/// In the parse table and not in the metadata, exactly as `--help` is: a spec does not declare
1068/// `--version`, so listing one would make the rendered page disagree with the spec it came from.
1069pub static VERSION_LONG: Flag<'static> = Flag {
1070 key: VERSION_LONG_KEY,
1071 name: "version",
1072 longs: &["version"],
1073 action: ArgAction::Version,
1074 ..Flag::BOOL
1075};
1076
1077/// `-V`, which clap also supplies.
1078pub static VERSION_SHORT: Flag<'static> = Flag {
1079 key: VERSION_SHORT_KEY,
1080 name: "version",
1081 shorts: b"V",
1082 action: ArgAction::Version,
1083 ..Flag::BOOL
1084};
1085
1086/// `-h`, which prints the shorter form.
1087pub static HELP_SHORT: Flag<'static> = Flag {
1088 key: HELP_SHORT_KEY,
1089 name: "help",
1090 shorts: b"h",
1091 action: ArgAction::HelpShort,
1092 ..Flag::BOOL
1093};
1094
1095/// A named subcommand of a given command, by name or alias.
1096///
1097/// Free rather than a method because `help` resolves a path *without* descending: the words
1098/// after it are a question about a command rather than a walk into one.
1099///
1100/// Names across every subcommand before any alias, the precedence the grammar states — and
1101/// the reason this is the only implementation of it on argv's side. `ex run` and `ex help run`
1102/// selecting different commands would be exactly the divergence this rule was written to end.
1103pub(crate) fn find_named<'t>(cmd: &'t Command<'t>, name: &[u8]) -> Option<&'t Command<'t>> {
1104 let subcommands = || cmd.subcommands.iter().copied();
1105 subcommands()
1106 .find(|c| c.name.as_bytes() == name)
1107 .or_else(|| subcommands().find(|c| c.aliases.iter().any(|a| a.as_bytes() == name)))
1108}
1109
1110/// What a caller should print for a parse failure, and what to exit with.
1111///
1112/// The one entry point a generated `parse()` reaches for, and the reason it exists here rather
1113/// than in the derive: whether the good rendering is available is a *feature of this crate* in
1114/// the adopter's dependency graph, and a `#[cfg]` written into generated code is evaluated in
1115/// the adopter's crate, where the feature is not theirs to see. That is how a metadata field
1116/// once got silently dropped; the answer is that the cfg lives beside the thing it gates.
1117///
1118/// With `diagnostics` on, this is the clap-shaped message. Without it, the error's `Debug`
1119/// form — which is still better than nothing and is what a parser-only build asked for.
1120///
1121/// [`Error::Help`] and [`Error::Version`] are not failures and must be handled before this.
1122#[cfg(feature = "diagnostics")]
1123pub fn render_failure(spec: &spec::Spec<'_>, argv: &[&OsStr], error: &Error<'_, '_>) -> String {
1124 diagnostic::render(spec, argv, error, diagnostic::Style::auto())
1125}
1126
1127/// A parse failure, never coloured.
1128///
1129/// [`render_failure`] asks the environment whether to colour, which is right for a process and
1130/// wrong for anything that keeps the string: a test that asserts on a message, or a snapshot of
1131/// one, would pass or fail by whether stderr happened to be a terminal. The renderer is the
1132/// same; only the answer to that question is fixed.
1133#[cfg(feature = "diagnostics")]
1134pub fn render_failure_plain(
1135 spec: &spec::Spec<'_>,
1136 argv: &[&OsStr],
1137 error: &Error<'_, '_>,
1138) -> String {
1139 diagnostic::render(spec, argv, error, diagnostic::Style::PLAIN)
1140}
1141
1142/// Render a failure through a spec-declared executable view.
1143///
1144/// `argv` is the original full argv, including the view executable as argv0.
1145#[cfg(feature = "diagnostics")]
1146pub fn render_failure_view<'a>(
1147 spec: &'a spec::Spec<'a>,
1148 argv: &[&OsStr],
1149 error: &Error<'_, '_>,
1150 view: &'a spec::ViewMeta<'a>,
1151) -> String {
1152 diagnostic::render_view(spec, argv, error, diagnostic::Style::auto(), view)
1153}
1154
1155/// What a caller should print for a parse failure, without the renderer that makes it readable.
1156///
1157/// See the other half. A caller that wants the clap-shaped message turns on `diagnostics`;
1158/// this is what a parser-only build asked for, and it still says which error it was.
1159#[cfg(all(feature = "spec", not(feature = "diagnostics")))]
1160pub fn render_failure(spec: &spec::Spec<'_>, argv: &[&OsStr], error: &Error<'_, '_>) -> String {
1161 let _ = (spec, argv);
1162 ::std::format!("error: {error:?}\n")
1163}
1164
1165/// A parse failure without the renderer, which is plain either way.
1166#[cfg(all(feature = "spec", not(feature = "diagnostics")))]
1167pub fn render_failure_plain(
1168 spec: &spec::Spec<'_>,
1169 argv: &[&OsStr],
1170 error: &Error<'_, '_>,
1171) -> String {
1172 render_failure(spec, argv, error)
1173}
1174
1175/// Render a failure through a declared view without the optional diagnostics renderer.
1176#[cfg(all(feature = "spec", not(feature = "diagnostics")))]
1177pub fn render_failure_view(
1178 spec: &spec::Spec<'_>,
1179 argv: &[&OsStr],
1180 error: &Error<'_, '_>,
1181 view: &spec::ViewMeta<'_>,
1182) -> String {
1183 let _ = (spec, argv, view);
1184 ::std::format!("error: {error:?}\n")
1185}
1186
1187/// Answer an error from a generated `parse()` the way a command-line program does, and leave.
1188///
1189/// A help request goes to stdout and exits 0; a request for help because nothing was given goes
1190/// to stderr and exits 2; any other failure is rendered to stderr and exits 2, which is clap's
1191/// status, so a script that checks for it keeps working. [`Error::Version`] is the caller's to
1192/// answer before this, since only the generated code knows the version it prints.
1193///
1194/// Here rather than expanded into every CLI: the body is the same for all of them, and only the
1195/// spec — which carries any computed identity the generated code has already evaluated — and
1196/// the tables differ. `all_argv` includes argv0, which a view's route needs; `argv` does not.
1197#[cfg(feature = "spec")]
1198#[doc(hidden)]
1199pub fn __usage_exit_on_error(
1200 error: Error<'_, '_>,
1201 spec: &spec::Spec<'_>,
1202 root: &Command<'_>,
1203 all_argv: &[&OsStr],
1204 argv: &[&OsStr],
1205 view: Option<&spec::ViewMeta<'_>>,
1206) -> ! {
1207 let (want, to_stderr, cmd) = match error {
1208 Error::Help { cmd, long } => {
1209 let want = if long {
1210 help::Page::Long
1211 } else {
1212 help::Page::Short
1213 };
1214 (want, false, cmd)
1215 }
1216 Error::MissingArgsHelp { cmd } => (help::Page::Short, true, cmd),
1217 Error::HelpAll { cmd } => (help::Page::All, false, cmd),
1218 error => {
1219 let failure = match view {
1220 Some(view) => render_failure_view(spec, all_argv, &error, view),
1221 None => render_failure(spec, argv, &error),
1222 };
1223 __usage_eprint(format_args!("{failure}"));
1224 __usage_process_exit(2);
1225 }
1226 };
1227 let style = if to_stderr {
1228 help::Style::auto_stderr()
1229 } else {
1230 help::Style::auto()
1231 };
1232 let page = match view {
1233 Some(view) => help::page_view(spec, root, all_argv, cmd, view, want, style),
1234 None => help::page(spec, root, argv, cmd, want, style),
1235 };
1236 // `None` is only reachable if the command came from another CLI's tables.
1237 if let Some(page) = page {
1238 if to_stderr {
1239 __usage_eprint(format_args!("{page}"));
1240 } else {
1241 __usage_print(format_args!("{page}"));
1242 }
1243 }
1244 __usage_process_exit(if to_stderr { 2 } else { 0 })
1245}
1246
1247/// Print `bin version` for a generated `parse()`, and leave.
1248///
1249/// The half of [`__usage_exit_on_error`] the generated code answers itself, because the
1250/// version is a declaration of that CLI; the printing and the exit are the same everywhere.
1251#[cfg(feature = "spec")]
1252#[doc(hidden)]
1253pub fn __usage_exit_version(
1254 bin: &str,
1255 view: Option<&spec::ViewMeta<'_>>,
1256 version: &dyn std::fmt::Display,
1257) -> ! {
1258 let bin = view.map_or(bin, |view| view.bin);
1259 __usage_print(format_args!("{bin} {version}\n"));
1260 __usage_process_exit(0)
1261}
1262
1263/// What a caller should print for the deprecations a command line used.
1264///
1265/// The same arrangement as [`render_failure`], and for the same reason: whether the coloured
1266/// rendering is available is a feature of *this* crate in the adopter's dependency graph, so the
1267/// `#[cfg]` lives beside the thing it gates rather than in generated code.
1268///
1269/// Warnings are not failures. A caller prints these to stderr and carries on.
1270#[cfg(feature = "diagnostics")]
1271pub fn render_warnings(warnings: &[warn::Warning<'_>]) -> String {
1272 diagnostic::render_warnings(warnings, diagnostic::Style::auto())
1273}
1274
1275/// The same wording without the renderer that colours it. See the other half.
1276#[cfg(all(feature = "spec", not(feature = "diagnostics")))]
1277pub fn render_warnings(warnings: &[warn::Warning<'_>]) -> String {
1278 warn::render_warnings(warnings)
1279}
1280
1281/// The word a tool sends to ask a binary for its own spec.
1282///
1283/// Not a flag and not a command: a spec request is not something this CLI *does*, so it is
1284/// answered before the parse and stays out of the tables — the same reason
1285/// `__complete_word__` is a word rather than a subcommand. It also keeps the endpoint from
1286/// perturbing the document it prints, which a declared flag would not.
1287pub const SPEC_REQUEST: &str = "__usage_spec__";
1288
1289/// Whether this argv asks for the spec rather than for the CLI to run.
1290///
1291/// Only the first word counts: `mycli build __usage_spec__` passes the word through as an
1292/// ordinary value, because a request is the whole invocation or it is nothing.
1293///
1294/// A root that declares a command of that name keeps it, which is the precedence the `help`
1295/// subcommand already has. The check is here rather than in the derive because a `Cli` derive
1296/// expands one struct and cannot see the variant names of a separate `Subcommands` enum — the
1297/// static tables can.
1298pub fn is_spec_request(root: &Command<'_>, argv: &[&OsStr]) -> bool {
1299 let [first, ..] = argv else { return false };
1300 first.as_encoded_bytes() == SPEC_REQUEST.as_bytes()
1301 && find_named(root, SPEC_REQUEST.as_bytes()).is_none()
1302}
1303
1304/// Whether a flag is one of the two the parser supplies rather than the CLI declaring it.
1305pub fn is_help_flag(flag: &Flag<'_>) -> bool {
1306 matches!(
1307 flag.action,
1308 ArgAction::Help | ArgAction::HelpShort | ArgAction::HelpLong | ArgAction::HelpAll
1309 )
1310}
1311
1312/// Whether a flag is one of the two the parser supplies for `--version`.
1313pub fn is_version_flag(flag: &Flag<'_>) -> bool {
1314 flag.action == ArgAction::Version
1315}
1316
1317/// Whether one exact root argument selects a declared or synthesized version action.
1318///
1319/// Declared flags are checked first because they shadow the built-in `--version` and `-V`
1320/// spellings. Executable views use this before projection so custom version spellings keep
1321/// reporting the package that owns the view.
1322pub fn is_version_arg(cmd: &Command<'_>, word: &OsStr) -> bool {
1323 let token = word.as_encoded_bytes();
1324 if let Some(long) = token.strip_prefix(b"--") {
1325 if let Some(flag) = cmd.flags.iter().find(|flag| {
1326 flag.longs
1327 .iter()
1328 .any(|spelling| spelling.as_bytes() == long)
1329 }) {
1330 return is_version_flag(flag);
1331 }
1332 if cmd.flags.iter().any(|flag| {
1333 flag.negate
1334 .is_some_and(|spelling| spelling.as_bytes() == long)
1335 }) {
1336 return false;
1337 }
1338 return long == b"version" && cmd.version && !cmd.disable_version_flag;
1339 }
1340 if let [b'-', short] = token {
1341 if let Some(flag) = cmd.flags.iter().find(|flag| flag.shorts.contains(short)) {
1342 return is_version_flag(flag);
1343 }
1344 return *short == b'V' && cmd.version && !cmd.disable_version_flag;
1345 }
1346 false
1347}
1348
1349/// Resolve a subcommand by name or alias, at compile time.
1350///
1351/// For [`Command::default_subcommand`], which names a command that a derive cannot see: the
1352/// variants of a subcommand enum are a different macro expansion, so the name is all the
1353/// parent has. Searching the list in a `const fn` closes that gap — the answer is the same
1354/// `&'static` the table already holds, found before the program runs.
1355///
1356/// A name no subcommand answers to is a **compile error**, since this panics during const
1357/// evaluation. That is the whole point of doing it here rather than at startup.
1358///
1359/// ```
1360/// use usage_argv::{find_subcommand, Command};
1361///
1362/// static RUN: Command = Command { name: "run", ..Command::EMPTY };
1363/// static SUBS: &[&Command] = &[&RUN];
1364/// static ROOT: Command = Command {
1365/// name: "ex",
1366/// subcommands: SUBS,
1367/// default_subcommand: Some(find_subcommand(SUBS, "run")),
1368/// ..Command::EMPTY
1369/// };
1370/// assert_eq!(ROOT.default_subcommand.unwrap().name, "run");
1371/// ```
1372pub const fn find_subcommand<'a>(
1373 subcommands: &'a [&'a Command<'a>],
1374 name: &str,
1375) -> &'a Command<'a> {
1376 // Names first, then aliases: a command's own name outranks another command's alias, so
1377 // the answer does not depend on the order the table happens to list them in. Checking
1378 // each candidate's name *and* aliases in one pass instead let whichever command came
1379 // first win, and usage-lib resolved the same spec to the last one.
1380 let mut i = 0;
1381 while i < subcommands.len() {
1382 if str_eq(subcommands[i].name, name) {
1383 return subcommands[i];
1384 }
1385 i += 1;
1386 }
1387 // Aliases answer too, because usage-lib resolves the name against names, aliases and
1388 // hidden aliases alike — so a spec may point `default_subcommand` at any of them.
1389 let mut i = 0;
1390 while i < subcommands.len() {
1391 let candidate = subcommands[i];
1392 let mut a = 0;
1393 while a < candidate.aliases.len() {
1394 if str_eq(candidate.aliases[a], name) {
1395 return candidate;
1396 }
1397 a += 1;
1398 }
1399 i += 1;
1400 }
1401 panic!("`default_subcommand` names a command that this one does not have")
1402}
1403
1404/// Refuse two subcommands that answer to the same name, aliases included.
1405///
1406/// A derive expansion can validate aliases written on one enum, but aliases may also live on
1407/// the independently expanded `Args` structs its variants wrap. This final, joined-table check
1408/// is where both declarations are visible.
1409pub const fn assert_unique_subcommand_names(subcommands: &[&Command<'_>]) {
1410 const fn form<'a>(cmd: &'a Command<'a>, at: usize) -> Option<&'a str> {
1411 if at == 0 {
1412 Some(cmd.name)
1413 } else if at <= cmd.aliases.len() {
1414 Some(cmd.aliases[at - 1])
1415 } else {
1416 None
1417 }
1418 }
1419
1420 let mut command = 0;
1421 while command < subcommands.len() {
1422 let mut at = 0;
1423 while let Some(name) = form(subcommands[command], at) {
1424 let mut other_command = command;
1425 while other_command < subcommands.len() {
1426 let mut other_at = if other_command == command { at + 1 } else { 0 };
1427 while let Some(other) = form(subcommands[other_command], other_at) {
1428 assert!(
1429 !str_eq(name, other),
1430 "two subcommands answer to the same name, counting aliases"
1431 );
1432 other_at += 1;
1433 }
1434 other_command += 1;
1435 }
1436 at += 1;
1437 }
1438 command += 1;
1439 }
1440}
1441
1442/// Assert that flags scoped to a clause do not reuse a command-level spelling.
1443///
1444/// Derive-generated command tables include scoped flags so the parser can find them. Entries
1445/// that share a key are therefore the same declaration and are ignored here; every other pair
1446/// must have disjoint long, short, alias, and negated spellings.
1447pub const fn assert_clause_flag_spellings(command: &Command<'_>) {
1448 let Some(clause) = command.clause else {
1449 return;
1450 };
1451 let mut scoped_at = 0;
1452 while scoped_at < clause.flags.len() {
1453 let scoped = clause.flags[scoped_at];
1454 let mut command_at = 0;
1455 while command_at < command.flags.len() {
1456 let flag = command.flags[command_at];
1457 if flag.key != scoped.key && const_flag_forms_overlap(flag, scoped) {
1458 panic!("a clause flag spelling conflicts with a command-level flag");
1459 }
1460 command_at += 1;
1461 }
1462 scoped_at += 1;
1463 }
1464}
1465
1466const fn const_flag_forms_overlap(a: &Flag<'_>, b: &Flag<'_>) -> bool {
1467 let mut i = 0;
1468 while i < a.longs.len() {
1469 let mut j = 0;
1470 while j < b.longs.len() {
1471 if const_bytes_eq(a.longs[i].as_bytes(), b.longs[j].as_bytes()) {
1472 return true;
1473 }
1474 j += 1;
1475 }
1476 if let Some(negate) = b.negate
1477 && const_bytes_eq(a.longs[i].as_bytes(), negate.as_bytes())
1478 {
1479 return true;
1480 }
1481 i += 1;
1482 }
1483 i = 0;
1484 while i < a.shorts.len() {
1485 let mut j = 0;
1486 while j < b.shorts.len() {
1487 if a.shorts[i] == b.shorts[j] {
1488 return true;
1489 }
1490 j += 1;
1491 }
1492 i += 1;
1493 }
1494 if let Some(negate) = a.negate {
1495 let mut j = 0;
1496 while j < b.longs.len() {
1497 if const_bytes_eq(negate.as_bytes(), b.longs[j].as_bytes()) {
1498 return true;
1499 }
1500 j += 1;
1501 }
1502 if let Some(other) = b.negate
1503 && const_bytes_eq(negate.as_bytes(), other.as_bytes())
1504 {
1505 return true;
1506 }
1507 }
1508 false
1509}
1510
1511const fn const_bytes_eq(a: &[u8], b: &[u8]) -> bool {
1512 if a.len() != b.len() {
1513 return false;
1514 }
1515 let mut i = 0;
1516 while i < a.len() {
1517 if a[i] != b[i] {
1518 return false;
1519 }
1520 i += 1;
1521 }
1522 true
1523}
1524
1525/// `==` on strings, in a `const fn`.
1526const fn str_eq(a: &str, b: &str) -> bool {
1527 let (a, b) = (a.as_bytes(), b.as_bytes());
1528 if a.len() != b.len() {
1529 return false;
1530 }
1531 let mut i = 0;
1532 while i < a.len() {
1533 if a[i] != b[i] {
1534 return false;
1535 }
1536 i += 1;
1537 }
1538 true
1539}
1540
1541/// Rebuild an [`OsString`] from bytes the parser handed back.
1542///
1543/// This is the reverse of [`OsStr::as_encoded_bytes`], and it is how a `PathBuf` field
1544/// receives a filename the operating system accepts but UTF-8 does not — `/tmp/\xff` stays
1545/// `/tmp/\xff` rather than becoming a *different* filename with `U+FFFD` in it.
1546///
1547/// Where the platform cannot hold those bytes, they are handed back in the `Err` — as
1548/// `String::from_utf8` does — so the caller can name the value in its error without this
1549/// having to copy it for a case that is nearly never taken.
1550///
1551/// # Why this is not `unsafe`, and why it is not lossless everywhere
1552///
1553/// On **Unix** an `OsString` is an arbitrary byte sequence, so the conversion is total and
1554/// uses the safe [`OsStringExt::from_vec`]. Every byte survives, which is the case that
1555/// matters: non-UTF-8 filenames are ordinary there.
1556///
1557/// [`OsStringExt::from_vec`]: std::os::unix::ffi::OsStringExt::from_vec
1558///
1559/// On **Windows** the encoding is WTF-8, where not every byte sequence is valid, and the only
1560/// constructor that accepts one is `OsString::from_encoded_bytes_unchecked` — whose
1561/// precondition this function cannot enforce. It takes a `Vec<u8>` from a safe caller, so
1562/// there is no way to know the bytes came from `as_encoded_bytes` rather than from anywhere
1563/// else, and a safe function with a precondition that can be violated is unsound however
1564/// carefully its callers behave today.
1565///
1566/// So on Windows the bytes go through UTF-8, and one that is not valid UTF-8 is refused
1567/// rather than assumed. What that gives up is a Windows argument containing an unpaired
1568/// surrogate, which is reported instead of accepted; what it buys is that this crate needs no
1569/// `unsafe` at all.
1570pub fn os_string_from_bytes(value: Vec<u8>) -> Result<OsString, Vec<u8>> {
1571 #[cfg(unix)]
1572 {
1573 Ok(std::os::unix::ffi::OsStringExt::from_vec(value))
1574 }
1575 #[cfg(not(unix))]
1576 {
1577 match String::from_utf8(value) {
1578 Ok(text) => Ok(OsString::from(text)),
1579 Err(bad) => Err(bad.into_bytes()),
1580 }
1581 }
1582}
1583
1584/// One [`Error::InvalidValue`], built out of line.
1585///
1586/// Cold and never inlined on purpose: this is the failure path of every value
1587/// conversion in every generated `build`, and inlining it there is what made
1588/// those functions large.
1589#[cold]
1590#[inline(never)]
1591pub(crate) fn invalid_value_error<'t, 'v>(
1592 name: &'t str,
1593 value: String,
1594 reason: String,
1595) -> Error<'t, 'v> {
1596 Error::InvalidValue(Box::new(InvalidValue {
1597 name,
1598 value,
1599 reason,
1600 }))
1601}
1602
1603/// One [`Error::InvalidValue`] for a word that was not UTF-8.
1604///
1605/// The error half of what a generated `build` does per text field: the check stays
1606/// inline at the field, and this — the lossy rendering and the allocations — lives
1607/// here once instead of once per field.
1608#[cold]
1609#[inline(never)]
1610pub fn invalid_utf8_value<'t, 'v>(name: &'t str, bad: std::string::FromUtf8Error) -> Error<'t, 'v> {
1611 invalid_value_error(
1612 name,
1613 String::from_utf8_lossy(bad.as_bytes()).into_owned(),
1614 bad.utf8_error().to_string(),
1615 )
1616}
1617
1618/// One [`Error::InvalidValue`] for a value whose type would not build from it.
1619///
1620/// Takes the reason as `&dyn Display` so one copy serves every `FromStr` error type.
1621#[cold]
1622#[inline(never)]
1623pub fn invalid_parsed_value<'t, 'v>(
1624 name: &'t str,
1625 value: String,
1626 reason: &dyn std::fmt::Display,
1627) -> Error<'t, 'v> {
1628 invalid_value_error(name, value, reason.to_string())
1629}
1630
1631/// One [`Error::InvalidValue`] for a word that is not one of a value enum's choices.
1632#[cold]
1633#[inline(never)]
1634pub fn invalid_choice_value<'t, 'v>(name: &'t str, value: String) -> Error<'t, 'v> {
1635 invalid_value_error(name, value, String::from("not one of the declared values"))
1636}
1637
1638/// One [`Error::InvalidValue`] for bytes the platform cannot hold in a path.
1639#[cold]
1640#[inline(never)]
1641pub fn invalid_os_value<'t, 'v>(name: &'t str, bytes: Vec<u8>) -> Error<'t, 'v> {
1642 invalid_value_error(
1643 name,
1644 String::from_utf8_lossy(&bytes).into_owned(),
1645 "this platform cannot hold these bytes in a path".to_string(),
1646 )
1647}
1648
1649/// Convert every repeated value of one text field, reporting `name` for the
1650/// first that is not UTF-8.
1651///
1652/// The shared body of what a generated `build` does per collecting text field:
1653/// one loop in the binary rather than one per field. Converts element by
1654/// element rather than with `collect` so the error can carry the value that
1655/// failed rather than only that one did.
1656///
1657/// The empty case is answered here rather than in the shared loop, and this is
1658/// what keeps sharing the loop free: a command at mise's scale declares dozens
1659/// of collecting fields and a command line names one or two of them, so most of
1660/// these calls have nothing to convert. Testing that at the field costs a branch;
1661/// reaching the loop to learn it costs the call.
1662#[inline]
1663pub fn utf8_values<'t, 'v>(
1664 values: Vec<Vec<u8>>,
1665 name: &'t str,
1666) -> Result<Vec<String>, Error<'t, 'v>> {
1667 if values.is_empty() {
1668 return Ok(Vec::new());
1669 }
1670 utf8_values_given(values, name)
1671}
1672
1673#[inline(never)]
1674fn utf8_values_given<'t, 'v>(
1675 values: Vec<Vec<u8>>,
1676 name: &'t str,
1677) -> Result<Vec<String>, Error<'t, 'v>> {
1678 let mut out = Vec::with_capacity(values.len());
1679 for value in values {
1680 match String::from_utf8(value) {
1681 Ok(text) => out.push(text),
1682 Err(bad) => return Err(invalid_utf8_value(name, bad)),
1683 }
1684 }
1685 Ok(out)
1686}
1687
1688/// Convert every repeated value of one field through
1689/// [`FromStr`](std::str::FromStr), reporting `name` for the first that fails.
1690///
1691/// Monomorphized once per target type rather than expanded once per field, and
1692/// the empty case is answered at the field for the reason [`utf8_values`] gives.
1693#[inline]
1694pub fn parsed_values<'t, 'v, T>(
1695 values: Vec<Vec<u8>>,
1696 name: &'t str,
1697) -> Result<Vec<T>, Error<'t, 'v>>
1698where
1699 T: std::str::FromStr,
1700 T::Err: std::fmt::Display,
1701{
1702 if values.is_empty() {
1703 return Ok(Vec::new());
1704 }
1705 parsed_values_given(values, name)
1706}
1707
1708#[inline(never)]
1709fn parsed_values_given<'t, 'v, T>(
1710 values: Vec<Vec<u8>>,
1711 name: &'t str,
1712) -> Result<Vec<T>, Error<'t, 'v>>
1713where
1714 T: std::str::FromStr,
1715 T::Err: std::fmt::Display,
1716{
1717 let mut out = Vec::with_capacity(values.len());
1718 for value in values {
1719 let text = match String::from_utf8(value) {
1720 Ok(text) => text,
1721 Err(bad) => return Err(invalid_utf8_value(name, bad)),
1722 };
1723 match text.parse() {
1724 Ok(parsed) => out.push(parsed),
1725 Err(reason) => return Err(invalid_parsed_value(name, text, &reason)),
1726 }
1727 }
1728 Ok(out)
1729}
1730
1731/// Convert every repeated value of one path-like field, reporting `name` for
1732/// the first the platform cannot hold.
1733///
1734/// `T` is what the field collects — [`PathBuf`](std::path::PathBuf) or
1735/// [`OsString`] — so one body serves both. The same platform note as
1736/// [`os_string_from_bytes`] applies: lossless on Unix, partial on Windows. The
1737/// empty case is answered at the field for the reason [`utf8_values`] gives.
1738#[inline]
1739pub fn os_values<'t, 'v, T: From<OsString>>(
1740 values: Vec<Vec<u8>>,
1741 name: &'t str,
1742) -> Result<Vec<T>, Error<'t, 'v>> {
1743 if values.is_empty() {
1744 return Ok(Vec::new());
1745 }
1746 os_values_given(values, name)
1747}
1748
1749#[inline(never)]
1750fn os_values_given<'t, 'v, T: From<OsString>>(
1751 values: Vec<Vec<u8>>,
1752 name: &'t str,
1753) -> Result<Vec<T>, Error<'t, 'v>> {
1754 let mut out = Vec::with_capacity(values.len());
1755 for value in values {
1756 match os_string_from_bytes(value) {
1757 Ok(os) => out.push(T::from(os)),
1758 Err(bytes) => return Err(invalid_os_value(name, bytes)),
1759 }
1760 }
1761 Ok(out)
1762}
1763
1764/// What a generated `apply` stores for a flag that takes one value: the word, or nothing
1765/// when the flag arrived without one.
1766///
1767/// Out of line because the copy is an allocation, a `memcpy`, and the drop of whatever the
1768/// field held before — a few dozen instructions that would otherwise be expanded at every
1769/// value-taking field of every command.
1770#[doc(hidden)]
1771#[inline(never)]
1772pub fn bind_text(slot: &mut Vec<u8>, value: Option<&[u8]>) {
1773 *slot = value.map(<[u8]>::to_vec).unwrap_or_default();
1774}
1775
1776/// [`bind_text`] for an `Option<T>` field, which a flag without a value still fills.
1777#[doc(hidden)]
1778#[inline(never)]
1779pub fn bind_some_text(slot: &mut Option<Vec<u8>>, value: Option<&[u8]>) {
1780 *slot = Some(value.map(<[u8]>::to_vec).unwrap_or_default());
1781}
1782
1783/// [`bind_text`] for an `Option<Option<T>>` field, where a flag given without a value is
1784/// the inner `None`.
1785#[doc(hidden)]
1786#[inline(never)]
1787pub fn bind_optional_text(slot: &mut Option<Vec<u8>>, value: Option<&[u8]>) {
1788 *slot = value.map(<[u8]>::to_vec);
1789}
1790
1791/// [`bind_text`] for a collecting field: one more occurrence, appended.
1792#[doc(hidden)]
1793#[inline(never)]
1794pub fn bind_push_text(slot: &mut Vec<Vec<u8>>, value: Option<&[u8]>) {
1795 slot.push(value.map(<[u8]>::to_vec).unwrap_or_default());
1796}
1797
1798/// [`bind_push_text`] for a field with a value delimiter: every part, appended in order.
1799#[doc(hidden)]
1800#[inline(never)]
1801pub fn bind_push_split(slot: &mut Vec<Vec<u8>>, value: Option<&[u8]>, delimiter: u8) {
1802 for part in value.unwrap_or_default().split(|b| *b == delimiter) {
1803 slot.push(part.to_vec());
1804 }
1805}
1806
1807/// A single binding pass over `argv`.
1808///
1809/// [`Command::default_subcommand_flags`] adds a read-only lookahead before binding
1810/// to choose an implicit command boundary.
1811///
1812/// Created with [`Parser::new`] and driven with [`Parser::next_event`].
1813pub struct Parser<'t, 'a, 'v> {
1814 argv: &'a [&'v OsStr],
1815 /// Index of the next token to read.
1816 pos: usize,
1817 /// The command currently in scope.
1818 cmd: &'t Command<'t>,
1819 /// The canonical root, used to hide root globals omitted by an executable view.
1820 #[cfg(feature = "spec")]
1821 root: &'t Command<'t>,
1822 /// The executable projection being parsed, if argv0 selected one.
1823 #[cfg(feature = "spec")]
1824 view: Option<&'t spec::ViewMeta<'t>>,
1825 /// What an unrecognized flag-like token means in the command currently in scope.
1826 ///
1827 /// Carried rather than looked up, because it is inherited: a command that states
1828 /// nothing keeps what the enclosing one said, and walking back up the ancestors on
1829 /// every unrecognized token would pay for the inheritance at the wrong moment.
1830 unknown_flags: UnknownFlags,
1831 /// Effective inherited trailing-delimiter policy.
1832 dont_delimit_trailing_values: bool,
1833 /// The chain above `cmd`, used to find inherited global flags. Fixed size so
1834 /// that nothing is allocated.
1835 ancestors: [Option<&'t Command<'t>>; MAX_DEPTH],
1836 depth: usize,
1837 /// Bytes left in a short-flag bundle, if one is partly read.
1838 bundle: &'v [u8],
1839 /// The whole token the current bundle came from, so an error raised part way
1840 /// through it can still name what the user typed.
1841 bundle_token: &'v [u8],
1842 /// A variadic flag that is still collecting values.
1843 collecting: Option<&'t Flag<'t>>,
1844 /// Where the command in scope began, as an index into `argv`.
1845 cmd_start: usize,
1846 /// Where each ancestor's own words began, in step with `ancestors`.
1847 starts: [usize; MAX_DEPTH],
1848 /// How many values it has taken, so a bound can stop it.
1849 collected: u32,
1850 /// Which of `cmd.args` is next to fill.
1851 arg_pos: usize,
1852 /// How many words the variadic at `arg_pos` has taken, for the same reason.
1853 arg_taken: u32,
1854 /// Whether any word has been bound to a positional of `cmd`. Once one has,
1855 /// no further word can select a subcommand.
1856 arg_filled: bool,
1857 /// Whether this command has bound any flag or positional. Unlike
1858 /// `arg_filled`, flags count because clap's command policy treats both as
1859 /// arguments that exclude a later subcommand.
1860 command_arg_found: bool,
1861 positional_arg_found: bool,
1862 /// Whether flag interpretation has stopped. A `--` does this, and so does an
1863 /// `automatic` argument taking a value.
1864 flags_stopped: bool,
1865 /// Whether a `--` was actually consumed as a separator.
1866 ///
1867 /// Tracked apart from `flags_stopped` because the two can differ: an
1868 /// `automatic` argument stops flag interpretation without any separator being
1869 /// typed, and a `preserve` argument keeps one as a value rather than
1870 /// consuming it. Callers asking this question want to know what the user
1871 /// wrote, not what state the parser reached.
1872 separator_seen: bool,
1873 /// Whether a root clause separator was consumed, so empty-default routing cannot fire.
1874 clause_separator_seen: bool,
1875 /// Whether the default subcommand has already been taken.
1876 ///
1877 /// Once, per parse: a default subcommand that itself declares one would otherwise
1878 /// descend on every word until the tree ran out.
1879 default_taken: bool,
1880 /// First default-only flag, when lookahead found no explicit sibling.
1881 default_flag_at: Option<usize>,
1882 /// Cursor just after the implicit boundary bundle, whose shorts keep parent ownership.
1883 default_bundle_end: usize,
1884 /// Set once a fatal error has been reported, so iteration stops.
1885 done: bool,
1886 /// Whether declared built-in actions stop parsing with their action error.
1887 ///
1888 /// Invocation parsing does; completion walking only needs the grammar position after the
1889 /// flag, and must not execute an action while inspecting a partial command line.
1890 action_errors: bool,
1891 /// The `argv` range the `help` *word* resolved as a command path, if one was typed.
1892 ///
1893 /// Empty for `--help`, which asks about wherever the parse had got to. For the word, the
1894 /// question is about a command deeper than the parse reached, and only this walk knows
1895 /// which tokens named it: a caller re-scanning `argv` would count a flag's detached value
1896 /// that happens to spell a sibling's name. Two indices rather than the commands
1897 /// themselves, so the parser keeps allocating nothing.
1898 help_span: (usize, usize),
1899 /// An implicit clause boundary follows the positional event that completed it.
1900 pending_clause_boundary: ::core::option::Option<&'t Clause<'t>>,
1901}
1902
1903impl<'t: 'v, 'a, 'v> Parser<'t, 'a, 'v> {
1904 /// Begin parsing `argv` against `root`.
1905 ///
1906 /// `argv` excludes the program name.
1907 pub fn new(root: &'t Command<'t>, argv: &'a [&'v OsStr]) -> Self {
1908 Self::with_action_errors(root, argv, true)
1909 }
1910
1911 /// Begin a non-executing parse for completion walking.
1912 #[cfg(feature = "complete")]
1913 pub(crate) fn for_completion(root: &'t Command<'t>, argv: &'a [&'v OsStr]) -> Self {
1914 Self::with_action_errors(root, argv, false)
1915 }
1916
1917 fn with_action_errors(
1918 root: &'t Command<'t>,
1919 argv: &'a [&'v OsStr],
1920 action_errors: bool,
1921 ) -> Self {
1922 let mut parser = Parser {
1923 argv,
1924 pos: 0,
1925 cmd: root,
1926 #[cfg(feature = "spec")]
1927 root,
1928 #[cfg(feature = "spec")]
1929 view: None,
1930 unknown_flags: match root.unknown_flags {
1931 ::core::option::Option::Some(mode) => mode,
1932 // Nothing above the root to inherit from, so the default stands.
1933 ::core::option::Option::None => UnknownFlags::Value,
1934 },
1935 dont_delimit_trailing_values: root.dont_delimit_trailing_values,
1936 ancestors: [None; MAX_DEPTH],
1937 depth: 0,
1938 bundle: &[],
1939 bundle_token: &[],
1940 collecting: None,
1941 cmd_start: 0,
1942 starts: [0; MAX_DEPTH],
1943 collected: 0,
1944 arg_pos: 0,
1945 arg_taken: 0,
1946 arg_filled: false,
1947 command_arg_found: false,
1948 positional_arg_found: false,
1949 flags_stopped: false,
1950 separator_seen: false,
1951 clause_separator_seen: false,
1952 default_taken: false,
1953 default_flag_at: None,
1954 default_bundle_end: 0,
1955 done: false,
1956 action_errors,
1957 help_span: (0, 0),
1958 pending_clause_boundary: ::core::option::Option::None,
1959 };
1960 if root.default_subcommand_flags {
1961 parser.default_flag_at = parser.default_flag_route();
1962 }
1963 parser
1964 }
1965
1966 /// Find the implicit boundary without binding anything. Parent spellings win while
1967 /// scanning the prefix; after the boundary the ordinary child/global scope applies.
1968 /// Unknown flags stop lookahead because their value arity cannot be guessed.
1969 fn default_flag_route(&self) -> Option<usize> {
1970 let default = self.cmd.default_subcommand?;
1971 let mut at = None;
1972 let mut i = 0;
1973 while let Some(token) = self.argv.get(i).map(bytes) {
1974 let scope = if at.is_some() { default } else { self.cmd };
1975 if token == b"--"
1976 || token == b"-"
1977 || scope.clause.is_some_and(|c| c.separator == Some(token))
1978 {
1979 return at;
1980 }
1981 if !is_flag_like(token) {
1982 // A sibling name before any default-only flag stays on the parent.
1983 // After one, later words are the default command's args (`-u pkg`
1984 // when `pkg` is also a command).
1985 if at.is_none()
1986 && (self.find_subcommand(token).is_some()
1987 || (token == b"help" && !self.cmd.disable_help_subcommand))
1988 {
1989 return None;
1990 }
1991 return at;
1992 }
1993 let mut value_flag = None;
1994 let mut attached = None;
1995 if let Some(body) = token.strip_prefix(b"--") {
1996 let end = body.iter().position(|b| *b == b'=').unwrap_or(body.len());
1997 let name = &body[..end];
1998 let parent = self.find_long(name).or_else(|| self.find_negation(name));
1999 if parent.is_none()
2000 && ((name == b"help" && !self.cmd.disable_help_flag)
2001 || (name == b"version"
2002 && self.cmd.version
2003 && !self.cmd.disable_version_flag))
2004 {
2005 i += 1;
2006 continue;
2007 }
2008 let flag = parent.or_else(|| {
2009 default.flags.iter().copied().find(|f| {
2010 f.longs.iter().any(|l| l.as_bytes() == name)
2011 || f.negate.is_some_and(|n| n.as_bytes() == name)
2012 })
2013 });
2014 let Some(flag) = flag else {
2015 return at;
2016 };
2017 if parent.is_none() {
2018 at.get_or_insert(i);
2019 }
2020 if flag.takes_value && flag.negate.is_none_or(|n| n.as_bytes() != name) {
2021 value_flag = Some(flag);
2022 attached = (end < body.len()).then(|| &body[end + 1..]);
2023 }
2024 } else {
2025 for (offset, byte) in token[1..].iter().enumerate() {
2026 let parent = self.find_short(*byte);
2027 let Some(flag) = parent.or_else(|| {
2028 default
2029 .flags
2030 .iter()
2031 .copied()
2032 .find(|f| f.shorts.contains(byte))
2033 }) else {
2034 return at;
2035 };
2036 if parent.is_none() {
2037 at.get_or_insert(i);
2038 }
2039 if flag.takes_value {
2040 value_flag = Some(flag);
2041 let rest = &token[offset + 2..];
2042 attached =
2043 (!rest.is_empty()).then_some(rest.strip_prefix(b"=").unwrap_or(rest));
2044 break;
2045 }
2046 }
2047 }
2048 i += 1;
2049 if let Some(flag) = value_flag {
2050 let scope = if at.is_some() { default } else { self.cmd };
2051 let is_separator =
2052 |next: &[u8]| scope.clause.is_some_and(|c| c.separator == Some(next));
2053 let first = if let Some(value) = attached {
2054 Some(value)
2055 } else if !flag.require_equals {
2056 self.argv
2057 .get(i)
2058 .map(bytes)
2059 .filter(|next| {
2060 !is_separator(next)
2061 && (flag.allow_hyphen_values
2062 || !is_flag_like(next)
2063 || (flag.allow_negative_numbers && is_negative_number(next)))
2064 })
2065 .inspect(|_| i += 1)
2066 } else {
2067 None
2068 };
2069 if first.is_none() && !flag.value_optional && flag.default_missing.is_none() {
2070 return at;
2071 }
2072 if flag.variadic {
2073 let mut count = first.map_or(0, |v| values_in(v, flag.delimiter));
2074 while flag.var_max.is_none_or(|max| count < max) {
2075 let Some(next) = self.argv.get(i).map(bytes) else {
2076 break;
2077 };
2078 if is_separator(next) {
2079 break;
2080 }
2081 if next == b"--" || flag.value_terminator.is_some_and(|end| end == next) {
2082 if next != b"--" {
2083 i += 1;
2084 }
2085 break;
2086 }
2087 if is_flag_like(next)
2088 && !(flag.allow_negative_numbers && is_negative_number(next))
2089 {
2090 break;
2091 }
2092 if self.cmd.subcommand_precedence_over_arg
2093 && self.find_subcommand(next).is_some()
2094 {
2095 break;
2096 }
2097 count += values_in(next, flag.delimiter);
2098 i += 1;
2099 }
2100 }
2101 }
2102 }
2103 at
2104 }
2105
2106 /// Restrict inherited root globals to those carried by an executable view.
2107 #[cfg(feature = "spec")]
2108 pub fn with_view(mut self, view: &'t spec::ViewMeta<'t>) -> Self {
2109 self.view = Some(view);
2110 self
2111 }
2112
2113 /// The command in scope: the root, or the deepest subcommand selected so far.
2114 pub fn command(&self) -> &'t Command<'t> {
2115 self.cmd
2116 }
2117
2118 /// Whether a `--` was consumed as a separator.
2119 ///
2120 /// False when flag interpretation stopped for another reason, such as an
2121 /// `automatic` argument taking a value, and false for a `--` that a
2122 /// `preserve` argument kept as a value.
2123 pub fn double_dash_seen(&self) -> bool {
2124 self.separator_seen
2125 }
2126
2127 /// Every command entered so far, and where each one's own words begin.
2128 ///
2129 /// The ancestors are already kept for flag scoping; this is the same chain with the offsets,
2130 /// which is what lets a completion hand a callback the words of *its* command rather than of
2131 /// the deepest one — a global flag is declared on an ancestor.
2132 pub fn command_path(&self) -> Vec<(&'t Command<'t>, usize)> {
2133 let mut out = Vec::with_capacity(self.depth + 1);
2134 for (i, ancestor) in self.ancestors[..self.depth].iter().enumerate() {
2135 if let Some(cmd) = ancestor {
2136 // An ancestor's own words start where the one before it descended, and the
2137 // root's start at the beginning.
2138 out.push((*cmd, self.starts[i]));
2139 }
2140 }
2141 out.push((self.cmd, self.cmd_start));
2142 out
2143 }
2144
2145 /// The `argv` range the `help` word resolved as a command path.
2146 ///
2147 /// Empty unless the word was typed. Every token in it named a subcommand of the one before
2148 /// it — the parser resolved them itself, so nothing here is a flag or a flag's value.
2149 pub fn help_span(&self) -> (usize, usize) {
2150 self.help_span
2151 }
2152
2153 /// Where the command in scope began: the index in `argv` just after its name, or at the
2154 /// unmatched word routed into a default subcommand.
2155 ///
2156 /// `argv[command_start()..]` is what that command was given, which is what a completion
2157 /// callback needs to be handed its own command's half-parsed struct rather than the root's.
2158 pub fn command_start(&self) -> usize {
2159 self.cmd_start
2160 }
2161
2162 /// Whether flag interpretation has stopped, for any reason.
2163 ///
2164 /// Wider than [`double_dash_seen`](Self::double_dash_seen), and the question completion
2165 /// asks: past a separator *or* past the first value of an `automatic` argument, a
2166 /// dash-prefixed word is a value, so there is no flag there to offer.
2167 pub fn flags_stopped(&self) -> bool {
2168 self.flags_stopped
2169 }
2170
2171 /// A variadic flag that is still claiming words.
2172 ///
2173 /// Asked *between* events, because the answer is gone by the end: the call that finds argv
2174 /// exhausted is the one that clears it. A completion needs it — the next word after
2175 /// `--tools a ⌶` is another tool, not the positional that follows.
2176 pub fn collecting(&self) -> Option<&'t Flag<'t>> {
2177 self.collecting
2178 }
2179
2180 /// The positional the next word would fill, if there is one left.
2181 ///
2182 /// A variadic stays here until it reaches its bound, which is what makes it the answer to
2183 /// "what could go where the cursor is" as many times as it can be filled.
2184 pub fn pending_arg(&self) -> Option<&'t Arg<'t>> {
2185 self.next_arg()
2186 }
2187
2188 /// Flags a word here could name: this command's own, then any ancestor's globals.
2189 ///
2190 /// The same set the parser itself would look in, so what is offered and what is accepted
2191 /// cannot disagree — including the shadowing rule, where a subcommand redeclaring an
2192 /// inherited name hides it.
2193 pub fn flags_in_scope(&self) -> impl Iterator<Item = &'t Flag<'t>> + '_ {
2194 self.in_scope()
2195 }
2196
2197 /// Read the next event.
2198 ///
2199 /// Returns `None` when `argv` is exhausted. An `Err` is terminal: the parse
2200 /// stops there, since continuing past a token that could not be understood
2201 /// would only produce bindings derived from a guess. Events already yielded
2202 /// before an error are therefore not a partial result to be used — a caller
2203 /// that assigned them into fields should discard the whole attempt.
2204 ///
2205 /// One case is stronger than that, because the grammar demands it: a short
2206 /// bundle containing an unrecognized letter yields the error *instead of*, not
2207 /// after, the letters that did match.
2208 #[allow(clippy::should_implement_trait)] // not an Iterator: items borrow from self's tables
2209 pub fn next_event(&mut self) -> Option<Result<Event<'t, 'a, 'v>, Error<'t, 'v>>> {
2210 if self.done {
2211 return None;
2212 }
2213 let event = self.step();
2214 if matches!(event, Some(Ok(Event::Flag { .. } | Event::Arg { .. }))) {
2215 self.command_arg_found = true;
2216 }
2217 if matches!(event, Some(Ok(Event::Arg { .. }))) {
2218 self.positional_arg_found = true;
2219 }
2220 if event.is_none()
2221 && self.depth == 0
2222 && self.cmd.default_subcommand_on_empty
2223 && !self.default_taken
2224 && !self.positional_arg_found
2225 && !self.separator_seen
2226 && !self.clause_separator_seen
2227 && let Some(default) = self.cmd.default_subcommand
2228 {
2229 if self.cmd.args_conflicts_with_subcommands && self.command_arg_found {
2230 self.done = true;
2231 return Some(Err(Error::SubcommandConflict {
2232 subcommand: default,
2233 }));
2234 }
2235 self.default_taken = true;
2236 match self.descend(default) {
2237 Ok(()) => return Some(Ok(Event::Command(default))),
2238 Err(error) => {
2239 self.done = true;
2240 return Some(Err(error));
2241 }
2242 }
2243 }
2244 if let Some(Err(_)) = event {
2245 self.done = true;
2246 }
2247 event
2248 }
2249
2250 /// Count parent flags in the shared boundary token before entering the child.
2251 fn default_bundle_has_parent_flag(&self, default: &Command<'_>) -> bool {
2252 let Some(token) = self.argv.get(self.pos).map(bytes) else {
2253 return false;
2254 };
2255 if !token.starts_with(b"-") || token.starts_with(b"--") {
2256 return false;
2257 }
2258 for byte in &token[1..] {
2259 if self.find_short(*byte).is_some() {
2260 return true;
2261 }
2262 match default.flags.iter().find(|f| f.shorts.contains(byte)) {
2263 Some(flag) if !flag.takes_value => {}
2264 _ => break,
2265 }
2266 }
2267 false
2268 }
2269
2270 fn step(&mut self) -> Option<Result<Event<'t, 'a, 'v>, Error<'t, 'v>>> {
2271 if self.default_flag_at == Some(self.pos) && self.bundle.is_empty() {
2272 self.default_flag_at = None;
2273 if let Some(default) = self.cmd.default_subcommand {
2274 if self.cmd.args_conflicts_with_subcommands
2275 && (self.command_arg_found || self.default_bundle_has_parent_flag(default))
2276 {
2277 return Some(Err(Error::SubcommandConflict {
2278 subcommand: default,
2279 }));
2280 }
2281 if self.argv.get(self.pos).is_some_and(|word| {
2282 let token = bytes(word);
2283 token.starts_with(b"-") && !token.starts_with(b"--")
2284 }) {
2285 self.default_bundle_end = self.pos + 1;
2286 }
2287 self.default_taken = true;
2288 return Some(self.descend(default).map(|()| Event::Command(default)));
2289 }
2290 }
2291 if let Some(clause) = self.pending_clause_boundary.take() {
2292 self.arg_pos = 0;
2293 self.arg_taken = 0;
2294 self.arg_filled = false;
2295 self.collecting = None;
2296 self.flags_stopped = false;
2297 return Some(Ok(Event::ClauseSeparator { clause: *clause }));
2298 }
2299 // A partly-read short bundle takes priority: its remaining bytes are
2300 // still part of the token being processed.
2301 if !self.bundle.is_empty() {
2302 return Some(self.short_flag());
2303 }
2304
2305 if self.cmd.subcommand_precedence_over_arg
2306 && !self.flags_stopped
2307 && let Some(token) = self.argv.get(self.pos).map(bytes)
2308 && let Some(sub) = self.find_subcommand(token)
2309 {
2310 if self.cmd.args_conflicts_with_subcommands && self.command_arg_found {
2311 return Some(Err(Error::SubcommandConflict { subcommand: sub }));
2312 }
2313 self.pos += 1;
2314 return Some(self.descend(sub).map(|()| Event::Command(sub)));
2315 }
2316
2317 // A variadic flag keeps claiming tokens until one of them could be
2318 // something else.
2319 if let Some(flag) = self.collecting {
2320 match self.argv.get(self.pos) {
2321 Some(next)
2322 if flag
2323 .value_terminator
2324 .is_some_and(|terminator| bytes(next) == terminator) =>
2325 {
2326 self.pos += 1;
2327 self.collecting = None;
2328 return self.step();
2329 }
2330 Some(next)
2331 if (!is_flag_like(bytes(next))
2332 || (flag.allow_negative_numbers && is_negative_number(bytes(next))))
2333 && self.clause_separator(bytes(next)).is_none()
2334 && bytes(next) != b"--" =>
2335 {
2336 self.pos += 1;
2337 self.collected += values_in(bytes(next), flag.delimiter);
2338 // Same rule as a positional: a bounded occurrence takes that many and
2339 // leaves the rest to whatever follows.
2340 if flag.var_max.is_some_and(|max| self.collected >= max) {
2341 self.collecting = None;
2342 }
2343 // Stopping is only the same as staying within the bound while one word
2344 // is one value. A delimited word can carry the occurrence past it in a
2345 // single step, and that word cannot be split between two owners, so the
2346 // overshoot is an error rather than a place to stop.
2347 if let Some(max) = flag.var_max.filter(|max| self.collected > *max) {
2348 return Some(Err(Error::VarTooMany {
2349 name: flag.name,
2350 max: max as usize,
2351 got: self.collected as usize,
2352 }));
2353 }
2354 return Some(Ok(Event::Flag {
2355 flag,
2356 value: Some(bytes(next)),
2357 negated: false,
2358 }));
2359 }
2360 // A token that could be something else ends the run — but the *end of argv*
2361 // decides nothing. Clearing there threw away the answer to "would the next
2362 // word be claimed?", which is the question a completion asks and no parse
2363 // ever does: once argv is exhausted there are no more events either way.
2364 Some(_) => self.collecting = None,
2365 None => {}
2366 }
2367 }
2368
2369 let token = bytes(self.argv.get(self.pos)?);
2370 self.pos += 1;
2371
2372 // A clause separator remains syntax after an automatic positional stopped flags.
2373 // Only an explicit `--` protects a literal separator.
2374 if let Some(clause) = self.clause_separator(token) {
2375 self.arg_pos = 0;
2376 self.arg_taken = 0;
2377 self.arg_filled = false;
2378 self.collecting = None;
2379 self.flags_stopped = false;
2380 if self.depth == 0 {
2381 self.clause_separator_seen = true;
2382 }
2383 return Some(Ok(Event::ClauseSeparator { clause: *clause }));
2384 }
2385
2386 // An automatic trailing argument stops flag interpretation without consuming an
2387 // explicit separator. A later `--` must still unlock a required trailing argument
2388 // (clap's `last`), while a separator already consumed makes every later `--` data.
2389 if self.flags_stopped && (token != b"--" || self.separator_seen) {
2390 return Some(self.word(token));
2391 }
2392
2393 if token == b"--" {
2394 // `preserve` wants the separator itself as a value, so ask the
2395 // argument that would receive it before treating it as syntax.
2396 if self
2397 .next_arg()
2398 .is_some_and(|a| a.double_dash == DoubleDash::Preserve)
2399 {
2400 return Some(self.word(token));
2401 }
2402 self.flags_stopped = true;
2403 self.separator_seen = true;
2404 // An explicit separator unlocks any argument that required one, even
2405 // if earlier arguments are still unfilled.
2406 if let Some(idx) = self.current_args()[self.arg_pos..]
2407 .iter()
2408 .position(|a| a.double_dash == DoubleDash::Required)
2409 {
2410 // The count belongs to the argument at `arg_pos`, so jumping past it has
2411 // to leave the count behind: a bounded variadic before the separator would
2412 // otherwise lend its total to the argument after it, which then stops
2413 // early or at once.
2414 self.arg_pos += idx;
2415 self.arg_taken = 0;
2416 }
2417 return self.step();
2418 }
2419
2420 if self.arg_taken > 0
2421 && self.next_arg().is_some_and(|arg| {
2422 arg.value_terminator
2423 .is_some_and(|terminator| token == terminator)
2424 })
2425 {
2426 self.advance_arg();
2427 return self.step();
2428 }
2429
2430 // An exact declared short outranks the numeric shape. This keeps ordinary negative
2431 // numbers available as values while allowing clap-compatible spellings such as fd's
2432 // `-0` / `--print0` switch.
2433 let declared_numeric_short = matches!(token, [b'-', short]
2434 if short.is_ascii_digit() && self.find_short(*short).is_some());
2435
2436 if !declared_numeric_short
2437 && is_negative_number(token)
2438 && self
2439 .next_arg()
2440 .is_some_and(|arg| arg.allow_negative_numbers)
2441 {
2442 return Some(self.word(token));
2443 }
2444
2445 if !declared_numeric_short
2446 && is_negative_number(token)
2447 && self.cmd.external_subcommand
2448 && !self.arg_filled
2449 {
2450 return Some(self.word(token));
2451 }
2452
2453 if is_flag_like(token) {
2454 if token.starts_with(b"--") {
2455 return Some(self.long_flag(token));
2456 }
2457 // Check the whole bundle before emitting anything from it. Events go
2458 // out one at a time, so discovering an unknown letter half way
2459 // through would mean the earlier letters had already been applied —
2460 // and the grammar rejects the entire token, not the tail of it.
2461 match self.check_bundle(token) {
2462 Ok(()) => {}
2463 // Unrecognized, so it is a word unless this command wants it refused.
2464 Err(e) if self.unknown_flags == UnknownFlags::Error => {
2465 return Some(Err(e));
2466 }
2467 Err(_) => return Some(self.word(token)),
2468 }
2469 self.bundle = &token[1..];
2470 self.bundle_token = token;
2471 return Some(self.short_flag());
2472 }
2473
2474 Some(self.word(token))
2475 }
2476
2477 fn long_flag(&mut self, token: &'v [u8]) -> Result<Event<'t, 'a, 'v>, Error<'t, 'v>> {
2478 let body = &token[2..];
2479 let (name, attached) = match body.iter().position(|&b| b == b'=') {
2480 Some(i) => (&body[..i], Some(&body[i + 1..])),
2481 None => (body, None),
2482 };
2483
2484 if let Some(flag) = self.find_long(name) {
2485 let value = if flag.takes_value {
2486 match attached {
2487 Some(v) => Some(v),
2488 None => self.take_detached_value(flag)?,
2489 }
2490 } else if flag.bool_value {
2491 validate_bool_value(flag, attached)?
2492 } else {
2493 None
2494 };
2495 if flag.variadic
2496 && let Some(value) = value
2497 {
2498 self.start_collecting(flag, value)?;
2499 }
2500 if let Some(error) = self.flag_action(flag, true) {
2501 return Err(error);
2502 }
2503 return Ok(Event::Flag {
2504 flag,
2505 value,
2506 negated: false,
2507 });
2508 }
2509
2510 if let Some(flag) = self.find_negation(name) {
2511 return Ok(Event::Flag {
2512 flag,
2513 value: if flag.bool_value {
2514 validate_bool_value(flag, attached)?
2515 } else {
2516 None
2517 },
2518 negated: true,
2519 });
2520 }
2521
2522 // Where the CLI declared a version, `--version` answers with it — asked after the
2523 // command's own flags, so a CLI declaring its own keeps it.
2524 let version_command = self.version_command();
2525 if name == b"version" && version_command.version && !version_command.disable_version_flag {
2526 return Ok(Event::Flag {
2527 flag: &VERSION_LONG,
2528 value: None,
2529 negated: false,
2530 });
2531 }
2532
2533 // Every CLI answers to `--help`, and none of them declares it. Asked *after* the
2534 // command's own flags, so a CLI that declares its own `--help` keeps it.
2535 if name == b"help" && !self.cmd.disable_help_flag {
2536 return Ok(Event::Flag {
2537 flag: &HELP_LONG,
2538 value: None,
2539 negated: false,
2540 });
2541 }
2542
2543 if self.unknown_flags == UnknownFlags::Error {
2544 return Err(Error::UnknownFlag { token });
2545 }
2546 // Not a flag here, so it is a word like any other.
2547 self.word(token)
2548 }
2549
2550 /// Walk a short-flag token without binding anything, to find out whether all
2551 /// of it is recognized.
2552 ///
2553 /// Scanning stops at the first letter whose flag takes a value, because
2554 /// everything after it is that value rather than more letters.
2555 fn check_bundle(&self, token: &'v [u8]) -> Result<(), Error<'t, 'v>> {
2556 let mut rest = &token[1..];
2557 while let Some((&byte, tail)) = rest.split_first() {
2558 match self.find_short(byte) {
2559 None => return Err(Error::UnknownFlag { token }),
2560 Some(flag) if flag.takes_value => return Ok(()),
2561 Some(_) => rest = tail,
2562 }
2563 }
2564 Ok(())
2565 }
2566
2567 fn short_flag(&mut self) -> Result<Event<'t, 'a, 'v>, Error<'t, 'v>> {
2568 let byte = self.bundle[0];
2569 let rest = &self.bundle[1..];
2570
2571 let Some(flag) = self.find_short(byte) else {
2572 // check_bundle already rejected any token containing an unrecognized
2573 // letter, so this is unreachable — but a parser should report rather
2574 // than panic if that ever stops being true.
2575 self.bundle = &[];
2576 return Err(Error::UnknownFlag {
2577 token: self.bundle_token,
2578 });
2579 };
2580
2581 if !flag.takes_value {
2582 self.bundle = rest;
2583 if let Some(error) = self.flag_action(flag, false) {
2584 self.bundle = &[];
2585 return Err(error);
2586 }
2587 return Ok(Event::Flag {
2588 flag,
2589 value: None,
2590 negated: false,
2591 });
2592 }
2593
2594 // A value-taking short ends the token: everything after it is the value,
2595 // less one separating `=`.
2596 self.bundle = &[];
2597 let value = if rest.is_empty() {
2598 self.take_detached_value(flag)?
2599 } else if rest[0] == b'=' {
2600 Some(&rest[1..])
2601 } else {
2602 Some(rest)
2603 };
2604 if flag.variadic
2605 && let Some(value) = value
2606 {
2607 self.start_collecting(flag, value)?;
2608 }
2609 if let Some(error) = self.flag_action(flag, false) {
2610 return Err(error);
2611 }
2612 Ok(Event::Flag {
2613 flag,
2614 value,
2615 negated: false,
2616 })
2617 }
2618
2619 fn flag_action(&self, flag: &'t Flag<'t>, long_spelling: bool) -> Option<Error<'t, 'v>> {
2620 if matches!(
2621 flag.key,
2622 HELP_LONG_KEY | HELP_SHORT_KEY | VERSION_LONG_KEY | VERSION_SHORT_KEY
2623 ) || !self.action_errors
2624 {
2625 return None;
2626 }
2627 match flag.action {
2628 ArgAction::Set => None,
2629 ArgAction::Help => Some(Error::Help {
2630 cmd: self.cmd,
2631 long: long_spelling,
2632 }),
2633 ArgAction::HelpShort => Some(Error::Help {
2634 cmd: self.cmd,
2635 long: false,
2636 }),
2637 ArgAction::HelpLong => Some(Error::Help {
2638 cmd: self.cmd,
2639 long: true,
2640 }),
2641 ArgAction::HelpAll => Some(Error::HelpAll { cmd: self.cmd }),
2642 ArgAction::Version => Some(Error::Version {
2643 long: long_spelling,
2644 }),
2645 }
2646 }
2647
2648 /// Take the following token as a flag's value.
2649 ///
2650 /// Refuses a flag-like token unless [`Flag::allow_hyphen_values`] is set:
2651 /// `--jobs --force` is far more likely a forgotten value than a deliberate
2652 /// one, and the attached form is available for the deliberate case. Declared,
2653 /// the next token is taken whatever it looks like, including `--`.
2654 fn take_detached_value(
2655 &mut self,
2656 flag: &'t Flag<'t>,
2657 ) -> Result<Option<&'v [u8]>, Error<'t, 'v>> {
2658 if flag.require_equals {
2659 return self.missing_or_default(flag);
2660 }
2661 match self.argv.get(self.pos) {
2662 Some(next)
2663 if self.clause_separator(bytes(next)).is_none()
2664 && (flag.allow_hyphen_values
2665 || !is_flag_like(bytes(next))
2666 || (flag.allow_negative_numbers && is_negative_number(bytes(next)))) =>
2667 {
2668 self.pos += 1;
2669 Ok(Some(bytes(next)))
2670 }
2671 _ => self.missing_or_default(flag),
2672 }
2673 }
2674
2675 fn missing_or_default(&self, flag: &'t Flag<'t>) -> Result<Option<&'v [u8]>, Error<'t, 'v>> {
2676 match flag.default_missing {
2677 Some(value) => Ok(Some(value)),
2678 None if flag.value_optional => Ok(None),
2679 None => Err(Error::MissingFlagValue { flag }),
2680 }
2681 }
2682
2683 fn clause_separator(&self, token: &[u8]) -> Option<&'t Clause<'t>> {
2684 (!self.separator_seen)
2685 .then_some(self.cmd.clause)
2686 .flatten()
2687 .filter(|clause| clause.separator.is_some_and(|separator| token == separator))
2688 }
2689
2690 fn word(&mut self, token: &'v [u8]) -> Result<Event<'t, 'a, 'v>, Error<'t, 'v>> {
2691 // Subcommands are only matched where descent is still possible: once a
2692 // positional of this command has taken a word, a later word that happens
2693 // to equal a subcommand name is just a value.
2694 if !self.arg_filled && !self.flags_stopped {
2695 if let Some(sub) = self.find_subcommand(token) {
2696 if self.cmd.args_conflicts_with_subcommands && self.command_arg_found {
2697 return Err(Error::SubcommandConflict { subcommand: sub });
2698 }
2699 self.descend(sub)?;
2700 return Ok(Event::Command(sub));
2701 }
2702
2703 // `ex help config ls` — the line every page with a Commands section has printed
2704 // all along ("help Print this message or the help of the given subcommand(s)"),
2705 // and which until now did nothing. The page is what decides the condition here:
2706 // it prints that line where there are subcommands, so that is where the word is
2707 // answered, and to a leaf `help` is a word like any other.
2708 //
2709 // Asked *after* the subcommand lookup, so a CLI that declares a `help` of its own
2710 // keeps it — the same rule the two help flags follow.
2711 //
2712 // The words after it name a command, resolved here rather than descended into:
2713 // descending would bind them, and they are a question rather than an invocation.
2714 if token == b"help"
2715 && !self.cmd.disable_help_subcommand
2716 && !self.cmd.subcommands.is_empty()
2717 {
2718 let mut cmd = self.cmd;
2719 let from = self.pos;
2720 while let Some(next) = self.argv.get(self.pos) {
2721 let Some(sub) = find_named(cmd, bytes(next)) else {
2722 break;
2723 };
2724 cmd = sub;
2725 self.pos += 1;
2726 }
2727 // Kept for `help::route_to`: which mount was asked about is not recoverable
2728 // from `cmd`, since two mounts of one `Subcommands` type are one address.
2729 self.help_span = (from, self.pos);
2730 // The long form, as `ex config --help` gives: someone who typed a whole word to
2731 // ask for help wants the fuller answer.
2732 return Err(Error::Help { cmd, long: true });
2733 }
2734
2735 // A word that names no subcommand goes to the default one, if there is one.
2736 //
2737 // Only a word, though. A dash-prefixed token that named no flag arrives here as a
2738 // value — that is what `unknown_flags = value` means — and it was never a
2739 // candidate to *select* anything, so it binds where it was typed. usage-lib stops
2740 // looking for subcommands at an unrecognised flag for the same reason. `--` is
2741 // excluded on the same grounds: it reaches this function only when a `preserve`
2742 // argument wants it as a value.
2743 //
2744 // The token is *not* consumed: the cursor steps back so the next event reads it
2745 // again, now against the command just descended into. That is what lets it be a
2746 // subcommand of the default (`mise build` where `build` is a task the mount
2747 // added) as easily as an argument of it, without this function having to decide
2748 // which — and without yielding two events for one word.
2749 if let Some(default) = self.cmd.default_subcommand {
2750 // `-` joins `--` in being excluded, and for the reason already written above:
2751 // a value was never a candidate to *select* anything. `is_flag_like` calls a
2752 // lone `-` a value — conventionally stdin — so it passed this guard and
2753 // descended, where mise's `run` has no positional and the parse failed.
2754 // usage-lib and clap both bind it to the root's own `[TASK]` instead.
2755 let default_accepts_negative = is_negative_number(token)
2756 && default
2757 .args
2758 .first()
2759 .is_some_and(|arg| arg.allow_negative_numbers);
2760 if !self.default_taken
2761 && (!is_flag_like(token) || default_accepts_negative)
2762 && token != b"--"
2763 && token != b"-"
2764 {
2765 self.default_taken = true;
2766 self.descend(default)?;
2767 self.pos -= 1;
2768 // Unlike an explicitly named command, the default command receives the
2769 // word that caused descent. Keep its argv boundary at that word so
2770 // command-level policies and completion callbacks see the same input the
2771 // command parser is about to re-read.
2772 self.cmd_start = self.pos;
2773 return Ok(Event::Command(default));
2774 }
2775 }
2776
2777 // Known subcommands and the default route keep precedence. A sigil positional
2778 // then claims its classified word before an external-subcommand catch-all can.
2779 if let Some((arg, sigil)) = self.match_sigil_arg(token) {
2780 if token.len() == sigil.len() {
2781 return Err(invalid_value_error(
2782 arg.name,
2783 as_str(token).unwrap_or_default().to_string(),
2784 format!(
2785 "expected a value after sigil {:?}",
2786 as_str(sigil).unwrap_or_default()
2787 ),
2788 ));
2789 }
2790 return Ok(Event::Arg {
2791 arg,
2792 value: &token[sigil.len()..],
2793 delimit: true,
2794 });
2795 }
2796
2797 // An unmatched word that names no subcommand is forwarded as an external
2798 // command: this word, then every token after it, including flags. Known
2799 // subcommands already won above, and a default_subcommand already caught.
2800 if self.cmd.external_subcommand
2801 && (!is_flag_like(token) || is_negative_number(token))
2802 && token != b"--"
2803 && token != b"-"
2804 {
2805 let from = self.pos - 1;
2806 self.pos = self.argv.len();
2807 return Ok(Event::External {
2808 values: &self.argv[from..],
2809 });
2810 }
2811 }
2812
2813 if self.arg_filled
2814 && !self.flags_stopped
2815 && let Some((arg, sigil)) = self.match_sigil_arg(token)
2816 {
2817 if token.len() == sigil.len() {
2818 return Err(invalid_value_error(
2819 arg.name,
2820 as_str(token).unwrap_or_default().to_string(),
2821 format!(
2822 "expected a value after sigil {:?}",
2823 as_str(sigil).unwrap_or_default()
2824 ),
2825 ));
2826 }
2827 return Ok(Event::Arg {
2828 arg,
2829 value: &token[sigil.len()..],
2830 delimit: true,
2831 });
2832 }
2833
2834 self.skip_sigil_args();
2835 self.reserve_for_required_positionals();
2836 let Some(arg) = self.next_arg() else {
2837 return Err(Error::UnexpectedArg { token });
2838 };
2839
2840 if arg.double_dash == DoubleDash::Required && !self.separator_seen {
2841 return Err(Error::ArgRequiresDoubleDash { arg });
2842 }
2843
2844 self.arg_filled = true;
2845 // An `automatic` argument stops flag interpretation from here on, as
2846 // though the caller had typed the separator themselves.
2847 let trailing_value = self.separator_seen || arg.double_dash == DoubleDash::Automatic;
2848 let delimit = !(self.dont_delimit_trailing_values && trailing_value);
2849 if arg.double_dash == DoubleDash::Automatic {
2850 self.flags_stopped = true;
2851 }
2852 // A variadic keeps taking values, so the cursor stays put — until it reaches its
2853 // bound, at which point the words after it belong to whatever comes next. That is
2854 // what makes `[a]… [b]` expressible at all.
2855 if arg.var {
2856 self.arg_taken += values_in(token, delimit.then_some(arg.delimiter).flatten());
2857 // Before advancing, which resets the count: as with a variadic flag, reaching
2858 // the bound and passing it are the same event once a word can carry several
2859 // values, and only the second is a mistake.
2860 if let Some(max) = arg.var_max.filter(|max| self.arg_taken > *max) {
2861 return Err(Error::VarTooMany {
2862 name: arg.name,
2863 max: max as usize,
2864 got: self.arg_taken as usize,
2865 });
2866 }
2867 if arg.var_max.is_some_and(|max| self.arg_taken >= max) {
2868 self.advance_arg();
2869 }
2870 } else {
2871 self.advance_arg();
2872 }
2873 if self
2874 .cmd
2875 .clause
2876 .is_some_and(|clause| clause.separator.is_none() && self.next_arg().is_none())
2877 {
2878 self.pending_clause_boundary = self.cmd.clause;
2879 }
2880 Ok(Event::Arg {
2881 arg,
2882 value: token,
2883 delimit,
2884 })
2885 }
2886
2887 fn descend(&mut self, sub: &'t Command<'t>) -> Result<(), Error<'t, 'v>> {
2888 if self.depth >= MAX_DEPTH {
2889 return Err(Error::TooDeep);
2890 }
2891 self.ancestors[self.depth] = Some(self.cmd);
2892 self.starts[self.depth] = self.cmd_start;
2893 self.depth += 1;
2894 self.cmd = sub;
2895 // Only a command that says something changes it, which is what inheriting means.
2896 if let ::core::option::Option::Some(mode) = sub.unknown_flags {
2897 self.unknown_flags = mode;
2898 }
2899 self.dont_delimit_trailing_values |= sub.dont_delimit_trailing_values;
2900 // Where this command's own words start, which is what lets a completion hand a callback
2901 // the half-parsed struct of the command it was declared on rather than of the root.
2902 self.cmd_start = self.pos;
2903 self.arg_pos = 0;
2904 self.arg_taken = 0;
2905 self.arg_filled = false;
2906 self.command_arg_found = false;
2907 Ok(())
2908 }
2909
2910 /// Move to the next positional, forgetting what the last one took.
2911 fn advance_arg(&mut self) {
2912 self.skip_sigil_args();
2913 self.arg_pos += 1;
2914 self.arg_taken = 0;
2915 self.skip_sigil_args();
2916 }
2917
2918 /// A variadic flag occurrence begins, counting from zero.
2919 ///
2920 /// The value it was given on the same token counts, which is why this starts at what
2921 /// that value holds: `--include a b` with `var_max=2` takes `a` and `b`, not three
2922 /// words — and `--include a,b` has already taken both on the one token.
2923 fn start_collecting(&mut self, flag: &'t Flag<'t>, first: &[u8]) -> Result<(), Error<'t, 'v>> {
2924 self.collected = values_in(first, flag.delimiter);
2925 if let Some(max) = flag.var_max.filter(|max| self.collected > *max) {
2926 return Err(Error::VarTooMany {
2927 name: flag.name,
2928 max: max as usize,
2929 got: self.collected as usize,
2930 });
2931 }
2932 self.collecting = if flag.var_max.is_some_and(|max| self.collected >= max) {
2933 None
2934 } else {
2935 Some(flag)
2936 };
2937 Ok(())
2938 }
2939
2940 fn next_arg(&self) -> Option<&'t Arg<'t>> {
2941 self.current_args()[self.arg_pos..]
2942 .iter()
2943 .find(|arg| arg.sigil.is_none())
2944 .copied()
2945 }
2946
2947 fn skip_sigil_args(&mut self) {
2948 while self
2949 .current_args()
2950 .get(self.arg_pos)
2951 .is_some_and(|arg| arg.sigil.is_some())
2952 {
2953 self.arg_pos += 1;
2954 }
2955 }
2956
2957 fn match_sigil_arg(&self, token: &[u8]) -> Option<(&'t Arg<'t>, &'t [u8])> {
2958 if self.flags_stopped {
2959 return None;
2960 }
2961 let own = self.current_args().iter().copied();
2962 let inherited = self.ancestors[..self.depth]
2963 .iter()
2964 .rev()
2965 .filter_map(|cmd| *cmd)
2966 .flat_map(|cmd| cmd.args.iter().copied());
2967 own.chain(inherited)
2968 .filter_map(|arg| {
2969 let sigil = arg.sigil?;
2970 (token.len() >= sigil.len() && token.starts_with(sigil)).then_some((arg, sigil))
2971 })
2972 .max_by_key(|(_, sigil)| sigil.len())
2973 }
2974
2975 fn current_args(&self) -> &'t [&'t Arg<'t>] {
2976 self.cmd
2977 .clause
2978 .map(|clause| clause.args)
2979 .unwrap_or(self.cmd.args)
2980 }
2981
2982 /// Skip empty optional positionals when every remaining value is needed by a later
2983 /// required positional. This is clap's opt-in `allow_missing_positional` policy.
2984 fn reserve_for_required_positionals(&mut self) {
2985 if !self.cmd.allow_missing_positional || self.arg_taken != 0 {
2986 return;
2987 }
2988 loop {
2989 let Some(current) = self.next_arg() else {
2990 return;
2991 };
2992 if current.required {
2993 return;
2994 }
2995 let required_after = self.current_args()[self.arg_pos + 1..]
2996 .iter()
2997 .filter(|arg| arg.required && arg.sigil.is_none())
2998 .count();
2999 if required_after == 0 {
3000 return;
3001 }
3002 let remaining_values = 1 + self.argv[self.pos..]
3003 .iter()
3004 .filter(|word| {
3005 (self.flags_stopped || !is_flag_like(bytes(word)))
3006 && self.match_sigil_arg(bytes(word)).is_none()
3007 })
3008 .count();
3009 if remaining_values > required_after {
3010 return;
3011 }
3012 self.advance_arg();
3013 }
3014 }
3015
3016 #[cfg(feature = "spec")]
3017 fn view_allows_own_flag(&self, flag: &Flag<'_>) -> bool {
3018 match self.view {
3019 None => true,
3020 // The promoted command keeps its own surface. While the injected path is
3021 // still at the host root, however, only explicitly carried globals belong
3022 // to the view; root-local flags are not part of the projected executable.
3023 Some(view) => {
3024 !core::ptr::eq(self.cmd, self.root) || is_version_flag(flag) || view.carries(flag)
3025 }
3026 }
3027 }
3028
3029 #[cfg(not(feature = "spec"))]
3030 fn view_allows_own_flag(&self, _flag: &Flag<'_>) -> bool {
3031 true
3032 }
3033
3034 #[cfg(feature = "spec")]
3035 fn view_allows_inherited_flag(&self, flag: &Flag<'_>) -> bool {
3036 match self.view {
3037 None => true,
3038 // A portable view carries selected host globals, not globals declared
3039 // by intermediate commands on a multi-segment promoted path.
3040 Some(view) => {
3041 self.root
3042 .flags
3043 .iter()
3044 .any(|root| core::ptr::eq(*root, flag))
3045 && (is_version_flag(flag) || view.carries(flag))
3046 }
3047 }
3048 }
3049
3050 #[cfg(not(feature = "spec"))]
3051 fn view_allows_inherited_flag(&self, _flag: &Flag<'_>) -> bool {
3052 true
3053 }
3054
3055 #[cfg(feature = "spec")]
3056 fn inherited_flag_is_in_scope(&self, flag: &Flag<'_>) -> bool {
3057 flag.global || (self.view.is_some() && is_version_flag(flag))
3058 }
3059
3060 #[cfg(not(feature = "spec"))]
3061 fn inherited_flag_is_in_scope(&self, flag: &Flag<'_>) -> bool {
3062 flag.global
3063 }
3064
3065 /// Flags in scope: this command's own, then any ancestor's globals.
3066 ///
3067 /// Own flags come first so that a subcommand redeclaring an inherited name
3068 /// shadows it, which is what mise relies on when it redeclares root globals
3069 /// on `run` with different shorts.
3070 fn in_scope(&self) -> impl Iterator<Item = &'t Flag<'t>> + '_ {
3071 let own = self
3072 .cmd
3073 .flags
3074 .iter()
3075 .copied()
3076 .filter(|flag| self.view_allows_own_flag(flag));
3077 let inherited = self.ancestors[..self.depth]
3078 .iter()
3079 .rev()
3080 .filter_map(|c| *c)
3081 .flat_map(|c| c.flags.iter().copied())
3082 .filter(|flag| self.inherited_flag_is_in_scope(flag))
3083 .filter(|flag| self.view_allows_inherited_flag(flag));
3084 own.chain(inherited)
3085 }
3086
3087 fn find_long(&self, name: &[u8]) -> Option<&'t Flag<'t>> {
3088 self.in_scope()
3089 .find(|f| f.longs.iter().any(|l| l.as_bytes() == name))
3090 }
3091
3092 fn find_negation(&self, name: &[u8]) -> Option<&'t Flag<'t>> {
3093 self.in_scope()
3094 .find(|f| f.negate.is_some_and(|n| n.as_bytes() == name))
3095 }
3096
3097 fn find_short(&self, byte: u8) -> Option<&'t Flag<'t>> {
3098 // Only the boundary token is shared. Later tokens use ordinary child/global scope.
3099 if self.default_taken
3100 && self.pos == self.default_bundle_end
3101 && let Some(flag) = self.ancestors[0].and_then(|parent| {
3102 parent
3103 .flags
3104 .iter()
3105 .copied()
3106 .find(|f| f.shorts.contains(&byte))
3107 })
3108 {
3109 return Some(flag);
3110 }
3111 self.in_scope()
3112 .find(|f| f.shorts.contains(&byte))
3113 // As for `--help`: supplied by the parser, and only where the command has not
3114 // declared a `-h` of its own.
3115 .or(if byte == b'h' && !self.cmd.disable_help_flag {
3116 Some(&HELP_SHORT)
3117 } else if byte == b'V'
3118 && self.version_command().version
3119 && !self.version_command().disable_version_flag
3120 {
3121 Some(&VERSION_SHORT)
3122 } else {
3123 None
3124 })
3125 }
3126
3127 #[cfg(feature = "spec")]
3128 fn version_command(&self) -> &'t Command<'t> {
3129 if self.view.is_some() {
3130 self.root
3131 } else {
3132 self.cmd
3133 }
3134 }
3135
3136 #[cfg(not(feature = "spec"))]
3137 fn version_command(&self) -> &'t Command<'t> {
3138 self.cmd
3139 }
3140
3141 fn find_subcommand(&self, name: &[u8]) -> Option<&'t Command<'t>> {
3142 // Shared with `help` rather than spelled out again, so descending into a command and
3143 // asking about one cannot drift apart.
3144 find_named(self.cmd, name)
3145 }
3146}
3147
3148/// View a token as bytes.
3149///
3150/// `as_encoded_bytes` is a plain accessor with no conversion and no allocation.
3151/// The reverse direction is the one with a cost — see [`os_string_from_bytes`] —
3152/// which is why values come back as bytes.
3153fn bytes<'v>(s: &&'v OsStr) -> &'v [u8] {
3154 s.as_encoded_bytes()
3155}
3156
3157/// How many values one word carries.
3158///
3159/// One, until a delimiter is declared — and then one per separator, counting the same way
3160/// splitting on it does: `a,b` is two, `a,` is two with an empty second, and `` is one.
3161/// Counted rather than split because binding only needs the number, and the split itself
3162/// belongs to the layer that owns the values.
3163fn values_in(word: &[u8], delimiter: ::core::option::Option<u8>) -> u32 {
3164 match delimiter {
3165 Some(d) => 1 + word.iter().filter(|b| **b == d).count() as u32,
3166 None => 1,
3167 }
3168}
3169
3170/// Whether a token should be read as a flag.
3171///
3172/// `-` alone is a value, conventionally stdin. Other dash-prefixed tokens are
3173/// flag-like; a field may make the narrower negative-number exception.
3174fn is_flag_like(token: &[u8]) -> bool {
3175 matches!(token, [b'-', rest @ ..] if !rest.is_empty())
3176}
3177
3178fn is_negative_number(token: &[u8]) -> bool {
3179 token.strip_prefix(b"-").is_some_and(is_number)
3180}
3181
3182/// Whether the text after a `-` is a number, so `-1`, `-2.5`, and `-1e5` are values
3183/// while `-1x` is a flag-shaped token that names nothing.
3184///
3185/// Digits, at most one `.`, and an optional exponent. Deliberately narrower than
3186/// `f64::from_str`, which also accepts `inf` and `NaN` — `-inf` is far likelier to be
3187/// a misspelled flag than a number somebody meant to pass.
3188///
3189/// usage-lib applies the same rule, and the corpus pins the edges so the two cannot
3190/// drift apart: they disagreed about `-1e5` when this was a hand-rolled scanner on
3191/// one side and a float parse on the other.
3192///
3193/// Written out rather than deferred to `f64::from_str` because this runs on the hot
3194/// path, and a parse would mean a UTF-8 check on a slice already decided by its
3195/// bytes.
3196fn is_number(rest: &[u8]) -> bool {
3197 let (mantissa, exponent) = match rest.iter().position(|b| matches!(b, b'e' | b'E')) {
3198 Some(at) => (&rest[..at], Some(&rest[at + 1..])),
3199 None => (rest, None),
3200 };
3201
3202 let mut seen_digit = false;
3203 let mut seen_dot = false;
3204 for &b in mantissa {
3205 match b {
3206 b'0'..=b'9' => seen_digit = true,
3207 b'.' if !seen_dot => seen_dot = true,
3208 _ => return false,
3209 }
3210 }
3211 if !seen_digit {
3212 return false;
3213 }
3214
3215 match exponent {
3216 None => true,
3217 // An exponent needs digits of its own, and may carry a sign.
3218 Some(exp) => {
3219 let digits = exp
3220 .strip_prefix(b"+")
3221 .or_else(|| exp.strip_prefix(b"-"))
3222 .unwrap_or(exp);
3223 !digits.is_empty() && digits.iter().all(|b| b.is_ascii_digit())
3224 }
3225 }
3226}
3227
3228fn validate_bool_value<'t, 'v>(
3229 flag: &'t Flag<'t>,
3230 value: Option<&'v [u8]>,
3231) -> Result<Option<&'v [u8]>, Error<'t, 'v>> {
3232 match value {
3233 None | Some(b"true" | b"false") => Ok(value),
3234 Some(_) => Err(Error::InvalidChoice {
3235 name: flag.name,
3236 choices: &["true", "false"],
3237 }),
3238 }
3239}
3240
3241#[cfg(test)]
3242mod tests {
3243 use super::*;
3244
3245 #[test]
3246 fn clause_separator_resets_inner_args_and_survives_automatic_mode() {
3247 static TASK: Arg = Arg {
3248 key: 91,
3249 name: "task",
3250 ..Arg::REQUIRED
3251 };
3252 static REST: Arg = Arg {
3253 key: 92,
3254 name: "args",
3255 double_dash: DoubleDash::Automatic,
3256 ..Arg::VAR
3257 };
3258 static ROOT: Command = Command {
3259 name: "ex",
3260 clause: Some(&Clause {
3261 key: 90,
3262 name: "tasks",
3263 separator: Some(b":::"),
3264 flags: &[],
3265 args: &[&TASK, &REST],
3266 }),
3267 ..Command::EMPTY
3268 };
3269 let argv = [
3270 OsStr::new("lint"),
3271 OsStr::new("--fix"),
3272 OsStr::new(":::"),
3273 OsStr::new("test"),
3274 ];
3275 let mut parser = Parser::new(&ROOT, &argv);
3276 let mut seen = Vec::new();
3277 while let Some(event) = parser.next_event() {
3278 match event.expect("valid clause") {
3279 Event::Arg { arg, value, .. } => {
3280 seen.push((arg.name, String::from_utf8_lossy(value).into_owned()))
3281 }
3282 Event::ClauseSeparator { clause } => seen.push((clause.name, ":::".into())),
3283 _ => {}
3284 }
3285 }
3286 assert_eq!(
3287 seen,
3288 [
3289 ("task", "lint".into()),
3290 ("args", "--fix".into()),
3291 ("tasks", ":::".into()),
3292 ("task", "test".into())
3293 ]
3294 );
3295 }
3296
3297 #[test]
3298 fn clause_separator_is_not_a_scoped_flag_value() {
3299 static POSTINSTALL: Flag = Flag {
3300 key: 93,
3301 name: "postinstall",
3302 longs: &["postinstall"],
3303 takes_value: true,
3304 variadic: true,
3305 ..Flag::BOOL
3306 };
3307 static TOOL: Arg = Arg {
3308 key: 94,
3309 name: "tool",
3310 ..Arg::REQUIRED
3311 };
3312 static ROOT: Command = Command {
3313 name: "ex",
3314 flags: &[&POSTINSTALL],
3315 clause: Some(&Clause {
3316 key: 95,
3317 name: "tools",
3318 separator: Some(b":::"),
3319 flags: &[&POSTINSTALL],
3320 args: &[&TOOL],
3321 }),
3322 ..Command::EMPTY
3323 };
3324 let argv = [OsStr::new("--postinstall"), OsStr::new(":::")];
3325 let mut parser = Parser::new(&ROOT, &argv);
3326 assert!(matches!(
3327 parser.next_event(),
3328 Some(Err(Error::MissingFlagValue { flag })) if flag.key == POSTINSTALL.key
3329 ));
3330
3331 let argv = [
3332 OsStr::new("--postinstall"),
3333 OsStr::new("setup"),
3334 OsStr::new(":::"),
3335 OsStr::new("tool"),
3336 ];
3337 let mut parser = Parser::new(&ROOT, &argv);
3338 assert!(matches!(
3339 parser.next_event(),
3340 Some(Ok(Event::Flag {
3341 value: Some(b"setup"),
3342 ..
3343 }))
3344 ));
3345 assert!(matches!(
3346 parser.next_event(),
3347 Some(Ok(Event::ClauseSeparator { clause })) if clause.key == 95
3348 ));
3349 assert!(matches!(
3350 parser.next_event(),
3351 Some(Ok(Event::Arg { value: b"tool", .. }))
3352 ));
3353 }
3354
3355 static FORCE: Flag = Flag {
3356 key: 1,
3357 longs: &["force"],
3358 shorts: b"f",
3359 ..Flag::BOOL
3360 };
3361 static EXPLICIT_BOOL: Flag = Flag {
3362 key: 20,
3363 name: "color",
3364 longs: &["color"],
3365 negate: Some("no-color"),
3366 bool_value: true,
3367 ..Flag::BOOL
3368 };
3369 static EXPLICIT_BOOL_ROOT: Command = Command {
3370 name: "ex",
3371 flags: &[&EXPLICIT_BOOL],
3372 ..Command::EMPTY
3373 };
3374 static JOBS: Flag = Flag {
3375 key: 2,
3376 longs: &["jobs"],
3377 shorts: b"j",
3378 allow_negative_numbers: true,
3379 ..Flag::VALUE
3380 };
3381 static COLOR: Flag = Flag {
3382 key: 3,
3383 longs: &["color"],
3384 negate: Some("no-color"),
3385 ..Flag::BOOL
3386 };
3387 static VERBOSE: Flag = Flag {
3388 key: 4,
3389 longs: &["verbose"],
3390 shorts: b"v",
3391 global: true,
3392 ..Flag::BOOL
3393 };
3394 static FILE: Arg = Arg {
3395 key: 10,
3396 name: "file",
3397 allow_negative_numbers: true,
3398 ..Arg::REQUIRED
3399 };
3400 static REST: Arg = Arg {
3401 key: 11,
3402 name: "rest",
3403 ..Arg::VAR
3404 };
3405 static INSTALL: Command = Command {
3406 name: "install",
3407 aliases: &["i"],
3408 flags: &[&FORCE],
3409 key: 100,
3410 ..Command::EMPTY
3411 };
3412 /// Same shape as ROOT, but a CLI that owns all of its flags. The subcommand says
3413 /// nothing and inherits it, which is the point: only the root declares the mode.
3414 static STRICT_INSTALL: Command = Command {
3415 name: "install",
3416 aliases: &["i"],
3417 flags: &[&FORCE],
3418 key: 100,
3419 ..Command::EMPTY
3420 };
3421 static STRICT: Command = Command {
3422 name: "ex",
3423 flags: &[&FORCE, &JOBS, &COLOR, &VERBOSE],
3424 args: &[&FILE, &REST],
3425 subcommands: &[&STRICT_INSTALL],
3426 unknown_flags: Some(UnknownFlags::Error),
3427 ..Command::EMPTY
3428 };
3429 static ROOT: Command = Command {
3430 name: "ex",
3431 flags: &[&FORCE, &JOBS, &COLOR, &VERBOSE],
3432 args: &[&FILE, &REST],
3433 subcommands: &[&INSTALL],
3434 ..Command::EMPTY
3435 };
3436 static ARGUMENT_CONFLICT: Command = Command {
3437 name: "ex",
3438 flags: &[&FORCE],
3439 subcommands: &[&INSTALL],
3440 args_conflicts_with_subcommands: true,
3441 ..Command::EMPTY
3442 };
3443
3444 // A CLI shaped exactly like mise's root: a default subcommand, a positional of its own,
3445 // and a subcommand under the default — which is the arrangement that tells routing from
3446 // a plain positional.
3447 static TASK: Arg = Arg {
3448 key: 20,
3449 name: "task",
3450 ..Arg::REQUIRED
3451 };
3452 static RUN_TASK: Arg = Arg {
3453 key: 21,
3454 name: "run_task",
3455 ..Arg::REQUIRED
3456 };
3457 static DEEP: Command = Command {
3458 name: "deep",
3459 args: &[&RUN_TASK],
3460 key: 203,
3461 ..Command::EMPTY
3462 };
3463 static LINT: Command = Command {
3464 name: "lint",
3465 subcommands: &[&DEEP],
3466 // A default of its own, so that a parse which forgot it had already taken one would
3467 // have somewhere to go. Nothing else in these fixtures can show the latch working.
3468 default_subcommand: Some(&DEEP),
3469 key: 202,
3470 ..Command::EMPTY
3471 };
3472 static RUN: Command = Command {
3473 name: "run",
3474 args: &[&RUN_TASK],
3475 subcommands: &[&LINT],
3476 key: 200,
3477 ..Command::EMPTY
3478 };
3479 static DEFAULTING: Command = Command {
3480 name: "mise",
3481 flags: &[&VERBOSE],
3482 args: &[&TASK],
3483 subcommands: &[&RUN, &INSTALL],
3484 default_subcommand: Some(find_subcommand(&[&RUN, &INSTALL], "run")),
3485 ..Command::EMPTY
3486 };
3487
3488 /// Collect every event, or the first error.
3489 fn parse<'t: 'v, 'v>(
3490 root: &'t Command<'t>,
3491 argv: &'v [&'v OsStr],
3492 ) -> Result<Vec<Event<'t, 'v, 'v>>, Error<'t, 'v>> {
3493 let mut parser = Parser::new(root, argv);
3494 let mut events = Vec::new();
3495 while let Some(event) = parser.next_event() {
3496 events.push(event?);
3497 }
3498 Ok(events)
3499 }
3500
3501 fn argv<const N: usize>(tokens: [&str; N]) -> [&OsStr; N] {
3502 tokens.map(OsStr::new)
3503 }
3504
3505 #[test]
3506 fn long_boolean() {
3507 let a = argv(["--force"]);
3508 assert_eq!(
3509 parse(&ROOT, &a).unwrap(),
3510 vec![Event::Flag {
3511 flag: &FORCE,
3512 value: None,
3513 negated: false
3514 }]
3515 );
3516 }
3517
3518 #[test]
3519 fn long_boolean_accepts_only_opted_in_attached_values() {
3520 for (token, negated, value) in [
3521 ("--color=false", false, b"false".as_slice()),
3522 ("--color=true", false, b"true".as_slice()),
3523 ("--no-color=false", true, b"false".as_slice()),
3524 ] {
3525 let a = argv([token]);
3526 assert_eq!(
3527 parse(&EXPLICIT_BOOL_ROOT, &a).unwrap(),
3528 vec![Event::Flag {
3529 flag: &EXPLICIT_BOOL,
3530 value: Some(value),
3531 negated,
3532 }]
3533 );
3534 }
3535
3536 let a = argv(["--color=maybe"]);
3537 assert!(matches!(
3538 parse(&EXPLICIT_BOOL_ROOT, &a),
3539 Err(Error::InvalidChoice { name: "color", .. })
3540 ));
3541
3542 let a = argv(["--force=false"]);
3543 assert_eq!(
3544 parse(&ROOT, &a).unwrap(),
3545 vec![Event::Flag {
3546 flag: &FORCE,
3547 value: None,
3548 negated: false,
3549 }]
3550 );
3551 }
3552
3553 #[test]
3554 fn long_value_forms() {
3555 for tokens in [vec!["--jobs=8"], vec!["--jobs", "8"]] {
3556 let a: Vec<&OsStr> = tokens.iter().map(|t| OsStr::new(*t)).collect();
3557 assert_eq!(
3558 parse(&ROOT, &a).unwrap(),
3559 vec![Event::Flag {
3560 flag: &JOBS,
3561 value: Some(b"8"),
3562 negated: false
3563 }],
3564 "{tokens:?}"
3565 );
3566 }
3567 }
3568
3569 #[test]
3570 fn long_value_keeps_later_equals() {
3571 let a = argv(["--jobs=a=b"]);
3572 let Event::Flag { value, .. } = parse(&ROOT, &a).unwrap()[0] else {
3573 panic!("expected a flag");
3574 };
3575 assert_eq!(value, Some(&b"a=b"[..]));
3576 }
3577
3578 #[test]
3579 fn long_value_attached_empty_is_empty_not_absent() {
3580 let a = argv(["--jobs="]);
3581 let Event::Flag { value, .. } = parse(&ROOT, &a).unwrap()[0] else {
3582 panic!("expected a flag");
3583 };
3584 assert_eq!(value, Some(&b""[..]));
3585 }
3586
3587 #[test]
3588 fn long_value_refuses_flaglike_next_word() {
3589 let a = argv(["--jobs", "--force"]);
3590 assert_eq!(
3591 parse(&ROOT, &a),
3592 Err(Error::MissingFlagValue { flag: &JOBS })
3593 );
3594 }
3595
3596 #[test]
3597 fn long_value_accepts_negative_number() {
3598 let a = argv(["--jobs", "-1"]);
3599 let Event::Flag { value, .. } = parse(&ROOT, &a).unwrap()[0] else {
3600 panic!("expected a flag");
3601 };
3602 assert_eq!(value, Some(&b"-1"[..]));
3603 }
3604
3605 #[test]
3606 fn missing_optional_positional_reserves_the_last_word() {
3607 static OPTIONAL: Arg = Arg {
3608 key: 90,
3609 name: "optional",
3610 required: false,
3611 ..Arg::REQUIRED
3612 };
3613 static REQUIRED: Arg = Arg {
3614 key: 91,
3615 name: "required",
3616 ..Arg::REQUIRED
3617 };
3618 static CMD: Command = Command {
3619 name: "ex",
3620 args: &[&OPTIONAL, &REQUIRED],
3621 allow_missing_positional: true,
3622 ..Command::EMPTY
3623 };
3624
3625 let one = argv(["value"]);
3626 assert_eq!(
3627 parse(&CMD, &one).unwrap(),
3628 vec![Event::Arg {
3629 arg: &REQUIRED,
3630 value: b"value",
3631 delimit: true
3632 }]
3633 );
3634 let two = argv(["optional", "required"]);
3635 assert_eq!(
3636 parse(&CMD, &two).unwrap(),
3637 vec![
3638 Event::Arg {
3639 arg: &OPTIONAL,
3640 value: b"optional",
3641 delimit: true
3642 },
3643 Event::Arg {
3644 arg: &REQUIRED,
3645 value: b"required",
3646 delimit: true
3647 },
3648 ]
3649 );
3650 }
3651
3652 #[test]
3653 fn negative_numbers_are_narrowly_opted_in() {
3654 static PLAIN: Flag = Flag {
3655 key: 90,
3656 name: "plain",
3657 longs: &["plain"],
3658 ..Flag::VALUE
3659 };
3660 static VALUE: Arg = Arg {
3661 key: 91,
3662 name: "value",
3663 ..Arg::REQUIRED
3664 };
3665 static CMD: Command = Command {
3666 name: "ex",
3667 flags: &[&PLAIN],
3668 args: &[&VALUE],
3669 unknown_flags: Some(UnknownFlags::Error),
3670 ..Command::EMPTY
3671 };
3672
3673 let flag = argv(["--plain", "-1"]);
3674 assert_eq!(
3675 parse(&CMD, &flag),
3676 Err(Error::MissingFlagValue { flag: &PLAIN })
3677 );
3678 let positional = argv(["-1"]);
3679 assert_eq!(
3680 parse(&CMD, &positional),
3681 Err(Error::UnknownFlag { token: b"-1" })
3682 );
3683 }
3684
3685 #[test]
3686 fn an_exact_declared_digit_short_outranks_a_negative_number() {
3687 static PRINT0: Flag = Flag {
3688 key: 92,
3689 name: "print0",
3690 shorts: b"0",
3691 ..Flag::BOOL
3692 };
3693 static VALUE: Arg = Arg {
3694 key: 93,
3695 name: "value",
3696 required: false,
3697 allow_negative_numbers: true,
3698 ..Arg::REQUIRED
3699 };
3700 static CMD: Command = Command {
3701 name: "fd",
3702 flags: &[&PRINT0],
3703 args: &[&VALUE],
3704 unknown_flags: Some(UnknownFlags::Error),
3705 ..Command::EMPTY
3706 };
3707
3708 assert_eq!(
3709 parse(&CMD, &argv(["-0"])),
3710 Ok(vec![Event::Flag {
3711 flag: &PRINT0,
3712 value: None,
3713 negated: false,
3714 }])
3715 );
3716 assert!(matches!(
3717 parse(&CMD, &argv(["-1"])),
3718 Ok(events) if matches!(events.as_slice(), [Event::Arg { value: b"-1", .. }])
3719 ));
3720 }
3721
3722 #[test]
3723 fn negation_of_value_flag_does_not_consume_a_value() {
3724 static MODE: Flag = Flag {
3725 key: 9,
3726 name: "mode",
3727 longs: &["mode"],
3728 negate: Some("no-mode"),
3729 ..Flag::VALUE
3730 };
3731 static NEGATED_VALUE: Command = Command {
3732 name: "ex",
3733 flags: &[&MODE],
3734 args: &[&FILE],
3735 ..Command::EMPTY
3736 };
3737
3738 let a = argv(["--no-mode", "input"]);
3739 assert_eq!(
3740 parse(&NEGATED_VALUE, &a).unwrap(),
3741 vec![
3742 Event::Flag {
3743 flag: &MODE,
3744 value: None,
3745 negated: true
3746 },
3747 Event::Arg {
3748 arg: &FILE,
3749 value: b"input",
3750 delimit: true,
3751 }
3752 ]
3753 );
3754 }
3755
3756 #[test]
3757 fn no_abbreviation() {
3758 // A prefix names no flag, so by default it is a value like any other word.
3759 let a = argv(["--forc"]);
3760 assert_eq!(
3761 parse(&ROOT, &a).unwrap(),
3762 vec![Event::Arg {
3763 arg: &FILE,
3764 value: b"--forc",
3765 delimit: true,
3766 }]
3767 );
3768
3769 // And a CLI that owns its flags hears about it, which is the whole reason
3770 // the strict mode exists.
3771 assert!(matches!(
3772 parse(&STRICT, &a),
3773 Err(Error::UnknownFlag { token: b"--forc" })
3774 ));
3775 }
3776
3777 #[test]
3778 fn an_unknown_flag_is_a_value_by_default() {
3779 // The default, and the case it is for: a command line being forwarded to
3780 // something whose flags this spec does not know.
3781 let a = argv(["--wat", "keep"]);
3782 assert_eq!(
3783 parse(&ROOT, &a).unwrap(),
3784 vec![
3785 Event::Arg {
3786 arg: &FILE,
3787 value: b"--wat",
3788 delimit: true,
3789 },
3790 Event::Arg {
3791 arg: &REST,
3792 value: b"keep",
3793 delimit: true,
3794 },
3795 ]
3796 );
3797
3798 // With nowhere to put it, it is an unexpected argument — the same error an
3799 // extra word gets, rather than a special one about flags.
3800 static ONE: Command = Command {
3801 name: "ex",
3802 args: &[&FILE],
3803 ..Command::EMPTY
3804 };
3805 let a = argv(["a", "--wat"]);
3806 assert_eq!(
3807 parse(&ONE, &a),
3808 Err(Error::UnexpectedArg { token: b"--wat" })
3809 );
3810 }
3811
3812 #[test]
3813 fn negation() {
3814 let a = argv(["--no-color"]);
3815 assert_eq!(
3816 parse(&ROOT, &a).unwrap(),
3817 vec![Event::Flag {
3818 flag: &COLOR,
3819 value: None,
3820 negated: true
3821 }]
3822 );
3823 }
3824
3825 #[test]
3826 fn short_bundle_and_attached_value() {
3827 let a = argv(["-fj8"]);
3828 assert_eq!(
3829 parse(&ROOT, &a).unwrap(),
3830 vec![
3831 Event::Flag {
3832 flag: &FORCE,
3833 value: None,
3834 negated: false
3835 },
3836 Event::Flag {
3837 flag: &JOBS,
3838 value: Some(b"8"),
3839 negated: false
3840 },
3841 ]
3842 );
3843 }
3844
3845 #[test]
3846 fn short_value_strips_one_equals() {
3847 for (tokens, want) in [(["-j=8"], &b"8"[..]), (["-j==8"], &b"=8"[..])] {
3848 let a = argv(tokens);
3849 let Event::Flag { value, .. } = parse(&ROOT, &a).unwrap()[0] else {
3850 panic!("expected a flag");
3851 };
3852 assert_eq!(value, Some(want), "{tokens:?}");
3853 }
3854 }
3855
3856 #[test]
3857 fn bare_dash_is_a_value() {
3858 let a = argv(["-"]);
3859 assert_eq!(
3860 parse(&ROOT, &a).unwrap(),
3861 vec![Event::Arg {
3862 arg: &FILE,
3863 value: b"-",
3864 delimit: true,
3865 }]
3866 );
3867 }
3868
3869 #[test]
3870 fn positionals_then_variadic() {
3871 let a = argv(["one", "two", "three"]);
3872 assert_eq!(
3873 parse(&ROOT, &a).unwrap(),
3874 vec![
3875 Event::Arg {
3876 arg: &FILE,
3877 value: b"one",
3878 delimit: true,
3879 },
3880 Event::Arg {
3881 arg: &REST,
3882 value: b"two",
3883 delimit: true,
3884 },
3885 Event::Arg {
3886 arg: &REST,
3887 value: b"three",
3888 delimit: true,
3889 },
3890 ]
3891 );
3892 }
3893
3894 #[test]
3895 fn subcommand_and_alias_route_the_same() {
3896 for token in ["install", "i"] {
3897 let a = argv([token]);
3898 assert_eq!(
3899 parse(&ROOT, &a).unwrap(),
3900 vec![Event::Command(&INSTALL)],
3901 "{token}"
3902 );
3903 }
3904 }
3905
3906 #[test]
3907 fn a_parent_argument_can_exclude_a_later_subcommand() {
3908 let a = argv(["--force", "install"]);
3909 assert!(matches!(
3910 parse(&ARGUMENT_CONFLICT, &a),
3911 Err(Error::SubcommandConflict { subcommand }) if subcommand.name == "install"
3912 ));
3913 }
3914
3915 #[test]
3916 fn subcommand_only_routes_before_a_positional_is_filled() {
3917 let a = argv(["other", "install"]);
3918 assert_eq!(
3919 parse(&ROOT, &a).unwrap(),
3920 vec![
3921 Event::Arg {
3922 arg: &FILE,
3923 value: b"other",
3924 delimit: true,
3925 },
3926 Event::Arg {
3927 arg: &REST,
3928 value: b"install",
3929 delimit: true,
3930 },
3931 ]
3932 );
3933 }
3934
3935 #[test]
3936 fn a_word_naming_no_subcommand_goes_to_the_default_one() {
3937 // usage-lib's answer, which this reproduces: `mise build` comes back as commands
3938 // `["mise", "run"]` with the word bound to *run's* argument — not to `mise`'s own
3939 // `[TASK]`, which is what makes this more than a synonym for a positional.
3940 let a = argv(["build"]);
3941 assert_eq!(
3942 parse(&DEFAULTING, &a).unwrap(),
3943 vec![
3944 Event::Command(&RUN),
3945 Event::Arg {
3946 arg: &RUN_TASK,
3947 value: b"build",
3948 delimit: true,
3949 },
3950 ]
3951 );
3952 }
3953
3954 // A shared table, as a flattened struct's would be. Declared outside the tests so both
3955 // can splice it, which is the arrangement it exists to model.
3956 static SHARED_QUIET: Flag = Flag {
3957 key: 300,
3958 name: "quiet",
3959 longs: &["quiet"],
3960 ..Flag::BOOL
3961 };
3962 static SHARED_FLAGS: &[&Flag] = &[&SHARED_QUIET];
3963 static SHARED_WHAT: Arg = Arg {
3964 key: 301,
3965 name: "what",
3966 ..Arg::REQUIRED
3967 };
3968 static SHARED_ARGS: &[&Arg] = &[&SHARED_WHAT];
3969
3970 #[test]
3971 fn concatenating_tables_keeps_the_order_they_were_given_in() {
3972 // The property positional arguments depend on: a flattened group lands where the
3973 // field was written, not at the end. `[&FILE], SHARED, [&REST]` has to stay in that
3974 // order or `ex a b c` binds the wrong words.
3975 const ARGS: &[&[&Arg]] = &[&[&FILE], SHARED_ARGS, &[&REST]];
3976 static TABLE: [&Arg; table_len(ARGS)] = concat_args(ARGS);
3977 assert_eq!(
3978 TABLE.iter().map(|a| a.name).collect::<Vec<_>>(),
3979 ["file", "what", "rest"]
3980 );
3981
3982 // Empty groups contribute nothing and disturb nothing, which is what lets the derive
3983 // emit a group per field without checking whether it is empty first.
3984 const WITH_GAPS: &[&[&Flag]] = &[&[], &[&FORCE], &[], SHARED_FLAGS, &[]];
3985 static FLAGS: [&Flag; table_len(WITH_GAPS)] = concat_flags(WITH_GAPS);
3986 // By long form: these fixtures do not all set `name`, and the placeholder's is also
3987 // empty — so comparing names could not tell a real entry from a leftover slot.
3988 assert_eq!(
3989 FLAGS.iter().map(|f| f.longs).collect::<Vec<_>>(),
3990 [&["force"], &["quiet"]]
3991 );
3992 }
3993
3994 #[test]
3995 fn a_concatenated_table_parses_like_a_declared_one() {
3996 // The point of doing this at compile time: what the parser walks is one flat slice,
3997 // indistinguishable from a command that declared everything itself.
3998 const FLAG_GROUPS: &[&[&Flag]] = &[&[&FORCE], SHARED_FLAGS];
3999 const ARG_GROUPS: &[&[&Arg]] = &[SHARED_ARGS, &[&REST]];
4000 static FLAGS: [&Flag; table_len(FLAG_GROUPS)] = concat_flags(FLAG_GROUPS);
4001 static ARGS: [&Arg; table_len(ARG_GROUPS)] = concat_args(ARG_GROUPS);
4002 static JOINED: Command = Command {
4003 name: "joined",
4004 flags: &FLAGS,
4005 args: &ARGS,
4006 ..Command::EMPTY
4007 };
4008
4009 let a = argv(["--quiet", "one", "two", "--force"]);
4010 assert_eq!(
4011 parse(&JOINED, &a).unwrap(),
4012 vec![
4013 Event::Flag {
4014 flag: &SHARED_QUIET,
4015 value: None,
4016 negated: false
4017 },
4018 Event::Arg {
4019 arg: &SHARED_WHAT,
4020 value: b"one",
4021 delimit: true,
4022 },
4023 Event::Arg {
4024 arg: &REST,
4025 value: b"two",
4026 delimit: true,
4027 },
4028 Event::Flag {
4029 flag: &FORCE,
4030 value: None,
4031 negated: false
4032 },
4033 ]
4034 );
4035 }
4036
4037 #[test]
4038 fn an_unknown_flag_is_not_routed() {
4039 // A dash-prefixed token that names no flag becomes a value here (the default for
4040 // `unknown_flags`), and it must not thereby become a *subcommand* word: usage-lib
4041 // stops looking for subcommands at an unrecognised flag, and binds it to the command
4042 // still in scope. Verified against usage-lib, where `ex --wat` comes back as commands
4043 // `["ex"]` with `ROOT_TASK = "--wat"`.
4044 for token in ["--wat", "-x"] {
4045 let a = argv([token]);
4046 assert_eq!(
4047 parse(&DEFAULTING, &a).unwrap(),
4048 vec![Event::Arg {
4049 arg: &TASK,
4050 value: token.as_bytes(),
4051 delimit: true,
4052 }],
4053 "{token} should bind where it was typed, not in the default subcommand"
4054 );
4055 }
4056 }
4057
4058 #[test]
4059 fn a_named_subcommand_is_not_routed() {
4060 // The default is for words that name nothing. A word that names a sibling still
4061 // selects it, and the root's own argument is still reachable behind one.
4062 let a = argv(["install"]);
4063 assert_eq!(
4064 parse(&DEFAULTING, &a).unwrap(),
4065 vec![Event::Command(&INSTALL)]
4066 );
4067 }
4068
4069 #[test]
4070 fn an_unmatched_word_is_forwarded_when_external_subcommand_is_set() {
4071 static CATCH: Command = Command {
4072 name: "ex",
4073 flags: &[&VERBOSE],
4074 subcommands: &[&INSTALL],
4075 external_subcommand: true,
4076 unknown_flags: Some(UnknownFlags::Error),
4077 ..Command::EMPTY
4078 };
4079 let a = argv(["foo", "--help", "bar"]);
4080 assert_eq!(
4081 parse(&CATCH, &a).unwrap(),
4082 vec![Event::External { values: &a[..] }]
4083 );
4084
4085 let a = argv(["install"]);
4086 assert_eq!(parse(&CATCH, &a).unwrap(), vec![Event::Command(&INSTALL)]);
4087
4088 let a = argv(["--verbose", "foo", "--verbose"]);
4089 assert_eq!(
4090 parse(&CATCH, &a).unwrap(),
4091 vec![
4092 Event::Flag {
4093 flag: &VERBOSE,
4094 value: None,
4095 negated: false
4096 },
4097 Event::External { values: &a[1..] }
4098 ]
4099 );
4100
4101 let a = argv(["--wat"]);
4102 assert_eq!(
4103 parse(&CATCH, &a),
4104 Err(Error::UnknownFlag { token: b"--wat" })
4105 );
4106
4107 // A negative number is a value, not a flag, so it can be the unmatched word.
4108 let a = argv(["-1", "rest"]);
4109 assert_eq!(
4110 parse(&CATCH, &a).unwrap(),
4111 vec![Event::External { values: &a[..] }]
4112 );
4113 }
4114
4115 #[test]
4116 fn a_default_subcommand_outranks_an_external_one() {
4117 static CATCH_DEFAULT: Command = Command {
4118 name: "ex",
4119 subcommands: &[&RUN],
4120 default_subcommand: Some(&RUN),
4121 external_subcommand: true,
4122 ..Command::EMPTY
4123 };
4124 let a = argv(["build"]);
4125 assert_eq!(
4126 parse(&CATCH_DEFAULT, &a).unwrap(),
4127 vec![
4128 Event::Command(&RUN),
4129 Event::Arg {
4130 arg: &RUN_TASK,
4131 value: b"build",
4132 delimit: true,
4133 }
4134 ]
4135 );
4136 }
4137
4138 #[test]
4139 fn a_default_subcommand_starts_at_the_word_it_receives() {
4140 let a = argv(["build"]);
4141 let mut parser = Parser::new(&DEFAULTING, &a);
4142 assert_eq!(parser.next_event(), Some(Ok(Event::Command(&RUN))));
4143 assert_eq!(parser.command_start(), 0);
4144 assert_eq!(
4145 parser.next_event(),
4146 Some(Ok(Event::Arg {
4147 arg: &RUN_TASK,
4148 value: b"build",
4149 delimit: true,
4150 }))
4151 );
4152 }
4153
4154 #[test]
4155 fn a_root_clause_separator_does_not_select_empty_default() {
4156 static CHILD: Command = Command {
4157 name: "run",
4158 ..Command::EMPTY
4159 };
4160 static ROOT: Command = Command {
4161 name: "ex",
4162 clause: Some(&Clause {
4163 key: 400,
4164 name: "items",
4165 separator: Some(b":::"),
4166 flags: &[],
4167 args: &[],
4168 }),
4169 subcommands: &[&CHILD],
4170 default_subcommand: Some(&CHILD),
4171 default_subcommand_on_empty: true,
4172 ..Command::EMPTY
4173 };
4174 let a = argv([":::"]);
4175 assert_eq!(
4176 parse(&ROOT, &a).unwrap(),
4177 vec![Event::ClauseSeparator {
4178 clause: *ROOT.clause.unwrap(),
4179 }]
4180 );
4181 }
4182
4183 #[test]
4184 fn the_default_can_be_named_by_an_alias() {
4185 // usage-lib resolves the name against subcommand names, aliases and hidden aliases
4186 // alike, so a spec may point `default_subcommand` at any of them.
4187 static BY_ALIAS: Command = Command {
4188 name: "mise",
4189 args: &[&TASK],
4190 subcommands: &[&INSTALL],
4191 // `INSTALL` answers to "i" as well as to its name.
4192 default_subcommand: Some(find_subcommand(&[&INSTALL], "i")),
4193 ..Command::EMPTY
4194 };
4195 assert!(::core::ptr::eq(
4196 BY_ALIAS.default_subcommand.expect("declared"),
4197 &INSTALL
4198 ));
4199 }
4200
4201 #[test]
4202 fn a_name_outranks_another_commands_alias() {
4203 // A spec `assert_unique_subcommand_names` would reject, resolved anyway: a parser
4204 // handed a table nothing validated still has to answer, and the answer is the
4205 // command whose own name it is. Both orders, because taking the first candidate
4206 // that matched on either name or alias made this depend on which was listed first
4207 // — and usage-lib, building a map, took the last.
4208 static ALPHA: Command = Command {
4209 name: "alpha",
4210 aliases: &["run"],
4211 key: 300,
4212 ..Command::EMPTY
4213 };
4214 static PLAIN_RUN: Command = Command {
4215 name: "run",
4216 key: 301,
4217 ..Command::EMPTY
4218 };
4219 for subcommands in [&[&ALPHA, &PLAIN_RUN] as &[&Command], &[&PLAIN_RUN, &ALPHA]] {
4220 assert!(::core::ptr::eq(
4221 find_subcommand(subcommands, "run"),
4222 &PLAIN_RUN
4223 ));
4224 let root: Command = Command {
4225 name: "ex",
4226 subcommands,
4227 ..Command::EMPTY
4228 };
4229 let a = argv(["run"]);
4230 assert_eq!(parse(&root, &a).unwrap(), vec![Event::Command(&PLAIN_RUN)]);
4231 // The alias still reaches its own command by every name it does not share.
4232 let a = argv(["alpha"]);
4233 assert_eq!(parse(&root, &a).unwrap(), vec![Event::Command(&ALPHA)]);
4234 // `ex help run` asks about the command `ex run` selects. These are separate
4235 // lookups — help resolves a path without descending — and answering differently
4236 // for a colliding word is the divergence this rule exists to end.
4237 let a = argv(["help", "run"]);
4238 match parse(&root, &a) {
4239 Err(Error::Help { cmd, .. }) => {
4240 assert!(
4241 ::core::ptr::eq(cmd, &PLAIN_RUN),
4242 "got help for {}",
4243 cmd.name
4244 )
4245 }
4246 other => panic!("expected a help request, got {other:?}"),
4247 }
4248 }
4249 }
4250
4251 #[test]
4252 #[should_panic(expected = "two subcommands answer to the same name")]
4253 fn an_alias_cannot_shadow_a_sibling_command() {
4254 static ADD: Command = Command {
4255 name: "add",
4256 aliases: &["install"],
4257 ..Command::EMPTY
4258 };
4259 assert_unique_subcommand_names(&[&INSTALL, &ADD]);
4260 }
4261
4262 #[test]
4263 fn the_word_is_re_examined_against_the_command_it_reached() {
4264 // The reason the cursor steps back rather than the token being consumed: `lint` names
4265 // nothing at the root, and once inside `run` it names a subcommand. mise's mounted
4266 // task names arrive exactly this way.
4267 let a = argv(["lint"]);
4268 assert_eq!(
4269 parse(&DEFAULTING, &a).unwrap(),
4270 vec![Event::Command(&RUN), Event::Command(&LINT)]
4271 );
4272 }
4273
4274 #[test]
4275 fn the_default_is_taken_at_most_once_per_parse() {
4276 // usage-lib latches this for the whole parse rather than per command, and the shape
4277 // that shows the difference needs two of them: `lint` routes through `run`, and `lint`
4278 // declares a default too. A second word there would descend again — walking a CLI
4279 // deeper than anything the user typed — so the answer is that it does not.
4280 let a = argv(["lint", "zzz"]);
4281 assert_eq!(
4282 parse(&DEFAULTING, &a),
4283 Err(Error::UnexpectedArg { token: b"zzz" }),
4284 "the second word must not reach `deep`"
4285 );
4286
4287 // Reached explicitly, the same command still takes it: the latch bounds routing, not
4288 // the tree.
4289 let a = argv(["lint", "deep", "zzz"]);
4290 assert_eq!(
4291 parse(&DEFAULTING, &a).unwrap(),
4292 vec![
4293 Event::Command(&RUN),
4294 Event::Command(&LINT),
4295 Event::Command(&DEEP),
4296 Event::Arg {
4297 arg: &RUN_TASK,
4298 value: b"zzz",
4299 delimit: true,
4300 },
4301 ]
4302 );
4303 }
4304
4305 #[test]
4306 fn a_flag_before_the_word_still_belongs_to_the_root() {
4307 // Routing happens at the word, so anything typed before it was addressed to the
4308 // command the user was actually at.
4309 let a = argv(["--verbose", "build"]);
4310 assert_eq!(
4311 parse(&DEFAULTING, &a).unwrap(),
4312 vec![
4313 Event::Flag {
4314 flag: &VERBOSE,
4315 value: None,
4316 negated: false
4317 },
4318 Event::Command(&RUN),
4319 Event::Arg {
4320 arg: &RUN_TASK,
4321 value: b"build",
4322 delimit: true,
4323 },
4324 ]
4325 );
4326 }
4327
4328 #[test]
4329 fn nothing_routes_after_the_separator() {
4330 // Past `--` there are no subcommands left to select, so there is no default to reach
4331 // either: the words are values of whatever the command declares.
4332 let a = argv(["--", "build"]);
4333 assert_eq!(
4334 parse(&DEFAULTING, &a).unwrap(),
4335 vec![Event::Arg {
4336 arg: &TASK,
4337 value: b"build",
4338 delimit: true,
4339 }]
4340 );
4341 }
4342
4343 #[test]
4344 fn globals_are_inherited_but_plain_flags_are_not() {
4345 let a = argv(["install", "--verbose"]);
4346 assert_eq!(
4347 parse(&ROOT, &a).unwrap(),
4348 vec![
4349 Event::Command(&INSTALL),
4350 Event::Flag {
4351 flag: &VERBOSE,
4352 value: None,
4353 negated: false
4354 }
4355 ]
4356 );
4357
4358 // `--jobs` belongs to the root and is not global, so it is not a flag here.
4359 // Strictly that is an unknown flag; leniently it is a word, and `install`
4360 // declares no argument to hold one — either way it is never read as the
4361 // root's flag, which is what this test is about.
4362 let a = argv(["install", "--jobs", "8"]);
4363 assert!(matches!(parse(&STRICT, &a), Err(Error::UnknownFlag { .. })));
4364 assert!(matches!(
4365 parse(&ROOT, &a),
4366 Err(Error::UnexpectedArg { token: b"--jobs" })
4367 ));
4368 }
4369
4370 #[test]
4371 fn double_dash_protects_flaglike_values() {
4372 let a = argv(["--", "--force", "-x"]);
4373 assert_eq!(
4374 parse(&ROOT, &a).unwrap(),
4375 vec![
4376 Event::Arg {
4377 arg: &FILE,
4378 value: b"--force",
4379 delimit: true,
4380 },
4381 Event::Arg {
4382 arg: &REST,
4383 value: b"-x",
4384 delimit: true,
4385 },
4386 ]
4387 );
4388 }
4389
4390 #[test]
4391 fn second_double_dash_is_a_value() {
4392 let a = argv(["--", "a", "--", "b"]);
4393 let values: Vec<&[u8]> = parse(&ROOT, &a)
4394 .unwrap()
4395 .iter()
4396 .filter_map(|e| match e {
4397 Event::Arg { value, .. } => Some(*value),
4398 _ => None,
4399 })
4400 .collect();
4401 assert_eq!(values, vec![&b"a"[..], &b"--"[..], &b"b"[..]]);
4402 }
4403
4404 #[test]
4405 fn allow_hyphen_values_takes_a_flaglike_detached_value() {
4406 static ARGS: Flag = Flag {
4407 key: 6,
4408 name: "args",
4409 longs: &["args"],
4410 shorts: b"a",
4411 takes_value: true,
4412 allow_hyphen_values: true,
4413 ..Flag::BOOL
4414 };
4415 static DIR: Flag = Flag {
4416 key: 7,
4417 name: "working-dir",
4418 longs: &["working-dir"],
4419 shorts: b"d",
4420 ..Flag::VALUE
4421 };
4422 static HYPHEN: Command = Command {
4423 name: "ex",
4424 flags: &[&ARGS, &DIR],
4425 args: &[&REST],
4426 ..Command::EMPTY
4427 };
4428
4429 let a = argv(["-a", "-destroy"]);
4430 assert_eq!(
4431 parse(&HYPHEN, &a).unwrap(),
4432 vec![Event::Flag {
4433 flag: &ARGS,
4434 value: Some(b"-destroy"),
4435 negated: false
4436 }]
4437 );
4438
4439 let a = argv(["--args", "--", "-x"]);
4440 assert_eq!(
4441 parse(&HYPHEN, &a).unwrap(),
4442 vec![
4443 Event::Flag {
4444 flag: &ARGS,
4445 value: Some(b"--"),
4446 negated: false
4447 },
4448 Event::Arg {
4449 arg: &REST,
4450 value: b"-x",
4451 delimit: true,
4452 },
4453 ]
4454 );
4455 }
4456
4457 #[test]
4458 fn require_equals_refuses_a_detached_value() {
4459 static INSPECT: Flag = Flag {
4460 key: 8,
4461 name: "inspect",
4462 longs: &["inspect"],
4463 shorts: b"i",
4464 takes_value: true,
4465 require_equals: true,
4466 ..Flag::BOOL
4467 };
4468 static EQ: Command = Command {
4469 name: "ex",
4470 flags: &[&INSPECT],
4471 ..Command::EMPTY
4472 };
4473
4474 let a = argv(["--inspect=9229"]);
4475 assert_eq!(
4476 parse(&EQ, &a).unwrap(),
4477 vec![Event::Flag {
4478 flag: &INSPECT,
4479 value: Some(b"9229"),
4480 negated: false
4481 }]
4482 );
4483
4484 let a = argv(["--inspect", "9229"]);
4485 assert!(matches!(
4486 parse(&EQ, &a),
4487 Err(Error::MissingFlagValue { .. })
4488 ));
4489
4490 let a = argv(["-i9229"]);
4491 assert_eq!(
4492 parse(&EQ, &a).unwrap(),
4493 vec![Event::Flag {
4494 flag: &INSPECT,
4495 value: Some(b"9229"),
4496 negated: false
4497 }]
4498 );
4499
4500 static ALL: Flag = Flag {
4501 key: 9,
4502 name: "all",
4503 longs: &["all"],
4504 shorts: b"a",
4505 ..Flag::BOOL
4506 };
4507 static BUNDLE: Command = Command {
4508 name: "ex",
4509 flags: &[&ALL, &INSPECT],
4510 ..Command::EMPTY
4511 };
4512 let a = argv(["-ai", "9229"]);
4513 assert!(
4514 matches!(parse(&BUNDLE, &a), Err(Error::MissingFlagValue { .. })),
4515 "a require_equals short reached through a bundle still refuses the following word"
4516 );
4517 }
4518
4519 #[test]
4520 fn default_missing_binds_when_the_value_is_left_off() {
4521 static COLOR: Flag = Flag {
4522 key: 9,
4523 name: "color",
4524 longs: &["color"],
4525 takes_value: true,
4526 default_missing: Some(b"always"),
4527 ..Flag::BOOL
4528 };
4529 static VERBOSE: Flag = Flag {
4530 key: 10,
4531 name: "verbose",
4532 longs: &["verbose"],
4533 ..Flag::BOOL
4534 };
4535 static MISSING: Command = Command {
4536 name: "ex",
4537 flags: &[&COLOR, &VERBOSE],
4538 ..Command::EMPTY
4539 };
4540
4541 let a = argv(["--color"]);
4542 assert_eq!(
4543 parse(&MISSING, &a).unwrap(),
4544 vec![Event::Flag {
4545 flag: &COLOR,
4546 value: Some(b"always"),
4547 negated: false
4548 }]
4549 );
4550
4551 let a = argv(["--color=never"]);
4552 assert_eq!(
4553 parse(&MISSING, &a).unwrap(),
4554 vec![Event::Flag {
4555 flag: &COLOR,
4556 value: Some(b"never"),
4557 negated: false
4558 }]
4559 );
4560
4561 let a = argv(["--color", "--verbose"]);
4562 assert_eq!(
4563 parse(&MISSING, &a).unwrap(),
4564 vec![
4565 Event::Flag {
4566 flag: &COLOR,
4567 value: Some(b"always"),
4568 negated: false
4569 },
4570 Event::Flag {
4571 flag: &VERBOSE,
4572 value: None,
4573 negated: false
4574 },
4575 ]
4576 );
4577
4578 let a = argv(["--color="]);
4579 assert_eq!(
4580 parse(&MISSING, &a).unwrap(),
4581 vec![Event::Flag {
4582 flag: &COLOR,
4583 value: Some(b""),
4584 negated: false
4585 }]
4586 );
4587 }
4588
4589 #[test]
4590 fn optional_flag_value_distinguishes_bare_and_explicit_forms() {
4591 static BUMP: Flag = Flag {
4592 key: 11,
4593 name: "bump",
4594 longs: &["bump"],
4595 takes_value: true,
4596 value_optional: true,
4597 ..Flag::BOOL
4598 };
4599 static OPTIONAL: Command = Command {
4600 name: "ex",
4601 flags: &[&BUMP],
4602 ..Command::EMPTY
4603 };
4604
4605 assert_eq!(parse(&OPTIONAL, &argv([])).unwrap(), vec![]);
4606 assert_eq!(
4607 parse(&OPTIONAL, &argv(["--bump"])).unwrap(),
4608 vec![Event::Flag {
4609 flag: &BUMP,
4610 value: None,
4611 negated: false,
4612 }]
4613 );
4614 assert_eq!(
4615 parse(&OPTIONAL, &argv(["--bump=5"])).unwrap(),
4616 vec![Event::Flag {
4617 flag: &BUMP,
4618 value: Some(b"5"),
4619 negated: false,
4620 }]
4621 );
4622
4623 static INCLUDE: Flag = Flag {
4624 key: 12,
4625 name: "include",
4626 longs: &["include"],
4627 takes_value: true,
4628 variadic: true,
4629 value_optional: true,
4630 ..Flag::BOOL
4631 };
4632 static VERBOSE: Flag = Flag {
4633 key: 13,
4634 name: "verbose",
4635 longs: &["verbose"],
4636 ..Flag::BOOL
4637 };
4638 static VARIADIC: Command = Command {
4639 name: "ex",
4640 flags: &[&INCLUDE, &VERBOSE],
4641 args: &[&REST],
4642 ..Command::EMPTY
4643 };
4644 assert_eq!(
4645 parse(&VARIADIC, &argv(["--include", "--verbose", "file"])).unwrap(),
4646 vec![
4647 Event::Flag {
4648 flag: &INCLUDE,
4649 value: None,
4650 negated: false,
4651 },
4652 Event::Flag {
4653 flag: &VERBOSE,
4654 value: None,
4655 negated: false,
4656 },
4657 Event::Arg {
4658 arg: &REST,
4659 value: b"file",
4660 delimit: true,
4661 },
4662 ]
4663 );
4664 }
4665
4666 #[test]
4667 fn default_missing_with_require_equals_leaves_the_following_word() {
4668 static INSPECT: Flag = Flag {
4669 key: 11,
4670 name: "inspect",
4671 longs: &["inspect"],
4672 takes_value: true,
4673 require_equals: true,
4674 default_missing: Some(b"9229"),
4675 ..Flag::BOOL
4676 };
4677 static BOTH: Command = Command {
4678 name: "ex",
4679 flags: &[&INSPECT],
4680 args: &[&REST],
4681 ..Command::EMPTY
4682 };
4683
4684 let a = argv(["--inspect"]);
4685 assert_eq!(
4686 parse(&BOTH, &a).unwrap(),
4687 vec![Event::Flag {
4688 flag: &INSPECT,
4689 value: Some(b"9229"),
4690 negated: false
4691 }]
4692 );
4693
4694 let a = argv(["--inspect", "80"]);
4695 assert_eq!(
4696 parse(&BOTH, &a).unwrap(),
4697 vec![
4698 Event::Flag {
4699 flag: &INSPECT,
4700 value: Some(b"9229"),
4701 negated: false
4702 },
4703 Event::Arg {
4704 arg: &REST,
4705 value: b"80",
4706 delimit: true,
4707 },
4708 ]
4709 );
4710
4711 let a = argv(["--inspect="]);
4712 assert_eq!(
4713 parse(&BOTH, &a).unwrap(),
4714 vec![Event::Flag {
4715 flag: &INSPECT,
4716 value: Some(b""),
4717 negated: false
4718 }]
4719 );
4720 }
4721
4722 #[test]
4723 fn variadic_flag_collects_until_a_flaglike_token() {
4724 static INCLUDE: Flag = Flag {
4725 key: 5,
4726 name: "include",
4727 longs: &["include"],
4728 shorts: b"i",
4729 takes_value: true,
4730 variadic: true,
4731 ..Flag::BOOL
4732 };
4733 static GREEDY: Command = Command {
4734 name: "ex",
4735 flags: &[&INCLUDE, &FORCE],
4736 args: &[&FILE],
4737 ..Command::EMPTY
4738 };
4739
4740 let a = argv(["--include", "x", "y", "--force"]);
4741 assert_eq!(
4742 parse(&GREEDY, &a).unwrap(),
4743 vec![
4744 Event::Flag {
4745 flag: &INCLUDE,
4746 value: Some(b"x"),
4747 negated: false
4748 },
4749 Event::Flag {
4750 flag: &INCLUDE,
4751 value: Some(b"y"),
4752 negated: false
4753 },
4754 Event::Flag {
4755 flag: &FORCE,
4756 value: None,
4757 negated: false
4758 },
4759 ]
4760 );
4761 }
4762
4763 #[test]
4764 fn value_terminators_end_variadic_owners_without_binding() {
4765 static INCLUDE: Flag = Flag {
4766 key: 92,
4767 name: "include",
4768 longs: &["include"],
4769 takes_value: true,
4770 variadic: true,
4771 value_terminator: Some(b";"),
4772 ..Flag::BOOL
4773 };
4774 static ITEMS: Arg = Arg {
4775 key: 93,
4776 name: "items",
4777 var: true,
4778 value_terminator: Some(b";"),
4779 ..Arg::REQUIRED
4780 };
4781 static AFTER: Arg = Arg {
4782 key: 94,
4783 name: "after",
4784 ..Arg::REQUIRED
4785 };
4786 static FLAG_CMD: Command = Command {
4787 name: "ex",
4788 flags: &[&INCLUDE],
4789 args: &[&AFTER],
4790 ..Command::EMPTY
4791 };
4792 static ARG_CMD: Command = Command {
4793 name: "ex",
4794 args: &[&ITEMS, &AFTER],
4795 ..Command::EMPTY
4796 };
4797
4798 let flag = argv(["--include", "a", ";", "tail"]);
4799 assert_eq!(
4800 parse(&FLAG_CMD, &flag).unwrap(),
4801 vec![
4802 Event::Flag {
4803 flag: &INCLUDE,
4804 value: Some(b"a"),
4805 negated: false,
4806 },
4807 Event::Arg {
4808 arg: &AFTER,
4809 value: b"tail",
4810 delimit: true,
4811 },
4812 ]
4813 );
4814
4815 let positional = argv(["a", ";", "tail"]);
4816 assert_eq!(
4817 parse(&ARG_CMD, &positional).unwrap(),
4818 vec![
4819 Event::Arg {
4820 arg: &ITEMS,
4821 value: b"a",
4822 delimit: true,
4823 },
4824 Event::Arg {
4825 arg: &AFTER,
4826 value: b"tail",
4827 delimit: true,
4828 },
4829 ]
4830 );
4831 }
4832
4833 #[test]
4834 fn a_non_variadic_flag_leaves_the_next_word_alone() {
4835 // The counterpart to the test above: a flag that takes one value must not
4836 // swallow the word after it, which would silently steal a positional.
4837 let a = argv(["--jobs", "8", "keep-me"]);
4838 assert_eq!(
4839 parse(&ROOT, &a).unwrap(),
4840 vec![
4841 Event::Flag {
4842 flag: &JOBS,
4843 value: Some(b"8"),
4844 negated: false
4845 },
4846 Event::Arg {
4847 arg: &FILE,
4848 value: b"keep-me",
4849 delimit: true,
4850 },
4851 ]
4852 );
4853 }
4854
4855 #[test]
4856 fn double_dash_seen_means_a_separator_was_typed() {
4857 static FILES: Arg = Arg {
4858 key: 23,
4859 name: "files",
4860 double_dash: DoubleDash::Automatic,
4861 ..Arg::VAR
4862 };
4863 static AUTO: Command = Command {
4864 name: "ex",
4865 flags: &[&FORCE],
4866 args: &[&FILES],
4867 ..Command::EMPTY
4868 };
4869
4870 let a = argv(["--", "x"]);
4871 let mut parser = Parser::new(&ROOT, &a);
4872 while parser.next_event().is_some() {}
4873 assert!(parser.double_dash_seen(), "a real separator was consumed");
4874
4875 // `automatic` stops flag interpretation without a separator being typed,
4876 // and reporting one would be a lie to any caller that forwards argv.
4877 let a = argv(["x", "--force"]);
4878 let mut parser = Parser::new(&AUTO, &a);
4879 while parser.next_event().is_some() {}
4880 assert!(
4881 !parser.double_dash_seen(),
4882 "automatic mode must not claim a separator was given"
4883 );
4884 }
4885
4886 #[test]
4887 fn a_wrapper_still_forwards_a_help_flag() {
4888 // Supplying `--help` must not take the two forwarding mechanisms away from a wrapper,
4889 // which is the one place a CLI means to hand the token on rather than answer it.
4890 static ARGS: Arg = Arg {
4891 key: 24,
4892 name: "args",
4893 ..Arg::VAR
4894 };
4895 static WRAP: Command = Command {
4896 name: "wrap",
4897 args: &[&ARGS],
4898 ..Command::EMPTY
4899 };
4900
4901 // A typed separator: everything after it is a value, `--help` included.
4902 let a = argv(["--", "--help", "-h"]);
4903 assert_eq!(
4904 parse(&WRAP, &a).unwrap(),
4905 vec![
4906 Event::Arg {
4907 arg: &ARGS,
4908 value: b"--help",
4909 delimit: true,
4910 },
4911 Event::Arg {
4912 arg: &ARGS,
4913 value: b"-h",
4914 delimit: true,
4915 },
4916 ]
4917 );
4918
4919 // And `automatic`, for the wrapper whose caller should not have to type one: the
4920 // first value stops flag interpretation, so the flags after it forward.
4921 static AUTO_ARGS: Arg = Arg {
4922 key: 25,
4923 name: "args",
4924 double_dash: DoubleDash::Automatic,
4925 ..Arg::VAR
4926 };
4927 static AUTO_WRAP: Command = Command {
4928 name: "wrap",
4929 args: &[&AUTO_ARGS],
4930 ..Command::EMPTY
4931 };
4932
4933 let a = argv(["node", "--help"]);
4934 assert_eq!(
4935 parse(&AUTO_WRAP, &a).unwrap(),
4936 vec![
4937 Event::Arg {
4938 arg: &AUTO_ARGS,
4939 value: b"node",
4940 delimit: true,
4941 },
4942 Event::Arg {
4943 arg: &AUTO_ARGS,
4944 value: b"--help",
4945 delimit: true,
4946 },
4947 ]
4948 );
4949
4950 // Before either takes effect, though, the wrapper's own help is what `--help` asks
4951 // for — `mise run --help` is a question about `run`, not a value for it.
4952 let a = argv(["--help"]);
4953 assert_eq!(
4954 parse(&AUTO_WRAP, &a).unwrap(),
4955 vec![Event::Flag {
4956 flag: &HELP_LONG,
4957 value: None,
4958 negated: false
4959 }]
4960 );
4961 }
4962
4963 #[test]
4964 fn double_dash_required_arg() {
4965 static CMD: Arg = Arg {
4966 key: 20,
4967 name: "cmd",
4968 double_dash: DoubleDash::Required,
4969 ..Arg::REQUIRED
4970 };
4971 static EXEC: Command = Command {
4972 name: "ex",
4973 args: &[&CMD],
4974 ..Command::EMPTY
4975 };
4976
4977 let a = argv(["--", "ls"]);
4978 assert_eq!(
4979 parse(&EXEC, &a).unwrap(),
4980 vec![Event::Arg {
4981 arg: &CMD,
4982 value: b"ls",
4983 delimit: true,
4984 }]
4985 );
4986
4987 let a = argv(["ls"]);
4988 assert_eq!(
4989 parse(&EXEC, &a),
4990 Err(Error::ArgRequiresDoubleDash { arg: &CMD })
4991 );
4992 }
4993
4994 #[test]
4995 fn double_dash_preserve_keeps_the_separator() {
4996 static ARGS: Arg = Arg {
4997 key: 21,
4998 name: "args",
4999 double_dash: DoubleDash::Preserve,
5000 ..Arg::VAR
5001 };
5002 static WRAP: Command = Command {
5003 name: "ex",
5004 args: &[&ARGS],
5005 ..Command::EMPTY
5006 };
5007
5008 let a = argv(["a", "--", "b"]);
5009 let values: Vec<&[u8]> = parse(&WRAP, &a)
5010 .unwrap()
5011 .iter()
5012 .filter_map(|e| match e {
5013 Event::Arg { value, .. } => Some(*value),
5014 _ => None,
5015 })
5016 .collect();
5017 assert_eq!(values, vec![&b"a"[..], &b"--"[..], &b"b"[..]]);
5018 }
5019
5020 #[test]
5021 fn double_dash_automatic_stops_flag_interpretation() {
5022 static FILES: Arg = Arg {
5023 key: 22,
5024 name: "files",
5025 double_dash: DoubleDash::Automatic,
5026 ..Arg::VAR
5027 };
5028 static AUTO: Command = Command {
5029 name: "ex",
5030 flags: &[&FORCE],
5031 args: &[&FILES],
5032 ..Command::EMPTY
5033 };
5034
5035 // The flag before the first value is still a flag; the one after it is a
5036 // value.
5037 let a = argv(["-f", "one", "--force"]);
5038 assert_eq!(
5039 parse(&AUTO, &a).unwrap(),
5040 vec![
5041 Event::Flag {
5042 flag: &FORCE,
5043 value: None,
5044 negated: false
5045 },
5046 Event::Arg {
5047 arg: &FILES,
5048 value: b"one",
5049 delimit: true,
5050 },
5051 Event::Arg {
5052 arg: &FILES,
5053 value: b"--force",
5054 delimit: true,
5055 },
5056 ]
5057 );
5058 }
5059
5060 #[test]
5061 fn too_many_words() {
5062 static ONE: Command = Command {
5063 name: "ex",
5064 args: &[&FILE],
5065 ..Command::EMPTY
5066 };
5067 let a = argv(["a", "b"]);
5068 assert_eq!(parse(&ONE, &a), Err(Error::UnexpectedArg { token: b"b" }));
5069 }
5070
5071 #[test]
5072 fn unknown_letter_rejects_the_whole_bundle() {
5073 // `-f` is real and `-z` is not. The first event must be the error: if the
5074 // flag event came out first, a caller would have applied `-f` from a
5075 // command line that was rejected.
5076 let a = argv(["-fz"]);
5077 let mut parser = Parser::new(&STRICT, &a);
5078 assert_eq!(
5079 parser.next_event(),
5080 Some(Err(Error::UnknownFlag { token: b"-fz" })),
5081 "an unknown letter must reject the token before any of it is applied"
5082 );
5083 assert!(parser.next_event().is_none());
5084
5085 // Leniently, the same token is a value — and `-f` is *not* applied, since
5086 // the token was never a bundle at all.
5087 let a = argv(["-fz"]);
5088 assert_eq!(
5089 parse(&ROOT, &a).unwrap(),
5090 vec![Event::Arg {
5091 arg: &FILE,
5092 value: b"-fz",
5093 delimit: true,
5094 }]
5095 );
5096 }
5097
5098 #[test]
5099 fn unknown_short_error_names_the_whole_token() {
5100 for (tokens, want) in [(["-z"], &b"-z"[..]), (["-fz"], &b"-fz"[..])] {
5101 let a = argv(tokens);
5102 assert_eq!(
5103 parse(&STRICT, &a),
5104 Err(Error::UnknownFlag { token: want }),
5105 "{tokens:?}"
5106 );
5107 }
5108 }
5109
5110 #[test]
5111 fn errors_are_terminal() {
5112 let a = argv(["--wat", "--force"]);
5113 let mut parser = Parser::new(&STRICT, &a);
5114 assert!(parser.next_event().unwrap().is_err());
5115 assert!(parser.next_event().is_none());
5116 }
5117
5118 #[test]
5119 fn non_utf8_values_still_parse() {
5120 // A value that is not valid UTF-8 binds; only converting it fails, and
5121 // only if a caller asks.
5122 let raw = OsStr::new("--force");
5123 let a = [raw];
5124 assert!(parse(&ROOT, &a).is_ok());
5125
5126 assert!(as_str(b"ok").is_ok());
5127 assert!(as_str(&[0xff, 0xfe]).is_err());
5128 }
5129
5130 #[test]
5131 fn a_multicall_applet_is_the_basename_unless_it_is_the_dispatcher() {
5132 assert_eq!(multicall_basename("/usr/bin/ls"), "ls");
5133 assert_eq!(multicall_basename(r"C:\busybox\ls.exe"), "ls");
5134 assert_eq!(
5135 multicall_applet("/usr/bin/ls", "busybox", Some("busybox")),
5136 Some("ls")
5137 );
5138 assert_eq!(
5139 multicall_applet("/usr/bin/busybox", "busybox", Some("busybox")),
5140 None
5141 );
5142 assert_eq!(
5143 multicall_applet("ls.exe", "busybox", Some("busybox")),
5144 Some("ls")
5145 );
5146 assert_eq!(
5147 multicall_applet("/usr/bin/busybox", "BusyBox", Some("/opt/bin/busybox")),
5148 None
5149 );
5150 assert_eq!(
5151 multicall_applet("busybox.exe", "BusyBox", Some("busybox.exe")),
5152 None
5153 );
5154 }
5155
5156 #[test]
5157 fn a_spec_request_is_the_first_word_and_nothing_else() {
5158 let request = [OsStr::new(SPEC_REQUEST)];
5159 assert!(is_spec_request(&ROOT, &request));
5160
5161 // Anywhere but the front it is an ordinary value, which is what makes the endpoint
5162 // safe for a CLI whose arguments are arbitrary text.
5163 let later = ["install", SPEC_REQUEST].map(OsStr::new);
5164 assert!(!is_spec_request(&ROOT, &later));
5165 assert!(!is_spec_request(&ROOT, &[]));
5166 assert!(!is_spec_request(&ROOT, &[OsStr::new("--help")]));
5167 }
5168
5169 #[test]
5170 fn a_declared_command_of_that_name_keeps_it() {
5171 static DECLARED: Command = Command {
5172 name: SPEC_REQUEST,
5173 key: 200,
5174 ..Command::EMPTY
5175 };
5176 static ALIASED: Command = Command {
5177 name: "describe",
5178 aliases: &[SPEC_REQUEST],
5179 key: 201,
5180 ..Command::EMPTY
5181 };
5182 static DECLARES_IT: Command = Command {
5183 name: "ex",
5184 subcommands: &[&DECLARED],
5185 ..Command::EMPTY
5186 };
5187 static ALIASES_IT: Command = Command {
5188 name: "ex",
5189 subcommands: &[&ALIASED],
5190 ..Command::EMPTY
5191 };
5192
5193 let request = [OsStr::new(SPEC_REQUEST)];
5194 assert!(!is_spec_request(&DECLARES_IT, &request));
5195 // An alias selects a command just as its name does, so it wins here too.
5196 assert!(!is_spec_request(&ALIASES_IT, &request));
5197 }
5198}