1 //! Minimal, flexible command-line parser
2 //!
3 //! As opposed to a declarative parser, this processes arguments as a stream of tokens. As lexing
4 //! a command-line is not context-free, we rely on the caller to decide how to interpret the
5 //! arguments.
6 //!
7 //! # Examples
8 //!
9 //! ```rust
10 //! use std::path::PathBuf;
11 //!
12 //! type BoxedError = Box<dyn std::error::Error + Send + Sync>;
13 //!
14 //! #[derive(Debug)]
15 //! struct Args {
16 //! paths: Vec<PathBuf>,
17 //! color: Color,
18 //! verbosity: usize,
19 //! }
20 //!
21 //! #[derive(Debug)]
22 //! enum Color {
23 //! Always,
24 //! Auto,
25 //! Never,
26 //! }
27 //!
28 //! impl Color {
29 //! fn parse(s: Option<&clap_lex::RawOsStr>) -> Result<Self, BoxedError> {
30 //! let s = s.map(|s| s.to_str().ok_or(s));
31 //! match s {
32 //! Some(Ok("always")) | Some(Ok("")) | None => {
33 //! Ok(Color::Always)
34 //! }
35 //! Some(Ok("auto")) => {
36 //! Ok(Color::Auto)
37 //! }
38 //! Some(Ok("never")) => {
39 //! Ok(Color::Never)
40 //! }
41 //! Some(invalid) => {
42 //! Err(format!("Invalid value for `--color`, {:?}", invalid).into())
43 //! }
44 //! }
45 //! }
46 //! }
47 //!
48 //! fn parse_args(
49 //! raw: impl IntoIterator<Item=impl Into<std::ffi::OsString>>
50 //! ) -> Result<Args, BoxedError> {
51 //! let mut args = Args {
52 //! paths: Vec::new(),
53 //! color: Color::Auto,
54 //! verbosity: 0,
55 //! };
56 //!
57 //! let raw = clap_lex::RawArgs::new(raw);
58 //! let mut cursor = raw.cursor();
59 //! raw.next(&mut cursor); // Skip the bin
60 //! while let Some(arg) = raw.next(&mut cursor) {
61 //! if arg.is_escape() {
62 //! args.paths.extend(raw.remaining(&mut cursor).map(PathBuf::from));
63 //! } else if arg.is_stdio() {
64 //! args.paths.push(PathBuf::from("-"));
65 //! } else if let Some((long, value)) = arg.to_long() {
66 //! match long {
67 //! Ok("verbose") => {
68 //! if let Some(value) = value {
69 //! return Err(format!("`--verbose` does not take a value, got `{:?}`", value).into());
70 //! }
71 //! args.verbosity += 1;
72 //! }
73 //! Ok("color") => {
74 //! args.color = Color::parse(value)?;
75 //! }
76 //! _ => {
77 //! return Err(
78 //! format!("Unexpected flag: --{}", arg.display()).into()
79 //! );
80 //! }
81 //! }
82 //! } else if let Some(mut shorts) = arg.to_short() {
83 //! while let Some(short) = shorts.next_flag() {
84 //! match short {
85 //! Ok('v') => {
86 //! args.verbosity += 1;
87 //! }
88 //! Ok('c') => {
89 //! let value = shorts.next_value_os();
90 //! args.color = Color::parse(value)?;
91 //! }
92 //! Ok(c) => {
93 //! return Err(format!("Unexpected flag: -{}", c).into());
94 //! }
95 //! Err(e) => {
96 //! return Err(format!("Unexpected flag: -{}", e.to_str_lossy()).into());
97 //! }
98 //! }
99 //! }
100 //! } else {
101 //! args.paths.push(PathBuf::from(arg.to_value_os().to_os_str().into_owned()));
102 //! }
103 //! }
104 //!
105 //! Ok(args)
106 //! }
107 //!
108 //! let args = parse_args(["bin", "--hello", "world"]);
109 //! println!("{:?}", args);
110 //! ```
111
112 use std::ffi::OsStr;
113 use std::ffi::OsString;
114
115 pub use std::io::SeekFrom;
116
117 pub use os_str_bytes::RawOsStr;
118 pub use os_str_bytes::RawOsString;
119
120 /// Command-line arguments
121 #[derive(Default, Clone, Debug, PartialEq, Eq)]
122 pub struct RawArgs {
123 items: Vec<OsString>,
124 }
125
126 impl RawArgs {
127 //// Create an argument list to parse
128 ///
129 /// **NOTE:** The argument returned will be the current binary.
130 ///
131 /// # Example
132 ///
133 /// ```rust,no_run
134 /// # use std::path::PathBuf;
135 /// let raw = clap_lex::RawArgs::from_args();
136 /// let mut cursor = raw.cursor();
137 /// let _bin = raw.next_os(&mut cursor);
138 ///
139 /// let mut paths = raw.remaining(&mut cursor).map(PathBuf::from).collect::<Vec<_>>();
140 /// println!("{:?}", paths);
141 /// ```
from_args() -> Self142 pub fn from_args() -> Self {
143 Self::new(std::env::args_os())
144 }
145
146 //// Create an argument list to parse
147 ///
148 /// # Example
149 ///
150 /// ```rust,no_run
151 /// # use std::path::PathBuf;
152 /// let raw = clap_lex::RawArgs::new(["bin", "foo.txt"]);
153 /// let mut cursor = raw.cursor();
154 /// let _bin = raw.next_os(&mut cursor);
155 ///
156 /// let mut paths = raw.remaining(&mut cursor).map(PathBuf::from).collect::<Vec<_>>();
157 /// println!("{:?}", paths);
158 /// ```
new(iter: impl IntoIterator<Item = impl Into<std::ffi::OsString>>) -> Self159 pub fn new(iter: impl IntoIterator<Item = impl Into<std::ffi::OsString>>) -> Self {
160 let iter = iter.into_iter();
161 Self::from(iter)
162 }
163
164 /// Create a cursor for walking the arguments
165 ///
166 /// # Example
167 ///
168 /// ```rust,no_run
169 /// # use std::path::PathBuf;
170 /// let raw = clap_lex::RawArgs::new(["bin", "foo.txt"]);
171 /// let mut cursor = raw.cursor();
172 /// let _bin = raw.next_os(&mut cursor);
173 ///
174 /// let mut paths = raw.remaining(&mut cursor).map(PathBuf::from).collect::<Vec<_>>();
175 /// println!("{:?}", paths);
176 /// ```
cursor(&self) -> ArgCursor177 pub fn cursor(&self) -> ArgCursor {
178 ArgCursor::new()
179 }
180
181 /// Advance the cursor, returning the next [`ParsedArg`]
next(&self, cursor: &mut ArgCursor) -> Option<ParsedArg<'_>>182 pub fn next(&self, cursor: &mut ArgCursor) -> Option<ParsedArg<'_>> {
183 self.next_os(cursor).map(ParsedArg::new)
184 }
185
186 /// Advance the cursor, returning a raw argument value.
next_os(&self, cursor: &mut ArgCursor) -> Option<&OsStr>187 pub fn next_os(&self, cursor: &mut ArgCursor) -> Option<&OsStr> {
188 let next = self.items.get(cursor.cursor).map(|s| s.as_os_str());
189 cursor.cursor = cursor.cursor.saturating_add(1);
190 next
191 }
192
193 /// Return the next [`ParsedArg`]
peek(&self, cursor: &ArgCursor) -> Option<ParsedArg<'_>>194 pub fn peek(&self, cursor: &ArgCursor) -> Option<ParsedArg<'_>> {
195 self.peek_os(cursor).map(ParsedArg::new)
196 }
197
198 /// Return a raw argument value.
peek_os(&self, cursor: &ArgCursor) -> Option<&OsStr>199 pub fn peek_os(&self, cursor: &ArgCursor) -> Option<&OsStr> {
200 self.items.get(cursor.cursor).map(|s| s.as_os_str())
201 }
202
203 /// Return all remaining raw arguments, advancing the cursor to the end
204 ///
205 /// # Example
206 ///
207 /// ```rust,no_run
208 /// # use std::path::PathBuf;
209 /// let raw = clap_lex::RawArgs::new(["bin", "foo.txt"]);
210 /// let mut cursor = raw.cursor();
211 /// let _bin = raw.next_os(&mut cursor);
212 ///
213 /// let mut paths = raw.remaining(&mut cursor).map(PathBuf::from).collect::<Vec<_>>();
214 /// println!("{:?}", paths);
215 /// ```
remaining(&self, cursor: &mut ArgCursor) -> impl Iterator<Item = &OsStr>216 pub fn remaining(&self, cursor: &mut ArgCursor) -> impl Iterator<Item = &OsStr> {
217 let remaining = self.items[cursor.cursor..].iter().map(|s| s.as_os_str());
218 cursor.cursor = self.items.len();
219 remaining
220 }
221
222 /// Adjust the cursor's position
seek(&self, cursor: &mut ArgCursor, pos: SeekFrom)223 pub fn seek(&self, cursor: &mut ArgCursor, pos: SeekFrom) {
224 let pos = match pos {
225 SeekFrom::Start(pos) => pos,
226 SeekFrom::End(pos) => (self.items.len() as i64).saturating_add(pos).max(0) as u64,
227 SeekFrom::Current(pos) => (cursor.cursor as i64).saturating_add(pos).max(0) as u64,
228 };
229 let pos = (pos as usize).min(self.items.len());
230 cursor.cursor = pos;
231 }
232
233 /// Inject arguments before the [`RawArgs::next`]
insert( &mut self, cursor: &ArgCursor, insert_items: impl IntoIterator<Item = impl Into<OsString>>, )234 pub fn insert(
235 &mut self,
236 cursor: &ArgCursor,
237 insert_items: impl IntoIterator<Item = impl Into<OsString>>,
238 ) {
239 self.items.splice(
240 cursor.cursor..cursor.cursor,
241 insert_items.into_iter().map(Into::into),
242 );
243 }
244
245 /// Any remaining args?
is_end(&self, cursor: &ArgCursor) -> bool246 pub fn is_end(&self, cursor: &ArgCursor) -> bool {
247 self.peek_os(cursor).is_none()
248 }
249 }
250
251 impl<I, T> From<I> for RawArgs
252 where
253 I: Iterator<Item = T>,
254 T: Into<OsString>,
255 {
from(val: I) -> Self256 fn from(val: I) -> Self {
257 Self {
258 items: val.map(|x| x.into()).collect(),
259 }
260 }
261 }
262
263 /// Position within [`RawArgs`]
264 #[derive(Clone, Debug, PartialEq, Eq, PartialOrd, Ord)]
265 pub struct ArgCursor {
266 cursor: usize,
267 }
268
269 impl ArgCursor {
new() -> Self270 fn new() -> Self {
271 Self { cursor: 0 }
272 }
273 }
274
275 /// Command-line Argument
276 #[derive(Clone, Debug, PartialEq, Eq, PartialOrd, Ord, Hash)]
277 pub struct ParsedArg<'s> {
278 inner: std::borrow::Cow<'s, RawOsStr>,
279 utf8: Option<&'s str>,
280 }
281
282 impl<'s> ParsedArg<'s> {
new(inner: &'s OsStr) -> Self283 fn new(inner: &'s OsStr) -> Self {
284 let utf8 = inner.to_str();
285 let inner = RawOsStr::new(inner);
286 Self { inner, utf8 }
287 }
288
289 /// Argument is length of 0
is_empty(&self) -> bool290 pub fn is_empty(&self) -> bool {
291 self.inner.as_ref().is_empty()
292 }
293
294 /// Does the argument look like a stdio argument (`-`)
is_stdio(&self) -> bool295 pub fn is_stdio(&self) -> bool {
296 self.inner.as_ref() == "-"
297 }
298
299 /// Does the argument look like an argument escape (`--`)
is_escape(&self) -> bool300 pub fn is_escape(&self) -> bool {
301 self.inner.as_ref() == "--"
302 }
303
304 /// Does the argument look like a number
is_number(&self) -> bool305 pub fn is_number(&self) -> bool {
306 self.to_value()
307 .map(|s| s.parse::<f64>().is_ok())
308 .unwrap_or_default()
309 }
310
311 /// Treat as a long-flag
to_long(&self) -> Option<(Result<&str, &RawOsStr>, Option<&RawOsStr>)>312 pub fn to_long(&self) -> Option<(Result<&str, &RawOsStr>, Option<&RawOsStr>)> {
313 if let Some(raw) = self.utf8 {
314 let remainder = raw.strip_prefix("--")?;
315 if remainder.is_empty() {
316 debug_assert!(self.is_escape());
317 return None;
318 }
319
320 let (flag, value) = if let Some((p0, p1)) = remainder.split_once('=') {
321 (p0, Some(p1))
322 } else {
323 (remainder, None)
324 };
325 let flag = Ok(flag);
326 let value = value.map(RawOsStr::from_str);
327 Some((flag, value))
328 } else {
329 let raw = self.inner.as_ref();
330 let remainder = raw.strip_prefix("--")?;
331 if remainder.is_empty() {
332 debug_assert!(self.is_escape());
333 return None;
334 }
335
336 let (flag, value) = if let Some((p0, p1)) = remainder.split_once('=') {
337 (p0, Some(p1))
338 } else {
339 (remainder, None)
340 };
341 let flag = flag.to_str().ok_or(flag);
342 Some((flag, value))
343 }
344 }
345
346 /// Can treat as a long-flag
is_long(&self) -> bool347 pub fn is_long(&self) -> bool {
348 self.inner.as_ref().starts_with("--") && !self.is_escape()
349 }
350
351 /// Treat as a short-flag
to_short(&self) -> Option<ShortFlags<'_>>352 pub fn to_short(&self) -> Option<ShortFlags<'_>> {
353 if let Some(remainder_os) = self.inner.as_ref().strip_prefix('-') {
354 if remainder_os.starts_with('-') {
355 None
356 } else if remainder_os.is_empty() {
357 debug_assert!(self.is_stdio());
358 None
359 } else {
360 let remainder = self.utf8.map(|s| &s[1..]);
361 Some(ShortFlags::new(remainder_os, remainder))
362 }
363 } else {
364 None
365 }
366 }
367
368 /// Can treat as a short-flag
is_short(&self) -> bool369 pub fn is_short(&self) -> bool {
370 self.inner.as_ref().starts_with('-')
371 && !self.is_stdio()
372 && !self.inner.as_ref().starts_with("--")
373 }
374
375 /// Treat as a value
376 ///
377 /// **NOTE:** May return a flag or an escape.
to_value_os(&self) -> &RawOsStr378 pub fn to_value_os(&self) -> &RawOsStr {
379 self.inner.as_ref()
380 }
381
382 /// Treat as a value
383 ///
384 /// **NOTE:** May return a flag or an escape.
to_value(&self) -> Result<&str, &RawOsStr>385 pub fn to_value(&self) -> Result<&str, &RawOsStr> {
386 self.utf8.ok_or_else(|| self.inner.as_ref())
387 }
388
389 /// Safely print an argument that may contain non-UTF8 content
390 ///
391 /// This may perform lossy conversion, depending on the platform. If you would like an implementation which escapes the path please use Debug instead.
display(&self) -> impl std::fmt::Display + '_392 pub fn display(&self) -> impl std::fmt::Display + '_ {
393 self.inner.to_str_lossy()
394 }
395 }
396
397 /// Walk through short flags within a [`ParsedArg`]
398 #[derive(Clone, Debug)]
399 pub struct ShortFlags<'s> {
400 inner: &'s RawOsStr,
401 utf8_prefix: std::str::CharIndices<'s>,
402 invalid_suffix: Option<&'s RawOsStr>,
403 }
404
405 impl<'s> ShortFlags<'s> {
new(inner: &'s RawOsStr, utf8: Option<&'s str>) -> Self406 fn new(inner: &'s RawOsStr, utf8: Option<&'s str>) -> Self {
407 let (utf8_prefix, invalid_suffix) = if let Some(utf8) = utf8 {
408 (utf8, None)
409 } else {
410 split_nonutf8_once(inner)
411 };
412 let utf8_prefix = utf8_prefix.char_indices();
413 Self {
414 inner,
415 utf8_prefix,
416 invalid_suffix,
417 }
418 }
419
420 /// Move the iterator forward by `n` short flags
advance_by(&mut self, n: usize) -> Result<(), usize>421 pub fn advance_by(&mut self, n: usize) -> Result<(), usize> {
422 for i in 0..n {
423 self.next().ok_or(i)?.map_err(|_| i)?;
424 }
425 Ok(())
426 }
427
428 /// No short flags left
is_empty(&self) -> bool429 pub fn is_empty(&self) -> bool {
430 self.invalid_suffix.is_none() && self.utf8_prefix.as_str().is_empty()
431 }
432
433 /// Does the short flag look like a number
434 ///
435 /// Ideally call this before doing any iterator
is_number(&self) -> bool436 pub fn is_number(&self) -> bool {
437 self.invalid_suffix.is_none() && self.utf8_prefix.as_str().parse::<f64>().is_ok()
438 }
439
440 /// Advance the iterator, returning the next short flag on success
441 ///
442 /// On error, returns the invalid-UTF8 value
next_flag(&mut self) -> Option<Result<char, &'s RawOsStr>>443 pub fn next_flag(&mut self) -> Option<Result<char, &'s RawOsStr>> {
444 if let Some((_, flag)) = self.utf8_prefix.next() {
445 return Some(Ok(flag));
446 }
447
448 if let Some(suffix) = self.invalid_suffix {
449 self.invalid_suffix = None;
450 return Some(Err(suffix));
451 }
452
453 None
454 }
455
456 /// Advance the iterator, returning everything left as a value
next_value_os(&mut self) -> Option<&'s RawOsStr>457 pub fn next_value_os(&mut self) -> Option<&'s RawOsStr> {
458 if let Some((index, _)) = self.utf8_prefix.next() {
459 self.utf8_prefix = "".char_indices();
460 self.invalid_suffix = None;
461 return Some(&self.inner[index..]);
462 }
463
464 if let Some(suffix) = self.invalid_suffix {
465 self.invalid_suffix = None;
466 return Some(suffix);
467 }
468
469 None
470 }
471 }
472
473 impl<'s> Iterator for ShortFlags<'s> {
474 type Item = Result<char, &'s RawOsStr>;
475
next(&mut self) -> Option<Self::Item>476 fn next(&mut self) -> Option<Self::Item> {
477 self.next_flag()
478 }
479 }
480
split_nonutf8_once(b: &RawOsStr) -> (&str, Option<&RawOsStr>)481 fn split_nonutf8_once(b: &RawOsStr) -> (&str, Option<&RawOsStr>) {
482 match std::str::from_utf8(b.as_raw_bytes()) {
483 Ok(s) => (s, None),
484 Err(err) => {
485 let (valid, after_valid) = b.split_at(err.valid_up_to());
486 let valid = std::str::from_utf8(valid.as_raw_bytes()).unwrap();
487 (valid, Some(after_valid))
488 }
489 }
490 }
491