1 | use std::ffi::OsStr; |
2 | |
3 | /// String-like methods for [`OsStr`] |
4 | pub trait OsStrExt: private::Sealed { |
5 | /// Converts to a string slice. |
6 | /// |
7 | /// The `Utf8Error` is guaranteed to have a valid UTF8 boundary |
8 | /// in its `valid_up_to()` |
9 | fn try_str(&self) -> Result<&str, std::str::Utf8Error>; |
10 | /// Returns `true` if the given pattern matches a sub-slice of |
11 | /// this string slice. |
12 | /// |
13 | /// Returns `false` if it does not. |
14 | /// |
15 | /// # Examples |
16 | /// |
17 | /// ```rust |
18 | /// use clap_lex::OsStrExt as _; |
19 | /// let bananas = std::ffi::OsStr::new("bananas" ); |
20 | /// |
21 | /// assert!(bananas.contains("nana" )); |
22 | /// assert!(!bananas.contains("apples" )); |
23 | /// ``` |
24 | fn contains(&self, needle: &str) -> bool; |
25 | /// Returns the byte index of the first character of this string slice that |
26 | /// matches the pattern. |
27 | /// |
28 | /// Returns [`None`] if the pattern doesn't match. |
29 | /// |
30 | /// # Examples |
31 | /// |
32 | /// ```rust |
33 | /// use clap_lex::OsStrExt as _; |
34 | /// let s = std::ffi::OsStr::new("Löwe 老虎 Léopard Gepardi" ); |
35 | /// |
36 | /// assert_eq!(s.find("L" ), Some(0)); |
37 | /// assert_eq!(s.find("é" ), Some(14)); |
38 | /// assert_eq!(s.find("par" ), Some(17)); |
39 | /// ``` |
40 | /// |
41 | /// Not finding the pattern: |
42 | /// |
43 | /// ```rust |
44 | /// use clap_lex::OsStrExt as _; |
45 | /// let s = std::ffi::OsStr::new("Löwe 老虎 Léopard" ); |
46 | /// |
47 | /// assert_eq!(s.find("1" ), None); |
48 | /// ``` |
49 | fn find(&self, needle: &str) -> Option<usize>; |
50 | /// Returns a string slice with the prefix removed. |
51 | /// |
52 | /// If the string starts with the pattern `prefix`, returns substring after the prefix, wrapped |
53 | /// in `Some`. |
54 | /// |
55 | /// If the string does not start with `prefix`, returns `None`. |
56 | /// |
57 | /// # Examples |
58 | /// |
59 | /// ``` |
60 | /// use std::ffi::OsStr; |
61 | /// use clap_lex::OsStrExt as _; |
62 | /// assert_eq!(OsStr::new("foo:bar" ).strip_prefix("foo:" ), Some(OsStr::new("bar" ))); |
63 | /// assert_eq!(OsStr::new("foo:bar" ).strip_prefix("bar" ), None); |
64 | /// assert_eq!(OsStr::new("foofoo" ).strip_prefix("foo" ), Some(OsStr::new("foo" ))); |
65 | /// ``` |
66 | fn strip_prefix(&self, prefix: &str) -> Option<&OsStr>; |
67 | /// Returns `true` if the given pattern matches a prefix of this |
68 | /// string slice. |
69 | /// |
70 | /// Returns `false` if it does not. |
71 | /// |
72 | /// # Examples |
73 | /// |
74 | /// ``` |
75 | /// use clap_lex::OsStrExt as _; |
76 | /// let bananas = std::ffi::OsStr::new("bananas" ); |
77 | /// |
78 | /// assert!(bananas.starts_with("bana" )); |
79 | /// assert!(!bananas.starts_with("nana" )); |
80 | /// ``` |
81 | fn starts_with(&self, prefix: &str) -> bool; |
82 | /// An iterator over substrings of this string slice, separated by |
83 | /// characters matched by a pattern. |
84 | /// |
85 | /// # Examples |
86 | /// |
87 | /// Simple patterns: |
88 | /// |
89 | /// ``` |
90 | /// use std::ffi::OsStr; |
91 | /// use clap_lex::OsStrExt as _; |
92 | /// let v: Vec<_> = OsStr::new("Mary had a little lamb" ).split(" " ).collect(); |
93 | /// assert_eq!(v, [OsStr::new("Mary" ), OsStr::new("had" ), OsStr::new("a" ), OsStr::new("little" ), OsStr::new("lamb" )]); |
94 | /// |
95 | /// let v: Vec<_> = OsStr::new("" ).split("X" ).collect(); |
96 | /// assert_eq!(v, [OsStr::new("" )]); |
97 | /// |
98 | /// let v: Vec<_> = OsStr::new("lionXXtigerXleopard" ).split("X" ).collect(); |
99 | /// assert_eq!(v, [OsStr::new("lion" ), OsStr::new("" ), OsStr::new("tiger" ), OsStr::new("leopard" )]); |
100 | /// |
101 | /// let v: Vec<_> = OsStr::new("lion::tiger::leopard" ).split("::" ).collect(); |
102 | /// assert_eq!(v, [OsStr::new("lion" ), OsStr::new("tiger" ), OsStr::new("leopard" )]); |
103 | /// ``` |
104 | /// |
105 | /// If a string contains multiple contiguous separators, you will end up |
106 | /// with empty strings in the output: |
107 | /// |
108 | /// ``` |
109 | /// use std::ffi::OsStr; |
110 | /// use clap_lex::OsStrExt as _; |
111 | /// let x = OsStr::new("||||a||b|c" ); |
112 | /// let d: Vec<_> = x.split("|" ).collect(); |
113 | /// |
114 | /// assert_eq!(d, &[OsStr::new("" ), OsStr::new("" ), OsStr::new("" ), OsStr::new("" ), OsStr::new("a" ), OsStr::new("" ), OsStr::new("b" ), OsStr::new("c" )]); |
115 | /// ``` |
116 | /// |
117 | /// Contiguous separators are separated by the empty string. |
118 | /// |
119 | /// ``` |
120 | /// use std::ffi::OsStr; |
121 | /// use clap_lex::OsStrExt as _; |
122 | /// let x = OsStr::new("(///)" ); |
123 | /// let d: Vec<_> = x.split("/" ).collect(); |
124 | /// |
125 | /// assert_eq!(d, &[OsStr::new("(" ), OsStr::new("" ), OsStr::new("" ), OsStr::new(")" )]); |
126 | /// ``` |
127 | /// |
128 | /// Separators at the start or end of a string are neighbored |
129 | /// by empty strings. |
130 | /// |
131 | /// ``` |
132 | /// use std::ffi::OsStr; |
133 | /// use clap_lex::OsStrExt as _; |
134 | /// let d: Vec<_> = OsStr::new("010" ).split("0" ).collect(); |
135 | /// assert_eq!(d, &[OsStr::new("" ), OsStr::new("1" ), OsStr::new("" )]); |
136 | /// ``` |
137 | /// |
138 | /// When the empty string is used as a separator, it panics |
139 | /// |
140 | /// ```should_panic |
141 | /// use std::ffi::OsStr; |
142 | /// use clap_lex::OsStrExt as _; |
143 | /// let f: Vec<_> = OsStr::new("rust" ).split("" ).collect(); |
144 | /// assert_eq!(f, &[OsStr::new("" ), OsStr::new("r" ), OsStr::new("u" ), OsStr::new("s" ), OsStr::new("t" ), OsStr::new("" )]); |
145 | /// ``` |
146 | /// |
147 | /// Contiguous separators can lead to possibly surprising behavior |
148 | /// when whitespace is used as the separator. This code is correct: |
149 | /// |
150 | /// ``` |
151 | /// use std::ffi::OsStr; |
152 | /// use clap_lex::OsStrExt as _; |
153 | /// let x = OsStr::new(" a b c" ); |
154 | /// let d: Vec<_> = x.split(" " ).collect(); |
155 | /// |
156 | /// assert_eq!(d, &[OsStr::new("" ), OsStr::new("" ), OsStr::new("" ), OsStr::new("" ), OsStr::new("a" ), OsStr::new("" ), OsStr::new("b" ), OsStr::new("c" )]); |
157 | /// ``` |
158 | /// |
159 | /// It does _not_ give you: |
160 | /// |
161 | /// ```,ignore |
162 | /// assert_eq!(d, &[OsStr::new("a" ), OsStr::new("b" ), OsStr::new("c" )]); |
163 | /// ``` |
164 | /// |
165 | /// Use [`split_whitespace`] for this behavior. |
166 | /// |
167 | /// [`split_whitespace`]: str::split_whitespace |
168 | fn split<'s, 'n>(&'s self, needle: &'n str) -> Split<'s, 'n>; |
169 | /// Splits the string on the first occurrence of the specified delimiter and |
170 | /// returns prefix before delimiter and suffix after delimiter. |
171 | /// |
172 | /// # Examples |
173 | /// |
174 | /// ``` |
175 | /// use std::ffi::OsStr; |
176 | /// use clap_lex::OsStrExt as _; |
177 | /// assert_eq!(OsStr::new("cfg" ).split_once("=" ), None); |
178 | /// assert_eq!(OsStr::new("cfg=" ).split_once("=" ), Some((OsStr::new("cfg" ), OsStr::new("" )))); |
179 | /// assert_eq!(OsStr::new("cfg=foo" ).split_once("=" ), Some((OsStr::new("cfg" ), OsStr::new("foo" )))); |
180 | /// assert_eq!(OsStr::new("cfg=foo=bar" ).split_once("=" ), Some((OsStr::new("cfg" ), OsStr::new("foo=bar" )))); |
181 | /// ``` |
182 | fn split_once(&self, needle: &'_ str) -> Option<(&OsStr, &OsStr)>; |
183 | } |
184 | |
185 | impl OsStrExt for OsStr { |
186 | fn try_str(&self) -> Result<&str, std::str::Utf8Error> { |
187 | let bytes = self.as_encoded_bytes(); |
188 | std::str::from_utf8(bytes) |
189 | } |
190 | |
191 | fn contains(&self, needle: &str) -> bool { |
192 | self.find(needle).is_some() |
193 | } |
194 | |
195 | fn find(&self, needle: &str) -> Option<usize> { |
196 | let bytes = self.as_encoded_bytes(); |
197 | (0..=self.len().checked_sub(needle.len())?) |
198 | .find(|&x| bytes[x..].starts_with(needle.as_bytes())) |
199 | } |
200 | |
201 | fn strip_prefix(&self, prefix: &str) -> Option<&OsStr> { |
202 | let bytes = self.as_encoded_bytes(); |
203 | bytes.strip_prefix(prefix.as_bytes()).map(|s| { |
204 | // SAFETY: |
205 | // - This came from `as_encoded_bytes` |
206 | // - Since `prefix` is `&str`, any split will be along UTF-8 boundary |
207 | unsafe { OsStr::from_encoded_bytes_unchecked(s) } |
208 | }) |
209 | } |
210 | fn starts_with(&self, prefix: &str) -> bool { |
211 | let bytes = self.as_encoded_bytes(); |
212 | bytes.starts_with(prefix.as_bytes()) |
213 | } |
214 | |
215 | fn split<'s, 'n>(&'s self, needle: &'n str) -> Split<'s, 'n> { |
216 | assert_ne!(needle, "" ); |
217 | Split { |
218 | haystack: Some(self), |
219 | needle, |
220 | } |
221 | } |
222 | |
223 | fn split_once(&self, needle: &'_ str) -> Option<(&OsStr, &OsStr)> { |
224 | let start = self.find(needle)?; |
225 | let end = start + needle.len(); |
226 | let haystack = self.as_encoded_bytes(); |
227 | let first = &haystack[0..start]; |
228 | let second = &haystack[end..]; |
229 | // SAFETY: |
230 | // - This came from `as_encoded_bytes` |
231 | // - Since `needle` is `&str`, any split will be along UTF-8 boundary |
232 | unsafe { |
233 | Some(( |
234 | OsStr::from_encoded_bytes_unchecked(first), |
235 | OsStr::from_encoded_bytes_unchecked(second), |
236 | )) |
237 | } |
238 | } |
239 | } |
240 | |
241 | mod private { |
242 | pub trait Sealed {} |
243 | |
244 | impl Sealed for std::ffi::OsStr {} |
245 | } |
246 | |
247 | pub struct Split<'s, 'n> { |
248 | haystack: Option<&'s OsStr>, |
249 | needle: &'n str, |
250 | } |
251 | |
252 | impl<'s> Iterator for Split<'s, '_> { |
253 | type Item = &'s OsStr; |
254 | |
255 | fn next(&mut self) -> Option<Self::Item> { |
256 | let haystack: &'s OsStr = self.haystack?; |
257 | if let Some((first: &OsStr, second: &OsStr)) = haystack.split_once(self.needle) { |
258 | if !haystack.is_empty() { |
259 | debug_assert_ne!(haystack, second); |
260 | } |
261 | self.haystack = Some(second); |
262 | Some(first) |
263 | } else { |
264 | self.haystack = None; |
265 | Some(haystack) |
266 | } |
267 | } |
268 | } |
269 | |
270 | /// Split an `OsStr` |
271 | /// |
272 | /// # Safety |
273 | /// |
274 | /// `index` must be at a valid UTF-8 boundary |
275 | pub(crate) unsafe fn split_at(os: &OsStr, index: usize) -> (&OsStr, &OsStr) { |
276 | unsafe { |
277 | let bytes: &[u8] = os.as_encoded_bytes(); |
278 | let (first: &[u8], second: &[u8]) = bytes.split_at(mid:index); |
279 | ( |
280 | OsStr::from_encoded_bytes_unchecked(bytes:first), |
281 | OsStr::from_encoded_bytes_unchecked(bytes:second), |
282 | ) |
283 | } |
284 | } |
285 | |