gix_path/
convert.rs

1use std::{
2    borrow::Cow,
3    ffi::{OsStr, OsString},
4    path::{Component, Path, PathBuf},
5};
6
7use bstr::{BStr, BString};
8
9#[derive(Debug)]
10/// The error type returned by [`into_bstr()`] and others may suffer from failed conversions from or to bytes.
11pub struct Utf8Error;
12
13impl std::fmt::Display for Utf8Error {
14    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
15        f.write_str("Could not convert to UTF8 or from UTF8 due to ill-formed input")
16    }
17}
18
19impl std::error::Error for Utf8Error {}
20
21/// Like [`into_bstr()`], but takes `OsStr` as input for a lossless, but fallible, conversion.
22pub fn os_str_into_bstr(path: &OsStr) -> Result<&BStr, Utf8Error> {
23    let path = try_into_bstr(Cow::Borrowed(path.as_ref()))?;
24    match path {
25        Cow::Borrowed(path) => Ok(path),
26        Cow::Owned(_) => unreachable!("borrowed cows stay borrowed"),
27    }
28}
29
30/// Like [`into_bstr()`], but takes `OsString` as input for a lossless, but fallible, conversion.
31pub fn os_string_into_bstring(path: OsString) -> Result<BString, Utf8Error> {
32    let path = try_into_bstr(Cow::Owned(path.into()))?;
33    match path {
34        Cow::Borrowed(_path) => unreachable!("borrowed cows stay borrowed"),
35        Cow::Owned(path) => Ok(path),
36    }
37}
38
39/// Like [`into_bstr()`], but takes `Cow<OsStr>` as input for a lossless, but fallible, conversion.
40pub fn try_os_str_into_bstr(path: Cow<'_, OsStr>) -> Result<Cow<'_, BStr>, Utf8Error> {
41    match path {
42        Cow::Borrowed(path) => os_str_into_bstr(path).map(Cow::Borrowed),
43        Cow::Owned(path) => os_string_into_bstring(path).map(Cow::Owned),
44    }
45}
46
47/// Convert the given path either into its raw bytes on Unix or its UTF-8 encoded counterpart on non-Unix platforms.
48///
49/// On non-Unix platforms, if the source `Path`` contains ill-formed, lone surrogates, the UTF-8 conversion will fail
50/// causing `Utf8Error` to be returned.
51pub fn try_into_bstr<'a>(path: impl Into<Cow<'a, Path>>) -> Result<Cow<'a, BStr>, Utf8Error> {
52    let path = path.into();
53    let path_str = match path {
54        Cow::Owned(path) => Cow::Owned({
55            #[cfg(unix)]
56            let p: BString = {
57                use std::os::unix::ffi::OsStringExt;
58                path.into_os_string().into_vec().into()
59            };
60            #[cfg(not(unix))]
61            let p: BString = path.into_os_string().into_string().map_err(|_| Utf8Error)?.into();
62            p
63        }),
64        Cow::Borrowed(path) => Cow::Borrowed({
65            #[cfg(unix)]
66            let p: &BStr = {
67                use std::os::unix::ffi::OsStrExt;
68                path.as_os_str().as_bytes().into()
69            };
70            #[cfg(not(unix))]
71            let p: &BStr = path.to_str().ok_or(Utf8Error)?.as_bytes().into();
72            p
73        }),
74    };
75    Ok(path_str)
76}
77
78/// Similar to [`try_into_bstr()`] but **panics** if malformed surrogates are encountered on Windows.
79pub fn into_bstr<'a>(path: impl Into<Cow<'a, Path>>) -> Cow<'a, BStr> {
80    try_into_bstr(path).expect("prefix path doesn't contain ill-formed UTF-8")
81}
82
83/// Join `path` to `base` such that they are separated with a `/`, i.e. `base/path`.
84pub fn join_bstr_unix_pathsep<'a, 'b>(base: impl Into<Cow<'a, BStr>>, path: impl Into<&'b BStr>) -> Cow<'a, BStr> {
85    let mut base = base.into();
86    if !base.is_empty() && base.last() != Some(&b'/') {
87        base.to_mut().push(b'/');
88    }
89    base.to_mut().extend_from_slice(path.into());
90    base
91}
92
93/// Given `input` bytes, produce a `Path` from them ignoring encoding entirely if on Unix.
94///
95/// On non-Unix platforms, the input is required to be valid UTF-8, which is guaranteed if we wrote it before.
96/// There are some potential Git versions and Windows installations which produce malformed UTF-16
97/// if certain emojis are in the path. It's as rare as it sounds, but possible.
98pub fn try_from_byte_slice(input: &[u8]) -> Result<&Path, Utf8Error> {
99    #[cfg(unix)]
100    let p = {
101        use std::os::unix::ffi::OsStrExt;
102        OsStr::from_bytes(input).as_ref()
103    };
104    #[cfg(not(unix))]
105    let p = Path::new(std::str::from_utf8(input).map_err(|_| Utf8Error)?);
106    Ok(p)
107}
108
109/// Similar to [`from_byte_slice()`], but takes either borrowed or owned `input`.
110pub fn try_from_bstr<'a>(input: impl Into<Cow<'a, BStr>>) -> Result<Cow<'a, Path>, Utf8Error> {
111    let input = input.into();
112    match input {
113        Cow::Borrowed(input) => try_from_byte_slice(input).map(Cow::Borrowed),
114        Cow::Owned(input) => try_from_bstring(input).map(Cow::Owned),
115    }
116}
117
118/// Similar to [`try_from_bstr()`], but **panics** if malformed surrogates are encountered on Windows.
119pub fn from_bstr<'a>(input: impl Into<Cow<'a, BStr>>) -> Cow<'a, Path> {
120    try_from_bstr(input).expect("prefix path doesn't contain ill-formed UTF-8")
121}
122
123/// Similar to [`try_from_bstr()`], but takes and produces owned data.
124pub fn try_from_bstring(input: impl Into<BString>) -> Result<PathBuf, Utf8Error> {
125    let input = input.into();
126    #[cfg(unix)]
127    let p = {
128        use std::os::unix::ffi::OsStringExt;
129        std::ffi::OsString::from_vec(input.into()).into()
130    };
131    #[cfg(not(unix))]
132    let p = {
133        use bstr::ByteVec;
134        PathBuf::from(
135            {
136                let v: Vec<_> = input.into();
137                v
138            }
139            .into_string()
140            .map_err(|_| Utf8Error)?,
141        )
142    };
143    Ok(p)
144}
145
146/// Similar to [`try_from_bstring()`], but will **panic** if there is ill-formed UTF-8 in the `input`.
147pub fn from_bstring(input: impl Into<BString>) -> PathBuf {
148    try_from_bstring(input).expect("well-formed UTF-8 on windows")
149}
150
151/// Similar to [`try_from_byte_slice()`], but will **panic** if there is ill-formed UTF-8 in the `input`.
152pub fn from_byte_slice(input: &[u8]) -> &Path {
153    try_from_byte_slice(input).expect("well-formed UTF-8 on windows")
154}
155
156fn replace<'a>(path: impl Into<Cow<'a, BStr>>, find: u8, replace: u8) -> Cow<'a, BStr> {
157    let path = path.into();
158    match path {
159        Cow::Owned(mut path) => {
160            for b in path.iter_mut().filter(|b| **b == find) {
161                *b = replace;
162            }
163            path.into()
164        }
165        Cow::Borrowed(path) => {
166            if !path.contains(&find) {
167                return path.into();
168            }
169            let mut path = path.to_owned();
170            for b in path.iter_mut().filter(|b| **b == find) {
171                *b = replace;
172            }
173            path.into()
174        }
175    }
176}
177
178/// Assures the given bytes use the native path separator.
179pub fn to_native_separators<'a>(path: impl Into<Cow<'a, BStr>>) -> Cow<'a, BStr> {
180    #[cfg(not(windows))]
181    let p = to_unix_separators(path);
182    #[cfg(windows)]
183    let p = to_windows_separators(path);
184    p
185}
186
187/// Convert paths with slashes to backslashes on Windows and do nothing on Unix,
188/// but **panic** if unpaired surrogates are encountered on Windows.
189pub fn to_native_path_on_windows<'a>(path: impl Into<Cow<'a, BStr>>) -> Cow<'a, std::path::Path> {
190    #[cfg(not(windows))]
191    {
192        crate::from_bstr(path)
193    }
194    #[cfg(windows)]
195    {
196        crate::from_bstr(to_windows_separators(path))
197    }
198}
199
200/// Replace Windows path separators with slashes, but only do so on Windows.
201pub fn to_unix_separators_on_windows<'a>(path: impl Into<Cow<'a, BStr>>) -> Cow<'a, BStr> {
202    #[cfg(windows)]
203    {
204        to_unix_separators(path)
205    }
206    #[cfg(not(windows))]
207    {
208        path.into()
209    }
210}
211
212/// Replace Windows path separators with slashes, which typically resembles a Unix path, unconditionally.
213///
214/// **Note** Do not use these and prefer the conditional versions of this method.
215pub fn to_unix_separators<'a>(path: impl Into<Cow<'a, BStr>>) -> Cow<'a, BStr> {
216    replace(path, b'\\', b'/')
217}
218
219/// Find slashes and replace them with backslashes, unconditionally.
220///
221/// **Note** Do not use these and prefer the conditional versions of this method.
222pub fn to_windows_separators<'a>(path: impl Into<Cow<'a, BStr>>) -> Cow<'a, BStr> {
223    replace(path, b'/', b'\\')
224}
225
226/// Resolve relative components virtually, eliminating intermediate `..` without accessing the filesystem.
227///
228/// For example, this turns `a/./b/c/.././..` into `a`, and turns `/a/../b/..` into `/`.
229///
230/// ```
231/// # fn main() {
232/// # use std::path::Path;
233/// # use gix_path::normalize;
234/// for (input, expected) in [
235///     ("a/./b/c/.././..", "a"),
236///     ("/a/../b/..", "/"),
237///     ("/base/a/..", "/base"),
238///     ("./a/..", "."),
239///     ("./a/../..", "/"),
240///     (".///", ".///"),
241///     ("a//b", "a//b"),
242///     ("/base/../base", "/base"),
243/// ] {
244///     let input = Path::new(input);
245///     let expected = Path::new(expected);
246///     assert_eq!(normalize(input.into(), Path::new("/cwd")), Some(expected.into()));
247/// }
248/// # }
249/// ```
250///
251/// Leading `.` components as well as duplicate separators are left untouched.
252///
253/// This is particularly useful when manipulating paths that are based on user input, and not
254/// resolving intermediate symlinks keeps the path similar to what the user provided. If that's not
255/// desirable, use [`realpath()`](crate::realpath()) instead.
256///
257/// Note that we will use the `current_dir` if we run out of path components to pop off, which
258/// is expected to be absolute as typical return value of `std::env::current_dir()` or
259/// `gix_fs::current_dir(…)` when `core.precomposeUnicode` is known. As a `current_dir` like `/c`
260/// can be exhausted by paths like `../../r`, `None` will be returned to indicate the inability to
261/// produce a logically consistent path.
262pub fn normalize<'a>(path: Cow<'a, Path>, current_dir: &Path) -> Option<Cow<'a, Path>> {
263    use std::path::Component::ParentDir;
264
265    if !path.components().any(|c| matches!(c, ParentDir)) {
266        return Some(path);
267    }
268    let mut current_dir_opt = Some(current_dir);
269    let was_relative = path.is_relative();
270    let components = path.components();
271    let mut path = PathBuf::new();
272    for component in components {
273        if let ParentDir = component {
274            let path_was_dot = path == Path::new(".");
275            if path.as_os_str().is_empty() || path_was_dot {
276                path.push(current_dir_opt.take()?);
277            }
278            if !path.pop() {
279                return None;
280            }
281        } else {
282            path.push(component);
283        }
284    }
285
286    if (path.as_os_str().is_empty() || path == current_dir) && was_relative {
287        Cow::Borrowed(Path::new("."))
288    } else {
289        path.into()
290    }
291    .into()
292}
293
294/// Rebuild the worktree-relative `relative_path` to be relative to `prefix`, which is the
295/// worktree-relative path equivalent to the position of the user, or current working directory.
296///
297/// This is a no-op if `prefix` is empty.
298///
299/// Note that both `relative_path` and `prefix` are assumed to be [normalized](normalize()), and
300/// failure to do so will lead to incorrect results.
301///
302/// Note that both input paths are expected to be equal in terms of case too, as comparisons will
303/// be case-sensitive.
304pub fn relativize_with_prefix<'a>(relative_path: &'a Path, prefix: &Path) -> Cow<'a, Path> {
305    if prefix.as_os_str().is_empty() {
306        return Cow::Borrowed(relative_path);
307    }
308    debug_assert!(
309        relative_path.components().all(|c| matches!(c, Component::Normal(_))),
310        "BUG: all input is expected to be normalized, but relative_path was not"
311    );
312    debug_assert!(
313        prefix.components().all(|c| matches!(c, Component::Normal(_))),
314        "BUG: all input is expected to be normalized, but prefix was not"
315    );
316
317    let mut buf = PathBuf::new();
318    let mut rpc = relative_path.components().peekable();
319    let mut equal_thus_far = true;
320    for pcomp in prefix.components() {
321        if equal_thus_far {
322            if let (Component::Normal(pname), Some(Component::Normal(rpname))) = (pcomp, rpc.peek()) {
323                if &pname == rpname {
324                    rpc.next();
325                    continue;
326                } else {
327                    equal_thus_far = false;
328                }
329            }
330        }
331        buf.push(Component::ParentDir);
332    }
333    buf.extend(rpc);
334    if buf.as_os_str().is_empty() {
335        Cow::Borrowed(Path::new("."))
336    } else {
337        Cow::Owned(buf)
338    }
339}
gix_path/convert.rs

gix_path/
convert.rs
