1 // Copyright 2012 The Rust Project Developers. See the COPYRIGHT
2 // file at the top-level directory of this distribution and at
3 // http://rust-lang.org/COPYRIGHT.
5 // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
6 // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
7 // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
8 // option. This file may not be copied, modified, or distributed
9 // except according to those terms.
11 //! C-string manipulation and management
13 //! This modules provides the basic methods for creating and manipulating
14 //! null-terminated strings for use with FFI calls (back to C). Most C APIs require
15 //! that the string being passed to them is null-terminated, and by default rust's
16 //! string types are *not* null terminated.
18 //! The other problem with translating Rust strings to C strings is that Rust
19 //! strings can validly contain a null-byte in the middle of the string (0 is a
20 //! valid Unicode codepoint). This means that not all Rust strings can actually be
21 //! translated to C strings.
23 //! # Creation of a C string
25 //! A C string is managed through the `CString` type defined in this module. It
26 //! "owns" the internal buffer of characters and will automatically deallocate the
27 //! buffer when the string is dropped. The `ToCStr` trait is implemented for `&str`
28 //! and `&[u8]`, but the conversions can fail due to some of the limitations
31 //! This also means that currently whenever a C string is created, an allocation
32 //! must be performed to place the data elsewhere (the lifetime of the C string is
33 //! not tied to the lifetime of the original string/data buffer). If C strings are
34 //! heavily used in applications, then caching may be advisable to prevent
35 //! unnecessary amounts of allocations.
37 //! Be carefull to remember that the memory is managed by C allocator API and not
38 //! by Rust allocator API.
39 //! That means that the CString pointers should be freed with C allocator API
40 //! if you intend to do that on your own, as the behaviour if you free them with
41 //! Rust's allocator API is not well defined
43 //! An example of creating and using a C string would be:
46 //! extern crate libc;
49 //! fn puts(s: *const libc::c_char);
53 //! let my_string = "Hello, world!";
55 //! // Allocate the C string with an explicit local that owns the string. The
56 //! // `c_buffer` pointer will be deallocated when `my_c_string` goes out of scope.
57 //! let my_c_string = my_string.to_c_str();
59 //! puts(my_c_string.as_ptr());
62 //! // Don't save/return the pointer to the C string, the `c_buffer` will be
63 //! // deallocated when this block returns!
64 //! my_string.with_c_str(|c_buffer| {
65 //! unsafe { puts(c_buffer); }
70 use collections::string::String;
71 use collections::hash;
73 use core::kinds::{Sized, marker};
75 use core::prelude::{Clone, Drop, Eq, Iterator};
76 use core::prelude::{SlicePrelude, None, Option, Ordering, PartialEq};
77 use core::prelude::{PartialOrd, RawPtr, Some, StrPrelude, range};
84 /// The representation of a C String.
86 /// This structure wraps a `*libc::c_char`, and will automatically free the
87 /// memory it is pointing to when it goes out of scope.
89 buf: *const libc::c_char,
93 impl Clone for CString {
94 /// Clone this CString into a new, uniquely owned CString. For safety
95 /// reasons, this is always a deep clone with the memory allocated
96 /// with C's allocator API, rather than the usual shallow clone.
97 fn clone(&self) -> CString {
98 let len = self.len() + 1;
99 let buf = unsafe { libc::malloc(len as libc::size_t) } as *mut libc::c_char;
100 if buf.is_null() { ::alloc::oom() }
101 unsafe { ptr::copy_nonoverlapping_memory(buf, self.buf, len); }
102 CString { buf: buf as *const libc::c_char, owns_buffer_: true }
106 impl PartialEq for CString {
107 fn eq(&self, other: &CString) -> bool {
108 // Check if the two strings share the same buffer
109 if self.buf as uint == other.buf as uint {
113 libc::strcmp(self.buf, other.buf) == 0
119 impl PartialOrd for CString {
121 fn partial_cmp(&self, other: &CString) -> Option<Ordering> {
122 self.as_bytes().partial_cmp(other.as_bytes())
126 impl Eq for CString {}
128 impl<S: hash::Writer> hash::Hash<S> for CString {
130 fn hash(&self, state: &mut S) {
131 self.as_bytes().hash(state)
136 /// Create a C String from a pointer, with memory managed by C's allocator
137 /// API, so avoid calling it with a pointer to memory managed by Rust's
138 /// allocator API, as the behaviour would not be well defined.
142 /// Panics if `buf` is null
143 pub unsafe fn new(buf: *const libc::c_char, owns_buffer: bool) -> CString {
144 assert!(!buf.is_null());
145 CString { buf: buf, owns_buffer_: owns_buffer }
148 /// Return a pointer to the NUL-terminated string data.
150 /// `.as_ptr` returns an internal pointer into the `CString`, and
151 /// may be invalidated when the `CString` falls out of scope (the
152 /// destructor will run, freeing the allocation if there is
156 /// let foo = "some string";
159 /// let x = foo.to_c_str();
160 /// let p = x.as_ptr();
162 /// // wrong (the CString will be freed, invalidating `p`)
163 /// let p = foo.to_c_str().as_ptr();
169 /// extern crate libc;
172 /// let c_str = "foo bar".to_c_str();
174 /// libc::puts(c_str.as_ptr());
178 pub fn as_ptr(&self) -> *const libc::c_char {
182 /// Return a mutable pointer to the NUL-terminated string data.
184 /// `.as_mut_ptr` returns an internal pointer into the `CString`, and
185 /// may be invalidated when the `CString` falls out of scope (the
186 /// destructor will run, freeing the allocation if there is
190 /// let foo = "some string";
193 /// let mut x = foo.to_c_str();
194 /// let p = x.as_mut_ptr();
196 /// // wrong (the CString will be freed, invalidating `p`)
197 /// let p = foo.to_c_str().as_mut_ptr();
199 pub fn as_mut_ptr(&mut self) -> *mut libc::c_char {
203 /// Returns whether or not the `CString` owns the buffer.
204 pub fn owns_buffer(&self) -> bool {
208 /// Converts the CString into a `&[u8]` without copying.
209 /// Includes the terminating NUL byte.
211 pub fn as_bytes<'a>(&'a self) -> &'a [u8] {
213 mem::transmute(Slice { data: self.buf, len: self.len() + 1 })
217 /// Converts the CString into a `&[u8]` without copying.
218 /// Does not include the terminating NUL byte.
220 pub fn as_bytes_no_nul<'a>(&'a self) -> &'a [u8] {
222 mem::transmute(Slice { data: self.buf, len: self.len() })
226 /// Converts the CString into a `&str` without copying.
227 /// Returns None if the CString is not UTF-8.
229 pub fn as_str<'a>(&'a self) -> Option<&'a str> {
230 let buf = self.as_bytes_no_nul();
234 /// Return a CString iterator.
235 pub fn iter<'a>(&'a self) -> CChars<'a> {
238 marker: marker::ContravariantLifetime,
242 /// Unwraps the wrapped `*libc::c_char` from the `CString` wrapper.
244 /// Any ownership of the buffer by the `CString` wrapper is
245 /// forgotten, meaning that the backing allocation of this
246 /// `CString` is not automatically freed if it owns the
247 /// allocation. In this case, a user of `.unwrap()` should ensure
248 /// the allocation is freed, to avoid leaking memory. You should
249 /// use libc's memory allocator in this case.
251 /// Prefer `.as_ptr()` when just retrieving a pointer to the
252 /// string data, as that does not relinquish ownership.
253 pub unsafe fn into_inner(mut self) -> *const libc::c_char {
254 self.owns_buffer_ = false;
258 /// Deprecated, use into_inner() instead
259 #[deprecated = "renamed to into_inner()"]
260 pub unsafe fn unwrap(self) -> *const libc::c_char { self.into_inner() }
262 /// Return the number of bytes in the CString (not including the NUL
265 pub fn len(&self) -> uint {
266 unsafe { libc::strlen(self.buf) as uint }
269 /// Returns if there are no bytes in this string
271 pub fn is_empty(&self) -> bool { self.len() == 0 }
274 impl Drop for CString {
276 if self.owns_buffer_ {
278 libc::free(self.buf as *mut libc::c_void)
284 impl fmt::Show for CString {
285 fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
286 String::from_utf8_lossy(self.as_bytes_no_nul()).fmt(f)
290 /// A generic trait for converting a value to a CString.
291 pub trait ToCStr for Sized? {
292 /// Copy the receiver into a CString.
296 /// Panics the task if the receiver has an interior null.
297 fn to_c_str(&self) -> CString;
299 /// Unsafe variant of `to_c_str()` that doesn't check for nulls.
300 unsafe fn to_c_str_unchecked(&self) -> CString;
302 /// Work with a temporary CString constructed from the receiver.
303 /// The provided `*libc::c_char` will be freed immediately upon return.
308 /// extern crate libc;
311 /// let s = "PATH".with_c_str(|path| unsafe {
312 /// libc::getenv(path)
319 /// Panics the task if the receiver has an interior null.
321 fn with_c_str<T>(&self, f: |*const libc::c_char| -> T) -> T {
322 let c_str = self.to_c_str();
326 /// Unsafe variant of `with_c_str()` that doesn't check for nulls.
328 unsafe fn with_c_str_unchecked<T>(&self, f: |*const libc::c_char| -> T) -> T {
329 let c_str = self.to_c_str_unchecked();
334 impl ToCStr for str {
336 fn to_c_str(&self) -> CString {
337 self.as_bytes().to_c_str()
341 unsafe fn to_c_str_unchecked(&self) -> CString {
342 self.as_bytes().to_c_str_unchecked()
346 fn with_c_str<T>(&self, f: |*const libc::c_char| -> T) -> T {
347 self.as_bytes().with_c_str(f)
351 unsafe fn with_c_str_unchecked<T>(&self, f: |*const libc::c_char| -> T) -> T {
352 self.as_bytes().with_c_str_unchecked(f)
356 impl ToCStr for String {
358 fn to_c_str(&self) -> CString {
359 self.as_bytes().to_c_str()
363 unsafe fn to_c_str_unchecked(&self) -> CString {
364 self.as_bytes().to_c_str_unchecked()
368 fn with_c_str<T>(&self, f: |*const libc::c_char| -> T) -> T {
369 self.as_bytes().with_c_str(f)
373 unsafe fn with_c_str_unchecked<T>(&self, f: |*const libc::c_char| -> T) -> T {
374 self.as_bytes().with_c_str_unchecked(f)
378 // The length of the stack allocated buffer for `vec.with_c_str()`
379 const BUF_LEN: uint = 128;
381 impl ToCStr for [u8] {
382 fn to_c_str(&self) -> CString {
383 let mut cs = unsafe { self.to_c_str_unchecked() };
384 check_for_null(self, cs.as_mut_ptr());
388 unsafe fn to_c_str_unchecked(&self) -> CString {
389 let self_len = self.len();
390 let buf = libc::malloc(self_len as libc::size_t + 1) as *mut u8;
391 if buf.is_null() { ::alloc::oom() }
393 ptr::copy_memory(buf, self.as_ptr(), self_len);
394 *buf.offset(self_len as int) = 0;
396 CString::new(buf as *const libc::c_char, true)
399 fn with_c_str<T>(&self, f: |*const libc::c_char| -> T) -> T {
400 unsafe { with_c_str(self, true, f) }
403 unsafe fn with_c_str_unchecked<T>(&self, f: |*const libc::c_char| -> T) -> T {
404 with_c_str(self, false, f)
408 impl<'a, Sized? T: ToCStr> ToCStr for &'a T {
410 fn to_c_str(&self) -> CString {
415 unsafe fn to_c_str_unchecked(&self) -> CString {
416 (**self).to_c_str_unchecked()
420 fn with_c_str<T>(&self, f: |*const libc::c_char| -> T) -> T {
421 (**self).with_c_str(f)
425 unsafe fn with_c_str_unchecked<T>(&self, f: |*const libc::c_char| -> T) -> T {
426 (**self).with_c_str_unchecked(f)
430 // Unsafe function that handles possibly copying the &[u8] into a stack array.
431 unsafe fn with_c_str<T>(v: &[u8], checked: bool,
432 f: |*const libc::c_char| -> T) -> T {
433 let c_str = if v.len() < BUF_LEN {
434 let mut buf: [u8, .. BUF_LEN] = mem::uninitialized();
435 slice::bytes::copy_memory(&mut buf, v);
438 let buf = buf.as_mut_ptr();
440 check_for_null(v, buf as *mut libc::c_char);
443 return f(buf as *const libc::c_char)
447 v.to_c_str_unchecked()
454 fn check_for_null(v: &[u8], buf: *mut libc::c_char) {
455 for i in range(0, v.len()) {
457 let p = buf.offset(i as int);
463 /// External iterator for a CString's bytes.
465 /// Use with the `std::iter` module.
466 pub struct CChars<'a> {
467 ptr: *const libc::c_char,
468 marker: marker::ContravariantLifetime<'a>,
471 impl<'a> Iterator<libc::c_char> for CChars<'a> {
472 fn next(&mut self) -> Option<libc::c_char> {
473 let ch = unsafe { *self.ptr };
477 self.ptr = unsafe { self.ptr.offset(1) };
483 /// Parses a C "multistring", eg windows env values or
484 /// the req->ptr result in a uv_fs_readdir() call.
486 /// Optionally, a `count` can be passed in, limiting the
487 /// parsing to only being done `count`-times.
489 /// The specified closure is invoked with each string that
490 /// is found, and the number of strings found is returned.
491 pub unsafe fn from_c_multistring(buf: *const libc::c_char,
493 f: |&CString|) -> uint {
495 let mut curr_ptr: uint = buf as uint;
497 let (limited_count, limit) = match count {
498 Some(limit) => (true, limit),
501 while ((limited_count && ctr < limit) || !limited_count)
502 && *(curr_ptr as *const libc::c_char) != 0 as libc::c_char {
503 let cstr = CString::new(curr_ptr as *const libc::c_char, false);
505 curr_ptr += cstr.len() + 1;
521 fn test_str_multistring_parsing() {
523 let input = b"zero\0one\0\0";
524 let ptr = input.as_ptr();
525 let expected = ["zero", "one"];
526 let mut it = expected.iter();
527 let result = from_c_multistring(ptr as *const libc::c_char, None, |c| {
528 let cbytes = c.as_bytes_no_nul();
529 assert_eq!(cbytes, it.next().unwrap().as_bytes());
531 assert_eq!(result, 2);
532 assert!(it.next().is_none());
537 fn test_str_to_c_str() {
538 let c_str = "".to_c_str();
540 assert_eq!(*c_str.as_ptr().offset(0), 0);
543 let c_str = "hello".to_c_str();
544 let buf = c_str.as_ptr();
546 assert_eq!(*buf.offset(0), 'h' as libc::c_char);
547 assert_eq!(*buf.offset(1), 'e' as libc::c_char);
548 assert_eq!(*buf.offset(2), 'l' as libc::c_char);
549 assert_eq!(*buf.offset(3), 'l' as libc::c_char);
550 assert_eq!(*buf.offset(4), 'o' as libc::c_char);
551 assert_eq!(*buf.offset(5), 0);
556 fn test_vec_to_c_str() {
558 let c_str = b.to_c_str();
560 assert_eq!(*c_str.as_ptr().offset(0), 0);
563 let c_str = b"hello".to_c_str();
564 let buf = c_str.as_ptr();
566 assert_eq!(*buf.offset(0), 'h' as libc::c_char);
567 assert_eq!(*buf.offset(1), 'e' as libc::c_char);
568 assert_eq!(*buf.offset(2), 'l' as libc::c_char);
569 assert_eq!(*buf.offset(3), 'l' as libc::c_char);
570 assert_eq!(*buf.offset(4), 'o' as libc::c_char);
571 assert_eq!(*buf.offset(5), 0);
574 let c_str = b"foo\xFF".to_c_str();
575 let buf = c_str.as_ptr();
577 assert_eq!(*buf.offset(0), 'f' as libc::c_char);
578 assert_eq!(*buf.offset(1), 'o' as libc::c_char);
579 assert_eq!(*buf.offset(2), 'o' as libc::c_char);
580 assert_eq!(*buf.offset(3), 0xffu8 as i8);
581 assert_eq!(*buf.offset(4), 0);
587 let c_str = "hello".to_c_str();
588 unsafe { libc::free(c_str.unwrap() as *mut libc::c_void) }
593 let c_str = "hello".to_c_str();
594 let len = unsafe { libc::strlen(c_str.as_ptr()) };
600 let c_str = "".to_c_str();
601 let mut iter = c_str.iter();
602 assert_eq!(iter.next(), None);
604 let c_str = "hello".to_c_str();
605 let mut iter = c_str.iter();
606 assert_eq!(iter.next(), Some('h' as libc::c_char));
607 assert_eq!(iter.next(), Some('e' as libc::c_char));
608 assert_eq!(iter.next(), Some('l' as libc::c_char));
609 assert_eq!(iter.next(), Some('l' as libc::c_char));
610 assert_eq!(iter.next(), Some('o' as libc::c_char));
611 assert_eq!(iter.next(), None);
615 fn test_to_c_str_fail() {
616 assert!(task::try(proc() { "he\x00llo".to_c_str() }).is_err());
620 fn test_to_c_str_unchecked() {
622 let c_string = "he\x00llo".to_c_str_unchecked();
623 let buf = c_string.as_ptr();
624 assert_eq!(*buf.offset(0), 'h' as libc::c_char);
625 assert_eq!(*buf.offset(1), 'e' as libc::c_char);
626 assert_eq!(*buf.offset(2), 0);
627 assert_eq!(*buf.offset(3), 'l' as libc::c_char);
628 assert_eq!(*buf.offset(4), 'l' as libc::c_char);
629 assert_eq!(*buf.offset(5), 'o' as libc::c_char);
630 assert_eq!(*buf.offset(6), 0);
636 let c_str = "hello".to_c_str();
637 assert_eq!(c_str.as_bytes(), b"hello\0");
638 let c_str = "".to_c_str();
639 assert_eq!(c_str.as_bytes(), b"\0");
640 let c_str = b"foo\xFF".to_c_str();
641 assert_eq!(c_str.as_bytes(), b"foo\xFF\0");
645 fn test_as_bytes_no_nul() {
646 let c_str = "hello".to_c_str();
647 assert_eq!(c_str.as_bytes_no_nul(), b"hello");
648 let c_str = "".to_c_str();
649 let exp: &[u8] = &[];
650 assert_eq!(c_str.as_bytes_no_nul(), exp);
651 let c_str = b"foo\xFF".to_c_str();
652 assert_eq!(c_str.as_bytes_no_nul(), b"foo\xFF");
657 let c_str = "hello".to_c_str();
658 assert_eq!(c_str.as_str(), Some("hello"));
659 let c_str = "".to_c_str();
660 assert_eq!(c_str.as_str(), Some(""));
661 let c_str = b"foo\xFF".to_c_str();
662 assert_eq!(c_str.as_str(), None);
668 let _c_str = unsafe { CString::new(ptr::null(), false) };
673 let a = "hello".to_c_str();
679 fn test_clone_noleak() {
680 fn foo(f: |c: &CString|) {
681 let s = "test".to_string();
682 let c = s.to_c_str();
683 // give the closure a non-owned CString
684 let mut c_ = unsafe { CString::new(c.as_ptr(), false) };
686 // muck with the buffer for later printing
687 unsafe { *c_.as_mut_ptr() = 'X' as libc::c_char }
690 let mut c_: Option<CString> = None;
692 c_ = Some(c.clone());
694 // force a copy, reading the memory
695 c.as_bytes().to_vec();
697 let c_ = c_.unwrap();
698 // force a copy, reading the memory
699 c_.as_bytes().to_vec();
710 fn check(s: &str, c_str: *const libc::c_char) {
711 let s_buf = s.as_ptr();
712 for i in range(0, s.len()) {
715 *s_buf.offset(i as int) as libc::c_char,
716 *c_str.offset(i as int));
721 static S_SHORT: &'static str = "Mary";
722 static S_MEDIUM: &'static str = "Mary had a little lamb";
723 static S_LONG: &'static str = "\
724 Mary had a little lamb, Little lamb
725 Mary had a little lamb, Little lamb
726 Mary had a little lamb, Little lamb
727 Mary had a little lamb, Little lamb
728 Mary had a little lamb, Little lamb
729 Mary had a little lamb, Little lamb";
731 fn bench_to_string(b: &mut Bencher, s: &str) {
733 let c_str = s.to_c_str();
734 check(s, c_str.as_ptr());
739 fn bench_to_c_str_short(b: &mut Bencher) {
740 bench_to_string(b, S_SHORT)
744 fn bench_to_c_str_medium(b: &mut Bencher) {
745 bench_to_string(b, S_MEDIUM)
749 fn bench_to_c_str_long(b: &mut Bencher) {
750 bench_to_string(b, S_LONG)
753 fn bench_to_c_str_unchecked(b: &mut Bencher, s: &str) {
755 let c_str = unsafe { s.to_c_str_unchecked() };
756 check(s, c_str.as_ptr())
761 fn bench_to_c_str_unchecked_short(b: &mut Bencher) {
762 bench_to_c_str_unchecked(b, S_SHORT)
766 fn bench_to_c_str_unchecked_medium(b: &mut Bencher) {
767 bench_to_c_str_unchecked(b, S_MEDIUM)
771 fn bench_to_c_str_unchecked_long(b: &mut Bencher) {
772 bench_to_c_str_unchecked(b, S_LONG)
775 fn bench_with_c_str(b: &mut Bencher, s: &str) {
777 s.with_c_str(|c_str_buf| check(s, c_str_buf))
782 fn bench_with_c_str_short(b: &mut Bencher) {
783 bench_with_c_str(b, S_SHORT)
787 fn bench_with_c_str_medium(b: &mut Bencher) {
788 bench_with_c_str(b, S_MEDIUM)
792 fn bench_with_c_str_long(b: &mut Bencher) {
793 bench_with_c_str(b, S_LONG)
796 fn bench_with_c_str_unchecked(b: &mut Bencher, s: &str) {
799 s.with_c_str_unchecked(|c_str_buf| check(s, c_str_buf))
805 fn bench_with_c_str_unchecked_short(b: &mut Bencher) {
806 bench_with_c_str_unchecked(b, S_SHORT)
810 fn bench_with_c_str_unchecked_medium(b: &mut Bencher) {
811 bench_with_c_str_unchecked(b, S_MEDIUM)
815 fn bench_with_c_str_unchecked_long(b: &mut Bencher) {
816 bench_with_c_str_unchecked(b, S_LONG)