percent_encoding/
lib.rs

Help
1// Copyright 2013-2016 The rust-url developers.
2//
3// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
4// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
5// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
6// option. This file may not be copied, modified, or distributed
7// except according to those terms.
8
9//! URLs use special characters to indicate the parts of the request.
10//! For example, a `?` question mark marks the end of a path and the start of a query string.
11//! In order for that character to exist inside a path, it needs to be encoded differently.
12//!
13//! Percent encoding replaces reserved characters with the `%` escape character
14//! followed by a byte value as two hexadecimal digits.
15//! For example, an ASCII space is replaced with `%20`.
16//!
17//! When encoding, the set of characters that can (and should, for readability) be left alone
18//! depends on the context.
19//! The `?` question mark mentioned above is not a separator when used literally
20//! inside of a query string, and therefore does not need to be encoded.
21//! The [`AsciiSet`] parameter of [`percent_encode`] and [`utf8_percent_encode`]
22//! lets callers configure this.
23//!
24//! This crate deliberately does not provide many different sets.
25//! Users should consider in what context the encoded string will be used,
26//! read relevant specifications, and define their own set.
27//! This is done by using the `add` method of an existing set.
28//!
29//! # Examples
30//!
31//! ```
32//! use percent_encoding::{utf8_percent_encode, AsciiSet, CONTROLS};
33//!
34//! /// https://url.spec.whatwg.org/#fragment-percent-encode-set
35//! const FRAGMENT: &AsciiSet = &CONTROLS.add(b' ').add(b'"').add(b'<').add(b'>').add(b'`');
36//!
37//! assert_eq!(utf8_percent_encode("foo <bar>", FRAGMENT).to_string(), "foo%20%3Cbar%3E");
38//! ```
39
40#![no_std]
41#[cfg(feature = "alloc")]
42extern crate alloc;
43
44#[cfg(feature = "alloc")]
45use alloc::{
46    borrow::{Cow, ToOwned},
47    string::String,
48    vec::Vec,
49};
50use core::{fmt, mem, slice, str};
51
52/// Represents a set of characters or bytes in the ASCII range.
53///
54/// This is used in [`percent_encode`] and [`utf8_percent_encode`].
55/// This is similar to [percent-encode sets](https://url.spec.whatwg.org/#percent-encoded-bytes).
56///
57/// Use the `add` method of an existing set to define a new set. For example:
58///
59/// ```
60/// use percent_encoding::{AsciiSet, CONTROLS};
61///
62/// /// https://url.spec.whatwg.org/#fragment-percent-encode-set
63/// const FRAGMENT: &AsciiSet = &CONTROLS.add(b' ').add(b'"').add(b'<').add(b'>').add(b'`');
64/// ```
65pub struct AsciiSet {
66    mask: [Chunk; ASCII_RANGE_LEN / BITS_PER_CHUNK],
67}
68
69type Chunk = u32;
70
71const ASCII_RANGE_LEN: usize = 0x80;
72
73const BITS_PER_CHUNK: usize = 8 * mem::size_of::<Chunk>();
74
75impl AsciiSet {
76    /// Called with UTF-8 bytes rather than code points.
77    /// Not used for non-ASCII bytes.
78    const fn contains(&self, byte: u8) -> bool {
79        let chunk = self.mask[byte as usize / BITS_PER_CHUNK];
80        let mask = 1 << (byte as usize % BITS_PER_CHUNK);
81        (chunk & mask) != 0
82    }
83
84    fn should_percent_encode(&self, byte: u8) -> bool {
85        !byte.is_ascii() || self.contains(byte)
86    }
87
88    pub const fn add(&self, byte: u8) -> Self {
89        let mut mask = self.mask;
90        mask[byte as usize / BITS_PER_CHUNK] |= 1 << (byte as usize % BITS_PER_CHUNK);
91        AsciiSet { mask }
92    }
93
94    pub const fn remove(&self, byte: u8) -> Self {
95        let mut mask = self.mask;
96        mask[byte as usize / BITS_PER_CHUNK] &= !(1 << (byte as usize % BITS_PER_CHUNK));
97        AsciiSet { mask }
98    }
99}
100
101/// The set of 0x00 to 0x1F (C0 controls), and 0x7F (DEL).
102///
103/// Note that this includes the newline and tab characters, but not the space 0x20.
104///
105/// <https://url.spec.whatwg.org/#c0-control-percent-encode-set>
106pub const CONTROLS: &AsciiSet = &AsciiSet {
107    mask: [
108        !0_u32, // C0: 0x00 to 0x1F (32 bits set)
109        0,
110        0,
111        1 << (0x7F_u32 % 32), // DEL: 0x7F (one bit set)
112    ],
113};
114
115macro_rules! static_assert {
116    ($( $bool: expr, )+) => {
117        fn _static_assert() {
118            $(
119                let _ = mem::transmute::<[u8; $bool as usize], u8>;
120            )+
121        }
122    }
123}
124
125static_assert! {
126    CONTROLS.contains(0x00),
127    CONTROLS.contains(0x1F),
128    !CONTROLS.contains(0x20),
129    !CONTROLS.contains(0x7E),
130    CONTROLS.contains(0x7F),
131}
132
133/// Everything that is not an ASCII letter or digit.
134///
135/// This is probably more eager than necessary in any context.
136pub const NON_ALPHANUMERIC: &AsciiSet = &CONTROLS
137    .add(b' ')
138    .add(b'!')
139    .add(b'"')
140    .add(b'#')
141    .add(b'$')
142    .add(b'%')
143    .add(b'&')
144    .add(b'\'')
145    .add(b'(')
146    .add(b')')
147    .add(b'*')
148    .add(b'+')
149    .add(b',')
150    .add(b'-')
151    .add(b'.')
152    .add(b'/')
153    .add(b':')
154    .add(b';')
155    .add(b'<')
156    .add(b'=')
157    .add(b'>')
158    .add(b'?')
159    .add(b'@')
160    .add(b'[')
161    .add(b'\\')
162    .add(b']')
163    .add(b'^')
164    .add(b'_')
165    .add(b'`')
166    .add(b'{')
167    .add(b'|')
168    .add(b'}')
169    .add(b'~');
170
171/// Return the percent-encoding of the given byte.
172///
173/// This is unconditional, unlike `percent_encode()` which has an `AsciiSet` parameter.
174///
175/// # Examples
176///
177/// ```
178/// use percent_encoding::percent_encode_byte;
179///
180/// assert_eq!("foo bar".bytes().map(percent_encode_byte).collect::<String>(),
181///            "%66%6F%6F%20%62%61%72");
182/// ```
183pub fn percent_encode_byte(byte: u8) -> &'static str {
184    let index = usize::from(byte) * 3;
185    &"\
186      %00%01%02%03%04%05%06%07%08%09%0A%0B%0C%0D%0E%0F\
187      %10%11%12%13%14%15%16%17%18%19%1A%1B%1C%1D%1E%1F\
188      %20%21%22%23%24%25%26%27%28%29%2A%2B%2C%2D%2E%2F\
189      %30%31%32%33%34%35%36%37%38%39%3A%3B%3C%3D%3E%3F\
190      %40%41%42%43%44%45%46%47%48%49%4A%4B%4C%4D%4E%4F\
191      %50%51%52%53%54%55%56%57%58%59%5A%5B%5C%5D%5E%5F\
192      %60%61%62%63%64%65%66%67%68%69%6A%6B%6C%6D%6E%6F\
193      %70%71%72%73%74%75%76%77%78%79%7A%7B%7C%7D%7E%7F\
194      %80%81%82%83%84%85%86%87%88%89%8A%8B%8C%8D%8E%8F\
195      %90%91%92%93%94%95%96%97%98%99%9A%9B%9C%9D%9E%9F\
196      %A0%A1%A2%A3%A4%A5%A6%A7%A8%A9%AA%AB%AC%AD%AE%AF\
197      %B0%B1%B2%B3%B4%B5%B6%B7%B8%B9%BA%BB%BC%BD%BE%BF\
198      %C0%C1%C2%C3%C4%C5%C6%C7%C8%C9%CA%CB%CC%CD%CE%CF\
199      %D0%D1%D2%D3%D4%D5%D6%D7%D8%D9%DA%DB%DC%DD%DE%DF\
200      %E0%E1%E2%E3%E4%E5%E6%E7%E8%E9%EA%EB%EC%ED%EE%EF\
201      %F0%F1%F2%F3%F4%F5%F6%F7%F8%F9%FA%FB%FC%FD%FE%FF\
202      "[index..index + 3]
203}
204
205/// Percent-encode the given bytes with the given set.
206///
207/// Non-ASCII bytes and bytes in `ascii_set` are encoded.
208///
209/// The return type:
210///
211/// * Implements `Iterator<Item = &str>` and therefore has a `.collect::<String>()` method,
212/// * Implements `Display` and therefore has a `.to_string()` method,
213/// * Implements `Into<Cow<str>>` borrowing `input` when none of its bytes are encoded.
214///
215/// # Examples
216///
217/// ```
218/// use percent_encoding::{percent_encode, NON_ALPHANUMERIC};
219///
220/// assert_eq!(percent_encode(b"foo bar?", NON_ALPHANUMERIC).to_string(), "foo%20bar%3F");
221/// ```
222#[inline]
223pub fn percent_encode<'a>(input: &'a [u8], ascii_set: &'static AsciiSet) -> PercentEncode<'a> {
224    PercentEncode {
225        bytes: input,
226        ascii_set,
227    }
228}
229
230/// Percent-encode the UTF-8 encoding of the given string.
231///
232/// See [`percent_encode`] regarding the return type.
233///
234/// # Examples
235///
236/// ```
237/// use percent_encoding::{utf8_percent_encode, NON_ALPHANUMERIC};
238///
239/// assert_eq!(utf8_percent_encode("foo bar?", NON_ALPHANUMERIC).to_string(), "foo%20bar%3F");
240/// ```
241#[inline]
242pub fn utf8_percent_encode<'a>(input: &'a str, ascii_set: &'static AsciiSet) -> PercentEncode<'a> {
243    percent_encode(input.as_bytes(), ascii_set)
244}
245
246/// The return type of [`percent_encode`] and [`utf8_percent_encode`].
247#[derive(Clone)]
248pub struct PercentEncode<'a> {
249    bytes: &'a [u8],
250    ascii_set: &'static AsciiSet,
251}
252
253impl<'a> Iterator for PercentEncode<'a> {
254    type Item = &'a str;
255
256    fn next(&mut self) -> Option<&'a str> {
257        if let Some((&first_byte, remaining)) = self.bytes.split_first() {
258            if self.ascii_set.should_percent_encode(first_byte) {
259                self.bytes = remaining;
260                Some(percent_encode_byte(first_byte))
261            } else {
262                // The unsafe blocks here are appropriate because the bytes are
263                // confirmed as a subset of UTF-8 in should_percent_encode.
264                for (i, &byte) in remaining.iter().enumerate() {
265                    if self.ascii_set.should_percent_encode(byte) {
266                        // 1 for first_byte + i for previous iterations of this loop
267                        let (unchanged_slice, remaining) = self.bytes.split_at(1 + i);
268                        self.bytes = remaining;
269                        return Some(unsafe { str::from_utf8_unchecked(unchanged_slice) });
270                    }
271                }
272                let unchanged_slice = self.bytes;
273                self.bytes = &[][..];
274                Some(unsafe { str::from_utf8_unchecked(unchanged_slice) })
275            }
276        } else {
277            None
278        }
279    }
280
281    fn size_hint(&self) -> (usize, Option<usize>) {
282        if self.bytes.is_empty() {
283            (0, Some(0))
284        } else {
285            (1, Some(self.bytes.len()))
286        }
287    }
288}
289
290impl<'a> fmt::Display for PercentEncode<'a> {
291    fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result {
292        for c in (*self).clone() {
293            formatter.write_str(c)?
294        }
295        Ok(())
296    }
297}
298
299#[cfg(feature = "alloc")]
300impl<'a> From<PercentEncode<'a>> for Cow<'a, str> {
301    fn from(mut iter: PercentEncode<'a>) -> Self {
302        match iter.next() {
303            None => "".into(),
304            Some(first) => match iter.next() {
305                None => first.into(),
306                Some(second) => {
307                    let mut string = first.to_owned();
308                    string.push_str(second);
309                    string.extend(iter);
310                    string.into()
311                }
312            },
313        }
314    }
315}
316
317/// Percent-decode the given string.
318///
319/// <https://url.spec.whatwg.org/#string-percent-decode>
320///
321/// See [`percent_decode`] regarding the return type.
322#[inline]
323pub fn percent_decode_str(input: &str) -> PercentDecode<'_> {
324    percent_decode(input.as_bytes())
325}
326
327/// Percent-decode the given bytes.
328///
329/// <https://url.spec.whatwg.org/#percent-decode>
330///
331/// Any sequence of `%` followed by two hexadecimal digits is decoded.
332/// The return type:
333///
334/// * Implements `Into<Cow<u8>>` borrowing `input` when it contains no percent-encoded sequence,
335/// * Implements `Iterator<Item = u8>` and therefore has a `.collect::<Vec<u8>>()` method,
336/// * Has `decode_utf8()` and `decode_utf8_lossy()` methods.
337///
338/// # Examples
339///
340/// ```
341/// use percent_encoding::percent_decode;
342///
343/// assert_eq!(percent_decode(b"foo%20bar%3f").decode_utf8().unwrap(), "foo bar?");
344/// ```
345#[inline]
346pub fn percent_decode(input: &[u8]) -> PercentDecode<'_> {
347    PercentDecode {
348        bytes: input.iter(),
349    }
350}
351
352/// The return type of [`percent_decode`].
353#[derive(Clone, Debug)]
354pub struct PercentDecode<'a> {
355    bytes: slice::Iter<'a, u8>,
356}
357
358fn after_percent_sign(iter: &mut slice::Iter<'_, u8>) -> Option<u8> {
359    let mut cloned_iter = iter.clone();
360    let h = char::from(*cloned_iter.next()?).to_digit(16)?;
361    let l = char::from(*cloned_iter.next()?).to_digit(16)?;
362    *iter = cloned_iter;
363    Some(h as u8 * 0x10 + l as u8)
364}
365
366impl<'a> Iterator for PercentDecode<'a> {
367    type Item = u8;
368
369    fn next(&mut self) -> Option<u8> {
370        self.bytes.next().map(|&byte| {
371            if byte == b'%' {
372                after_percent_sign(&mut self.bytes).unwrap_or(byte)
373            } else {
374                byte
375            }
376        })
377    }
378
379    fn size_hint(&self) -> (usize, Option<usize>) {
380        let bytes = self.bytes.len();
381        ((bytes + 2) / 3, Some(bytes))
382    }
383}
384
385#[cfg(feature = "alloc")]
386impl<'a> From<PercentDecode<'a>> for Cow<'a, [u8]> {
387    fn from(iter: PercentDecode<'a>) -> Self {
388        match iter.if_any() {
389            Some(vec) => Cow::Owned(vec),
390            None => Cow::Borrowed(iter.bytes.as_slice()),
391        }
392    }
393}
394
395impl<'a> PercentDecode<'a> {
396    /// If the percent-decoding is different from the input, return it as a new bytes vector.
397    #[cfg(feature = "alloc")]
398    fn if_any(&self) -> Option<Vec<u8>> {
399        let mut bytes_iter = self.bytes.clone();
400        while bytes_iter.any(|&b| b == b'%') {
401            if let Some(decoded_byte) = after_percent_sign(&mut bytes_iter) {
402                let initial_bytes = self.bytes.as_slice();
403                let unchanged_bytes_len = initial_bytes.len() - bytes_iter.len() - 3;
404                let mut decoded = initial_bytes[..unchanged_bytes_len].to_owned();
405                decoded.push(decoded_byte);
406                decoded.extend(PercentDecode { bytes: bytes_iter });
407                return Some(decoded);
408            }
409        }
410        // Nothing to decode
411        None
412    }
413
414    /// Decode the result of percent-decoding as UTF-8.
415    ///
416    /// This is return `Err` when the percent-decoded bytes are not well-formed in UTF-8.
417    #[cfg(feature = "alloc")]
418    pub fn decode_utf8(self) -> Result<Cow<'a, str>, str::Utf8Error> {
419        match self.clone().into() {
420            Cow::Borrowed(bytes) => match str::from_utf8(bytes) {
421                Ok(s) => Ok(s.into()),
422                Err(e) => Err(e),
423            },
424            Cow::Owned(bytes) => match String::from_utf8(bytes) {
425                Ok(s) => Ok(s.into()),
426                Err(e) => Err(e.utf8_error()),
427            },
428        }
429    }
430
431    /// Decode the result of percent-decoding as UTF-8, lossily.
432    ///
433    /// Invalid UTF-8 percent-encoded byte sequences will be replaced � U+FFFD,
434    /// the replacement character.
435    #[cfg(feature = "alloc")]
436    pub fn decode_utf8_lossy(self) -> Cow<'a, str> {
437        decode_utf8_lossy(self.clone().into())
438    }
439}
440
441#[cfg(feature = "alloc")]
442fn decode_utf8_lossy(input: Cow<'_, [u8]>) -> Cow<'_, str> {
443    // Note: This function is duplicated in `form_urlencoded/src/query_encoding.rs`.
444    match input {
445        Cow::Borrowed(bytes) => String::from_utf8_lossy(bytes),
446        Cow::Owned(bytes) => {
447            match String::from_utf8_lossy(&bytes) {
448                Cow::Borrowed(utf8) => {
449                    // If from_utf8_lossy returns a Cow::Borrowed, then we can
450                    // be sure our original bytes were valid UTF-8. This is because
451                    // if the bytes were invalid UTF-8 from_utf8_lossy would have
452                    // to allocate a new owned string to back the Cow so it could
453                    // replace invalid bytes with a placeholder.
454
455                    // First we do a debug_assert to confirm our description above.
456                    let raw_utf8: *const [u8] = utf8.as_bytes();
457                    debug_assert!(raw_utf8 == &*bytes as *const [u8]);
458
459                    // Given we know the original input bytes are valid UTF-8,
460                    // and we have ownership of those bytes, we re-use them and
461                    // return a Cow::Owned here.
462                    Cow::Owned(unsafe { String::from_utf8_unchecked(bytes) })
463                }
464                Cow::Owned(s) => Cow::Owned(s),
465            }
466        }
467    }
468}
percent_encoding/lib.rs

percent_encoding/
lib.rs
