iri_string/
convert.rs

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
//! Conversion between URI/IRI types.

use core::fmt;

#[cfg(feature = "alloc")]
use alloc::collections::TryReserveError;
#[cfg(all(feature = "alloc", not(feature = "std")))]
use alloc::string::String;

#[cfg(feature = "alloc")]
use crate::format::{ToDedicatedString, ToStringFallible};
use crate::spec::Spec;
use crate::types::{
    RiAbsoluteStr, RiFragmentStr, RiQueryStr, RiReferenceStr, RiRelativeStr, RiStr,
};
#[cfg(feature = "alloc")]
use crate::types::{
    RiAbsoluteString, RiFragmentString, RiQueryString, RiReferenceString, RiRelativeString,
    RiString,
};
#[cfg(feature = "alloc")]
use crate::types::{
    UriAbsoluteString, UriFragmentString, UriQueryString, UriReferenceString, UriRelativeString,
    UriString,
};

/// Hexadecimal digits for a nibble.
const HEXDIGITS: [u8; 16] = [
    b'0', b'1', b'2', b'3', b'4', b'5', b'6', b'7', b'8', b'9', b'A', b'B', b'C', b'D', b'E', b'F',
];

/// A resource identifier mapped to a URI of some kind.
///
/// Supported `Src` type are:
///
/// * IRIs:
///     + [`IriAbsoluteStr`] (alias of `RiAbsoluteStr<IriSpec>`)
///     + [`IriReferenceStr`] (alias of `RiReferenceStr<IriSpec>`)
///     + [`IriRelativeStr`] (alias of `RiRelativeStr<IriSpec>`)
///     + [`IriStr`] (alias of `RiStr<IriSpec>`)
/// * URIs:
///     + [`UriAbsoluteStr`] (alias of `RiAbsoluteStr<UriSpec>`)
///     + [`UriReferenceStr`] (alias of `RiReferenceStr<UriSpec>`)
///     + [`UriRelativeStr`] (alias of `RiRelativeStr<UriSpec>`)
///     + [`UriStr`] (alias of `RiStr<UriSpec>`)
///
/// # Examples
///
/// ```
/// use iri_string::convert::MappedToUri;
/// use iri_string::types::{IriStr, UriStr};
///
/// let src = IriStr::new("http://example.com/?alpha=\u{03B1}")?;
/// // The type is `MappedToUri<IriStr>`, but you usually don't need to specify.
/// let mapped = MappedToUri::from(src).to_string();
/// assert_eq!(mapped, "http://example.com/?alpha=%CE%B1");
/// # Ok::<_, iri_string::validate::Error>(())
/// ```
///
/// [`IriAbsoluteStr`]: crate::types::IriAbsoluteStr
/// [`IriReferenceStr`]: crate::types::IriReferenceStr
/// [`IriRelativeStr`]: crate::types::IriRelativeStr
/// [`IriStr`]: crate::types::IriStr
/// [`UriAbsoluteStr`]: crate::types::UriAbsoluteStr
/// [`UriReferenceStr`]: crate::types::UriReferenceStr
/// [`UriRelativeStr`]: crate::types::UriRelativeStr
/// [`UriStr`]: crate::types::UriStr
#[derive(Debug, Clone, Copy)]
pub struct MappedToUri<'a, Src: ?Sized>(&'a Src);

/// Implement conversions for an IRI string type.
macro_rules! impl_for_iri {
    ($borrowed:ident, $owned:ident, $owned_uri:ident) => {
        impl<S: Spec> fmt::Display for MappedToUri<'_, $borrowed<S>> {
            fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
                write_percent_encoded(f, self.0.as_str())
            }
        }

        #[cfg(feature = "alloc")]
        impl<S: Spec> ToDedicatedString for MappedToUri<'_, $borrowed<S>> {
            type Target = $owned_uri;

            fn try_to_dedicated_string(&self) -> Result<Self::Target, TryReserveError> {
                let s = self.try_to_string()?;
                Ok(TryFrom::try_from(s)
                    .expect("[validity] the IRI must be encoded into a valid URI"))
            }
        }

        impl<'a, S: Spec> From<&'a $borrowed<S>> for MappedToUri<'a, $borrowed<S>> {
            #[inline]
            fn from(iri: &'a $borrowed<S>) -> Self {
                Self(iri)
            }
        }

        #[cfg(feature = "alloc")]
        impl<'a, S: Spec> From<&'a $owned<S>> for MappedToUri<'a, $borrowed<S>> {
            #[inline]
            fn from(iri: &'a $owned<S>) -> Self {
                Self(iri.as_slice())
            }
        }
    };
}

impl_for_iri!(RiReferenceStr, RiReferenceString, UriReferenceString);
impl_for_iri!(RiStr, RiString, UriString);
impl_for_iri!(RiAbsoluteStr, RiAbsoluteString, UriAbsoluteString);
impl_for_iri!(RiRelativeStr, RiRelativeString, UriRelativeString);
impl_for_iri!(RiQueryStr, RiQueryString, UriQueryString);
impl_for_iri!(RiFragmentStr, RiFragmentString, UriFragmentString);

/// Percent-encodes and writes the IRI string using the given buffer.
fn write_percent_encoded(f: &mut fmt::Formatter<'_>, mut s: &str) -> fmt::Result {
    while !s.is_empty() {
        // Skip ASCII characters.
        let non_ascii_pos = s.bytes().position(|b| !b.is_ascii()).unwrap_or(s.len());
        let (ascii, rest) = s.split_at(non_ascii_pos);
        if !ascii.is_empty() {
            f.write_str(ascii)?;
            s = rest;
        }

        if s.is_empty() {
            return Ok(());
        }

        // Search for the next ASCII character.
        let nonascii_end = s.bytes().position(|b| b.is_ascii()).unwrap_or(s.len());
        let (nonasciis, rest) = s.split_at(nonascii_end);
        debug_assert!(
            !nonasciis.is_empty(),
            "string without non-ASCII characters should have caused early return"
        );
        s = rest;

        // Escape non-ASCII characters as percent-encoded bytes.
        //
        // RFC 3987 (section 3.1 step 2) says "for each character in
        // 'ucschar' or 'iprivate'", but this simply means "for each
        // non-ASCII characters" since any non-ASCII characters that can
        // appear in an IRI match `ucschar` or `iprivate`.
        /// Number of source bytes to encode at once.
        const NUM_BYTES_AT_ONCE: usize = 21;
        percent_encode_bytes(f, nonasciis, &mut [0_u8; NUM_BYTES_AT_ONCE * 3])?;
    }

    Ok(())
}

/// Percent-encode the string and pass the encoded chunks to the given function.
///
/// `buf` is used as a temporary working buffer. It is initialized by this
/// function, so users can pass any mutable byte slice with enough size.
///
/// # Precondition
///
/// The length of `buf` must be 3 bytes or more.
fn percent_encode_bytes(f: &mut fmt::Formatter<'_>, s: &str, buf: &mut [u8]) -> fmt::Result {
    /// Fill the buffer by percent-encoded bytes.
    ///
    /// Note that this function applies percent-encoding to every characters,
    /// even if it is ASCII alphabet.
    ///
    /// # Precondition
    ///
    /// * The length of `buf` must be 3 bytes or more.
    /// * All of the `buf[i * 3]` elements should already be set to `b'%'`.
    // This function have many preconditions and I don't want checks for them
    // to be mandatory, so make this nested inner function.
    fn fill_by_percent_encoded<'a>(buf: &'a mut [u8], bytes: &mut core::str::Bytes<'_>) -> &'a str {
        let src_len = bytes.len();
        // `<[u8; N]>::array_chunks_mut` is unstable as of Rust 1.58.1.
        for (dest, byte) in buf.chunks_exact_mut(3).zip(bytes.by_ref()) {
            debug_assert_eq!(
                dest.len(),
                3,
                "[validity] `chunks_exact()` must return a slice with the exact length"
            );
            debug_assert_eq!(
                dest[0], b'%',
                "[precondition] the buffer must be properly initialized"
            );

            let upper = byte >> 4;
            let lower = byte & 0b1111;
            dest[1] = HEXDIGITS[usize::from(upper)];
            dest[2] = HEXDIGITS[usize::from(lower)];
        }
        let num_dest_written = (src_len - bytes.len()) * 3;
        let buf_filled = &buf[..num_dest_written];
        // SAFETY: `b'%'` and `HEXDIGITS[_]` are all ASCII characters, so
        // `buf_filled` is filled with ASCII characters and is valid UTF-8 bytes.
        unsafe {
            debug_assert!(core::str::from_utf8(buf_filled).is_ok());
            core::str::from_utf8_unchecked(buf_filled)
        }
    }

    assert!(
        buf.len() >= 3,
        "[precondition] length of `buf` must be 3 bytes or more"
    );

    // Drop the elements that will never be used.
    // The length to be used is always a multiple of three.
    let buf_len = buf.len() / 3 * 3;
    let buf = &mut buf[..buf_len];

    // Fill some bytes with `%`.
    // This will be vectorized by optimization (especially for long buffers),
    // so no need to selectively set `buf[i * 3]`.
    buf.fill(b'%');

    let mut bytes = s.bytes();
    // `<core::str::Bytes as ExactSizeIterator>::is_empty` is unstable as of Rust 1.58.1.
    while bytes.len() != 0 {
        let encoded = fill_by_percent_encoded(buf, &mut bytes);
        f.write_str(encoded)?;
    }

    Ok(())
}

/// Percent-encodes the given IRI using the given buffer.
#[cfg(feature = "alloc")]
pub(crate) fn try_percent_encode_iri_inline(
    iri: &mut String,
) -> Result<(), alloc::collections::TryReserveError> {
    // Calculate the result length and extend the buffer.
    let num_nonascii = count_nonascii(iri);
    if num_nonascii == 0 {
        // No need to escape.
        return Ok(());
    }
    let additional = num_nonascii * 2;
    iri.try_reserve(additional)?;
    let src_len = iri.len();

    // Temporarily take the ownership of the internal buffer.
    let mut buf = core::mem::take(iri).into_bytes();
    // `b'\0'` cannot appear in a valid IRI, so this default value would be
    // useful in case of debugging.
    buf.extend(core::iter::repeat(b'\0').take(additional));

    // Fill the buffer from the tail to the head.
    let mut dest_end = buf.len();
    let mut src_end = src_len;
    let mut rest_nonascii = num_nonascii;
    while rest_nonascii > 0 {
        debug_assert!(
            src_end > 0,
            "[validity] the source position should not overrun"
        );
        debug_assert!(
            dest_end > 0,
            "[validity] the destination position should not overrun"
        );
        src_end -= 1;
        dest_end -= 1;
        let byte = buf[src_end];
        if byte.is_ascii() {
            buf[dest_end] = byte;
            // Use the ASCII character directly.
        } else {
            // Percent-encode the byte.
            dest_end -= 2;
            buf[dest_end] = b'%';
            let upper = byte >> 4;
            let lower = byte & 0b1111;
            buf[dest_end + 1] = HEXDIGITS[usize::from(upper)];
            buf[dest_end + 2] = HEXDIGITS[usize::from(lower)];
            rest_nonascii -= 1;
        }
    }

    // Move the result from the temporary buffer to the destination.
    let s = String::from_utf8(buf).expect("[consistency] the encoding result is an ASCII string");
    *iri = s;
    Ok(())
}

/// Returns the number of non-ASCII characters.
#[cfg(feature = "alloc")]
#[inline]
#[must_use]
fn count_nonascii(s: &str) -> usize {
    s.bytes().filter(|b| !b.is_ascii()).count()
}