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
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
// Copyright 2015 Google Inc. All rights reserved.
//
// Permission is hereby granted, free of charge, to any person obtaining a copy
// of this software and associated documentation files (the "Software"), to deal
// in the Software without restriction, including without limitation the rights
// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
// copies of the Software, and to permit persons to whom the Software is
// furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included in
// all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
// THE SOFTWARE.

//! Utility functions for HTML escaping. Only useful when building your own
//! HTML renderer.

use std::fmt::{self, Arguments};
use std::io::{self, Write};
use std::str::from_utf8;

#[rustfmt::skip]
static HREF_SAFE: [u8; 128] = [
    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    0, 1, 0, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1,
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 1,
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1,
    0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0,
];

static HEX_CHARS: &[u8] = b"0123456789ABCDEF";
static AMP_ESCAPE: &str = "&";
static SINGLE_QUOTE_ESCAPE: &str = "'";

/// This wrapper exists because we can't have both a blanket implementation
/// for all types implementing `Write` and types of the for `&mut W` where
/// `W: StrWrite`. Since we need the latter a lot, we choose to wrap
/// `Write` types.
#[derive(Debug)]
pub struct IoWriter<W>(pub W);

/// Trait that allows writing string slices. This is basically an extension
/// of `std::io::Write` in order to include `String`.
pub trait StrWrite {
    type Error;

    fn write_str(&mut self, s: &str) -> Result<(), Self::Error>;
    fn write_fmt(&mut self, args: Arguments) -> Result<(), Self::Error>;
}

impl<W> StrWrite for IoWriter<W>
where
    W: Write,
{
    type Error = io::Error;

    #[inline]
    fn write_str(&mut self, s: &str) -> io::Result<()> {
        self.0.write_all(s.as_bytes())
    }

    #[inline]
    fn write_fmt(&mut self, args: Arguments) -> io::Result<()> {
        self.0.write_fmt(args)
    }
}

/// This wrapper exists because we can't have both a blanket implementation
/// for all types implementing `io::Write` and types of the form `&mut W` where
/// `W: StrWrite`. Since we need the latter a lot, we choose to wrap
/// `Write` types.
#[derive(Debug)]
pub struct FmtWriter<W>(pub W);

impl<W> StrWrite for FmtWriter<W>
where
    W: fmt::Write,
{
    type Error = fmt::Error;

    #[inline]
    fn write_str(&mut self, s: &str) -> fmt::Result {
        self.0.write_str(s)
    }

    #[inline]
    fn write_fmt(&mut self, args: Arguments) -> fmt::Result {
        self.0.write_fmt(args)
    }
}

impl StrWrite for String {
    type Error = fmt::Error;

    #[inline]
    fn write_str(&mut self, s: &str) -> fmt::Result {
        self.push_str(s);
        Ok(())
    }

    #[inline]
    fn write_fmt(&mut self, args: Arguments) -> fmt::Result {
        fmt::Write::write_fmt(self, args)
    }
}

impl<W> StrWrite for &'_ mut W
where
    W: StrWrite,
{
    type Error = W::Error;

    #[inline]
    fn write_str(&mut self, s: &str) -> Result<(), Self::Error> {
        (**self).write_str(s)
    }

    #[inline]
    fn write_fmt(&mut self, args: Arguments) -> Result<(), Self::Error> {
        (**self).write_fmt(args)
    }
}

/// Writes an href to the buffer, escaping href unsafe bytes.
pub fn escape_href<W>(mut w: W, s: &str) -> Result<(), W::Error>
where
    W: StrWrite,
{
    let bytes = s.as_bytes();
    let mut mark = 0;
    for i in 0..bytes.len() {
        let c = bytes[i];
        if c >= 0x80 || HREF_SAFE[c as usize] == 0 {
            // character needing escape

            // write partial substring up to mark
            if mark < i {
                w.write_str(&s[mark..i])?;
            }
            match c {
                b'&' => {
                    w.write_str(AMP_ESCAPE)?;
                }
                b'\'' => {
                    w.write_str(SINGLE_QUOTE_ESCAPE)?;
                }
                _ => {
                    let mut buf = [0u8; 3];
                    buf[0] = b'%';
                    buf[1] = HEX_CHARS[((c as usize) >> 4) & 0xF];
                    buf[2] = HEX_CHARS[(c as usize) & 0xF];
                    let escaped = from_utf8(&buf).unwrap();
                    w.write_str(escaped)?;
                }
            }
            mark = i + 1; // all escaped characters are ASCII
        }
    }
    w.write_str(&s[mark..])
}

const fn create_html_escape_table(body: bool) -> [u8; 256] {
    let mut table = [0; 256];
    table[b'&' as usize] = 1;
    table[b'<' as usize] = 2;
    table[b'>' as usize] = 3;
    if !body {
        table[b'"' as usize] = 4;
        table[b'\'' as usize] = 5;
    }
    table
}

static HTML_ESCAPE_TABLE: [u8; 256] = create_html_escape_table(false);
static HTML_BODY_TEXT_ESCAPE_TABLE: [u8; 256] = create_html_escape_table(true);

static HTML_ESCAPES: [&str; 6] = ["", "&amp;", "&lt;", "&gt;", "&quot;", "&#39;"];

/// Writes the given string to the Write sink, replacing special HTML bytes
/// (<, >, &, ", ') by escape sequences.
///
/// Use this function to write output to quoted HTML attributes.
/// Since this function doesn't escape spaces, unquoted attributes
/// cannot be used. For example:
///
/// ```rust
/// let mut value = String::new();
/// pulldown_cmark_escape::escape_html(&mut value, "two words")
///     .expect("writing to a string is infallible");
/// // This is okay.
/// let ok = format!("<a title='{value}'>test</a>");
/// // This is not okay.
/// //let not_ok = format!("<a title={value}>test</a>");
/// ````
pub fn escape_html<W: StrWrite>(w: W, s: &str) -> Result<(), W::Error> {
    #[cfg(all(target_arch = "x86_64", feature = "simd"))]
    {
        simd::escape_html(w, s, &HTML_ESCAPE_TABLE)
    }
    #[cfg(not(all(target_arch = "x86_64", feature = "simd")))]
    {
        escape_html_scalar(w, s, &HTML_ESCAPE_TABLE)
    }
}

/// For use in HTML body text, writes the given string to the Write sink,
/// replacing special HTML bytes (<, >, &) by escape sequences.
///
/// <div class="warning">
///
/// This function should be used for escaping text nodes, not attributes.
/// In the below example, the word "foo" is an attribute, and the word
/// "bar" is an text node. The word "bar" could be escaped by this function,
/// but the word "foo" must be escaped using [`escape_html`].
///
/// ```html
/// <span class="foo">bar</span>
/// ```
///
/// If you aren't sure what the difference is, use [`escape_html`].
/// It should always be correct, but will produce larger output.
///
/// </div>
pub fn escape_html_body_text<W: StrWrite>(w: W, s: &str) -> Result<(), W::Error> {
    #[cfg(all(target_arch = "x86_64", feature = "simd"))]
    {
        simd::escape_html(w, s, &HTML_BODY_TEXT_ESCAPE_TABLE)
    }
    #[cfg(not(all(target_arch = "x86_64", feature = "simd")))]
    {
        escape_html_scalar(w, s, &HTML_BODY_TEXT_ESCAPE_TABLE)
    }
}

fn escape_html_scalar<W: StrWrite>(
    mut w: W,
    s: &str,
    table: &'static [u8; 256],
) -> Result<(), W::Error> {
    let bytes = s.as_bytes();
    let mut mark = 0;
    let mut i = 0;
    while i < s.len() {
        match bytes[i..].iter().position(|&c| table[c as usize] != 0) {
            Some(pos) => {
                i += pos;
            }
            None => break,
        }
        let c = bytes[i];
        let escape = table[c as usize];
        let escape_seq = HTML_ESCAPES[escape as usize];
        w.write_str(&s[mark..i])?;
        w.write_str(escape_seq)?;
        i += 1;
        mark = i; // all escaped characters are ASCII
    }
    w.write_str(&s[mark..])
}

#[cfg(all(target_arch = "x86_64", feature = "simd"))]
mod simd {
    use super::StrWrite;
    use std::arch::x86_64::*;
    use std::mem::size_of;

    const VECTOR_SIZE: usize = size_of::<__m128i>();

    pub(super) fn escape_html<W: StrWrite>(
        mut w: W,
        s: &str,
        table: &'static [u8; 256],
    ) -> Result<(), W::Error> {
        // The SIMD accelerated code uses the PSHUFB instruction, which is part
        // of the SSSE3 instruction set. Further, we can only use this code if
        // the buffer is at least one VECTOR_SIZE in length to prevent reading
        // out of bounds. If either of these conditions is not met, we fall back
        // to scalar code.
        if is_x86_feature_detected!("ssse3") && s.len() >= VECTOR_SIZE {
            let bytes = s.as_bytes();
            let mut mark = 0;

            unsafe {
                foreach_special_simd(bytes, 0, |i| {
                    let escape_ix = *bytes.get_unchecked(i) as usize;
                    let entry = table[escape_ix] as usize;
                    w.write_str(s.get_unchecked(mark..i))?;
                    mark = i + 1; // all escaped characters are ASCII
                    if entry == 0 {
                        w.write_str(s.get_unchecked(i..mark))
                    } else {
                        let replacement = super::HTML_ESCAPES[entry];
                        w.write_str(replacement)
                    }
                })?;
                w.write_str(s.get_unchecked(mark..))
            }
        } else {
            super::escape_html_scalar(w, s, table)
        }
    }

    /// Creates the lookup table for use in `compute_mask`.
    const fn create_lookup() -> [u8; 16] {
        let mut table = [0; 16];
        table[(b'<' & 0x0f) as usize] = b'<';
        table[(b'>' & 0x0f) as usize] = b'>';
        table[(b'&' & 0x0f) as usize] = b'&';
        table[(b'"' & 0x0f) as usize] = b'"';
        table[(b'\'' & 0x0f) as usize] = b'\'';
        table[0] = 0b0111_1111;
        table
    }

    #[target_feature(enable = "ssse3")]
    /// Computes a byte mask at given offset in the byte buffer. Its first 16 (least significant)
    /// bits correspond to whether there is an HTML special byte (&, <, ", >) at the 16 bytes
    /// `bytes[offset..]`. For example, the mask `(1 << 3)` states that there is an HTML byte
    /// at `offset + 3`. It is only safe to call this function when
    /// `bytes.len() >= offset + VECTOR_SIZE`.
    unsafe fn compute_mask(bytes: &[u8], offset: usize) -> i32 {
        debug_assert!(bytes.len() >= offset + VECTOR_SIZE);

        let table = create_lookup();
        let lookup = _mm_loadu_si128(table.as_ptr() as *const __m128i);
        let raw_ptr = bytes.as_ptr().add(offset) as *const __m128i;

        // Load the vector from memory.
        let vector = _mm_loadu_si128(raw_ptr);
        // We take the least significant 4 bits of every byte and use them as indices
        // to map into the lookup vector.
        // Note that shuffle maps bytes with their most significant bit set to lookup[0].
        // Bytes that share their lower nibble with an HTML special byte get mapped to that
        // corresponding special byte. Note that all HTML special bytes have distinct lower
        // nibbles. Other bytes either get mapped to 0 or 127.
        let expected = _mm_shuffle_epi8(lookup, vector);
        // We compare the original vector to the mapped output. Bytes that shared a lower
        // nibble with an HTML special byte match *only* if they are that special byte. Bytes
        // that have either a 0 lower nibble or their most significant bit set were mapped to
        // 127 and will hence never match. All other bytes have non-zero lower nibbles but
        // were mapped to 0 and will therefore also not match.
        let matches = _mm_cmpeq_epi8(expected, vector);

        // Translate matches to a bitmask, where every 1 corresponds to a HTML special character
        // and a 0 is a non-HTML byte.
        _mm_movemask_epi8(matches)
    }

    /// Calls the given function with the index of every byte in the given byteslice
    /// that is either ", &, <, or > and for no other byte.
    /// Make sure to only call this when `bytes.len() >= 16`, undefined behaviour may
    /// occur otherwise.
    #[target_feature(enable = "ssse3")]
    unsafe fn foreach_special_simd<E, F>(
        bytes: &[u8],
        mut offset: usize,
        mut callback: F,
    ) -> Result<(), E>
    where
        F: FnMut(usize) -> Result<(), E>,
    {
        // The strategy here is to walk the byte buffer in chunks of VECTOR_SIZE (16)
        // bytes at a time starting at the given offset. For each chunk, we compute a
        // a bitmask indicating whether the corresponding byte is a HTML special byte.
        // We then iterate over all the 1 bits in this mask and call the callback function
        // with the corresponding index in the buffer.
        // When the number of HTML special bytes in the buffer is relatively low, this
        // allows us to quickly go through the buffer without a lookup and for every
        // single byte.

        debug_assert!(bytes.len() >= VECTOR_SIZE);
        let upperbound = bytes.len() - VECTOR_SIZE;
        while offset < upperbound {
            let mut mask = compute_mask(bytes, offset);
            while mask != 0 {
                let ix = mask.trailing_zeros();
                callback(offset + ix as usize)?;
                mask ^= mask & -mask;
            }
            offset += VECTOR_SIZE;
        }

        // Final iteration. We align the read with the end of the slice and
        // shift off the bytes at start we have already scanned.
        let mut mask = compute_mask(bytes, upperbound);
        mask >>= offset - upperbound;
        while mask != 0 {
            let ix = mask.trailing_zeros();
            callback(offset + ix as usize)?;
            mask ^= mask & -mask;
        }
        Ok(())
    }

    #[cfg(test)]
    mod html_scan_tests {
        #[test]
        fn multichunk() {
            let mut vec = Vec::new();
            unsafe {
                super::foreach_special_simd("&aXaaaa.a'aa9a<>aab&".as_bytes(), 0, |ix| {
                    #[allow(clippy::unit_arg)]
                    Ok::<_, std::fmt::Error>(vec.push(ix))
                })
                .unwrap();
            }
            assert_eq!(vec, vec![0, 9, 14, 15, 19]);
        }

        // only match these bytes, and when we match them, match them VECTOR_SIZE times
        #[test]
        fn only_right_bytes_matched() {
            for b in 0..255u8 {
                let right_byte = b == b'&' || b == b'<' || b == b'>' || b == b'"' || b == b'\'';
                let vek = vec![b; super::VECTOR_SIZE];
                let mut match_count = 0;
                unsafe {
                    super::foreach_special_simd(&vek, 0, |_| {
                        match_count += 1;
                        Ok::<_, std::fmt::Error>(())
                    })
                    .unwrap();
                }
                assert!((match_count > 0) == (match_count == super::VECTOR_SIZE));
                assert_eq!(
                    (match_count == super::VECTOR_SIZE),
                    right_byte,
                    "match_count: {}, byte: {:?}",
                    match_count,
                    b as char
                );
            }
        }
    }
}

#[cfg(test)]
mod test {
    pub use super::{escape_href, escape_html, escape_html_body_text};

    #[test]
    fn check_href_escape() {
        let mut s = String::new();
        escape_href(&mut s, "&^_").unwrap();
        assert_eq!(s.as_str(), "&amp;^_");
    }

    #[test]
    fn check_attr_escape() {
        let mut s = String::new();
        escape_html(&mut s, r##"&^"'_"##).unwrap();
        assert_eq!(s.as_str(), "&amp;^&quot;&#39;_");
    }

    #[test]
    fn check_body_escape() {
        let mut s = String::new();
        escape_html_body_text(&mut s, r##"&^"'_"##).unwrap();
        assert_eq!(s.as_str(), r##"&amp;^"'_"##);
    }
}