Skip to content

Commit 2bb101b

Browse files
author
Daniel Lemire
committed
Code reformatting.
1 parent 26baf70 commit 2bb101b

2 files changed

Lines changed: 24 additions & 64 deletions

File tree

‎src/fallback/dom_parser_implementation.cpp‎

Lines changed: 14 additions & 37 deletions
Original file line numberDiff line numberDiff line change
@@ -243,17 +243,17 @@ WARN_UNUSED error_code implementation::minify(const uint8_t *buf, size_t len, ui
243243
// for fear of aliasing
244244
return SUCCESS;
245245
}
246+
247+
// credit: based on code from Google Fuchsia (Apache Licensed)
246248
WARN_UNUSED bool implementation::validate_utf8(const char *buf, size_t len) const noexcept {
247249
const uint8_t *data = (const uint8_t *)buf;
248250
uint64_t pos = 0;
249251
uint64_t next_pos = 0;
250252
uint32_t code_point = 0;
251253
while (pos < len) {
252-
253254
// check of the next 8 bytes are ascii.
254255
next_pos = pos + 16;
255-
if (next_pos <=
256-
len) { // if it is safe to read 8 more bytes, check that they are ascii
256+
if (next_pos <= len) { // if it is safe to read 8 more bytes, check that they are ascii
257257
uint64_t v1;
258258
memcpy(&v1, data + pos, sizeof(uint64_t));
259259
uint64_t v2;
@@ -265,34 +265,21 @@ WARN_UNUSED bool implementation::validate_utf8(const char *buf, size_t len) cons
265265
}
266266
}
267267
unsigned char byte = data[pos];
268-
269268
if (byte < 0b10000000) {
270269
pos++;
271270
continue;
272271
} else if ((byte & 0b11100000) == 0b11000000) {
273272
next_pos = pos + 2;
274-
if (next_pos > len) {
275-
return false;
276-
}
277-
if ((data[pos + 1] & 0b11000000) != 0b10000000) {
278-
return false;
279-
}
273+
if (next_pos > len) { return false; }
274+
if ((data[pos + 1] & 0b11000000) != 0b10000000) { return false; }
280275
// range check
281276
code_point = (byte & 0b00011111) << 6 | (data[pos + 1] & 0b00111111);
282-
if (code_point < 0x80 || 0x7ff < code_point) {
283-
return false;
284-
}
277+
if (code_point < 0x80 || 0x7ff < code_point) { return false; }
285278
} else if ((byte & 0b11110000) == 0b11100000) {
286279
next_pos = pos + 3;
287-
if (next_pos > len) {
288-
return false;
289-
}
290-
if ((data[pos + 1] & 0b11000000) != 0b10000000) {
291-
return false;
292-
}
293-
if ((data[pos + 2] & 0b11000000) != 0b10000000) {
294-
return false;
295-
}
280+
if (next_pos > len) { return false; }
281+
if ((data[pos + 1] & 0b11000000) != 0b10000000) { return false; }
282+
if ((data[pos + 2] & 0b11000000) != 0b10000000) { return false; }
296283
// range check
297284
code_point = (byte & 0b00001111) << 12 |
298285
(data[pos + 1] & 0b00111111) << 6 |
@@ -303,25 +290,15 @@ WARN_UNUSED bool implementation::validate_utf8(const char *buf, size_t len) cons
303290
}
304291
} else if ((byte & 0b11111000) == 0b11110000) { // 0b11110000
305292
next_pos = pos + 4;
306-
if (next_pos > len) {
307-
return false;
308-
}
309-
if ((data[pos + 1] & 0b11000000) != 0b10000000) {
310-
return false;
311-
}
312-
if ((data[pos + 2] & 0b11000000) != 0b10000000) {
313-
return false;
314-
}
315-
if ((data[pos + 3] & 0b11000000) != 0b10000000) {
316-
return false;
317-
}
293+
if (next_pos > len) { return false; }
294+
if ((data[pos + 1] & 0b11000000) != 0b10000000) { return false; }
295+
if ((data[pos + 2] & 0b11000000) != 0b10000000) { return false; }
296+
if ((data[pos + 3] & 0b11000000) != 0b10000000) { return false; }
318297
// range check
319298
code_point =
320299
(byte & 0b00000111) << 18 | (data[pos + 1] & 0b00111111) << 12 |
321300
(data[pos + 2] & 0b00111111) << 6 | (data[pos + 3] & 0b00111111);
322-
if (code_point < 0xffff || 0x10ffff < code_point) {
323-
return false;
324-
}
301+
if (code_point < 0xffff || 0x10ffff < code_point) { return false; }
325302
} else {
326303
// we may have a continuation
327304
return false;

‎tests/unicode_tests.cpp‎

Lines changed: 10 additions & 27 deletions
Original file line numberDiff line numberDiff line change
@@ -98,6 +98,7 @@ std::vector<uint8_t> RandomUTF8::generate(size_t output_bytes, long seed) {
9898
return generate(output_bytes);
9999
}
100100

101+
// credit: based on code from Google Fuchsia (Apache Licensed)
101102
WARN_UNUSED bool basic_validate_utf8(const char *buf, size_t len) noexcept {
102103
const uint8_t *data = (const uint8_t *)buf;
103104
uint64_t pos = 0;
@@ -110,28 +111,18 @@ WARN_UNUSED bool basic_validate_utf8(const char *buf, size_t len) noexcept {
110111
continue;
111112
} else if ((byte & 0b11100000) == 0b11000000) {
112113
next_pos = pos + 2;
113-
if (next_pos > len) {
114-
return false;
115-
}
114+
if (next_pos > len) { return false; }
116115
if ((data[pos + 1] & 0b11000000) != 0b10000000) {
117116
return false;
118117
}
119118
// range check
120119
code_point = (byte & 0b00011111) << 6 | (data[pos + 1] & 0b00111111);
121-
if (code_point < 0x80 || 0x7ff < code_point) {
122-
return false;
123-
}
120+
if (code_point < 0x80 || 0x7ff < code_point) { return false; }
124121
} else if ((byte & 0b11110000) == 0b11100000) {
125122
next_pos = pos + 3;
126-
if (next_pos > len) {
127-
return false;
128-
}
129-
if ((data[pos + 1] & 0b11000000) != 0b10000000) {
130-
return false;
131-
}
132-
if ((data[pos + 2] & 0b11000000) != 0b10000000) {
133-
return false;
134-
}
123+
if (next_pos > len) { return false; }
124+
if ((data[pos + 1] & 0b11000000) != 0b10000000) { return false; }
125+
if ((data[pos + 2] & 0b11000000) != 0b10000000) { return false; }
135126
// range check
136127
code_point = (byte & 0b00001111) << 12 |
137128
(data[pos + 1] & 0b00111111) << 6 |
@@ -142,18 +133,10 @@ WARN_UNUSED bool basic_validate_utf8(const char *buf, size_t len) noexcept {
142133
}
143134
} else if ((byte & 0b11111000) == 0b11110000) { // 0b11110000
144135
next_pos = pos + 4;
145-
if (next_pos > len) {
146-
return false;
147-
}
148-
if ((data[pos + 1] & 0b11000000) != 0b10000000) {
149-
return false;
150-
}
151-
if ((data[pos + 2] & 0b11000000) != 0b10000000) {
152-
return false;
153-
}
154-
if ((data[pos + 3] & 0b11000000) != 0b10000000) {
155-
return false;
156-
}
136+
if (next_pos > len) { return false; }
137+
if ((data[pos + 1] & 0b11000000) != 0b10000000) { return false; }
138+
if ((data[pos + 2] & 0b11000000) != 0b10000000) { return false; }
139+
if ((data[pos + 3] & 0b11000000) != 0b10000000) { return false; }
157140
// range check
158141
code_point =
159142
(byte & 0b00000111) << 18 | (data[pos + 1] & 0b00111111) << 12 |

0 commit comments

Comments
 (0)