xref: /aosp_15_r20/external/cronet/third_party/abseil-cpp/absl/strings/ascii.cc (revision 6777b5387eb2ff775bb5750e3f5d96f37fb7352b)
1 // Copyright 2017 The Abseil Authors.
2 //
3 // Licensed under the Apache License, Version 2.0 (the "License");
4 // you may not use this file except in compliance with the License.
5 // You may obtain a copy of the License at
6 //
7 //      https://www.apache.org/licenses/LICENSE-2.0
8 //
9 // Unless required by applicable law or agreed to in writing, software
10 // distributed under the License is distributed on an "AS IS" BASIS,
11 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12 // See the License for the specific language governing permissions and
13 // limitations under the License.
14 
15 #include "absl/strings/ascii.h"
16 
17 #include <climits>
18 #include <cstddef>
19 #include <cstring>
20 #include <string>
21 
22 #include "absl/base/attributes.h"
23 #include "absl/base/config.h"
24 #include "absl/base/nullability.h"
25 #include "absl/base/optimization.h"
26 
27 namespace absl {
28 ABSL_NAMESPACE_BEGIN
29 namespace ascii_internal {
30 
31 // # Table generated by this Python code (bit 0x02 is currently unused):
32 // TODO(mbar) Move Python code for generation of table to BUILD and link here.
33 
34 // NOTE: The kAsciiPropertyBits table used within this code was generated by
35 // Python code of the following form. (Bit 0x02 is currently unused and
36 // available.)
37 //
38 // def Hex2(n):
39 //   return '0x' + hex(n/16)[2:] + hex(n%16)[2:]
40 // def IsPunct(ch):
41 //   return (ord(ch) >= 32 and ord(ch) < 127 and
42 //           not ch.isspace() and not ch.isalnum())
43 // def IsBlank(ch):
44 //   return ch in ' \t'
45 // def IsCntrl(ch):
46 //   return ord(ch) < 32 or ord(ch) == 127
47 // def IsXDigit(ch):
48 //   return ch.isdigit() or ch.lower() in 'abcdef'
49 // for i in range(128):
50 //   ch = chr(i)
51 //   mask = ((ch.isalpha() and 0x01 or 0) |
52 //           (ch.isalnum() and 0x04 or 0) |
53 //           (ch.isspace() and 0x08 or 0) |
54 //           (IsPunct(ch) and 0x10 or 0) |
55 //           (IsBlank(ch) and 0x20 or 0) |
56 //           (IsCntrl(ch) and 0x40 or 0) |
57 //           (IsXDigit(ch) and 0x80 or 0))
58 //   print Hex2(mask) + ',',
59 //   if i % 16 == 7:
60 //     print ' //', Hex2(i & 0x78)
61 //   elif i % 16 == 15:
62 //     print
63 
64 // clang-format off
65 // Array of bitfields holding character information. Each bit value corresponds
66 // to a particular character feature. For readability, and because the value
67 // of these bits is tightly coupled to this implementation, the individual bits
68 // are not named. Note that bitfields for all characters above ASCII 127 are
69 // zero-initialized.
70 ABSL_DLL const unsigned char kPropertyBits[256] = {
71     0x40, 0x40, 0x40, 0x40, 0x40, 0x40, 0x40, 0x40,  // 0x00
72     0x40, 0x68, 0x48, 0x48, 0x48, 0x48, 0x40, 0x40,
73     0x40, 0x40, 0x40, 0x40, 0x40, 0x40, 0x40, 0x40,  // 0x10
74     0x40, 0x40, 0x40, 0x40, 0x40, 0x40, 0x40, 0x40,
75     0x28, 0x10, 0x10, 0x10, 0x10, 0x10, 0x10, 0x10,  // 0x20
76     0x10, 0x10, 0x10, 0x10, 0x10, 0x10, 0x10, 0x10,
77     0x84, 0x84, 0x84, 0x84, 0x84, 0x84, 0x84, 0x84,  // 0x30
78     0x84, 0x84, 0x10, 0x10, 0x10, 0x10, 0x10, 0x10,
79     0x10, 0x85, 0x85, 0x85, 0x85, 0x85, 0x85, 0x05,  // 0x40
80     0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05,
81     0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05,  // 0x50
82     0x05, 0x05, 0x05, 0x10, 0x10, 0x10, 0x10, 0x10,
83     0x10, 0x85, 0x85, 0x85, 0x85, 0x85, 0x85, 0x05,  // 0x60
84     0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05,
85     0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05,  // 0x70
86     0x05, 0x05, 0x05, 0x10, 0x10, 0x10, 0x10, 0x40,
87 };
88 
89 // Array of characters for the ascii_tolower() function. For values 'A'
90 // through 'Z', return the lower-case character; otherwise, return the
91 // identity of the passed character.
92 ABSL_DLL const char kToLower[256] = {
93   '\x00', '\x01', '\x02', '\x03', '\x04', '\x05', '\x06', '\x07',
94   '\x08', '\x09', '\x0a', '\x0b', '\x0c', '\x0d', '\x0e', '\x0f',
95   '\x10', '\x11', '\x12', '\x13', '\x14', '\x15', '\x16', '\x17',
96   '\x18', '\x19', '\x1a', '\x1b', '\x1c', '\x1d', '\x1e', '\x1f',
97   '\x20', '\x21', '\x22', '\x23', '\x24', '\x25', '\x26', '\x27',
98   '\x28', '\x29', '\x2a', '\x2b', '\x2c', '\x2d', '\x2e', '\x2f',
99   '\x30', '\x31', '\x32', '\x33', '\x34', '\x35', '\x36', '\x37',
100   '\x38', '\x39', '\x3a', '\x3b', '\x3c', '\x3d', '\x3e', '\x3f',
101   '\x40',    'a',    'b',    'c',    'd',    'e',    'f',    'g',
102      'h',    'i',    'j',    'k',    'l',    'm',    'n',    'o',
103      'p',    'q',    'r',    's',    't',    'u',    'v',    'w',
104      'x',    'y',    'z', '\x5b', '\x5c', '\x5d', '\x5e', '\x5f',
105   '\x60', '\x61', '\x62', '\x63', '\x64', '\x65', '\x66', '\x67',
106   '\x68', '\x69', '\x6a', '\x6b', '\x6c', '\x6d', '\x6e', '\x6f',
107   '\x70', '\x71', '\x72', '\x73', '\x74', '\x75', '\x76', '\x77',
108   '\x78', '\x79', '\x7a', '\x7b', '\x7c', '\x7d', '\x7e', '\x7f',
109   '\x80', '\x81', '\x82', '\x83', '\x84', '\x85', '\x86', '\x87',
110   '\x88', '\x89', '\x8a', '\x8b', '\x8c', '\x8d', '\x8e', '\x8f',
111   '\x90', '\x91', '\x92', '\x93', '\x94', '\x95', '\x96', '\x97',
112   '\x98', '\x99', '\x9a', '\x9b', '\x9c', '\x9d', '\x9e', '\x9f',
113   '\xa0', '\xa1', '\xa2', '\xa3', '\xa4', '\xa5', '\xa6', '\xa7',
114   '\xa8', '\xa9', '\xaa', '\xab', '\xac', '\xad', '\xae', '\xaf',
115   '\xb0', '\xb1', '\xb2', '\xb3', '\xb4', '\xb5', '\xb6', '\xb7',
116   '\xb8', '\xb9', '\xba', '\xbb', '\xbc', '\xbd', '\xbe', '\xbf',
117   '\xc0', '\xc1', '\xc2', '\xc3', '\xc4', '\xc5', '\xc6', '\xc7',
118   '\xc8', '\xc9', '\xca', '\xcb', '\xcc', '\xcd', '\xce', '\xcf',
119   '\xd0', '\xd1', '\xd2', '\xd3', '\xd4', '\xd5', '\xd6', '\xd7',
120   '\xd8', '\xd9', '\xda', '\xdb', '\xdc', '\xdd', '\xde', '\xdf',
121   '\xe0', '\xe1', '\xe2', '\xe3', '\xe4', '\xe5', '\xe6', '\xe7',
122   '\xe8', '\xe9', '\xea', '\xeb', '\xec', '\xed', '\xee', '\xef',
123   '\xf0', '\xf1', '\xf2', '\xf3', '\xf4', '\xf5', '\xf6', '\xf7',
124   '\xf8', '\xf9', '\xfa', '\xfb', '\xfc', '\xfd', '\xfe', '\xff',
125 };
126 
127 // Array of characters for the ascii_toupper() function. For values 'a'
128 // through 'z', return the upper-case character; otherwise, return the
129 // identity of the passed character.
130 ABSL_DLL const char kToUpper[256] = {
131   '\x00', '\x01', '\x02', '\x03', '\x04', '\x05', '\x06', '\x07',
132   '\x08', '\x09', '\x0a', '\x0b', '\x0c', '\x0d', '\x0e', '\x0f',
133   '\x10', '\x11', '\x12', '\x13', '\x14', '\x15', '\x16', '\x17',
134   '\x18', '\x19', '\x1a', '\x1b', '\x1c', '\x1d', '\x1e', '\x1f',
135   '\x20', '\x21', '\x22', '\x23', '\x24', '\x25', '\x26', '\x27',
136   '\x28', '\x29', '\x2a', '\x2b', '\x2c', '\x2d', '\x2e', '\x2f',
137   '\x30', '\x31', '\x32', '\x33', '\x34', '\x35', '\x36', '\x37',
138   '\x38', '\x39', '\x3a', '\x3b', '\x3c', '\x3d', '\x3e', '\x3f',
139   '\x40', '\x41', '\x42', '\x43', '\x44', '\x45', '\x46', '\x47',
140   '\x48', '\x49', '\x4a', '\x4b', '\x4c', '\x4d', '\x4e', '\x4f',
141   '\x50', '\x51', '\x52', '\x53', '\x54', '\x55', '\x56', '\x57',
142   '\x58', '\x59', '\x5a', '\x5b', '\x5c', '\x5d', '\x5e', '\x5f',
143   '\x60',    'A',    'B',    'C',    'D',    'E',    'F',    'G',
144      'H',    'I',    'J',    'K',    'L',    'M',    'N',    'O',
145      'P',    'Q',    'R',    'S',    'T',    'U',    'V',    'W',
146      'X',    'Y',    'Z', '\x7b', '\x7c', '\x7d', '\x7e', '\x7f',
147   '\x80', '\x81', '\x82', '\x83', '\x84', '\x85', '\x86', '\x87',
148   '\x88', '\x89', '\x8a', '\x8b', '\x8c', '\x8d', '\x8e', '\x8f',
149   '\x90', '\x91', '\x92', '\x93', '\x94', '\x95', '\x96', '\x97',
150   '\x98', '\x99', '\x9a', '\x9b', '\x9c', '\x9d', '\x9e', '\x9f',
151   '\xa0', '\xa1', '\xa2', '\xa3', '\xa4', '\xa5', '\xa6', '\xa7',
152   '\xa8', '\xa9', '\xaa', '\xab', '\xac', '\xad', '\xae', '\xaf',
153   '\xb0', '\xb1', '\xb2', '\xb3', '\xb4', '\xb5', '\xb6', '\xb7',
154   '\xb8', '\xb9', '\xba', '\xbb', '\xbc', '\xbd', '\xbe', '\xbf',
155   '\xc0', '\xc1', '\xc2', '\xc3', '\xc4', '\xc5', '\xc6', '\xc7',
156   '\xc8', '\xc9', '\xca', '\xcb', '\xcc', '\xcd', '\xce', '\xcf',
157   '\xd0', '\xd1', '\xd2', '\xd3', '\xd4', '\xd5', '\xd6', '\xd7',
158   '\xd8', '\xd9', '\xda', '\xdb', '\xdc', '\xdd', '\xde', '\xdf',
159   '\xe0', '\xe1', '\xe2', '\xe3', '\xe4', '\xe5', '\xe6', '\xe7',
160   '\xe8', '\xe9', '\xea', '\xeb', '\xec', '\xed', '\xee', '\xef',
161   '\xf0', '\xf1', '\xf2', '\xf3', '\xf4', '\xf5', '\xf6', '\xf7',
162   '\xf8', '\xf9', '\xfa', '\xfb', '\xfc', '\xfd', '\xfe', '\xff',
163 };
164 // clang-format on
165 
166 // Returns whether `c` is in the a-z/A-Z range (w.r.t. `ToUpper`).
167 // Implemented by:
168 //  1. Pushing the a-z/A-Z range to [SCHAR_MIN, SCHAR_MIN + 26).
169 //  2. Comparing to SCHAR_MIN + 26.
170 template <bool ToUpper>
AsciiInAZRange(unsigned char c)171 constexpr bool AsciiInAZRange(unsigned char c) {
172   constexpr unsigned char sub = (ToUpper ? 'a' : 'A') - SCHAR_MIN;
173   constexpr signed char threshold = SCHAR_MIN + 26;  // 26 = alphabet size.
174   // Using unsigned arithmetic as overflows/underflows are well defined.
175   unsigned char u = c - sub;
176   // Using signed cmp, as SIMD unsigned cmp isn't available in many platforms.
177   return static_cast<signed char>(u) < threshold;
178 }
179 
180 // Force-inline so the compiler won't merge the short and long implementations.
181 template <bool ToUpper>
AsciiStrCaseFoldImpl(absl::Nonnull<char * > p,size_t size)182 ABSL_ATTRIBUTE_ALWAYS_INLINE inline constexpr void AsciiStrCaseFoldImpl(
183     absl::Nonnull<char*> p, size_t size) {
184   // The upper- and lowercase versions of ASCII characters differ by only 1 bit.
185   // When we need to flip the case, we can xor with this bit to achieve the
186   // desired result. Note that the choice of 'a' and 'A' here is arbitrary. We
187   // could have chosen 'z' and 'Z', or any other pair of characters as they all
188   // have the same single bit difference.
189   constexpr unsigned char kAsciiCaseBitFlip = 'a' ^ 'A';
190 
191   for (size_t i = 0; i < size; ++i) {
192     unsigned char v = static_cast<unsigned char>(p[i]);
193     v ^= AsciiInAZRange<ToUpper>(v) ? kAsciiCaseBitFlip : 0;
194     p[i] = static_cast<char>(v);
195   }
196 }
197 
198 // The string size threshold for starting using the long string version.
199 constexpr size_t kCaseFoldThreshold = 16;
200 
201 // No-inline so the compiler won't merge the short and long implementations.
202 template <bool ToUpper>
AsciiStrCaseFoldLong(absl::Nonnull<char * > p,size_t size)203 ABSL_ATTRIBUTE_NOINLINE constexpr void AsciiStrCaseFoldLong(
204     absl::Nonnull<char*> p, size_t size) {
205   ABSL_ASSUME(size >= kCaseFoldThreshold);
206   AsciiStrCaseFoldImpl<ToUpper>(p, size);
207 }
208 
209 // Splitting to short and long strings to allow vectorization decisions
210 // to be made separately in the long and short cases.
211 template <bool ToUpper>
AsciiStrCaseFold(absl::Nonnull<char * > p,size_t size)212 constexpr void AsciiStrCaseFold(absl::Nonnull<char*> p, size_t size) {
213   size < kCaseFoldThreshold ? AsciiStrCaseFoldImpl<ToUpper>(p, size)
214                             : AsciiStrCaseFoldLong<ToUpper>(p, size);
215 }
216 
ValidateAsciiCasefold()217 static constexpr size_t ValidateAsciiCasefold() {
218   constexpr size_t num_chars = 1 + CHAR_MAX - CHAR_MIN;
219   size_t incorrect_index = 0;
220   char lowered[num_chars] = {};
221   char uppered[num_chars] = {};
222   for (unsigned int i = 0; i < num_chars; ++i) {
223     uppered[i] = lowered[i] = static_cast<char>(i);
224   }
225   AsciiStrCaseFold<false>(&lowered[0], num_chars);
226   AsciiStrCaseFold<true>(&uppered[0], num_chars);
227   for (size_t i = 0; i < num_chars; ++i) {
228     const char ch = static_cast<char>(i),
229                ch_upper = ('a' <= ch && ch <= 'z' ? 'A' + (ch - 'a') : ch),
230                ch_lower = ('A' <= ch && ch <= 'Z' ? 'a' + (ch - 'A') : ch);
231     if (uppered[i] != ch_upper || lowered[i] != ch_lower) {
232       incorrect_index = i > 0 ? i : num_chars;
233       break;
234     }
235   }
236   return incorrect_index;
237 }
238 
239 static_assert(ValidateAsciiCasefold() == 0, "error in case conversion");
240 
241 }  // namespace ascii_internal
242 
AsciiStrToLower(absl::Nonnull<std::string * > s)243 void AsciiStrToLower(absl::Nonnull<std::string*> s) {
244   return ascii_internal::AsciiStrCaseFold<false>(&(*s)[0], s->size());
245 }
246 
AsciiStrToUpper(absl::Nonnull<std::string * > s)247 void AsciiStrToUpper(absl::Nonnull<std::string*> s) {
248   return ascii_internal::AsciiStrCaseFold<true>(&(*s)[0], s->size());
249 }
250 
RemoveExtraAsciiWhitespace(absl::Nonnull<std::string * > str)251 void RemoveExtraAsciiWhitespace(absl::Nonnull<std::string*> str) {
252   auto stripped = StripAsciiWhitespace(*str);
253 
254   if (stripped.empty()) {
255     str->clear();
256     return;
257   }
258 
259   auto input_it = stripped.begin();
260   auto input_end = stripped.end();
261   auto output_it = &(*str)[0];
262   bool is_ws = false;
263 
264   for (; input_it < input_end; ++input_it) {
265     if (is_ws) {
266       // Consecutive whitespace?  Keep only the last.
267       is_ws = absl::ascii_isspace(static_cast<unsigned char>(*input_it));
268       if (is_ws) --output_it;
269     } else {
270       is_ws = absl::ascii_isspace(static_cast<unsigned char>(*input_it));
271     }
272 
273     *output_it = *input_it;
274     ++output_it;
275   }
276 
277   str->erase(static_cast<size_t>(output_it - &(*str)[0]));
278 }
279 
280 ABSL_NAMESPACE_END
281 }  // namespace absl
282