1 /*
2 * Copyright 2014 Google Inc. All rights reserved.
3 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 * http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16
17 #ifndef FLATBUFFERS_UTIL_H_
18 #define FLATBUFFERS_UTIL_H_
19
20 #include <ctype.h>
21 #include <errno.h>
22
23 #include "flatbuffers/base.h"
24 #include "flatbuffers/stl_emulation.h"
25
26 #ifndef FLATBUFFERS_PREFER_PRINTF
27 # include <iomanip>
28 # include <sstream>
29 #else // FLATBUFFERS_PREFER_PRINTF
30 # include <float.h>
31 # include <stdio.h>
32 #endif // FLATBUFFERS_PREFER_PRINTF
33
34 #include <cmath>
35 #include <limits>
36 #include <string>
37
38 namespace flatbuffers {
39
40 // @locale-independent functions for ASCII characters set.
41
42 // Fast checking that character lies in closed range: [a <= x <= b]
43 // using one compare (conditional branch) operator.
check_ascii_range(char x,char a,char b)44 inline bool check_ascii_range(char x, char a, char b) {
45 FLATBUFFERS_ASSERT(a <= b);
46 // (Hacker's Delight): `a <= x <= b` <=> `(x-a) <={u} (b-a)`.
47 // The x, a, b will be promoted to int and subtracted without overflow.
48 return static_cast<unsigned int>(x - a) <= static_cast<unsigned int>(b - a);
49 }
50
51 // Case-insensitive isalpha
is_alpha(char c)52 inline bool is_alpha(char c) {
53 // ASCII only: alpha to upper case => reset bit 0x20 (~0x20 = 0xDF).
54 return check_ascii_range(c & 0xDF, 'a' & 0xDF, 'z' & 0xDF);
55 }
56
57 // Check for uppercase alpha
is_alpha_upper(char c)58 inline bool is_alpha_upper(char c) { return check_ascii_range(c, 'A', 'Z'); }
59
60 // Check (case-insensitive) that `c` is equal to alpha.
is_alpha_char(char c,char alpha)61 inline bool is_alpha_char(char c, char alpha) {
62 FLATBUFFERS_ASSERT(is_alpha(alpha));
63 // ASCII only: alpha to upper case => reset bit 0x20 (~0x20 = 0xDF).
64 return ((c & 0xDF) == (alpha & 0xDF));
65 }
66
67 // https://en.cppreference.com/w/cpp/string/byte/isxdigit
68 // isdigit and isxdigit are the only standard narrow character classification
69 // functions that are not affected by the currently installed C locale. although
70 // some implementations (e.g. Microsoft in 1252 codepage) may classify
71 // additional single-byte characters as digits.
is_digit(char c)72 inline bool is_digit(char c) { return check_ascii_range(c, '0', '9'); }
73
is_xdigit(char c)74 inline bool is_xdigit(char c) {
75 // Replace by look-up table.
76 return is_digit(c) || check_ascii_range(c & 0xDF, 'a' & 0xDF, 'f' & 0xDF);
77 }
78
79 // Case-insensitive isalnum
is_alnum(char c)80 inline bool is_alnum(char c) { return is_alpha(c) || is_digit(c); }
81
CharToUpper(char c)82 inline char CharToUpper(char c) {
83 return static_cast<char>(::toupper(static_cast<unsigned char>(c)));
84 }
85
CharToLower(char c)86 inline char CharToLower(char c) {
87 return static_cast<char>(::tolower(static_cast<unsigned char>(c)));
88 }
89
90 // @end-locale-independent functions for ASCII character set
91
92 #ifdef FLATBUFFERS_PREFER_PRINTF
IntToDigitCount(T t)93 template<typename T> size_t IntToDigitCount(T t) {
94 size_t digit_count = 0;
95 // Count the sign for negative numbers
96 if (t < 0) digit_count++;
97 // Count a single 0 left of the dot for fractional numbers
98 if (-1 < t && t < 1) digit_count++;
99 // Count digits until fractional part
100 T eps = std::numeric_limits<T>::epsilon();
101 while (t <= (-1 + eps) || (1 - eps) <= t) {
102 t /= 10;
103 digit_count++;
104 }
105 return digit_count;
106 }
107
108 template<typename T> size_t NumToStringWidth(T t, int precision = 0) {
109 size_t string_width = IntToDigitCount(t);
110 // Count the dot for floating point numbers
111 if (precision) string_width += (precision + 1);
112 return string_width;
113 }
114
115 template<typename T>
116 std::string NumToStringImplWrapper(T t, const char *fmt, int precision = 0) {
117 size_t string_width = NumToStringWidth(t, precision);
118 std::string s(string_width, 0x00);
119 // Allow snprintf to use std::string trailing null to detect buffer overflow
120 snprintf(const_cast<char *>(s.data()), (s.size() + 1), fmt, string_width, t);
121 return s;
122 }
123 #endif // FLATBUFFERS_PREFER_PRINTF
124
125 // Convert an integer or floating point value to a string.
126 // In contrast to std::stringstream, "char" values are
127 // converted to a string of digits, and we don't use scientific notation.
NumToString(T t)128 template<typename T> std::string NumToString(T t) {
129 // clang-format off
130
131 #ifndef FLATBUFFERS_PREFER_PRINTF
132 std::stringstream ss;
133 ss << t;
134 return ss.str();
135 #else // FLATBUFFERS_PREFER_PRINTF
136 auto v = static_cast<long long>(t);
137 return NumToStringImplWrapper(v, "%.*lld");
138 #endif // FLATBUFFERS_PREFER_PRINTF
139 // clang-format on
140 }
141 // Avoid char types used as character data.
142 template<> inline std::string NumToString<signed char>(signed char t) {
143 return NumToString(static_cast<int>(t));
144 }
145 template<> inline std::string NumToString<unsigned char>(unsigned char t) {
146 return NumToString(static_cast<int>(t));
147 }
148 template<> inline std::string NumToString<char>(char t) {
149 return NumToString(static_cast<int>(t));
150 }
151
152 // Special versions for floats/doubles.
FloatToString(T t,int precision)153 template<typename T> std::string FloatToString(T t, int precision) {
154 // clang-format off
155
156 #ifndef FLATBUFFERS_PREFER_PRINTF
157 // to_string() prints different numbers of digits for floats depending on
158 // platform and isn't available on Android, so we use stringstream
159 std::stringstream ss;
160 // Use std::fixed to suppress scientific notation.
161 ss << std::fixed;
162 // Default precision is 6, we want that to be higher for doubles.
163 ss << std::setprecision(precision);
164 ss << t;
165 auto s = ss.str();
166 #else // FLATBUFFERS_PREFER_PRINTF
167 auto v = static_cast<double>(t);
168 auto s = NumToStringImplWrapper(v, "%0.*f", precision);
169 #endif // FLATBUFFERS_PREFER_PRINTF
170 // clang-format on
171 // Sadly, std::fixed turns "1" into "1.00000", so here we undo that.
172 auto p = s.find_last_not_of('0');
173 if (p != std::string::npos) {
174 // Strip trailing zeroes. If it is a whole number, keep one zero.
175 s.resize(p + (s[p] == '.' ? 2 : 1));
176 }
177 return s;
178 }
179
180 template<> inline std::string NumToString<double>(double t) {
181 return FloatToString(t, 12);
182 }
183 template<> inline std::string NumToString<float>(float t) {
184 return FloatToString(t, 6);
185 }
186
187 // Convert an integer value to a hexadecimal string.
188 // The returned string length is always xdigits long, prefixed by 0 digits.
189 // For example, IntToStringHex(0x23, 8) returns the string "00000023".
IntToStringHex(int i,int xdigits)190 inline std::string IntToStringHex(int i, int xdigits) {
191 FLATBUFFERS_ASSERT(i >= 0);
192 // clang-format off
193
194 #ifndef FLATBUFFERS_PREFER_PRINTF
195 std::stringstream ss;
196 ss << std::setw(xdigits) << std::setfill('0') << std::hex << std::uppercase
197 << i;
198 return ss.str();
199 #else // FLATBUFFERS_PREFER_PRINTF
200 return NumToStringImplWrapper(i, "%.*X", xdigits);
201 #endif // FLATBUFFERS_PREFER_PRINTF
202 // clang-format on
203 }
204
205 // clang-format off
206 // Use locale independent functions {strtod_l, strtof_l, strtoll_l, strtoull_l}.
207 #if defined(FLATBUFFERS_LOCALE_INDEPENDENT) && (FLATBUFFERS_LOCALE_INDEPENDENT > 0)
208 class ClassicLocale {
209 #ifdef _MSC_VER
210 typedef _locale_t locale_type;
211 #else
212 typedef locale_t locale_type; // POSIX.1-2008 locale_t type
213 #endif
214 ClassicLocale();
215 ~ClassicLocale();
216 locale_type locale_;
217 static ClassicLocale instance_;
218 public:
Get()219 static locale_type Get() { return instance_.locale_; }
220 };
221
222 #ifdef _MSC_VER
223 #define __strtoull_impl(s, pe, b) _strtoui64_l(s, pe, b, ClassicLocale::Get())
224 #define __strtoll_impl(s, pe, b) _strtoi64_l(s, pe, b, ClassicLocale::Get())
225 #define __strtod_impl(s, pe) _strtod_l(s, pe, ClassicLocale::Get())
226 #define __strtof_impl(s, pe) _strtof_l(s, pe, ClassicLocale::Get())
227 #else
228 #define __strtoull_impl(s, pe, b) strtoull_l(s, pe, b, ClassicLocale::Get())
229 #define __strtoll_impl(s, pe, b) strtoll_l(s, pe, b, ClassicLocale::Get())
230 #define __strtod_impl(s, pe) strtod_l(s, pe, ClassicLocale::Get())
231 #define __strtof_impl(s, pe) strtof_l(s, pe, ClassicLocale::Get())
232 #endif
233 #else
234 #define __strtod_impl(s, pe) strtod(s, pe)
235 #define __strtof_impl(s, pe) static_cast<float>(strtod(s, pe))
236 #ifdef _MSC_VER
237 #define __strtoull_impl(s, pe, b) _strtoui64(s, pe, b)
238 #define __strtoll_impl(s, pe, b) _strtoi64(s, pe, b)
239 #else
240 #define __strtoull_impl(s, pe, b) strtoull(s, pe, b)
241 #define __strtoll_impl(s, pe, b) strtoll(s, pe, b)
242 #endif
243 #endif
244
strtoval_impl(int64_t * val,const char * str,char ** endptr,int base)245 inline void strtoval_impl(int64_t *val, const char *str, char **endptr,
246 int base) {
247 *val = __strtoll_impl(str, endptr, base);
248 }
249
strtoval_impl(uint64_t * val,const char * str,char ** endptr,int base)250 inline void strtoval_impl(uint64_t *val, const char *str, char **endptr,
251 int base) {
252 *val = __strtoull_impl(str, endptr, base);
253 }
254
strtoval_impl(double * val,const char * str,char ** endptr)255 inline void strtoval_impl(double *val, const char *str, char **endptr) {
256 *val = __strtod_impl(str, endptr);
257 }
258
259 // UBSAN: double to float is safe if numeric_limits<float>::is_iec559 is true.
260 FLATBUFFERS_SUPPRESS_UBSAN("float-cast-overflow")
strtoval_impl(float * val,const char * str,char ** endptr)261 inline void strtoval_impl(float *val, const char *str, char **endptr) {
262 *val = __strtof_impl(str, endptr);
263 }
264 #undef __strtoull_impl
265 #undef __strtoll_impl
266 #undef __strtod_impl
267 #undef __strtof_impl
268 // clang-format on
269
270 // Adaptor for strtoull()/strtoll().
271 // Flatbuffers accepts numbers with any count of leading zeros (-009 is -9),
272 // while strtoll with base=0 interprets first leading zero as octal prefix.
273 // In future, it is possible to add prefixed 0b0101.
274 // 1) Checks errno code for overflow condition (out of range).
275 // 2) If base <= 0, function try to detect base of number by prefix.
276 //
277 // Return value (like strtoull and strtoll, but reject partial result):
278 // - If successful, an integer value corresponding to the str is returned.
279 // - If full string conversion can't be performed, 0 is returned.
280 // - If the converted value falls out of range of corresponding return type, a
281 // range error occurs. In this case value MAX(T)/MIN(T) is returned.
282 template<typename T>
283 inline bool StringToIntegerImpl(T *val, const char *const str,
284 const int base = 0,
285 const bool check_errno = true) {
286 // T is int64_t or uint64_T
287 FLATBUFFERS_ASSERT(str);
288 if (base <= 0) {
289 auto s = str;
290 while (*s && !is_digit(*s)) s++;
291 if (s[0] == '0' && is_alpha_char(s[1], 'X'))
292 return StringToIntegerImpl(val, str, 16, check_errno);
293 // if a prefix not match, try base=10
294 return StringToIntegerImpl(val, str, 10, check_errno);
295 } else {
296 if (check_errno) errno = 0; // clear thread-local errno
297 auto endptr = str;
298 strtoval_impl(val, str, const_cast<char **>(&endptr), base);
299 if ((*endptr != '\0') || (endptr == str)) {
300 *val = 0; // erase partial result
301 return false; // invalid string
302 }
303 // errno is out-of-range, return MAX/MIN
304 if (check_errno && errno) return false;
305 return true;
306 }
307 }
308
309 template<typename T>
StringToFloatImpl(T * val,const char * const str)310 inline bool StringToFloatImpl(T *val, const char *const str) {
311 // Type T must be either float or double.
312 FLATBUFFERS_ASSERT(str && val);
313 auto end = str;
314 strtoval_impl(val, str, const_cast<char **>(&end));
315 auto done = (end != str) && (*end == '\0');
316 if (!done) *val = 0; // erase partial result
317 if (done && std::isnan(*val)) { *val = std::numeric_limits<T>::quiet_NaN(); }
318 return done;
319 }
320
321 // Convert a string to an instance of T.
322 // Return value (matched with StringToInteger64Impl and strtod):
323 // - If successful, a numeric value corresponding to the str is returned.
324 // - If full string conversion can't be performed, 0 is returned.
325 // - If the converted value falls out of range of corresponding return type, a
326 // range error occurs. In this case value MAX(T)/MIN(T) is returned.
StringToNumber(const char * s,T * val)327 template<typename T> inline bool StringToNumber(const char *s, T *val) {
328 // Assert on `unsigned long` and `signed long` on LP64.
329 // If it is necessary, it could be solved with flatbuffers::enable_if<B,T>.
330 static_assert(sizeof(T) < sizeof(int64_t), "unexpected type T");
331 FLATBUFFERS_ASSERT(s && val);
332 int64_t i64;
333 // The errno check isn't needed, will return MAX/MIN on overflow.
334 if (StringToIntegerImpl(&i64, s, 0, false)) {
335 const int64_t max = (flatbuffers::numeric_limits<T>::max)();
336 const int64_t min = flatbuffers::numeric_limits<T>::lowest();
337 if (i64 > max) {
338 *val = static_cast<T>(max);
339 return false;
340 }
341 if (i64 < min) {
342 // For unsigned types return max to distinguish from
343 // "no conversion can be performed" when 0 is returned.
344 *val = static_cast<T>(flatbuffers::is_unsigned<T>::value ? max : min);
345 return false;
346 }
347 *val = static_cast<T>(i64);
348 return true;
349 }
350 *val = 0;
351 return false;
352 }
353
354 template<> inline bool StringToNumber<int64_t>(const char *str, int64_t *val) {
355 return StringToIntegerImpl(val, str);
356 }
357
358 template<>
359 inline bool StringToNumber<uint64_t>(const char *str, uint64_t *val) {
360 if (!StringToIntegerImpl(val, str)) return false;
361 // The strtoull accepts negative numbers:
362 // If the minus sign was part of the input sequence, the numeric value
363 // calculated from the sequence of digits is negated as if by unary minus
364 // in the result type, which applies unsigned integer wraparound rules.
365 // Fix this behaviour (except -0).
366 if (*val) {
367 auto s = str;
368 while (*s && !is_digit(*s)) s++;
369 s = (s > str) ? (s - 1) : s; // step back to one symbol
370 if (*s == '-') {
371 // For unsigned types return the max to distinguish from
372 // "no conversion can be performed".
373 *val = (flatbuffers::numeric_limits<uint64_t>::max)();
374 return false;
375 }
376 }
377 return true;
378 }
379
StringToNumber(const char * s,float * val)380 template<> inline bool StringToNumber(const char *s, float *val) {
381 return StringToFloatImpl(val, s);
382 }
383
StringToNumber(const char * s,double * val)384 template<> inline bool StringToNumber(const char *s, double *val) {
385 return StringToFloatImpl(val, s);
386 }
387
388 inline int64_t StringToInt(const char *s, int base = 10) {
389 int64_t val;
390 return StringToIntegerImpl(&val, s, base) ? val : 0;
391 }
392
393 inline uint64_t StringToUInt(const char *s, int base = 10) {
394 uint64_t val;
395 return StringToIntegerImpl(&val, s, base) ? val : 0;
396 }
397
StringIsFlatbufferNan(const std::string & s)398 inline bool StringIsFlatbufferNan(const std::string &s) {
399 return s == "nan" || s == "+nan" || s == "-nan";
400 }
401
StringIsFlatbufferPositiveInfinity(const std::string & s)402 inline bool StringIsFlatbufferPositiveInfinity(const std::string &s) {
403 return s == "inf" || s == "+inf" || s == "infinity" || s == "+infinity";
404 }
405
StringIsFlatbufferNegativeInfinity(const std::string & s)406 inline bool StringIsFlatbufferNegativeInfinity(const std::string &s) {
407 return s == "-inf" || s == "-infinity";
408 }
409
410 typedef bool (*LoadFileFunction)(const char *filename, bool binary,
411 std::string *dest);
412 typedef bool (*FileExistsFunction)(const char *filename);
413
414 LoadFileFunction SetLoadFileFunction(LoadFileFunction load_file_function);
415
416 FileExistsFunction SetFileExistsFunction(
417 FileExistsFunction file_exists_function);
418
419 // Check if file "name" exists.
420 bool FileExists(const char *name);
421
422 // Check if "name" exists and it is also a directory.
423 bool DirExists(const char *name);
424
425 // Load file "name" into "buf" returning true if successful
426 // false otherwise. If "binary" is false data is read
427 // using ifstream's text mode, otherwise data is read with
428 // no transcoding.
429 bool LoadFile(const char *name, bool binary, std::string *buf);
430
431 // Save data "buf" of length "len" bytes into a file
432 // "name" returning true if successful, false otherwise.
433 // If "binary" is false data is written using ifstream's
434 // text mode, otherwise data is written with no
435 // transcoding.
436 bool SaveFile(const char *name, const char *buf, size_t len, bool binary);
437
438 // Save data "buf" into file "name" returning true if
439 // successful, false otherwise. If "binary" is false
440 // data is written using ifstream's text mode, otherwise
441 // data is written with no transcoding.
SaveFile(const char * name,const std::string & buf,bool binary)442 inline bool SaveFile(const char *name, const std::string &buf, bool binary) {
443 return SaveFile(name, buf.c_str(), buf.size(), binary);
444 }
445
446 // Functionality for minimalistic portable path handling.
447
448 // The functions below behave correctly regardless of whether posix ('/') or
449 // Windows ('/' or '\\') separators are used.
450
451 // Any new separators inserted are always posix.
452 FLATBUFFERS_CONSTEXPR char kPathSeparator = '/';
453
454 // Returns the path with the extension, if any, removed.
455 std::string StripExtension(const std::string &filepath);
456
457 // Returns the extension, if any.
458 std::string GetExtension(const std::string &filepath);
459
460 // Return the last component of the path, after the last separator.
461 std::string StripPath(const std::string &filepath);
462
463 // Strip the last component of the path + separator.
464 std::string StripFileName(const std::string &filepath);
465
466 std::string StripPrefix(const std::string &filepath,
467 const std::string &prefix_to_remove);
468
469 // Concatenates a path with a filename, regardless of whether the path
470 // ends in a separator or not.
471 std::string ConCatPathFileName(const std::string &path,
472 const std::string &filename);
473
474 // Replaces any '\\' separators with '/'
475 std::string PosixPath(const char *path);
476 std::string PosixPath(const std::string &path);
477
478 // This function ensure a directory exists, by recursively
479 // creating dirs for any parts of the path that don't exist yet.
480 void EnsureDirExists(const std::string &filepath);
481
482 // Obtains the relative or absolute path.
483 std::string FilePath(const std::string &project,
484 const std::string &filePath,
485 bool absolute);
486
487 // Obtains the absolute path from any other path.
488 // Returns the input path if the absolute path couldn't be resolved.
489 std::string AbsolutePath(const std::string &filepath);
490
491 // Returns files relative to the --project_root path, prefixed with `//`.
492 std::string RelativeToRootPath(const std::string &project,
493 const std::string &filepath);
494
495 // To and from UTF-8 unicode conversion functions
496
497 // Convert a unicode code point into a UTF-8 representation by appending it
498 // to a string. Returns the number of bytes generated.
ToUTF8(uint32_t ucc,std::string * out)499 inline int ToUTF8(uint32_t ucc, std::string *out) {
500 FLATBUFFERS_ASSERT(!(ucc & 0x80000000)); // Top bit can't be set.
501 // 6 possible encodings: http://en.wikipedia.org/wiki/UTF-8
502 for (int i = 0; i < 6; i++) {
503 // Max bits this encoding can represent.
504 uint32_t max_bits = 6 + i * 5 + static_cast<int>(!i);
505 if (ucc < (1u << max_bits)) { // does it fit?
506 // Remaining bits not encoded in the first byte, store 6 bits each
507 uint32_t remain_bits = i * 6;
508 // Store first byte:
509 (*out) += static_cast<char>((0xFE << (max_bits - remain_bits)) |
510 (ucc >> remain_bits));
511 // Store remaining bytes:
512 for (int j = i - 1; j >= 0; j--) {
513 (*out) += static_cast<char>(((ucc >> (j * 6)) & 0x3F) | 0x80);
514 }
515 return i + 1; // Return the number of bytes added.
516 }
517 }
518 FLATBUFFERS_ASSERT(0); // Impossible to arrive here.
519 return -1;
520 }
521
522 // Converts whatever prefix of the incoming string corresponds to a valid
523 // UTF-8 sequence into a unicode code. The incoming pointer will have been
524 // advanced past all bytes parsed.
525 // returns -1 upon corrupt UTF-8 encoding (ignore the incoming pointer in
526 // this case).
FromUTF8(const char ** in)527 inline int FromUTF8(const char **in) {
528 int len = 0;
529 // Count leading 1 bits.
530 for (int mask = 0x80; mask >= 0x04; mask >>= 1) {
531 if (**in & mask) {
532 len++;
533 } else {
534 break;
535 }
536 }
537 if ((static_cast<unsigned char>(**in) << len) & 0x80)
538 return -1; // Bit after leading 1's must be 0.
539 if (!len) return *(*in)++;
540 // UTF-8 encoded values with a length are between 2 and 4 bytes.
541 if (len < 2 || len > 4) { return -1; }
542 // Grab initial bits of the code.
543 int ucc = *(*in)++ & ((1 << (7 - len)) - 1);
544 for (int i = 0; i < len - 1; i++) {
545 if ((**in & 0xC0) != 0x80) return -1; // Upper bits must 1 0.
546 ucc <<= 6;
547 ucc |= *(*in)++ & 0x3F; // Grab 6 more bits of the code.
548 }
549 // UTF-8 cannot encode values between 0xD800 and 0xDFFF (reserved for
550 // UTF-16 surrogate pairs).
551 if (ucc >= 0xD800 && ucc <= 0xDFFF) { return -1; }
552 // UTF-8 must represent code points in their shortest possible encoding.
553 switch (len) {
554 case 2:
555 // Two bytes of UTF-8 can represent code points from U+0080 to U+07FF.
556 if (ucc < 0x0080 || ucc > 0x07FF) { return -1; }
557 break;
558 case 3:
559 // Three bytes of UTF-8 can represent code points from U+0800 to U+FFFF.
560 if (ucc < 0x0800 || ucc > 0xFFFF) { return -1; }
561 break;
562 case 4:
563 // Four bytes of UTF-8 can represent code points from U+10000 to U+10FFFF.
564 if (ucc < 0x10000 || ucc > 0x10FFFF) { return -1; }
565 break;
566 }
567 return ucc;
568 }
569
570 #ifndef FLATBUFFERS_PREFER_PRINTF
571 // Wraps a string to a maximum length, inserting new lines where necessary. Any
572 // existing whitespace will be collapsed down to a single space. A prefix or
573 // suffix can be provided, which will be inserted before or after a wrapped
574 // line, respectively.
WordWrap(const std::string in,size_t max_length,const std::string wrapped_line_prefix,const std::string wrapped_line_suffix)575 inline std::string WordWrap(const std::string in, size_t max_length,
576 const std::string wrapped_line_prefix,
577 const std::string wrapped_line_suffix) {
578 std::istringstream in_stream(in);
579 std::string wrapped, line, word;
580
581 in_stream >> word;
582 line = word;
583
584 while (in_stream >> word) {
585 if ((line.length() + 1 + word.length() + wrapped_line_suffix.length()) <
586 max_length) {
587 line += " " + word;
588 } else {
589 wrapped += line + wrapped_line_suffix + "\n";
590 line = wrapped_line_prefix + word;
591 }
592 }
593 wrapped += line;
594
595 return wrapped;
596 }
597 #endif // !FLATBUFFERS_PREFER_PRINTF
598
EscapeString(const char * s,size_t length,std::string * _text,bool allow_non_utf8,bool natural_utf8)599 inline bool EscapeString(const char *s, size_t length, std::string *_text,
600 bool allow_non_utf8, bool natural_utf8) {
601 std::string &text = *_text;
602 text += "\"";
603 for (uoffset_t i = 0; i < length; i++) {
604 char c = s[i];
605 switch (c) {
606 case '\n': text += "\\n"; break;
607 case '\t': text += "\\t"; break;
608 case '\r': text += "\\r"; break;
609 case '\b': text += "\\b"; break;
610 case '\f': text += "\\f"; break;
611 case '\"': text += "\\\""; break;
612 case '\\': text += "\\\\"; break;
613 default:
614 if (c >= ' ' && c <= '~') {
615 text += c;
616 } else {
617 // Not printable ASCII data. Let's see if it's valid UTF-8 first:
618 const char *utf8 = s + i;
619 int ucc = FromUTF8(&utf8);
620 if (ucc < 0) {
621 if (allow_non_utf8) {
622 text += "\\x";
623 text += IntToStringHex(static_cast<uint8_t>(c), 2);
624 } else {
625 // There are two cases here:
626 //
627 // 1) We reached here by parsing an IDL file. In that case,
628 // we previously checked for non-UTF-8, so we shouldn't reach
629 // here.
630 //
631 // 2) We reached here by someone calling GenText()
632 // on a previously-serialized flatbuffer. The data might have
633 // non-UTF-8 Strings, or might be corrupt.
634 //
635 // In both cases, we have to give up and inform the caller
636 // they have no JSON.
637 return false;
638 }
639 } else {
640 if (natural_utf8) {
641 // utf8 points to past all utf-8 bytes parsed
642 text.append(s + i, static_cast<size_t>(utf8 - s - i));
643 } else if (ucc <= 0xFFFF) {
644 // Parses as Unicode within JSON's \uXXXX range, so use that.
645 text += "\\u";
646 text += IntToStringHex(ucc, 4);
647 } else if (ucc <= 0x10FFFF) {
648 // Encode Unicode SMP values to a surrogate pair using two \u
649 // escapes.
650 uint32_t base = ucc - 0x10000;
651 auto high_surrogate = (base >> 10) + 0xD800;
652 auto low_surrogate = (base & 0x03FF) + 0xDC00;
653 text += "\\u";
654 text += IntToStringHex(high_surrogate, 4);
655 text += "\\u";
656 text += IntToStringHex(low_surrogate, 4);
657 }
658 // Skip past characters recognized.
659 i = static_cast<uoffset_t>(utf8 - s - 1);
660 }
661 }
662 break;
663 }
664 }
665 text += "\"";
666 return true;
667 }
668
BufferToHexText(const void * buffer,size_t buffer_size,size_t max_length,const std::string & wrapped_line_prefix,const std::string & wrapped_line_suffix)669 inline std::string BufferToHexText(const void *buffer, size_t buffer_size,
670 size_t max_length,
671 const std::string &wrapped_line_prefix,
672 const std::string &wrapped_line_suffix) {
673 std::string text = wrapped_line_prefix;
674 size_t start_offset = 0;
675 const char *s = reinterpret_cast<const char *>(buffer);
676 for (size_t i = 0; s && i < buffer_size; i++) {
677 // Last iteration or do we have more?
678 bool have_more = i + 1 < buffer_size;
679 text += "0x";
680 text += IntToStringHex(static_cast<uint8_t>(s[i]), 2);
681 if (have_more) { text += ','; }
682 // If we have more to process and we reached max_length
683 if (have_more &&
684 text.size() + wrapped_line_suffix.size() >= start_offset + max_length) {
685 text += wrapped_line_suffix;
686 text += '\n';
687 start_offset = text.size();
688 text += wrapped_line_prefix;
689 }
690 }
691 text += wrapped_line_suffix;
692 return text;
693 }
694
695 // Remove paired quotes in a string: "text"|'text' -> text.
696 std::string RemoveStringQuotes(const std::string &s);
697
698 // Change th global C-locale to locale with name <locale_name>.
699 // Returns an actual locale name in <_value>, useful if locale_name is "" or
700 // null.
701 bool SetGlobalTestLocale(const char *locale_name,
702 std::string *_value = nullptr);
703
704 // Read (or test) a value of environment variable.
705 bool ReadEnvironmentVariable(const char *var_name,
706 std::string *_value = nullptr);
707
708 enum class Case {
709 kUnknown = 0,
710 // TheQuickBrownFox
711 kUpperCamel = 1,
712 // theQuickBrownFox
713 kLowerCamel = 2,
714 // the_quick_brown_fox
715 kSnake = 3,
716 // THE_QUICK_BROWN_FOX
717 kScreamingSnake = 4,
718 // THEQUICKBROWNFOX
719 kAllUpper = 5,
720 // thequickbrownfox
721 kAllLower = 6,
722 // the-quick-brown-fox
723 kDasher = 7,
724 // THEQuiCKBr_ownFox (or whatever you want, we won't change it)
725 kKeep = 8,
726 // the_quick_brown_fox123 (as opposed to the_quick_brown_fox_123)
727 kSnake2 = 9,
728 };
729
730 // Convert the `input` string of case `input_case` to the specified
731 // `output_case`.
732 std::string ConvertCase(const std::string &input, Case output_case,
733 Case input_case = Case::kSnake);
734
735 } // namespace flatbuffers
736
737 #endif // FLATBUFFERS_UTIL_H_
738