1 | // __ _____ _____ _____ |
2 | // __| | __| | | | JSON for Modern C++ |
3 | // | | |__ | | | | | | version 3.11.3 |
4 | // |_____|_____|_____|_|___| https://github.com/nlohmann/json |
5 | // |
6 | // SPDX-FileCopyrightText: 2008-2009 Björn Hoehrmann <bjoern@hoehrmann.de> |
7 | // SPDX-FileCopyrightText: 2013-2023 Niels Lohmann <https://nlohmann.me> |
8 | // SPDX-License-Identifier: MIT |
9 | |
10 | #pragma once |
11 | |
12 | #include <algorithm> // reverse, remove, fill, find, none_of |
13 | #include <array> // array |
14 | #include <clocale> // localeconv, lconv |
15 | #include <cmath> // labs, isfinite, isnan, signbit |
16 | #include <cstddef> // size_t, ptrdiff_t |
17 | #include <cstdint> // uint8_t |
18 | #include <cstdio> // snprintf |
19 | #include <limits> // numeric_limits |
20 | #include <string> // string, char_traits |
21 | #include <iomanip> // setfill, setw |
22 | #include <type_traits> // is_same |
23 | #include <utility> // move |
24 | |
25 | #include <nlohmann/detail/conversions/to_chars.hpp> |
26 | #include <nlohmann/detail/exceptions.hpp> |
27 | #include <nlohmann/detail/macro_scope.hpp> |
28 | #include <nlohmann/detail/meta/cpp_future.hpp> |
29 | #include <nlohmann/detail/output/binary_writer.hpp> |
30 | #include <nlohmann/detail/output/output_adapters.hpp> |
31 | #include <nlohmann/detail/string_concat.hpp> |
32 | #include <nlohmann/detail/value_t.hpp> |
33 | |
34 | NLOHMANN_JSON_NAMESPACE_BEGIN |
35 | namespace detail |
36 | { |
37 | |
38 | /////////////////// |
39 | // serialization // |
40 | /////////////////// |
41 | |
42 | /// how to treat decoding errors |
43 | enum class error_handler_t |
44 | { |
45 | strict, ///< throw a type_error exception in case of invalid UTF-8 |
46 | replace, ///< replace invalid UTF-8 sequences with U+FFFD |
47 | ignore ///< ignore invalid UTF-8 sequences |
48 | }; |
49 | |
50 | template<typename BasicJsonType> |
51 | class serializer |
52 | { |
53 | using string_t = typename BasicJsonType::string_t; |
54 | using number_float_t = typename BasicJsonType::number_float_t; |
55 | using number_integer_t = typename BasicJsonType::number_integer_t; |
56 | using number_unsigned_t = typename BasicJsonType::number_unsigned_t; |
57 | using binary_char_t = typename BasicJsonType::binary_t::value_type; |
58 | static constexpr std::uint8_t UTF8_ACCEPT = 0; |
59 | static constexpr std::uint8_t UTF8_REJECT = 1; |
60 | |
61 | public: |
62 | /*! |
63 | @param[in] s output stream to serialize to |
64 | @param[in] ichar indentation character to use |
65 | @param[in] error_handler_ how to react on decoding errors |
66 | */ |
67 | serializer(output_adapter_t<char> s, const char ichar, |
68 | error_handler_t error_handler_ = error_handler_t::strict) |
69 | : o(std::move(s)) |
70 | , loc(std::localeconv()) |
71 | , thousands_sep(loc->thousands_sep == nullptr ? '\0' : std::char_traits<char>::to_char_type(c: * (loc->thousands_sep))) |
72 | , decimal_point(loc->decimal_point == nullptr ? '\0' : std::char_traits<char>::to_char_type(c: * (loc->decimal_point))) |
73 | , indent_char(ichar) |
74 | , indent_string(512, indent_char) |
75 | , error_handler(error_handler_) |
76 | {} |
77 | |
78 | // delete because of pointer members |
79 | serializer(const serializer&) = delete; |
80 | serializer& operator=(const serializer&) = delete; |
81 | serializer(serializer&&) = delete; |
82 | serializer& operator=(serializer&&) = delete; |
83 | ~serializer() = default; |
84 | |
85 | /*! |
86 | @brief internal implementation of the serialization function |
87 | |
88 | This function is called by the public member function dump and organizes |
89 | the serialization internally. The indentation level is propagated as |
90 | additional parameter. In case of arrays and objects, the function is |
91 | called recursively. |
92 | |
93 | - strings and object keys are escaped using `escape_string()` |
94 | - integer numbers are converted implicitly via `operator<<` |
95 | - floating-point numbers are converted to a string using `"%g"` format |
96 | - binary values are serialized as objects containing the subtype and the |
97 | byte array |
98 | |
99 | @param[in] val value to serialize |
100 | @param[in] pretty_print whether the output shall be pretty-printed |
101 | @param[in] ensure_ascii If @a ensure_ascii is true, all non-ASCII characters |
102 | in the output are escaped with `\uXXXX` sequences, and the result consists |
103 | of ASCII characters only. |
104 | @param[in] indent_step the indent level |
105 | @param[in] current_indent the current indent level (only used internally) |
106 | */ |
107 | void dump(const BasicJsonType& val, |
108 | const bool pretty_print, |
109 | const bool ensure_ascii, |
110 | const unsigned int indent_step, |
111 | const unsigned int current_indent = 0) |
112 | { |
113 | switch (val.m_data.m_type) |
114 | { |
115 | case value_t::object: |
116 | { |
117 | if (val.m_data.m_value.object->empty()) |
118 | { |
119 | o->write_characters(s: "{}" , length: 2); |
120 | return; |
121 | } |
122 | |
123 | if (pretty_print) |
124 | { |
125 | o->write_characters(s: "{\n" , length: 2); |
126 | |
127 | // variable to hold indentation for recursive calls |
128 | const auto new_indent = current_indent + indent_step; |
129 | if (JSON_HEDLEY_UNLIKELY(indent_string.size() < new_indent)) |
130 | { |
131 | indent_string.resize(indent_string.size() * 2, ' '); |
132 | } |
133 | |
134 | // first n-1 elements |
135 | auto i = val.m_data.m_value.object->cbegin(); |
136 | for (std::size_t cnt = 0; cnt < val.m_data.m_value.object->size() - 1; ++cnt, ++i) |
137 | { |
138 | o->write_characters(s: indent_string.c_str(), length: new_indent); |
139 | o->write_character(c: '\"'); |
140 | dump_escaped(s: i->first, ensure_ascii); |
141 | o->write_characters(s: "\": " , length: 3); |
142 | dump(val: i->second, pretty_print: true, ensure_ascii, indent_step, current_indent: new_indent); |
143 | o->write_characters(s: ",\n" , length: 2); |
144 | } |
145 | |
146 | // last element |
147 | JSON_ASSERT(i != val.m_data.m_value.object->cend()); |
148 | JSON_ASSERT(std::next(i) == val.m_data.m_value.object->cend()); |
149 | o->write_characters(s: indent_string.c_str(), length: new_indent); |
150 | o->write_character(c: '\"'); |
151 | dump_escaped(s: i->first, ensure_ascii); |
152 | o->write_characters(s: "\": " , length: 3); |
153 | dump(val: i->second, pretty_print: true, ensure_ascii, indent_step, current_indent: new_indent); |
154 | |
155 | o->write_character(c: '\n'); |
156 | o->write_characters(s: indent_string.c_str(), length: current_indent); |
157 | o->write_character(c: '}'); |
158 | } |
159 | else |
160 | { |
161 | o->write_character(c: '{'); |
162 | |
163 | // first n-1 elements |
164 | auto i = val.m_data.m_value.object->cbegin(); |
165 | for (std::size_t cnt = 0; cnt < val.m_data.m_value.object->size() - 1; ++cnt, ++i) |
166 | { |
167 | o->write_character(c: '\"'); |
168 | dump_escaped(s: i->first, ensure_ascii); |
169 | o->write_characters(s: "\":" , length: 2); |
170 | dump(val: i->second, pretty_print: false, ensure_ascii, indent_step, current_indent); |
171 | o->write_character(c: ','); |
172 | } |
173 | |
174 | // last element |
175 | JSON_ASSERT(i != val.m_data.m_value.object->cend()); |
176 | JSON_ASSERT(std::next(i) == val.m_data.m_value.object->cend()); |
177 | o->write_character(c: '\"'); |
178 | dump_escaped(s: i->first, ensure_ascii); |
179 | o->write_characters(s: "\":" , length: 2); |
180 | dump(val: i->second, pretty_print: false, ensure_ascii, indent_step, current_indent); |
181 | |
182 | o->write_character(c: '}'); |
183 | } |
184 | |
185 | return; |
186 | } |
187 | |
188 | case value_t::array: |
189 | { |
190 | if (val.m_data.m_value.array->empty()) |
191 | { |
192 | o->write_characters(s: "[]" , length: 2); |
193 | return; |
194 | } |
195 | |
196 | if (pretty_print) |
197 | { |
198 | o->write_characters(s: "[\n" , length: 2); |
199 | |
200 | // variable to hold indentation for recursive calls |
201 | const auto new_indent = current_indent + indent_step; |
202 | if (JSON_HEDLEY_UNLIKELY(indent_string.size() < new_indent)) |
203 | { |
204 | indent_string.resize(indent_string.size() * 2, ' '); |
205 | } |
206 | |
207 | // first n-1 elements |
208 | for (auto i = val.m_data.m_value.array->cbegin(); |
209 | i != val.m_data.m_value.array->cend() - 1; ++i) |
210 | { |
211 | o->write_characters(s: indent_string.c_str(), length: new_indent); |
212 | dump(val: *i, pretty_print: true, ensure_ascii, indent_step, current_indent: new_indent); |
213 | o->write_characters(s: ",\n" , length: 2); |
214 | } |
215 | |
216 | // last element |
217 | JSON_ASSERT(!val.m_data.m_value.array->empty()); |
218 | o->write_characters(s: indent_string.c_str(), length: new_indent); |
219 | dump(val: val.m_data.m_value.array->back(), pretty_print: true, ensure_ascii, indent_step, current_indent: new_indent); |
220 | |
221 | o->write_character(c: '\n'); |
222 | o->write_characters(s: indent_string.c_str(), length: current_indent); |
223 | o->write_character(c: ']'); |
224 | } |
225 | else |
226 | { |
227 | o->write_character(c: '['); |
228 | |
229 | // first n-1 elements |
230 | for (auto i = val.m_data.m_value.array->cbegin(); |
231 | i != val.m_data.m_value.array->cend() - 1; ++i) |
232 | { |
233 | dump(val: *i, pretty_print: false, ensure_ascii, indent_step, current_indent); |
234 | o->write_character(c: ','); |
235 | } |
236 | |
237 | // last element |
238 | JSON_ASSERT(!val.m_data.m_value.array->empty()); |
239 | dump(val: val.m_data.m_value.array->back(), pretty_print: false, ensure_ascii, indent_step, current_indent); |
240 | |
241 | o->write_character(c: ']'); |
242 | } |
243 | |
244 | return; |
245 | } |
246 | |
247 | case value_t::string: |
248 | { |
249 | o->write_character(c: '\"'); |
250 | dump_escaped(s: *val.m_data.m_value.string, ensure_ascii); |
251 | o->write_character(c: '\"'); |
252 | return; |
253 | } |
254 | |
255 | case value_t::binary: |
256 | { |
257 | if (pretty_print) |
258 | { |
259 | o->write_characters(s: "{\n" , length: 2); |
260 | |
261 | // variable to hold indentation for recursive calls |
262 | const auto new_indent = current_indent + indent_step; |
263 | if (JSON_HEDLEY_UNLIKELY(indent_string.size() < new_indent)) |
264 | { |
265 | indent_string.resize(indent_string.size() * 2, ' '); |
266 | } |
267 | |
268 | o->write_characters(s: indent_string.c_str(), length: new_indent); |
269 | |
270 | o->write_characters(s: "\"bytes\": [" , length: 10); |
271 | |
272 | if (!val.m_data.m_value.binary->empty()) |
273 | { |
274 | for (auto i = val.m_data.m_value.binary->cbegin(); |
275 | i != val.m_data.m_value.binary->cend() - 1; ++i) |
276 | { |
277 | dump_integer(*i); |
278 | o->write_characters(s: ", " , length: 2); |
279 | } |
280 | dump_integer(val.m_data.m_value.binary->back()); |
281 | } |
282 | |
283 | o->write_characters(s: "],\n" , length: 3); |
284 | o->write_characters(s: indent_string.c_str(), length: new_indent); |
285 | |
286 | o->write_characters(s: "\"subtype\": " , length: 11); |
287 | if (val.m_data.m_value.binary->has_subtype()) |
288 | { |
289 | dump_integer(val.m_data.m_value.binary->subtype()); |
290 | } |
291 | else |
292 | { |
293 | o->write_characters(s: "null" , length: 4); |
294 | } |
295 | o->write_character(c: '\n'); |
296 | o->write_characters(s: indent_string.c_str(), length: current_indent); |
297 | o->write_character(c: '}'); |
298 | } |
299 | else |
300 | { |
301 | o->write_characters(s: "{\"bytes\":[" , length: 10); |
302 | |
303 | if (!val.m_data.m_value.binary->empty()) |
304 | { |
305 | for (auto i = val.m_data.m_value.binary->cbegin(); |
306 | i != val.m_data.m_value.binary->cend() - 1; ++i) |
307 | { |
308 | dump_integer(*i); |
309 | o->write_character(c: ','); |
310 | } |
311 | dump_integer(val.m_data.m_value.binary->back()); |
312 | } |
313 | |
314 | o->write_characters(s: "],\"subtype\":" , length: 12); |
315 | if (val.m_data.m_value.binary->has_subtype()) |
316 | { |
317 | dump_integer(val.m_data.m_value.binary->subtype()); |
318 | o->write_character(c: '}'); |
319 | } |
320 | else |
321 | { |
322 | o->write_characters(s: "null}" , length: 5); |
323 | } |
324 | } |
325 | return; |
326 | } |
327 | |
328 | case value_t::boolean: |
329 | { |
330 | if (val.m_data.m_value.boolean) |
331 | { |
332 | o->write_characters(s: "true" , length: 4); |
333 | } |
334 | else |
335 | { |
336 | o->write_characters(s: "false" , length: 5); |
337 | } |
338 | return; |
339 | } |
340 | |
341 | case value_t::number_integer: |
342 | { |
343 | dump_integer(val.m_data.m_value.number_integer); |
344 | return; |
345 | } |
346 | |
347 | case value_t::number_unsigned: |
348 | { |
349 | dump_integer(val.m_data.m_value.number_unsigned); |
350 | return; |
351 | } |
352 | |
353 | case value_t::number_float: |
354 | { |
355 | dump_float(val.m_data.m_value.number_float); |
356 | return; |
357 | } |
358 | |
359 | case value_t::discarded: |
360 | { |
361 | o->write_characters(s: "<discarded>" , length: 11); |
362 | return; |
363 | } |
364 | |
365 | case value_t::null: |
366 | { |
367 | o->write_characters(s: "null" , length: 4); |
368 | return; |
369 | } |
370 | |
371 | default: // LCOV_EXCL_LINE |
372 | JSON_ASSERT(false); // NOLINT(cert-dcl03-c,hicpp-static-assert,misc-static-assert) LCOV_EXCL_LINE |
373 | } |
374 | } |
375 | |
376 | JSON_PRIVATE_UNLESS_TESTED: |
377 | /*! |
378 | @brief dump escaped string |
379 | |
380 | Escape a string by replacing certain special characters by a sequence of an |
381 | escape character (backslash) and another character and other control |
382 | characters by a sequence of "\u" followed by a four-digit hex |
383 | representation. The escaped string is written to output stream @a o. |
384 | |
385 | @param[in] s the string to escape |
386 | @param[in] ensure_ascii whether to escape non-ASCII characters with |
387 | \uXXXX sequences |
388 | |
389 | @complexity Linear in the length of string @a s. |
390 | */ |
391 | void dump_escaped(const string_t& s, const bool ensure_ascii) |
392 | { |
393 | std::uint32_t codepoint{}; |
394 | std::uint8_t state = UTF8_ACCEPT; |
395 | std::size_t bytes = 0; // number of bytes written to string_buffer |
396 | |
397 | // number of bytes written at the point of the last valid byte |
398 | std::size_t bytes_after_last_accept = 0; |
399 | std::size_t undumped_chars = 0; |
400 | |
401 | for (std::size_t i = 0; i < s.size(); ++i) |
402 | { |
403 | const auto byte = static_cast<std::uint8_t>(s[i]); |
404 | |
405 | switch (decode(state, codep&: codepoint, byte)) |
406 | { |
407 | case UTF8_ACCEPT: // decode found a new code point |
408 | { |
409 | switch (codepoint) |
410 | { |
411 | case 0x08: // backspace |
412 | { |
413 | string_buffer[bytes++] = '\\'; |
414 | string_buffer[bytes++] = 'b'; |
415 | break; |
416 | } |
417 | |
418 | case 0x09: // horizontal tab |
419 | { |
420 | string_buffer[bytes++] = '\\'; |
421 | string_buffer[bytes++] = 't'; |
422 | break; |
423 | } |
424 | |
425 | case 0x0A: // newline |
426 | { |
427 | string_buffer[bytes++] = '\\'; |
428 | string_buffer[bytes++] = 'n'; |
429 | break; |
430 | } |
431 | |
432 | case 0x0C: // formfeed |
433 | { |
434 | string_buffer[bytes++] = '\\'; |
435 | string_buffer[bytes++] = 'f'; |
436 | break; |
437 | } |
438 | |
439 | case 0x0D: // carriage return |
440 | { |
441 | string_buffer[bytes++] = '\\'; |
442 | string_buffer[bytes++] = 'r'; |
443 | break; |
444 | } |
445 | |
446 | case 0x22: // quotation mark |
447 | { |
448 | string_buffer[bytes++] = '\\'; |
449 | string_buffer[bytes++] = '\"'; |
450 | break; |
451 | } |
452 | |
453 | case 0x5C: // reverse solidus |
454 | { |
455 | string_buffer[bytes++] = '\\'; |
456 | string_buffer[bytes++] = '\\'; |
457 | break; |
458 | } |
459 | |
460 | default: |
461 | { |
462 | // escape control characters (0x00..0x1F) or, if |
463 | // ensure_ascii parameter is used, non-ASCII characters |
464 | if ((codepoint <= 0x1F) || (ensure_ascii && (codepoint >= 0x7F))) |
465 | { |
466 | if (codepoint <= 0xFFFF) |
467 | { |
468 | // NOLINTNEXTLINE(cppcoreguidelines-pro-type-vararg,hicpp-vararg) |
469 | static_cast<void>((std::snprintf)(s: string_buffer.data() + bytes, maxlen: 7, format: "\\u%04x" , |
470 | static_cast<std::uint16_t>(codepoint))); |
471 | bytes += 6; |
472 | } |
473 | else |
474 | { |
475 | // NOLINTNEXTLINE(cppcoreguidelines-pro-type-vararg,hicpp-vararg) |
476 | static_cast<void>((std::snprintf)(s: string_buffer.data() + bytes, maxlen: 13, format: "\\u%04x\\u%04x" , |
477 | static_cast<std::uint16_t>(0xD7C0u + (codepoint >> 10u)), |
478 | static_cast<std::uint16_t>(0xDC00u + (codepoint & 0x3FFu)))); |
479 | bytes += 12; |
480 | } |
481 | } |
482 | else |
483 | { |
484 | // copy byte to buffer (all previous bytes |
485 | // been copied have in default case above) |
486 | string_buffer[bytes++] = s[i]; |
487 | } |
488 | break; |
489 | } |
490 | } |
491 | |
492 | // write buffer and reset index; there must be 13 bytes |
493 | // left, as this is the maximal number of bytes to be |
494 | // written ("\uxxxx\uxxxx\0") for one code point |
495 | if (string_buffer.size() - bytes < 13) |
496 | { |
497 | o->write_characters(s: string_buffer.data(), length: bytes); |
498 | bytes = 0; |
499 | } |
500 | |
501 | // remember the byte position of this accept |
502 | bytes_after_last_accept = bytes; |
503 | undumped_chars = 0; |
504 | break; |
505 | } |
506 | |
507 | case UTF8_REJECT: // decode found invalid UTF-8 byte |
508 | { |
509 | switch (error_handler) |
510 | { |
511 | case error_handler_t::strict: |
512 | { |
513 | JSON_THROW(type_error::create(316, concat("invalid UTF-8 byte at index " , std::to_string(i), ": 0x" , hex_bytes(byte | 0)), nullptr)); |
514 | } |
515 | |
516 | case error_handler_t::ignore: |
517 | case error_handler_t::replace: |
518 | { |
519 | // in case we saw this character the first time, we |
520 | // would like to read it again, because the byte |
521 | // may be OK for itself, but just not OK for the |
522 | // previous sequence |
523 | if (undumped_chars > 0) |
524 | { |
525 | --i; |
526 | } |
527 | |
528 | // reset length buffer to the last accepted index; |
529 | // thus removing/ignoring the invalid characters |
530 | bytes = bytes_after_last_accept; |
531 | |
532 | if (error_handler == error_handler_t::replace) |
533 | { |
534 | // add a replacement character |
535 | if (ensure_ascii) |
536 | { |
537 | string_buffer[bytes++] = '\\'; |
538 | string_buffer[bytes++] = 'u'; |
539 | string_buffer[bytes++] = 'f'; |
540 | string_buffer[bytes++] = 'f'; |
541 | string_buffer[bytes++] = 'f'; |
542 | string_buffer[bytes++] = 'd'; |
543 | } |
544 | else |
545 | { |
546 | string_buffer[bytes++] = detail::binary_writer<BasicJsonType, char>::to_char_type('\xEF'); |
547 | string_buffer[bytes++] = detail::binary_writer<BasicJsonType, char>::to_char_type('\xBF'); |
548 | string_buffer[bytes++] = detail::binary_writer<BasicJsonType, char>::to_char_type('\xBD'); |
549 | } |
550 | |
551 | // write buffer and reset index; there must be 13 bytes |
552 | // left, as this is the maximal number of bytes to be |
553 | // written ("\uxxxx\uxxxx\0") for one code point |
554 | if (string_buffer.size() - bytes < 13) |
555 | { |
556 | o->write_characters(s: string_buffer.data(), length: bytes); |
557 | bytes = 0; |
558 | } |
559 | |
560 | bytes_after_last_accept = bytes; |
561 | } |
562 | |
563 | undumped_chars = 0; |
564 | |
565 | // continue processing the string |
566 | state = UTF8_ACCEPT; |
567 | break; |
568 | } |
569 | |
570 | default: // LCOV_EXCL_LINE |
571 | JSON_ASSERT(false); // NOLINT(cert-dcl03-c,hicpp-static-assert,misc-static-assert) LCOV_EXCL_LINE |
572 | } |
573 | break; |
574 | } |
575 | |
576 | default: // decode found yet incomplete multi-byte code point |
577 | { |
578 | if (!ensure_ascii) |
579 | { |
580 | // code point will not be escaped - copy byte to buffer |
581 | string_buffer[bytes++] = s[i]; |
582 | } |
583 | ++undumped_chars; |
584 | break; |
585 | } |
586 | } |
587 | } |
588 | |
589 | // we finished processing the string |
590 | if (JSON_HEDLEY_LIKELY(state == UTF8_ACCEPT)) |
591 | { |
592 | // write buffer |
593 | if (bytes > 0) |
594 | { |
595 | o->write_characters(s: string_buffer.data(), length: bytes); |
596 | } |
597 | } |
598 | else |
599 | { |
600 | // we finish reading, but do not accept: string was incomplete |
601 | switch (error_handler) |
602 | { |
603 | case error_handler_t::strict: |
604 | { |
605 | JSON_THROW(type_error::create(316, concat("incomplete UTF-8 string; last byte: 0x" , hex_bytes(static_cast<std::uint8_t>(s.back() | 0))), nullptr)); |
606 | } |
607 | |
608 | case error_handler_t::ignore: |
609 | { |
610 | // write all accepted bytes |
611 | o->write_characters(s: string_buffer.data(), length: bytes_after_last_accept); |
612 | break; |
613 | } |
614 | |
615 | case error_handler_t::replace: |
616 | { |
617 | // write all accepted bytes |
618 | o->write_characters(s: string_buffer.data(), length: bytes_after_last_accept); |
619 | // add a replacement character |
620 | if (ensure_ascii) |
621 | { |
622 | o->write_characters(s: "\\ufffd" , length: 6); |
623 | } |
624 | else |
625 | { |
626 | o->write_characters(s: "\xEF\xBF\xBD" , length: 3); |
627 | } |
628 | break; |
629 | } |
630 | |
631 | default: // LCOV_EXCL_LINE |
632 | JSON_ASSERT(false); // NOLINT(cert-dcl03-c,hicpp-static-assert,misc-static-assert) LCOV_EXCL_LINE |
633 | } |
634 | } |
635 | } |
636 | |
637 | private: |
638 | /*! |
639 | @brief count digits |
640 | |
641 | Count the number of decimal (base 10) digits for an input unsigned integer. |
642 | |
643 | @param[in] x unsigned integer number to count its digits |
644 | @return number of decimal digits |
645 | */ |
646 | inline unsigned int count_digits(number_unsigned_t x) noexcept |
647 | { |
648 | unsigned int n_digits = 1; |
649 | for (;;) |
650 | { |
651 | if (x < 10) |
652 | { |
653 | return n_digits; |
654 | } |
655 | if (x < 100) |
656 | { |
657 | return n_digits + 1; |
658 | } |
659 | if (x < 1000) |
660 | { |
661 | return n_digits + 2; |
662 | } |
663 | if (x < 10000) |
664 | { |
665 | return n_digits + 3; |
666 | } |
667 | x = x / 10000u; |
668 | n_digits += 4; |
669 | } |
670 | } |
671 | |
672 | /*! |
673 | * @brief convert a byte to a uppercase hex representation |
674 | * @param[in] byte byte to represent |
675 | * @return representation ("00".."FF") |
676 | */ |
677 | static std::string hex_bytes(std::uint8_t byte) |
678 | { |
679 | std::string result = "FF" ; |
680 | constexpr const char* nibble_to_hex = "0123456789ABCDEF" ; |
681 | result[0] = nibble_to_hex[byte / 16]; |
682 | result[1] = nibble_to_hex[byte % 16]; |
683 | return result; |
684 | } |
685 | |
686 | // templates to avoid warnings about useless casts |
687 | template <typename NumberType, enable_if_t<std::is_signed<NumberType>::value, int> = 0> |
688 | bool is_negative_number(NumberType x) |
689 | { |
690 | return x < 0; |
691 | } |
692 | |
693 | template < typename NumberType, enable_if_t <std::is_unsigned<NumberType>::value, int > = 0 > |
694 | bool is_negative_number(NumberType /*unused*/) |
695 | { |
696 | return false; |
697 | } |
698 | |
699 | /*! |
700 | @brief dump an integer |
701 | |
702 | Dump a given integer to output stream @a o. Works internally with |
703 | @a number_buffer. |
704 | |
705 | @param[in] x integer number (signed or unsigned) to dump |
706 | @tparam NumberType either @a number_integer_t or @a number_unsigned_t |
707 | */ |
708 | template < typename NumberType, detail::enable_if_t < |
709 | std::is_integral<NumberType>::value || |
710 | std::is_same<NumberType, number_unsigned_t>::value || |
711 | std::is_same<NumberType, number_integer_t>::value || |
712 | std::is_same<NumberType, binary_char_t>::value, |
713 | int > = 0 > |
714 | void dump_integer(NumberType x) |
715 | { |
716 | static constexpr std::array<std::array<char, 2>, 100> digits_to_99 |
717 | { |
718 | ._M_elems: { |
719 | {._M_elems: {'0', '0'}}, {._M_elems: {'0', '1'}}, {._M_elems: {'0', '2'}}, {._M_elems: {'0', '3'}}, {._M_elems: {'0', '4'}}, {._M_elems: {'0', '5'}}, {._M_elems: {'0', '6'}}, {._M_elems: {'0', '7'}}, {._M_elems: {'0', '8'}}, {._M_elems: {'0', '9'}}, |
720 | {._M_elems: {'1', '0'}}, {._M_elems: {'1', '1'}}, {._M_elems: {'1', '2'}}, {._M_elems: {'1', '3'}}, {._M_elems: {'1', '4'}}, {._M_elems: {'1', '5'}}, {._M_elems: {'1', '6'}}, {._M_elems: {'1', '7'}}, {._M_elems: {'1', '8'}}, {._M_elems: {'1', '9'}}, |
721 | {._M_elems: {'2', '0'}}, {._M_elems: {'2', '1'}}, {._M_elems: {'2', '2'}}, {._M_elems: {'2', '3'}}, {._M_elems: {'2', '4'}}, {._M_elems: {'2', '5'}}, {._M_elems: {'2', '6'}}, {._M_elems: {'2', '7'}}, {._M_elems: {'2', '8'}}, {._M_elems: {'2', '9'}}, |
722 | {._M_elems: {'3', '0'}}, {._M_elems: {'3', '1'}}, {._M_elems: {'3', '2'}}, {._M_elems: {'3', '3'}}, {._M_elems: {'3', '4'}}, {._M_elems: {'3', '5'}}, {._M_elems: {'3', '6'}}, {._M_elems: {'3', '7'}}, {._M_elems: {'3', '8'}}, {._M_elems: {'3', '9'}}, |
723 | {._M_elems: {'4', '0'}}, {._M_elems: {'4', '1'}}, {._M_elems: {'4', '2'}}, {._M_elems: {'4', '3'}}, {._M_elems: {'4', '4'}}, {._M_elems: {'4', '5'}}, {._M_elems: {'4', '6'}}, {._M_elems: {'4', '7'}}, {._M_elems: {'4', '8'}}, {._M_elems: {'4', '9'}}, |
724 | {._M_elems: {'5', '0'}}, {._M_elems: {'5', '1'}}, {._M_elems: {'5', '2'}}, {._M_elems: {'5', '3'}}, {._M_elems: {'5', '4'}}, {._M_elems: {'5', '5'}}, {._M_elems: {'5', '6'}}, {._M_elems: {'5', '7'}}, {._M_elems: {'5', '8'}}, {._M_elems: {'5', '9'}}, |
725 | {._M_elems: {'6', '0'}}, {._M_elems: {'6', '1'}}, {._M_elems: {'6', '2'}}, {._M_elems: {'6', '3'}}, {._M_elems: {'6', '4'}}, {._M_elems: {'6', '5'}}, {._M_elems: {'6', '6'}}, {._M_elems: {'6', '7'}}, {._M_elems: {'6', '8'}}, {._M_elems: {'6', '9'}}, |
726 | {._M_elems: {'7', '0'}}, {._M_elems: {'7', '1'}}, {._M_elems: {'7', '2'}}, {._M_elems: {'7', '3'}}, {._M_elems: {'7', '4'}}, {._M_elems: {'7', '5'}}, {._M_elems: {'7', '6'}}, {._M_elems: {'7', '7'}}, {._M_elems: {'7', '8'}}, {._M_elems: {'7', '9'}}, |
727 | {._M_elems: {'8', '0'}}, {._M_elems: {'8', '1'}}, {._M_elems: {'8', '2'}}, {._M_elems: {'8', '3'}}, {._M_elems: {'8', '4'}}, {._M_elems: {'8', '5'}}, {._M_elems: {'8', '6'}}, {._M_elems: {'8', '7'}}, {._M_elems: {'8', '8'}}, {._M_elems: {'8', '9'}}, |
728 | {._M_elems: {'9', '0'}}, {._M_elems: {'9', '1'}}, {._M_elems: {'9', '2'}}, {._M_elems: {'9', '3'}}, {._M_elems: {'9', '4'}}, {._M_elems: {'9', '5'}}, {._M_elems: {'9', '6'}}, {._M_elems: {'9', '7'}}, {._M_elems: {'9', '8'}}, {._M_elems: {'9', '9'}}, |
729 | } |
730 | }; |
731 | |
732 | // special case for "0" |
733 | if (x == 0) |
734 | { |
735 | o->write_character(c: '0'); |
736 | return; |
737 | } |
738 | |
739 | // use a pointer to fill the buffer |
740 | auto buffer_ptr = number_buffer.begin(); // NOLINT(llvm-qualified-auto,readability-qualified-auto,cppcoreguidelines-pro-type-vararg,hicpp-vararg) |
741 | |
742 | number_unsigned_t abs_value; |
743 | |
744 | unsigned int n_chars{}; |
745 | |
746 | if (is_negative_number(x)) |
747 | { |
748 | *buffer_ptr = '-'; |
749 | abs_value = remove_sign(static_cast<number_integer_t>(x)); |
750 | |
751 | // account one more byte for the minus sign |
752 | n_chars = 1 + count_digits(x: abs_value); |
753 | } |
754 | else |
755 | { |
756 | abs_value = static_cast<number_unsigned_t>(x); |
757 | n_chars = count_digits(x: abs_value); |
758 | } |
759 | |
760 | // spare 1 byte for '\0' |
761 | JSON_ASSERT(n_chars < number_buffer.size() - 1); |
762 | |
763 | // jump to the end to generate the string from backward, |
764 | // so we later avoid reversing the result |
765 | buffer_ptr += n_chars; |
766 | |
767 | // Fast int2ascii implementation inspired by "Fastware" talk by Andrei Alexandrescu |
768 | // See: https://www.youtube.com/watch?v=o4-CwDo2zpg |
769 | while (abs_value >= 100) |
770 | { |
771 | const auto digits_index = static_cast<unsigned>((abs_value % 100)); |
772 | abs_value /= 100; |
773 | *(--buffer_ptr) = digits_to_99[digits_index][1]; |
774 | *(--buffer_ptr) = digits_to_99[digits_index][0]; |
775 | } |
776 | |
777 | if (abs_value >= 10) |
778 | { |
779 | const auto digits_index = static_cast<unsigned>(abs_value); |
780 | *(--buffer_ptr) = digits_to_99[digits_index][1]; |
781 | *(--buffer_ptr) = digits_to_99[digits_index][0]; |
782 | } |
783 | else |
784 | { |
785 | *(--buffer_ptr) = static_cast<char>('0' + abs_value); |
786 | } |
787 | |
788 | o->write_characters(s: number_buffer.data(), length: n_chars); |
789 | } |
790 | |
791 | /*! |
792 | @brief dump a floating-point number |
793 | |
794 | Dump a given floating-point number to output stream @a o. Works internally |
795 | with @a number_buffer. |
796 | |
797 | @param[in] x floating-point number to dump |
798 | */ |
799 | void dump_float(number_float_t x) |
800 | { |
801 | // NaN / inf |
802 | if (!std::isfinite(x)) |
803 | { |
804 | o->write_characters(s: "null" , length: 4); |
805 | return; |
806 | } |
807 | |
808 | // If number_float_t is an IEEE-754 single or double precision number, |
809 | // use the Grisu2 algorithm to produce short numbers which are |
810 | // guaranteed to round-trip, using strtof and strtod, resp. |
811 | // |
812 | // NB: The test below works if <long double> == <double>. |
813 | static constexpr bool is_ieee_single_or_double |
814 | = (std::numeric_limits<number_float_t>::is_iec559 && std::numeric_limits<number_float_t>::digits == 24 && std::numeric_limits<number_float_t>::max_exponent == 128) || |
815 | (std::numeric_limits<number_float_t>::is_iec559 && std::numeric_limits<number_float_t>::digits == 53 && std::numeric_limits<number_float_t>::max_exponent == 1024); |
816 | |
817 | dump_float(x, std::integral_constant<bool, is_ieee_single_or_double>()); |
818 | } |
819 | |
820 | void dump_float(number_float_t x, std::true_type /*is_ieee_single_or_double*/) |
821 | { |
822 | auto* begin = number_buffer.data(); |
823 | auto* end = ::nlohmann::detail::to_chars(begin, begin + number_buffer.size(), x); |
824 | |
825 | o->write_characters(s: begin, length: static_cast<size_t>(end - begin)); |
826 | } |
827 | |
828 | void dump_float(number_float_t x, std::false_type /*is_ieee_single_or_double*/) |
829 | { |
830 | // get number of digits for a float -> text -> float round-trip |
831 | static constexpr auto d = std::numeric_limits<number_float_t>::max_digits10; |
832 | |
833 | // the actual conversion |
834 | // NOLINTNEXTLINE(cppcoreguidelines-pro-type-vararg,hicpp-vararg) |
835 | std::ptrdiff_t len = (std::snprintf)(s: number_buffer.data(), maxlen: number_buffer.size(), format: "%.*g" , d, x); |
836 | |
837 | // negative value indicates an error |
838 | JSON_ASSERT(len > 0); |
839 | // check if buffer was large enough |
840 | JSON_ASSERT(static_cast<std::size_t>(len) < number_buffer.size()); |
841 | |
842 | // erase thousands separator |
843 | if (thousands_sep != '\0') |
844 | { |
845 | // NOLINTNEXTLINE(readability-qualified-auto,llvm-qualified-auto): std::remove returns an iterator, see https://github.com/nlohmann/json/issues/3081 |
846 | const auto end = std::remove(first: number_buffer.begin(), last: number_buffer.begin() + len, value: thousands_sep); |
847 | std::fill(first: end, last: number_buffer.end(), value: '\0'); |
848 | JSON_ASSERT((end - number_buffer.begin()) <= len); |
849 | len = (end - number_buffer.begin()); |
850 | } |
851 | |
852 | // convert decimal point to '.' |
853 | if (decimal_point != '\0' && decimal_point != '.') |
854 | { |
855 | // NOLINTNEXTLINE(readability-qualified-auto,llvm-qualified-auto): std::find returns an iterator, see https://github.com/nlohmann/json/issues/3081 |
856 | const auto dec_pos = std::find(first: number_buffer.begin(), last: number_buffer.end(), val: decimal_point); |
857 | if (dec_pos != number_buffer.end()) |
858 | { |
859 | *dec_pos = '.'; |
860 | } |
861 | } |
862 | |
863 | o->write_characters(s: number_buffer.data(), length: static_cast<std::size_t>(len)); |
864 | |
865 | // determine if we need to append ".0" |
866 | const bool value_is_int_like = |
867 | std::none_of(number_buffer.begin(), number_buffer.begin() + len + 1, |
868 | [](char c) |
869 | { |
870 | return c == '.' || c == 'e'; |
871 | }); |
872 | |
873 | if (value_is_int_like) |
874 | { |
875 | o->write_characters(s: ".0" , length: 2); |
876 | } |
877 | } |
878 | |
879 | /*! |
880 | @brief check whether a string is UTF-8 encoded |
881 | |
882 | The function checks each byte of a string whether it is UTF-8 encoded. The |
883 | result of the check is stored in the @a state parameter. The function must |
884 | be called initially with state 0 (accept). State 1 means the string must |
885 | be rejected, because the current byte is not allowed. If the string is |
886 | completely processed, but the state is non-zero, the string ended |
887 | prematurely; that is, the last byte indicated more bytes should have |
888 | followed. |
889 | |
890 | @param[in,out] state the state of the decoding |
891 | @param[in,out] codep codepoint (valid only if resulting state is UTF8_ACCEPT) |
892 | @param[in] byte next byte to decode |
893 | @return new state |
894 | |
895 | @note The function has been edited: a std::array is used. |
896 | |
897 | @copyright Copyright (c) 2008-2009 Bjoern Hoehrmann <bjoern@hoehrmann.de> |
898 | @sa http://bjoern.hoehrmann.de/utf-8/decoder/dfa/ |
899 | */ |
900 | static std::uint8_t decode(std::uint8_t& state, std::uint32_t& codep, const std::uint8_t byte) noexcept |
901 | { |
902 | static const std::array<std::uint8_t, 400> utf8d = |
903 | { |
904 | ._M_elems: { |
905 | 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, // 00..1F |
906 | 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, // 20..3F |
907 | 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, // 40..5F |
908 | 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, // 60..7F |
909 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, // 80..9F |
910 | 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, // A0..BF |
911 | 8, 8, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, // C0..DF |
912 | 0xA, 0x3, 0x3, 0x3, 0x3, 0x3, 0x3, 0x3, 0x3, 0x3, 0x3, 0x3, 0x3, 0x4, 0x3, 0x3, // E0..EF |
913 | 0xB, 0x6, 0x6, 0x6, 0x5, 0x8, 0x8, 0x8, 0x8, 0x8, 0x8, 0x8, 0x8, 0x8, 0x8, 0x8, // F0..FF |
914 | 0x0, 0x1, 0x2, 0x3, 0x5, 0x8, 0x7, 0x1, 0x1, 0x1, 0x4, 0x6, 0x1, 0x1, 0x1, 0x1, // s0..s0 |
915 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 0, 1, 1, 1, 1, 1, 1, // s1..s2 |
916 | 1, 2, 1, 1, 1, 1, 1, 2, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1, // s3..s4 |
917 | 1, 2, 1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 3, 1, 3, 1, 1, 1, 1, 1, 1, // s5..s6 |
918 | 1, 3, 1, 1, 1, 1, 1, 3, 1, 3, 1, 1, 1, 1, 1, 1, 1, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 // s7..s8 |
919 | } |
920 | }; |
921 | |
922 | JSON_ASSERT(byte < utf8d.size()); |
923 | const std::uint8_t type = utf8d[byte]; |
924 | |
925 | codep = (state != UTF8_ACCEPT) |
926 | ? (byte & 0x3fu) | (codep << 6u) |
927 | : (0xFFu >> type) & (byte); |
928 | |
929 | const std::size_t index = 256u + static_cast<size_t>(state) * 16u + static_cast<size_t>(type); |
930 | JSON_ASSERT(index < utf8d.size()); |
931 | state = utf8d[index]; |
932 | return state; |
933 | } |
934 | |
935 | /* |
936 | * Overload to make the compiler happy while it is instantiating |
937 | * dump_integer for number_unsigned_t. |
938 | * Must never be called. |
939 | */ |
940 | number_unsigned_t remove_sign(number_unsigned_t x) |
941 | { |
942 | JSON_ASSERT(false); // NOLINT(cert-dcl03-c,hicpp-static-assert,misc-static-assert) LCOV_EXCL_LINE |
943 | return x; // LCOV_EXCL_LINE |
944 | } |
945 | |
946 | /* |
947 | * Helper function for dump_integer |
948 | * |
949 | * This function takes a negative signed integer and returns its absolute |
950 | * value as unsigned integer. The plus/minus shuffling is necessary as we can |
951 | * not directly remove the sign of an arbitrary signed integer as the |
952 | * absolute values of INT_MIN and INT_MAX are usually not the same. See |
953 | * #1708 for details. |
954 | */ |
955 | inline number_unsigned_t remove_sign(number_integer_t x) noexcept |
956 | { |
957 | JSON_ASSERT(x < 0 && x < (std::numeric_limits<number_integer_t>::max)()); // NOLINT(misc-redundant-expression) |
958 | return static_cast<number_unsigned_t>(-(x + 1)) + 1; |
959 | } |
960 | |
961 | private: |
962 | /// the output of the serializer |
963 | output_adapter_t<char> o = nullptr; |
964 | |
965 | /// a (hopefully) large enough character buffer |
966 | std::array<char, 64> number_buffer{._M_elems: {}}; |
967 | |
968 | /// the locale |
969 | const std::lconv* loc = nullptr; |
970 | /// the locale's thousand separator character |
971 | const char thousands_sep = '\0'; |
972 | /// the locale's decimal point character |
973 | const char decimal_point = '\0'; |
974 | |
975 | /// string buffer |
976 | std::array<char, 512> string_buffer{._M_elems: {}}; |
977 | |
978 | /// the indentation character |
979 | const char indent_char; |
980 | /// the indentation string |
981 | string_t indent_string; |
982 | |
983 | /// error_handler how to react on decoding errors |
984 | const error_handler_t error_handler; |
985 | }; |
986 | |
987 | } // namespace detail |
988 | NLOHMANN_JSON_NAMESPACE_END |
989 | |