/src/jsoncons/include/jsoncons_ext/csv/csv_parser.hpp
Line | Count | Source |
1 | | // Copyright 2013-2025 Daniel Parker |
2 | | // Distributed under the Boost license, Version 1.0. |
3 | | // (See accompanying file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) |
4 | | |
5 | | // See https://github.com/danielaparker/jsoncons for latest version |
6 | | |
7 | | #ifndef JSONCONS_CSV_CSV_PARSER_HPP |
8 | | #define JSONCONS_CSV_CSV_PARSER_HPP |
9 | | |
10 | | #include <cctype> |
11 | | #include <cstddef> |
12 | | #include <cstdint> |
13 | | #include <functional> |
14 | | #include <memory> // std::allocator |
15 | | #include <sstream> |
16 | | #include <string> |
17 | | #include <system_error> |
18 | | #include <vector> |
19 | | |
20 | | #include <jsoncons/config/compiler_support.hpp> |
21 | | #include <jsoncons/config/jsoncons_config.hpp> |
22 | | #include <jsoncons/utility/read_number.hpp> |
23 | | #include <jsoncons/json_exception.hpp> |
24 | | #include <jsoncons/json_filter.hpp> |
25 | | #include <jsoncons/json_reader.hpp> |
26 | | #include <jsoncons/json_type.hpp> |
27 | | #include <jsoncons/json_visitor.hpp> |
28 | | #include <jsoncons/semantic_tag.hpp> |
29 | | #include <jsoncons/ser_util.hpp> |
30 | | #include <jsoncons/staj_event.hpp> |
31 | | |
32 | | #include <jsoncons_ext/csv/csv_error.hpp> |
33 | | #include <jsoncons_ext/csv/csv_options.hpp> |
34 | | |
35 | | namespace jsoncons { |
36 | | namespace csv { |
37 | | |
38 | | enum class csv_mode |
39 | | { |
40 | | initial, |
41 | | header, |
42 | | data, |
43 | | subfields |
44 | | }; |
45 | | |
46 | | enum class csv_parse_state |
47 | | { |
48 | | start, |
49 | | cr, |
50 | | expect_comment_or_record, |
51 | | expect_record, |
52 | | end_record, |
53 | | no_more_records, |
54 | | comment, |
55 | | between_values, |
56 | | quoted_string, |
57 | | unquoted_string, |
58 | | before_unquoted_string, |
59 | | escaped_value, |
60 | | minus, |
61 | | zero, |
62 | | integer, |
63 | | fraction, |
64 | | exp1, |
65 | | exp2, |
66 | | exp3, |
67 | | accept, |
68 | | before_unquoted_field, |
69 | | before_unquoted_field_tail, |
70 | | before_unquoted_field_tail1, |
71 | | before_last_unquoted_field, |
72 | | before_last_unquoted_field_tail, |
73 | | before_unquoted_subfield, |
74 | | before_unquoted_subfield_tail, |
75 | | before_quoted_subfield, |
76 | | before_quoted_subfield_tail, |
77 | | before_quoted_field, |
78 | | before_quoted_field_tail, |
79 | | before_last_quoted_field, |
80 | | before_last_quoted_field_tail, |
81 | | done |
82 | | }; |
83 | | |
84 | | enum class cached_state |
85 | | { |
86 | | begin_object, |
87 | | end_object, |
88 | | begin_array, |
89 | | end_array, |
90 | | name, |
91 | | item, |
92 | | done |
93 | | }; |
94 | | |
95 | | struct default_csv_parsing |
96 | | { |
97 | | bool operator()(csv_errc, const ser_context&) noexcept |
98 | 0 | { |
99 | 0 | return false; |
100 | 0 | } |
101 | | }; |
102 | | |
103 | | namespace detail { |
104 | | |
105 | | template <typename CharT,typename TempAlloc > |
106 | | class parse_event |
107 | | { |
108 | | using temp_allocator_type = TempAlloc; |
109 | | using string_view_type = typename basic_json_visitor<CharT>::string_view_type; |
110 | | using char_allocator_type = typename std::allocator_traits<temp_allocator_type>:: template rebind_alloc<CharT>; |
111 | | using byte_allocator_type = typename std::allocator_traits<temp_allocator_type>:: template rebind_alloc<uint8_t>; |
112 | | using string_type = std::basic_string<CharT,std::char_traits<CharT>,char_allocator_type>; |
113 | | using byte_string_type = basic_byte_string<byte_allocator_type>; |
114 | | |
115 | | staj_event_type event_type; |
116 | | string_type string_value; |
117 | | byte_string_type byte_string_value; |
118 | | union |
119 | | { |
120 | | bool bool_value; |
121 | | int64_t int64_value; |
122 | | uint64_t uint64_value; |
123 | | double double_value; |
124 | | }; |
125 | | semantic_tag tag; |
126 | | public: |
127 | | parse_event(staj_event_type event_type, semantic_tag tag, const TempAlloc& alloc) |
128 | 0 | : event_type(event_type), |
129 | 0 | string_value(alloc), |
130 | 0 | byte_string_value(alloc), |
131 | 0 | tag(tag) |
132 | 0 | { |
133 | 0 | } |
134 | | |
135 | | parse_event(const string_view_type& value, semantic_tag tag, const TempAlloc& alloc) |
136 | 0 | : event_type(staj_event_type::string_value), |
137 | 0 | string_value(value.data(),value.length(),alloc), |
138 | 0 | byte_string_value(alloc), |
139 | 0 | tag(tag) |
140 | 0 | { |
141 | 0 | } |
142 | | |
143 | | parse_event(const byte_string_view& value, semantic_tag tag, const TempAlloc& alloc) |
144 | 0 | : event_type(staj_event_type::byte_string_value), |
145 | 0 | string_value(alloc), |
146 | 0 | byte_string_value(value.data(),value.size(),alloc), |
147 | 0 | tag(tag) |
148 | 0 | { |
149 | 0 | } |
150 | | |
151 | | parse_event(bool value, semantic_tag tag, const TempAlloc& alloc) |
152 | 0 | : event_type(staj_event_type::bool_value), |
153 | 0 | string_value(alloc), |
154 | 0 | byte_string_value(alloc), |
155 | 0 | bool_value(value), |
156 | 0 | tag(tag) |
157 | 0 | { |
158 | 0 | } |
159 | | |
160 | | parse_event(int64_t value, semantic_tag tag, const TempAlloc& alloc) |
161 | 0 | : event_type(staj_event_type::int64_value), |
162 | 0 | string_value(alloc), |
163 | 0 | byte_string_value(alloc), |
164 | 0 | int64_value(value), |
165 | 0 | tag(tag) |
166 | 0 | { |
167 | 0 | } |
168 | | |
169 | | parse_event(uint64_t value, semantic_tag tag, const TempAlloc& alloc) |
170 | 0 | : event_type(staj_event_type::uint64_value), |
171 | 0 | string_value(alloc), |
172 | 0 | byte_string_value(alloc), |
173 | 0 | uint64_value(value), |
174 | 0 | tag(tag) |
175 | 0 | { |
176 | 0 | } |
177 | | |
178 | | parse_event(double value, semantic_tag tag, const TempAlloc& alloc) |
179 | 0 | : event_type(staj_event_type::double_value), |
180 | 0 | string_value(alloc), |
181 | 0 | byte_string_value(alloc), |
182 | 0 | double_value(value), |
183 | 0 | tag(tag) |
184 | 0 | { |
185 | 0 | } |
186 | | |
187 | | parse_event(const parse_event&) = default; |
188 | 0 | parse_event(parse_event&&) = default; |
189 | | parse_event& operator=(const parse_event&) = default; |
190 | | parse_event& operator=(parse_event&&) = default; |
191 | | |
192 | | void replay(basic_json_visitor<CharT>& visitor) const |
193 | 0 | { |
194 | 0 | switch (event_type) |
195 | 0 | { |
196 | 0 | case staj_event_type::begin_array: |
197 | 0 | visitor.begin_array(tag, ser_context()); |
198 | 0 | break; |
199 | 0 | case staj_event_type::end_array: |
200 | 0 | visitor.end_array(ser_context()); |
201 | 0 | break; |
202 | 0 | case staj_event_type::string_value: |
203 | 0 | visitor.string_value(string_value, tag, ser_context()); |
204 | 0 | break; |
205 | 0 | case staj_event_type::byte_string_value: |
206 | 0 | visitor.byte_string_value(byte_string_value, tag, ser_context()); |
207 | 0 | break; |
208 | 0 | case staj_event_type::null_value: |
209 | 0 | visitor.null_value(tag, ser_context()); |
210 | 0 | break; |
211 | 0 | case staj_event_type::bool_value: |
212 | 0 | visitor.bool_value(bool_value, tag, ser_context()); |
213 | 0 | break; |
214 | 0 | case staj_event_type::int64_value: |
215 | 0 | visitor.int64_value(int64_value, tag, ser_context()); |
216 | 0 | break; |
217 | 0 | case staj_event_type::uint64_value: |
218 | 0 | visitor.uint64_value(uint64_value, tag, ser_context()); |
219 | 0 | break; |
220 | 0 | case staj_event_type::double_value: |
221 | 0 | visitor.double_value(double_value, tag, ser_context()); |
222 | 0 | break; |
223 | 0 | default: |
224 | 0 | break; |
225 | 0 | } |
226 | 0 | } |
227 | | }; |
228 | | |
229 | | template <typename CharT,typename TempAlloc > |
230 | | class m_columns_filter : public basic_json_visitor<CharT> |
231 | | { |
232 | | public: |
233 | | using string_view_type = typename basic_json_visitor<CharT>::string_view_type; |
234 | | using char_type = CharT; |
235 | | using temp_allocator_type = TempAlloc; |
236 | | |
237 | | using char_allocator_type = typename std::allocator_traits<temp_allocator_type>:: template rebind_alloc<CharT>; |
238 | | using string_type = std::basic_string<CharT,std::char_traits<CharT>,char_allocator_type>; |
239 | | |
240 | | using string_allocator_type = typename std::allocator_traits<temp_allocator_type>:: template rebind_alloc<string_type>; |
241 | | using parse_event_allocator_type = typename std::allocator_traits<temp_allocator_type>:: template rebind_alloc<parse_event<CharT,TempAlloc>>; |
242 | | using parse_event_vector_type = std::vector<parse_event<CharT,TempAlloc>, parse_event_allocator_type>; |
243 | | using parse_event_vector_allocator_type = typename std::allocator_traits<temp_allocator_type>:: template rebind_alloc<parse_event_vector_type>; |
244 | | private: |
245 | | TempAlloc alloc_; |
246 | | std::size_t name_index_{0}; |
247 | | int level_{0}; |
248 | | int level2_{0}; |
249 | | cached_state state_{cached_state::begin_object}; |
250 | | std::size_t column_index_{0}; |
251 | | std::size_t row_index_{0}; |
252 | | |
253 | | std::vector<string_type, string_allocator_type> column_names_; |
254 | | std::vector<parse_event_vector_type,parse_event_vector_allocator_type> cached_events_; |
255 | | public: |
256 | | |
257 | | m_columns_filter(const TempAlloc& alloc) |
258 | 2.52k | : alloc_(alloc), |
259 | 2.52k | column_names_(alloc), |
260 | 2.52k | cached_events_(alloc) |
261 | 2.52k | { |
262 | 2.52k | } |
263 | | |
264 | | void reset() |
265 | | { |
266 | | name_index_ = 0; |
267 | | level_ = 0; |
268 | | level2_ = 0; |
269 | | state_ = cached_state::begin_object; |
270 | | column_index_ = 0; |
271 | | row_index_ = 0; |
272 | | column_names_.clear(); |
273 | | cached_events_.clear(); |
274 | | } |
275 | | |
276 | | bool done() const |
277 | 0 | { |
278 | 0 | return state_ == cached_state::done; |
279 | 0 | } |
280 | | |
281 | | void initialize(const std::vector<string_type, string_allocator_type>& column_names) |
282 | 0 | { |
283 | 0 | for (const auto& name : column_names) |
284 | 0 | { |
285 | 0 | column_names_.push_back(name); |
286 | 0 | cached_events_.emplace_back(alloc_); |
287 | 0 | } |
288 | 0 | name_index_ = 0; |
289 | 0 | level_ = 0; |
290 | 0 | level2_ = 0; |
291 | 0 | column_index_ = 0; |
292 | 0 | row_index_ = 0; |
293 | 0 | state_ = cached_state::begin_object; |
294 | 0 | } |
295 | | |
296 | | void skip_column() |
297 | 0 | { |
298 | 0 | ++name_index_; |
299 | 0 | } |
300 | | |
301 | | int level() const |
302 | | { |
303 | | return static_cast<int>(level_); |
304 | | } |
305 | | |
306 | | bool replay_parse_events(basic_json_visitor<CharT>& visitor, bool cursor_mode, int mark_level) |
307 | 0 | { |
308 | 0 | bool more = true; |
309 | 0 | while (more) |
310 | 0 | { |
311 | 0 | switch (state_) |
312 | 0 | { |
313 | 0 | case cached_state::begin_object: |
314 | 0 | visitor.begin_object(semantic_tag::none, ser_context()); |
315 | 0 | ++level_; |
316 | 0 | more = !cursor_mode; |
317 | 0 | column_index_ = 0; |
318 | 0 | state_ = cached_state::name; |
319 | 0 | break; |
320 | 0 | case cached_state::end_object: |
321 | 0 | visitor.end_object(ser_context()); |
322 | 0 | more = !cursor_mode; |
323 | 0 | if (level_ == mark_level) |
324 | 0 | { |
325 | 0 | more = false; |
326 | 0 | } |
327 | 0 | --level_; |
328 | 0 | state_ = cached_state::done; |
329 | 0 | break; |
330 | 0 | case cached_state::name: |
331 | 0 | if (column_index_ < column_names_.size()) |
332 | 0 | { |
333 | 0 | visitor.key(column_names_[column_index_], ser_context()); |
334 | 0 | more = !cursor_mode; |
335 | 0 | state_ = cached_state::begin_array; |
336 | 0 | } |
337 | 0 | else |
338 | 0 | { |
339 | 0 | state_ = cached_state::end_object; |
340 | 0 | } |
341 | 0 | break; |
342 | 0 | case cached_state::begin_array: |
343 | 0 | visitor.begin_array(semantic_tag::none, ser_context()); |
344 | 0 | ++level_; |
345 | 0 | more = !cursor_mode; |
346 | 0 | row_index_ = 0; |
347 | 0 | state_ = cached_state::item; |
348 | 0 | break; |
349 | 0 | case cached_state::end_array: |
350 | 0 | visitor.end_array(ser_context()); |
351 | 0 | more = !cursor_mode; |
352 | 0 | if (level_ == mark_level) |
353 | 0 | { |
354 | 0 | more = false; |
355 | 0 | } |
356 | 0 | --level_; |
357 | 0 | ++column_index_; |
358 | 0 | state_ = cached_state::name; |
359 | 0 | break; |
360 | 0 | case cached_state::item: |
361 | 0 | if (row_index_ < cached_events_[column_index_].size()) |
362 | 0 | { |
363 | 0 | cached_events_[column_index_][row_index_].replay(visitor); |
364 | 0 | more = !cursor_mode; |
365 | 0 | ++row_index_; |
366 | 0 | } |
367 | 0 | else |
368 | 0 | { |
369 | 0 | state_ = cached_state::end_array; |
370 | 0 | } |
371 | 0 | break; |
372 | 0 | default: |
373 | 0 | more = false; |
374 | 0 | break; |
375 | 0 | } |
376 | 0 | } |
377 | 0 | return more; |
378 | 0 | } |
379 | | |
380 | | void visit_flush() override |
381 | 0 | { |
382 | 0 | } |
383 | | |
384 | | JSONCONS_VISITOR_RETURN_TYPE visit_begin_object(semantic_tag, const ser_context&, std::error_code& ec) override |
385 | 0 | { |
386 | 0 | ec = csv_errc::invalid_parse_state; |
387 | 0 | JSONCONS_VISITOR_RETURN; |
388 | 0 | } |
389 | | |
390 | | JSONCONS_VISITOR_RETURN_TYPE visit_end_object(const ser_context&, std::error_code& ec) override |
391 | 0 | { |
392 | 0 | ec = csv_errc::invalid_parse_state; |
393 | 0 | JSONCONS_VISITOR_RETURN; |
394 | 0 | } |
395 | | |
396 | | JSONCONS_VISITOR_RETURN_TYPE visit_begin_array(semantic_tag tag, const ser_context&, std::error_code&) override |
397 | 0 | { |
398 | 0 | if (name_index_ < column_names_.size()) |
399 | 0 | { |
400 | 0 | cached_events_[name_index_].emplace_back(staj_event_type::begin_array, tag, alloc_); |
401 | | |
402 | 0 | ++level2_; |
403 | 0 | } |
404 | 0 | JSONCONS_VISITOR_RETURN; |
405 | 0 | } |
406 | | |
407 | | JSONCONS_VISITOR_RETURN_TYPE visit_end_array(const ser_context&, std::error_code&) override |
408 | 0 | { |
409 | 0 | if (level2_ > 0) |
410 | 0 | { |
411 | 0 | cached_events_[name_index_].emplace_back(staj_event_type::end_array, semantic_tag::none, alloc_); |
412 | 0 | ++name_index_; |
413 | 0 | --level2_; |
414 | 0 | } |
415 | 0 | else |
416 | 0 | { |
417 | 0 | name_index_ = 0; |
418 | 0 | } |
419 | 0 | JSONCONS_VISITOR_RETURN; |
420 | 0 | } |
421 | | |
422 | | JSONCONS_VISITOR_RETURN_TYPE visit_key(const string_view_type&, const ser_context&, std::error_code& ec) override |
423 | 0 | { |
424 | 0 | ec = csv_errc::invalid_parse_state; |
425 | 0 | JSONCONS_VISITOR_RETURN; |
426 | 0 | } |
427 | | |
428 | | JSONCONS_VISITOR_RETURN_TYPE visit_null(semantic_tag tag, const ser_context&, std::error_code&) override |
429 | 0 | { |
430 | 0 | if (name_index_ < column_names_.size()) |
431 | 0 | { |
432 | 0 | cached_events_[name_index_].emplace_back(staj_event_type::null_value, tag, alloc_); |
433 | 0 | if (level2_ == 0) |
434 | 0 | { |
435 | 0 | ++name_index_; |
436 | 0 | } |
437 | 0 | } |
438 | 0 | JSONCONS_VISITOR_RETURN; |
439 | 0 | } |
440 | | |
441 | | JSONCONS_VISITOR_RETURN_TYPE visit_string(const string_view_type& value, semantic_tag tag, const ser_context&, std::error_code&) override |
442 | 0 | { |
443 | 0 | if (name_index_ < column_names_.size()) |
444 | 0 | { |
445 | 0 | cached_events_[name_index_].emplace_back(value, tag, alloc_); |
446 | |
|
447 | 0 | if (level2_ == 0) |
448 | 0 | { |
449 | 0 | ++name_index_; |
450 | 0 | } |
451 | 0 | } |
452 | 0 | JSONCONS_VISITOR_RETURN; |
453 | 0 | } |
454 | | |
455 | | JSONCONS_VISITOR_RETURN_TYPE visit_byte_string(const byte_string_view& value, |
456 | | semantic_tag tag, |
457 | | const ser_context&, |
458 | | std::error_code&) override |
459 | 0 | { |
460 | 0 | if (name_index_ < column_names_.size()) |
461 | 0 | { |
462 | 0 | cached_events_[name_index_].emplace_back(value, tag, alloc_); |
463 | 0 | if (level2_ == 0) |
464 | 0 | { |
465 | 0 | ++name_index_; |
466 | 0 | } |
467 | 0 | } |
468 | 0 | JSONCONS_VISITOR_RETURN; |
469 | 0 | } |
470 | | |
471 | | JSONCONS_VISITOR_RETURN_TYPE visit_double(double value, |
472 | | semantic_tag tag, |
473 | | const ser_context&, |
474 | | std::error_code&) override |
475 | 0 | { |
476 | 0 | if (name_index_ < column_names_.size()) |
477 | 0 | { |
478 | 0 | cached_events_[name_index_].emplace_back(value, tag, alloc_); |
479 | 0 | if (level2_ == 0) |
480 | 0 | { |
481 | 0 | ++name_index_; |
482 | 0 | } |
483 | 0 | } |
484 | 0 | JSONCONS_VISITOR_RETURN; |
485 | 0 | } |
486 | | |
487 | | JSONCONS_VISITOR_RETURN_TYPE visit_int64(int64_t value, |
488 | | semantic_tag tag, |
489 | | const ser_context&, |
490 | | std::error_code&) override |
491 | 0 | { |
492 | 0 | if (name_index_ < column_names_.size()) |
493 | 0 | { |
494 | 0 | cached_events_[name_index_].emplace_back(value, tag, alloc_); |
495 | 0 | if (level2_ == 0) |
496 | 0 | { |
497 | 0 | ++name_index_; |
498 | 0 | } |
499 | 0 | } |
500 | 0 | JSONCONS_VISITOR_RETURN; |
501 | 0 | } |
502 | | |
503 | | JSONCONS_VISITOR_RETURN_TYPE visit_uint64(uint64_t value, |
504 | | semantic_tag tag, |
505 | | const ser_context&, |
506 | | std::error_code&) override |
507 | 0 | { |
508 | 0 | if (name_index_ < column_names_.size()) |
509 | 0 | { |
510 | 0 | cached_events_[name_index_].emplace_back(value, tag, alloc_); |
511 | 0 | if (level2_ == 0) |
512 | 0 | { |
513 | 0 | ++name_index_; |
514 | 0 | } |
515 | 0 | } |
516 | 0 | JSONCONS_VISITOR_RETURN; |
517 | 0 | } |
518 | | |
519 | | JSONCONS_VISITOR_RETURN_TYPE visit_bool(bool value, semantic_tag tag, const ser_context&, std::error_code&) override |
520 | 0 | { |
521 | 0 | if (name_index_ < column_names_.size()) |
522 | 0 | { |
523 | 0 | cached_events_[name_index_].emplace_back(value, tag, alloc_); |
524 | 0 | if (level2_ == 0) |
525 | 0 | { |
526 | 0 | ++name_index_; |
527 | 0 | } |
528 | 0 | } |
529 | 0 | JSONCONS_VISITOR_RETURN; |
530 | 0 | } |
531 | | }; |
532 | | |
533 | | } // namespace detail |
534 | | |
535 | | template <typename CharT,typename TempAlloc =std::allocator<char>> |
536 | | class basic_csv_parser : public ser_context |
537 | | { |
538 | | public: |
539 | | using string_view_type = jsoncons::basic_string_view<CharT>; |
540 | | using char_type = CharT; |
541 | | private: |
542 | | struct string_maps_to_double |
543 | | { |
544 | | string_view_type s; |
545 | | |
546 | | bool operator()(const std::pair<string_view_type,double>& val) const |
547 | 0 | { |
548 | 0 | return val.first == s; |
549 | 0 | } |
550 | | }; |
551 | | |
552 | | using temp_allocator_type = TempAlloc; |
553 | | typedef typename std::allocator_traits<temp_allocator_type>:: template rebind_alloc<CharT> char_allocator_type; |
554 | | using string_type = std::basic_string<CharT,std::char_traits<CharT>,char_allocator_type>; |
555 | | typedef typename std::allocator_traits<temp_allocator_type>:: template rebind_alloc<string_type> string_allocator_type; |
556 | | typedef typename std::allocator_traits<temp_allocator_type>:: template rebind_alloc<csv_mode> csv_mode_allocator_type; |
557 | | typedef typename std::allocator_traits<temp_allocator_type>:: template rebind_alloc<csv_type_info> csv_type_info_allocator_type; |
558 | | typedef typename std::allocator_traits<temp_allocator_type>:: template rebind_alloc<std::vector<string_type,string_allocator_type>> string_vector_allocator_type; |
559 | | typedef typename std::allocator_traits<temp_allocator_type>:: template rebind_alloc<csv_parse_state> csv_parse_state_allocator_type; |
560 | | |
561 | | static constexpr int default_depth = 3; |
562 | | |
563 | | temp_allocator_type alloc_; |
564 | | csv_parse_state state_; |
565 | | std::function<bool(csv_errc,const ser_context&)> err_handler_; |
566 | | std::size_t column_{1}; |
567 | | std::size_t line_{1}; |
568 | | int nesting_depth_{default_depth}; |
569 | | bool assume_header_; |
570 | | char_type comment_starter_; |
571 | | char_type field_delimiter_; |
572 | | std::size_t header_lines_; |
573 | | bool ignore_empty_values_; |
574 | | bool ignore_empty_lines_; |
575 | | bool infer_types_; |
576 | | bool lossless_number_; |
577 | | csv_mapping_kind mapping_kind_; |
578 | | std::size_t max_lines_; |
579 | | char_type quote_char_; |
580 | | char_type quote_escape_char_; |
581 | | char_type subfield_delimiter_; |
582 | | bool trim_leading_; |
583 | | bool trim_leading_inside_quotes_; |
584 | | bool trim_trailing_; |
585 | | bool trim_trailing_inside_quotes_; |
586 | | bool unquoted_empty_value_is_null_; |
587 | | std::size_t min_column_names_{0}; |
588 | | std::size_t column_index_{0}; |
589 | | int level_{0}; |
590 | | std::size_t depth_{0}; |
591 | | std::size_t offset_{0}; |
592 | | const CharT* begin_input_{nullptr}; |
593 | | const CharT* input_end_{nullptr}; |
594 | | const CharT* input_ptr_{nullptr}; |
595 | | bool more_{true}; |
596 | | std::size_t header_line_{1}; |
597 | | bool cursor_mode_{false}; |
598 | | bool actual_cursor_mode_{false}; |
599 | | int mark_level_{0}; |
600 | | std::size_t header_line_offset_{0}; |
601 | | |
602 | | detail::m_columns_filter<CharT,TempAlloc> m_columns_filter_; |
603 | | std::vector<csv_mode,csv_mode_allocator_type> stack_; |
604 | | std::vector<string_type,string_allocator_type> column_names_; |
605 | | std::vector<csv_type_info,csv_type_info_allocator_type> column_types_; |
606 | | std::vector<string_type,string_allocator_type> column_defaults_; |
607 | | std::vector<csv_parse_state,csv_parse_state_allocator_type> state_stack_; |
608 | | string_type buffer_; |
609 | | std::vector<std::pair<std::basic_string<char_type>,double>> string_double_map_; |
610 | | |
611 | | public: |
612 | | basic_csv_parser() |
613 | | : basic_csv_parser(basic_csv_decode_options<CharT>()) |
614 | | { |
615 | | } |
616 | | |
617 | | explicit basic_csv_parser(const TempAlloc& alloc) |
618 | | : basic_csv_parser(basic_csv_decode_options<CharT>(), |
619 | | default_csv_parsing(), |
620 | | alloc) |
621 | | { |
622 | | } |
623 | | |
624 | | explicit basic_csv_parser(const basic_csv_decode_options<CharT>& options) |
625 | | : basic_csv_parser(options, TempAlloc{}) |
626 | | { |
627 | | } |
628 | | |
629 | | basic_csv_parser(const basic_csv_decode_options<CharT>& options, |
630 | | const TempAlloc& alloc) |
631 | 2.52k | : alloc_(alloc), |
632 | 2.52k | state_(csv_parse_state::start), |
633 | 2.52k | err_handler_(default_csv_parsing()), |
634 | 2.52k | assume_header_(options.assume_header()), |
635 | 2.52k | comment_starter_(options.comment_starter()), |
636 | 2.52k | field_delimiter_(options.field_delimiter()), |
637 | 2.52k | header_lines_(options.header_lines()), |
638 | 2.52k | ignore_empty_values_(options.ignore_empty_values()), |
639 | 2.52k | ignore_empty_lines_(options.ignore_empty_lines()), |
640 | 2.52k | infer_types_(options.infer_types()), |
641 | 2.52k | lossless_number_(options.lossless_number()), |
642 | 2.52k | mapping_kind_(options.mapping_kind()), |
643 | 2.52k | max_lines_(options.max_lines()), |
644 | 2.52k | quote_char_(options.quote_char()), |
645 | 2.52k | quote_escape_char_(options.quote_escape_char()), |
646 | 2.52k | subfield_delimiter_(options.subfield_delimiter()), |
647 | 2.52k | trim_leading_(options.trim_leading()), |
648 | 2.52k | trim_leading_inside_quotes_(options.trim_leading_inside_quotes()), |
649 | 2.52k | trim_trailing_(options.trim_trailing()), |
650 | 2.52k | trim_trailing_inside_quotes_(options.trim_trailing_inside_quotes()), |
651 | 2.52k | unquoted_empty_value_is_null_(options.unquoted_empty_value_is_null()), |
652 | 2.52k | m_columns_filter_(alloc), |
653 | 2.52k | stack_(alloc), |
654 | 2.52k | column_names_(alloc), |
655 | 2.52k | column_types_(alloc), |
656 | 2.52k | column_defaults_(alloc), |
657 | 2.52k | state_stack_(alloc), |
658 | 2.52k | buffer_(alloc) |
659 | 2.52k | { |
660 | 2.52k | if (options.enable_str_to_nan()) |
661 | 0 | { |
662 | 0 | string_double_map_.emplace_back(options.nan_to_str(),std::nan("")); |
663 | 0 | } |
664 | 2.52k | if (options.enable_str_to_inf()) |
665 | 0 | { |
666 | 0 | string_double_map_.emplace_back(options.inf_to_str(),std::numeric_limits<double>::infinity()); |
667 | 0 | } |
668 | 2.52k | if (options.enable_str_to_neginf()) |
669 | 0 | { |
670 | 0 | string_double_map_.emplace_back(options.neginf_to_str(),-std::numeric_limits<double>::infinity()); |
671 | 0 | } |
672 | | |
673 | 2.52k | jsoncons::csv::detail::parse_column_types(options.column_types(), column_types_); |
674 | 2.52k | jsoncons::csv::detail::parse_column_names(options.column_defaults(), column_defaults_); |
675 | 2.52k | jsoncons::csv::detail::parse_column_names(options.column_names(), column_names_); |
676 | 2.52k | min_column_names_ = column_names_.size(); |
677 | 2.52k | initialize(); |
678 | 2.52k | } |
679 | | |
680 | | #if !defined(JSONCONS_NO_DEPRECATED) |
681 | | basic_csv_parser(std::function<bool(csv_errc,const ser_context&)> err_handler, |
682 | | const TempAlloc& alloc = TempAlloc()) |
683 | | : basic_csv_parser(basic_csv_decode_options<CharT>(), |
684 | | err_handler, |
685 | | alloc) |
686 | | { |
687 | | } |
688 | | |
689 | | basic_csv_parser(const basic_csv_decode_options<CharT>& options, |
690 | | std::function<bool(csv_errc,const ser_context&)> err_handler, |
691 | | const TempAlloc& alloc = TempAlloc()) |
692 | | : alloc_(alloc), |
693 | | state_(csv_parse_state::start), |
694 | | err_handler_(err_handler), |
695 | | assume_header_(options.assume_header()), |
696 | | comment_starter_(options.comment_starter()), |
697 | | field_delimiter_(options.field_delimiter()), |
698 | | header_lines_(options.header_lines()), |
699 | | ignore_empty_values_(options.ignore_empty_values()), |
700 | | ignore_empty_lines_(options.ignore_empty_lines()), |
701 | | infer_types_(options.infer_types()), |
702 | | lossless_number_(options.lossless_number()), |
703 | | mapping_kind_(options.mapping_kind()), |
704 | | max_lines_(options.max_lines()), |
705 | | quote_char_(options.quote_char()), |
706 | | quote_escape_char_(options.quote_escape_char()), |
707 | | subfield_delimiter_(options.subfield_delimiter()), |
708 | | trim_leading_(options.trim_leading()), |
709 | | trim_leading_inside_quotes_(options.trim_leading_inside_quotes()), |
710 | | trim_trailing_(options.trim_trailing()), |
711 | | trim_trailing_inside_quotes_(options.trim_trailing_inside_quotes()), |
712 | | unquoted_empty_value_is_null_(options.unquoted_empty_value_is_null()), |
713 | | m_columns_filter_(alloc), |
714 | | stack_(alloc), |
715 | | column_names_(alloc), |
716 | | column_types_(alloc), |
717 | | column_defaults_(alloc), |
718 | | state_stack_(alloc), |
719 | | buffer_(alloc) |
720 | | { |
721 | | if (options.enable_str_to_nan()) |
722 | | { |
723 | | string_double_map_.emplace_back(options.nan_to_str(),std::nan("")); |
724 | | } |
725 | | if (options.enable_str_to_inf()) |
726 | | { |
727 | | string_double_map_.emplace_back(options.inf_to_str(),std::numeric_limits<double>::infinity()); |
728 | | } |
729 | | if (options.enable_str_to_neginf()) |
730 | | { |
731 | | string_double_map_.emplace_back(options.neginf_to_str(),-std::numeric_limits<double>::infinity()); |
732 | | } |
733 | | |
734 | | jsoncons::csv::detail::parse_column_types(options.column_types(), column_types_); |
735 | | jsoncons::csv::detail::parse_column_names(options.column_defaults(), column_defaults_); |
736 | | jsoncons::csv::detail::parse_column_names(options.column_names(), column_names_); |
737 | | min_column_names_ = column_names_.size(); |
738 | | initialize(); |
739 | | } |
740 | | #endif |
741 | | |
742 | | ~basic_csv_parser() noexcept |
743 | 2.52k | { |
744 | 2.52k | } |
745 | | |
746 | | void cursor_mode(bool value) |
747 | | { |
748 | | actual_cursor_mode_ = value; |
749 | | cursor_mode_ = (mapping_kind_ == csv_mapping_kind::m_columns) ? false : value; |
750 | | } |
751 | | |
752 | | int level() const |
753 | 3.14k | { |
754 | 3.14k | return level_; |
755 | 3.14k | } |
756 | | |
757 | | int mark_level() const |
758 | | { |
759 | | return mark_level_; |
760 | | } |
761 | | |
762 | | void mark_level(int value) |
763 | | { |
764 | | mark_level_ = value; |
765 | | } |
766 | | |
767 | | bool done() const |
768 | | { |
769 | | return state_ == csv_parse_state::done; |
770 | | } |
771 | | |
772 | | bool accept() const |
773 | | { |
774 | | return state_ == csv_parse_state::accept || state_ == csv_parse_state::done; |
775 | | } |
776 | | |
777 | | bool stopped() const |
778 | 18.9k | { |
779 | 18.9k | return !more_; |
780 | 18.9k | } |
781 | | |
782 | | bool source_exhausted() const |
783 | 16.4k | { |
784 | 16.4k | return input_ptr_ == input_end_; |
785 | 16.4k | } |
786 | | |
787 | | const std::vector<string_type,string_allocator_type>& column_labels() const |
788 | | { |
789 | | return column_names_; |
790 | | } |
791 | | |
792 | | void reinitialize() |
793 | | { |
794 | | state_ = csv_parse_state::start; |
795 | | column_ = 1; |
796 | | line_ = 1; |
797 | | nesting_depth_ = default_depth; |
798 | | column_index_ = 0; |
799 | | level_ = 0; |
800 | | depth_ = 0; |
801 | | offset_ = 0; |
802 | | begin_input_ = nullptr; |
803 | | input_end_ = nullptr; |
804 | | input_ptr_ = nullptr; |
805 | | more_ = true; |
806 | | header_line_ = 1; |
807 | | m_columns_filter_.reset(); |
808 | | stack_.clear(); |
809 | | column_names_.erase(column_names_.begin() + min_column_names_, column_names_.end()); |
810 | | state_stack_.clear(); |
811 | | buffer_.clear(); |
812 | | |
813 | | initialize(); |
814 | | } |
815 | | |
816 | | void restart() |
817 | | { |
818 | | more_ = true; |
819 | | } |
820 | | |
821 | | void parse_some(basic_json_visitor<CharT>& visitor) |
822 | | { |
823 | | std::error_code ec; |
824 | | parse_some(visitor, ec); |
825 | | if (JSONCONS_UNLIKELY(ec)) |
826 | | { |
827 | | JSONCONS_THROW(ser_error(ec,line_,column_)); |
828 | | } |
829 | | } |
830 | | |
831 | | void parse_some(basic_json_visitor<CharT>& visitor, std::error_code& ec) |
832 | 16.4k | { |
833 | 16.4k | basic_json_visitor<CharT>& local_visitor = (mapping_kind_ == csv_mapping_kind::m_columns) |
834 | 16.4k | ? m_columns_filter_ : visitor; |
835 | | |
836 | 16.4k | switch (mapping_kind_) |
837 | 16.4k | { |
838 | 0 | case csv_mapping_kind::m_columns: |
839 | 0 | cursor_mode_ = false; |
840 | 0 | break; |
841 | 16.4k | default: |
842 | 16.4k | break; |
843 | 16.4k | } |
844 | | |
845 | 16.4k | const CharT* local_input_end = input_end_; |
846 | | |
847 | 16.4k | if (input_ptr_ == local_input_end && more_) |
848 | 13.9k | { |
849 | 13.9k | switch (state_) |
850 | 13.9k | { |
851 | 1 | case csv_parse_state::start: |
852 | 1 | ec = csv_errc::source_error; |
853 | 1 | more_ = false; |
854 | 1 | return; |
855 | 2.09k | case csv_parse_state::before_unquoted_field: |
856 | 2.09k | case csv_parse_state::before_last_unquoted_field: |
857 | 2.09k | end_unquoted_string_value(local_visitor, ec); |
858 | 2.09k | state_ = csv_parse_state::before_last_unquoted_field_tail; |
859 | 2.09k | break; |
860 | 2.09k | case csv_parse_state::before_last_unquoted_field_tail: |
861 | 2.09k | if (stack_.back() == csv_mode::subfields) |
862 | 0 | { |
863 | 0 | stack_.pop_back(); |
864 | 0 | local_visitor.end_array(*this, ec); |
865 | 0 | more_ = !cursor_mode_; |
866 | 0 | if (mapping_kind_ != csv_mapping_kind::m_columns) |
867 | 0 | { |
868 | 0 | if (level() == mark_level_) |
869 | 0 | { |
870 | 0 | more_ = false; |
871 | 0 | } |
872 | 0 | } |
873 | 0 | --level_; |
874 | 0 | } |
875 | 2.09k | ++column_index_; |
876 | 2.09k | state_ = csv_parse_state::end_record; |
877 | 2.09k | break; |
878 | 270 | case csv_parse_state::before_unquoted_string: |
879 | 270 | buffer_.clear(); |
880 | 270 | JSONCONS_FALLTHROUGH; |
881 | 2.09k | case csv_parse_state::unquoted_string: |
882 | 2.09k | if (trim_leading_ || trim_trailing_) |
883 | 0 | { |
884 | 0 | trim_string_buffer(trim_leading_,trim_trailing_); |
885 | 0 | } |
886 | 2.09k | if (ignore_empty_values_ && buffer_.empty()) |
887 | 0 | { |
888 | 0 | state_ = csv_parse_state::end_record; |
889 | 0 | } |
890 | 2.09k | else |
891 | 2.09k | { |
892 | 2.09k | before_value(local_visitor, ec); |
893 | 2.09k | state_ = csv_parse_state::before_unquoted_field; |
894 | 2.09k | } |
895 | 2.09k | break; |
896 | 31 | case csv_parse_state::before_last_quoted_field: |
897 | 31 | end_quoted_string_value(local_visitor, ec); |
898 | 31 | ++column_index_; |
899 | 31 | state_ = csv_parse_state::end_record; |
900 | 31 | break; |
901 | 31 | case csv_parse_state::escaped_value: |
902 | 31 | if (quote_escape_char_ == quote_char_) |
903 | 31 | { |
904 | 31 | if (!(ignore_empty_values_ && buffer_.empty())) |
905 | 31 | { |
906 | 31 | before_value(local_visitor, ec); |
907 | 31 | ++column_; |
908 | 31 | state_ = csv_parse_state::before_last_quoted_field; |
909 | 31 | } |
910 | 0 | else |
911 | 0 | { |
912 | 0 | state_ = csv_parse_state::end_record; |
913 | 0 | } |
914 | 31 | } |
915 | 0 | else |
916 | 0 | { |
917 | 0 | ec = csv_errc::invalid_escaped_char; |
918 | 0 | more_ = false; |
919 | 0 | return; |
920 | 0 | } |
921 | 31 | break; |
922 | 2.43k | case csv_parse_state::end_record: |
923 | 2.43k | if (column_index_ > 0) |
924 | 2.13k | { |
925 | 2.13k | end_record(local_visitor, ec); |
926 | 2.13k | } |
927 | 2.43k | state_ = csv_parse_state::no_more_records; |
928 | 2.43k | break; |
929 | 2.43k | case csv_parse_state::no_more_records: |
930 | 2.43k | switch (stack_.back()) |
931 | 2.43k | { |
932 | 118 | case csv_mode::header: |
933 | 118 | stack_.pop_back(); |
934 | 118 | break; |
935 | 2.31k | case csv_mode::data: |
936 | 2.31k | stack_.pop_back(); |
937 | 2.31k | break; |
938 | 0 | default: |
939 | 0 | break; |
940 | 2.43k | } |
941 | 2.43k | local_visitor.end_array(*this, ec); |
942 | 2.43k | more_ = !cursor_mode_; |
943 | 2.43k | if (mapping_kind_ != csv_mapping_kind::m_columns) |
944 | 2.43k | { |
945 | 2.43k | if (level() == mark_level_) |
946 | 0 | { |
947 | 0 | more_ = false; |
948 | 0 | } |
949 | 2.43k | } |
950 | 2.43k | --level_; |
951 | | |
952 | 2.43k | if (mapping_kind_ == csv_mapping_kind::m_columns) |
953 | 0 | { |
954 | 0 | if (!m_columns_filter_.done()) |
955 | 0 | { |
956 | 0 | more_ = m_columns_filter_.replay_parse_events(visitor, actual_cursor_mode_, mark_level_); |
957 | 0 | } |
958 | 0 | else |
959 | 0 | { |
960 | 0 | state_ = csv_parse_state::accept; |
961 | 0 | } |
962 | 0 | } |
963 | 2.43k | else |
964 | 2.43k | { |
965 | 2.43k | state_ = csv_parse_state::accept; |
966 | 2.43k | } |
967 | 2.43k | break; |
968 | 2.43k | case csv_parse_state::accept: |
969 | 2.43k | if (!(stack_.size() == 1 && stack_.back() == csv_mode::initial)) |
970 | 0 | { |
971 | 0 | err_handler_(csv_errc::unexpected_eof, *this); |
972 | 0 | ec = csv_errc::unexpected_eof; |
973 | 0 | more_ = false; |
974 | 0 | return; |
975 | 0 | } |
976 | 2.43k | stack_.pop_back(); |
977 | 2.43k | local_visitor.flush(); |
978 | 2.43k | state_ = csv_parse_state::done; |
979 | 2.43k | more_ = false; |
980 | 2.43k | return; |
981 | 313 | default: |
982 | 313 | state_ = csv_parse_state::end_record; |
983 | 313 | break; |
984 | 13.9k | } |
985 | 13.9k | } |
986 | | |
987 | 166M | for (; (input_ptr_ < local_input_end) && more_;) |
988 | 166M | { |
989 | 166M | CharT curr_char = *input_ptr_; |
990 | | |
991 | 166M | switch (state_) |
992 | 166M | { |
993 | 4.34M | case csv_parse_state::cr: |
994 | 4.34M | ++line_; |
995 | 4.34M | column_ = 1; |
996 | 4.34M | switch (*input_ptr_) |
997 | 4.34M | { |
998 | 326 | case '\n': |
999 | 326 | ++input_ptr_; |
1000 | 326 | state_ = pop_state(); |
1001 | 326 | break; |
1002 | 4.34M | default: |
1003 | 4.34M | state_ = pop_state(); |
1004 | 4.34M | break; |
1005 | 4.34M | } |
1006 | 4.34M | break; |
1007 | 4.34M | case csv_parse_state::start: |
1008 | 2.51k | if (mapping_kind_ != csv_mapping_kind::m_columns) |
1009 | 2.51k | { |
1010 | 2.51k | local_visitor.begin_array(semantic_tag::none, *this, ec); |
1011 | 2.51k | more_ = !cursor_mode_; |
1012 | 2.51k | ++level_; |
1013 | 2.51k | } |
1014 | 2.51k | if (assume_header_ && mapping_kind_ == csv_mapping_kind::n_rows && !column_names_.empty()) |
1015 | 0 | { |
1016 | 0 | column_index_ = 0; |
1017 | 0 | local_visitor.begin_array(semantic_tag::none, *this, ec); |
1018 | 0 | ++level_; |
1019 | 0 | more_ = !cursor_mode_; |
1020 | 0 | state_ = csv_parse_state::expect_comment_or_record; |
1021 | 0 | } |
1022 | 2.51k | else |
1023 | 2.51k | { |
1024 | 2.51k | state_ = csv_parse_state::expect_comment_or_record; |
1025 | 2.51k | } |
1026 | 2.51k | break; |
1027 | 9.87M | case csv_parse_state::comment: |
1028 | 9.87M | switch (curr_char) |
1029 | 9.87M | { |
1030 | 1.81M | case '\n': |
1031 | 1.81M | { |
1032 | 1.81M | ++line_; |
1033 | 1.81M | if (stack_.back() == csv_mode::header) |
1034 | 1.66M | { |
1035 | 1.66M | ++header_line_offset_; |
1036 | 1.66M | } |
1037 | 1.81M | column_ = 1; |
1038 | 1.81M | state_ = csv_parse_state::expect_comment_or_record; |
1039 | 1.81M | break; |
1040 | 0 | } |
1041 | 15.1k | case '\r': |
1042 | 15.1k | ++line_; |
1043 | 15.1k | if (stack_.back() == csv_mode::header) |
1044 | 525 | { |
1045 | 525 | ++header_line_offset_; |
1046 | 525 | } |
1047 | 15.1k | column_ = 1; |
1048 | 15.1k | state_ = csv_parse_state::expect_comment_or_record; |
1049 | 15.1k | push_state(state_); |
1050 | 15.1k | state_ = csv_parse_state::cr; |
1051 | 15.1k | break; |
1052 | 8.03M | default: |
1053 | 8.03M | ++column_; |
1054 | 8.03M | break; |
1055 | 9.87M | } |
1056 | 9.87M | ++input_ptr_; |
1057 | 9.87M | break; |
1058 | | |
1059 | 5.27M | case csv_parse_state::expect_comment_or_record: |
1060 | 5.27M | buffer_.clear(); |
1061 | 5.27M | if (curr_char == comment_starter_) |
1062 | 1.83M | { |
1063 | 1.83M | state_ = csv_parse_state::comment; |
1064 | 1.83M | ++column_; |
1065 | 1.83M | ++input_ptr_; |
1066 | 1.83M | } |
1067 | 3.43M | else |
1068 | 3.43M | { |
1069 | 3.43M | state_ = csv_parse_state::expect_record; |
1070 | 3.43M | } |
1071 | 5.27M | break; |
1072 | 2.02M | case csv_parse_state::quoted_string: |
1073 | 2.02M | { |
1074 | 2.02M | if (curr_char == quote_escape_char_) |
1075 | 2.47k | { |
1076 | 2.47k | state_ = csv_parse_state::escaped_value; |
1077 | 2.47k | } |
1078 | 2.02M | else if (curr_char == quote_char_) |
1079 | 0 | { |
1080 | 0 | state_ = csv_parse_state::between_values; |
1081 | 0 | } |
1082 | 2.02M | else |
1083 | 2.02M | { |
1084 | 2.02M | buffer_.push_back(static_cast<CharT>(curr_char)); |
1085 | 2.02M | } |
1086 | 2.02M | } |
1087 | 2.02M | ++column_; |
1088 | 2.02M | ++input_ptr_; |
1089 | 2.02M | break; |
1090 | 2.44k | case csv_parse_state::escaped_value: |
1091 | 2.44k | { |
1092 | 2.44k | if (curr_char == quote_char_) |
1093 | 417 | { |
1094 | 417 | buffer_.push_back(static_cast<CharT>(curr_char)); |
1095 | 417 | state_ = csv_parse_state::quoted_string; |
1096 | 417 | ++column_; |
1097 | 417 | ++input_ptr_; |
1098 | 417 | } |
1099 | 2.03k | else if (quote_escape_char_ == quote_char_) |
1100 | 2.03k | { |
1101 | 2.03k | state_ = csv_parse_state::between_values; |
1102 | 2.03k | } |
1103 | 0 | else |
1104 | 0 | { |
1105 | 0 | ec = csv_errc::invalid_escaped_char; |
1106 | 0 | more_ = false; |
1107 | 0 | return; |
1108 | 0 | } |
1109 | 2.44k | } |
1110 | 2.44k | break; |
1111 | 2.54k | case csv_parse_state::between_values: |
1112 | 2.54k | switch (curr_char) |
1113 | 2.54k | { |
1114 | 272 | case '\r': |
1115 | 1.08k | case '\n': |
1116 | 1.08k | { |
1117 | 1.08k | if (trim_leading_ || trim_trailing_) |
1118 | 0 | { |
1119 | 0 | trim_string_buffer(trim_leading_,trim_trailing_); |
1120 | 0 | } |
1121 | 1.08k | if (!(ignore_empty_values_ && buffer_.empty())) |
1122 | 1.08k | { |
1123 | 1.08k | before_value(local_visitor, ec); |
1124 | 1.08k | state_ = csv_parse_state::before_last_quoted_field; |
1125 | 1.08k | } |
1126 | 0 | else |
1127 | 0 | { |
1128 | 0 | state_ = csv_parse_state::end_record; |
1129 | 0 | } |
1130 | 1.08k | break; |
1131 | 272 | } |
1132 | 1.46k | default: |
1133 | 1.46k | if (curr_char == field_delimiter_) |
1134 | 857 | { |
1135 | 857 | if (trim_leading_ || trim_trailing_) |
1136 | 0 | { |
1137 | 0 | trim_string_buffer(trim_leading_,trim_trailing_); |
1138 | 0 | } |
1139 | 857 | before_value(local_visitor, ec); |
1140 | 857 | state_ = csv_parse_state::before_quoted_field; |
1141 | 857 | } |
1142 | 607 | else if (subfield_delimiter_ != char_type() && curr_char == subfield_delimiter_) |
1143 | 0 | { |
1144 | 0 | if (trim_leading_ || trim_trailing_) |
1145 | 0 | { |
1146 | 0 | trim_string_buffer(trim_leading_,trim_trailing_); |
1147 | 0 | } |
1148 | 0 | before_value(local_visitor, ec); |
1149 | 0 | state_ = csv_parse_state::before_quoted_subfield; |
1150 | 0 | } |
1151 | 607 | else if (curr_char == ' ' || curr_char == '\t') |
1152 | 529 | { |
1153 | 529 | ++column_; |
1154 | 529 | ++input_ptr_; |
1155 | 529 | } |
1156 | 78 | else |
1157 | 78 | { |
1158 | 78 | ec = csv_errc::unexpected_char_between_fields; |
1159 | 78 | more_ = false; |
1160 | 78 | return; |
1161 | 78 | } |
1162 | 1.38k | break; |
1163 | 2.54k | } |
1164 | 2.46k | break; |
1165 | 23.9M | case csv_parse_state::before_unquoted_string: |
1166 | 23.9M | { |
1167 | 23.9M | buffer_.clear(); |
1168 | 23.9M | state_ = csv_parse_state::unquoted_string; |
1169 | 23.9M | break; |
1170 | 2.54k | } |
1171 | 23.9M | case csv_parse_state::before_unquoted_field: |
1172 | 23.9M | end_unquoted_string_value(local_visitor, ec); |
1173 | 23.9M | state_ = csv_parse_state::before_unquoted_field_tail; |
1174 | 23.9M | break; |
1175 | 23.9M | case csv_parse_state::before_unquoted_field_tail: |
1176 | 23.9M | { |
1177 | 23.9M | if (stack_.back() == csv_mode::subfields) |
1178 | 0 | { |
1179 | 0 | stack_.pop_back(); |
1180 | 0 | local_visitor.end_array(*this, ec); |
1181 | 0 | more_ = !cursor_mode_; |
1182 | 0 | if (mapping_kind_ != csv_mapping_kind::m_columns) |
1183 | 0 | { |
1184 | 0 | if (level() == mark_level_) |
1185 | 0 | { |
1186 | 0 | more_ = false; |
1187 | 0 | } |
1188 | 0 | } |
1189 | 0 | --level_; |
1190 | 0 | } |
1191 | 23.9M | ++column_index_; |
1192 | 23.9M | state_ = csv_parse_state::before_unquoted_string; |
1193 | 23.9M | ++column_; |
1194 | 23.9M | ++input_ptr_; |
1195 | 23.9M | break; |
1196 | 2.54k | } |
1197 | 0 | case csv_parse_state::before_unquoted_field_tail1: |
1198 | 0 | { |
1199 | 0 | if (stack_.back() == csv_mode::subfields) |
1200 | 0 | { |
1201 | 0 | stack_.pop_back(); |
1202 | 0 | local_visitor.end_array(*this, ec); |
1203 | 0 | more_ = !cursor_mode_; |
1204 | 0 | if (mapping_kind_ != csv_mapping_kind::m_columns) |
1205 | 0 | { |
1206 | 0 | if (level() == mark_level_) |
1207 | 0 | { |
1208 | 0 | more_ = false; |
1209 | 0 | } |
1210 | 0 | } |
1211 | 0 | --level_; |
1212 | 0 | } |
1213 | 0 | state_ = csv_parse_state::end_record; |
1214 | 0 | ++column_; |
1215 | 0 | ++input_ptr_; |
1216 | 0 | break; |
1217 | 2.54k | } |
1218 | | |
1219 | 2.38M | case csv_parse_state::before_last_unquoted_field: |
1220 | 2.38M | end_unquoted_string_value(local_visitor, ec); |
1221 | 2.38M | state_ = csv_parse_state::before_last_unquoted_field_tail; |
1222 | 2.38M | break; |
1223 | | |
1224 | 2.38M | case csv_parse_state::before_last_unquoted_field_tail: |
1225 | 2.38M | if (stack_.back() == csv_mode::subfields) |
1226 | 0 | { |
1227 | 0 | stack_.pop_back(); |
1228 | 0 | local_visitor.end_array(*this, ec); |
1229 | 0 | more_ = !cursor_mode_; |
1230 | 0 | if (mapping_kind_ != csv_mapping_kind::m_columns) |
1231 | 0 | { |
1232 | 0 | if (level() == mark_level_) |
1233 | 0 | { |
1234 | 0 | more_ = false; |
1235 | 0 | } |
1236 | 0 | } |
1237 | 0 | --level_; |
1238 | 0 | } |
1239 | 2.38M | ++column_index_; |
1240 | 2.38M | state_ = csv_parse_state::end_record; |
1241 | 2.38M | break; |
1242 | | |
1243 | 0 | case csv_parse_state::before_unquoted_subfield: |
1244 | 0 | if (stack_.back() == csv_mode::data) |
1245 | 0 | { |
1246 | 0 | stack_.push_back(csv_mode::subfields); |
1247 | 0 | local_visitor.begin_array(semantic_tag::none, *this, ec); |
1248 | 0 | more_ = !cursor_mode_; |
1249 | 0 | ++level_; |
1250 | 0 | } |
1251 | 0 | state_ = csv_parse_state::before_unquoted_subfield_tail; |
1252 | 0 | break; |
1253 | 0 | case csv_parse_state::before_unquoted_subfield_tail: |
1254 | 0 | end_unquoted_string_value(local_visitor, ec); |
1255 | 0 | state_ = csv_parse_state::before_unquoted_string; |
1256 | 0 | ++column_; |
1257 | 0 | ++input_ptr_; |
1258 | 0 | break; |
1259 | 857 | case csv_parse_state::before_quoted_field: |
1260 | 857 | end_quoted_string_value(local_visitor, ec); |
1261 | 857 | state_ = csv_parse_state::before_unquoted_field_tail; // return to unquoted |
1262 | 857 | break; |
1263 | 0 | case csv_parse_state::before_quoted_subfield: |
1264 | 0 | if (stack_.back() == csv_mode::data) |
1265 | 0 | { |
1266 | 0 | stack_.push_back(csv_mode::subfields); |
1267 | 0 | local_visitor.begin_array(semantic_tag::none, *this, ec); |
1268 | 0 | more_ = !cursor_mode_; |
1269 | 0 | ++level_; |
1270 | 0 | } |
1271 | 0 | state_ = csv_parse_state::before_quoted_subfield_tail; |
1272 | 0 | break; |
1273 | 0 | case csv_parse_state::before_quoted_subfield_tail: |
1274 | 0 | end_quoted_string_value(local_visitor, ec); |
1275 | 0 | state_ = csv_parse_state::before_unquoted_string; |
1276 | 0 | ++column_; |
1277 | 0 | ++input_ptr_; |
1278 | 0 | break; |
1279 | 1.08k | case csv_parse_state::before_last_quoted_field: |
1280 | 1.08k | end_quoted_string_value(local_visitor, ec); |
1281 | 1.08k | state_ = csv_parse_state::before_last_quoted_field_tail; |
1282 | 1.08k | break; |
1283 | 1.08k | case csv_parse_state::before_last_quoted_field_tail: |
1284 | 1.08k | if (stack_.back() == csv_mode::subfields) |
1285 | 0 | { |
1286 | 0 | stack_.pop_back(); |
1287 | 0 | local_visitor.end_array(*this, ec); |
1288 | 0 | more_ = !cursor_mode_; |
1289 | 0 | if (mapping_kind_ != csv_mapping_kind::m_columns) |
1290 | 0 | { |
1291 | 0 | if (level() == mark_level_) |
1292 | 0 | { |
1293 | 0 | more_ = false; |
1294 | 0 | } |
1295 | 0 | } |
1296 | 0 | --level_; |
1297 | 0 | } |
1298 | 1.08k | ++column_index_; |
1299 | 1.08k | state_ = csv_parse_state::end_record; |
1300 | 1.08k | break; |
1301 | 60.4M | case csv_parse_state::unquoted_string: |
1302 | 60.4M | { |
1303 | 60.4M | switch (curr_char) |
1304 | 60.4M | { |
1305 | 689k | case '\n': |
1306 | 2.38M | case '\r': |
1307 | 2.38M | { |
1308 | 2.38M | if (trim_leading_ || trim_trailing_) |
1309 | 0 | { |
1310 | 0 | trim_string_buffer(trim_leading_,trim_trailing_); |
1311 | 0 | } |
1312 | 2.38M | if (!(ignore_empty_values_ && buffer_.empty())) |
1313 | 2.38M | { |
1314 | 2.38M | before_value(local_visitor, ec); |
1315 | 2.38M | state_ = csv_parse_state::before_last_unquoted_field; |
1316 | 2.38M | } |
1317 | 0 | else |
1318 | 0 | { |
1319 | 0 | state_ = csv_parse_state::end_record; |
1320 | 0 | } |
1321 | 2.38M | break; |
1322 | 689k | } |
1323 | 58.0M | default: |
1324 | 58.0M | if (curr_char == field_delimiter_) |
1325 | 23.9M | { |
1326 | 23.9M | if (trim_leading_ || trim_trailing_) |
1327 | 0 | { |
1328 | 0 | trim_string_buffer(trim_leading_,trim_trailing_); |
1329 | 0 | } |
1330 | 23.9M | before_value(local_visitor, ec); |
1331 | 23.9M | state_ = csv_parse_state::before_unquoted_field; |
1332 | 23.9M | } |
1333 | 34.0M | else if (subfield_delimiter_ != char_type() && curr_char == subfield_delimiter_) |
1334 | 0 | { |
1335 | 0 | if (trim_leading_ || trim_trailing_) |
1336 | 0 | { |
1337 | 0 | trim_string_buffer(trim_leading_,trim_trailing_); |
1338 | 0 | } |
1339 | 0 | before_value(local_visitor, ec); |
1340 | 0 | state_ = csv_parse_state::before_unquoted_subfield; |
1341 | 0 | } |
1342 | 34.0M | else if (curr_char == quote_char_) |
1343 | 907 | { |
1344 | 907 | buffer_.clear(); |
1345 | 907 | state_ = csv_parse_state::quoted_string; |
1346 | 907 | ++column_; |
1347 | 907 | ++input_ptr_; |
1348 | 907 | } |
1349 | 34.0M | else |
1350 | 34.0M | { |
1351 | 34.0M | buffer_.push_back(static_cast<CharT>(curr_char)); |
1352 | 34.0M | ++column_; |
1353 | 34.0M | ++input_ptr_; |
1354 | 34.0M | } |
1355 | 58.0M | break; |
1356 | 60.4M | } |
1357 | 60.4M | break; |
1358 | 60.4M | } |
1359 | 60.4M | case csv_parse_state::expect_record: |
1360 | 6.07M | { |
1361 | 6.07M | switch (curr_char) |
1362 | 6.07M | { |
1363 | 1.05M | case '\n': |
1364 | 1.05M | { |
1365 | 1.05M | if (!ignore_empty_lines_) |
1366 | 0 | { |
1367 | 0 | begin_record(local_visitor, ec); |
1368 | 0 | state_ = csv_parse_state::end_record; |
1369 | 0 | } |
1370 | 1.05M | else |
1371 | 1.05M | { |
1372 | 1.05M | ++line_; |
1373 | 1.05M | column_ = 1; |
1374 | 1.05M | state_ = csv_parse_state::expect_comment_or_record; |
1375 | 1.05M | ++input_ptr_; |
1376 | 1.05M | } |
1377 | 1.05M | break; |
1378 | 0 | } |
1379 | 2.64M | case '\r': |
1380 | 2.64M | if (!ignore_empty_lines_) |
1381 | 0 | { |
1382 | 0 | begin_record(local_visitor, ec); |
1383 | 0 | state_ = csv_parse_state::end_record; |
1384 | 0 | } |
1385 | 2.64M | else |
1386 | 2.64M | { |
1387 | 2.64M | ++input_ptr_; |
1388 | 2.64M | push_state(state_); |
1389 | 2.64M | state_ = csv_parse_state::cr; |
1390 | 2.64M | } |
1391 | 2.64M | break; |
1392 | 700 | case ' ': |
1393 | 45.4k | case '\t': |
1394 | 45.4k | if (!trim_leading_) |
1395 | 45.4k | { |
1396 | 45.4k | buffer_.push_back(static_cast<CharT>(curr_char)); |
1397 | 45.4k | begin_record(local_visitor, ec); |
1398 | 45.4k | state_ = csv_parse_state::unquoted_string; |
1399 | 45.4k | } |
1400 | 45.4k | ++column_; |
1401 | 45.4k | ++input_ptr_; |
1402 | 45.4k | break; |
1403 | 2.34M | default: |
1404 | 2.34M | begin_record(local_visitor, ec); |
1405 | 2.34M | if (curr_char == quote_char_) |
1406 | 1.24k | { |
1407 | 1.24k | buffer_.clear(); |
1408 | 1.24k | state_ = csv_parse_state::quoted_string; |
1409 | 1.24k | ++column_; |
1410 | 1.24k | ++input_ptr_; |
1411 | 1.24k | } |
1412 | 2.33M | else |
1413 | 2.33M | { |
1414 | 2.33M | state_ = csv_parse_state::unquoted_string; |
1415 | 2.33M | } |
1416 | 2.34M | break; |
1417 | 6.07M | } |
1418 | 6.07M | break; |
1419 | 6.07M | } |
1420 | 6.07M | case csv_parse_state::end_record: |
1421 | 2.38M | { |
1422 | 2.38M | switch (curr_char) |
1423 | 2.38M | { |
1424 | 690k | case '\n': |
1425 | 690k | { |
1426 | 690k | ++line_; |
1427 | 690k | column_ = 1; |
1428 | 690k | state_ = csv_parse_state::expect_comment_or_record; |
1429 | 690k | end_record(local_visitor, ec); |
1430 | 690k | ++input_ptr_; |
1431 | 690k | break; |
1432 | 0 | } |
1433 | 1.69M | case '\r': |
1434 | 1.69M | ++line_; |
1435 | 1.69M | column_ = 1; |
1436 | 1.69M | state_ = csv_parse_state::expect_comment_or_record; |
1437 | 1.69M | end_record(local_visitor, ec); |
1438 | 1.69M | push_state(state_); |
1439 | 1.69M | state_ = csv_parse_state::cr; |
1440 | 1.69M | ++input_ptr_; |
1441 | 1.69M | break; |
1442 | 0 | case ' ': |
1443 | 0 | case '\t': |
1444 | 0 | ++column_; |
1445 | 0 | ++input_ptr_; |
1446 | 0 | break; |
1447 | 0 | default: |
1448 | 0 | err_handler_(csv_errc::syntax_error, *this); |
1449 | 0 | ec = csv_errc::syntax_error; |
1450 | 0 | more_ = false; |
1451 | 0 | return; |
1452 | 2.38M | } |
1453 | 2.38M | break; |
1454 | 2.38M | } |
1455 | 2.38M | default: |
1456 | 0 | err_handler_(csv_errc::invalid_parse_state, *this); |
1457 | 0 | ec = csv_errc::invalid_parse_state; |
1458 | 0 | more_ = false; |
1459 | 0 | return; |
1460 | 166M | } |
1461 | 166M | if (line_ > max_lines_) |
1462 | 0 | { |
1463 | 0 | state_ = csv_parse_state::done; |
1464 | 0 | more_ = false; |
1465 | 0 | } |
1466 | 166M | } |
1467 | 14.0k | } |
1468 | | |
1469 | | void finish_parse() |
1470 | | { |
1471 | | std::error_code ec; |
1472 | | finish_parse(ec); |
1473 | | if (JSONCONS_UNLIKELY(ec)) |
1474 | | { |
1475 | | JSONCONS_THROW(ser_error(ec,line_,column_)); |
1476 | | } |
1477 | | } |
1478 | | |
1479 | | void finish_parse(std::error_code& ec) |
1480 | | { |
1481 | | while (more_) |
1482 | | { |
1483 | | parse_some(ec); |
1484 | | } |
1485 | | } |
1486 | | |
1487 | | csv_parse_state state() const |
1488 | | { |
1489 | | return state_; |
1490 | | } |
1491 | | |
1492 | | void update(const string_view_type sv) |
1493 | | { |
1494 | | update(sv.data(),sv.length()); |
1495 | | } |
1496 | | |
1497 | | void update(const CharT* data, std::size_t length) |
1498 | 2.51k | { |
1499 | 2.51k | begin_input_ = data; |
1500 | 2.51k | input_end_ = data + length; |
1501 | 2.51k | input_ptr_ = begin_input_; |
1502 | 2.51k | } |
1503 | | |
1504 | | std::size_t line() const override |
1505 | 91 | { |
1506 | 91 | return line_; |
1507 | 91 | } |
1508 | | |
1509 | | std::size_t column() const override |
1510 | 91 | { |
1511 | 91 | return column_; |
1512 | 91 | } |
1513 | | |
1514 | | private: |
1515 | | void initialize() |
1516 | 2.52k | { |
1517 | 2.52k | stack_.reserve(default_depth); |
1518 | 2.52k | stack_.push_back(csv_mode::initial); |
1519 | 2.52k | stack_.push_back((header_lines_ > 0) ? csv_mode::header : csv_mode::data); |
1520 | 2.52k | } |
1521 | | |
1522 | | // name |
1523 | | void before_value(basic_json_visitor<CharT>& visitor, |
1524 | | std::error_code& ec) |
1525 | 26.3M | { |
1526 | 26.3M | switch (stack_.back()) |
1527 | 26.3M | { |
1528 | 11.0M | case csv_mode::header: |
1529 | 11.0M | if (trim_leading_inside_quotes_ || trim_trailing_inside_quotes_) |
1530 | 0 | { |
1531 | 0 | trim_string_buffer(trim_leading_inside_quotes_,trim_trailing_inside_quotes_); |
1532 | 0 | } |
1533 | 11.0M | if (line_ == (header_line_+header_line_offset_) && column_index_ >= min_column_names_) |
1534 | 11.0M | { |
1535 | 11.0M | column_names_.push_back(buffer_); |
1536 | 11.0M | if (assume_header_ && mapping_kind_ == csv_mapping_kind::n_rows) |
1537 | 11.0M | { |
1538 | 11.0M | visitor.string_value(buffer_, semantic_tag::none, *this, ec); |
1539 | 11.0M | more_ = !cursor_mode_; |
1540 | 11.0M | } |
1541 | 11.0M | } |
1542 | 11.0M | break; |
1543 | 15.2M | case csv_mode::data: |
1544 | 15.2M | if (mapping_kind_ == csv_mapping_kind::n_objects) |
1545 | 0 | { |
1546 | 0 | if (!(ignore_empty_values_ && buffer_.empty())) |
1547 | 0 | { |
1548 | 0 | if (column_index_ < column_names_.size() + offset_) |
1549 | 0 | { |
1550 | 0 | visitor.key(column_names_[column_index_ - offset_], *this, ec); |
1551 | 0 | more_ = !cursor_mode_; |
1552 | 0 | } |
1553 | 0 | } |
1554 | 0 | } |
1555 | 15.2M | break; |
1556 | 0 | default: |
1557 | 0 | break; |
1558 | 26.3M | } |
1559 | 26.3M | } |
1560 | | |
1561 | | // begin_array or begin_record |
1562 | | void begin_record(basic_json_visitor<CharT>& visitor, std::error_code& ec) |
1563 | 2.38M | { |
1564 | 2.38M | offset_ = 0; |
1565 | | |
1566 | 2.38M | if (stack_.back() == csv_mode::header && line_ > (header_lines_+header_line_offset_)) |
1567 | 1.63k | { |
1568 | 1.63k | stack_.back() = csv_mode::data; |
1569 | 1.63k | } |
1570 | 2.38M | switch (stack_.back()) |
1571 | 2.38M | { |
1572 | 812 | case csv_mode::header: |
1573 | 812 | switch (mapping_kind_) |
1574 | 812 | { |
1575 | 812 | case csv_mapping_kind::n_rows: |
1576 | 812 | if (assume_header_ && line_ == (header_line_+header_line_offset_)) |
1577 | 812 | { |
1578 | 812 | visitor.begin_array(semantic_tag::none, *this, ec); |
1579 | 812 | more_ = !cursor_mode_; |
1580 | 812 | ++level_; |
1581 | 812 | } |
1582 | 812 | break; |
1583 | 0 | default: |
1584 | 0 | break; |
1585 | 812 | } |
1586 | 812 | break; |
1587 | 2.38M | case csv_mode::data: |
1588 | 2.38M | switch (mapping_kind_) |
1589 | 2.38M | { |
1590 | 2.38M | case csv_mapping_kind::n_rows: |
1591 | 2.38M | visitor.begin_array(semantic_tag::none, *this, ec); |
1592 | 2.38M | more_ = !cursor_mode_; |
1593 | 2.38M | ++level_; |
1594 | 2.38M | break; |
1595 | 0 | case csv_mapping_kind::n_objects: |
1596 | 0 | visitor.begin_object(semantic_tag::none, *this, ec); |
1597 | 0 | more_ = !cursor_mode_; |
1598 | 0 | ++level_; |
1599 | 0 | break; |
1600 | 0 | case csv_mapping_kind::m_columns: |
1601 | 0 | break; |
1602 | 0 | default: |
1603 | 0 | break; |
1604 | 2.38M | } |
1605 | 2.38M | break; |
1606 | 2.38M | default: |
1607 | 0 | break; |
1608 | 2.38M | } |
1609 | 2.38M | } |
1610 | | |
1611 | | // end_array, begin_array, string_value (headers) |
1612 | | void end_record(basic_json_visitor<CharT>& visitor, std::error_code& ec) |
1613 | 2.38M | { |
1614 | 2.38M | if (!column_types_.empty()) |
1615 | 0 | { |
1616 | 0 | switch (mapping_kind_) |
1617 | 0 | { |
1618 | 0 | case csv_mapping_kind::n_rows: |
1619 | 0 | case csv_mapping_kind::n_objects: |
1620 | 0 | if (depth_ > 0) |
1621 | 0 | { |
1622 | 0 | visitor.end_array(*this, ec); |
1623 | 0 | more_ = !cursor_mode_; |
1624 | 0 | if (level_ == mark_level_) |
1625 | 0 | { |
1626 | 0 | more_ = false; |
1627 | 0 | } |
1628 | 0 | --level_; |
1629 | 0 | depth_ = 0; |
1630 | 0 | } |
1631 | 0 | break; |
1632 | 0 | case csv_mapping_kind::m_columns: |
1633 | 0 | if (depth_ > 0) |
1634 | 0 | { |
1635 | 0 | visitor.end_array(*this, ec); |
1636 | 0 | more_ = !cursor_mode_; |
1637 | 0 | --level_; |
1638 | 0 | depth_ = 0; |
1639 | 0 | } |
1640 | 0 | break; |
1641 | 0 | default: |
1642 | 0 | break; |
1643 | 0 | } |
1644 | 0 | } |
1645 | 2.38M | switch (stack_.back()) |
1646 | 2.38M | { |
1647 | 713 | case csv_mode::header: |
1648 | 713 | if (line_ >= header_lines_) |
1649 | 713 | { |
1650 | 713 | stack_.back() = csv_mode::data; |
1651 | 713 | } |
1652 | 713 | switch (mapping_kind_) |
1653 | 713 | { |
1654 | 713 | case csv_mapping_kind::n_rows: |
1655 | 713 | if (assume_header_) |
1656 | 713 | { |
1657 | 713 | visitor.end_array(*this, ec); |
1658 | 713 | more_ = !cursor_mode_; |
1659 | 713 | if (level() == mark_level_) |
1660 | 0 | { |
1661 | 0 | more_ = false; |
1662 | 0 | } |
1663 | 713 | --level_; |
1664 | 713 | } |
1665 | 713 | break; |
1666 | 0 | case csv_mapping_kind::m_columns: |
1667 | 0 | m_columns_filter_.initialize(column_names_); |
1668 | 0 | break; |
1669 | 0 | default: |
1670 | 0 | break; |
1671 | 713 | } |
1672 | 713 | break; |
1673 | 2.38M | case csv_mode::data: |
1674 | 2.38M | case csv_mode::subfields: |
1675 | 2.38M | { |
1676 | 2.38M | switch (mapping_kind_) |
1677 | 2.38M | { |
1678 | 2.38M | case csv_mapping_kind::n_rows: |
1679 | 2.38M | visitor.end_array(*this, ec); |
1680 | 2.38M | more_ = !cursor_mode_; |
1681 | 2.38M | if (level_ == mark_level_) |
1682 | 0 | { |
1683 | 0 | more_ = false; |
1684 | 0 | } |
1685 | 2.38M | --level_; |
1686 | 2.38M | break; |
1687 | 0 | case csv_mapping_kind::n_objects: |
1688 | 0 | visitor.end_object(*this, ec); |
1689 | 0 | more_ = !cursor_mode_; |
1690 | 0 | if (level_ == mark_level_) |
1691 | 0 | { |
1692 | 0 | more_ = false; |
1693 | 0 | } |
1694 | 0 | --level_; |
1695 | 0 | break; |
1696 | 0 | case csv_mapping_kind::m_columns: |
1697 | 0 | visitor.end_array(*this, ec); |
1698 | 0 | more_ = !cursor_mode_; |
1699 | 0 | --level_; |
1700 | 0 | break; |
1701 | 2.38M | } |
1702 | 2.38M | break; |
1703 | 2.38M | } |
1704 | 2.38M | default: |
1705 | 0 | break; |
1706 | 2.38M | } |
1707 | 2.38M | column_index_ = 0; |
1708 | 2.38M | } |
1709 | | |
1710 | | void trim_string_buffer(bool trim_leading, bool trim_trailing) |
1711 | 0 | { |
1712 | 0 | std::size_t start = 0; |
1713 | 0 | std::size_t length = buffer_.length(); |
1714 | 0 | if (trim_leading) |
1715 | 0 | { |
1716 | 0 | bool done = false; |
1717 | 0 | while (!done && start < buffer_.length()) |
1718 | 0 | { |
1719 | 0 | if ((buffer_[start] < 256) && std::isspace(buffer_[start])) |
1720 | 0 | { |
1721 | 0 | ++start; |
1722 | 0 | } |
1723 | 0 | else |
1724 | 0 | { |
1725 | 0 | done = true; |
1726 | 0 | } |
1727 | 0 | } |
1728 | 0 | } |
1729 | 0 | if (trim_trailing) |
1730 | 0 | { |
1731 | 0 | bool done = false; |
1732 | 0 | while (!done && length > 0) |
1733 | 0 | { |
1734 | 0 | if ((buffer_[length-1] < 256) && std::isspace(buffer_[length-1])) |
1735 | 0 | { |
1736 | 0 | --length; |
1737 | 0 | } |
1738 | 0 | else |
1739 | 0 | { |
1740 | 0 | done = true; |
1741 | 0 | } |
1742 | 0 | } |
1743 | 0 | } |
1744 | 0 | if (start != 0 || length != buffer_.size()) |
1745 | 0 | { |
1746 | | // Do not use buffer_.substr(...), as this won't preserve the allocator state. |
1747 | 0 | buffer_.resize(length); |
1748 | 0 | buffer_.erase(0, start); |
1749 | 0 | } |
1750 | 0 | } |
1751 | | |
1752 | | /* |
1753 | | end_array, begin_array, xxx_value (end_value) |
1754 | | */ |
1755 | | void end_unquoted_string_value(basic_json_visitor<CharT>& visitor, std::error_code& ec) |
1756 | 26.3M | { |
1757 | 26.3M | switch (stack_.back()) |
1758 | 26.3M | { |
1759 | 15.2M | case csv_mode::data: |
1760 | 15.2M | case csv_mode::subfields: |
1761 | 15.2M | switch (mapping_kind_) |
1762 | 15.2M | { |
1763 | 15.2M | case csv_mapping_kind::n_rows: |
1764 | 15.2M | if (unquoted_empty_value_is_null_ && buffer_.empty()) |
1765 | 0 | { |
1766 | 0 | visitor.null_value(semantic_tag::none, *this, ec); |
1767 | 0 | more_ = !cursor_mode_; |
1768 | 0 | } |
1769 | 15.2M | else |
1770 | 15.2M | { |
1771 | 15.2M | end_value(visitor, infer_types_, ec); |
1772 | 15.2M | } |
1773 | 15.2M | break; |
1774 | 0 | case csv_mapping_kind::n_objects: |
1775 | 0 | if (!(ignore_empty_values_ && buffer_.empty())) |
1776 | 0 | { |
1777 | 0 | if (column_index_ < column_names_.size() + offset_) |
1778 | 0 | { |
1779 | 0 | if (unquoted_empty_value_is_null_ && buffer_.empty()) |
1780 | 0 | { |
1781 | 0 | visitor.null_value(semantic_tag::none, *this, ec); |
1782 | 0 | more_ = !cursor_mode_; |
1783 | 0 | } |
1784 | 0 | else |
1785 | 0 | { |
1786 | 0 | end_value(visitor, infer_types_, ec); |
1787 | 0 | } |
1788 | 0 | } |
1789 | 0 | else if (depth_ > 0) |
1790 | 0 | { |
1791 | 0 | if (unquoted_empty_value_is_null_ && buffer_.empty()) |
1792 | 0 | { |
1793 | 0 | visitor.null_value(semantic_tag::none, *this, ec); |
1794 | 0 | more_ = !cursor_mode_; |
1795 | 0 | } |
1796 | 0 | else |
1797 | 0 | { |
1798 | 0 | end_value(visitor, infer_types_, ec); |
1799 | 0 | } |
1800 | 0 | } |
1801 | 0 | } |
1802 | 0 | break; |
1803 | 0 | case csv_mapping_kind::m_columns: |
1804 | 0 | if (!(ignore_empty_values_ && buffer_.empty())) |
1805 | 0 | { |
1806 | 0 | end_value(visitor, infer_types_, ec); |
1807 | 0 | } |
1808 | 0 | else |
1809 | 0 | { |
1810 | 0 | m_columns_filter_.skip_column(); |
1811 | 0 | } |
1812 | 0 | break; |
1813 | 15.2M | } |
1814 | 15.2M | break; |
1815 | 15.2M | default: |
1816 | 11.0M | break; |
1817 | 26.3M | } |
1818 | 26.3M | } |
1819 | | |
1820 | | void end_quoted_string_value(basic_json_visitor<CharT>& visitor, |
1821 | | std::error_code& ec) |
1822 | 1.96k | { |
1823 | 1.96k | switch (stack_.back()) |
1824 | 1.96k | { |
1825 | 1.17k | case csv_mode::data: |
1826 | 1.17k | case csv_mode::subfields: |
1827 | 1.17k | if (trim_leading_inside_quotes_ || trim_trailing_inside_quotes_) |
1828 | 0 | { |
1829 | 0 | trim_string_buffer(trim_leading_inside_quotes_,trim_trailing_inside_quotes_); |
1830 | 0 | } |
1831 | 1.17k | switch (mapping_kind_) |
1832 | 1.17k | { |
1833 | 1.17k | case csv_mapping_kind::n_rows: |
1834 | 1.17k | end_value(visitor, false, ec); |
1835 | 1.17k | break; |
1836 | 0 | case csv_mapping_kind::n_objects: |
1837 | 0 | if (!(ignore_empty_values_ && buffer_.empty())) |
1838 | 0 | { |
1839 | 0 | if (column_index_ < column_names_.size() + offset_) |
1840 | 0 | { |
1841 | 0 | if (unquoted_empty_value_is_null_ && buffer_.empty()) |
1842 | 0 | { |
1843 | 0 | visitor.null_value(semantic_tag::none, *this, ec); |
1844 | 0 | more_ = !cursor_mode_; |
1845 | 0 | } |
1846 | 0 | else |
1847 | 0 | { |
1848 | 0 | end_value(visitor, false, ec); |
1849 | 0 | } |
1850 | 0 | } |
1851 | 0 | else if (depth_ > 0) |
1852 | 0 | { |
1853 | 0 | if (unquoted_empty_value_is_null_ && buffer_.empty()) |
1854 | 0 | { |
1855 | 0 | visitor.null_value(semantic_tag::none, *this, ec); |
1856 | 0 | more_ = !cursor_mode_; |
1857 | 0 | } |
1858 | 0 | else |
1859 | 0 | { |
1860 | 0 | end_value(visitor, false, ec); |
1861 | 0 | } |
1862 | 0 | } |
1863 | 0 | } |
1864 | 0 | break; |
1865 | 0 | case csv_mapping_kind::m_columns: |
1866 | 0 | if (!(ignore_empty_values_ && buffer_.empty())) |
1867 | 0 | { |
1868 | 0 | end_value(visitor, false, ec); |
1869 | 0 | } |
1870 | 0 | else |
1871 | 0 | { |
1872 | 0 | m_columns_filter_.skip_column(); |
1873 | 0 | } |
1874 | 0 | break; |
1875 | 1.17k | } |
1876 | 1.17k | break; |
1877 | 1.17k | default: |
1878 | 794 | break; |
1879 | 1.96k | } |
1880 | 1.96k | } |
1881 | | |
1882 | | void end_value(basic_json_visitor<CharT>& visitor, |
1883 | | bool infer_types, std::error_code& ec) |
1884 | 15.2M | { |
1885 | 15.2M | auto it = std::find_if(string_double_map_.begin(), string_double_map_.end(), string_maps_to_double{ buffer_ }); |
1886 | 15.2M | if (it != string_double_map_.end()) |
1887 | 0 | { |
1888 | 0 | visitor.double_value((*it).second, semantic_tag::none, *this, ec); |
1889 | 0 | more_ = !cursor_mode_; |
1890 | 0 | } |
1891 | 15.2M | else if (column_index_ < column_types_.size() + offset_) |
1892 | 0 | { |
1893 | 0 | if (column_types_[column_index_ - offset_].col_type == csv_column_type::repeat_t) |
1894 | 0 | { |
1895 | 0 | offset_ = offset_ + column_types_[column_index_ - offset_].rep_count; |
1896 | 0 | if (column_index_ - offset_ + 1 < column_types_.size()) |
1897 | 0 | { |
1898 | 0 | if (column_index_ == offset_ || depth_ > column_types_[column_index_-offset_].level) |
1899 | 0 | { |
1900 | 0 | visitor.end_array(*this, ec); |
1901 | 0 | more_ = !cursor_mode_; |
1902 | 0 | if (mapping_kind_ != csv_mapping_kind::m_columns) |
1903 | 0 | { |
1904 | 0 | if (level() == mark_level_) |
1905 | 0 | { |
1906 | 0 | more_ = false; |
1907 | 0 | } |
1908 | 0 | } |
1909 | 0 | --level_; |
1910 | 0 | } |
1911 | 0 | depth_ = column_index_ == offset_ ? 0 : column_types_[column_index_ - offset_].level; |
1912 | 0 | } |
1913 | 0 | } |
1914 | 0 | if (depth_ < column_types_[column_index_ - offset_].level) |
1915 | 0 | { |
1916 | 0 | visitor.begin_array(semantic_tag::none, *this, ec); |
1917 | 0 | more_ = !cursor_mode_; |
1918 | 0 | depth_ = column_types_[column_index_ - offset_].level; |
1919 | 0 | ++level_; |
1920 | 0 | } |
1921 | 0 | else if (depth_ > column_types_[column_index_ - offset_].level) |
1922 | 0 | { |
1923 | 0 | visitor.end_array(*this, ec); |
1924 | 0 | more_ = !cursor_mode_; |
1925 | 0 | if (mapping_kind_ != csv_mapping_kind::m_columns) |
1926 | 0 | { |
1927 | 0 | if (level() == mark_level_) |
1928 | 0 | { |
1929 | 0 | more_ = false; |
1930 | 0 | } |
1931 | 0 | } |
1932 | 0 | --level_; |
1933 | 0 | depth_ = column_types_[column_index_ - offset_].level; |
1934 | 0 | } |
1935 | 0 | switch (column_types_[column_index_ - offset_].col_type) |
1936 | 0 | { |
1937 | 0 | case csv_column_type::integer_t: |
1938 | 0 | { |
1939 | 0 | std::basic_istringstream<CharT,std::char_traits<CharT>,char_allocator_type> iss{buffer_}; |
1940 | 0 | int64_t val; |
1941 | 0 | iss >> val; |
1942 | 0 | if (!iss.fail()) |
1943 | 0 | { |
1944 | 0 | visitor.int64_value(val, semantic_tag::none, *this, ec); |
1945 | 0 | more_ = !cursor_mode_; |
1946 | 0 | } |
1947 | 0 | else |
1948 | 0 | { |
1949 | 0 | if (column_index_ - offset_ < column_defaults_.size() && column_defaults_[column_index_ - offset_].length() > 0) |
1950 | 0 | { |
1951 | 0 | basic_json_parser<CharT,temp_allocator_type> parser(alloc_); |
1952 | 0 | parser.update(column_defaults_[column_index_ - offset_].data(),column_defaults_[column_index_ - offset_].length()); |
1953 | 0 | parser.parse_some(visitor); |
1954 | 0 | parser.finish_parse(visitor); |
1955 | 0 | } |
1956 | 0 | else |
1957 | 0 | { |
1958 | 0 | visitor.null_value(semantic_tag::none, *this, ec); |
1959 | 0 | more_ = !cursor_mode_; |
1960 | 0 | } |
1961 | 0 | } |
1962 | 0 | } |
1963 | 0 | break; |
1964 | 0 | case csv_column_type::float_t: |
1965 | 0 | { |
1966 | 0 | if (lossless_number_) |
1967 | 0 | { |
1968 | 0 | visitor.string_value(buffer_,semantic_tag::bigdec, *this, ec); |
1969 | 0 | more_ = !cursor_mode_; |
1970 | 0 | } |
1971 | 0 | else |
1972 | 0 | { |
1973 | 0 | std::basic_istringstream<CharT, std::char_traits<CharT>, char_allocator_type> iss{ buffer_ }; |
1974 | 0 | double val; |
1975 | 0 | iss >> val; |
1976 | 0 | if (!iss.fail()) |
1977 | 0 | { |
1978 | 0 | visitor.double_value(val, semantic_tag::none, *this, ec); |
1979 | 0 | more_ = !cursor_mode_; |
1980 | 0 | } |
1981 | 0 | else |
1982 | 0 | { |
1983 | 0 | if (column_index_ - offset_ < column_defaults_.size() && column_defaults_[column_index_ - offset_].length() > 0) |
1984 | 0 | { |
1985 | 0 | basic_json_parser<CharT,temp_allocator_type> parser(alloc_); |
1986 | 0 | parser.update(column_defaults_[column_index_ - offset_].data(),column_defaults_[column_index_ - offset_].length()); |
1987 | 0 | parser.parse_some(visitor); |
1988 | 0 | parser.finish_parse(visitor); |
1989 | 0 | } |
1990 | 0 | else |
1991 | 0 | { |
1992 | 0 | visitor.null_value(semantic_tag::none, *this, ec); |
1993 | 0 | more_ = !cursor_mode_; |
1994 | 0 | } |
1995 | 0 | } |
1996 | 0 | } |
1997 | 0 | } |
1998 | 0 | break; |
1999 | 0 | case csv_column_type::boolean_t: |
2000 | 0 | { |
2001 | 0 | if (buffer_.length() == 1 && buffer_[0] == '0') |
2002 | 0 | { |
2003 | 0 | visitor.bool_value(false, semantic_tag::none, *this, ec); |
2004 | 0 | more_ = !cursor_mode_; |
2005 | 0 | } |
2006 | 0 | else if (buffer_.length() == 1 && buffer_[0] == '1') |
2007 | 0 | { |
2008 | 0 | visitor.bool_value(true, semantic_tag::none, *this, ec); |
2009 | 0 | more_ = !cursor_mode_; |
2010 | 0 | } |
2011 | 0 | else if (buffer_.length() == 5 && ((buffer_[0] == 'f' || buffer_[0] == 'F') && (buffer_[1] == 'a' || buffer_[1] == 'A') && (buffer_[2] == 'l' || buffer_[2] == 'L') && (buffer_[3] == 's' || buffer_[3] == 'S') && (buffer_[4] == 'e' || buffer_[4] == 'E'))) |
2012 | 0 | { |
2013 | 0 | visitor.bool_value(false, semantic_tag::none, *this, ec); |
2014 | 0 | more_ = !cursor_mode_; |
2015 | 0 | } |
2016 | 0 | else if (buffer_.length() == 4 && ((buffer_[0] == 't' || buffer_[0] == 'T') && (buffer_[1] == 'r' || buffer_[1] == 'R') && (buffer_[2] == 'u' || buffer_[2] == 'U') && (buffer_[3] == 'e' || buffer_[3] == 'E'))) |
2017 | 0 | { |
2018 | 0 | visitor.bool_value(true, semantic_tag::none, *this, ec); |
2019 | 0 | more_ = !cursor_mode_; |
2020 | 0 | } |
2021 | 0 | else |
2022 | 0 | { |
2023 | 0 | if (column_index_ - offset_ < column_defaults_.size() && column_defaults_[column_index_ - offset_].length() > 0) |
2024 | 0 | { |
2025 | 0 | basic_json_parser<CharT,temp_allocator_type> parser(alloc_); |
2026 | 0 | parser.update(column_defaults_[column_index_ - offset_].data(),column_defaults_[column_index_ - offset_].length()); |
2027 | 0 | parser.parse_some(visitor); |
2028 | 0 | parser.finish_parse(visitor); |
2029 | 0 | } |
2030 | 0 | else |
2031 | 0 | { |
2032 | 0 | visitor.null_value(semantic_tag::none, *this, ec); |
2033 | 0 | more_ = !cursor_mode_; |
2034 | 0 | } |
2035 | 0 | } |
2036 | 0 | } |
2037 | 0 | break; |
2038 | 0 | default: |
2039 | 0 | if (buffer_.length() > 0) |
2040 | 0 | { |
2041 | 0 | visitor.string_value(buffer_, semantic_tag::none, *this, ec); |
2042 | 0 | more_ = !cursor_mode_; |
2043 | 0 | } |
2044 | 0 | else |
2045 | 0 | { |
2046 | 0 | if (column_index_ < column_defaults_.size() + offset_ && column_defaults_[column_index_ - offset_].length() > 0) |
2047 | 0 | { |
2048 | 0 | basic_json_parser<CharT,temp_allocator_type> parser(alloc_); |
2049 | 0 | parser.update(column_defaults_[column_index_ - offset_].data(),column_defaults_[column_index_ - offset_].length()); |
2050 | 0 | parser.parse_some(visitor); |
2051 | 0 | parser.finish_parse(visitor); |
2052 | 0 | } |
2053 | 0 | else |
2054 | 0 | { |
2055 | 0 | visitor.string_value(string_view_type(), semantic_tag::none, *this, ec); |
2056 | 0 | more_ = !cursor_mode_; |
2057 | 0 | } |
2058 | 0 | } |
2059 | 0 | break; |
2060 | 0 | } |
2061 | 0 | } |
2062 | 15.2M | else |
2063 | 15.2M | { |
2064 | 15.2M | if (infer_types) |
2065 | 15.2M | { |
2066 | 15.2M | end_value_with_numeric_check(visitor, ec); |
2067 | 15.2M | } |
2068 | 1.17k | else |
2069 | 1.17k | { |
2070 | 1.17k | visitor.string_value(buffer_, semantic_tag::none, *this, ec); |
2071 | 1.17k | more_ = !cursor_mode_; |
2072 | 1.17k | } |
2073 | 15.2M | } |
2074 | 15.2M | } |
2075 | | |
2076 | | enum class numeric_check_state |
2077 | | { |
2078 | | initial, |
2079 | | null, |
2080 | | boolean_true, |
2081 | | boolean_false, |
2082 | | minus, |
2083 | | zero, |
2084 | | integer, |
2085 | | fraction1, |
2086 | | fraction, |
2087 | | exp1, |
2088 | | exp, |
2089 | | not_a_number |
2090 | | }; |
2091 | | |
2092 | | /* |
2093 | | xxx_value |
2094 | | */ |
2095 | | void end_value_with_numeric_check(basic_json_visitor<CharT>& visitor, |
2096 | | std::error_code& ec) |
2097 | 15.2M | { |
2098 | 15.2M | numeric_check_state state = numeric_check_state::initial; |
2099 | 15.2M | bool is_negative = false; |
2100 | | //int precision = 0; |
2101 | | //uint8_t decimal_places = 0; |
2102 | | |
2103 | 15.2M | auto last = buffer_.end(); |
2104 | | |
2105 | 15.2M | std::string buffer; |
2106 | 31.3M | for (auto p = buffer_.begin(); state != numeric_check_state::not_a_number && p != last; ++p) |
2107 | 16.1M | { |
2108 | 16.1M | switch (state) |
2109 | 16.1M | { |
2110 | 2.29M | case numeric_check_state::initial: |
2111 | 2.29M | { |
2112 | 2.29M | switch (*p) |
2113 | 2.29M | { |
2114 | 64.1k | case 'n':case 'N': |
2115 | 64.1k | if ((last-p) == 4 && (p[1] == 'u' || p[1] == 'U') && (p[2] == 'l' || p[2] == 'L') && (p[3] == 'l' || p[3] == 'L')) |
2116 | 37.2k | { |
2117 | 37.2k | state = numeric_check_state::null; |
2118 | 37.2k | } |
2119 | 26.8k | else |
2120 | 26.8k | { |
2121 | 26.8k | state = numeric_check_state::not_a_number; |
2122 | 26.8k | } |
2123 | 64.1k | break; |
2124 | 123k | case 't':case 'T': |
2125 | 123k | if ((last-p) == 4 && (p[1] == 'r' || p[1] == 'R') && (p[2] == 'u' || p[2] == 'U') && (p[3] == 'e' || p[3] == 'U')) |
2126 | 110k | { |
2127 | 110k | state = numeric_check_state::boolean_true; |
2128 | 110k | } |
2129 | 13.0k | else |
2130 | 13.0k | { |
2131 | 13.0k | state = numeric_check_state::not_a_number; |
2132 | 13.0k | } |
2133 | 123k | break; |
2134 | 3.89k | case 'f':case 'F': |
2135 | 3.89k | if ((last-p) == 5 && (p[1] == 'a' || p[1] == 'A') && (p[2] == 'l' || p[2] == 'L') && (p[3] == 's' || p[3] == 'S') && (p[4] == 'e' || p[4] == 'E')) |
2136 | 967 | { |
2137 | 967 | state = numeric_check_state::boolean_false; |
2138 | 967 | } |
2139 | 2.93k | else |
2140 | 2.93k | { |
2141 | 2.93k | state = numeric_check_state::not_a_number; |
2142 | 2.93k | } |
2143 | 3.89k | break; |
2144 | 415k | case '-': |
2145 | 415k | is_negative = true; |
2146 | 415k | buffer.push_back(*p); |
2147 | 415k | state = numeric_check_state::minus; |
2148 | 415k | break; |
2149 | 133k | case '0': |
2150 | | //++precision; |
2151 | 133k | buffer.push_back(*p); |
2152 | 133k | state = numeric_check_state::zero; |
2153 | 133k | break; |
2154 | 167k | case '1':case '2':case '3':case '4':case '5':case '6':case '7':case '8':case '9': |
2155 | | //++precision; |
2156 | 167k | buffer.push_back(*p); |
2157 | 167k | state = numeric_check_state::integer; |
2158 | 167k | break; |
2159 | 1.38M | default: |
2160 | 1.38M | state = numeric_check_state::not_a_number; |
2161 | 1.38M | break; |
2162 | 2.29M | } |
2163 | 2.29M | break; |
2164 | 2.29M | } |
2165 | 2.29M | case numeric_check_state::zero: |
2166 | 97.3k | { |
2167 | 97.3k | switch (*p) |
2168 | 97.3k | { |
2169 | 77.8k | case '.': |
2170 | 77.8k | { |
2171 | 77.8k | buffer.push_back('.'); |
2172 | 77.8k | state = numeric_check_state::fraction1; |
2173 | 77.8k | } |
2174 | 77.8k | break; |
2175 | 8.19k | case 'e':case 'E': |
2176 | 8.19k | buffer.push_back(*p); |
2177 | 8.19k | state = numeric_check_state::exp1; |
2178 | 8.19k | break; |
2179 | 11.3k | default: |
2180 | 11.3k | state = numeric_check_state::not_a_number; |
2181 | 11.3k | break; |
2182 | 97.3k | } |
2183 | 97.3k | break; |
2184 | 97.3k | } |
2185 | 12.7M | case numeric_check_state::integer: |
2186 | 12.7M | { |
2187 | 12.7M | switch (*p) |
2188 | 12.7M | { |
2189 | 12.6M | case '0':case '1':case '2':case '3':case '4':case '5':case '6':case '7':case '8':case '9': |
2190 | | //++precision; |
2191 | 12.6M | buffer.push_back(*p); |
2192 | 12.6M | break; |
2193 | 18.4k | case '.': |
2194 | 18.4k | buffer.push_back('.'); |
2195 | 18.4k | state = numeric_check_state::fraction1; |
2196 | 18.4k | break; |
2197 | 9.93k | case 'e':case 'E': |
2198 | 9.93k | buffer.push_back(*p); |
2199 | 9.93k | state = numeric_check_state::exp1; |
2200 | 9.93k | break; |
2201 | 11.0k | default: |
2202 | 11.0k | state = numeric_check_state::not_a_number; |
2203 | 11.0k | break; |
2204 | 12.7M | } |
2205 | 12.7M | break; |
2206 | 12.7M | } |
2207 | 12.7M | case numeric_check_state::minus: |
2208 | 397k | { |
2209 | 397k | switch (*p) |
2210 | 397k | { |
2211 | 247k | case '0': |
2212 | | //++precision; |
2213 | 247k | buffer.push_back(*p); |
2214 | 247k | state = numeric_check_state::zero; |
2215 | 247k | break; |
2216 | 142k | case '1':case '2':case '3':case '4':case '5':case '6':case '7':case '8':case '9': |
2217 | | //++precision; |
2218 | 142k | buffer.push_back(*p); |
2219 | 142k | state = numeric_check_state::integer; |
2220 | 142k | break; |
2221 | 6.27k | default: |
2222 | 6.27k | state = numeric_check_state::not_a_number; |
2223 | 6.27k | break; |
2224 | 397k | } |
2225 | 397k | break; |
2226 | 397k | } |
2227 | 397k | case numeric_check_state::fraction1: |
2228 | 92.6k | { |
2229 | 92.6k | switch (*p) |
2230 | 92.6k | { |
2231 | 88.6k | case '0':case '1':case '2':case '3':case '4':case '5':case '6':case '7':case '8':case '9': |
2232 | | //++precision; |
2233 | | //++decimal_places; |
2234 | 88.6k | buffer.push_back(*p); |
2235 | 88.6k | state = numeric_check_state::fraction; |
2236 | 88.6k | break; |
2237 | 3.96k | default: |
2238 | 3.96k | state = numeric_check_state::not_a_number; |
2239 | 3.96k | break; |
2240 | 92.6k | } |
2241 | 92.6k | break; |
2242 | 92.6k | } |
2243 | 92.6k | case numeric_check_state::fraction: |
2244 | 14.3k | { |
2245 | 14.3k | switch (*p) |
2246 | 14.3k | { |
2247 | 12.4k | case '0':case '1':case '2':case '3':case '4':case '5':case '6':case '7':case '8':case '9': |
2248 | | //++precision; |
2249 | | //++decimal_places; |
2250 | 12.4k | buffer.push_back(*p); |
2251 | 12.4k | break; |
2252 | 927 | case 'e':case 'E': |
2253 | 927 | buffer.push_back(*p); |
2254 | 927 | state = numeric_check_state::exp1; |
2255 | 927 | break; |
2256 | 933 | default: |
2257 | 933 | state = numeric_check_state::not_a_number; |
2258 | 933 | break; |
2259 | 14.3k | } |
2260 | 14.3k | break; |
2261 | 14.3k | } |
2262 | 15.6k | case numeric_check_state::exp1: |
2263 | 15.6k | { |
2264 | 15.6k | switch (*p) |
2265 | 15.6k | { |
2266 | 1.34k | case '-': |
2267 | 1.34k | buffer.push_back(*p); |
2268 | 1.34k | break; |
2269 | 1.52k | case '+': |
2270 | 1.52k | break; |
2271 | 9.35k | case '0':case '1':case '2':case '3':case '4':case '5':case '6':case '7':case '8':case '9': |
2272 | 9.35k | state = numeric_check_state::exp; |
2273 | 9.35k | buffer.push_back(*p); |
2274 | 9.35k | break; |
2275 | 3.44k | default: |
2276 | 3.44k | state = numeric_check_state::not_a_number; |
2277 | 3.44k | break; |
2278 | 15.6k | } |
2279 | 15.6k | break; |
2280 | 15.6k | } |
2281 | 29.4k | case numeric_check_state::exp: |
2282 | 29.4k | { |
2283 | 29.4k | switch (*p) |
2284 | 29.4k | { |
2285 | 24.9k | case '0':case '1':case '2':case '3':case '4':case '5':case '6':case '7':case '8':case '9': |
2286 | 24.9k | buffer.push_back(*p); |
2287 | 24.9k | break; |
2288 | 4.48k | default: |
2289 | 4.48k | state = numeric_check_state::not_a_number; |
2290 | 4.48k | break; |
2291 | 29.4k | } |
2292 | 29.4k | break; |
2293 | 29.4k | } |
2294 | 447k | default: |
2295 | 447k | break; |
2296 | 16.1M | } |
2297 | 16.1M | } |
2298 | | |
2299 | 15.2M | switch (state) |
2300 | 15.2M | { |
2301 | 37.2k | case numeric_check_state::null: |
2302 | 37.2k | visitor.null_value(semantic_tag::none, *this, ec); |
2303 | 37.2k | more_ = !cursor_mode_; |
2304 | 37.2k | break; |
2305 | 110k | case numeric_check_state::boolean_true: |
2306 | 110k | visitor.bool_value(true, semantic_tag::none, *this, ec); |
2307 | 110k | more_ = !cursor_mode_; |
2308 | 110k | break; |
2309 | 967 | case numeric_check_state::boolean_false: |
2310 | 967 | visitor.bool_value(false, semantic_tag::none, *this, ec); |
2311 | 967 | more_ = !cursor_mode_; |
2312 | 967 | break; |
2313 | 284k | case numeric_check_state::zero: |
2314 | 554k | case numeric_check_state::integer: |
2315 | 554k | { |
2316 | 554k | if (is_negative) |
2317 | 378k | { |
2318 | 378k | int64_t val{ 0 }; |
2319 | 378k | auto result = jsoncons::utility::dec_to_integer(buffer_.data(), buffer_.length(), val); |
2320 | 378k | if (result) |
2321 | 375k | { |
2322 | 375k | visitor.int64_value(val, semantic_tag::none, *this, ec); |
2323 | 375k | more_ = !cursor_mode_; |
2324 | 375k | } |
2325 | 3.59k | else // Must be overflow |
2326 | 3.59k | { |
2327 | 3.59k | visitor.string_value(buffer_, semantic_tag::bigint, *this, ec); |
2328 | 3.59k | more_ = !cursor_mode_; |
2329 | 3.59k | } |
2330 | 378k | } |
2331 | 176k | else |
2332 | 176k | { |
2333 | 176k | uint64_t val{ 0 }; |
2334 | 176k | auto result = jsoncons::utility::dec_to_integer(buffer_.data(), buffer_.length(), val); |
2335 | 176k | if (result) |
2336 | 173k | { |
2337 | 173k | visitor.uint64_value(val, semantic_tag::none, *this, ec); |
2338 | 173k | more_ = !cursor_mode_; |
2339 | 173k | } |
2340 | 2.31k | else if (result.ec == std::errc::result_out_of_range) |
2341 | 2.31k | { |
2342 | 2.31k | visitor.string_value(buffer_, semantic_tag::bigint, *this, ec); |
2343 | 2.31k | more_ = !cursor_mode_; |
2344 | 2.31k | } |
2345 | 0 | else |
2346 | 0 | { |
2347 | 0 | ec = csv_errc::invalid_number; |
2348 | 0 | more_ = false; |
2349 | 0 | return; |
2350 | 0 | } |
2351 | 176k | } |
2352 | 554k | break; |
2353 | 554k | } |
2354 | 554k | case numeric_check_state::fraction: |
2355 | 91.6k | case numeric_check_state::exp: |
2356 | 91.6k | { |
2357 | 91.6k | if (lossless_number_) |
2358 | 0 | { |
2359 | 0 | visitor.string_value(buffer_,semantic_tag::bigdec, *this, ec); |
2360 | 0 | more_ = !cursor_mode_; |
2361 | 0 | } |
2362 | 91.6k | else |
2363 | 91.6k | { |
2364 | 91.6k | double d{0}; |
2365 | 91.6k | auto result = jsoncons::utility::decstr_to_double(buffer.c_str(), buffer.length(), d); |
2366 | 91.6k | if (result.ec == std::errc::result_out_of_range) |
2367 | 651 | { |
2368 | 651 | d = buffer.front() == '-' ? -HUGE_VAL : HUGE_VAL; |
2369 | 651 | } |
2370 | 91.0k | else if (result.ec == std::errc::invalid_argument) |
2371 | 6 | { |
2372 | 6 | ec = csv_errc::invalid_number; |
2373 | 6 | more_ = false; |
2374 | 6 | return; |
2375 | 6 | } |
2376 | 91.6k | visitor.double_value(d, semantic_tag::none, *this, ec); |
2377 | 91.6k | more_ = !cursor_mode_; |
2378 | 91.6k | } |
2379 | 91.6k | break; |
2380 | 91.6k | } |
2381 | 14.4M | default: |
2382 | 14.4M | { |
2383 | 14.4M | visitor.string_value(buffer_, semantic_tag::none, *this, ec); |
2384 | 14.4M | more_ = !cursor_mode_; |
2385 | 14.4M | break; |
2386 | 91.6k | } |
2387 | 15.2M | } |
2388 | 15.2M | } |
2389 | | |
2390 | | void push_state(csv_parse_state state) |
2391 | 4.34M | { |
2392 | 4.34M | state_stack_.push_back(state); |
2393 | 4.34M | } |
2394 | | |
2395 | | csv_parse_state pop_state() |
2396 | 4.34M | { |
2397 | 4.34M | JSONCONS_ASSERT(!state_stack_.empty()) |
2398 | 4.34M | csv_parse_state state = state_stack_.back(); |
2399 | 4.34M | state_stack_.pop_back(); |
2400 | 4.34M | return state; |
2401 | 4.34M | } |
2402 | | }; |
2403 | | |
2404 | | using csv_parser = basic_csv_parser<char>; |
2405 | | using wcsv_parser = basic_csv_parser<wchar_t>; |
2406 | | |
2407 | | }} |
2408 | | |
2409 | | #endif |
2410 | | |