/root/doris/be/src/util/jsonb_writer.h
Line | Count | Source |
1 | | /* |
2 | | * Copyright (c) 2014, Facebook, Inc. |
3 | | * All rights reserved. |
4 | | * |
5 | | * This source code is licensed under the BSD-style license found in the |
6 | | * LICENSE file in the root directory of this source tree. An additional grant |
7 | | * of patent rights can be found in the PATENTS file in the same directory. |
8 | | * |
9 | | */ |
10 | | |
11 | | /* |
12 | | * This file defines JsonbWriterT (template) and JsonbWriter. |
13 | | * |
14 | | * JsonbWriterT is a template class which implements an JSONB serializer. |
15 | | * Users call various write functions of JsonbWriterT object to write values |
16 | | * directly to JSONB packed bytes. All write functions of value or key return |
17 | | * the number of bytes written to JSONB, or 0 if there is an error. To write an |
18 | | * object, an array, or a string, you must call writeStart[..] before writing |
19 | | * values or key, and call writeEnd[..] after finishing at the end. |
20 | | * |
21 | | * By default, an JsonbWriterT object creates an output stream buffer. |
22 | | * Alternatively, you can also pass any output stream object to a writer, as |
23 | | * long as the stream object implements some basic functions of std::ostream |
24 | | * (such as JsonbOutStream, see JsonbStream.h). |
25 | | * |
26 | | * JsonbWriter specializes JsonbWriterT with JsonbOutStream type (see |
27 | | * JsonbStream.h). So unless you want to provide own a different output stream |
28 | | * type, use JsonbParser object. |
29 | | * |
30 | | * @author Tian Xia <tianx@fb.com> |
31 | | * this file is copied from |
32 | | * https://github.com/facebook/mysql-5.6/blob/fb-mysql-5.6.35/fbson/FbsonWriter.h |
33 | | * and modified by Doris |
34 | | */ |
35 | | |
36 | | #ifndef JSONB_JSONBWRITER_H |
37 | | #define JSONB_JSONBWRITER_H |
38 | | |
39 | | #include <glog/logging.h> |
40 | | |
41 | | #include <cstdint> |
42 | | #include <limits> |
43 | | #include <stack> |
44 | | #include <string> |
45 | | |
46 | | #include "common/status.h" |
47 | | #include "jsonb_document.h" |
48 | | #include "jsonb_stream.h" |
49 | | #include "vec/core/types.h" |
50 | | |
51 | | namespace doris { |
52 | | |
53 | | using int128_t = __int128; |
54 | | |
55 | | template <class OS_TYPE> |
56 | | class JsonbWriterT { |
57 | | /// TODO: maybe we should not use a template class here |
58 | | static_assert(std::is_same_v<OS_TYPE, JsonbOutStream>); |
59 | | |
60 | | public: |
61 | 175k | JsonbWriterT() : alloc_(true), hasHdr_(false), kvState_(WS_Value), str_pos_(0) { |
62 | 175k | os_ = new OS_TYPE(); |
63 | 175k | } |
64 | | |
65 | | explicit JsonbWriterT(OS_TYPE& os) |
66 | | : os_(&os), alloc_(false), hasHdr_(false), kvState_(WS_Value), str_pos_(0) {} |
67 | | |
68 | 175k | ~JsonbWriterT() { |
69 | 175k | if (alloc_) { |
70 | 175k | delete os_; |
71 | 175k | } |
72 | 175k | } |
73 | | |
74 | | JsonbWriterT<OS_TYPE>& operator=(JsonbWriterT<OS_TYPE>&& other) { |
75 | | if (this != &other) { |
76 | | if (alloc_) { |
77 | | delete os_; |
78 | | } |
79 | | os_ = other.os_; |
80 | | other.os_ = nullptr; |
81 | | alloc_ = other.alloc_; |
82 | | other.alloc_ = false; |
83 | | hasHdr_ = other.hasHdr_; |
84 | | kvState_ = other.kvState_; |
85 | | str_pos_ = other.str_pos_; |
86 | | first_ = other.first_; |
87 | | stack_ = std::move(other.stack_); |
88 | | } |
89 | | return *this; |
90 | | } |
91 | | |
92 | 58.0k | void reset() { |
93 | 58.0k | os_->clear(); |
94 | 58.0k | os_->seekp(0); |
95 | 58.0k | hasHdr_ = false; |
96 | 58.0k | kvState_ = WS_Value; |
97 | 58.0k | first_ = true; |
98 | 58.0k | for (; !stack_.empty(); stack_.pop()) { |
99 | 0 | ; |
100 | 0 | } |
101 | 58.0k | } |
102 | | |
103 | 17 | bool writeKey(const char* key) { return writeKey(key, strlen(key)); } |
104 | | |
105 | | // write a key string (or key id if an external dict is provided) |
106 | 9.26k | bool writeKey(const char* key, uint8_t len) { |
107 | 9.26k | if (!stack_.empty() && verifyKeyState()) { |
108 | 9.26k | os_->put(len); |
109 | 9.26k | if (len == 0) { |
110 | | // NOTE: we use sMaxKeyId to represent an empty key |
111 | 431 | JsonbKeyValue::keyid_type idx = JsonbKeyValue::sMaxKeyId; |
112 | 431 | os_->write((char*)&idx, sizeof(JsonbKeyValue::keyid_type)); |
113 | 8.83k | } else { |
114 | 8.83k | os_->write(key, len); |
115 | 8.83k | } |
116 | 9.26k | kvState_ = WS_Key; |
117 | 9.26k | return true; |
118 | 9.26k | } |
119 | | |
120 | 1 | return false; |
121 | 9.26k | } |
122 | | |
123 | 25.9k | bool writeValue(const JsonbValue* value) { |
124 | 25.9k | if (!value) { |
125 | 0 | return writeNull(); |
126 | 0 | } |
127 | | |
128 | 25.9k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
129 | 25.9k | if (!writeFirstHeader()) { |
130 | 0 | return false; |
131 | 0 | } |
132 | 25.9k | os_->write((char*)value, value->numPackedBytes()); |
133 | 25.9k | kvState_ = WS_Value; |
134 | 25.9k | return true; |
135 | 25.9k | } |
136 | 0 | return false; |
137 | 25.9k | } |
138 | | |
139 | 44 | bool writeValueSimple(const JsonbValue* value) { |
140 | 44 | DCHECK(value) << "value should not be nullptr"; |
141 | 44 | DCHECK(first_) << "only called at the beginning"; |
142 | 44 | DCHECK(stack_.empty()) << "only called at the beginning"; |
143 | 44 | DCHECK(!hasHdr_) << "only called at the beginning"; |
144 | 44 | first_ = false; |
145 | 44 | writeHeader(); |
146 | 44 | os_->write((char*)value, value->numPackedBytes()); |
147 | 44 | kvState_ = WS_Value; |
148 | 44 | return true; |
149 | 44 | } |
150 | | |
151 | | // write a key id |
152 | 18.2k | bool writeKey(JsonbKeyValue::keyid_type idx) { |
153 | 18.2k | if (!stack_.empty() && verifyKeyState()) { |
154 | 16.8k | os_->put(0); |
155 | 16.8k | os_->write((char*)&idx, sizeof(JsonbKeyValue::keyid_type)); |
156 | 16.8k | kvState_ = WS_Key; |
157 | 16.8k | return true; |
158 | 16.8k | } |
159 | | |
160 | 1.42k | return false; |
161 | 18.2k | } |
162 | | |
163 | 1.30M | bool writeFirstHeader() { |
164 | 1.30M | if (first_ && stack_.empty()) { |
165 | 106k | first_ = false; |
166 | | // if this is a new JSONB, write the header |
167 | 106k | if (!hasHdr_) { |
168 | 106k | writeHeader(); |
169 | 106k | return true; |
170 | 106k | } else { |
171 | 0 | return false; |
172 | 0 | } |
173 | 1.19M | } else { |
174 | 1.19M | return true; |
175 | 1.19M | } |
176 | 1.30M | } |
177 | | |
178 | 26.0k | bool writeNull() { |
179 | 26.0k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
180 | 26.0k | if (!writeFirstHeader()) { |
181 | 0 | return false; |
182 | 0 | } |
183 | 26.0k | os_->put((JsonbTypeUnder)JsonbType::T_Null); |
184 | 26.0k | kvState_ = WS_Value; |
185 | 26.0k | return true; |
186 | 26.0k | } |
187 | | |
188 | 0 | return false; |
189 | 26.0k | } |
190 | | |
191 | 20.7k | bool writeBool(bool b) { |
192 | 20.7k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
193 | 20.7k | if (!writeFirstHeader()) { |
194 | 0 | return false; |
195 | 0 | } |
196 | 20.7k | if (b) { |
197 | 10.3k | os_->put((JsonbTypeUnder)JsonbType::T_True); |
198 | 10.4k | } else { |
199 | 10.4k | os_->put((JsonbTypeUnder)JsonbType::T_False); |
200 | 10.4k | } |
201 | | |
202 | 20.7k | kvState_ = WS_Value; |
203 | 20.7k | return true; |
204 | 20.7k | } |
205 | | |
206 | 0 | return false; |
207 | 20.7k | } |
208 | | |
209 | | // This function is a helper. It will make use of smallest space to |
210 | | // write an int |
211 | 2 | bool writeInt(int64_t val) { |
212 | 2 | if (val >= std::numeric_limits<int8_t>::min() && |
213 | 2 | val <= std::numeric_limits<int8_t>::max()) { |
214 | 1 | return writeInt8((int8_t)val); |
215 | 1 | } else if (val >= std::numeric_limits<int16_t>::min() && |
216 | 1 | val <= std::numeric_limits<int16_t>::max()) { |
217 | 1 | return writeInt16((int16_t)val); |
218 | 1 | } else if (val >= std::numeric_limits<int32_t>::min() && |
219 | 0 | val <= std::numeric_limits<int32_t>::max()) { |
220 | 0 | return writeInt32((int32_t)val); |
221 | 0 | } else { |
222 | 0 | return writeInt64(val); |
223 | 0 | } |
224 | 2 | } |
225 | | |
226 | 60.8k | bool writeInt8(int8_t v) { |
227 | 60.8k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
228 | 60.8k | if (!writeFirstHeader()) { |
229 | 0 | return false; |
230 | 0 | } |
231 | 60.8k | os_->put((JsonbTypeUnder)JsonbType::T_Int8); |
232 | 60.8k | os_->put(v); |
233 | 60.8k | kvState_ = WS_Value; |
234 | 60.8k | return true; |
235 | 60.8k | } |
236 | | |
237 | 0 | return false; |
238 | 60.8k | } |
239 | | |
240 | 188 | bool writeInt16(int16_t v) { |
241 | 188 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
242 | 188 | if (!writeFirstHeader()) { |
243 | 0 | return false; |
244 | 0 | } |
245 | 188 | os_->put((JsonbTypeUnder)JsonbType::T_Int16); |
246 | 188 | os_->write((char*)&v, sizeof(int16_t)); |
247 | 188 | kvState_ = WS_Value; |
248 | 188 | return true; |
249 | 188 | } |
250 | | |
251 | 0 | return false; |
252 | 188 | } |
253 | | |
254 | 3.49k | bool writeInt32(int32_t v) { |
255 | 3.49k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
256 | 3.49k | if (!writeFirstHeader()) { |
257 | 0 | return false; |
258 | 0 | } |
259 | 3.49k | os_->put((JsonbTypeUnder)JsonbType::T_Int32); |
260 | 3.49k | os_->write((char*)&v, sizeof(int32_t)); |
261 | 3.49k | kvState_ = WS_Value; |
262 | 3.49k | return true; |
263 | 3.49k | } |
264 | | |
265 | 0 | return false; |
266 | 3.49k | } |
267 | | |
268 | 277k | bool writeInt64(int64_t v) { |
269 | 277k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
270 | 277k | if (!writeFirstHeader()) { |
271 | 0 | return false; |
272 | 0 | } |
273 | 277k | os_->put((JsonbTypeUnder)JsonbType::T_Int64); |
274 | 277k | os_->write((char*)&v, sizeof(int64_t)); |
275 | 277k | kvState_ = WS_Value; |
276 | 277k | return true; |
277 | 277k | } |
278 | | |
279 | 0 | return false; |
280 | 277k | } |
281 | | |
282 | 45.2k | bool writeInt128(int128_t v) { |
283 | 45.2k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
284 | 45.2k | if (!writeFirstHeader()) { |
285 | 0 | return false; |
286 | 0 | } |
287 | 45.2k | os_->put((JsonbTypeUnder)JsonbType::T_Int128); |
288 | 45.2k | os_->write((char*)&v, sizeof(int128_t)); |
289 | 45.2k | kvState_ = WS_Value; |
290 | 45.2k | return true; |
291 | 45.2k | } |
292 | | |
293 | 0 | return false; |
294 | 45.2k | } |
295 | | |
296 | 280k | bool writeDouble(double v) { |
297 | 280k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
298 | 280k | if (!writeFirstHeader()) { |
299 | 0 | return false; |
300 | 0 | } |
301 | 280k | os_->put((JsonbTypeUnder)JsonbType::T_Double); |
302 | 280k | os_->write((char*)&v, sizeof(double)); |
303 | 280k | kvState_ = WS_Value; |
304 | 280k | return true; |
305 | 280k | } |
306 | | |
307 | 0 | return false; |
308 | 280k | } |
309 | | |
310 | 23 | bool writeFloat(float v) { |
311 | 23 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
312 | 23 | if (!writeFirstHeader()) { |
313 | 0 | return false; |
314 | 0 | } |
315 | 23 | os_->put((JsonbTypeUnder)JsonbType::T_Float); |
316 | 23 | os_->write((char*)&v, sizeof(float)); |
317 | 23 | kvState_ = WS_Value; |
318 | 23 | return true; |
319 | 23 | } |
320 | | |
321 | 0 | return false; |
322 | 23 | } |
323 | | |
324 | | template <JsonbDecimalType T> |
325 | 25 | bool writeDecimal(const T& v, const uint32_t precision, const uint32_t scale) { |
326 | 25 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
327 | 25 | if (!writeFirstHeader()) { |
328 | 0 | return false; |
329 | 0 | } |
330 | | |
331 | 25 | if constexpr (std::same_as<T, vectorized::Decimal256>) { |
332 | 6 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal256); |
333 | 7 | } else if constexpr (std::same_as<T, vectorized::Decimal128V3>) { |
334 | 7 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal128); |
335 | 7 | } else if constexpr (std::same_as<T, vectorized::Decimal64>) { |
336 | 6 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal64); |
337 | 6 | } else { |
338 | 6 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal32); |
339 | 6 | } |
340 | | |
341 | 25 | os_->write(reinterpret_cast<const char*>(&precision), sizeof(uint32_t)); |
342 | 25 | os_->write(reinterpret_cast<const char*>(&scale), sizeof(uint32_t)); |
343 | 25 | os_->write((char*)(&(v.value)), sizeof(v.value)); |
344 | 25 | kvState_ = WS_Value; |
345 | 25 | return true; |
346 | 25 | } |
347 | | |
348 | 0 | return false; |
349 | 25 | } _ZN5doris12JsonbWriterTINS_14JsonbOutStreamEE12writeDecimalITk16JsonbDecimalTypeNS_10vectorized7DecimalIiEEEEbRKT_jj Line | Count | Source | 325 | 6 | bool writeDecimal(const T& v, const uint32_t precision, const uint32_t scale) { | 326 | 6 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { | 327 | 6 | if (!writeFirstHeader()) { | 328 | 0 | return false; | 329 | 0 | } | 330 | | | 331 | | if constexpr (std::same_as<T, vectorized::Decimal256>) { | 332 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal256); | 333 | | } else if constexpr (std::same_as<T, vectorized::Decimal128V3>) { | 334 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal128); | 335 | | } else if constexpr (std::same_as<T, vectorized::Decimal64>) { | 336 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal64); | 337 | 6 | } else { | 338 | 6 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal32); | 339 | 6 | } | 340 | | | 341 | 6 | os_->write(reinterpret_cast<const char*>(&precision), sizeof(uint32_t)); | 342 | 6 | os_->write(reinterpret_cast<const char*>(&scale), sizeof(uint32_t)); | 343 | 6 | os_->write((char*)(&(v.value)), sizeof(v.value)); | 344 | 6 | kvState_ = WS_Value; | 345 | 6 | return true; | 346 | 6 | } | 347 | | | 348 | 0 | return false; | 349 | 6 | } |
_ZN5doris12JsonbWriterTINS_14JsonbOutStreamEE12writeDecimalITk16JsonbDecimalTypeNS_10vectorized7DecimalIlEEEEbRKT_jj Line | Count | Source | 325 | 6 | bool writeDecimal(const T& v, const uint32_t precision, const uint32_t scale) { | 326 | 6 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { | 327 | 6 | if (!writeFirstHeader()) { | 328 | 0 | return false; | 329 | 0 | } | 330 | | | 331 | | if constexpr (std::same_as<T, vectorized::Decimal256>) { | 332 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal256); | 333 | | } else if constexpr (std::same_as<T, vectorized::Decimal128V3>) { | 334 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal128); | 335 | 6 | } else if constexpr (std::same_as<T, vectorized::Decimal64>) { | 336 | 6 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal64); | 337 | | } else { | 338 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal32); | 339 | | } | 340 | | | 341 | 6 | os_->write(reinterpret_cast<const char*>(&precision), sizeof(uint32_t)); | 342 | 6 | os_->write(reinterpret_cast<const char*>(&scale), sizeof(uint32_t)); | 343 | 6 | os_->write((char*)(&(v.value)), sizeof(v.value)); | 344 | 6 | kvState_ = WS_Value; | 345 | 6 | return true; | 346 | 6 | } | 347 | | | 348 | 0 | return false; | 349 | 6 | } |
_ZN5doris12JsonbWriterTINS_14JsonbOutStreamEE12writeDecimalITk16JsonbDecimalTypeNS_10vectorized12Decimal128V3EEEbRKT_jj Line | Count | Source | 325 | 7 | bool writeDecimal(const T& v, const uint32_t precision, const uint32_t scale) { | 326 | 7 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { | 327 | 7 | if (!writeFirstHeader()) { | 328 | 0 | return false; | 329 | 0 | } | 330 | | | 331 | | if constexpr (std::same_as<T, vectorized::Decimal256>) { | 332 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal256); | 333 | 7 | } else if constexpr (std::same_as<T, vectorized::Decimal128V3>) { | 334 | 7 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal128); | 335 | | } else if constexpr (std::same_as<T, vectorized::Decimal64>) { | 336 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal64); | 337 | | } else { | 338 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal32); | 339 | | } | 340 | | | 341 | 7 | os_->write(reinterpret_cast<const char*>(&precision), sizeof(uint32_t)); | 342 | 7 | os_->write(reinterpret_cast<const char*>(&scale), sizeof(uint32_t)); | 343 | 7 | os_->write((char*)(&(v.value)), sizeof(v.value)); | 344 | 7 | kvState_ = WS_Value; | 345 | 7 | return true; | 346 | 7 | } | 347 | | | 348 | 0 | return false; | 349 | 7 | } |
_ZN5doris12JsonbWriterTINS_14JsonbOutStreamEE12writeDecimalITk16JsonbDecimalTypeNS_10vectorized7DecimalIN4wide7integerILm256EiEEEEEEbRKT_jj Line | Count | Source | 325 | 6 | bool writeDecimal(const T& v, const uint32_t precision, const uint32_t scale) { | 326 | 6 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { | 327 | 6 | if (!writeFirstHeader()) { | 328 | 0 | return false; | 329 | 0 | } | 330 | | | 331 | 6 | if constexpr (std::same_as<T, vectorized::Decimal256>) { | 332 | 6 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal256); | 333 | | } else if constexpr (std::same_as<T, vectorized::Decimal128V3>) { | 334 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal128); | 335 | | } else if constexpr (std::same_as<T, vectorized::Decimal64>) { | 336 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal64); | 337 | | } else { | 338 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal32); | 339 | | } | 340 | | | 341 | 6 | os_->write(reinterpret_cast<const char*>(&precision), sizeof(uint32_t)); | 342 | 6 | os_->write(reinterpret_cast<const char*>(&scale), sizeof(uint32_t)); | 343 | 6 | os_->write((char*)(&(v.value)), sizeof(v.value)); | 344 | 6 | kvState_ = WS_Value; | 345 | 6 | return true; | 346 | 6 | } | 347 | | | 348 | 0 | return false; | 349 | 6 | } |
|
350 | | |
351 | | // must call writeStartString before writing a string val |
352 | 559k | bool writeStartString() { |
353 | 559k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
354 | 559k | if (!writeFirstHeader()) return 0; |
355 | 559k | os_->put((JsonbTypeUnder)JsonbType::T_String); |
356 | 559k | str_pos_ = os_->tellp(); |
357 | | |
358 | | // fill the size bytes with 0 for now |
359 | 559k | uint32_t size = 0; |
360 | 559k | os_->write((char*)&size, sizeof(uint32_t)); |
361 | | |
362 | 559k | kvState_ = WS_String; |
363 | 559k | return true; |
364 | 559k | } |
365 | | |
366 | 0 | return false; |
367 | 559k | } |
368 | | |
369 | | // finish writing a string val |
370 | 559k | bool writeEndString() { |
371 | 559k | if (kvState_ == WS_String) { |
372 | 559k | std::streampos cur_pos = os_->tellp(); |
373 | 559k | int32_t size = (int32_t)(cur_pos - str_pos_ - sizeof(uint32_t)); |
374 | 559k | assert(size >= 0); |
375 | | |
376 | 559k | os_->seekp(str_pos_); |
377 | 559k | os_->write((char*)&size, sizeof(uint32_t)); |
378 | 559k | os_->seekp(cur_pos); |
379 | | |
380 | 559k | kvState_ = WS_Value; |
381 | 559k | return true; |
382 | 559k | } |
383 | | |
384 | 0 | return false; |
385 | 559k | } |
386 | | |
387 | | // TODO: here changed length to uint64_t, as some api also need changed, But the thirdparty api is uint_32t |
388 | | // need consider a better way to handle case. |
389 | 548k | bool writeString(const char* str, uint64_t len) { |
390 | 548k | if (kvState_ == WS_String) { |
391 | 548k | os_->write(str, len); |
392 | 548k | return true; |
393 | 548k | } |
394 | | |
395 | 5 | return false; |
396 | 548k | } |
397 | | |
398 | 292k | bool writeString(const std::string& str) { return writeString(str.c_str(), str.size()); } |
399 | | bool writeString(char ch) { |
400 | | if (kvState_ == WS_String) { |
401 | | os_->put(ch); |
402 | | return true; |
403 | | } |
404 | | |
405 | | return false; |
406 | | } |
407 | | |
408 | | // must call writeStartBinary before writing a binary val |
409 | 4.36k | bool writeStartBinary() { |
410 | 4.36k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
411 | 4.36k | if (!writeFirstHeader()) return 0; |
412 | 4.36k | os_->put((JsonbTypeUnder)JsonbType::T_Binary); |
413 | 4.36k | str_pos_ = os_->tellp(); |
414 | | |
415 | | // fill the size bytes with 0 for now |
416 | 4.36k | uint32_t size = 0; |
417 | 4.36k | os_->write((char*)&size, sizeof(uint32_t)); |
418 | | |
419 | 4.36k | kvState_ = WS_Binary; |
420 | 4.36k | return true; |
421 | 4.36k | } |
422 | | |
423 | 0 | return false; |
424 | 4.36k | } |
425 | | |
426 | | // finish writing a binary val |
427 | 4.36k | bool writeEndBinary() { |
428 | 4.36k | if (kvState_ == WS_Binary) { |
429 | 4.36k | std::streampos cur_pos = os_->tellp(); |
430 | 4.36k | int32_t size = (int32_t)(cur_pos - str_pos_ - sizeof(uint32_t)); |
431 | 4.36k | assert(size >= 0); |
432 | | |
433 | 4.36k | os_->seekp(str_pos_); |
434 | 4.36k | os_->write((char*)&size, sizeof(uint32_t)); |
435 | 4.36k | os_->seekp(cur_pos); |
436 | | |
437 | 4.36k | kvState_ = WS_Value; |
438 | 4.36k | return true; |
439 | 4.36k | } |
440 | | |
441 | 0 | return false; |
442 | 4.36k | } |
443 | | |
444 | 4.36k | bool writeBinary(const char* bin, uint64_t len) { |
445 | 4.36k | if (kvState_ == WS_Binary) { |
446 | 4.36k | os_->write(bin, len); |
447 | 4.36k | return true; |
448 | 4.36k | } |
449 | | |
450 | 0 | return false; |
451 | 4.36k | } |
452 | | |
453 | | // must call writeStartObject before writing an object val |
454 | 5.20k | bool writeStartObject() { |
455 | 5.20k | if (stack_.empty() || verifyValueState()) { |
456 | 5.20k | if (stack_.empty()) { |
457 | | // if this is a new JSONB, write the header |
458 | 1.48k | if (!hasHdr_) { |
459 | 1.48k | writeHeader(); |
460 | 1.48k | } else |
461 | 0 | return false; |
462 | 1.48k | } |
463 | | |
464 | | // check if the object exceeds the maximum nesting level |
465 | 5.20k | if (stack_.size() >= MaxNestingLevel) return false; |
466 | | |
467 | 5.20k | os_->put((JsonbTypeUnder)JsonbType::T_Object); |
468 | | // save the size position |
469 | 5.20k | stack_.push(WriteInfo({WS_Object, os_->tellp()})); |
470 | | |
471 | | // fill the size bytes with 0 for now |
472 | 5.20k | uint32_t size = 0; |
473 | 5.20k | os_->write((char*)&size, sizeof(uint32_t)); |
474 | | |
475 | 5.20k | kvState_ = WS_Value; |
476 | 5.20k | return true; |
477 | 5.20k | } |
478 | | |
479 | 0 | return false; |
480 | 5.20k | } |
481 | | |
482 | | // finish writing an object val |
483 | 4.63k | bool writeEndObject() { |
484 | 4.63k | if (!stack_.empty() && stack_.top().state == WS_Object && kvState_ == WS_Value) { |
485 | 4.63k | WriteInfo& ci = stack_.top(); |
486 | 4.63k | std::streampos cur_pos = os_->tellp(); |
487 | 4.63k | auto size = (int32_t)(cur_pos - ci.sz_pos - sizeof(uint32_t)); |
488 | 4.63k | assert(size >= 0); |
489 | | |
490 | 4.63k | os_->seekp(ci.sz_pos); |
491 | 4.63k | os_->write((char*)&size, sizeof(uint32_t)); |
492 | 4.63k | os_->seekp(cur_pos); |
493 | 4.63k | stack_.pop(); |
494 | | |
495 | 4.63k | return true; |
496 | 4.63k | } |
497 | | |
498 | 1 | return false; |
499 | 4.63k | } |
500 | | |
501 | | // must call writeStartArray before writing an array val |
502 | 173k | bool writeStartArray() { |
503 | 173k | if (stack_.empty() || verifyValueState()) { |
504 | 173k | if (stack_.empty()) { |
505 | | // if this is a new JSONB, write the header |
506 | 73.1k | if (!hasHdr_) { |
507 | 73.1k | writeHeader(); |
508 | 73.1k | } else { |
509 | 0 | return false; |
510 | 0 | } |
511 | 73.1k | } |
512 | | |
513 | | // check if the array exceeds the maximum nesting level |
514 | 173k | if (stack_.size() >= MaxNestingLevel) { |
515 | 0 | return false; |
516 | 0 | } |
517 | | |
518 | 173k | os_->put((JsonbTypeUnder)JsonbType::T_Array); |
519 | | // save the size position |
520 | 173k | stack_.push(WriteInfo({WS_Array, os_->tellp()})); |
521 | | |
522 | | // fill the size bytes with 0 for now |
523 | 173k | uint32_t size = 0; |
524 | 173k | os_->write((char*)&size, sizeof(uint32_t)); |
525 | | |
526 | 173k | kvState_ = WS_Value; |
527 | 173k | return true; |
528 | 173k | } |
529 | | |
530 | 0 | return false; |
531 | 173k | } |
532 | | |
533 | | // finish writing an array val |
534 | 172k | bool writeEndArray() { |
535 | 172k | if (!stack_.empty() && stack_.top().state == WS_Array && kvState_ == WS_Value) { |
536 | 172k | WriteInfo& ci = stack_.top(); |
537 | 172k | std::streampos cur_pos = os_->tellp(); |
538 | 172k | auto size = (int32_t)(cur_pos - ci.sz_pos - sizeof(uint32_t)); |
539 | 172k | assert(size >= 0); |
540 | | |
541 | 172k | os_->seekp(ci.sz_pos); |
542 | 172k | os_->write((char*)&size, sizeof(uint32_t)); |
543 | 172k | os_->seekp(cur_pos); |
544 | 172k | stack_.pop(); |
545 | | |
546 | 172k | return true; |
547 | 172k | } |
548 | | |
549 | 0 | return false; |
550 | 172k | } |
551 | | |
552 | 360k | OS_TYPE* getOutput() { return os_; } |
553 | 1 | JsonbDocument* getDocument() { |
554 | 1 | JsonbDocument* doc = nullptr; |
555 | 1 | THROW_IF_ERROR(JsonbDocument::checkAndCreateDocument(getOutput()->getBuffer(), |
556 | 1 | getOutput()->getSize(), &doc)); |
557 | 1 | return doc; |
558 | 1 | } |
559 | | |
560 | 12 | JsonbValue* getValue() { |
561 | 12 | return JsonbDocument::createValue(getOutput()->getBuffer(), getOutput()->getSize()); |
562 | 12 | } |
563 | | |
564 | | bool writeEnd() { |
565 | | while (!stack_.empty()) { |
566 | | bool ok = false; |
567 | | switch (stack_.top().state) { |
568 | | case WS_Array: |
569 | | ok = writeEndArray(); |
570 | | break; |
571 | | case WS_Object: |
572 | | ok = writeEndObject(); |
573 | | break; |
574 | | case WS_String: |
575 | | ok = writeEndString(); |
576 | | break; |
577 | | case WS_Binary: |
578 | | ok = writeEndBinary(); |
579 | | break; |
580 | | default: |
581 | | ok = false; |
582 | | break; |
583 | | } |
584 | | if (!ok) { |
585 | | return false; |
586 | | } |
587 | | } |
588 | | return true; |
589 | | } |
590 | | |
591 | | private: |
592 | | // verify we are in the right state before writing a value |
593 | 1.30M | bool verifyValueState() { |
594 | 1.30M | assert(!stack_.empty()); |
595 | | // The document can only be an Object or an Array which follows |
596 | | // the standard. |
597 | 1.30M | return (stack_.top().state == WS_Object && kvState_ == WS_Key) || |
598 | 1.30M | (stack_.top().state == WS_Array && kvState_ == WS_Value); |
599 | 1.30M | } |
600 | | |
601 | | // verify we are in the right state before writing a key |
602 | 27.4k | bool verifyKeyState() { |
603 | 27.4k | assert(!stack_.empty()); |
604 | 27.4k | return stack_.top().state == WS_Object && kvState_ == WS_Value; |
605 | 27.4k | } |
606 | | |
607 | 180k | void writeHeader() { |
608 | 180k | os_->put(JSONB_VER); |
609 | 180k | hasHdr_ = true; |
610 | 180k | } |
611 | | |
612 | | enum WriteState { |
613 | | WS_NONE, |
614 | | WS_Array, |
615 | | WS_Object, |
616 | | WS_Key, |
617 | | WS_Value, |
618 | | WS_String, |
619 | | WS_Binary, |
620 | | }; |
621 | | |
622 | | struct WriteInfo { |
623 | | WriteState state; |
624 | | std::streampos sz_pos; |
625 | | }; |
626 | | |
627 | | OS_TYPE* os_ = nullptr; |
628 | | bool alloc_; |
629 | | bool hasHdr_; |
630 | | WriteState kvState_; // key or value state |
631 | | std::streampos str_pos_; |
632 | | std::stack<WriteInfo> stack_; |
633 | | bool first_ = true; |
634 | | }; |
635 | | |
636 | | using JsonbWriter = JsonbWriterT<JsonbOutStream>; |
637 | | |
638 | | } // namespace doris |
639 | | |
640 | | #endif // JSONB_JSONBWRITER_H |