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 "core/types.h" |
48 | | #include "util/jsonb_document.h" |
49 | | #include "util/jsonb_stream.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 | 3.37M | JsonbWriterT() : alloc_(true), hasHdr_(false), kvState_(WS_Value), str_pos_(0) { |
62 | 3.37M | os_ = new OS_TYPE(); |
63 | 3.37M | } |
64 | | |
65 | | explicit JsonbWriterT(OS_TYPE& os) |
66 | | : os_(&os), alloc_(false), hasHdr_(false), kvState_(WS_Value), str_pos_(0) {} |
67 | | |
68 | 3.36M | ~JsonbWriterT() { |
69 | 3.37M | if (alloc_) { |
70 | 3.37M | delete os_; |
71 | 3.37M | } |
72 | 3.36M | } |
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 | 322k | void reset() { |
93 | 322k | os_->clear(); |
94 | 322k | os_->seekp(0); |
95 | 322k | hasHdr_ = false; |
96 | 322k | kvState_ = WS_Value; |
97 | 322k | first_ = true; |
98 | 322k | for (; !stack_.empty(); stack_.pop()) { |
99 | 9 | ; |
100 | 9 | } |
101 | 322k | } |
102 | | |
103 | | // write a key string (or key id if an external dict is provided) |
104 | 765k | bool writeKey(const char* key, uint8_t len) { |
105 | 765k | if (!stack_.empty() && verifyKeyState()) { |
106 | 765k | os_->put(len); |
107 | 765k | if (len == 0) { |
108 | | // NOTE: we use sMaxKeyId to represent an empty key |
109 | 456 | JsonbKeyValue::keyid_type idx = JsonbKeyValue::sMaxKeyId; |
110 | 456 | os_->write((char*)&idx, sizeof(JsonbKeyValue::keyid_type)); |
111 | 765k | } else { |
112 | 765k | os_->write(key, len); |
113 | 765k | } |
114 | 765k | kvState_ = WS_Key; |
115 | 765k | return true; |
116 | 765k | } |
117 | | |
118 | 18.4E | return false; |
119 | 765k | } |
120 | | |
121 | 27.3k | bool writeValue(const JsonbValue* value) { |
122 | 27.3k | if (!value) { |
123 | 11 | return writeNull(); |
124 | 11 | } |
125 | | |
126 | 27.3k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
127 | 27.3k | if (!writeFirstHeader()) { |
128 | 0 | return false; |
129 | 0 | } |
130 | 27.3k | os_->write((char*)value, value->numPackedBytes()); |
131 | 27.3k | kvState_ = WS_Value; |
132 | 27.3k | return true; |
133 | 27.3k | } |
134 | 0 | return false; |
135 | 27.3k | } |
136 | | |
137 | 117k | bool writeValueSimple(const JsonbValue* value) { |
138 | 18.4E | DCHECK(value) << "value should not be nullptr"; |
139 | 18.4E | DCHECK(first_) << "only called at the beginning"; |
140 | 18.4E | DCHECK(stack_.empty()) << "only called at the beginning"; |
141 | 18.4E | DCHECK(!hasHdr_) << "only called at the beginning"; |
142 | 117k | first_ = false; |
143 | 117k | writeHeader(); |
144 | 117k | os_->write((char*)value, value->numPackedBytes()); |
145 | 117k | kvState_ = WS_Value; |
146 | 117k | return true; |
147 | 117k | } |
148 | | |
149 | | // write a key id |
150 | 44.0M | bool writeKey(JsonbKeyValue::keyid_type idx) { |
151 | 44.3M | if (!stack_.empty() && verifyKeyState()) { |
152 | 23.9M | os_->put(0); |
153 | 23.9M | os_->write((char*)&idx, sizeof(JsonbKeyValue::keyid_type)); |
154 | 23.9M | kvState_ = WS_Key; |
155 | 23.9M | return true; |
156 | 23.9M | } |
157 | | |
158 | 20.1M | return false; |
159 | 44.0M | } |
160 | | |
161 | 27.2M | bool writeFirstHeader() { |
162 | 27.2M | if (first_ && stack_.empty()) { |
163 | 1.18M | first_ = false; |
164 | | // if this is a new JSONB, write the header |
165 | 1.18M | if (!hasHdr_) { |
166 | 1.18M | writeHeader(); |
167 | 1.18M | return true; |
168 | 1.18M | } else { |
169 | 0 | return false; |
170 | 0 | } |
171 | 26.0M | } else { |
172 | 26.0M | return true; |
173 | 26.0M | } |
174 | 27.2M | } |
175 | | |
176 | 31.1k | bool writeNull() { |
177 | 31.1k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
178 | 31.1k | if (!writeFirstHeader()) { |
179 | 0 | return false; |
180 | 0 | } |
181 | 31.1k | os_->put((JsonbTypeUnder)JsonbType::T_Null); |
182 | 31.1k | kvState_ = WS_Value; |
183 | 31.1k | return true; |
184 | 31.1k | } |
185 | | |
186 | 1 | return false; |
187 | 31.1k | } |
188 | | |
189 | 167k | bool writeBool(bool b) { |
190 | 167k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
191 | 167k | if (!writeFirstHeader()) { |
192 | 0 | return false; |
193 | 0 | } |
194 | 167k | if (b) { |
195 | 117k | os_->put((JsonbTypeUnder)JsonbType::T_True); |
196 | 117k | } else { |
197 | 49.7k | os_->put((JsonbTypeUnder)JsonbType::T_False); |
198 | 49.7k | } |
199 | | |
200 | 167k | kvState_ = WS_Value; |
201 | 167k | return true; |
202 | 167k | } |
203 | | |
204 | 18.4E | return false; |
205 | 167k | } |
206 | | |
207 | | // This function is a helper. It will make use of smallest space to |
208 | | // write an int |
209 | | bool writeInt(int64_t val) { |
210 | | if (val >= std::numeric_limits<int8_t>::min() && |
211 | | val <= std::numeric_limits<int8_t>::max()) { |
212 | | return writeInt8((int8_t)val); |
213 | | } else if (val >= std::numeric_limits<int16_t>::min() && |
214 | | val <= std::numeric_limits<int16_t>::max()) { |
215 | | return writeInt16((int16_t)val); |
216 | | } else if (val >= std::numeric_limits<int32_t>::min() && |
217 | | val <= std::numeric_limits<int32_t>::max()) { |
218 | | return writeInt32((int32_t)val); |
219 | | } else { |
220 | | return writeInt64(val); |
221 | | } |
222 | | } |
223 | | |
224 | 3.25M | bool writeInt8(int8_t v) { |
225 | 3.26M | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
226 | 3.26M | if (!writeFirstHeader()) { |
227 | 0 | return false; |
228 | 0 | } |
229 | 3.26M | os_->put((JsonbTypeUnder)JsonbType::T_Int8); |
230 | 3.26M | os_->put(v); |
231 | 3.26M | kvState_ = WS_Value; |
232 | 3.26M | return true; |
233 | 3.26M | } |
234 | | |
235 | 18.4E | return false; |
236 | 3.25M | } |
237 | | |
238 | 1.24M | bool writeInt16(int16_t v) { |
239 | 1.24M | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
240 | 1.24M | if (!writeFirstHeader()) { |
241 | 0 | return false; |
242 | 0 | } |
243 | 1.24M | os_->put((JsonbTypeUnder)JsonbType::T_Int16); |
244 | 1.24M | os_->write((char*)&v, sizeof(int16_t)); |
245 | 1.24M | kvState_ = WS_Value; |
246 | 1.24M | return true; |
247 | 1.24M | } |
248 | | |
249 | 18.4E | return false; |
250 | 1.24M | } |
251 | | |
252 | 2.80M | bool writeInt32(int32_t v) { |
253 | 2.81M | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
254 | 2.81M | if (!writeFirstHeader()) { |
255 | 0 | return false; |
256 | 0 | } |
257 | 2.81M | os_->put((JsonbTypeUnder)JsonbType::T_Int32); |
258 | 2.81M | os_->write((char*)&v, sizeof(int32_t)); |
259 | 2.81M | kvState_ = WS_Value; |
260 | 2.81M | return true; |
261 | 2.81M | } |
262 | | |
263 | 18.4E | return false; |
264 | 2.80M | } |
265 | | |
266 | 7.63M | bool writeInt64(int64_t v) { |
267 | 7.64M | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
268 | 7.64M | if (!writeFirstHeader()) { |
269 | 0 | return false; |
270 | 0 | } |
271 | 7.64M | os_->put((JsonbTypeUnder)JsonbType::T_Int64); |
272 | 7.64M | os_->write((char*)&v, sizeof(int64_t)); |
273 | 7.64M | kvState_ = WS_Value; |
274 | 7.64M | return true; |
275 | 7.64M | } |
276 | | |
277 | 18.4E | return false; |
278 | 7.63M | } |
279 | | |
280 | 4.31M | bool writeInt128(int128_t v) { |
281 | 4.31M | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
282 | 4.31M | if (!writeFirstHeader()) { |
283 | 0 | return false; |
284 | 0 | } |
285 | 4.31M | os_->put((JsonbTypeUnder)JsonbType::T_Int128); |
286 | 4.31M | os_->write((char*)&v, sizeof(int128_t)); |
287 | 4.31M | kvState_ = WS_Value; |
288 | 4.31M | return true; |
289 | 4.31M | } |
290 | | |
291 | 18.4E | return false; |
292 | 4.31M | } |
293 | | |
294 | 1.50M | bool writeDouble(double v) { |
295 | 1.50M | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
296 | 1.50M | if (!writeFirstHeader()) { |
297 | 0 | return false; |
298 | 0 | } |
299 | 1.50M | os_->put((JsonbTypeUnder)JsonbType::T_Double); |
300 | 1.50M | os_->write((char*)&v, sizeof(double)); |
301 | 1.50M | kvState_ = WS_Value; |
302 | 1.50M | return true; |
303 | 1.50M | } |
304 | | |
305 | 18.4E | return false; |
306 | 1.50M | } |
307 | | |
308 | 1.19M | bool writeFloat(float v) { |
309 | 1.19M | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
310 | 1.19M | if (!writeFirstHeader()) { |
311 | 0 | return false; |
312 | 0 | } |
313 | 1.19M | os_->put((JsonbTypeUnder)JsonbType::T_Float); |
314 | 1.19M | os_->write((char*)&v, sizeof(float)); |
315 | 1.19M | kvState_ = WS_Value; |
316 | 1.19M | return true; |
317 | 1.19M | } |
318 | | |
319 | 18.4E | return false; |
320 | 1.19M | } |
321 | | |
322 | | template <JsonbDecimalType T> |
323 | 115 | bool writeDecimal(const T& v, const uint32_t precision, const uint32_t scale) { |
324 | 115 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
325 | 115 | if (!writeFirstHeader()) { |
326 | 0 | return false; |
327 | 0 | } |
328 | | |
329 | 115 | if constexpr (std::same_as<T, Decimal256>) { |
330 | 6 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal256); |
331 | 31 | } else if constexpr (std::same_as<T, Decimal128V3>) { |
332 | 31 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal128); |
333 | 40 | } else if constexpr (std::same_as<T, Decimal64>) { |
334 | 40 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal64); |
335 | 40 | } else { |
336 | 38 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal32); |
337 | 38 | } |
338 | | |
339 | 115 | os_->write(reinterpret_cast<const char*>(&precision), sizeof(uint32_t)); |
340 | 115 | os_->write(reinterpret_cast<const char*>(&scale), sizeof(uint32_t)); |
341 | 115 | os_->write((char*)(&(v.value)), sizeof(v.value)); |
342 | 115 | kvState_ = WS_Value; |
343 | 115 | return true; |
344 | 115 | } |
345 | | |
346 | 0 | return false; |
347 | 115 | } _ZN5doris12JsonbWriterTINS_14JsonbOutStreamEE12writeDecimalITk16JsonbDecimalTypeNS_7DecimalIiEEEEbRKT_jj Line | Count | Source | 323 | 38 | bool writeDecimal(const T& v, const uint32_t precision, const uint32_t scale) { | 324 | 38 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { | 325 | 38 | if (!writeFirstHeader()) { | 326 | 0 | return false; | 327 | 0 | } | 328 | | | 329 | | if constexpr (std::same_as<T, Decimal256>) { | 330 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal256); | 331 | | } else if constexpr (std::same_as<T, Decimal128V3>) { | 332 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal128); | 333 | | } else if constexpr (std::same_as<T, Decimal64>) { | 334 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal64); | 335 | 38 | } else { | 336 | 38 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal32); | 337 | 38 | } | 338 | | | 339 | 38 | os_->write(reinterpret_cast<const char*>(&precision), sizeof(uint32_t)); | 340 | 38 | os_->write(reinterpret_cast<const char*>(&scale), sizeof(uint32_t)); | 341 | 38 | os_->write((char*)(&(v.value)), sizeof(v.value)); | 342 | 38 | kvState_ = WS_Value; | 343 | 38 | return true; | 344 | 38 | } | 345 | | | 346 | 0 | return false; | 347 | 38 | } |
_ZN5doris12JsonbWriterTINS_14JsonbOutStreamEE12writeDecimalITk16JsonbDecimalTypeNS_7DecimalIlEEEEbRKT_jj Line | Count | Source | 323 | 40 | bool writeDecimal(const T& v, const uint32_t precision, const uint32_t scale) { | 324 | 40 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { | 325 | 40 | if (!writeFirstHeader()) { | 326 | 0 | return false; | 327 | 0 | } | 328 | | | 329 | | if constexpr (std::same_as<T, Decimal256>) { | 330 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal256); | 331 | | } else if constexpr (std::same_as<T, Decimal128V3>) { | 332 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal128); | 333 | 40 | } else if constexpr (std::same_as<T, Decimal64>) { | 334 | 40 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal64); | 335 | | } else { | 336 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal32); | 337 | | } | 338 | | | 339 | 40 | os_->write(reinterpret_cast<const char*>(&precision), sizeof(uint32_t)); | 340 | 40 | os_->write(reinterpret_cast<const char*>(&scale), sizeof(uint32_t)); | 341 | 40 | os_->write((char*)(&(v.value)), sizeof(v.value)); | 342 | 40 | kvState_ = WS_Value; | 343 | 40 | return true; | 344 | 40 | } | 345 | | | 346 | 0 | return false; | 347 | 40 | } |
_ZN5doris12JsonbWriterTINS_14JsonbOutStreamEE12writeDecimalITk16JsonbDecimalTypeNS_12Decimal128V3EEEbRKT_jj Line | Count | Source | 323 | 31 | bool writeDecimal(const T& v, const uint32_t precision, const uint32_t scale) { | 324 | 31 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { | 325 | 31 | if (!writeFirstHeader()) { | 326 | 0 | return false; | 327 | 0 | } | 328 | | | 329 | | if constexpr (std::same_as<T, Decimal256>) { | 330 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal256); | 331 | 31 | } else if constexpr (std::same_as<T, Decimal128V3>) { | 332 | 31 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal128); | 333 | | } else if constexpr (std::same_as<T, Decimal64>) { | 334 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal64); | 335 | | } else { | 336 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal32); | 337 | | } | 338 | | | 339 | 31 | os_->write(reinterpret_cast<const char*>(&precision), sizeof(uint32_t)); | 340 | 31 | os_->write(reinterpret_cast<const char*>(&scale), sizeof(uint32_t)); | 341 | 31 | os_->write((char*)(&(v.value)), sizeof(v.value)); | 342 | 31 | kvState_ = WS_Value; | 343 | 31 | return true; | 344 | 31 | } | 345 | | | 346 | 0 | return false; | 347 | 31 | } |
_ZN5doris12JsonbWriterTINS_14JsonbOutStreamEE12writeDecimalITk16JsonbDecimalTypeNS_7DecimalIN4wide7integerILm256EiEEEEEEbRKT_jj Line | Count | Source | 323 | 6 | bool writeDecimal(const T& v, const uint32_t precision, const uint32_t scale) { | 324 | 6 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { | 325 | 6 | if (!writeFirstHeader()) { | 326 | 0 | return false; | 327 | 0 | } | 328 | | | 329 | 6 | if constexpr (std::same_as<T, Decimal256>) { | 330 | 6 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal256); | 331 | | } else if constexpr (std::same_as<T, Decimal128V3>) { | 332 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal128); | 333 | | } else if constexpr (std::same_as<T, Decimal64>) { | 334 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal64); | 335 | | } else { | 336 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal32); | 337 | | } | 338 | | | 339 | 6 | os_->write(reinterpret_cast<const char*>(&precision), sizeof(uint32_t)); | 340 | 6 | os_->write(reinterpret_cast<const char*>(&scale), sizeof(uint32_t)); | 341 | 6 | os_->write((char*)(&(v.value)), sizeof(v.value)); | 342 | 6 | kvState_ = WS_Value; | 343 | 6 | return true; | 344 | 6 | } | 345 | | | 346 | 0 | return false; | 347 | 6 | } |
|
348 | | |
349 | | // must call writeStartString before writing a string val |
350 | 1.63M | bool writeStartString() { |
351 | 1.63M | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
352 | 1.63M | if (!writeFirstHeader()) return 0; |
353 | 1.63M | os_->put((JsonbTypeUnder)JsonbType::T_String); |
354 | 1.63M | str_pos_ = os_->tellp(); |
355 | | |
356 | | // fill the size bytes with 0 for now |
357 | 1.63M | uint32_t size = 0; |
358 | 1.63M | os_->write((char*)&size, sizeof(uint32_t)); |
359 | | |
360 | 1.63M | kvState_ = WS_String; |
361 | 1.63M | return true; |
362 | 1.63M | } |
363 | | |
364 | 18.4E | return false; |
365 | 1.63M | } |
366 | | |
367 | | // finish writing a string val |
368 | 1.63M | bool writeEndString() { |
369 | 1.63M | if (kvState_ == WS_String) { |
370 | 1.63M | std::streampos cur_pos = os_->tellp(); |
371 | 1.63M | int32_t size = (int32_t)(cur_pos - str_pos_ - sizeof(uint32_t)); |
372 | 1.63M | assert(size >= 0); |
373 | | |
374 | 1.63M | os_->seekp(str_pos_); |
375 | 1.63M | os_->write((char*)&size, sizeof(uint32_t)); |
376 | 1.63M | os_->seekp(cur_pos); |
377 | | |
378 | 1.63M | kvState_ = WS_Value; |
379 | 1.63M | return true; |
380 | 1.63M | } |
381 | | |
382 | 18.4E | return false; |
383 | 1.63M | } |
384 | | |
385 | | // TODO: here changed length to uint64_t, as some api also need changed, But the thirdparty api is uint_32t |
386 | | // need consider a better way to handle case. |
387 | 1.61M | bool writeString(const char* str, uint64_t len) { |
388 | 1.61M | if (kvState_ == WS_String) { |
389 | 1.61M | os_->write(str, len); |
390 | 1.61M | return true; |
391 | 1.61M | } |
392 | | |
393 | 18.4E | return false; |
394 | 1.61M | } |
395 | | |
396 | 827k | bool writeString(const std::string& str) { return writeString(str.c_str(), str.size()); } |
397 | | bool writeString(char ch) { |
398 | | if (kvState_ == WS_String) { |
399 | | os_->put(ch); |
400 | | return true; |
401 | | } |
402 | | |
403 | | return false; |
404 | | } |
405 | | |
406 | | // must call writeStartBinary before writing a binary val |
407 | 3.79M | bool writeStartBinary() { |
408 | 3.81M | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
409 | 3.81M | if (!writeFirstHeader()) return 0; |
410 | 3.81M | os_->put((JsonbTypeUnder)JsonbType::T_Binary); |
411 | 3.81M | str_pos_ = os_->tellp(); |
412 | | |
413 | | // fill the size bytes with 0 for now |
414 | 3.81M | uint32_t size = 0; |
415 | 3.81M | os_->write((char*)&size, sizeof(uint32_t)); |
416 | | |
417 | 3.81M | kvState_ = WS_Binary; |
418 | 3.81M | return true; |
419 | 3.81M | } |
420 | | |
421 | 18.4E | return false; |
422 | 3.79M | } |
423 | | |
424 | | // finish writing a binary val |
425 | 3.80M | bool writeEndBinary() { |
426 | 3.80M | if (kvState_ == WS_Binary) { |
427 | 3.80M | std::streampos cur_pos = os_->tellp(); |
428 | 3.80M | int32_t size = (int32_t)(cur_pos - str_pos_ - sizeof(uint32_t)); |
429 | 3.80M | assert(size >= 0); |
430 | | |
431 | 3.80M | os_->seekp(str_pos_); |
432 | 3.80M | os_->write((char*)&size, sizeof(uint32_t)); |
433 | 3.80M | os_->seekp(cur_pos); |
434 | | |
435 | 3.80M | kvState_ = WS_Value; |
436 | 3.80M | return true; |
437 | 3.80M | } |
438 | | |
439 | 18.4E | return false; |
440 | 3.80M | } |
441 | | |
442 | 3.81M | bool writeBinary(const char* bin, uint64_t len) { |
443 | 3.81M | if (kvState_ == WS_Binary) { |
444 | 3.81M | os_->write(bin, len); |
445 | 3.81M | return true; |
446 | 3.81M | } |
447 | | |
448 | 18.4E | return false; |
449 | 3.81M | } |
450 | | |
451 | | // must call writeStartObject before writing an object val |
452 | 2.12M | bool writeStartObject() { |
453 | 2.12M | if (stack_.empty() || verifyValueState()) { |
454 | 2.12M | if (stack_.empty()) { |
455 | | // if this is a new JSONB, write the header |
456 | 1.95M | if (!hasHdr_) { |
457 | 1.95M | writeHeader(); |
458 | 1.95M | } else |
459 | 18.4E | return false; |
460 | 1.95M | } |
461 | | |
462 | | // check if the object exceeds the maximum nesting level |
463 | 2.12M | if (stack_.size() >= MaxNestingLevel) return false; |
464 | | |
465 | 2.12M | os_->put((JsonbTypeUnder)JsonbType::T_Object); |
466 | | // save the size position |
467 | 2.12M | stack_.push(WriteInfo({WS_Object, os_->tellp()})); |
468 | | |
469 | | // fill the size bytes with 0 for now |
470 | 2.12M | uint32_t size = 0; |
471 | 2.12M | os_->write((char*)&size, sizeof(uint32_t)); |
472 | | |
473 | 2.12M | kvState_ = WS_Value; |
474 | 2.12M | return true; |
475 | 2.12M | } |
476 | | |
477 | 18.4E | return false; |
478 | 2.12M | } |
479 | | |
480 | | // finish writing an object val |
481 | 2.12M | bool writeEndObject() { |
482 | 2.13M | if (!stack_.empty() && stack_.top().state == WS_Object && kvState_ == WS_Value) { |
483 | 2.13M | WriteInfo& ci = stack_.top(); |
484 | 2.13M | std::streampos cur_pos = os_->tellp(); |
485 | 2.13M | auto size = (int32_t)(cur_pos - ci.sz_pos - sizeof(uint32_t)); |
486 | 2.13M | assert(size >= 0); |
487 | | |
488 | 2.12M | os_->seekp(ci.sz_pos); |
489 | 2.12M | os_->write((char*)&size, sizeof(uint32_t)); |
490 | 2.12M | os_->seekp(cur_pos); |
491 | 2.12M | stack_.pop(); |
492 | | |
493 | 2.12M | return true; |
494 | 2.13M | } |
495 | | |
496 | 18.4E | return false; |
497 | 2.12M | } |
498 | | |
499 | | // must call writeStartArray before writing an array val |
500 | 461k | bool writeStartArray() { |
501 | 462k | if (stack_.empty() || verifyValueState()) { |
502 | 462k | if (stack_.empty()) { |
503 | | // if this is a new JSONB, write the header |
504 | 239k | if (!hasHdr_) { |
505 | 239k | writeHeader(); |
506 | 18.4E | } else { |
507 | 18.4E | return false; |
508 | 18.4E | } |
509 | 239k | } |
510 | | |
511 | | // check if the array exceeds the maximum nesting level |
512 | 462k | if (stack_.size() >= MaxNestingLevel) { |
513 | 0 | return false; |
514 | 0 | } |
515 | | |
516 | 462k | os_->put((JsonbTypeUnder)JsonbType::T_Array); |
517 | | // save the size position |
518 | 462k | stack_.push(WriteInfo({WS_Array, os_->tellp()})); |
519 | | |
520 | | // fill the size bytes with 0 for now |
521 | 462k | uint32_t size = 0; |
522 | 462k | os_->write((char*)&size, sizeof(uint32_t)); |
523 | | |
524 | 462k | kvState_ = WS_Value; |
525 | 462k | return true; |
526 | 462k | } |
527 | | |
528 | 18.4E | return false; |
529 | 461k | } |
530 | | |
531 | | // finish writing an array val |
532 | 461k | bool writeEndArray() { |
533 | 461k | if (!stack_.empty() && stack_.top().state == WS_Array && kvState_ == WS_Value) { |
534 | 461k | WriteInfo& ci = stack_.top(); |
535 | 461k | std::streampos cur_pos = os_->tellp(); |
536 | 461k | auto size = (int32_t)(cur_pos - ci.sz_pos - sizeof(uint32_t)); |
537 | 461k | assert(size >= 0); |
538 | | |
539 | 461k | os_->seekp(ci.sz_pos); |
540 | 461k | os_->write((char*)&size, sizeof(uint32_t)); |
541 | 461k | os_->seekp(cur_pos); |
542 | 461k | stack_.pop(); |
543 | | |
544 | 461k | return true; |
545 | 461k | } |
546 | | |
547 | 18.4E | return false; |
548 | 461k | } |
549 | | |
550 | 6.97M | OS_TYPE* getOutput() { return os_; } |
551 | | |
552 | | #ifdef BE_TEST |
553 | | const JsonbDocument* getDocument() { |
554 | | const JsonbDocument* doc = nullptr; |
555 | | THROW_IF_ERROR(JsonbDocument::checkAndCreateDocument(getOutput()->getBuffer(), |
556 | | getOutput()->getSize(), &doc)); |
557 | | return doc; |
558 | | } |
559 | | |
560 | | const JsonbValue* getValue() { |
561 | | return JsonbDocument::createValue(getOutput()->getBuffer(), getOutput()->getSize()); |
562 | | } |
563 | | #endif |
564 | | |
565 | | bool writeEnd() { |
566 | | while (!stack_.empty()) { |
567 | | bool ok = false; |
568 | | switch (stack_.top().state) { |
569 | | case WS_Array: |
570 | | ok = writeEndArray(); |
571 | | break; |
572 | | case WS_Object: |
573 | | ok = writeEndObject(); |
574 | | break; |
575 | | case WS_String: |
576 | | ok = writeEndString(); |
577 | | break; |
578 | | case WS_Binary: |
579 | | ok = writeEndBinary(); |
580 | | break; |
581 | | default: |
582 | | ok = false; |
583 | | break; |
584 | | } |
585 | | if (!ok) { |
586 | | return false; |
587 | | } |
588 | | } |
589 | | return true; |
590 | | } |
591 | | |
592 | | private: |
593 | | // verify we are in the right state before writing a value |
594 | 25.9M | bool verifyValueState() { |
595 | 25.9M | assert(!stack_.empty()); |
596 | | // The document can only be an Object or an Array which follows |
597 | | // the standard. |
598 | 25.9M | return (stack_.top().state == WS_Object && kvState_ == WS_Key) || |
599 | 25.9M | (stack_.top().state == WS_Array && kvState_ == WS_Value); |
600 | 25.9M | } |
601 | | |
602 | | // verify we are in the right state before writing a key |
603 | 44.9M | bool verifyKeyState() { |
604 | 44.9M | assert(!stack_.empty()); |
605 | 46.3M | return stack_.top().state == WS_Object && kvState_ == WS_Value; |
606 | 44.9M | } |
607 | | |
608 | 3.49M | void writeHeader() { |
609 | 3.49M | os_->put(JSONB_VER); |
610 | 3.49M | hasHdr_ = true; |
611 | 3.49M | } |
612 | | |
613 | | enum WriteState { |
614 | | WS_NONE, |
615 | | WS_Array, |
616 | | WS_Object, |
617 | | WS_Key, |
618 | | WS_Value, |
619 | | WS_String, |
620 | | WS_Binary, |
621 | | }; |
622 | | |
623 | | struct WriteInfo { |
624 | | WriteState state; |
625 | | std::streampos sz_pos; |
626 | | }; |
627 | | |
628 | | OS_TYPE* os_ = nullptr; |
629 | | bool alloc_; |
630 | | bool hasHdr_; |
631 | | WriteState kvState_; // key or value state |
632 | | std::streampos str_pos_; |
633 | | std::stack<WriteInfo> stack_; |
634 | | bool first_ = true; |
635 | | }; |
636 | | |
637 | | using JsonbWriter = JsonbWriterT<JsonbOutStream>; |
638 | | |
639 | | } // namespace doris |
640 | | |
641 | | #endif // JSONB_JSONBWRITER_H |