/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 <cstdint> |
40 | | #include <limits> |
41 | | #include <stack> |
42 | | #include <string> |
43 | | |
44 | | #include "common/exception.h" |
45 | | #include "common/status.h" |
46 | | #include "jsonb_document.h" |
47 | | #include "jsonb_stream.h" |
48 | | #include "runtime/define_primitive_type.h" |
49 | | #include "runtime/primitive_type.h" |
50 | | #include "vec/core/types.h" |
51 | | |
52 | | namespace doris { |
53 | | |
54 | | using int128_t = __int128; |
55 | | |
56 | | template <class OS_TYPE> |
57 | | class JsonbWriterT { |
58 | | /// TODO: maybe we should not use a template class here |
59 | | static_assert(std::is_same_v<OS_TYPE, JsonbOutStream>); |
60 | | |
61 | | public: |
62 | 3.79k | JsonbWriterT() : alloc_(true), hasHdr_(false), kvState_(WS_Value), str_pos_(0) { |
63 | 3.79k | os_ = new OS_TYPE(); |
64 | 3.79k | } |
65 | | |
66 | | explicit JsonbWriterT(OS_TYPE& os) |
67 | | : os_(&os), alloc_(false), hasHdr_(false), kvState_(WS_Value), str_pos_(0) {} |
68 | | |
69 | 3.79k | ~JsonbWriterT() { |
70 | 3.79k | if (alloc_) { |
71 | 3.79k | delete os_; |
72 | 3.79k | } |
73 | 3.79k | } |
74 | | |
75 | 0 | JsonbWriterT<OS_TYPE>& operator=(JsonbWriterT<OS_TYPE>&& other) { |
76 | 0 | if (this != &other) { |
77 | 0 | if (alloc_) { |
78 | 0 | delete os_; |
79 | 0 | } |
80 | 0 | os_ = other.os_; |
81 | 0 | other.os_ = nullptr; |
82 | 0 | alloc_ = other.alloc_; |
83 | 0 | other.alloc_ = false; |
84 | 0 | hasHdr_ = other.hasHdr_; |
85 | 0 | kvState_ = other.kvState_; |
86 | 0 | str_pos_ = other.str_pos_; |
87 | 0 | first_ = other.first_; |
88 | 0 | stack_ = std::move(other.stack_); |
89 | 0 | } |
90 | 0 | return *this; |
91 | 0 | } |
92 | | |
93 | 5.17k | void reset() { |
94 | 5.17k | os_->clear(); |
95 | 5.17k | os_->seekp(0); |
96 | 5.17k | hasHdr_ = false; |
97 | 5.17k | kvState_ = WS_Value; |
98 | 5.17k | first_ = true; |
99 | 5.17k | for (; !stack_.empty(); stack_.pop()) { |
100 | 0 | ; |
101 | 0 | } |
102 | 5.17k | } |
103 | | |
104 | 17 | bool writeKey(const char* key) { return writeKey(key, strlen(key)); } |
105 | | |
106 | | // write a key string (or key id if an external dict is provided) |
107 | 6.48k | bool writeKey(const char* key, uint8_t len) { |
108 | 6.48k | if (!stack_.empty() && verifyKeyState()) { |
109 | 6.48k | os_->put(len); |
110 | 6.48k | if (len == 0) { |
111 | | // NOTE: we use sMaxKeyId to represent an empty key |
112 | 250 | JsonbKeyValue::keyid_type idx = JsonbKeyValue::sMaxKeyId; |
113 | 250 | os_->write((char*)&idx, sizeof(JsonbKeyValue::keyid_type)); |
114 | 6.23k | } else { |
115 | 6.23k | os_->write(key, len); |
116 | 6.23k | } |
117 | 6.48k | kvState_ = WS_Key; |
118 | 6.48k | return true; |
119 | 6.48k | } |
120 | | |
121 | 1 | return false; |
122 | 6.48k | } |
123 | | |
124 | 63 | bool writeValue(const JsonbValue* value) { |
125 | 63 | if (!value) { |
126 | 0 | return writeNull(); |
127 | 0 | } |
128 | | |
129 | 63 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
130 | 63 | if (!writeFirstHeader()) { |
131 | 0 | return false; |
132 | 0 | } |
133 | 63 | os_->write((char*)value, value->numPackedBytes()); |
134 | 63 | kvState_ = WS_Value; |
135 | 63 | return true; |
136 | 63 | } |
137 | 0 | return false; |
138 | 63 | } |
139 | | |
140 | | // write a key id |
141 | 18.2k | bool writeKey(JsonbKeyValue::keyid_type idx) { |
142 | 18.2k | if (!stack_.empty() && verifyKeyState()) { |
143 | 16.8k | os_->put(0); |
144 | 16.8k | os_->write((char*)&idx, sizeof(JsonbKeyValue::keyid_type)); |
145 | 16.8k | kvState_ = WS_Key; |
146 | 16.8k | return true; |
147 | 16.8k | } |
148 | | |
149 | 1.42k | return false; |
150 | 18.2k | } |
151 | | |
152 | 525k | bool writeFirstHeader() { |
153 | 525k | if (first_ && stack_.empty()) { |
154 | 753 | first_ = false; |
155 | | // if this is a new JSONB, write the header |
156 | 753 | if (!hasHdr_) { |
157 | 753 | writeHeader(); |
158 | 753 | return true; |
159 | 753 | } else { |
160 | 0 | return false; |
161 | 0 | } |
162 | 524k | } else { |
163 | 524k | return true; |
164 | 524k | } |
165 | 525k | } |
166 | | |
167 | 14.4k | bool writeNull() { |
168 | 14.4k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
169 | 14.4k | if (!writeFirstHeader()) { |
170 | 0 | return false; |
171 | 0 | } |
172 | 14.4k | os_->put((JsonbTypeUnder)JsonbType::T_Null); |
173 | 14.4k | kvState_ = WS_Value; |
174 | 14.4k | return true; |
175 | 14.4k | } |
176 | | |
177 | 0 | return false; |
178 | 14.4k | } |
179 | | |
180 | 20.7k | bool writeBool(bool b) { |
181 | 20.7k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
182 | 20.7k | if (!writeFirstHeader()) { |
183 | 0 | return false; |
184 | 0 | } |
185 | 20.7k | if (b) { |
186 | 10.3k | os_->put((JsonbTypeUnder)JsonbType::T_True); |
187 | 10.3k | } else { |
188 | 10.3k | os_->put((JsonbTypeUnder)JsonbType::T_False); |
189 | 10.3k | } |
190 | | |
191 | 20.7k | kvState_ = WS_Value; |
192 | 20.7k | return true; |
193 | 20.7k | } |
194 | | |
195 | 0 | return false; |
196 | 20.7k | } |
197 | | |
198 | | // This function is a helper. It will make use of smallest space to |
199 | | // write an int |
200 | 2 | bool writeInt(int64_t val) { |
201 | 2 | if (val >= std::numeric_limits<int8_t>::min() && |
202 | 2 | val <= std::numeric_limits<int8_t>::max()) { |
203 | 1 | return writeInt8((int8_t)val); |
204 | 1 | } else if (val >= std::numeric_limits<int16_t>::min() && |
205 | 1 | val <= std::numeric_limits<int16_t>::max()) { |
206 | 1 | return writeInt16((int16_t)val); |
207 | 1 | } else if (val >= std::numeric_limits<int32_t>::min() && |
208 | 0 | val <= std::numeric_limits<int32_t>::max()) { |
209 | 0 | return writeInt32((int32_t)val); |
210 | 0 | } else { |
211 | 0 | return writeInt64(val); |
212 | 0 | } |
213 | 2 | } |
214 | | |
215 | 48.8k | bool writeInt8(int8_t v) { |
216 | 48.8k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
217 | 48.8k | if (!writeFirstHeader()) { |
218 | 0 | return false; |
219 | 0 | } |
220 | 48.8k | os_->put((JsonbTypeUnder)JsonbType::T_Int8); |
221 | 48.8k | os_->put(v); |
222 | 48.8k | kvState_ = WS_Value; |
223 | 48.8k | return true; |
224 | 48.8k | } |
225 | | |
226 | 0 | return false; |
227 | 48.8k | } |
228 | | |
229 | 184 | bool writeInt16(int16_t v) { |
230 | 184 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
231 | 184 | if (!writeFirstHeader()) { |
232 | 0 | return false; |
233 | 0 | } |
234 | 184 | os_->put((JsonbTypeUnder)JsonbType::T_Int16); |
235 | 184 | os_->write((char*)&v, sizeof(int16_t)); |
236 | 184 | kvState_ = WS_Value; |
237 | 184 | return true; |
238 | 184 | } |
239 | | |
240 | 0 | return false; |
241 | 184 | } |
242 | | |
243 | 3.48k | bool writeInt32(int32_t v) { |
244 | 3.48k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
245 | 3.48k | if (!writeFirstHeader()) { |
246 | 0 | return false; |
247 | 0 | } |
248 | 3.48k | os_->put((JsonbTypeUnder)JsonbType::T_Int32); |
249 | 3.48k | os_->write((char*)&v, sizeof(int32_t)); |
250 | 3.48k | kvState_ = WS_Value; |
251 | 3.48k | return true; |
252 | 3.48k | } |
253 | | |
254 | 0 | return false; |
255 | 3.48k | } |
256 | | |
257 | 62.8k | bool writeInt64(int64_t v) { |
258 | 62.8k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
259 | 62.8k | if (!writeFirstHeader()) { |
260 | 0 | return false; |
261 | 0 | } |
262 | 62.8k | os_->put((JsonbTypeUnder)JsonbType::T_Int64); |
263 | 62.8k | os_->write((char*)&v, sizeof(int64_t)); |
264 | 62.8k | kvState_ = WS_Value; |
265 | 62.8k | return true; |
266 | 62.8k | } |
267 | | |
268 | 0 | return false; |
269 | 62.8k | } |
270 | | |
271 | 4.17k | bool writeInt128(int128_t v) { |
272 | 4.17k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
273 | 4.17k | if (!writeFirstHeader()) { |
274 | 0 | return false; |
275 | 0 | } |
276 | 4.17k | os_->put((JsonbTypeUnder)JsonbType::T_Int128); |
277 | 4.17k | os_->write((char*)&v, sizeof(int128_t)); |
278 | 4.17k | kvState_ = WS_Value; |
279 | 4.17k | return true; |
280 | 4.17k | } |
281 | | |
282 | 0 | return false; |
283 | 4.17k | } |
284 | | |
285 | 115k | bool writeDouble(double v) { |
286 | 115k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
287 | 115k | if (!writeFirstHeader()) { |
288 | 0 | return false; |
289 | 0 | } |
290 | 115k | os_->put((JsonbTypeUnder)JsonbType::T_Double); |
291 | 115k | os_->write((char*)&v, sizeof(double)); |
292 | 115k | kvState_ = WS_Value; |
293 | 115k | return true; |
294 | 115k | } |
295 | | |
296 | 0 | return false; |
297 | 115k | } |
298 | | |
299 | 19 | bool writeFloat(float v) { |
300 | 19 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
301 | 19 | if (!writeFirstHeader()) { |
302 | 0 | return false; |
303 | 0 | } |
304 | 19 | os_->put((JsonbTypeUnder)JsonbType::T_Float); |
305 | 19 | os_->write((char*)&v, sizeof(float)); |
306 | 19 | kvState_ = WS_Value; |
307 | 19 | return true; |
308 | 19 | } |
309 | | |
310 | 0 | return false; |
311 | 19 | } |
312 | | |
313 | | template <JsonbDecimalType T> |
314 | 9 | bool writeDecimal(const T& v, const uint32_t precision, const uint32_t scale) { |
315 | 9 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
316 | 9 | if (!writeFirstHeader()) { |
317 | 0 | return false; |
318 | 0 | } |
319 | | |
320 | 9 | if constexpr (std::same_as<T, vectorized::Decimal256>) { |
321 | 2 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal256); |
322 | 3 | } else if constexpr (std::same_as<T, vectorized::Decimal128V3>) { |
323 | 3 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal128); |
324 | 3 | } else if constexpr (std::same_as<T, vectorized::Decimal64>) { |
325 | 2 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal64); |
326 | 2 | } else { |
327 | 2 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal32); |
328 | 2 | } |
329 | | |
330 | 9 | os_->write(reinterpret_cast<const char*>(&precision), sizeof(uint32_t)); |
331 | 9 | os_->write(reinterpret_cast<const char*>(&scale), sizeof(uint32_t)); |
332 | 9 | os_->write((char*)(&(v.value)), sizeof(v.value)); |
333 | 9 | kvState_ = WS_Value; |
334 | 9 | return true; |
335 | 9 | } |
336 | | |
337 | 0 | return false; |
338 | 9 | } _ZN5doris12JsonbWriterTINS_14JsonbOutStreamEE12writeDecimalITk16JsonbDecimalTypeNS_10vectorized7DecimalIiEEEEbRKT_jj Line | Count | Source | 314 | 2 | bool writeDecimal(const T& v, const uint32_t precision, const uint32_t scale) { | 315 | 2 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { | 316 | 2 | if (!writeFirstHeader()) { | 317 | 0 | return false; | 318 | 0 | } | 319 | | | 320 | | if constexpr (std::same_as<T, vectorized::Decimal256>) { | 321 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal256); | 322 | | } else if constexpr (std::same_as<T, vectorized::Decimal128V3>) { | 323 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal128); | 324 | | } else if constexpr (std::same_as<T, vectorized::Decimal64>) { | 325 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal64); | 326 | 2 | } else { | 327 | 2 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal32); | 328 | 2 | } | 329 | | | 330 | 2 | os_->write(reinterpret_cast<const char*>(&precision), sizeof(uint32_t)); | 331 | 2 | os_->write(reinterpret_cast<const char*>(&scale), sizeof(uint32_t)); | 332 | 2 | os_->write((char*)(&(v.value)), sizeof(v.value)); | 333 | 2 | kvState_ = WS_Value; | 334 | 2 | return true; | 335 | 2 | } | 336 | | | 337 | 0 | return false; | 338 | 2 | } |
_ZN5doris12JsonbWriterTINS_14JsonbOutStreamEE12writeDecimalITk16JsonbDecimalTypeNS_10vectorized7DecimalIlEEEEbRKT_jj Line | Count | Source | 314 | 2 | bool writeDecimal(const T& v, const uint32_t precision, const uint32_t scale) { | 315 | 2 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { | 316 | 2 | if (!writeFirstHeader()) { | 317 | 0 | return false; | 318 | 0 | } | 319 | | | 320 | | if constexpr (std::same_as<T, vectorized::Decimal256>) { | 321 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal256); | 322 | | } else if constexpr (std::same_as<T, vectorized::Decimal128V3>) { | 323 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal128); | 324 | 2 | } else if constexpr (std::same_as<T, vectorized::Decimal64>) { | 325 | 2 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal64); | 326 | | } else { | 327 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal32); | 328 | | } | 329 | | | 330 | 2 | os_->write(reinterpret_cast<const char*>(&precision), sizeof(uint32_t)); | 331 | 2 | os_->write(reinterpret_cast<const char*>(&scale), sizeof(uint32_t)); | 332 | 2 | os_->write((char*)(&(v.value)), sizeof(v.value)); | 333 | 2 | kvState_ = WS_Value; | 334 | 2 | return true; | 335 | 2 | } | 336 | | | 337 | 0 | return false; | 338 | 2 | } |
_ZN5doris12JsonbWriterTINS_14JsonbOutStreamEE12writeDecimalITk16JsonbDecimalTypeNS_10vectorized12Decimal128V3EEEbRKT_jj Line | Count | Source | 314 | 3 | bool writeDecimal(const T& v, const uint32_t precision, const uint32_t scale) { | 315 | 3 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { | 316 | 3 | if (!writeFirstHeader()) { | 317 | 0 | return false; | 318 | 0 | } | 319 | | | 320 | | if constexpr (std::same_as<T, vectorized::Decimal256>) { | 321 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal256); | 322 | 3 | } else if constexpr (std::same_as<T, vectorized::Decimal128V3>) { | 323 | 3 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal128); | 324 | | } else if constexpr (std::same_as<T, vectorized::Decimal64>) { | 325 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal64); | 326 | | } else { | 327 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal32); | 328 | | } | 329 | | | 330 | 3 | os_->write(reinterpret_cast<const char*>(&precision), sizeof(uint32_t)); | 331 | 3 | os_->write(reinterpret_cast<const char*>(&scale), sizeof(uint32_t)); | 332 | 3 | os_->write((char*)(&(v.value)), sizeof(v.value)); | 333 | 3 | kvState_ = WS_Value; | 334 | 3 | return true; | 335 | 3 | } | 336 | | | 337 | 0 | return false; | 338 | 3 | } |
_ZN5doris12JsonbWriterTINS_14JsonbOutStreamEE12writeDecimalITk16JsonbDecimalTypeNS_10vectorized7DecimalIN4wide7integerILm256EiEEEEEEbRKT_jj Line | Count | Source | 314 | 2 | bool writeDecimal(const T& v, const uint32_t precision, const uint32_t scale) { | 315 | 2 | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { | 316 | 2 | if (!writeFirstHeader()) { | 317 | 0 | return false; | 318 | 0 | } | 319 | | | 320 | 2 | if constexpr (std::same_as<T, vectorized::Decimal256>) { | 321 | 2 | os_->put((JsonbTypeUnder)JsonbType::T_Decimal256); | 322 | | } else if constexpr (std::same_as<T, vectorized::Decimal128V3>) { | 323 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal128); | 324 | | } else if constexpr (std::same_as<T, vectorized::Decimal64>) { | 325 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal64); | 326 | | } else { | 327 | | os_->put((JsonbTypeUnder)JsonbType::T_Decimal32); | 328 | | } | 329 | | | 330 | 2 | os_->write(reinterpret_cast<const char*>(&precision), sizeof(uint32_t)); | 331 | 2 | os_->write(reinterpret_cast<const char*>(&scale), sizeof(uint32_t)); | 332 | 2 | os_->write((char*)(&(v.value)), sizeof(v.value)); | 333 | 2 | kvState_ = WS_Value; | 334 | 2 | return true; | 335 | 2 | } | 336 | | | 337 | 0 | return false; | 338 | 2 | } |
|
339 | | |
340 | | // must call writeStartString before writing a string val |
341 | 250k | bool writeStartString() { |
342 | 250k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
343 | 250k | if (!writeFirstHeader()) return 0; |
344 | 250k | os_->put((JsonbTypeUnder)JsonbType::T_String); |
345 | 250k | str_pos_ = os_->tellp(); |
346 | | |
347 | | // fill the size bytes with 0 for now |
348 | 250k | uint32_t size = 0; |
349 | 250k | os_->write((char*)&size, sizeof(uint32_t)); |
350 | | |
351 | 250k | kvState_ = WS_String; |
352 | 250k | return true; |
353 | 250k | } |
354 | | |
355 | 0 | return false; |
356 | 250k | } |
357 | | |
358 | | // finish writing a string val |
359 | 250k | bool writeEndString() { |
360 | 250k | if (kvState_ == WS_String) { |
361 | 250k | std::streampos cur_pos = os_->tellp(); |
362 | 250k | int32_t size = (int32_t)(cur_pos - str_pos_ - sizeof(uint32_t)); |
363 | 250k | assert(size >= 0); |
364 | | |
365 | 250k | os_->seekp(str_pos_); |
366 | 250k | os_->write((char*)&size, sizeof(uint32_t)); |
367 | 250k | os_->seekp(cur_pos); |
368 | | |
369 | 250k | kvState_ = WS_Value; |
370 | 250k | return true; |
371 | 250k | } |
372 | | |
373 | 0 | return false; |
374 | 250k | } |
375 | | |
376 | | // TODO: here changed length to uint64_t, as some api also need changed, But the thirdparty api is uint_32t |
377 | | // need consider a better way to handle case. |
378 | 239k | bool writeString(const char* str, uint64_t len) { |
379 | 239k | if (kvState_ == WS_String) { |
380 | 239k | os_->write(str, len); |
381 | 239k | return true; |
382 | 239k | } |
383 | | |
384 | 5 | return false; |
385 | 239k | } |
386 | | |
387 | 19 | bool writeString(const std::string& str) { return writeString(str.c_str(), str.size()); } |
388 | | bool writeString(char ch) { |
389 | | if (kvState_ == WS_String) { |
390 | | os_->put(ch); |
391 | | return true; |
392 | | } |
393 | | |
394 | | return false; |
395 | | } |
396 | | |
397 | | // must call writeStartBinary before writing a binary val |
398 | 4.36k | bool writeStartBinary() { |
399 | 4.36k | if ((first_ && stack_.empty()) || (!stack_.empty() && verifyValueState())) { |
400 | 4.36k | if (!writeFirstHeader()) return 0; |
401 | 4.36k | os_->put((JsonbTypeUnder)JsonbType::T_Binary); |
402 | 4.36k | str_pos_ = os_->tellp(); |
403 | | |
404 | | // fill the size bytes with 0 for now |
405 | 4.36k | uint32_t size = 0; |
406 | 4.36k | os_->write((char*)&size, sizeof(uint32_t)); |
407 | | |
408 | 4.36k | kvState_ = WS_Binary; |
409 | 4.36k | return true; |
410 | 4.36k | } |
411 | | |
412 | 0 | return false; |
413 | 4.36k | } |
414 | | |
415 | | // finish writing a binary val |
416 | 4.36k | bool writeEndBinary() { |
417 | 4.36k | if (kvState_ == WS_Binary) { |
418 | 4.36k | std::streampos cur_pos = os_->tellp(); |
419 | 4.36k | int32_t size = (int32_t)(cur_pos - str_pos_ - sizeof(uint32_t)); |
420 | 4.36k | assert(size >= 0); |
421 | | |
422 | 4.36k | os_->seekp(str_pos_); |
423 | 4.36k | os_->write((char*)&size, sizeof(uint32_t)); |
424 | 4.36k | os_->seekp(cur_pos); |
425 | | |
426 | 4.36k | kvState_ = WS_Value; |
427 | 4.36k | return true; |
428 | 4.36k | } |
429 | | |
430 | 0 | return false; |
431 | 4.36k | } |
432 | | |
433 | 4.36k | bool writeBinary(const char* bin, uint64_t len) { |
434 | 4.36k | if (kvState_ == WS_Binary) { |
435 | 4.36k | os_->write(bin, len); |
436 | 4.36k | return true; |
437 | 4.36k | } |
438 | | |
439 | 0 | return false; |
440 | 4.36k | } |
441 | | |
442 | | // must call writeStartObject before writing an object val |
443 | 4.16k | bool writeStartObject() { |
444 | 4.16k | if (stack_.empty() || verifyValueState()) { |
445 | 4.16k | if (stack_.empty()) { |
446 | | // if this is a new JSONB, write the header |
447 | 1.49k | if (!hasHdr_) { |
448 | 1.49k | writeHeader(); |
449 | 1.49k | } else |
450 | 0 | return false; |
451 | 1.49k | } |
452 | | |
453 | | // check if the object exceeds the maximum nesting level |
454 | 4.16k | if (stack_.size() >= MaxNestingLevel) return false; |
455 | | |
456 | 4.16k | os_->put((JsonbTypeUnder)JsonbType::T_Object); |
457 | | // save the size position |
458 | 4.16k | stack_.push(WriteInfo({WS_Object, os_->tellp()})); |
459 | | |
460 | | // fill the size bytes with 0 for now |
461 | 4.16k | uint32_t size = 0; |
462 | 4.16k | os_->write((char*)&size, sizeof(uint32_t)); |
463 | | |
464 | 4.16k | kvState_ = WS_Value; |
465 | 4.16k | return true; |
466 | 4.16k | } |
467 | | |
468 | 0 | return false; |
469 | 4.16k | } |
470 | | |
471 | | // finish writing an object val |
472 | 3.80k | bool writeEndObject() { |
473 | 3.80k | if (!stack_.empty() && stack_.top().state == WS_Object && kvState_ == WS_Value) { |
474 | 3.80k | WriteInfo& ci = stack_.top(); |
475 | 3.80k | std::streampos cur_pos = os_->tellp(); |
476 | 3.80k | auto size = (int32_t)(cur_pos - ci.sz_pos - sizeof(uint32_t)); |
477 | 3.80k | assert(size >= 0); |
478 | | |
479 | 3.80k | os_->seekp(ci.sz_pos); |
480 | 3.80k | os_->write((char*)&size, sizeof(uint32_t)); |
481 | 3.80k | os_->seekp(cur_pos); |
482 | 3.80k | stack_.pop(); |
483 | | |
484 | 3.80k | return true; |
485 | 3.80k | } |
486 | | |
487 | 1 | return false; |
488 | 3.80k | } |
489 | | |
490 | | // must call writeStartArray before writing an array val |
491 | 51.8k | bool writeStartArray() { |
492 | 51.8k | if (stack_.empty() || verifyValueState()) { |
493 | 51.8k | if (stack_.empty()) { |
494 | | // if this is a new JSONB, write the header |
495 | 1.56k | if (!hasHdr_) { |
496 | 1.56k | writeHeader(); |
497 | 1.56k | } else { |
498 | 0 | return false; |
499 | 0 | } |
500 | 1.56k | } |
501 | | |
502 | | // check if the array exceeds the maximum nesting level |
503 | 51.8k | if (stack_.size() >= MaxNestingLevel) { |
504 | 0 | return false; |
505 | 0 | } |
506 | | |
507 | 51.8k | os_->put((JsonbTypeUnder)JsonbType::T_Array); |
508 | | // save the size position |
509 | 51.8k | stack_.push(WriteInfo({WS_Array, os_->tellp()})); |
510 | | |
511 | | // fill the size bytes with 0 for now |
512 | 51.8k | uint32_t size = 0; |
513 | 51.8k | os_->write((char*)&size, sizeof(uint32_t)); |
514 | | |
515 | 51.8k | kvState_ = WS_Value; |
516 | 51.8k | return true; |
517 | 51.8k | } |
518 | | |
519 | 0 | return false; |
520 | 51.8k | } |
521 | | |
522 | | // finish writing an array val |
523 | 51.6k | bool writeEndArray() { |
524 | 51.6k | if (!stack_.empty() && stack_.top().state == WS_Array && kvState_ == WS_Value) { |
525 | 51.6k | WriteInfo& ci = stack_.top(); |
526 | 51.6k | std::streampos cur_pos = os_->tellp(); |
527 | 51.6k | auto size = (int32_t)(cur_pos - ci.sz_pos - sizeof(uint32_t)); |
528 | 51.6k | assert(size >= 0); |
529 | | |
530 | 51.6k | os_->seekp(ci.sz_pos); |
531 | 51.6k | os_->write((char*)&size, sizeof(uint32_t)); |
532 | 51.6k | os_->seekp(cur_pos); |
533 | 51.6k | stack_.pop(); |
534 | | |
535 | 51.6k | return true; |
536 | 51.6k | } |
537 | | |
538 | 0 | return false; |
539 | 51.6k | } |
540 | | |
541 | 7.05k | OS_TYPE* getOutput() { return os_; } |
542 | 1 | JsonbDocument* getDocument() { |
543 | 1 | JsonbDocument* doc = nullptr; |
544 | 1 | THROW_IF_ERROR(JsonbDocument::checkAndCreateDocument(getOutput()->getBuffer(), |
545 | 1 | getOutput()->getSize(), &doc)); |
546 | 1 | return doc; |
547 | 1 | } |
548 | | |
549 | | JsonbValue* getValue() { |
550 | | return JsonbDocument::createValue(getOutput()->getBuffer(), getOutput()->getSize()); |
551 | | } |
552 | | |
553 | | bool writeEnd() { |
554 | | while (!stack_.empty()) { |
555 | | bool ok = false; |
556 | | switch (stack_.top().state) { |
557 | | case WS_Array: |
558 | | ok = writeEndArray(); |
559 | | break; |
560 | | case WS_Object: |
561 | | ok = writeEndObject(); |
562 | | break; |
563 | | case WS_String: |
564 | | ok = writeEndString(); |
565 | | break; |
566 | | case WS_Binary: |
567 | | ok = writeEndBinary(); |
568 | | break; |
569 | | default: |
570 | | ok = false; |
571 | | break; |
572 | | } |
573 | | if (!ok) { |
574 | | return false; |
575 | | } |
576 | | } |
577 | | return true; |
578 | | } |
579 | | |
580 | | private: |
581 | | // verify we are in the right state before writing a value |
582 | 577k | bool verifyValueState() { |
583 | 577k | assert(!stack_.empty()); |
584 | | // The document can only be an Object or an Array which follows |
585 | | // the standard. |
586 | 577k | return (stack_.top().state == WS_Object && kvState_ == WS_Key) || |
587 | 577k | (stack_.top().state == WS_Array && kvState_ == WS_Value); |
588 | 577k | } |
589 | | |
590 | | // verify we are in the right state before writing a key |
591 | 24.7k | bool verifyKeyState() { |
592 | 24.7k | assert(!stack_.empty()); |
593 | 24.7k | return stack_.top().state == WS_Object && kvState_ == WS_Value; |
594 | 24.7k | } |
595 | | |
596 | 3.81k | void writeHeader() { |
597 | 3.81k | os_->put(JSONB_VER); |
598 | 3.81k | hasHdr_ = true; |
599 | 3.81k | } |
600 | | |
601 | | enum WriteState { |
602 | | WS_NONE, |
603 | | WS_Array, |
604 | | WS_Object, |
605 | | WS_Key, |
606 | | WS_Value, |
607 | | WS_String, |
608 | | WS_Binary, |
609 | | }; |
610 | | |
611 | | struct WriteInfo { |
612 | | WriteState state; |
613 | | std::streampos sz_pos; |
614 | | }; |
615 | | |
616 | | OS_TYPE* os_ = nullptr; |
617 | | bool alloc_; |
618 | | bool hasHdr_; |
619 | | WriteState kvState_; // key or value state |
620 | | std::streampos str_pos_; |
621 | | std::stack<WriteInfo> stack_; |
622 | | bool first_ = true; |
623 | | }; |
624 | | |
625 | | using JsonbWriter = JsonbWriterT<JsonbOutStream>; |
626 | | |
627 | | } // namespace doris |
628 | | |
629 | | #endif // JSONB_JSONBWRITER_H |