be/src/util/jsonb_utils.h

Source
/*
 *  Copyright (c) 2014, Facebook, Inc.
 *  All rights reserved.
 *
 *  This source code is licensed under the BSD-style license found in the
 *  LICENSE file in the root directory of this source tree. An additional grant
 *  of patent rights can be found in the PATENTS file in the same directory.
 *
 */

/*
 * This header file defines miscellaneous utility classes.
 *
 * @author Tian Xia <tianx@fb.com>
 *
 *  this file is copied from 
 * https://github.com/facebook/mysql-5.6/blob/fb-mysql-5.6.35/fbson/FbsonUtil.h
 * and modified by Doris
 */

#ifndef JSONB_JSONBUTIL_H
#define JSONB_JSONBUTIL_H

#include "common/status.h"
#include "util/jsonb_document.h"
#include "util/jsonb_stream.h"

namespace doris {

#define OUT_BUF_SIZE 1024

/*
 * JsonbToJson converts an JsonbValue object to a JSON string.
 */
class JsonbToJson {
public:
    JsonbToJson() : os_(buffer_, OUT_BUF_SIZE) {}

    // get json string
    std::string to_json_string(const char* data, size_t size) {
        const JsonbDocument* pdoc;
        THROW_IF_ERROR(doris::JsonbDocument::checkAndCreateDocument(data, size, &pdoc));
        return to_json_string(pdoc->getValue());
    }

    std::string to_json_string(const JsonbValue* val) {
        os_.clear();
        os_.seekp(0);

        if (val) {
            intern_json(val);
        }

        os_.put(0);

        std::string json_string {os_.getBuffer()};
        return json_string;
    }

    static std::string jsonb_to_json_string(const char* data, size_t size) {
        JsonbToJson jsonb_to_json;
        return jsonb_to_json.to_json_string(data, size);
    }

private:
    // recursively convert JsonbValue
    void intern_json(const JsonbValue* val) {
        switch (val->type) {
        case JsonbType::T_Null: {
            os_.write("null", 4);
            break;
        }
        case JsonbType::T_True: {
            os_.write("true", 4);
            break;
        }
        case JsonbType::T_False: {
            os_.write("false", 5);
            break;
        }
        case JsonbType::T_Int8: {
            os_.write(val->unpack<JsonbInt8Val>()->val());
            break;
        }
        case JsonbType::T_Int16: {
            os_.write(val->unpack<JsonbInt16Val>()->val());
            break;
        }
        case JsonbType::T_Int32: {
            os_.write(val->unpack<JsonbInt32Val>()->val());
            break;
        }
        case JsonbType::T_Int64: {
            os_.write(val->unpack<JsonbInt64Val>()->val());
            break;
        }
        case JsonbType::T_Double: {
            os_.write(val->unpack<JsonbDoubleVal>()->val());
            break;
        }
        case JsonbType::T_Float: {
            os_.write(val->unpack<JsonbFloatVal>()->val());
            break;
        }
        case JsonbType::T_Int128: {
            os_.write(val->unpack<JsonbInt128Val>()->val());
            break;
        }
        case JsonbType::T_String: {
            // getBlobLen() is the stored payload length. length() would drop a
            // trailing NUL that belongs to the value, so it cannot be used here;
            // the writer stores strings by their exact length and never pads.
            string_to_json(val->unpack<JsonbStringVal>()->getBlob(),
                           val->unpack<JsonbStringVal>()->getBlobLen());
            break;
        }
        case JsonbType::T_Binary: {
            os_.write("\"<BINARY>", 9);
            os_.write(val->unpack<JsonbBinaryVal>()->getBlob(),
                      val->unpack<JsonbBinaryVal>()->getBlobLen());
            os_.write("<BINARY>\"", 9);
            break;
        }
        case JsonbType::T_Object: {
            object_to_json(val->unpack<ObjectVal>());
            break;
        }
        case JsonbType::T_Array: {
            array_to_json(val->unpack<ArrayVal>());
            break;
        }
        case JsonbType::T_Decimal32: {
            const auto* decimal_val = val->unpack<JsonbDecimal32>();
            decimal_to_json(Decimal32 {decimal_val->val()}, decimal_val->precision,
                            decimal_val->scale);
            break;
        }
        case JsonbType::T_Decimal64: {
            const auto* decimal_val = val->unpack<JsonbDecimal64>();
            decimal_to_json(Decimal64 {decimal_val->val()}, decimal_val->precision,
                            decimal_val->scale);
            break;
        }
        case JsonbType::T_Decimal128: {
            const auto* decimal_val = val->unpack<JsonbDecimal128>();
            decimal_to_json(Decimal128V3 {decimal_val->val()}, decimal_val->precision,
                            decimal_val->scale);
            break;
        }
        case JsonbType::T_Decimal256: {
            const auto* decimal_val = val->unpack<JsonbDecimal256>();
            decimal_to_json(Decimal256 {decimal_val->val()}, decimal_val->precision,
                            decimal_val->scale);
            break;
        }
        default:
            break;
        }
    }

    void string_to_json(const char* str, size_t len) {
        os_.put('"');
        if (nullptr == str) {
            os_.put('"');
            return;
        }
        char char_buffer[16];
        // A JSON string may legally contain U+0000, so the loop must be bounded by
        // the length only; the NUL itself is escaped as \u0000 by the default branch.
        for (const char* ptr = str; ptr != str + len; ++ptr) {
            if ((unsigned char)*ptr > 31 && *ptr != '\"' && *ptr != '\\') {
                os_.put(*ptr);
            } else {
                os_.put('\\');
                unsigned char token;
                switch (token = *ptr) {
                case '\\':
                    os_.put('\\');
                    break;
                case '\"':
                    os_.put('\"');
                    break;
                case '\b':
                    os_.put('b');
                    break;
                case '\f':
                    os_.put('f');
                    break;
                case '\n':
                    os_.put('n');
                    break;
                case '\r':
                    os_.put('r');
                    break;
                case '\t':
                    os_.put('t');
                    break;
                default: {
                    int char_num = snprintf(char_buffer, sizeof(char_buffer), "u%04x", token);
                    os_.write(char_buffer, char_num);
                    break;
                }
                }
            }
        }
        os_.put('"');
    }

    // convert object
    void object_to_json(const ObjectVal* val) {
        os_.put('{');

        auto iter = val->begin();
        auto iter_fence = val->end();

        while (iter < iter_fence) {
            // write key
            if (iter->klen()) {
                string_to_json(iter->getKeyStr(), iter->klen());
            } else {
                // NOTE: we use sMaxKeyId to represent an empty key. see jsonb_writer.h
                if (iter->getKeyId() == JsonbKeyValue::sMaxKeyId) {
                    string_to_json(nullptr, 0);
                } else {
                    os_.write(iter->getKeyId());
                }
            }
            os_.put(':');

            // convert value
            intern_json(iter->value());

            ++iter;
            if (iter != iter_fence) {
                os_.put(',');
            }
        }

        assert(iter == iter_fence);

        os_.put('}');
    }

    // convert array to json
    void array_to_json(const ArrayVal* val) {
        os_.put('[');

        auto iter = val->begin();
        auto iter_fence = val->end();

        while (iter != iter_fence) {
            // convert value
            intern_json((const JsonbValue*)iter);
            ++iter;
            if (iter != iter_fence) {
                os_.put(',');
            }
        }

        assert(iter == iter_fence);

        os_.put(']');
    }

    template <JsonbDecimalType T>
    void decimal_to_json(const T& value, const uint32_t precision, const uint32_t scale);

    JsonbOutStream os_;
    char buffer_[OUT_BUF_SIZE];
};

} // namespace doris

#endif // JSONB_JSONBUTIL_H

Coverage Report

Created: 2026-09-23 12:30

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 header file defines miscellaneous utility classes.
13		*
14		* @author Tian Xia <tianx@fb.com>
15		*
16		* this file is copied from
17		* https://github.com/facebook/mysql-5.6/blob/fb-mysql-5.6.35/fbson/FbsonUtil.h
18		* and modified by Doris
19		*/
20
21		#ifndef JSONB_JSONBUTIL_H
22		#define JSONB_JSONBUTIL_H
23
24		#include "common/status.h"
25		#include "util/jsonb_document.h"
26		#include "util/jsonb_stream.h"
27
28		namespace doris {
29
30	39.3k	#define OUT_BUF_SIZE 1024
31
32		/*
33		* JsonbToJson converts an JsonbValue object to a JSON string.
34		*/
35		class JsonbToJson {
36		public:
37	39.3k	JsonbToJson() : os_(buffer_, OUT_BUF_SIZE) {}
38
39		// get json string
40	37.8k	std::string to_json_string(const char* data, size_t size) {
41	37.8k	const JsonbDocument* pdoc;
42	37.8k	THROW_IF_ERROR(doris::JsonbDocument::checkAndCreateDocument(data, size, &pdoc));
43	37.8k	return to_json_string(pdoc->getValue());
44	37.8k	}
45
46	39.3k	std::string to_json_string(const JsonbValue* val) {
47	39.3k	os_.clear();
48	39.3k	os_.seekp(0);
49
50	39.3k	if (val) {
51	39.3k	intern_json(val);
52	39.3k	}
53
54	39.3k	os_.put(0);
55
56	39.3k	std::string json_string {os_.getBuffer()};
57	39.3k	return json_string;
58	39.3k	}
59
60	37.8k	static std::string jsonb_to_json_string(const char* data, size_t size) {
61	37.8k	JsonbToJson jsonb_to_json;
62	37.8k	return jsonb_to_json.to_json_string(data, size);
63	37.8k	}
64
65		private:
66		// recursively convert JsonbValue
67	111k	void intern_json(const JsonbValue* val) {
68	111k	switch (val->type) {
69	3.21k	case JsonbType::T_Null: {
70	3.21k	os_.write("null", 4);
71	3.21k	break;
72	0	}
73	3.20k	case JsonbType::T_True: {
74	3.20k	os_.write("true", 4);
75	3.20k	break;
76	0	}
77	3.06k	case JsonbType::T_False: {
78	3.06k	os_.write("false", 5);
79	3.06k	break;
80	0	}
81	12.6k	case JsonbType::T_Int8: {
82	12.6k	os_.write(val->unpack<JsonbInt8Val>()->val());
83	12.6k	break;
84	0	}
85	13.0k	case JsonbType::T_Int16: {
86	13.0k	os_.write(val->unpack<JsonbInt16Val>()->val());
87	13.0k	break;
88	0	}
89	1.96k	case JsonbType::T_Int32: {
90	1.96k	os_.write(val->unpack<JsonbInt32Val>()->val());
91	1.96k	break;
92	0	}
93	2.58k	case JsonbType::T_Int64: {
94	2.58k	os_.write(val->unpack<JsonbInt64Val>()->val());
95	2.58k	break;
96	0	}
97	6.74k	case JsonbType::T_Double: {
98	6.74k	os_.write(val->unpack<JsonbDoubleVal>()->val());
99	6.74k	break;
100	0	}
101	2	case JsonbType::T_Float: {
102	2	os_.write(val->unpack<JsonbFloatVal>()->val());
103	2	break;
104	0	}
105	2.07k	case JsonbType::T_Int128: {
106	2.07k	os_.write(val->unpack<JsonbInt128Val>()->val());
107	2.07k	break;
108	0	}
109	31.4k	case JsonbType::T_String: {
110		// getBlobLen() is the stored payload length. length() would drop a
111		// trailing NUL that belongs to the value, so it cannot be used here;
112		// the writer stores strings by their exact length and never pads.
113	31.4k	string_to_json(val->unpack<JsonbStringVal>()->getBlob(),
114	31.4k	val->unpack<JsonbStringVal>()->getBlobLen());
115	31.4k	break;
116	0	}
117	0	case JsonbType::T_Binary: {
118	0	os_.write("\"<BINARY>", 9);
119	0	os_.write(val->unpack<JsonbBinaryVal>()->getBlob(),
120	0	val->unpack<JsonbBinaryVal>()->getBlobLen());
121	0	os_.write("<BINARY>\"", 9);
122	0	break;
123	0	}
124	19.3k	case JsonbType::T_Object: {
125	19.3k	object_to_json(val->unpack<ObjectVal>());
126	19.3k	break;
127	0	}
128	11.6k	case JsonbType::T_Array: {
129	11.6k	array_to_json(val->unpack<ArrayVal>());
130	11.6k	break;
131	0	}
132	30	case JsonbType::T_Decimal32: {
133	30	const auto* decimal_val = val->unpack<JsonbDecimal32>();
134	30	decimal_to_json(Decimal32 {decimal_val->val()}, decimal_val->precision,
135	30	decimal_val->scale);
136	30	break;
137	0	}
138	31	case JsonbType::T_Decimal64: {
139	31	const auto* decimal_val = val->unpack<JsonbDecimal64>();
140	31	decimal_to_json(Decimal64 {decimal_val->val()}, decimal_val->precision,
141	31	decimal_val->scale);
142	31	break;
143	0	}
144	40	case JsonbType::T_Decimal128: {
145	40	const auto* decimal_val = val->unpack<JsonbDecimal128>();
146	40	decimal_to_json(Decimal128V3 {decimal_val->val()}, decimal_val->precision,
147	40	decimal_val->scale);
148	40	break;
149	0	}
150	1	case JsonbType::T_Decimal256: {
151	1	const auto* decimal_val = val->unpack<JsonbDecimal256>();
152	1	decimal_to_json(Decimal256 {decimal_val->val()}, decimal_val->precision,
153	1	decimal_val->scale);
154	1	break;
155	0	}
156	0	default:
157	0	break;
158	111k	}
159	111k	}
160
161	1.78k	void string_to_json(const char* str, size_t len) {
162	1.78k	os_.put('"');
163	1.78k	if (nullptr == str) {
164	3	os_.put('"');
165	3	return;
166	3	}
167	1.77k	char char_buffer[16];
168		// A JSON string may legally contain U+0000, so the loop must be bounded by
169		// the length only; the NUL itself is escaped as \u0000 by the default branch.
170	2.05M	for (const char* ptr = str; ptr != str + len; ++ptr) {
171	2.05M	if ((unsigned char)ptr > 31 && ptr != '\"' && *ptr != '\\') {
172	2.05M	os_.put(*ptr);
173	2.05M	} else {
174	13	os_.put('\\');
175	13	unsigned char token;
176	13	switch (token = *ptr) {
177	0	case '\\':
178	0	os_.put('\\');
179	0	break;
180	1	case '\"':
181	1	os_.put('\"');
182	1	break;
183	1	case '\b':
184	1	os_.put('b');
185	1	break;
186	1	case '\f':
187	1	os_.put('f');
188	1	break;
189	1	case '\n':
190	1	os_.put('n');
191	1	break;
192	1	case '\r':
193	1	os_.put('r');
194	1	break;
195	1	case '\t':
196	1	os_.put('t');
197	1	break;
198	7	default: {
199	7	int char_num = snprintf(char_buffer, sizeof(char_buffer), "u%04x", token);
200	7	os_.write(char_buffer, char_num);
201	7	break;
202	0	}
203	13	}
204	13	}
205	2.05M	}
206	1.77k	os_.put('"');
207	1.77k	}
208
209		// convert object
210	19.3k	void object_to_json(const ObjectVal* val) {
211	19.3k	os_.put('{');
212
213	19.3k	auto iter = val->begin();
214	19.3k	auto iter_fence = val->end();
215
216	57.5k	while (iter < iter_fence) {
217		// write key
218	38.1k	if (iter->klen()) {
219	35.9k	string_to_json(iter->getKeyStr(), iter->klen());
220	35.9k	} else {
221		// NOTE: we use sMaxKeyId to represent an empty key. see jsonb_writer.h
222	2.20k	if (iter->getKeyId() == JsonbKeyValue::sMaxKeyId) {
223	2.20k	string_to_json(nullptr, 0);
224	2.20k	} else {
225	0	os_.write(iter->getKeyId());
226	0	}
227	2.20k	}
228	38.1k	os_.put(':');
229
230		// convert value
231	38.1k	intern_json(iter->value());
232
233	38.1k	++iter;
234	38.1k	if (iter != iter_fence) {
235	20.4k	os_.put(',');
236	20.4k	}
237	38.1k	}
238
239	19.3k	assert(iter == iter_fence);
240
241	19.3k	os_.put('}');
242	19.3k	}
243
244		// convert array to json
245	11.6k	void array_to_json(const ArrayVal* val) {
246	11.6k	os_.put('[');
247
248	11.6k	auto iter = val->begin();
249	11.6k	auto iter_fence = val->end();
250
251	45.1k	while (iter != iter_fence) {
252		// convert value
253	33.5k	intern_json((const JsonbValue*)iter);
254	33.5k	++iter;
255	33.5k	if (iter != iter_fence) {
256	23.3k	os_.put(',');
257	23.3k	}
258	33.5k	}
259
260	11.6k	assert(iter == iter_fence);
261
262	11.6k	os_.put(']');
263	11.6k	}
264
265		template <JsonbDecimalType T>
266		void decimal_to_json(const T& value, const uint32_t precision, const uint32_t scale);
267
268		JsonbOutStream os_;
269		char buffer_[OUT_BUF_SIZE];
270		};
271
272		} // namespace doris
273
274		#endif // JSONB_JSONBUTIL_H