| 1 | #pragma once |
| 2 | |
| 3 | #include "config_functions.h" |
| 4 | #if USE_SIMDJSON |
| 5 | |
| 6 | #include <common/StringRef.h> |
| 7 | #include <Common/Exception.h> |
| 8 | #include <Core/Types.h> |
| 9 | |
| 10 | #include <simdjson/jsonparser.h> |
| 11 | |
| 12 | |
| 13 | namespace DB |
| 14 | { |
| 15 | namespace ErrorCodes |
| 16 | { |
| 17 | extern const int CANNOT_ALLOCATE_MEMORY; |
| 18 | } |
| 19 | |
| 20 | /// This class can be used as an argument for the template class FunctionJSON. |
| 21 | /// It provides ability to parse JSONs using simdjson library. |
| 22 | struct SimdJSONParser |
| 23 | { |
| 24 | static constexpr bool need_preallocate = true; |
| 25 | |
| 26 | void preallocate(size_t max_size) |
| 27 | { |
| 28 | if (!pj.allocate_capacity(max_size)) |
| 29 | throw Exception{"Can not allocate memory for " + std::to_string(max_size) + " units when parsing JSON" , |
| 30 | ErrorCodes::CANNOT_ALLOCATE_MEMORY}; |
| 31 | } |
| 32 | |
| 33 | bool parse(const StringRef & json) { return !json_parse(json.data, json.size, pj); } |
| 34 | |
| 35 | using Iterator = simdjson::ParsedJson::Iterator; |
| 36 | Iterator getRoot() { return Iterator{pj}; } |
| 37 | |
| 38 | static bool isInt64(const Iterator & it) { return it.is_integer(); } |
| 39 | static bool isUInt64(const Iterator &) { return false; /* See https://github.com/lemire/simdjson/issues/68 */ } |
| 40 | static bool isDouble(const Iterator & it) { return it.is_double(); } |
| 41 | static bool isString(const Iterator & it) { return it.is_string(); } |
| 42 | static bool isArray(const Iterator & it) { return it.is_array(); } |
| 43 | static bool isObject(const Iterator & it) { return it.is_object(); } |
| 44 | static bool isBool(const Iterator & it) { return it.get_type() == 't' || it.get_type() == 'f'; } |
| 45 | static bool isNull(const Iterator & it) { return it.is_null(); } |
| 46 | |
| 47 | static Int64 getInt64(const Iterator & it) { return it.get_integer(); } |
| 48 | static UInt64 getUInt64(const Iterator &) { return 0; /* isUInt64() never returns true */ } |
| 49 | static double getDouble(const Iterator & it) { return it.get_double(); } |
| 50 | static bool getBool(const Iterator & it) { return it.get_type() == 't'; } |
| 51 | static StringRef getString(const Iterator & it) { return StringRef{it.get_string(), it.get_string_length()}; } |
| 52 | |
| 53 | static size_t sizeOfArray(const Iterator & it) |
| 54 | { |
| 55 | size_t size = 0; |
| 56 | Iterator it2 = it; |
| 57 | if (it2.down()) |
| 58 | { |
| 59 | do |
| 60 | ++size; |
| 61 | while (it2.next()); |
| 62 | } |
| 63 | return size; |
| 64 | } |
| 65 | |
| 66 | static bool firstArrayElement(Iterator & it) { return it.down(); } |
| 67 | |
| 68 | static bool arrayElementByIndex(Iterator & it, size_t index) |
| 69 | { |
| 70 | if (!it.down()) |
| 71 | return false; |
| 72 | while (index--) |
| 73 | if (!it.next()) |
| 74 | return false; |
| 75 | return true; |
| 76 | } |
| 77 | |
| 78 | static bool nextArrayElement(Iterator & it) { return it.next(); } |
| 79 | |
| 80 | static size_t sizeOfObject(const Iterator & it) |
| 81 | { |
| 82 | size_t size = 0; |
| 83 | Iterator it2 = it; |
| 84 | if (it2.down()) |
| 85 | { |
| 86 | do |
| 87 | ++size; |
| 88 | while (it2.next() && it2.next()); //-V501 |
| 89 | } |
| 90 | return size; |
| 91 | } |
| 92 | |
| 93 | static bool firstObjectMember(Iterator & it) { return it.down() && it.next(); } |
| 94 | |
| 95 | static bool firstObjectMember(Iterator & it, StringRef & first_key) |
| 96 | { |
| 97 | if (!it.down()) |
| 98 | return false; |
| 99 | first_key.data = it.get_string(); |
| 100 | first_key.size = it.get_string_length(); |
| 101 | return it.next(); |
| 102 | } |
| 103 | |
| 104 | static bool objectMemberByIndex(Iterator & it, size_t index) |
| 105 | { |
| 106 | if (!it.down()) |
| 107 | return false; |
| 108 | while (index--) |
| 109 | if (!it.next() || !it.next()) //-V501 |
| 110 | return false; |
| 111 | return it.next(); |
| 112 | } |
| 113 | |
| 114 | static bool objectMemberByName(Iterator & it, const StringRef & name) { return it.move_to_key(name.data); } |
| 115 | static bool nextObjectMember(Iterator & it) { return it.next() && it.next(); } //-V501 |
| 116 | |
| 117 | static bool nextObjectMember(Iterator & it, StringRef & next_key) |
| 118 | { |
| 119 | if (!it.next()) |
| 120 | return false; |
| 121 | next_key.data = it.get_string(); |
| 122 | next_key.size = it.get_string_length(); |
| 123 | return it.next(); |
| 124 | } |
| 125 | |
| 126 | static bool isObjectMember(const Iterator & it) { return it.get_scope_type() == '{'; } |
| 127 | |
| 128 | static StringRef getKey(const Iterator & it) |
| 129 | { |
| 130 | Iterator it2 = it; |
| 131 | it2.prev(); |
| 132 | return StringRef{it2.get_string(), it2.get_string_length()}; |
| 133 | } |
| 134 | |
| 135 | private: |
| 136 | simdjson::ParsedJson pj; |
| 137 | }; |
| 138 | |
| 139 | } |
| 140 | #endif |
| 141 | |