Felipe97/llama-cpp-compiled
01.1k
1#pragma once2 3#include "json.h"4 5#include <cstdint>6#include <initializer_list>7#include <map>8#include <memory>9#include <string>10#include <vector>11 12// JSON schema, covering the subset that json_schema_to_grammar() can convert.13 14struct common_chat_schema {15 enum node_kind {16 KIND_ANY,17 KIND_REF,18 KIND_ANY_OF,19 KIND_ALL_OF,20 KIND_CONST,21 KIND_ENUM,22 KIND_NULL,23 KIND_BOOLEAN,24 KIND_NUMBER,25 KIND_INTEGER,26 KIND_STRING,27 KIND_ARRAY,28 KIND_TUPLE,29 KIND_OBJECT,30 };31 32 enum value_type {33 TYPE_NULL,34 TYPE_BOOLEAN,35 TYPE_NUMBER,36 TYPE_INTEGER,37 TYPE_STRING,38 TYPE_ARRAY,39 TYPE_OBJECT,40 };41 42 enum string_format {43 FORMAT_NONE,44 FORMAT_UUID, // uuid, uuid1 .. uuid545 FORMAT_DATE,46 FORMAT_TIME,47 FORMAT_DATE_TIME,48 };49 50 class type_set {51 uint32_t mask_ = 0;52 53 public:54 type_set() = default;55 type_set(std::initializer_list<value_type> types) {56 for (auto type : types) {57 add(type);58 }59 }60 61 static type_set all() {62 return { TYPE_NULL, TYPE_BOOLEAN, TYPE_NUMBER, TYPE_INTEGER, TYPE_STRING, TYPE_ARRAY, TYPE_OBJECT };63 }64 65 void add(value_type type) { mask_ |= 1u << type; }66 67 bool has(value_type type) const { return (mask_ & (1u << type)) != 0; }68 bool is_only(value_type type) const { return mask_ == (1u << type); }69 bool empty() const { return mask_ == 0; }70 71 type_set & operator|=(const type_set & other) { mask_ |= other.mask_; return *this; }72 type_set & operator&=(const type_set & other) { mask_ &= other.mask_; return *this; }73 74 bool operator==(const type_set & other) const { return mask_ == other.mask_; }75 bool operator!=(const type_set & other) const { return mask_ != other.mask_; }76 };77 78 virtual ~common_chat_schema() = default;79 virtual node_kind kind() const = 0;80 81 type_set value_types() const;82 83 // Whether a value matching the schema may be a string, through any branch of it.84 bool may_be_string() const;85 86 static const char * kind_name(node_kind kind);87 static const char * type_name(value_type type);88};89 90using common_chat_schema_ptr = std::unique_ptr<common_chat_schema>;91 92struct common_chat_schema_any : common_chat_schema {93 node_kind kind() const override { return KIND_ANY; }94};95 96// {"$ref": "#/..."}, only references into the same document are supported97struct common_chat_schema_ref : common_chat_schema {98 std::string ref;99 const common_chat_schema * target = nullptr; // owned by common_chat_schema_document::refs100 101 explicit common_chat_schema_ref(std::string ref) : ref(std::move(ref)) {}102 103 node_kind kind() const override { return KIND_REF; }104};105 106// oneOf / anyOf, or a "type" array expanded to one alternative per type107struct common_chat_schema_any_of : common_chat_schema {108 std::vector<common_chat_schema_ptr> children;109 110 node_kind kind() const override { return KIND_ANY_OF; }111};112 113struct common_chat_schema_all_of : common_chat_schema {114 std::vector<common_chat_schema_ptr> children;115 116 node_kind kind() const override { return KIND_ALL_OF; }117};118 119struct common_chat_schema_const : common_chat_schema {120 common_json value;121 122 explicit common_chat_schema_const(common_json value) : value(std::move(value)) {}123 124 node_kind kind() const override { return KIND_CONST; }125};126 127struct common_chat_schema_enum : common_chat_schema {128 std::vector<common_json> values;129 130 node_kind kind() const override { return KIND_ENUM; }131};132 133struct common_chat_schema_null : common_chat_schema {134 node_kind kind() const override { return KIND_NULL; }135};136 137struct common_chat_schema_boolean : common_chat_schema {138 node_kind kind() const override { return KIND_BOOLEAN; }139};140 141struct common_chat_schema_number : common_chat_schema {142 node_kind kind() const override { return KIND_NUMBER; }143};144 145// bounds are inclusive, exclusiveMinimum / exclusiveMaximum are folded in146struct common_chat_schema_integer : common_chat_schema {147 int64_t minimum = INT64_MIN; // INT64_MIN for unbounded148 int64_t maximum = INT64_MAX; // INT64_MAX for unbounded149 150 node_kind kind() const override { return KIND_INTEGER; }151};152 153struct common_chat_schema_string : common_chat_schema {154 std::string pattern; // empty when absent155 string_format format = FORMAT_NONE;156 int min_length = 0;157 int max_length = -1; // -1 for unbounded158 159 node_kind kind() const override { return KIND_STRING; }160};161 162struct common_chat_schema_array : common_chat_schema {163 common_chat_schema_ptr items; // a common_chat_schema_any when "items" is absent164 int min_items = 0;165 int max_items = -1; // -1 for unbounded166 167 node_kind kind() const override { return KIND_ARRAY; }168};169 170struct common_chat_schema_tuple : common_chat_schema {171 std::vector<common_chat_schema_ptr> items;172 173 node_kind kind() const override { return KIND_TUPLE; }174};175 176struct common_chat_schema_property {177 std::string name;178 common_chat_schema_ptr schema;179 bool required = false;180};181 182struct common_chat_schema_object : common_chat_schema {183 std::vector<common_chat_schema_property> properties; // in schema order184 common_chat_schema_ptr additional_properties; // null when not allowed185 186 node_kind kind() const override { return KIND_OBJECT; }187};188 189struct common_chat_schema_document {190 common_chat_schema_ptr root;191 std::map<std::string, common_chat_schema_ptr> refs;192};193 194// A document shared by the PEG parsers built from its nodes, which it keeps alive195using common_chat_schema_document_ptr = std::shared_ptr<const common_chat_schema_document>;196 197// Throws std::runtime_error when the schema falls outside the supported subset.198common_chat_schema_document common_chat_schema_from_json(const common_json & schema);199 