schema

package
v0.0.0-...-f09cf9b Latest Latest
Warning

This package is not in the latest version of its module.

Go to latest
Published: Jan 13, 2025 License: BSD-3-Clause Imports: 11 Imported by: 0

Documentation

Index

Constants

View Source
const (
	ErrorNoSuchClass    string = "no such class with name '%s' found in the schema. Check your schema files for which classes are available"
	ErrorNoSuchProperty string = "" /* 135-byte string literal not displayed */
	ErrorNoSuchDatatype string = "given value-DataType does not exist."
)
View Source
const (
	ClassNameRegexCore = `[A-Z][_0-9A-Za-z]{0,254}`
	ShardNameRegexCore = `[A-Za-z0-9\-\_]{1,64}`

	PropertyNameRegex = `[_A-Za-z][_0-9A-Za-z]{0,230}`
	// Nested properties names are not used to build directory names (yet),
	// no max length restriction is imposed
	NestedPropertyNameRegex = `[_A-Za-z][_0-9A-Za-z]*`
	// Target vector names must be GraphQL compliant names no longer then 230 characters
	TargetVectorNameMaxLength = 230
	TargetVectorNameRegex     = `[_A-Za-z][_0-9A-Za-z]{0,229}`
)

Variables

View Source
var DeprecatedPrimitiveDataTypes []DataType = []DataType{

	DataTypeString, DataTypeStringArray,
}
View Source
var ErrRefToNonexistentClass = errors.New("reference property to nonexistent class")

Functions

func ActivityStatus

func ActivityStatus(status string) string

func AutoTenantActivationEnabled

func AutoTenantActivationEnabled(class *models.Class) bool

func AutoTenantCreationEnabled

func AutoTenantCreationEnabled(class *models.Class) bool

func CollectionToClass

func CollectionToClass(c Collection) models.Class

CollectionToClass returns a models.Class from c. If the original models.Class from which c was created had nil pointers they will be replaced with default initialized structs in models.Class.

func DedupProperties

func DedupProperties(oldProps, newProps []*models.Property) []*models.Property

DedupProperties removes from newProps slice properties already present in oldProps slice. If property of nested type (object/object[]) is present in both slices, diff is calculated to contain only those nested properties that are missing from old property model

func DiffRecursivelyNestedProperties

func DiffRecursivelyNestedProperties(nestPropsOld, nestPropsNew []*models.NestedProperty,
) []*models.NestedProperty

Determines diff between nestPropsNew and nestPropsOld slices

func GetClassByName

func GetClassByName(s *models.Schema, className string) (*models.Class, error)

GetClassByName returns the class by its name

func GetNestedPropertyByName

func GetNestedPropertyByName[P PropertyInterface](p P, propName string) (*models.NestedProperty, error)

func GetPropertyByName

func GetPropertyByName(c *models.Class, propName string) (*models.Property, error)

GetPropertyByName returns the class by its name

func GetPropertyNamesFromClass

func GetPropertyNamesFromClass(class *models.Class, includeRef bool) []string

func IsArrayDataType

func IsArrayDataType(dt []string) bool

func IsBlobDataType

func IsBlobDataType(dt []string) bool

func IsNested

func IsNested(dataType DataType) bool

func IsPropertyLength

func IsPropertyLength(propName string, offset int) (string, bool)

IsPropertyLength returns if a string is a filters for property length. They have the form len(*PROPNAME*)

func IsRefDataType

func IsRefDataType(dt []string) bool

func IsValidValueDataType

func IsValidValueDataType(dt string) bool

IsValidValueDataType checks whether the given string is a valid data type

func LowercaseAllPropertyNames

func LowercaseAllPropertyNames(props []*models.Property) []*models.Property

func LowercaseFirstLetter

func LowercaseFirstLetter(name string) string

func LowercaseFirstLetterOfStrings

func LowercaseFirstLetterOfStrings(in []string) []string

func MergeRecursivelyNestedProperties

func MergeRecursivelyNestedProperties(nestPropsOld, nestPropsNew []*models.NestedProperty,
) ([]*models.NestedProperty, bool)

Merges nestPropsNew with nestPropsOld Returns new slice without changing input ones and bool indicating whether merged slice is different than the old one

func MultiTenancyEnabled

func MultiTenancyEnabled(class *models.Class) bool

func NestedPropertyToModel

func NestedPropertyToModel(n NestedProperty) models.NestedProperty

NestedPropertyToModel returns a models.NestedProperty from n. If the original models.NestedProperty from which n was created had nil pointers they will be replaced with default initialized structs in NestedProperty.

func PropertyToModel

func PropertyToModel(p Property) models.Property

PropertyToModel returns a models.Property from p. If the original models.Property from which n was created had nil pointers they will be replaced with default initialized structs in Property.

func ShardingConfigToModel

func ShardingConfigToModel(s ShardingConfig) interface{}

ShardingConfigToModel returns an interface{} containing a sharding.Config from s.

func UppercaseClassName

func UppercaseClassName(name string) string

func UppercaseClassesNames

func UppercaseClassesNames(names ...string) []string

func ValidateNestedPropertyName

func ValidateNestedPropertyName(name, prefix string) error

ValidateNestedPropertyName validates that this string is a valid nested property name

func ValidateReservedPropertyName

func ValidateReservedPropertyName(name string) error

ValidateReservedPropertyName validates that a string is not a reserved property name

Types

type BM25Config

type BM25Config struct {
	K1 float64
	B  float64
}

type ClassAndProperty

type ClassAndProperty struct {
	ClassName    ClassName
	PropertyName PropertyName
}

type ClassName

type ClassName string

Newtype to denote that this string is used as a Class name

func AssertValidClassName

func AssertValidClassName(name string) ClassName

AssertValidClassName assert that this string is a valid class name or panics and should therefore most likely not be used

func ValidateClassName

func ValidateClassName(name string) (ClassName, error)

ValidateClassName validates that this string is a valid class name (format wise)

func (ClassName) String

func (c ClassName) String() string

type Collection

type Collection struct {
	Name        string `json:"name,omitempty"`
	Description string `json:"description,omitempty"`

	// inverted index config
	InvertedIndexConfig InvertedIndexConfig `json:"invertedIndex,omitempty"`

	// TODO-RAFT START
	// Can we also get rid of the interface{} in the value side ?
	// Configuration specific to modules this Weaviate instance has installed
	ModuleConfig map[string]interface{} `json:"moduleConfig,omitempty"`

	// multi tenancy config
	MultiTenancyConfig MultiTenancyConfig `json:"multiTenancyConfig,omitempty"`

	// The properties of the class.
	Properties []Property `json:"properties"`

	// replication config
	ReplicationConfig ReplicationConfig `json:"replicationConfig,omitempty"`

	// Manage how the index should be sharded and distributed in the cluster
	ShardingConfig ShardingConfig `json:"shardingConfig,omitempty"`

	// VectorIndexType which vector index to use
	VectorIndexType VectorIndexType `json:"vectorIndexType,omitempty"`

	// VectorIndexConfig underlying implementation depends on VectorIndexType
	VectorIndexConfig config.VectorIndexConfig `json:"vectorIndexConfig,omitempty"`

	// Specify how the vectors for this class should be determined. The options are either 'none' - this means you have to import a vector
	// with each object yourself - or the name of a module that provides vectorization capabilities, such as 'text2vec-contextionary'. If
	// left empty, it will use the globally configured default which can itself either be 'none' or a specific module.
	Vectorizer string `json:"vectorizer,omitempty"`
}

func CollectionFromClass

func CollectionFromClass(m models.Class) (Collection, error)

CollectionFromClass returns a Collection copied from m.

type DataType

type DataType string
const (
	// DataTypeCRef The data type is a cross-reference, it is starting with a capital letter
	DataTypeCRef DataType = "cref"
	// DataTypeText The data type is a value of type string
	DataTypeText DataType = "text"
	// DataTypeInt The data type is a value of type int
	DataTypeInt DataType = "int"
	// DataTypeNumber The data type is a value of type number/float
	DataTypeNumber DataType = "number"
	// DataTypeBoolean The data type is a value of type boolean
	DataTypeBoolean DataType = "boolean"
	// DataTypeDate The data type is a value of type date
	DataTypeDate DataType = "date"
	// DataTypeGeoCoordinates is used to represent geo coordinates, i.e. latitude
	// and longitude pairs of locations on earth
	DataTypeGeoCoordinates DataType = "geoCoordinates"
	// DataTypePhoneNumber represents a parsed/to-be-parsed phone number
	DataTypePhoneNumber DataType = "phoneNumber"
	// DataTypeBlob represents a base64 encoded data
	DataTypeBlob DataType = "blob"
	// DataTypeTextArray The data type is a value of type string array
	DataTypeTextArray DataType = "text[]"
	// DataTypeIntArray The data type is a value of type int array
	DataTypeIntArray DataType = "int[]"
	// DataTypeNumberArray The data type is a value of type number/float array
	DataTypeNumberArray DataType = "number[]"
	// DataTypeBooleanArray The data type is a value of type boolean array
	DataTypeBooleanArray DataType = "boolean[]"
	// DataTypeDateArray The data type is a value of type date array
	DataTypeDateArray DataType = "date[]"
	// DataTypeUUID is a native UUID data type. It is stored in it's raw byte
	// representation and therefore takes up less space than storing a UUID as a
	// string
	DataTypeUUID DataType = "uuid"
	// DataTypeUUIDArray is the array version of DataTypeUUID
	DataTypeUUIDArray DataType = "uuid[]"

	DataTypeObject      DataType = "object"
	DataTypeObjectArray DataType = "object[]"

	// deprecated as of v1.19, replaced by DataTypeText + relevant tokenization setting
	// DataTypeString The data type is a value of type string
	DataTypeString DataType = "string"
	// deprecated as of v1.19, replaced by DataTypeTextArray + relevant tokenization setting
	// DataTypeArrayString The data type is a value of type string array
	DataTypeStringArray DataType = "string[]"
)

func AsNested

func AsNested(dataType []string) (DataType, bool)

func AsPrimitive

func AsPrimitive(dataType []string) (DataType, bool)

func GetNestedPropertyDataType

func GetNestedPropertyDataType[P PropertyInterface](p P, propertyName string) (*DataType, error)

func GetPropertyDataType

func GetPropertyDataType(class *models.Class, propertyName string) (*DataType, error)

GetPropertyDataType checks whether the given string is a valid data type

func GetValueDataTypeFromString

func GetValueDataTypeFromString(dt string) (*DataType, error)

GetValueDataTypeFromString checks whether the given string is a valid data type

func IsArrayType

func IsArrayType(dt DataType) (DataType, bool)

func (DataType) AsName

func (dt DataType) AsName() string

func (DataType) PropString

func (dt DataType) PropString() []string

func (DataType) String

func (dt DataType) String() string

type InvertedIndexConfig

type InvertedIndexConfig struct {
	BM25                   BM25Config
	Stopwords              models.StopwordConfig
	CleanupIntervalSeconds uint64
	IndexTimestamps        bool
	IndexNullState         bool
	IndexPropertyLength    bool
}

type MultiTenancyConfig

type MultiTenancyConfig struct {
	Enabled bool `json:"enabled"`
}

type NestedProperty

type NestedProperty struct {
	// name
	Name string `json:"name,omitempty"`
	// description
	Description string `json:"description,omitempty"`
	// data type
	DataType []string `json:"data_type"`

	// index filterable
	IndexFilterable bool `json:"index_filterable,omitempty"`

	// index searchable
	IndexSearchable bool `json:"index_searchable,omitempty"`

	// index range filters
	IndexRangeFilters bool `json:"index_range_filters,omitempty"`

	// nested properties
	NestedProperties []NestedProperty `json:"nested_properties,omitempty"`

	// tokenization
	// Enum: [word lowercase whitespace field]
	Tokenization string `json:"tokenization,omitempty"`
}

func NestedPropertyFromModel

func NestedPropertyFromModel(m models.NestedProperty) NestedProperty

NestedPropertyFromModel returns a NestedProperty copied from m.

type Property

type Property struct {
	// Name of the property as URI relative to the schema URL.
	Name string `json:"name,omitempty"`

	// Description of the property.
	Description string `json:"description,omitempty"`

	// Can be a reference to another type when it starts with a capital (for example Person), otherwise "string" or "int".
	// TODO-RAFT: Can we make DataType a slice of interface where other type and native type implements it ?
	DataType []string `json:"data_type"`

	// Optional. Should this property be indexed in the inverted index. Defaults to true. If you choose false, you will not be able to use this property in where filters. This property has no affect on vectorization decisions done by modules
	IndexFilterable bool `json:"indexFilterable,omitempty"`

	// Optional. Should this property be indexed in the inverted index. Defaults to true. If you choose false, you will not be able to use this property in where filters, bm25 or hybrid search. This property has no affect on vectorization decisions done by modules (deprecated as of v1.19; use indexFilterable or/and indexSearchable instead)
	IndexInverted bool `json:"indexInverted,omitempty"`

	// Optional. Should this property be indexed in the inverted index. Defaults to true. Applicable only to properties of data type text and text[]. If you choose false, you will not be able to use this property in bm25 or hybrid search. This property has no affect on vectorization decisions done by modules
	IndexSearchable bool `json:"indexSearchable,omitempty"`

	// Optional. Should this property be indexed in the inverted index. Defaults to false. Provides better performance for range queries compared to filterable index in large datasets. Applicable only to properties of data type int, number, date."
	IndexRangeFilters bool `json:"indexRangeFilters,omitempty"`

	// Configuration specific to modules this Weaviate instance has installed
	ModuleConfig map[string]interface{} `json:"moduleConfig,omitempty"`

	// The properties of the nested object(s). Applies to object and object[] data types.
	NestedProperties []NestedProperty `json:"nestedProperties,omitempty"`

	// Determines tokenization of the property as separate words or whole field. Optional. Applies to text and text[] data types. Allowed values are `word` (default; splits on any non-alphanumerical, lowercases), `lowercase` (splits on white spaces, lowercases), `whitespace` (splits on white spaces), `field` (trims). Not supported for remaining data types
	// Enum: [word lowercase whitespace field]
	Tokenization string `json:"tokenization,omitempty"`
}

func PropertyFromModel

func PropertyFromModel(m models.Property) Property

PropertyFromModel returns a Property copied from m.

type PropertyDataType

type PropertyDataType interface {
	Kind() PropertyKind
	IsPrimitive() bool
	AsPrimitive() DataType
	IsReference() bool
	Classes() []ClassName
	ContainsClass(name ClassName) bool
	IsNested() bool
	AsNested() DataType
}

func FindPropertyDataTypeWithRefs

func FindPropertyDataTypeWithRefs(authorizedGetClass func(string) *models.Class, dataType []string, relaxCrossRefValidation bool, beloningToClass ClassName) (PropertyDataType, error)

FindPropertyDataTypeWithRefs is a no auth wrapper for FindPropertyDataTypeWithRefsAndAuth

func FindPropertyDataTypeWithRefsAndAuth

func FindPropertyDataTypeWithRefsAndAuth(authorizedGetClass func(string) (*models.Class, error), dataType []string, relaxCrossRefValidation bool, beloningToClass ClassName) (PropertyDataType, error)

FindPropertyDataTypeWithRefsAndAuth Based on the schema, return a valid description of the defined datatype If relaxCrossRefValidation is set, there is no check if the referenced class exists in the schema. This can be helpful in scenarios, such as restoring from a backup where we have no guarantee over the order of class creation. If belongingToClass is set and equal to referenced class, check whether class exists in the schema is skipped. This is done to allow creating class schema with properties referencing to itself. Previously such properties had to be created separately only after creation of class schema

type PropertyInterface

type PropertyInterface interface {
	GetName() string
	GetNestedProperties() []*models.NestedProperty
}

type PropertyKind

type PropertyKind int
const (
	PropertyKindPrimitive PropertyKind = 1
	PropertyKindRef       PropertyKind = 2
	PropertyKindNested    PropertyKind = 3
)

type PropertyName

type PropertyName string

Newtype to denote that this string is used as a Property name

func AssertValidPropertyName

func AssertValidPropertyName(name string) PropertyName

AssertValidPropertyName asserts that this string is a valid property name or panics and should therefore most likely never be used.

func ValidatePropertyName

func ValidatePropertyName(name string) (PropertyName, error)

ValidatePropertyName validates that this string is a valid property name

func (PropertyName) String

func (p PropertyName) String() string

type ReplicationConfig

type ReplicationConfig struct {
	// Factor represent replication factor
	Factor int64 `json:"factor,omitempty"`
}

type Schema

type Schema struct {
	Objects *models.Schema
}

Describes the schema that is used in Weaviate.

func Empty

func Empty() Schema

func (*Schema) FindClassByName

func (s *Schema) FindClassByName(className ClassName) *models.Class

FindClassByName will find either a Thing or Class by name.

func (*Schema) FindPropertyDataType

func (s *Schema) FindPropertyDataType(dataType []string) (PropertyDataType, error)

Based on the schema, return a valid description of the defined datatype

Note that this function will error if referenced classes do not exist. If you don't want such validation, use [Schema.FindPropertyDataTypeRelaxedRefs] instead and set relax to true

func (*Schema) GetClass

func (s *Schema) GetClass(className string) *models.Class

func (*Schema) GetProperty

func (s *Schema) GetProperty(className ClassName, propName PropertyName) (*models.Property, error)

func (*Schema) GetPropsOfType

func (s *Schema) GetPropsOfType(propType string) []ClassAndProperty

func (*Schema) SemanticSchemaFor

func (s *Schema) SemanticSchemaFor() *models.Schema

Return one of the semantic schema's

type ShardingConfig

type ShardingConfig struct {
	VirtualPerPhysical  int    `json:"virtualPerPhysical"`
	DesiredCount        int    `json:"desiredCount"`
	ActualCount         int    `json:"actualCount"`
	DesiredVirtualCount int    `json:"desiredVirtualCount"`
	ActualVirtualCount  int    `json:"actualVirtualCount"`
	Key                 string `json:"key"`
	Strategy            string `json:"strategy"`
	Function            string `json:"function"`
}

func ShardingConfigFromModel

func ShardingConfigFromModel(m interface{}) ShardingConfig

ShardingConfigFromModel returns a ShardingConfig copied from m. If m isn't a sharding.Config underneath the interface{}, a default initialized ShardingConfig will be returned.

type VectorIndexType

type VectorIndexType int
const (
	// VectorIndexTypeEmpty is used when we parse an unexpected index type
	VectorIndexTypeEmpty VectorIndexType = iota
	VectorIndexTypeHNSW
	VectorIndexTypeFlat
)

Directories

Path Synopsis

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL