pg

package
v0.10.0 Latest Latest
Warning

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

 Go to latest Published: Oct 28, 2026 License: MIT Imports: 35 Imported by: 17

Documentation

Index

Constants

View Source 
const (
	DriverName = "pg"
)
View Source 
const (
	LargeNodeUpdateThreshold = 1_000_000
)

Variables

This section is empty.

Functions

func AfterPooledConnectionEstablished added in v0.5.4 

func AfterPooledConnectionEstablished(ctx context.Context, conn *pgx.Conn) error

func AfterPooledConnectionRelease added in v0.5.4 

func AfterPooledConnectionRelease(conn *pgx.Conn) bool

func IsPostgreSQLGraph 

func IsPostgreSQLGraph(db graph.Database) bool

func NewPool 

func NewPool(poolCfg *pgxpool.Config) (*pgxpool.Pool, error)

pgx pool config

func NewValueMapper 

func NewValueMapper(ctx context.Context, kindMapper KindMapper) graph.ValueMapper

func OptimizedTranslationEnabled added in v0.8.0 

func OptimizedTranslationEnabled() bool

OptimizedTranslationEnabled reports whether PostgreSQL Cypher translation uses optimization and the translation cache for newly started compilations.

func OptionSetQueryExecMode 

func OptionSetQueryExecMode(queryExecMode pgx.QueryExecMode) graph.TransactionOption

func SetOptimizedTranslation added in v0.8.0 

func SetOptimizedTranslation(enabled bool) bool

SetOptimizedTranslation enables or disables optimized PostgreSQL Cypher translation for the current process. It returns the previous setting.

Disabling optimization bypasses the translation cache and uses the baseline translation path for calls that begin after the setting changes. Existing cached entries remain available if optimization is enabled again.

Types

type Config 

type Config struct {
	Options            pgx.TxOptions
	QueryExecMode      pgx.QueryExecMode
	QueryResultFormats pgx.QueryResultFormats
	BatchWriteSize     int
}

type Driver 

type Driver struct {
	*SchemaManager
	// contains filtered or unexported fields
}

func NewDriver 

func NewDriver(graphQueryMemoryLimit size.Size, pool *pgxpool.Pool) *Driver

func NewDriverWithOptions added in v0.8.0 

func NewDriverWithOptions(graphQueryMemoryLimit size.Size, pool *pgxpool.Pool, options DriverOptions) *Driver

NewDriverWithOptions constructs a PostgreSQL driver with driver-wide options.

func (*Driver) AssertSchema 

func (s *Driver) AssertSchema(ctx context.Context, schema graph.Schema) error

func (*Driver) BatchOperation 

func (s *Driver) BatchOperation(ctx context.Context, batchDelegate graph.BatchDelegate, options ...graph.BatchOption) error

func (*Driver) Close 

func (s *Driver) Close(ctx context.Context) error

func (*Driver) DeleteNodesByKinds added in v0.7.0 

func (s *Driver) DeleteNodesByKinds(ctx context.Context, includeAny graph.Kinds, excludeAny graph.Kinds) error

DeleteNodesByKinds performs a server-side, set-based delete of nodes using the kind_ids GIN index instead of streaming node IDs through the application. A node is deleted when its kind_ids overlap includeAny (or, when includeAny is empty, for every node) and do not overlap excludeAny. Deleting nodes fires the delete_node_edges trigger, cascading the attached edge deletes.

includeAny is mapped to kind IDs tolerantly: include kinds that are not defined in the database map to no IDs and therefore match no nodes, so a request that targets only undefined kinds is a safe no-op rather than an accidental full delete. excludeAny is mapped fail-closed: if any exclude kind is undefined the delete is refused, because silently dropping an exclusion would widen the delete and could remove protected nodes (e.g. an unresolved MigrationData would turn a guarded wipe into an unguarded delete from node).

func (*Driver) DeleteRelationshipsByKinds added in v0.7.0 

func (s *Driver) DeleteRelationshipsByKinds(ctx context.Context, kinds graph.Kinds) error

DeleteRelationshipsByKinds performs a server-side, set-based delete of relationships whose kind_id matches any of the given kinds, using the edge_kind_index for kind filtering instead of streaming relationship IDs through the application.

kinds are mapped to kind IDs tolerantly: kinds that are not defined in the database map to no IDs. An empty kinds argument, or one that maps entirely to undefined kinds, deletes nothing rather than every relationship.

func (*Driver) FetchKinds 

func (s *Driver) FetchKinds(_ context.Context) (graph.Kinds, error)

func (*Driver) FetchSchema 

func (s *Driver) FetchSchema(ctx context.Context) (graph.Schema, error)

func (*Driver) KindMapper 

func (s *Driver) KindMapper() KindMapper

func (*Driver) OptimizeStorage added in v0.6.0 

func (s *Driver) OptimizeStorage(ctx context.Context) error

func (*Driver) RefreshKinds 

func (s *Driver) RefreshKinds(ctx context.Context) error

func (*Driver) Run 

func (s *Driver) Run(ctx context.Context, query string, parameters map[string]any) error

func (*Driver) SetBatchWriteSize 

func (s *Driver) SetBatchWriteSize(size int)

func (*Driver) SetDefaultGraph 

func (s *Driver) SetDefaultGraph(ctx context.Context, graphSchema graph.Graph) error

func (*Driver) SetWriteFlushSize 

func (s *Driver) SetWriteFlushSize(size int)

func (*Driver) TranslationCacheStats added in v0.8.0 

func (s *Driver) TranslationCacheStats() TranslationCacheStats

TranslationCacheStats returns aggregate PostgreSQL translation-cache counters. It never exposes cached query text, SQL, or parameter data.

func (*Driver) WipeGraph added in v0.7.0 

func (s *Driver) WipeGraph(ctx context.Context, retain graph.TransactionDelegate) error

WipeGraph truncates the partitioned node and edge tables, removing every node and edge across all graphs in a single statement that bypasses the per-row edge cascade trigger. The optional retain delegate runs within the same transaction after the truncate, allowing callers to atomically recreate any data that must survive the wipe. If retain returns an error the transaction is rolled back and the graph is left untouched.

type DriverOptions added in v0.8.0 

type DriverOptions struct {
	TranslationCacheEntries int
}

DriverOptions configures driver-wide behavior. TranslationCacheEntries is a count, not a per-connection setting; zero disables translation caching.

func DefaultDriverOptions added in v0.8.0 

func DefaultDriverOptions() DriverOptions

DefaultDriverOptions returns the driver-wide PostgreSQL settings used by NewDriver.

type Int2ArrayEncoder 

type Int2ArrayEncoder struct {
	// contains filtered or unexported fields
}

func (*Int2ArrayEncoder) Encode 

func (s *Int2ArrayEncoder) Encode(values []int16) string

type IterationOptions 

type IterationOptions interface {
	Once()
}

type KindMapper 

type KindMapper interface {
	MapKindID(ctx context.Context, kindID int16) (graph.Kind, error)
	MapKindIDs(ctx context.Context, kindIDs []int16) (graph.Kinds, error)
	MapKind(ctx context.Context, kind graph.Kind) (int16, error)
	MapKinds(ctx context.Context, kinds graph.Kinds) ([]int16, error)
	AssertKinds(ctx context.Context, kinds graph.Kinds) ([]int16, error)
}

func KindMapperFromGraphDatabase 

func KindMapperFromGraphDatabase(graphDB graph.Database) (KindMapper, error)

type LargeNodeUpdateRows added in v0.4.12 

type LargeNodeUpdateRows struct {
	// contains filtered or unexported fields
}

LargeNodeUpdateRows accumulates encoded node rows for bulk loading via COPY FROM. The column order matches sql.NodeUpdateStagingColumns.

func NewLargeNodeUpdateRows added in v0.4.12 

func NewLargeNodeUpdateRows(size int) *LargeNodeUpdateRows

func (*LargeNodeUpdateRows) Append added in v0.4.12 

func (s *LargeNodeUpdateRows) Append(ctx context.Context, node *graph.Node, schemaManager *SchemaManager, kindIDEncoder Int2ArrayEncoder) error

func (*LargeNodeUpdateRows) AppendAll added in v0.4.12 

func (s *LargeNodeUpdateRows) AppendAll(ctx context.Context, nodes []*graph.Node, schemaManager *SchemaManager, kindIDEncoder Int2ArrayEncoder) error

AppendAll encodes every node in the slice and appends its row to the accumulator.

func (*LargeNodeUpdateRows) Rows added in v0.4.12 

func (s *LargeNodeUpdateRows) Rows() [][]any

type NodeUpdateParameters added in v0.4.12 

type NodeUpdateParameters struct {
	NodeIDs           []graph.ID
	KindSlices        []string
	DeletedKindSlices []string
	Properties        []pgtype.JSONB
	DeletedProperties []string
}

func NewNodeUpdateParameters added in v0.4.12 

func NewNodeUpdateParameters(size int) *NodeUpdateParameters

func (*NodeUpdateParameters) Append added in v0.4.12 

func (s *NodeUpdateParameters) Append(ctx context.Context, node *graph.Node, schemaManager *SchemaManager, kindIDEncoder Int2ArrayEncoder) error

func (*NodeUpdateParameters) AppendAll added in v0.4.12 

func (s *NodeUpdateParameters) AppendAll(ctx context.Context, nodes []*graph.Node, schemaManager *SchemaManager, kindIDEncoder Int2ArrayEncoder) error

func (*NodeUpdateParameters) Format added in v0.4.12 

func (s *NodeUpdateParameters) Format() []any

type NodeUpsertParameters 

type NodeUpsertParameters struct {
	IDFutures    []*sql.Future[graph.ID]
	KindIDSlices []string
	Properties   []pgtype.JSONB
}

func NewNodeUpsertParameters 

func NewNodeUpsertParameters(size int) *NodeUpsertParameters

func (*NodeUpsertParameters) Append 

func (s *NodeUpsertParameters) Append(ctx context.Context, update *sql.NodeUpdate, schemaManager *SchemaManager, kindIDEncoder Int2ArrayEncoder) error

func (*NodeUpsertParameters) AppendAll 

func (s *NodeUpsertParameters) AppendAll(ctx context.Context, updates *sql.NodeUpdateBatch, schemaManager *SchemaManager, kindIDEncoder Int2ArrayEncoder) error

func (*NodeUpsertParameters) Format 

func (s *NodeUpsertParameters) Format(graphTarget model.Graph) []any

type QueryHook 

type QueryHook interface {
	OnStatementMatch(statement string) QueryHookOptions
	OnStatementRegex(re *regexp.Regexp) QueryHookOptions
}

type QueryHookOptions 

type QueryHookOptions interface {
	Trace() IterationOptions
}

type QueryPathInspector 

type QueryPathInspector interface {
	Hook() QueryHook
}

func Inspector 

func Inspector() QueryPathInspector

type RelationshipUpdateByParameters 

type RelationshipUpdateByParameters struct {
	StartIDs   []graph.ID
	EndIDs     []graph.ID
	KindIDs    []int16
	Properties []pgtype.JSONB
}

func NewRelationshipUpdateByParameters 

func NewRelationshipUpdateByParameters(size int) *RelationshipUpdateByParameters

func (*RelationshipUpdateByParameters) Append 

func (s *RelationshipUpdateByParameters) Append(ctx context.Context, update *sql.RelationshipUpdate, schemaManager *SchemaManager) error

func (*RelationshipUpdateByParameters) AppendAll 

func (s *RelationshipUpdateByParameters) AppendAll(ctx context.Context, updates *sql.RelationshipUpdateBatch, schemaManager *SchemaManager) error

func (*RelationshipUpdateByParameters) Format 

func (s *RelationshipUpdateByParameters) Format(graphTarget model.Graph) []any

type SQLState 

type SQLState string
const (
	StateObjectDoesNotExist SQLState = "42704"
)

func (SQLState) ErrorMatches 

func (s SQLState) ErrorMatches(err error) bool

func (SQLState) String 

func (s SQLState) String() string

type SchemaManager 

type SchemaManager struct {
	// contains filtered or unexported fields
}

func NewSchemaManager 

func NewSchemaManager(pool *pgxpool.Pool, graphQueryMemoryLimit size.Size) *SchemaManager

func NewSchemaManagerWithOptions added in v0.8.0 

func NewSchemaManagerWithOptions(pool *pgxpool.Pool, graphQueryMemoryLimit size.Size, options DriverOptions) *SchemaManager

NewSchemaManagerWithOptions constructs the shared compilation service with a bounded translation cache.

func NewSchemaManagerWithTranslationCache added in v0.8.0 

func NewSchemaManagerWithTranslationCache(pool *pgxpool.Pool, graphQueryMemoryLimit size.Size, translationCacheEntries int) *SchemaManager

NewSchemaManagerWithTranslationCache permits an application to disable the compilation cache (zero), or to choose a bounded driver-wide entry capacity. Negative values disable the cache.

func (*SchemaManager) AssertDefaultGraph 

func (s *SchemaManager) AssertDefaultGraph(ctx context.Context, schema graph.Graph) error

func (*SchemaManager) AssertGraph 

func (s *SchemaManager) AssertGraph(tx graph.Transaction, schema graph.Graph) (model.Graph, error)

func (*SchemaManager) AssertKinds 

func (s *SchemaManager) AssertKinds(ctx context.Context, kinds graph.Kinds) ([]int16, error)

func (*SchemaManager) AssertSchema 

func (s *SchemaManager) AssertSchema(ctx context.Context, schema graph.Schema) error

func (*SchemaManager) DefaultGraph 

func (s *SchemaManager) DefaultGraph() (model.Graph, bool)

func (*SchemaManager) Fetch 

func (s *SchemaManager) Fetch(ctx context.Context) error

func (*SchemaManager) GetKindIDsByKind 

func (s *SchemaManager) GetKindIDsByKind() map[int16]graph.Kind

func (*SchemaManager) MapKind 

func (s *SchemaManager) MapKind(ctx context.Context, kind graph.Kind) (int16, error)

func (*SchemaManager) MapKindID 

func (s *SchemaManager) MapKindID(ctx context.Context, kindID int16) (graph.Kind, error)

func (*SchemaManager) MapKindIDs 

func (s *SchemaManager) MapKindIDs(ctx context.Context, kindIDs []int16) (graph.Kinds, error)

func (*SchemaManager) MapKinds 

func (s *SchemaManager) MapKinds(ctx context.Context, kinds graph.Kinds) ([]int16, error)

func (*SchemaManager) ReadTransaction 

func (s *SchemaManager) ReadTransaction(ctx context.Context, txDelegate graph.TransactionDelegate, options ...graph.TransactionOption) error

func (*SchemaManager) SetDefaultGraph 

func (s *SchemaManager) SetDefaultGraph(ctx context.Context, schema graph.Graph) error

func (*SchemaManager) WriteTransaction 

func (s *SchemaManager) WriteTransaction(ctx context.Context, txDelegate graph.TransactionDelegate, options ...graph.TransactionOption) error

type TranslationCacheStats added in v0.8.0 

type TranslationCacheStats struct {
	Hits                    int64
	Misses                  int64
	Coalesced               int64
	Bypasses                int64
	Insertions              int64
	Evictions               int64
	BindingFailures         int64
	BuildFailures           int64
	UnoptimizedCompilations int64
	Size                    int64
	Capacity                int
	Generation              uint64
}

TranslationCacheStats contains aggregate counters only. It intentionally exposes no source text, SQL, parameter names, values, or connection data.

Source Files

View all Source files

Directories

Path Synopsis

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.