relation

type Builder ¶

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

Builder accumulates nodes and rows, validating each addition against the vocabulary. Insertion order is irrelevant: Build sorts everything into canonical order, and rows may be added before the nodes they reference (referential integrity is checked at Build).

func NewBuilder() *Builder

func (b *Builder) AddNode(n Node) error

func (b *Builder) AddRow(r Row) error

func (b *Builder) Build() (*Relation, error)

type Edge ¶

type Edge struct {
	Src string
	Dst string
}

Edge is one algorithm-graph edge: a distinct (src, dst) pair, as serialized IDs.

type Node ¶

type Node struct {
	Kind  vocab.NodeKind
	ID    NodeID
	Attrs map[string]Value
}

Node is one node-table entry: (node_kind, id, attrs).

type NodeID ¶

type NodeID struct {
	// Lang is the profile prefix: "py" | "go" | "ts".
	Lang string
	// Member is the workspace member name, or "_" for a single-project
	// (non-workspace) scan.
	Member string
	// Module is the module's logical name (SPEC.md section 2.2).
	Module string
	// Chain is the container chain from module scope inward; empty for
	// module-kind nodes.
	Chain []Segment
}

NodeID is the hierarchical qualified name of a node (schema/SPEC.md section 2). Location is never part of identity. The serialized form is

<lang>:<member>:<module>:<container-chain>

Serialized IDs are opaque strings for consumers; the structured form is authoritative.

func (id NodeID) String() string

func (id NodeID) Validate() error

type Relation ¶

type Relation struct {
	Nodes []Node
	Rows  []Row
}

Relation is the frozen, canonically ordered relation: the node table sorted by (node_kind, id) and the rows sorted by the total key (row_kind, src_id, dst_id, file, span_start, span_end, attr-tuple).

func (rel *Relation) AlgorithmGraph(kind vocab.RowKind) []Edge

func (rel *Relation) CanonicalForm() []byte

func (rel *Relation) Hash() [32]byte

func (rel *Relation) SiteFeed(kind vocab.RowKind) []Row

type Row ¶

type Row struct {
	Kind  vocab.RowKind
	Src   NodeID
	Dst   NodeID
	File  string
	Span  Span
	Attrs map[string]Value
}

Row is one interaction-relation row: (row_kind, src_node, dst_node, file, span, attrs).

type Segment ¶

type Segment struct {
	// Name is the unit's name; for anonymous units it is the name hint, or
	// "anon" when no hint is syntactically derivable.
	Name string
	// Overload is the 0-based source-order index among same-name siblings in
	// the same container (SPEC.md section 2.4). Serialized as #<n>, omitted
	// for 0.
	Overload int
	// Anonymous marks a synthesized segment (SPEC.md section 2.3):
	// <name-hint|anon>~<ordinal>~<fp8>.
	Anonymous bool
	// Ordinal is the 0-based source-order index among same-hint anonymous
	// siblings in the same parent. Anonymous only.
	Ordinal int
	// Fingerprint is the first 8 hex chars of SHA-256 over the unit's
	// normalized signature text. Anonymous only. It exists so ordinal drift
	// is detectable, never silently misattributed.
	Fingerprint string
}

Segment is one element of the container chain.

type Span ¶

type Span struct {
	Start uint32
	End   uint32
}

Span is a byte range over the file's LF-normalized UTF-8 bytes (schema/SPEC.md section 3). Line/column are derived at output time, never stored as truth.

type Value ¶

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

Value is one typed attribute value. Enum-typed attributes carry their value as a string; legality against the vocabulary enum is checked by the relation builder.

func BoolValue(b bool) Value

func IntValue(i int64) Value

func StringValue(s string) Value

func (v Value) AsBool() (bool, bool)

func (v Value) AsInt() (int64, bool)

func (v Value) AsString() (string, bool)

func (v Value) Kind() ValueKind

type ValueKind ¶

type ValueKind int

ValueKind is the type of an attribute value.

const (
	ValueString ValueKind = iota
	ValueInt
	ValueBool
)