datahub

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

type AliasConfig struct {
	// Source is the fully-qualified source table name.
	Source string `yaml:"source"`

	// Targets are glob patterns matching target table names.
	Targets []string `yaml:"targets"`

	// ColumnMapping provides explicit column name mappings.
	// Key: target column, Value: source column
	ColumnMapping map[string]string `yaml:"column_mapping,omitempty"`
}

type Client interface {
	SearchAcrossEntities(ctx context.Context, query string, opts ...dhclient.SearchOption) (*types.SearchResult, error)
	SemanticSearch(ctx context.Context, query string, opts ...dhclient.SearchOption) (*types.SearchResult, error)
	SearchDocuments(ctx context.Context, query string, opts ...dhclient.SearchOption) ([]types.Document, error)
	GetRelatedDocuments(ctx context.Context, urn string) ([]types.Document, error)
	GetDocument(ctx context.Context, urn string) (*types.Document, error)
	GetEntity(ctx context.Context, urn string) (*types.Entity, error)
	GetSchema(ctx context.Context, urn string) (*types.SchemaMetadata, error)
	GetSchemas(ctx context.Context, urns []string) (map[string]*types.SchemaMetadata, error)
	GetLineage(ctx context.Context, urn string, opts ...dhclient.LineageOption) (*types.LineageResult, error)
	GetColumnLineage(ctx context.Context, urn string) (*types.ColumnLineage, error)
	GetGlossaryTerm(ctx context.Context, urn string) (*types.GlossaryTerm, error)
	GetQueries(ctx context.Context, urn string) (*types.QueryList, error)
	ListTags(ctx context.Context, filter string) ([]types.Tag, error)
	ListDomains(ctx context.Context) ([]types.Domain, error)
	Ping(ctx context.Context) error
	Close() error
}

type ColumnTransformConfig struct {
	// TargetPattern is a glob pattern matching target dataset names.
	TargetPattern string `yaml:"target_pattern"`

	// StripPrefix removes this prefix from target column names.
	StripPrefix string `yaml:"strip_prefix,omitempty"`

	// StripSuffix removes this suffix from target column names.
	StripSuffix string `yaml:"strip_suffix,omitempty"`
}

type Config struct {
	URL      string
	Token    string
	Platform string // Default platform for URN building (e.g., "trino", "postgres")
	Timeout  time.Duration
	Debug    bool // Enable debug logging

	// CatalogMapping maps query engine catalog names to metadata catalog names.
	// For example: {"rdbms": "warehouse"} means the Trino "rdbms" catalog
	// corresponds to the "warehouse" catalog in DataHub URNs.
	CatalogMapping map[string]string

	// Lineage configuration for inheritance-aware column resolution.
	Lineage LineageConfig
}

type LineageConfig struct {
	// Enabled activates lineage traversal for missing documentation.
	Enabled bool `yaml:"enabled"`

	// MaxHops limits upstream traversal depth. Range: 1-5. Default: 2.
	MaxHops int `yaml:"max_hops"`

	// Inherit specifies which metadata types to inherit.
	// Valid: "glossary_terms", "descriptions", "tags"
	Inherit []string `yaml:"inherit"`

	// ConflictResolution determines behavior when multiple upstreams
	// define metadata for the same column.
	// Values: "nearest" (closest upstream wins), "all" (merge), "skip" (no inheritance on conflict)
	ConflictResolution string `yaml:"conflict_resolution"`

	// PreferColumnLineage uses DataHub's column-level lineage edges when available.
	PreferColumnLineage bool `yaml:"prefer_column_lineage"`

	// ColumnTransforms defines path normalization rules.
	ColumnTransforms []ColumnTransformConfig `yaml:"column_transforms"`

	// Aliases defines explicit source-target mappings that bypass lineage lookup.
	Aliases []AliasConfig `yaml:"aliases"`

	// CacheTTL for lineage graphs.
	CacheTTL time.Duration `yaml:"cache_ttl"`

	// Timeout for the entire inheritance operation.
	Timeout time.Duration `yaml:"timeout"`
}