Package net.thevpc.naru.api.model

Interface NaruCacheableContext

All Known Implementing Classes:

NaruCacheableContext.DefaultNaruCacheableContext

public interface NaruCacheableContext

An outgoing request expressed as an ordered, explicitly-labelled decomposition of its stable and volatile parts.

This is an opt-in view of a request. Callers that do not implement it keep passing a plain NaruModelRequest and providers that do not consume it keep seeing exactly what they saw before; nothing here is on the mandatory path.

Its purpose is to let a caller say what is stable without knowing how any provider expresses caching. toRequest(Map) collapses the segments back into the ordinary flat request, which is what a provider with no caching support — or any provider whose mode is NaruCachingMode.AUTOMATIC_PREFIX — actually consumes. Ordering is preserved exactly, because for automatic prefix caching the ordering is the contract.

Segments are expected in stable-to-volatile order. NARU does not reorder them: silently moving a caller's "latest user turn" ahead of their indexed context would change the model's view of the conversation, and a caching layer is the last place that should be reordering prompts.

Nested Class Summary

Nested Classes

Modifier and Type	Interface	Description
static final class	NaruCacheableContext.Builder	Fluent assembler.
static final class	NaruCacheableContext.DefaultNaruCacheableContext

Method Summary

Modifier and Type	Method	Description
static NaruCacheableContext.Builder	builder()
default boolean	isEmpty()
static NaruCacheableContext	of(List<NaruContextSegment> segments)
static NaruCacheableContext	of(List<NaruContextSegment> segments, List<NaruToolDefinition> tools)	Convenience for the common case: a stable head and a volatile tail, with the tool definitions folded in as their own cacheable segment.
List<NaruContextSegment>	segments()
default NaruModelRequest	toRequest(Map<String,NElement> env)	Collapse to the plain request shape.

Method Details

segments

List<NaruContextSegment> segments()

isEmpty

default boolean isEmpty()

toRequest

default NaruModelRequest toRequest(Map<String,NElement> env)

Collapse to the plain request shape. Cacheability is dropped here on purpose: the flat form carries no cache information at all, so any provider that does not implement caching translation gets byte-identical behaviour to a request built without segmentation.

of

static NaruCacheableContext of(List<NaruContextSegment> segments)

builder

static NaruCacheableContext.Builder builder()

of

static NaruCacheableContext of(List<NaruContextSegment> segments,
 List<NaruToolDefinition> tools)

Convenience for the common case: a stable head and a volatile tail, with the tool definitions folded in as their own cacheable segment.