Class BaseReRankerAbstract

constructor

• new BaseReRanker(): BaseReRanker

  Returns BaseReRanker

CostReporter

AbstractDriverClass

• get DriverClass(): string

  The DriverClass key used by @RegisterClass(BaseReRanker, 'DriverClassName') so the SearchEngine can resolve this subclass via ClassFactory from the scope's ScopeConfig.reRanker.driverClass value.

  Returns string

Name

• get Name(): string

  Human-friendly display name. Defaults to DriverClass so subclasses don't have to override unless they want a different label in UI dropdowns / cost reports.

  Returns string

Version

• get Version(): string

  Reranker semantic version (independent of the package version). Bump this on a subclass when the prompt, scoring formula, or upstream model identifier changes in a way that invalidates cached scores. Defaults to '1'.

  Returns string

EstimateCostCents

• EstimateCostCents(resultCount: number): number

  Pre-call cost estimate in cents for reranking the given candidate count. Used by the budget guard to short-circuit BEFORE making the call when the projected cost exceeds SearchScope.RerankerBudgetCents.

  Default: 0 (free / local). Real-provider subclasses override based on their pricing.

  Parameters

  • resultCount: number

    The number of candidates to be reranked.

  Returns number

  Estimated cost in cents (whole + fractional, e.g. 0.25 for ¼¢).

GetMaxResultCount

• GetMaxResultCount(): number

  Maximum number of candidates a single Rerank call can score. Real providers cap this (Cohere: 1000, Voyage: 1000, OpenAI: varies). The SearchEngine should respect this cap by chunking large candidate lists. Defaults to Number.MAX_SAFE_INTEGER (no cap) — subclasses should override.

  Returns number

ReRank

• ReRank(
      query: string,
      candidates: SearchResultItem[],
      topN: number,
      contextUser: UserInfo,
      config?: Record<string, unknown>,
  ): Promise<SearchResultItem[]>

  Score and re-order candidates against the query.

  Default implementation:

  1. Returns empty when topN <= 0 or candidates is empty.
  2. Resolves an AI-layer reranker via getAIReranker(). If none, slices to topN.
  3. Maps SearchResultItem[] to RerankDocument[] (preserving each original item in metadata.__mjSearchItem).
  4. Calls the AI reranker's Rerank() — inherits validation, timing, error handling.
  5. On success, maps the RerankResult[] back to SearchResultItem[], replacing Score with relevanceScore and augmenting ScoreBreakdown with a ReRank key.
  6. On failure, logs and falls back to the unchanged top-N slice.

  Subclasses that bypass the AI layer (e.g. NoopReRanker) may override this method.

  Parameters

  • query: string

    The original query text.

  • candidates: SearchResultItem[]

    The fused candidate list (post-RRF, pre-dedup).

  • topN: number

    Maximum number of candidates to return after re-ranking.

  • contextUser: UserInfo

    The calling user (for auth / tenant propagation).

  • Optionalconfig: Record<string, unknown>

    Optional provider-specific config from ScopeConfig.reRanker.config.

  Returns Promise<SearchResultItem[]>

StaticGetAvailableRerankers

• GetAvailableRerankers(): RegisteredReRankerInfo[]

  Enumerate every reranker currently registered with ClassFactory under BaseReRanker. Designed for the SearchScope form's reranker dropdown (P2D.7) — the form populates the dropdown from this single call rather than hardcoding a list, so any ClassFactory-registered reranker (including third-party ones published as separate packages) shows up automatically.

  Each entry includes the driver-class registration key, the friendly Name, Version, and a HasCost flag (true when EstimateCostCents(1) > 0). Sorted by Name for stable UI ordering.

  Returns RegisteredReRankerInfo[]