searchsocket 0.5.0 → 0.6.0

package/dist/{types-Dk43uz25.d.cts → types-029hl6P2.d.cts}

@@ -1,5 +1,10 @@
+type Awaitable<T> = T | Promise<T>;
 type ScopeMode = "fixed" | "git" | "env";
 type SourceMode = "static-output" | "crawl" | "content-files" | "build";
+interface OutgoingLink {
+    url: string;
+    anchorText: string;
+}
 interface SearchSocketConfig {
     project?: {
         id?: string;
@@ -43,6 +48,7 @@ interface SearchSocketConfig {
         dropSelectors?: string[];
         ignoreAttr?: string;
         noindexAttr?: string;
+        imageDescAttr?: string;
         respectRobotsNoindex?: boolean;
     };
     transform?: {
@@ -59,34 +65,47 @@ interface SearchSocketConfig {
         dontSplitInside?: Array<"code" | "table" | "blockquote">;
         prependTitle?: boolean;
         pageSummaryChunk?: boolean;
+        weightHeadings?: boolean;
     };
     upstash?: {
         url?: string;
         token?: string;
         urlEnv?: string;
         tokenEnv?: string;
+        namespaces?: {
+            pages?: string;
+            chunks?: string;
+        };
+    };
+    embedding?: {
+        model?: string;
+        dimensions?: number;
+        taskType?: string;
+        batchSize?: number;
     };
     search?: {
-        semanticWeight?: number;
-        inputEnrichment?: boolean;
-        reranking?: boolean;
         dualSearch?: boolean;
         pageSearchWeight?: number;
     };
     ranking?: {
         enableIncomingLinkBoost?: boolean;
         enableDepthBoost?: boolean;
+        enableFreshnessBoost?: boolean;
+        freshnessDecayRate?: number;
+        enableAnchorTextBoost?: boolean;
         pageWeights?: Record<string, number>;
         aggregationCap?: number;
         aggregationDecay?: number;
         minChunkScoreRatio?: number;
-        minScore?: number;
+        minScoreRatio?: number;
         scoreGapThreshold?: number;
         weights?: {
             incomingLinks?: number;
             depth?: number;
             aggregation?: number;
             titleMatch?: number;
+            freshness?: number;
+            anchorText?: number;
         };
     };
     api?: {
@@ -101,12 +120,28 @@ interface SearchSocketConfig {
     };
     mcp?: {
         enable?: boolean;
+        access?: "public" | "private";
         transport?: "stdio" | "http";
         http?: {
             port?: number;
             path?: string;
+            apiKey?: string;
+            apiKeyEnv?: string;
+        };
+        handle?: {
+            path?: string;
+            apiKey?: string;
+            enableJsonResponse?: boolean;
         };
     };
+    llmsTxt?: {
+        enable?: boolean;
+        outputPath?: string;
+        title?: string;
+        description?: string;
+        generateFull?: boolean;
+        serveMarkdownVariants?: boolean;
+    };
     state?: {
         dir?: string;
     };
@@ -154,6 +189,7 @@ interface ResolvedSearchSocketConfig {
         dropSelectors: string[];
         ignoreAttr: string;
         noindexAttr: string;
+        imageDescAttr: string;
         respectRobotsNoindex: boolean;
     };
     transform: {
@@ -170,34 +206,47 @@ interface ResolvedSearchSocketConfig {
         dontSplitInside: Array<"code" | "table" | "blockquote">;
         prependTitle: boolean;
         pageSummaryChunk: boolean;
+        weightHeadings: boolean;
     };
     upstash: {
         url?: string;
         token?: string;
         urlEnv: string;
         tokenEnv: string;
+        namespaces: {
+            pages: string;
+            chunks: string;
+        };
+    };
+    embedding: {
+        model: string;
+        dimensions: number;
+        taskType: string;
+        batchSize: number;
     };
     search: {
-        semanticWeight: number;
-        inputEnrichment: boolean;
-        reranking: boolean;
         dualSearch: boolean;
         pageSearchWeight: number;
     };
     ranking: {
         enableIncomingLinkBoost: boolean;
         enableDepthBoost: boolean;
+        enableFreshnessBoost: boolean;
+        freshnessDecayRate: number;
+        enableAnchorTextBoost: boolean;
         pageWeights: Record<string, number>;
         aggregationCap: number;
         aggregationDecay: number;
         minChunkScoreRatio: number;
-        minScore: number;
+        minScoreRatio: number;
         scoreGapThreshold: number;
         weights: {
             incomingLinks: number;
             depth: number;
             aggregation: number;
             titleMatch: number;
+            freshness: number;
+            anchorText: number;
         };
     };
     api: {
@@ -212,11 +261,27 @@ interface ResolvedSearchSocketConfig {
     };
     mcp: {
         enable: boolean;
+        access: "public" | "private";
         transport: "stdio" | "http";
         http: {
             port: number;
             path: string;
+            apiKey?: string;
+            apiKeyEnv?: string;
         };
+        handle: {
+            path: string;
+            apiKey?: string;
+            enableJsonResponse: boolean;
+        };
+    };
+    llmsTxt: {
+        enable: boolean;
+        outputPath: string;
+        title?: string;
+        description?: string;
+        generateFull: boolean;
+        serveMarkdownVariants: boolean;
     };
     state: {
         dir: string;
@@ -227,6 +292,19 @@ interface Scope {
     scopeName: string;
     scopeId: string;
 }
+interface ExtractedPage {
+    url: string;
+    title: string;
+    markdown: string;
+    outgoingLinks: OutgoingLink[];
+    noindex: boolean;
+    tags: string[];
+    description?: string;
+    keywords?: string[];
+    weight?: number;
+    publishedAt?: number;
+    meta?: Record<string, string | number | boolean | string[]>;
+}
 interface Chunk {
     chunkKey: string;
     ordinal: number;
@@ -234,6 +312,7 @@ interface Chunk {
     path: string;
     title: string;
     sectionTitle?: string;
+    headingLevel?: number;
     headingPath: string[];
     chunkText: string;
     snippet: string;
@@ -244,6 +323,9 @@ interface Chunk {
     contentHash: string;
     description?: string;
     keywords?: string[];
+    publishedAt?: number;
+    incomingAnchorText?: string;
+    meta?: Record<string, string | number | boolean | string[]>;
 }
 interface VectorHit {
     id: string;
@@ -264,8 +346,13 @@ interface VectorHit {
         incomingLinks: number;
         routeFile: string;
         tags: string[];
+        type?: "chunk" | "page" | "image";
         description?: string;
         keywords?: string[];
+        incomingAnchorText?: string;
+        imageSrc?: string;
+        imageAlt?: string;
+        publishedAt?: number;
     };
 }
 interface PageRecord {
@@ -278,12 +365,16 @@ interface PageRecord {
     routeResolution: "exact" | "best-effort";
     incomingLinks: number;
     outgoingLinks: number;
+    outgoingLinkUrls?: string[];
     depth: number;
     tags: string[];
     indexedAt: string;
     summary?: string;
     description?: string;
     keywords?: string[];
+    contentHash?: string;
+    publishedAt?: number;
+    meta?: Record<string, string | number | boolean | string[]>;
 }
 interface PageHit {
     id: string;
@@ -295,6 +386,7 @@ interface PageHit {
     depth: number;
     incomingLinks: number;
     routeFile: string;
+    publishedAt?: number;
 }
 interface ScopeInfo {
     projectId: string;
@@ -302,17 +394,50 @@ interface ScopeInfo {
     lastIndexedAt: string;
     documentCount?: number;
 }
+interface RankingOverrides {
+    ranking?: {
+        enableIncomingLinkBoost?: boolean;
+        enableDepthBoost?: boolean;
+        aggregationCap?: number;
+        aggregationDecay?: number;
+        minChunkScoreRatio?: number;
+        minScoreRatio?: number;
+        scoreGapThreshold?: number;
+        weights?: {
+            incomingLinks?: number;
+            depth?: number;
+            aggregation?: number;
+            titleMatch?: number;
+        };
+    };
+    search?: {
+        pageSearchWeight?: number;
+    };
+}
 interface SearchRequest {
     q: string;
     topK?: number;
     scope?: string;
     pathPrefix?: string;
     tags?: string[];
+    filters?: Record<string, string | number | boolean>;
     groupBy?: "page" | "chunk";
+    maxSubResults?: number;
+    debug?: boolean;
+    rankingOverrides?: RankingOverrides;
+}
+interface ScoreBreakdown {
+    baseScore: number;
+    incomingLinkBoost: number;
+    depthBoost: number;
+    titleMatchBoost: number;
+    freshnessBoost: number;
+    anchorTextMatchBoost: number;
 }
 interface SearchResultChunk {
     sectionTitle?: string;
     snippet: string;
+    chunkText?: string;
     headingPath: string[];
     score: number;
 }
@@ -321,9 +446,11 @@ interface SearchResult {
     title: string;
     sectionTitle?: string;
     snippet: string;
+    chunkText?: string;
     score: number;
     routeFile: string;
     chunks?: SearchResultChunk[];
+    breakdown?: ScoreBreakdown;
 }
 interface SearchResponse {
     q: string;
@@ -338,6 +465,8 @@ interface SearchResponse {
 }
 interface IndexStats {
     pagesProcessed: number;
+    pagesChanged: number;
+    pagesDeleted: number;
     chunksTotal: number;
     chunksChanged: number;
     documentsUpserted: number;
@@ -346,6 +475,20 @@ interface IndexStats {
     routeBestEffort: number;
     stageTimingsMs: Record<string, number>;
 }
+interface IndexingHooks {
+    transformPage?: (page: ExtractedPage) => Awaitable<ExtractedPage | null>;
+    transformChunk?: (chunk: Chunk) => Awaitable<Chunk | null>;
+    beforeIndex?: (chunks: Chunk[]) => Awaitable<Chunk[]>;
+    afterIndex?: (stats: IndexStats) => Awaitable<void>;
+}
+interface CustomRecord {
+    url: string;
+    title: string;
+    content: string;
+    metadata?: Record<string, string>;
+    tags?: string[];
+    weight?: number;
+}
 interface IndexOptions {
     scopeOverride?: string;
     changedOnly?: boolean;
@@ -355,6 +498,34 @@ interface IndexOptions {
     maxPages?: number;
     maxChunks?: number;
     verbose?: boolean;
+    customRecords?: CustomRecord[];
+}
+interface SiteTreeNode {
+    url: string;
+    title: string;
+    depth: number;
+    routeFile: string;
+    isIndexed: boolean;
+    childCount: number;
+    children: SiteTreeNode[];
+}
+interface SiteStructureResult {
+    root: SiteTreeNode;
+    totalPages: number;
+    truncated: boolean;
+}
+type RelationshipType = "outgoing_link" | "incoming_link" | "sibling" | "semantic";
+interface RelatedPage {
+    url: string;
+    title: string;
+    score: number;
+    relationshipType: RelationshipType;
+    routeFile: string;
+}
+interface RelatedPagesResult {
+    sourceUrl: string;
+    scope: string;
+    relatedPages: RelatedPage[];
 }
 
-export type { Chunk as C, IndexOptions as I, PageHit as P, ResolvedSearchSocketConfig as R, SearchRequest as S, VectorHit as V, SearchResponse as a, SearchResult as b, SearchSocketConfig as c, Scope as d, ScopeInfo as e, PageRecord as f, IndexStats as g };
+export type { Awaitable as A, Chunk as C, IndexingHooks as I, PageHit as P, ResolvedSearchSocketConfig as R, SearchRequest as S, VectorHit as V, SearchResponse as a, SearchResult as b, SearchSocketConfig as c, Scope as d, ScopeInfo as e, PageRecord as f, IndexOptions as g, IndexStats as h, SiteStructureResult as i, RelatedPagesResult as j, CustomRecord as k };