searchsocket 0.2.0

package/dist/types-D1K46vwd.d.ts

@@ -0,0 +1,403 @@
+type ScopeMode = "fixed" | "git" | "env";
+type SourceMode = "static-output" | "crawl" | "content-files" | "build";
+type EmbeddingProvider = "openai";
+type RerankProvider = "none" | "jina";
+interface SearchSocketConfig {
+    project?: {
+        id?: string;
+        baseUrl?: string;
+    };
+    scope?: {
+        mode?: ScopeMode;
+        fixed?: string;
+        envVar?: string;
+        sanitize?: boolean;
+    };
+    source?: {
+        mode?: SourceMode;
+        staticOutputDir?: string;
+        strictRouteMapping?: boolean;
+        crawl?: {
+            baseUrl: string;
+            routes?: string[];
+            sitemapUrl?: string;
+        };
+        contentFiles?: {
+            globs: string[];
+            baseDir?: string;
+        };
+        build?: {
+            outputDir?: string;
+            paramValues?: Record<string, string[]>;
+            exclude?: string[];
+            previewTimeout?: number;
+        };
+    };
+    extract?: {
+        mainSelector?: string;
+        dropTags?: string[];
+        dropSelectors?: string[];
+        ignoreAttr?: string;
+        noindexAttr?: string;
+        respectRobotsNoindex?: boolean;
+    };
+    transform?: {
+        output?: "markdown";
+        preserveCodeBlocks?: boolean;
+        preserveTables?: boolean;
+    };
+    chunking?: {
+        strategy?: "hybrid";
+        maxChars?: number;
+        overlapChars?: number;
+        minChars?: number;
+        headingPathDepth?: number;
+        dontSplitInside?: Array<"code" | "table" | "blockquote">;
+        prependTitle?: boolean;
+        pageSummaryChunk?: boolean;
+    };
+    embeddings?: {
+        provider?: EmbeddingProvider;
+        model?: string;
+        apiKeyEnv?: string;
+        batchSize?: number;
+        concurrency?: number;
+        pricePer1kTokens?: number;
+    };
+    vector?: {
+        dimension?: number;
+        turso?: {
+            urlEnv?: string;
+            authTokenEnv?: string;
+            localPath?: string;
+        };
+    };
+    rerank?: {
+        provider?: RerankProvider;
+        topN?: number;
+        jina?: {
+            apiKeyEnv?: string;
+            model?: string;
+        };
+    };
+    ranking?: {
+        enableIncomingLinkBoost?: boolean;
+        enableDepthBoost?: boolean;
+        pageWeights?: Record<string, number>;
+        aggregationCap?: number;
+        aggregationDecay?: number;
+        minChunkScoreRatio?: number;
+        weights?: {
+            incomingLinks?: number;
+            depth?: number;
+            rerank?: number;
+            aggregation?: number;
+        };
+    };
+    api?: {
+        path?: string;
+        cors?: {
+            allowOrigins?: string[];
+        };
+        rateLimit?: {
+            windowMs?: number;
+            max?: number;
+        };
+    };
+    mcp?: {
+        enable?: boolean;
+        transport?: "stdio" | "http";
+        http?: {
+            port?: number;
+            path?: string;
+        };
+    };
+    state?: {
+        dir?: string;
+        writeMirror?: boolean;
+    };
+}
+interface ResolvedSearchSocketConfig {
+    project: {
+        id: string;
+        baseUrl?: string;
+    };
+    scope: {
+        mode: ScopeMode;
+        fixed: string;
+        envVar: string;
+        sanitize: boolean;
+    };
+    source: {
+        mode: SourceMode;
+        staticOutputDir: string;
+        strictRouteMapping: boolean;
+        crawl?: {
+            baseUrl: string;
+            routes: string[];
+            sitemapUrl?: string;
+        };
+        contentFiles?: {
+            globs: string[];
+            baseDir: string;
+        };
+        build?: {
+            outputDir: string;
+            paramValues: Record<string, string[]>;
+            exclude: string[];
+            previewTimeout: number;
+        };
+    };
+    extract: {
+        mainSelector: string;
+        dropTags: string[];
+        dropSelectors: string[];
+        ignoreAttr: string;
+        noindexAttr: string;
+        respectRobotsNoindex: boolean;
+    };
+    transform: {
+        output: "markdown";
+        preserveCodeBlocks: boolean;
+        preserveTables: boolean;
+    };
+    chunking: {
+        strategy: "hybrid";
+        maxChars: number;
+        overlapChars: number;
+        minChars: number;
+        headingPathDepth: number;
+        dontSplitInside: Array<"code" | "table" | "blockquote">;
+        prependTitle: boolean;
+        pageSummaryChunk: boolean;
+    };
+    embeddings: {
+        provider: EmbeddingProvider;
+        model: string;
+        apiKeyEnv: string;
+        batchSize: number;
+        concurrency: number;
+        pricePer1kTokens?: number;
+    };
+    vector: {
+        dimension?: number;
+        turso: {
+            urlEnv: string;
+            authTokenEnv: string;
+            localPath: string;
+        };
+    };
+    rerank: {
+        provider: RerankProvider;
+        topN: number;
+        jina: {
+            apiKeyEnv: string;
+            model: string;
+        };
+    };
+    ranking: {
+        enableIncomingLinkBoost: boolean;
+        enableDepthBoost: boolean;
+        pageWeights: Record<string, number>;
+        aggregationCap: number;
+        aggregationDecay: number;
+        minChunkScoreRatio: number;
+        weights: {
+            incomingLinks: number;
+            depth: number;
+            rerank: number;
+            aggregation: number;
+        };
+    };
+    api: {
+        path: string;
+        cors: {
+            allowOrigins: string[];
+        };
+        rateLimit?: {
+            windowMs: number;
+            max: number;
+        };
+    };
+    mcp: {
+        enable: boolean;
+        transport: "stdio" | "http";
+        http: {
+            port: number;
+            path: string;
+        };
+    };
+    state: {
+        dir: string;
+        writeMirror: boolean;
+    };
+}
+interface Scope {
+    projectId: string;
+    scopeName: string;
+    scopeId: string;
+}
+interface Chunk {
+    chunkKey: string;
+    ordinal: number;
+    url: string;
+    path: string;
+    title: string;
+    sectionTitle?: string;
+    headingPath: string[];
+    chunkText: string;
+    snippet: string;
+    depth: number;
+    incomingLinks: number;
+    routeFile: string;
+    tags: string[];
+    contentHash: string;
+}
+interface VectorRecord {
+    id: string;
+    vector: number[];
+    metadata: {
+        projectId: string;
+        scopeName: string;
+        url: string;
+        path: string;
+        title: string;
+        sectionTitle: string;
+        headingPath: string[];
+        snippet: string;
+        contentHash: string;
+        modelId: string;
+        depth: number;
+        incomingLinks: number;
+        routeFile: string;
+        tags: string[];
+    };
+}
+interface QueryOpts {
+    topK: number;
+    pathPrefix?: string;
+    tags?: string[];
+}
+interface VectorHit {
+    id: string;
+    score: number;
+    metadata: VectorRecord["metadata"];
+}
+interface PageRecord {
+    url: string;
+    title: string;
+    markdown: string;
+    projectId: string;
+    scopeName: string;
+    routeFile: string;
+    routeResolution: "exact" | "best-effort";
+    incomingLinks: number;
+    outgoingLinks: number;
+    depth: number;
+    tags: string[];
+    indexedAt: string;
+}
+interface ScopeInfo {
+    projectId: string;
+    scopeName: string;
+    modelId: string;
+    lastIndexedAt: string;
+    vectorCount?: number;
+    lastEstimateTokens?: number;
+    lastEstimateCostUSD?: number;
+    lastEstimateChangedChunks?: number;
+}
+interface VectorStore {
+    upsert(records: VectorRecord[], scope: Scope): Promise<void>;
+    query(queryVector: number[], opts: QueryOpts, scope: Scope): Promise<VectorHit[]>;
+    deleteByIds(ids: string[], scope: Scope): Promise<void>;
+    deleteScope(scope: Scope): Promise<void>;
+    listScopes(scopeProjectId: string): Promise<ScopeInfo[]>;
+    recordScope(info: ScopeInfo): Promise<void>;
+    health(): Promise<{
+        ok: boolean;
+        details?: string;
+    }>;
+    getContentHashes(scope: Scope): Promise<Map<string, string>>;
+    upsertPages(pages: PageRecord[], scope: Scope): Promise<void>;
+    getPage(url: string, scope: Scope): Promise<PageRecord | null>;
+    deletePages(scope: Scope): Promise<void>;
+    getScopeModelId(scope: Scope): Promise<string | null>;
+}
+interface EmbeddingsProvider {
+    embedTexts(texts: string[], modelId: string): Promise<number[][]>;
+    estimateTokens(text: string): number;
+}
+interface RerankCandidate {
+    id: string;
+    text: string;
+}
+interface Reranker {
+    rerank(query: string, candidates: RerankCandidate[], topN?: number): Promise<Array<{
+        id: string;
+        score: number;
+    }>>;
+}
+interface SearchRequest {
+    q: string;
+    topK?: number;
+    scope?: string;
+    pathPrefix?: string;
+    tags?: string[];
+    rerank?: boolean;
+    groupBy?: "page" | "chunk";
+}
+interface SearchResultChunk {
+    sectionTitle?: string;
+    snippet: string;
+    headingPath: string[];
+    score: number;
+}
+interface SearchResult {
+    url: string;
+    title: string;
+    sectionTitle?: string;
+    snippet: string;
+    score: number;
+    routeFile: string;
+    chunks?: SearchResultChunk[];
+}
+interface SearchResponse {
+    q: string;
+    scope: string;
+    results: SearchResult[];
+    meta: {
+        timingsMs: {
+            embed: number;
+            vector: number;
+            rerank: number;
+            total: number;
+        };
+        usedRerank: boolean;
+        modelId: string;
+    };
+}
+interface IndexStats {
+    pagesProcessed: number;
+    chunksTotal: number;
+    chunksChanged: number;
+    newEmbeddings: number;
+    deletes: number;
+    estimatedTokens: number;
+    estimatedCostUSD: number;
+    routeExact: number;
+    routeBestEffort: number;
+    stageTimingsMs: Record<string, number>;
+}
+interface IndexOptions {
+    scopeOverride?: string;
+    changedOnly?: boolean;
+    force?: boolean;
+    dryRun?: boolean;
+    sourceOverride?: SourceMode;
+    maxPages?: number;
+    maxChunks?: number;
+    verbose?: boolean;
+}
+
+export type { Chunk as C, EmbeddingsProvider as E, IndexOptions as I, QueryOpts as Q, ResolvedSearchSocketConfig as R, SearchRequest as S, VectorStore as V, SearchResponse as a, SearchSocketConfig as b, Scope as c, Reranker as d, RerankCandidate as e, IndexStats as f, VectorHit as g, VectorRecord as h };

package/package.json

@@ -0,0 +1,86 @@
+{
+  "name": "searchsocket",
+  "version": "0.2.0",
+  "description": "Semantic site search and MCP retrieval for SvelteKit static sites",
+  "license": "MIT",
+  "author": "Greg Priday <greg@siteorigin.com>",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/gregpriday/searchsocket.git"
+  },
+  "homepage": "https://github.com/gregpriday/searchsocket",
+  "bugs": {
+    "url": "https://github.com/gregpriday/searchsocket/issues"
+  },
+  "keywords": [
+    "search",
+    "semantic-search",
+    "sveltekit",
+    "mcp",
+    "embeddings",
+    "vector-search",
+    "site-search",
+    "static-site"
+  ],
+  "type": "module",
+  "files": [
+    "dist",
+    "README.md"
+  ],
+  "bin": {
+    "searchsocket": "dist/cli.js"
+  },
+  "exports": {
+    ".": {
+      "types": "./dist/index.d.ts",
+      "import": "./dist/index.js",
+      "require": "./dist/index.cjs"
+    },
+    "./sveltekit": {
+      "types": "./dist/sveltekit.d.ts",
+      "import": "./dist/sveltekit.js",
+      "require": "./dist/sveltekit.cjs"
+    },
+    "./client": {
+      "types": "./dist/client.d.ts",
+      "import": "./dist/client.js",
+      "require": "./dist/client.cjs"
+    }
+  },
+  "scripts": {
+    "build": "tsup",
+    "clean": "rm -rf dist",
+    "typecheck": "tsc --noEmit",
+    "test": "vitest run",
+    "test:watch": "vitest"
+  },
+  "engines": {
+    "node": ">=20"
+  },
+  "packageManager": "pnpm@10.29.2",
+  "dependencies": {
+    "@libsql/client": "^0.17.0",
+    "@modelcontextprotocol/sdk": "^1.26.0",
+    "cheerio": "^1.2.0",
+    "chokidar": "^5.0.0",
+    "commander": "^14.0.3",
+    "dotenv": "^17.3.1",
+    "express": "^5.2.1",
+    "fast-glob": "^3.3.3",
+    "gray-matter": "^4.0.3",
+    "jiti": "^2.6.1",
+    "openai": "^6.19.0",
+    "p-limit": "^7.3.0",
+    "turndown": "^7.2.2",
+    "turndown-plugin-gfm": "^1.0.2",
+    "zod": "^4.3.6"
+  },
+  "devDependencies": {
+    "@types/express": "^5.0.6",
+    "@types/node": "^25.2.2",
+    "@types/turndown": "^5.0.6",
+    "tsup": "^8.5.1",
+    "typescript": "^5.9.3",
+    "vitest": "^4.0.18"
+  }
+}