@endday/search-mcp 1.0.0 → 1.0.2

package/src/mcp/local/search.js

@@ -1,233 +0,0 @@
-import { env } from "../../../envs.js";
-import { ApiError } from "../../core/errors.js";
-import { searchAllWithMeta } from "../../search/gateway.js";
-
-const LATIN_QUERY_RE = /^[\p{Script=Latin}\p{Number}\s'".,!?():/_+-]+$/u;
-const DEFAULT_NEWS_ENGINES = ["bing", "brave", "yahoo"];
-
-function normalizeSourceTypeList(value) {
-  const items = Array.isArray(value)
-    ? value
-    : String(value || "")
-        .split(",")
-        .map((item) => item.trim());
-
-  return [
-    ...new Set(
-      items
-        .map((item) => String(item || "").trim().toLowerCase())
-        .filter(Boolean)
-    ),
-  ];
-}
-
-function normalizeMinAuthorityScore(value) {
-  if (value === undefined || value === null || value === "") {
-    return null;
-  }
-
-  const parsed = Number.parseFloat(value);
-  return Number.isFinite(parsed) ? parsed : null;
-}
-
-function normalizeSourceFilters(options) {
-  const includeSourceTypes = normalizeSourceTypeList(options.include_source_types);
-  const excludeSourceTypes = normalizeSourceTypeList(options.exclude_source_types);
-  const minAuthorityScore = normalizeMinAuthorityScore(options.min_authority_score);
-
-  return {
-    include_source_types: includeSourceTypes,
-    exclude_source_types: excludeSourceTypes,
-    min_authority_score: minAuthorityScore,
-    active:
-      includeSourceTypes.length > 0 ||
-      excludeSourceTypes.length > 0 ||
-      minAuthorityScore !== null,
-  };
-}
-
-function resultMatchesSourceFilters(result, filters) {
-  if (!filters.active) {
-    return true;
-  }
-
-  const sourceType = String(result.source_type || "unknown").toLowerCase();
-  const authorityScore = Number.isFinite(result.authority_score)
-    ? result.authority_score
-    : 0;
-
-  if (
-    filters.include_source_types.length > 0 &&
-    !filters.include_source_types.includes(sourceType)
-  ) {
-    return false;
-  }
-
-  if (filters.exclude_source_types.includes(sourceType)) {
-    return false;
-  }
-
-  if (
-    filters.min_authority_score !== null &&
-    authorityScore < filters.min_authority_score
-  ) {
-    return false;
-  }
-
-  return true;
-}
-
-function applySourceFilters(response, filters) {
-  if (!filters.active) {
-    return response;
-  }
-
-  const results = response.results.filter((result) =>
-    resultMatchesSourceFilters(result, filters)
-  );
-
-  return {
-    ...response,
-    number_of_results: results.length,
-    source_filters: filters,
-    results,
-  };
-}
-
-function inferLanguageFromQuery(query, fallbackLanguage) {
-  const normalizedQuery = String(query || "");
-
-  if (/[\u3040-\u30ff]/u.test(normalizedQuery)) {
-    return "ja-JP";
-  }
-
-  if (/[\uac00-\ud7af]/u.test(normalizedQuery)) {
-    return "ko-KR";
-  }
-
-  if (/[\u3400-\u4dbf\u4e00-\u9fff\uf900-\ufaff]/u.test(normalizedQuery)) {
-    return "zh-CN";
-  }
-
-  if (LATIN_QUERY_RE.test(normalizedQuery) && /[a-z]/i.test(normalizedQuery)) {
-    return "en-US";
-  }
-
-  return fallbackLanguage;
-}
-
-function resolveLanguage(query, language) {
-  return language || inferLanguageFromQuery(query, env.DEFAULT_LANGUAGE);
-}
-
-function isChineseLanguage(language) {
-  return String(language || "").trim().toLowerCase().startsWith("zh");
-}
-
-function resolveRequestedEngines(engines, language, vertical = "web") {
-  if (Array.isArray(engines) && engines.length > 0) {
-    return engines;
-  }
-
-  if (vertical === "news") {
-    return DEFAULT_NEWS_ENGINES;
-  }
-
-  return isChineseLanguage(language)
-    ? env.DEFAULT_ENGINES_ZH
-    : env.DEFAULT_ENGINES_NON_ZH;
-}
-
-function appendLocationToQuery(query, location) {
-  const normalizedLocation = String(location || "").trim();
-  if (!normalizedLocation) {
-    return query;
-  }
-
-  if (String(query).toLowerCase().includes(normalizedLocation.toLowerCase())) {
-    return query;
-  }
-
-  return `${query} ${normalizedLocation}`;
-}
-
-function createRuntimeContext(query, clientId) {
-  return {
-    request: new Request(
-      `https://mcp.local/search?q=${encodeURIComponent(query)}`,
-      {
-        method: "GET",
-        headers: {
-          "x-mcp-client-id": clientId,
-        },
-      }
-    ),
-  };
-}
-
-function paginateResponse(response, options) {
-  const offset = Math.max(0, Number.parseInt(options.offset ?? "0", 10) || 0);
-  const count =
-    options.count === undefined || options.count === null || options.count === ""
-      ? null
-      : Math.max(1, Number.parseInt(options.count, 10) || 0);
-
-  if (offset === 0 && count === null) {
-    return response;
-  }
-
-  const results =
-    count === null
-      ? response.results.slice(offset)
-      : response.results.slice(offset, offset + count);
-
-  return {
-    ...response,
-    number_of_results: results.length,
-    offset,
-    count,
-    results,
-  };
-}
-
-export async function searchLocal(query, engines = null, options = {}) {
-  const normalizedQuery = String(query || "").trim();
-  if (!normalizedQuery) {
-    throw new ApiError({
-      status: 400,
-      code: "MISSING_QUERY",
-      category: "validation",
-      message: "query required",
-    });
-  }
-
-  const vertical = String(options.vertical || "web").trim().toLowerCase() || "web";
-  const language = resolveLanguage(
-    normalizedQuery,
-    options.search_lang || options.ui_lang || options.language
-  );
-  const requestedEngines = resolveRequestedEngines(engines, language, vertical);
-  const effectiveQuery = appendLocationToQuery(normalizedQuery, options.location);
-  const filters = normalizeSourceFilters(options);
-  const clientId = String(options.clientId || "mcp-local").trim() || "mcp-local";
-  const runtimeContext = createRuntimeContext(effectiveQuery, clientId);
-
-  const { response } = await searchAllWithMeta({
-    vertical,
-    query: effectiveQuery,
-    engines: requestedEngines,
-    language,
-    time_range: options.time_range,
-    pageno: options.pageno,
-    clientId,
-    runtimeContext,
-  });
-
-  return {
-    ...paginateResponse(applySourceFilters(response, filters), options),
-    query: normalizedQuery,
-    effective_query: effectiveQuery,
-    location: options.location || null,
-    location_source: options.location ? "explicit" : "disabled",
-  };
-}

package/src/mcp/schemas.js

@@ -1,132 +0,0 @@
-import { MAX_CHARS_DEFAULT, MAX_CHARS_MAX, MAX_CHARS_MIN } from "./format.js";
-
-function createSearchProperties(allEngines, options = {}) {
-  const properties = {
-    query: {
-      type: "string",
-      description: "Search query",
-    },
-    engines: {
-      type: "array",
-      items: {
-        type: "string",
-        enum: allEngines,
-      },
-      description: `Engines to query. Default: ${allEngines.join(", ")}.`,
-    },
-    language: {
-      type: "string",
-      description: "Language hint (e.g. en, zh-CN)",
-    },
-    search_lang: {
-      type: "string",
-      description: "Search language hint (e.g. en, zh-CN)",
-    },
-    ui_lang: {
-      type: "string",
-      description: "UI locale hint (e.g. en-US, zh-CN)",
-    },
-    time_range: {
-      type: "string",
-      enum: ["day", "week", "month", "year"],
-      description: "Time range filter",
-    },
-    count: {
-      type: "integer",
-      minimum: 1,
-      description: "Max number of results to return after ranking",
-    },
-    offset: {
-      type: "integer",
-      minimum: 0,
-      description: "Result offset to apply after ranking",
-    },
-    min_authority_score: {
-      type: "number",
-      description: "Min source authority score",
-    },
-    include_source_types: {
-      type: "array",
-      items: { type: "string" },
-      description: "Source types to include",
-    },
-    exclude_source_types: {
-      type: "array",
-      items: { type: "string" },
-      description: "Source types to exclude",
-    },
-  };
-
-  if (options.includeLocation !== false) {
-    properties.location = {
-      type: "string",
-      description: "Location hint for query enrichment",
-    };
-  }
-
-  if (options.includePage !== false) {
-    properties.pageno = {
-      type: "integer",
-      minimum: 0,
-      description: "Page number (0-based)",
-    };
-  }
-
-  return properties;
-}
-
-export function createSearchInputSchema(allEngines) {
-  return {
-    type: "object",
-    properties: createSearchProperties(allEngines, {
-      includeLocation: true,
-      includePage: true,
-    }),
-    required: ["query"],
-  };
-}
-
-export function createNewsSearchInputSchema(allEngines) {
-  return {
-    type: "object",
-    properties: createSearchProperties(allEngines, {
-      includeLocation: false,
-      includePage: false,
-    }),
-    required: ["query"],
-  };
-}
-
-export const CONTENT_INPUT_SCHEMA = {
-  type: "object",
-  properties: {
-    url: {
-      type: "string",
-      description: "URL to extract content from",
-    },
-    max_chars: {
-      type: "integer",
-      minimum: MAX_CHARS_MIN,
-      maximum: MAX_CHARS_MAX,
-      description: `Max text chars to return. Default: ${MAX_CHARS_DEFAULT}.`,
-    },
-  },
-  required: ["url"],
-};
-
-export const JINA_CONTENT_INPUT_SCHEMA = {
-  type: "object",
-  properties: {
-    url: {
-      type: "string",
-      description: "URL to extract content from via Jina AI reader",
-    },
-    max_chars: {
-      type: "integer",
-      minimum: MAX_CHARS_MIN,
-      maximum: MAX_CHARS_MAX,
-      description: `Max text chars to return. Default: ${MAX_CHARS_DEFAULT}.`,
-    },
-  },
-  required: ["url"],
-};

package/src/mcp/server.js

@@ -1,97 +0,0 @@
-import { Server } from "@modelcontextprotocol/sdk/server/index.js";
-import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
-import {
-  CallToolRequestSchema,
-  ListToolsRequestSchema,
-} from "@modelcontextprotocol/sdk/types.js";
-
-import {
-  CONTENT_INPUT_SCHEMA,
-  createNewsSearchInputSchema,
-  createSearchInputSchema,
-  JINA_CONTENT_INPUT_SCHEMA,
-} from "./schemas.js";
-import { getSupportedEnginesForVertical } from "../search/engineRegistry.js";
-import { handleWebSearch } from "./tools/webSearch.js";
-import { handleNewsSearch } from "./tools/newsSearch.js";
-import { handleContent } from "./tools/content.js";
-import { handleJinaContent } from "./tools/jinaContent.js";
-
-export function createServer(config) {
-  const newsEngineSet = new Set(getSupportedEnginesForVertical("news"));
-  const newsEngines =
-    config.newsEngines ||
-    config.allEngines.filter((engineName) => newsEngineSet.has(engineName));
-  const server = new Server(
-    { name: "search-mcp", version: "2.0.0" },
-    { capabilities: { tools: {} } }
-  );
-
-  const TOOL_LIST = {
-    tools: [
-      {
-        name: "web_search",
-        description:
-          "Search the web with local-first MCP execution. Returns deduplicated results with snippets and source authority scores.",
-        inputSchema: createSearchInputSchema(config.allEngines),
-      },
-      {
-        name: "news_search",
-        description:
-          "Search recent news with an explicit news-search capability. Use this instead of web_search when the caller needs news results rather than general web pages.",
-        inputSchema: createNewsSearchInputSchema(newsEngines),
-      },
-      {
-        name: "content",
-        description:
-          "Extract readable text from a URL with local-first MCP execution. Returns title, URL, and extracted content.",
-        inputSchema: CONTENT_INPUT_SCHEMA,
-      },
-      {
-        name: "jina_content",
-        description:
-          "Extract readable text from a URL using Jina AI reader. Per-user rate limit unless JINA_API_KEY is configured.",
-        inputSchema: JINA_CONTENT_INPUT_SCHEMA,
-      },
-    ],
-  };
-
-  const VALID_TOOL_NAMES = new Set([
-    "web_search",
-    "news_search",
-    "content",
-    "jina_content",
-  ]);
-
-  server.setRequestHandler(ListToolsRequestSchema, async () => TOOL_LIST);
-
-  server.setRequestHandler(CallToolRequestSchema, async (request) => {
-    const { name } = request.params;
-    if (!VALID_TOOL_NAMES.has(name)) {
-      throw new Error(`Unknown tool: ${name}`);
-    }
-
-    const args = request.params.arguments || {};
-
-    if (name === "web_search") {
-      return handleWebSearch(config, args);
-    }
-
-    if (name === "news_search") {
-      return handleNewsSearch(config, args);
-    }
-
-    if (name === "content") {
-      return handleContent(config, args);
-    }
-
-    return handleJinaContent(config, args);
-  });
-
-  return server;
-}
-
-export async function startServer(server) {
-  const transport = new StdioServerTransport();
-  await server.connect(transport);
-}

package/src/mcp/tools/content.js

@@ -1,31 +0,0 @@
-import {
-  formatContentResponse,
-  normalizeMaxChars,
-} from "../format.js";
-import { contentLocal, requireUrl } from "../local/content.js";
-
-function estimateContentMaxBytes(maxChars) {
-  return Math.max(1000000, maxChars * 16 + 200000);
-}
-
-async function executeContent(config, url, estimatedMaxBytes) {
-  return contentLocal(url, { max_bytes: estimatedMaxBytes });
-}
-
-export async function handleContent(config, args) {
-  const url = requireUrl(args);
-  const maxChars = normalizeMaxChars(args.max_chars);
-  const estimatedMaxBytes = estimateContentMaxBytes(maxChars);
-
-  try {
-    const result = await executeContent(config, url, estimatedMaxBytes);
-    return {
-      content: [{ type: "text", text: formatContentResponse(result, maxChars) }],
-    };
-  } catch (error) {
-    return {
-      content: [{ type: "text", text: `Content failed: ${error.message}` }],
-      isError: true,
-    };
-  }
-}

package/src/mcp/tools/jinaContent.js

@@ -1,38 +0,0 @@
-import { normalizeMaxChars, truncateText } from "../format.js";
-import { requireUrl } from "../local/content.js";
-
-async function jinaContentAPI(config, targetUrl) {
-  const encodedUrl = encodeURIComponent(targetUrl);
-  const response = await fetch(`${config.jinaBaseUrl}${encodedUrl}`, {
-    headers: config.jinaApiKey
-      ? { Authorization: `Bearer ${config.jinaApiKey}` }
-      : {},
-  });
-
-  if (!response.ok) {
-    const body = await response.text().catch(() => "");
-    throw new Error(`Jina failed: ${response.status} ${body}`.trim());
-  }
-
-  return {
-    url: targetUrl,
-    text: await response.text(),
-  };
-}
-
-export async function handleJinaContent(config, args) {
-  const url = requireUrl(args);
-  const maxChars = normalizeMaxChars(args.max_chars);
-
-  try {
-    const result = await jinaContentAPI(config, url);
-    return {
-      content: [{ type: "text", text: truncateText(String(result.text || ""), maxChars) }],
-    };
-  } catch (error) {
-    return {
-      content: [{ type: "text", text: error.message }],
-      isError: true,
-    };
-  }
-}

package/src/mcp/tools/newsSearch.js

@@ -1,22 +0,0 @@
-import { formatSearchResponse } from "../format.js";
-import { executeSearch } from "./webSearch.js";
-
-export async function handleNewsSearch(config, args) {
-  if (!args.query || typeof args.query !== "string") {
-    throw new Error("query required");
-  }
-
-  try {
-    const result = await executeSearch(config, args, {
-      vertical: "news",
-    });
-    return {
-      content: [{ type: "text", text: formatSearchResponse(result) }],
-    };
-  } catch (error) {
-    return {
-      content: [{ type: "text", text: `News search failed: ${error.message}` }],
-      isError: true,
-    };
-  }
-}

package/src/mcp/tools/webSearch.js

@@ -1,57 +0,0 @@
-import { formatSearchResponse } from "../format.js";
-import { searchLocal } from "../local/search.js";
-
-export function buildSearchOptions(args, overrides = {}) {
-  const searchOptions = {
-    vertical: overrides.vertical || "web",
-    language: args.language,
-    search_lang: args.search_lang,
-    ui_lang: args.ui_lang,
-    location: args.location,
-    time_range: args.time_range,
-    pageno: args.pageno,
-    count: args.count,
-    offset: args.offset,
-    clientId: args.client_id,
-  };
-
-  if (Number.isFinite(args.min_authority_score)) {
-    searchOptions.min_authority_score = args.min_authority_score;
-  }
-  if (Array.isArray(args.include_source_types)) {
-    searchOptions.include_source_types = args.include_source_types;
-  }
-  if (Array.isArray(args.exclude_source_types)) {
-    searchOptions.exclude_source_types = args.exclude_source_types;
-  }
-
-  return searchOptions;
-}
-
-export async function executeSearch(config, args, overrides = {}) {
-  const engines = Array.isArray(args.engines) && args.engines.length > 0 ? args.engines : null;
-  const searchOptions = {
-    ...buildSearchOptions(args, overrides),
-    clientId: args.client_id || config.localClientId,
-  };
-
-  return searchLocal(args.query, engines, searchOptions);
-}
-
-export async function handleWebSearch(config, args) {
-  if (!args.query || typeof args.query !== "string") {
-    throw new Error("query required");
-  }
-
-  try {
-    const result = await executeSearch(config, args);
-    return {
-      content: [{ type: "text", text: formatSearchResponse(result) }],
-    };
-  } catch (error) {
-    return {
-      content: [{ type: "text", text: `Search failed: ${error.message}` }],
-      isError: true,
-    };
-  }
-}