@tool-bridge/server 0.21.1 → 0.23.0

package/dist/opengraph_io-ZQ2FGLHX.js

@@ -0,0 +1,362 @@
+import {
+  createProviderHttpClient
+} from "./chunk-VWRUGUSU.js";
+import {
+  asJsonObject,
+  booleanValue,
+  compactDefined,
+  integerValue,
+  trimmedText
+} from "./chunk-JXHMDBAY.js";
+import {
+  createProviderPlugin,
+  requireApiKey,
+  upstreamError
+} from "./chunk-WJHG6OK7.js";
+import {
+  assertPublicHttpUrl
+} from "./chunk-2JNAI4OC.js";
+import "./chunk-BADVCN6K.js";
+import {
+  TBError
+} from "./chunk-5ICRY3VU.js";
+import "./chunk-U67V476Y.js";
+
+// ../plugins/src/opengraph_io/api.ts
+var SERVICE = "opengraph_io";
+var API_BASE = "https://opengraph.io";
+var http = createProviderHttpClient({ baseUrl: `${API_BASE}/`, service: SERVICE });
+function errorMessage(payload) {
+  if (typeof payload === "string" && payload.trim() !== "") return payload;
+  const body = asJsonObject(payload);
+  if (body === void 0) return "OpenGraph.io request failed";
+  for (const key of ["error_description", "error_message", "error", "message", "detail"]) {
+    const found = trimmedText(body[key]);
+    if (found !== void 0) return found;
+  }
+  return "OpenGraph.io request failed";
+}
+async function request(ctx, path, query) {
+  const apiKey = requireApiKey(ctx, SERVICE);
+  const requestQuery = [["app_id", apiKey], ...Object.entries(query)];
+  const { data } = await http.request({
+    path,
+    query: requestQuery,
+    headers: { accept: "application/json" },
+    invalidJson: "text",
+    mapError: ({ data: payload, status }) => upstreamError(status, errorMessage(payload)),
+    mapTransportError: ({ message }) => upstreamError(
+      502,
+      message === void 0 ? "OpenGraph.io request failed" : `OpenGraph.io request failed: ${message}`
+    )
+  });
+  return data ?? null;
+}
+function unwrap(payload) {
+  const body = asJsonObject(payload);
+  if (body === void 0) return payload;
+  return typeof body.successful === "boolean" && body.data !== void 0 ? body.data : payload;
+}
+function requireObject(payload, context) {
+  const body = asJsonObject(unwrap(payload));
+  if (body === void 0) throw upstreamError(502, `OpenGraph.io ${context} response was not a JSON object`);
+  return body;
+}
+function normalizeRequestInfo(payload) {
+  const body = asJsonObject(payload);
+  if (body === void 0) return void 0;
+  const normalized = compactDefined({
+    host: trimmedText(body.host),
+    redirects: integerValue(body.redirects),
+    responseCode: integerValue(body.responseCode ?? body.response_code),
+    responseContentType: trimmedText(body.responseContentType ?? body.response_content_type)
+  });
+  return Object.keys(normalized).length > 0 ? normalized : void 0;
+}
+function normalizeSiteResult(payload) {
+  const body = requireObject(payload, "site");
+  const tags = Array.isArray(body.tags) ? body.tags.map((item) => asJsonObject(item)).filter((item) => item !== void 0) : void 0;
+  return compactDefined({
+    hybridGraph: asJsonObject(body.hybridGraph ?? body.hybrid_graph),
+    openGraph: asJsonObject(body.openGraph ?? body.open_graph),
+    twitterCard: asJsonObject(body.twitterCard ?? body.twitter_card),
+    htmlInferred: asJsonObject(body.htmlInferred ?? body.html_inferred),
+    oEmbed: asJsonObject(body.oEmbed ?? body.oembed),
+    requestUrl: trimmedText(body.requestUrl ?? body.request_url),
+    requestInfo: normalizeRequestInfo(body.requestInfo ?? body.request_info),
+    cached: booleanValue(body.cached),
+    // createdAt 的 null 是有信息的("非缓存结果"),不能被 compact 当缺失丢掉。
+    createdAt: (body.createdAt ?? body.created_at) === null ? null : trimmedText(body.createdAt ?? body.created_at),
+    retryInfo: asJsonObject(body.retryInfo ?? body.retry_info),
+    aiSafety: asJsonObject(body.aiSafety ?? body.ai_safety ?? body.AI_SAFETY),
+    domain: trimmedText(body.domain),
+    tags: tags !== void 0 && tags.length > 0 ? tags : void 0
+  });
+}
+function targetPath(family, target, field) {
+  let normalized;
+  try {
+    normalized = assertPublicHttpUrl(target).toString();
+  } catch {
+    throw new TBError("invalid_argument", `${field} \u5FC5\u987B\u662F\u516C\u7F51\u53EF\u8FBE\u7684 http(s) \u5730\u5740`);
+  }
+  return `/api/1.1/${family}/${encodeURIComponent(normalized)}`;
+}
+function siteQuery(input) {
+  return {
+    cache_ok: input.cacheOk,
+    full_render: input.fullRender,
+    use_proxy: input.useProxy,
+    use_premium: input.usePremium,
+    use_superior: input.useSuperior,
+    use_ai: input.useAi,
+    max_cache_age: input.maxCacheAge,
+    accept_lang: input.acceptLang,
+    auto_proxy: input.autoProxy,
+    auto_render: input.autoRender,
+    retry: input.retry,
+    max_retries: input.maxRetries,
+    retry_escalate: input.retryEscalate,
+    proxy_country: input.proxyCountry
+  };
+}
+async function extractSiteMetadata(input, ctx) {
+  const payload = await request(ctx, targetPath("site", input.site, "site"), siteQuery(input));
+  return normalizeSiteResult(payload);
+}
+async function extractSite(input, ctx) {
+  return extractSiteMetadata(input, ctx);
+}
+async function scrapeSite(input, ctx) {
+  return extractSiteMetadata(input, ctx);
+}
+async function scrapeUrl(input, ctx) {
+  const payload = await request(ctx, targetPath("scrape", input.url, "url"), {
+    cache_ok: input.cacheOk,
+    full_render: input.fullRender,
+    use_proxy: input.useProxy,
+    use_premium: input.usePremium,
+    use_superior: input.useSuperior,
+    accept_lang: input.acceptLang,
+    auto_proxy: input.autoProxy,
+    auto_render: input.autoRender,
+    retry: input.retry
+  });
+  const unwrapped = unwrap(payload);
+  if (typeof unwrapped === "string") return { htmlContent: unwrapped };
+  const body = requireObject(unwrapped, "scrape");
+  const htmlContent = trimmedText(body.htmlContent ?? body.html_content ?? body.html);
+  if (htmlContent === void 0) {
+    throw upstreamError(502, "OpenGraph.io scrape response did not include html content");
+  }
+  return compactDefined({
+    htmlContent,
+    requestInfo: normalizeRequestInfo(body.requestInfo ?? body.request_info),
+    retryInfo: asJsonObject(body.retryInfo ?? body.retry_info)
+  });
+}
+async function captureScreenshot(input, ctx) {
+  const payload = await request(ctx, targetPath("screenshot", input.url, "url"), {
+    format: input.format,
+    quality: input.quality,
+    cache_ok: input.cacheOk,
+    selector: input.selector,
+    dark_mode: input.darkMode,
+    full_page: input.fullPage,
+    use_proxy: input.useProxy,
+    dimensions: input.dimensions,
+    capture_delay: input.captureDelay,
+    exclude_selectors: input.excludeSelectors,
+    navigation_timeout: input.navigationTimeout,
+    block_cookie_banner: input.blockCookieBanner
+  });
+  const body = requireObject(payload, "screenshot");
+  const screenshotUrl = trimmedText(body.screenshotUrl ?? body.screenshot_url);
+  if (screenshotUrl === void 0) {
+    throw upstreamError(502, "OpenGraph.io screenshot response did not include screenshotUrl");
+  }
+  const dimensions = asJsonObject(body.dimensions);
+  const width = integerValue(dimensions?.width);
+  const height = integerValue(dimensions?.height);
+  if (width === void 0 || height === void 0) {
+    throw upstreamError(502, "OpenGraph.io screenshot response did not include dimensions");
+  }
+  return compactDefined({
+    screenshotUrl,
+    dimensions: { width, height },
+    requestInfo: normalizeRequestInfo(body.requestInfo ?? body.request_info)
+  });
+}
+
+// ../plugins/src/opengraph_io/schema.ts
+import { z } from "zod/v4";
+var extractSiteInput = z.strictObject({
+  site: z.url().describe("The site URL to inspect."),
+  cacheOk: z.boolean().describe("Whether cached results may be returned when available.").optional(),
+  fullRender: z.boolean().describe("Whether the page should be rendered in a browser before extraction.").optional(),
+  useProxy: z.boolean().describe("Whether a proxy should be used for the request.").optional(),
+  usePremium: z.boolean().describe("Whether a residential proxy should be used when available.").optional(),
+  useSuperior: z.boolean().describe("Whether a mobile-grade proxy should be used when available.").optional(),
+  useAi: z.boolean().describe("Whether AI-enhanced metadata extraction should be enabled.").optional(),
+  maxCacheAge: z.int().min(0).describe("The maximum accepted cache age in seconds.").optional(),
+  acceptLang: z.string().min(1).describe("The Accept-Language header value sent to the target site.").optional(),
+  autoProxy: z.boolean().describe("Whether OpenGraph.io may automatically decide whether to use a proxy.").optional(),
+  autoRender: z.boolean().describe("Whether OpenGraph.io may automatically decide whether full rendering is needed.").optional(),
+  retry: z.boolean().describe("Whether OpenGraph.io should retry with fallback transport strategies.").optional(),
+  maxRetries: z.int().min(0).describe("The maximum number of retry attempts.").optional(),
+  retryEscalate: z.boolean().describe("Whether retries may escalate to more expensive fallback strategies.").optional(),
+  proxyCountry: z.string().min(1).describe("The ISO 3166-1 alpha-2 country code to use for proxy egress.").optional()
+}).describe("The input payload for extracting OpenGraph.io site metadata.");
+var extractSiteOutput = z.looseObject({
+  hybridGraph: z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.").optional(),
+  openGraph: z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.").optional(),
+  twitterCard: z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.").optional(),
+  htmlInferred: z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.").optional(),
+  oEmbed: z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.").optional(),
+  requestUrl: z.string().describe("The final URL that OpenGraph.io resolved after redirects.").optional(),
+  requestInfo: z.looseObject({
+    host: z.string().describe("The host that responded to the request.").optional(),
+    redirects: z.int().describe("The number of redirects followed while fetching the URL.").optional(),
+    responseCode: z.int().describe("The HTTP response code returned upstream.").optional(),
+    responseContentType: z.string().describe("The upstream response Content-Type header when available.").optional()
+  }).describe("Information about the upstream request performed by OpenGraph.io.").optional(),
+  cached: z.boolean().describe("Whether the result came from cache.").optional(),
+  createdAt: z.string().describe("The ISO 8601 timestamp when the cached result was created.").nullable().optional(),
+  retryInfo: z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.").optional(),
+  aiSafety: z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.").optional(),
+  domain: z.string().describe("The domain extracted from the requested URL.").optional(),
+  tags: z.array(z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.")).describe("Additional tag-level extraction results when returned by the API.").optional()
+}).describe("The normalized metadata payload returned by the OpenGraph.io Site endpoint.");
+var scrapeSiteInput = z.strictObject({
+  site: z.url().describe("The site URL to inspect."),
+  cacheOk: z.boolean().describe("Whether cached results may be returned when available.").optional(),
+  fullRender: z.boolean().describe("Whether the page should be rendered in a browser before extraction.").optional(),
+  useProxy: z.boolean().describe("Whether a proxy should be used for the request.").optional(),
+  usePremium: z.boolean().describe("Whether a residential proxy should be used when available.").optional(),
+  useSuperior: z.boolean().describe("Whether a mobile-grade proxy should be used when available.").optional(),
+  useAi: z.boolean().describe("Whether AI-enhanced metadata extraction should be enabled.").optional(),
+  maxCacheAge: z.int().min(0).describe("The maximum accepted cache age in seconds.").optional(),
+  acceptLang: z.string().min(1).describe("The Accept-Language header value sent to the target site.").optional(),
+  autoProxy: z.boolean().describe("Whether OpenGraph.io may automatically decide whether to use a proxy.").optional(),
+  autoRender: z.boolean().describe("Whether OpenGraph.io may automatically decide whether full rendering is needed.").optional(),
+  retry: z.boolean().describe("Whether OpenGraph.io should retry with fallback transport strategies.").optional(),
+  maxRetries: z.int().min(0).describe("The maximum number of retry attempts.").optional(),
+  retryEscalate: z.boolean().describe("Whether retries may escalate to more expensive fallback strategies.").optional(),
+  proxyCountry: z.string().min(1).describe("The ISO 3166-1 alpha-2 country code to use for proxy egress.").optional(),
+  scrape: z.boolean().describe("A reserved compatibility flag. The current official Site endpoint does not expose a separate scrape mode, so this value is ignored.").optional()
+}).describe("The input payload for retrieving OpenGraph.io site metadata.");
+var scrapeSiteOutput = z.looseObject({
+  hybridGraph: z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.").optional(),
+  openGraph: z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.").optional(),
+  twitterCard: z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.").optional(),
+  htmlInferred: z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.").optional(),
+  oEmbed: z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.").optional(),
+  requestUrl: z.string().describe("The final URL that OpenGraph.io resolved after redirects.").optional(),
+  requestInfo: z.looseObject({
+    host: z.string().describe("The host that responded to the request.").optional(),
+    redirects: z.int().describe("The number of redirects followed while fetching the URL.").optional(),
+    responseCode: z.int().describe("The HTTP response code returned upstream.").optional(),
+    responseContentType: z.string().describe("The upstream response Content-Type header when available.").optional()
+  }).describe("Information about the upstream request performed by OpenGraph.io.").optional(),
+  cached: z.boolean().describe("Whether the result came from cache.").optional(),
+  createdAt: z.string().describe("The ISO 8601 timestamp when the cached result was created.").nullable().optional(),
+  retryInfo: z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.").optional(),
+  aiSafety: z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.").optional(),
+  domain: z.string().describe("The domain extracted from the requested URL.").optional(),
+  tags: z.array(z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.")).describe("Additional tag-level extraction results when returned by the API.").optional()
+}).describe("The normalized metadata payload returned by the OpenGraph.io Site endpoint.");
+var scrapeUrlInput = z.strictObject({
+  url: z.url().describe("The page URL to scrape for raw HTML."),
+  cacheOk: z.boolean().describe("Whether cached results may be returned when available.").optional(),
+  fullRender: z.boolean().describe("Whether the page should be rendered in a browser before scraping.").optional(),
+  useProxy: z.boolean().describe("Whether a proxy should be used for the request.").optional(),
+  usePremium: z.boolean().describe("Whether a residential proxy should be used when available.").optional(),
+  useSuperior: z.boolean().describe("Whether a mobile-grade proxy should be used when available.").optional(),
+  acceptLang: z.string().min(1).describe("The Accept-Language header value sent to the target site.").optional(),
+  autoProxy: z.boolean().describe("Whether OpenGraph.io may automatically decide whether to use a proxy.").optional(),
+  autoRender: z.boolean().describe("Whether OpenGraph.io may automatically decide whether full rendering is needed.").optional(),
+  retry: z.boolean().describe("Whether OpenGraph.io should retry with fallback transport strategies.").optional()
+}).describe("The input payload for scraping a URL with OpenGraph.io.");
+var scrapeUrlOutput = z.looseObject({
+  htmlContent: z.string().describe("The raw HTML content returned for the page.").optional(),
+  requestInfo: z.looseObject({
+    host: z.string().describe("The host that responded to the request.").optional(),
+    redirects: z.int().describe("The number of redirects followed while fetching the URL.").optional(),
+    responseCode: z.int().describe("The HTTP response code returned upstream.").optional(),
+    responseContentType: z.string().describe("The upstream response Content-Type header when available.").optional()
+  }).describe("Information about the upstream request performed by OpenGraph.io.").optional(),
+  retryInfo: z.looseObject({}).describe("A loose JSON object returned by OpenGraph.io.").optional()
+}).describe("The normalized output payload returned by the OpenGraph.io Scrape endpoint.");
+var captureScreenshotInput = z.strictObject({
+  url: z.url().describe("The page URL to capture as an image."),
+  format: z.enum(["jpeg", "png", "webp"]).describe("The output image format.").optional(),
+  quality: z.int().min(10).max(80).describe("The image quality from 10 to 80.").optional(),
+  cacheOk: z.boolean().describe("Whether cached screenshots may be returned when available.").optional(),
+  selector: z.string().min(1).describe("An optional CSS selector that limits the capture to a specific element.").optional(),
+  darkMode: z.boolean().describe("Whether the page should be rendered with a dark color-scheme preference.").optional(),
+  fullPage: z.boolean().describe("Whether the entire scrollable page should be captured.").optional(),
+  useProxy: z.boolean().describe("Whether a proxy should be used for the request.").optional(),
+  dimensions: z.enum(["xs", "sm", "md", "lg"]).describe("A viewport size preset.").optional(),
+  captureDelay: z.int().min(0).max(1e4).describe("The delay in milliseconds before taking the screenshot.").optional(),
+  excludeSelectors: z.string().min(1).describe("Comma-separated CSS selectors for elements that should be hidden.").optional(),
+  navigationTimeout: z.int().min(1e3).max(6e4).describe("The navigation timeout in milliseconds.").optional(),
+  blockCookieBanner: z.boolean().describe("Whether known cookie consent banners should be blocked.").optional()
+}).describe("The input payload for capturing a screenshot with OpenGraph.io.");
+var captureScreenshotOutput = z.looseObject({
+  screenshotUrl: z.url().describe("The URL of the generated screenshot image.").optional(),
+  dimensions: z.strictObject({
+    width: z.int().describe("The screenshot width in pixels.").optional(),
+    height: z.int().describe("The screenshot height in pixels.").optional()
+  }).describe("The dimensions of the captured screenshot.").optional(),
+  requestInfo: z.looseObject({
+    host: z.string().describe("The host that responded to the request.").optional(),
+    redirects: z.int().describe("The number of redirects followed while fetching the URL.").optional(),
+    responseCode: z.int().describe("The HTTP response code returned upstream.").optional(),
+    responseContentType: z.string().describe("The upstream response Content-Type header when available.").optional()
+  }).describe("Information about the upstream request performed by OpenGraph.io.").optional()
+}).describe("The normalized output payload returned by the OpenGraph.io Screenshot endpoint.");
+var opengraphIoActions = {
+  extract_site: {
+    description: "Extract Open Graph, Twitter Card, oEmbed, and inferred metadata for a site through the OpenGraph.io Site endpoint.",
+    effect: "write",
+    inputSchema: extractSiteInput,
+    outputSchema: z.toJSONSchema(extractSiteOutput, { io: "output", unrepresentable: "any" })
+  },
+  scrape_site: {
+    description: "Retrieve a site's metadata through the OpenGraph.io Site endpoint with cache, proxy, render, and retry controls.",
+    effect: "write",
+    inputSchema: scrapeSiteInput,
+    outputSchema: z.toJSONSchema(scrapeSiteOutput, { io: "output", unrepresentable: "any" })
+  },
+  scrape_url: {
+    description: "Fetch the raw HTML for a page through the OpenGraph.io Scrape endpoint with optional render and proxy controls.",
+    effect: "write",
+    inputSchema: scrapeUrlInput,
+    outputSchema: z.toJSONSchema(scrapeUrlOutput, { io: "output", unrepresentable: "any" })
+  },
+  capture_screenshot: {
+    description: "Capture a webpage screenshot through the OpenGraph.io Screenshot endpoint with configurable viewport, delay, and element selection.",
+    effect: "write",
+    inputSchema: captureScreenshotInput,
+    outputSchema: z.toJSONSchema(captureScreenshotOutput, { io: "output", unrepresentable: "any" })
+  }
+};
+
+// ../plugins/src/opengraph_io/index.ts
+function createOpengraphIoPlugin() {
+  return createProviderPlugin({
+    description: "OpenGraph.io",
+    actions: opengraphIoActions,
+    handlers: {
+      extract_site: extractSite,
+      scrape_site: scrapeSite,
+      scrape_url: scrapeUrl,
+      capture_screenshot: captureScreenshot
+    }
+  });
+}
+var opengraph_io_default = createOpengraphIoPlugin();
+export {
+  createOpengraphIoPlugin,
+  opengraph_io_default as default
+};