scrapebadger 0.24.2 → 0.26.0

package/dist/index.js

@@ -137,6 +137,39 @@ var BaseClient = class _BaseClient {
     const { data } = await this.requestRaw(path, options);
     return data;
   }
+  /**
+   * POST and return the undecoded response body.
+   *
+   * For endpoints that answer with something other than JSON — currently
+   * `/v1/web/scrape` with `raw_content: true`, which returns the scraped body
+   * itself. The normal path funnels a non-JSON response into
+   * `{ detail: await response.text() }`, which both loses the result and, for
+   * a binary payload, corrupts it: `text()` decodes bytes as UTF-8.
+   *
+   * Returns the raw bytes plus the response headers.
+   */
+  async postBinary(path, options = {}) {
+    const url = new URL(path, this.config.baseUrl);
+    const { body, headers = {} } = options;
+    const response = await this.fetchWithTimeout(url.toString(), {
+      method: "POST",
+      headers: {
+        "Content-Type": "application/json",
+        "X-API-Key": this.config.apiKey,
+        "User-Agent": `scrapebadger-node/${SDK_VERSION}`,
+        ...headers
+      },
+      body: body ? JSON.stringify(body) : void 0
+    });
+    if (!response.ok) {
+      await this.handleResponse(response);
+    }
+    return {
+      bytes: new Uint8Array(await response.arrayBuffer()),
+      headers: response.headers,
+      status: response.status
+    };
+  }
   /**
    * Make an HTTP request and return both data and rate limit headers.
    */
@@ -2481,11 +2514,54 @@ var WebClient = class {
     if (options.aiPrompt !== void 0) body.ai_prompt = options.aiPrompt;
     if (options.rawContent !== void 0) body.raw_content = options.rawContent;
     if (options.skipBotDetection !== void 0) body.skip_bot_detection = options.skipBotDetection;
+    if (options.rawContent) {
+      return this.scrapeRaw(body);
+    }
     return this.client.request("/v1/web/scrape", {
       method: "POST",
       body
     });
   }
+  /**
+   * Run a `rawContent` scrape, whose response is not JSON.
+   *
+   * The normal path funnels a non-JSON response into `{ detail: text }`, so a
+   * raw scrape returned a result with no content — and for a binary target,
+   * `response.text()` decoded the bytes as UTF-8 and destroyed them. Read the
+   * body as bytes and rebuild the metadata from the `X-Scrape-*` headers the
+   * server sends in this mode.
+   */
+  async scrapeRaw(body) {
+    const { bytes, headers, status } = await this.client.postBinary("/v1/web/scrape", {
+      body
+    });
+    const int = (name) => {
+      const parsed = Number.parseInt(headers.get(name) ?? "", 10);
+      return Number.isNaN(parsed) ? 0 : parsed;
+    };
+    const mediaType = ((headers.get("content-type") ?? "").split(";")[0] ?? "").trim().toLowerCase();
+    const isText = mediaType.startsWith("text/") || ["application/json", "application/xml", "image/svg+xml"].includes(mediaType);
+    return {
+      success: headers.get("x-scrape-success") !== "0",
+      url: headers.get("x-scrape-url") ?? (typeof body.url === "string" ? body.url : ""),
+      status_code: int("x-scrape-status-code") || status,
+      content: isText ? new TextDecoder().decode(bytes) : null,
+      content_bytes: bytes,
+      content_base64: null,
+      is_binary: !isText,
+      content_type: mediaType || null,
+      format: headers.get("x-scrape-format") ?? "html",
+      engine_used: headers.get("x-scrape-engine") ?? "",
+      credits_used: int("x-credits-used"),
+      duration_ms: int("x-scrape-duration-ms"),
+      retries_used: int("x-scrape-retries"),
+      content_length: int("x-scrape-content-length") || bytes.length,
+      screenshot_url: null,
+      video_url: null,
+      headers: {},
+      blocking_detected: false
+    };
+  }
   /**
    * Extract structured data from a web page using AI.
    *
@@ -6750,6 +6826,150 @@ var LinkedInClient = class {
   }
 };
 
+// src/chatgpt/ask.ts
+var AskClient = class {
+  client;
+  constructor(client) {
+    this.client = client;
+  }
+  /**
+   * Ask ChatGPT a question and get the answer with its sources.
+   *
+   * Costs 20 credits. Typical latency is 20-25s ungrounded, 30-70s with web search.
+   *
+   * @param params - Ask parameters.
+   * @param params.prompt - The prompt to send (max 4096 characters).
+   * @param params.country - ISO-3166 alpha-2 egress country (default: "US").
+   * @param params.web_search - "auto", "force", or "off" (default: "auto").
+   * @returns The answer, its citations, and the full retrieved search set.
+   *
+   * @example
+   * ```typescript
+   * const result = await client.chatgpt.ask.ask({
+   *   prompt: "what is the best CRM for a 10-person startup?",
+   *   country: "GB",
+   *   web_search: "force",
+   * });
+   * console.log(result.web_search_triggered, result.model);
+   * for (const source of result.search_results) {
+   *   console.log(`${source.cited ? "*" : " "} ${source.url}`);
+   * }
+   * ```
+   */
+  async ask(params) {
+    return this.client.request("/v1/chatgpt/ask", {
+      params: {
+        prompt: params.prompt,
+        country: params.country,
+        web_search: params.web_search
+      }
+    });
+  }
+};
+
+// src/chatgpt/brand.ts
+var BrandClient = class {
+  client;
+  constructor(client) {
+    this.client = client;
+  }
+  /**
+   * Analyse how a brand shows up in ChatGPT's answer to a prompt.
+   *
+   * Costs 25 credits.
+   *
+   * @param params - Brand-visibility parameters.
+   * @param params.prompt - The prompt to send (max 4096 characters).
+   * @param params.brand - The brand name to look for in the answer.
+   * @param params.domain - The brand's domain, used to detect brand citations.
+   * @param params.aliases - Other spellings that should count as mentions.
+   * @param params.competitors - Competitors to measure share of voice against.
+   * @param params.country - ISO-3166 alpha-2 egress country (default: "US").
+   * @param params.web_search - "auto", "force", or "off" (default: "force").
+   * @returns The brand analysis plus the answer and its citations.
+   *
+   * @example
+   * ```typescript
+   * const result = await client.chatgpt.brand.visibility({
+   *   prompt: "which proxy provider should I use?",
+   *   brand: "ScrapeBadger",
+   *   domain: "scrapebadger.com",
+   *   aliases: ["Scrape Badger"],
+   *   competitors: ["Bright Data", "Oxylabs"],
+   *   country: "DE",
+   * });
+   * console.log(`position score: ${result.position_score}`);
+   * for (const competitor of result.competitors) {
+   *   console.log(`${competitor.name}: ${competitor.mention_count}`);
+   * }
+   * ```
+   */
+  async visibility(params) {
+    return this.client.request("/v1/chatgpt/brand-visibility", {
+      params: {
+        prompt: params.prompt,
+        brand: params.brand,
+        domain: params.domain,
+        aliases: params.aliases?.length ? params.aliases.join(",") : void 0,
+        competitors: params.competitors?.length ? params.competitors.join(",") : void 0,
+        country: params.country,
+        web_search: params.web_search
+      }
+    });
+  }
+};
+
+// src/chatgpt/reference.ts
+var ReferenceClient11 = class {
+  client;
+  constructor(client) {
+    this.client = client;
+  }
+  /**
+   * Get the models chatgpt.com currently offers.
+   *
+   * Costs 1 credit.
+   *
+   * @param params - Optional parameters.
+   * @param params.country - ISO-3166 alpha-2 egress country (default: "US").
+   * @returns The available models.
+   *
+   * @example
+   * ```typescript
+   * const result = await client.chatgpt.reference.models({ country: "GB" });
+   * console.log(`${result.count} models`);
+   * for (const model of result.models) {
+   *   console.log(`${model.slug}: ${model.max_tokens} tokens`);
+   * }
+   * ```
+   */
+  async models(params = {}) {
+    return this.client.request("/v1/chatgpt/models", {
+      params: { country: params.country }
+    });
+  }
+};
+
+// src/chatgpt/client.ts
+var ChatGPTClient = class {
+  /** Client for asking ChatGPT a question */
+  ask;
+  /** Client for AEO/GEO brand-visibility analysis */
+  brand;
+  /** Client for reference data (available models) */
+  reference;
+  /**
+   * Create a new ChatGPT client.
+   *
+   * @param client - The base HTTP client for making requests.
+   */
+  constructor(client) {
+    this.ask = new AskClient(client);
+    this.brand = new BrandClient(client);
+    this.reference = new ReferenceClient11(client);
+  }
+};
+
 // src/client.ts
 var ScrapeBadger = class {
   baseClient;
@@ -6789,6 +7009,8 @@ var ScrapeBadger = class {
   depop;
   /** LinkedIn scraper API client — 11 no-auth endpoints (jobs, company, school, profile, post, article, learning, geo) */
   linkedin;
+  /** ChatGPT scraper API client — ask, brand visibility, models (the real chatgpt.com, anonymous) */
+  chatgpt;
   /**
    * Create a new ScrapeBadger client.
    *
@@ -6841,6 +7063,7 @@ var ScrapeBadger = class {
     this.loopnet = new LoopNetClient(this.baseClient);
     this.depop = new DepopClient(this.baseClient);
     this.linkedin = new LinkedInClient(this.baseClient);
+    this.chatgpt = new ChatGPTClient(this.baseClient);
   }
 };
 
@@ -6852,6 +7075,10 @@ exports.AmazonReferenceClient = ReferenceClient2;
 exports.AmazonSearchClient = SearchClient4;
 exports.AmazonSellersClient = SellersClient;
 exports.AuthenticationError = AuthenticationError;
+exports.ChatGPTAskClient = AskClient;
+exports.ChatGPTBrandClient = BrandClient;
+exports.ChatGPTClient = ChatGPTClient;
+exports.ChatGPTReferenceClient = ReferenceClient11;
 exports.CommunitiesClient = CommunitiesClient;
 exports.ConflictError = ConflictError;
 exports.DepopClient = DepopClient;