@mendable/firecrawl 1.25.6 → 1.29.0

package/dist/index.cjs

@@ -35,7 +35,7 @@ var require_package = __commonJS({
   "package.json"(exports2, module2) {
     module2.exports = {
       name: "@mendable/firecrawl-js",
-      version: "1.25.6",
+      version: "1.29.0",
       description: "JavaScript SDK for Firecrawl API",
       main: "dist/index.js",
       types: "dist/index.d.ts",
@@ -109,7 +109,7 @@ __export(src_exports, {
 });
 module.exports = __toCommonJS(src_exports);
 var import_axios = __toESM(require("axios"), 1);
-var zt = __toESM(require("zod"), 1);
+var zt = require("zod");
 var import_zod_to_json_schema = require("zod-to-json-schema");
 
 // node_modules/typescript-event-target/dist/index.mjs
@@ -713,10 +713,12 @@ var FirecrawlApp = class {
     try {
       if (!params?.schema) {
         jsonSchema = void 0;
-      } else if (typeof params.schema === "object" && params.schema !== null && Object.getPrototypeOf(params.schema)?.constructor?.name?.startsWith("Zod")) {
-        jsonSchema = (0, import_zod_to_json_schema.zodToJsonSchema)(params.schema);
       } else {
-        jsonSchema = params.schema;
+        try {
+          jsonSchema = (0, import_zod_to_json_schema.zodToJsonSchema)(params.schema);
+        } catch (_) {
+          jsonSchema = params.schema;
+        }
       }
     } catch (error) {
       throw new FirecrawlError("Invalid schema. Schema must be either a valid Zod schema or JSON schema object.", 400);
@@ -773,10 +775,14 @@ var FirecrawlApp = class {
     let jsonData = { urls, ...params };
     let jsonSchema;
     try {
-      if (params?.schema instanceof zt.ZodType) {
-        jsonSchema = (0, import_zod_to_json_schema.zodToJsonSchema)(params.schema);
+      if (!params?.schema) {
+        jsonSchema = void 0;
       } else {
-        jsonSchema = params?.schema;
+        try {
+          jsonSchema = (0, import_zod_to_json_schema.zodToJsonSchema)(params.schema);
+        } catch (_) {
+          jsonSchema = params.schema;
+        }
       }
     } catch (error) {
       throw new FirecrawlError("Invalid schema. Schema must be either a valid Zod schema or JSON schema object.", 400);

package/dist/index.d.cts

@@ -120,6 +120,7 @@ interface CrawlScrapeOptions {
     proxy?: "basic" | "stealth" | "auto";
     storeInCache?: boolean;
     maxAge?: number;
+    parsePDF?: boolean;
 }
 type Action = {
     type: "wait";
@@ -132,6 +133,7 @@ type Action = {
 } | {
     type: "screenshot";
     fullPage?: boolean;
+    quality?: number;
 } | {
     type: "write";
     text: string;
@@ -199,6 +201,7 @@ interface CrawlParams {
     maxDiscoveryDepth?: number;
     limit?: number;
     allowBackwardLinks?: boolean;
+    crawlEntireDomain?: boolean;
     allowExternalLinks?: boolean;
     ignoreSitemap?: boolean;
     scrapeOptions?: CrawlScrapeOptions;
@@ -216,6 +219,7 @@ interface CrawlParams {
      * If not provided, the crawler may use the robots.txt crawl delay if available.
      */
     delay?: number;
+    allowSubdomains?: boolean;
     maxConcurrency?: number;
 }
 /**

package/dist/index.d.ts

@@ -120,6 +120,7 @@ interface CrawlScrapeOptions {
     proxy?: "basic" | "stealth" | "auto";
     storeInCache?: boolean;
     maxAge?: number;
+    parsePDF?: boolean;
 }
 type Action = {
     type: "wait";
@@ -132,6 +133,7 @@ type Action = {
 } | {
     type: "screenshot";
     fullPage?: boolean;
+    quality?: number;
 } | {
     type: "write";
     text: string;
@@ -199,6 +201,7 @@ interface CrawlParams {
     maxDiscoveryDepth?: number;
     limit?: number;
     allowBackwardLinks?: boolean;
+    crawlEntireDomain?: boolean;
     allowExternalLinks?: boolean;
     ignoreSitemap?: boolean;
     scrapeOptions?: CrawlScrapeOptions;
@@ -216,6 +219,7 @@ interface CrawlParams {
      * If not provided, the crawler may use the robots.txt crawl delay if available.
      */
     delay?: number;
+    allowSubdomains?: boolean;
     maxConcurrency?: number;
 }
 /**

package/dist/index.js

@@ -1,6 +1,6 @@
 // src/index.ts
 import axios, { AxiosError } from "axios";
-import * as zt from "zod";
+import "zod";
 import { zodToJsonSchema } from "zod-to-json-schema";
 
 // node_modules/typescript-event-target/dist/index.mjs
@@ -29,7 +29,7 @@ var FirecrawlApp = class {
   }
   async getVersion() {
     try {
-      const packageJson = await import("./package-AP4AUZKP.js");
+      const packageJson = await import("./package-SROKDQ7E.js");
       return packageJson.default.version;
     } catch (error) {
       console.error("Error getting version:", error);
@@ -604,10 +604,12 @@ var FirecrawlApp = class {
     try {
       if (!params?.schema) {
         jsonSchema = void 0;
-      } else if (typeof params.schema === "object" && params.schema !== null && Object.getPrototypeOf(params.schema)?.constructor?.name?.startsWith("Zod")) {
-        jsonSchema = zodToJsonSchema(params.schema);
       } else {
-        jsonSchema = params.schema;
+        try {
+          jsonSchema = zodToJsonSchema(params.schema);
+        } catch (_) {
+          jsonSchema = params.schema;
+        }
       }
     } catch (error) {
       throw new FirecrawlError("Invalid schema. Schema must be either a valid Zod schema or JSON schema object.", 400);
@@ -664,10 +666,14 @@ var FirecrawlApp = class {
     let jsonData = { urls, ...params };
     let jsonSchema;
     try {
-      if (params?.schema instanceof zt.ZodType) {
-        jsonSchema = zodToJsonSchema(params.schema);
+      if (!params?.schema) {
+        jsonSchema = void 0;
       } else {
-        jsonSchema = params?.schema;
+        try {
+          jsonSchema = zodToJsonSchema(params.schema);
+        } catch (_) {
+          jsonSchema = params.schema;
+        }
       }
     } catch (error) {
       throw new FirecrawlError("Invalid schema. Schema must be either a valid Zod schema or JSON schema object.", 400);

package/dist/{package-AP4AUZKP.js → package-SROKDQ7E.js}

@@ -1,6 +1,6 @@
 // package.json
 var name = "@mendable/firecrawl-js";
-var version = "1.25.6";
+var version = "1.29.0";
 var description = "JavaScript SDK for Firecrawl API";
 var main = "dist/index.js";
 var types = "dist/index.d.ts";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mendable/firecrawl",
-  "version": "1.25.6",
+  "version": "1.29.0",
   "description": "JavaScript SDK for Firecrawl API",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",

package/src/__tests__/v1/e2e_withAuth/index.test.ts

@@ -103,6 +103,32 @@ describe('FirecrawlApp E2E Tests', () => {
     expect(response?.markdown).toContain('We present spectrophotometric observations of the Broad Line Radio Galaxy');
   }, 30000); // 30 seconds timeout
 
+  test.concurrent('should return successful response for valid scrape with PDF file and parsePDF true', async () => {
+    const app = new FirecrawlApp({ apiKey: TEST_API_KEY, apiUrl: API_URL });
+    const response = await app.scrapeUrl('https://arxiv.org/pdf/astro-ph/9301001.pdf', {
+      parsePDF: true
+    });
+    if (!response.success) {
+      throw new Error(response.error);
+    }
+
+    expect(response).not.toBeNull();
+    expect(response?.markdown).toContain('We present spectrophotometric observations of the Broad Line Radio Galaxy');
+  }, 30000); // 30 seconds timeout
+
+  test.concurrent('should return successful response for valid scrape with PDF file and parsePDF false', async () => {
+    const app = new FirecrawlApp({ apiKey: TEST_API_KEY, apiUrl: API_URL });
+    const response = await app.scrapeUrl('https://arxiv.org/pdf/astro-ph/9301001.pdf', {
+      parsePDF: false
+    });
+    if (!response.success) {
+      throw new Error(response.error);
+    }
+
+    expect(response).not.toBeNull();
+    expect(response?.markdown).toMatch(/^[A-Za-z0-9+/]+=*$/);
+  }, 30000); // 30 seconds timeout
+
   test.concurrent('should throw error for invalid API key on crawl', async () => {
     if (API_URL.includes('api.firecrawl.dev')) {
       const invalidApp = new FirecrawlApp({ apiKey: "invalid_api_key", apiUrl: API_URL });

package/src/index.ts

@@ -125,6 +125,7 @@ export interface CrawlScrapeOptions {
   proxy?: "basic" | "stealth" | "auto";
   storeInCache?: boolean;
   maxAge?: number;
+  parsePDF?: boolean;
 }
 
 export type Action = {
@@ -138,6 +139,7 @@ export type Action = {
 } | {
   type: "screenshot",
   fullPage?: boolean,
+  quality?: number,
 } | {
   type: "write",
   text: string,
@@ -209,6 +211,7 @@ export interface CrawlParams {
   maxDiscoveryDepth?: number;
   limit?: number;
   allowBackwardLinks?: boolean;
+  crawlEntireDomain?: boolean;
   allowExternalLinks?: boolean;
   ignoreSitemap?: boolean;
   scrapeOptions?: CrawlScrapeOptions;
@@ -226,6 +229,7 @@ export interface CrawlParams {
    * If not provided, the crawler may use the robots.txt crawl delay if available.
    */
   delay?: number;
+  allowSubdomains?: boolean;
   maxConcurrency?: number;
 }
 
@@ -1243,10 +1247,12 @@ export default class FirecrawlApp {
     try {
       if (!params?.schema) {
         jsonSchema = undefined;
-      } else if (typeof params.schema === "object" && params.schema !== null && Object.getPrototypeOf(params.schema)?.constructor?.name?.startsWith("Zod")) {
-        jsonSchema = zodToJsonSchema(params.schema as zt.ZodType);
       } else {
-        jsonSchema = params.schema;
+        try {
+          jsonSchema = zodToJsonSchema(params.schema as zt.ZodType);
+        } catch (_) {
+          jsonSchema = params.schema;
+        }
       }
     } catch (error: any) {
       throw new FirecrawlError("Invalid schema. Schema must be either a valid Zod schema or JSON schema object.", 400);
@@ -1311,10 +1317,14 @@ export default class FirecrawlApp {
     let jsonSchema: any;
 
     try {
-      if (params?.schema instanceof zt.ZodType) {
-        jsonSchema = zodToJsonSchema(params.schema);
+      if (!params?.schema) {
+        jsonSchema = undefined;
       } else {
-        jsonSchema = params?.schema;
+        try {
+          jsonSchema = zodToJsonSchema(params.schema as zt.ZodType);
+        } catch (_) {
+          jsonSchema = params.schema;
+        }
       }
     } catch (error: any) {
       throw new FirecrawlError("Invalid schema. Schema must be either a valid Zod schema or JSON schema object.", 400);