Merge pull request #858 from mendableai/nsc/new-actions

Support for the 2 new actions

apps/api/src/controllers/v1/types.ts

@@ -61,8 +61,14 @@ export type ExtractOptions = z.infer<typeof extractOptions>;
 export const actionsSchema = z.array(z.union([
   z.object({
     type: z.literal("wait"),
-    milliseconds: z.number().int().positive().finite(),
-  }),
+    milliseconds: z.number().int().positive().finite().optional(),
+    selector: z.string().optional(),
+  }).refine(
+    (data) => (data.milliseconds !== undefined || data.selector !== undefined) && !(data.milliseconds !== undefined && data.selector !== undefined),
+    {
+      message: "Either 'milliseconds' or 'selector' must be provided, but not both.",
+    }
+  ),
   z.object({
     type: z.literal("click"),
     selector: z.string(),
@@ -83,6 +89,9 @@ export const actionsSchema = z.array(z.union([
     type: z.literal("scroll"),
     direction: z.enum(["up", "down"]),
   }),
+  z.object({
+    type: z.literal("scrape"),
+  }),
 ]));
 
 export const scrapeOptions = z.object({

apps/api/src/lib/entities.ts

@@ -12,7 +12,8 @@ export interface Progress {
 
 export type Action = {
   type: "wait",
-  milliseconds: number,
+  milliseconds?: number,
+  selector?: string,
 } | {
   type: "click",
   selector: string,
@@ -28,7 +29,9 @@ export type Action = {
 } | {
   type: "scroll",
   direction: "up" | "down"
-};
+} | {
+  type: "scrape",
+}
 
 export type PageOptions = {
   includeMarkdown?: boolean;
@@ -163,11 +166,17 @@ export class SearchResult {
   }
 }
 
+export interface ScrapeActionContent {
+  url: string;
+  html: string;
+}
+
 export interface FireEngineResponse {
   html: string;
   screenshots?: string[];
   pageStatusCode?: number;
   pageError?: string;
+  scrapeActionContent?: ScrapeActionContent[];
 }
 
 

apps/api/src/scraper/WebScraper/scrapers/fireEngine.ts

@@ -200,11 +200,13 @@ export async function scrapWithFireEngine({
       logParams.html = data.content ?? "";
       logParams.response_code = data.pageStatusCode;
       logParams.error_message = data.pageError ?? data.error;
+
       return {
         html: data.content ?? "",
         screenshots: data.screenshots ?? [data.screenshot] ?? [],
         pageStatusCode: data.pageStatusCode,
         pageError: data.pageError ?? data.error,
+        scrapeActionContent: data?.actionContent ?? [],
       };
     }
   } catch (error) {

apps/api/src/scraper/WebScraper/single_url.ts

@@ -21,6 +21,7 @@ import { extractLinks } from "./utils/utils";
 import { Logger } from "../../lib/logger";
 import { ScrapeEvents } from "../../lib/scrape-events";
 import { clientSideError } from "../../strings";
+import { ScrapeActionContent } from "../../lib/entities";
 
 dotenv.config();
 
@@ -180,7 +181,8 @@ export async function scrapSingleUrl(
       text: string;
       screenshot: string;
       actions?: {
-        screenshots: string[];
+        screenshots?: string[];
+        scrapes?: ScrapeActionContent[];
       };
       metadata: { pageStatusCode?: number; pageError?: string | null };
     } = { text: "", screenshot: "", metadata: {} };
@@ -259,6 +261,7 @@ export async function scrapSingleUrl(
           if (pageOptions.actions) {
             scraperResponse.actions = {
               screenshots: response.screenshots ?? [],
+              scrapes: response.scrapeActionContent ?? [],
             };
           }
           scraperResponse.metadata.pageStatusCode = response.pageStatusCode;

apps/js-sdk/firecrawl/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mendable/firecrawl-js",
-  "version": "1.7.2",
+  "version": "1.7.3",
   "description": "JavaScript SDK for Firecrawl API",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",

apps/js-sdk/firecrawl/src/index.ts

@@ -90,7 +90,8 @@ export interface CrawlScrapeOptions {
 
 export type Action = {
   type: "wait",
-  milliseconds: number,
+  milliseconds?: number,
+  selector?: string,
 } | {
   type: "click",
   selector: string,
@@ -106,6 +107,8 @@ export type Action = {
 } | {
   type: "scroll",
   direction: "up" | "down",
+} | {
+  type: "scrape",
 };
 
 export interface ScrapeParams<LLMSchema extends zt.ZodSchema = any, ActionsSchema extends (Action[] | undefined) = undefined> extends CrawlScrapeOptions {