From 92bb30d8b7edf7234d9c8a913ad77ad26de4bcc5 Mon Sep 17 00:00:00 2001
From: karishmas6 <carishmashukla28@gmail.com>
Date: Thu, 1 Aug 2024 19:16:13 +0530
Subject: [PATCH] fix(core): script injection for scrape | scrapeSchema

---
 maxun-core/src/interpret.ts | 48 ++++++++++++++++++++++++-------------
 1 file changed, 31 insertions(+), 17 deletions(-)

diff --git a/maxun-core/src/interpret.ts b/maxun-core/src/interpret.ts
index 9ac32b0e..c7cedc02 100644
--- a/maxun-core/src/interpret.ts
+++ b/maxun-core/src/interpret.ts
@@ -277,25 +277,39 @@ export default class Interpreter extends EventEmitter {
         await page.close();
       },
       scrape: async (selector?: string) => {
-        const scrapeResults: Record<string, string>[] = <any>await page
-          // eslint-disable-next-line
-          // @ts-ignore
-          .evaluate((s) => scrape(s ?? null), selector);
+        // Check if 'scrape' function is available in the page context
+        const isScrapeAvailable = await page.evaluate(() => typeof window.scrape === 'function');
+    
+        if (!isScrapeAvailable) {
+            // Inject the script that defines the 'scrape' function
+            await page.addScriptTag({ path: path.join(__dirname, 'browserSide', 'scraper.js') });
+        }
+    
+        const scrapeResults: Record<string, string>[] = await page.evaluate((s) => window.scrape(s ?? null), selector);
         await this.options.serializableCallback(scrapeResults);
-      },
-      scrapeSchema: async (schema: Record<string, string>) => {
+    },
+    
+    scrapeSchema: async (schema: Record<string, string>) => {
+        // Check if 'scrapeSchema' function is available in the page context
+        const isScrapeSchemaAvailable = await page.evaluate(() => typeof window.scrapeSchema === 'function');
+    
+        if (!isScrapeSchemaAvailable) {
+            // Inject the script that defines the 'scrapeSchema' function
+            await page.addScriptTag({ path: path.join(__dirname, 'browserSide', 'scraper.js') });
+        }
+    
         const handleLists = await Promise.all(
-          Object.values(schema).map((selector) => page.$$(selector)),
+            Object.values(schema).map((selector) => page.$$(selector)),
         );
-
+    
         const namedHandleLists = Object.fromEntries(
-          Object.keys(schema).map((key, i) => [key, handleLists[i]]),
+            Object.keys(schema).map((key, i) => [key, handleLists[i]]),
         );
-
-        const scrapeResult = await page.evaluate((n) => scrapeSchema(n), namedHandleLists);
-
-        this.options.serializableCallback(scrapeResult);
-      },
+    
+        const scrapeResult = await page.evaluate((n) => window.scrapeSchema(n), namedHandleLists);
+        await this.options.serializableCallback(scrapeResult);
+    },
+    
       scroll: async (pages?: number) => {
         await page.evaluate(async (pagesInternal) => {
           for (let i = 1; i <= (pagesInternal ?? 1); i += 1) {
@@ -433,9 +447,9 @@ export default class Interpreter extends EventEmitter {
     this.initializedWorkflow = Preprocessor.initWorkflow(this.workflow, params);
 
     // @ts-ignore
-    if (await page.evaluate(() => !<any>window.scrape)) {
-      page.context().addInitScript({ path: path.join(__dirname, 'browserSide', 'scraper.js') });
-    }
+    // if (await page.evaluate(() => !<any>window.scrape)) {
+    //   page.context().addInitScript({ path: path.join(__dirname, 'browserSide', 'scraper.js') });
+    // }
 
     this.stopper = () => {
       this.stopper = null;