Fixing child links only bug

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

@@ -88,6 +88,10 @@ export class WebCrawler {
           return false;
         }
 
+        if (!this.initialUrl.includes(link)) {
+          return false;
+        }
+
         return true;
       })
       .slice(0, limit);
@@ -109,7 +113,7 @@ export class WebCrawler {
 
     const sitemapLinks = await this.tryFetchSitemapLinks(this.initialUrl);
     if (sitemapLinks.length > 0) {
-      const filteredLinks = this.filterLinks(sitemapLinks, limit, maxDepth);
+      let filteredLinks = this.filterLinks(sitemapLinks, limit, maxDepth);
       return filteredLinks.map(link => ({ url: link, html: "" }));
     }
 

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

@@ -145,12 +145,18 @@ export class WebScraperDataProvider {
 
     let links = await crawler.start(inProgress, 5, this.limit, this.maxCrawledDepth);
 
-    const allLinks = links.map((e) => e.url);
+    let allLinks = links.map((e) => e.url);
     const allHtmls = links.map((e)=> e.html);
 
     if (this.returnOnlyUrls) {
       return this.returnOnlyUrlsResponse(allLinks , inProgress);
     }
+
+    allLinks = allLinks.filter(link => {
+      const normalizedInitialUrl = this.urls[0].endsWith('/') ? this.urls[0] : `${this.urls[0]}/`;
+      const normalizedLink = link.endsWith('/') ? link : `${link}/`;
+      return !normalizedLink.startsWith(normalizedInitialUrl) || normalizedLink !== normalizedInitialUrl;
+    });
     
     let documents = [];
     // check if fast mode is enabled and there is html inside the links
@@ -175,6 +181,12 @@ export class WebScraperDataProvider {
     inProgress?: (progress: Progress) => void
   ): Promise<Document[]> {
     let links = await getLinksFromSitemap(this.urls[0]);
+    links = links.filter(link => {
+      const normalizedInitialUrl = this.urls[0].endsWith('/') ? this.urls[0] : `${this.urls[0]}/`;
+      const normalizedLink = link.endsWith('/') ? link : `${link}/`;
+      return !normalizedLink.startsWith(normalizedInitialUrl) || normalizedLink !== normalizedInitialUrl;
+    });
+
     if (this.returnOnlyUrls) {
       return this.returnOnlyUrlsResponse(links, inProgress);
     }

apps/test-suite/data/crawl.json

@@ -2,7 +2,7 @@
   {
     "website": "https://mendable.ai/pricing",
     "expected_min_num_of_pages": 29,
-    "expected_crawled_pages": [
+    "expected_not_crawled_pages": [
       "https://mendable.ai/",
       "https://mendable.ai/blog",
       "https://mendable.ai/signin",
@@ -34,7 +34,9 @@
       "https://www.agentops.ai/blog/streamlining-hr-with-saas",
       "https://www.agentops.ai/blog/simplify-hr-with-modern-saas-solutions",
       "https://www.agentops.ai/blog/efficient-hr-operations-with-saas",
-      "https://www.agentops.ai/blog/hr-made-simple-with-saas",
+      "https://www.agentops.ai/blog/hr-made-simple-with-saas"
+    ],
+    "expected_not_crawled_pages": [
       "https://www.agentops.ai/about-us",
       "https://www.agentops.ai/contact-us"
     ]
@@ -69,7 +71,7 @@
   {
     "website": "https://en.wikipedia.org/wiki/T._N._Seshan",
     "expected_min_num_of_pages": 100,
-    "expected_crawled_pages": [
+    "expected_not_crawled_pages": [
       "https://en.wikipedia.org/wiki/Wikipedia:Contents",
       "https://en.wikipedia.org/wiki/Wikipedia:Contact_us",
       "https://en.wikipedia.org/wiki/V._S._Ramadevi",
@@ -79,15 +81,10 @@
       "https://en.wikipedia.org/wiki/File:T.N._Seshan_in_1994.jpg"
     ]
   },
-  {
-    "website": "https://mendable.ai/blog",
-    "expected_min_num_of_pages": 0,
-    "expected_crawled_pages": [""]
-  },
   {
     "website": "https://www.framer.com/pricing",
     "expected_min_num_of_pages": 58,
-    "expected_crawled_pages": [
+    "expected_not_crawled_pages": [
       "https://www.framer.com/features/navigation/",
       "https://www.framer.com/contact/",
       "https://www.framer.com/add-ons/",
@@ -101,7 +98,7 @@
   {
     "website": "https://fly.io/docs/gpus/gpu-quickstart",
     "expected_min_num_of_pages": 39,
-    "expected_crawled_pages": [
+    "expected_not_crawled_pages": [
       "https://fly.io/docs/getting-started/",
       "https://fly.io/docs/hands-on/",
       "https://fly.io/docs/about/support/",
@@ -118,8 +115,8 @@
     "expected_crawled_pages": [""]
   },
   {
-    "website": "https://www.instructables.com",
+    "website": "https://www.instructables.com/circuits",
-    "expected_min_num_of_pages": 78,
+    "expected_min_num_of_pages": 12,
     "expected_crawled_pages": [
       "https://www.instructables.com/circuits/",
       "https://www.instructables.com/circuits/apple/projects/",

apps/test-suite/tests/crawl.test.ts

@@ -62,6 +62,7 @@ describe("Crawling Checkup (E2E)", () => {
             // fail the test
             console.log('No response');
             continue;
+            // continue;
           }
 
           if (!completedResponse.body || completedResponse.body.status !== "completed") {
@@ -72,7 +73,7 @@ describe("Crawling Checkup (E2E)", () => {
               actual_output: 'FAILURE',
               error: `Crawl job did not complete successfully.`
             });
-            return null;
+            continue;
           }
 
           // check how many webpages were crawled successfully
@@ -85,11 +86,11 @@ describe("Crawling Checkup (E2E)", () => {
               actual_output: `FAILURE: ${completedResponse.body.data.length}`,
               error: `Expected at least ${websiteData.expected_min_num_of_pages} webpages, but got ${completedResponse.body.data.length}`
             });
-            return null;
+            continue;
           }
 
           // checks if crawled pages contain expected_crawled_pages
-          if (websiteData.expected_crawled_pages.some(page => !completedResponse.body.data.some((d: { url: string }) => d.url === page))) {
+          if (websiteData.expected_crawled_pages && websiteData.expected_crawled_pages.length > 0 && websiteData.expected_crawled_pages.some(page => !completedResponse.body.data?.some((d: { url: string }) => d.url === page))) {
             errorLog.push({
               website: websiteData.website,
               prompt: 'CRAWL',
@@ -97,7 +98,19 @@ describe("Crawling Checkup (E2E)", () => {
               actual_output: `FAILURE: ${completedResponse.body.data}`,
               error: `Expected crawled pages to contain ${websiteData.expected_crawled_pages}, but got ${completedResponse.body.data}`
             });
-            return null;
+            continue;
+          }
+
+          // checks if crawled pages not contain expected_not_crawled_pages
+          if (websiteData.expected_not_crawled_pages && websiteData.expected_not_crawled_pages.length > 0 && websiteData.expected_not_crawled_pages.filter(page => completedResponse.body.data.some((d: { url: string }) => d.url === page)).length > 0) {
+            errorLog.push({
+              website: websiteData.website,
+              prompt: 'CRAWL',
+              expected_output: `SUCCESS: ${websiteData.expected_not_crawled_pages}`,
+              actual_output: `FAILURE: ${completedResponse.body.data}`,
+              error: `Expected crawled pages to not contain ${websiteData.expected_not_crawled_pages}, but got ${completedResponse.body.data}`
+            });
+            continue;
           }
 
           passedTests++;
@@ -110,6 +123,7 @@ describe("Crawling Checkup (E2E)", () => {
             actual_output: 'FAILURE',
             error: `Error processing ${websiteData.website}: ${error}`
           });
+          continue;
         }
       }