From 35cb998ae5fb19653c562eaa2185925b120161fc Mon Sep 17 00:00:00 2001
From: Muhammad Azeem <37941410+AzeemSup@users.noreply.github.com>
Date: Thu, 31 Oct 2024 15:48:27 +0500
Subject: [PATCH] Improve URL handling for external links in index.js

Improved URL handling with error checks for malformed links in `getUrlInfo`, refined domain extraction in `extractMainDomainFromUrl`, and ensured external link attributes apply correctly. Enhanced logging for easier debugging.

Signed-off-by: Muhammad Azeem <37941410+AzeemSup@users.noreply.github.com>
---
 lib/index.js | 116 ++++++++++++++++++++++++++-------------------------
 1 file changed, 60 insertions(+), 56 deletions(-)

diff --git a/lib/index.js b/lib/index.js
index 4faee7a60..814e6faaa 100755
--- a/lib/index.js
+++ b/lib/index.js
@@ -141,74 +141,78 @@ exports.parseFromDirectory = contentPath => {
 
 	const mainUrl = 'remoteintech.company'
 
-	function addTargetBlankAndExternalLinkIcons (el) {
-		if (el.type === 'tag') {
-			const anchorTagElements = el.children.filter(element => element.name === 'a')
-			if (anchorTagElements.length > 0) {
-				anchorTagElements.forEach(element => {
-					const url = element.attribs.href
-					const urlInfo = getUrlInfo(url)
-
-					if (urlInfo.is_email || urlInfo.is_internal) {
-						return
-					}
-
-					element.attribs.target = '_blank'
-					
-					$element = $( element )
-					$element.append('<span style="vertical-align: text-top;"> <img src="/assets/external-link.svg" /> </span>')
-				})
-			}
-
-			if (el.children && el.children.length) {
-				el.children.forEach(element => {
-					addTargetBlankAndExternalLinkIcons(element)
-				})
-			}
-		}
+	function addTargetBlankAndExternalLinkIcons(el) {
+	    if (el.type === 'tag') {
+	        const anchorTagElements = el.children.filter(element => element.name === 'a');
+	        
+	        anchorTagElements.forEach(element => {
+	            const url = element.attribs.href;
+	            const urlInfo = getUrlInfo(url);
+	
+	            if (url && !urlInfo.is_email && !urlInfo.is_internal) {
+	                element.attribs.target = '_blank';
+	                element.attribs.rel = 'noopener noreferrer';
+	                
+	                const $element = $(element);
+	                $element.append('<span style="vertical-align: text-top;"> <img src="/assets/external-link.svg" /> </span>');
+	            }
+	        });
+	
+	        // Recursively process child elements
+	        el.children.forEach(child => addTargetBlankAndExternalLinkIcons(child));
+	    }
 	}
 	
 	/**
 	 * Getting info about the url. It includes checking isEmail of isInternal
 	 * @param {*} url 
 	 */
-	function getUrlInfo (url) {
-		const data = {}
-
-		if (url.match(/^mailto:/)) { // checking url email or not
-			data.is_email = true
-			return data
-		}
-
-		const mainDomainFromGivenUrl = extractMainDomainFromUrl(url)
-
-		// checking url is email or not	
-		if (mainDomainFromGivenUrl !== mainUrl) { 
-			data.is_internal = false
-			return data
-		} else {
-			data.is_internal = true
-		}
-
-		return data
+	function getUrlInfo(url) {
+	    const data = {};
+	
+	    if (!url || typeof url !== 'string') {
+	        data.is_email = false;
+	        data.is_internal = false;
+	        return data;
+	    }
+	
+	    if (url.startsWith('mailto:')) {
+	        data.is_email = true;
+	        return data;
+	    }
+	
+	    try {
+	        const mainDomainFromGivenUrl = extractMainDomainFromUrl(url);
+	        data.is_internal = mainDomainFromGivenUrl === mainUrl;
+	    } catch (error) {
+	        console.error("Error processing URL:", url, error);
+	        data.is_internal = false;
+	    }
+	
+	    return data;
 	}
 
 	/**
 	 * Extracting main domain from the url
 	 * @param {*} url 
 	 */
-	function extractMainDomainFromUrl (url) {
-		const domainRe = /(https?:\/\/){0,1}((?:[\w\d-]+\.)+[\w\d]{2,})/i; // taken example from https://stackoverflow.com/questions/6238351/fastest-way-to-detect-external-urls
-
-		const data = domainRe.exec(url)
-
-		const splittedDomain = data[2].split('.')
-
-		if (splittedDomain.length === 2) { // check extra subdomain is present or not
-			return data[2]
-		}
-
-		return splittedDomain[splittedDomain.length - 2] + '.' + splittedDomain[splittedDomain.length - 1] // return only main domain address
+	function extractMainDomainFromUrl(url) {
+	    try {
+	        const domainRe = /(https?:\/\/)?(([\w\d-]+\.)+[\w\d]{2,})/i; // taken example from https://stackoverflow.com/questions/6238351/fastest-way-to-detect-external-urls
+	        const data = domainRe.exec(url);
+	
+	        if (!data || !data[2]) {
+	            console.warn("Invalid URL format:", url);
+	            return ''; // Return empty if domain extraction fails
+	        }
+	
+	        const domainParts = data[2].split('.');
+	        return domainParts.length === 2 ? data[2] : 
+	               domainParts.slice(-2).join('.');
+	    } catch (error) {
+	        console.error("Error extracting main domain:", url, error);
+	        return ''; // Safe fallback for unexpected input
+	    }
 	}
 
 	let lastCompanyName = null;