mirror
/
learndb
kopia lustrzana https://github.com/learn-awesome/learndb
1
0
Forkuj 0
Kod Zgłoszenia Packages Projekty Wydania Wiki Aktywność

modified simplifyContent()

pull/76/head
Huda Joad 2023-12-04 23:43:59 +03:00
rodzic 9d8827ab52
commit 5b3c6578aa
1 zmienionych plików z 24 dodań i 12 usunięć
Pokaż statystyki Ściągnij plik aktualizacji Ściągnij plik porównania Expand all files Collapse all files

36
netlify/functions/handleMetadata.js
Wyświetl plik

@ -17,19 +17,29 @@ async function fetchContentFromURL(url) {
} }
function simplifyContent(content) { function simplifyContent(content) {
// Keep title and body tags but remove their attributes // Preserve the title tag and its content
content = content.replace(/<title.*?>(.*?)<\/title>/gms, '<title>$1</title>'); let title = content.match(/<title.*?>(.*?)<\/title>/i);
content = content.replace(/<body.*?>(.*?)<\/body>/gms, '<body>$1</body>'); title = title ? title[1] : '';
// Extract the body content, if present
let bodyContent = '';
const bodyMatch = content.match(/<body.*?>([\s\S]*)<\/body>/i);
if (bodyMatch) {
bodyContent = bodyMatch[1];
} else {
// If no body tag, assume entire content is body
bodyContent = content;
}
// Remove script and style elements and their content // Remove script and style elements and their content
let simplifiedContent = content.replace(/<script.*?>.*?<\/script>/gms, ''); bodyContent = bodyContent.replace(/<script.*?>.*?<\/script>/gms, '');
simplifiedContent = simplifiedContent.replace(/<style.*?>.*?<\/style>/gms, ''); bodyContent = bodyContent.replace(/<style.*?>.*?<\/style>/gms, '');
// Remove all remaining HTML tags except for title and body, leaving the inner text // Remove all remaining HTML tags, except for title and body
simplifiedContent = simplifiedContent.replace(/<(?!title|body)[^>]+>/g, ''); bodyContent = bodyContent.replace(/<(?!\/?title|\/?body)([^>]+)>/g, '');
// Manually replace common HTML entities // Manually replace common HTML entities
simplifiedContent = simplifiedContent bodyContent = bodyContent
.replace(/&amp;/g, '&') .replace(/&amp;/g, '&')
.replace(/&lt;/g, '<') .replace(/&lt;/g, '<')
.replace(/&gt;/g, '>') .replace(/&gt;/g, '>')
@ -37,15 +47,17 @@ function simplifyContent(content) {
.replace(/&#39;/g, "'"); .replace(/&#39;/g, "'");
// Remove inline CSS and JavaScript event handlers // Remove inline CSS and JavaScript event handlers
simplifiedContent = simplifiedContent.replace(/style\s*=\s*'.*?'/gi, ''); bodyContent = bodyContent.replace(/style\s*=\s*'.*?'/gi, '');
simplifiedContent = simplifiedContent.replace(/on\w+\s*=\s*".*?"/gi, ''); bodyContent = bodyContent.replace(/on\w+\s*=\s*".*?"/gi, '');
// Normalize whitespace without removing sentence punctuation // Normalize whitespace without removing sentence punctuation
simplifiedContent = simplifiedContent.replace(/\s+/g, ' ').trim(); bodyContent = bodyContent.replace(/\s+/g, ' ').trim();
// Condense multiple line breaks into a single one // Condense multiple line breaks into a single one
simplifiedContent = simplifiedContent.replace(/(\r\n|\r|\n){2,}/g, '\n'); bodyContent = bodyContent.replace(/(\r\n|\r|\n){2,}/g, '\n');
// Reconstruct content with title and body
const simplifiedContent = `<title>${title}</title><body>${bodyContent}</body>`;
return simplifiedContent; return simplifiedContent;
} }