static void MinifyHtml(string filePath)
{
IHtmlDocument htmlDoc = new HtmlParser().ParseDocument(File.ReadAllText(filePath));
string[] selectors =
{
"table#myTable", // Contains photographs only
"script", "noscript", "style",
"link[rel='stylesheet']",
"head > link[rel='apple-touch-icon']",
"head > link[rel='shortcut icon']",
"head > link[rel='manifest']",
"head > link[rel='publisher']",
"head > link[rel='alternate'][type='application/rss+xml']",
"head > meta[property^='og:']",
"head > meta[property^='fb:']",
"head > meta[property^='dc:']",
"head > meta[property='article:modified_time']",
"head > meta[name='robots']",
"head > meta[name='copyright']",
"head > meta[name='viewport']",
"head > meta[name='language']",
"head > meta[name='google-translate-customization']",
"head > meta[http-equiv='expires']",
"head > meta[http-equiv='Content-Security-Policy']",
"ins.adsbygoogle",
"span[id^='ezoic-pub-ad-placeholder-']",
"span.ezoic-ad",
"form#form",
"section#header-bg",
"div#topbar",
"div#breadcrumb", "section#breadcrumb",
"div#refbox",
"div#fb-root",
"div.widget",
"h3#comment",
"div.mobile-hide",
"div.comments",
"div.push.push-5", "div.push.push-10", "div.push.push-20","div.push.push-30","div.push.push-40",
"div#crt-1.ezflad-47", "div#crt-4.ezflad-47",
"a.back-to-top-button",
"div.fb-like", "div.fb-quote",
"div#dmo1",
"header", "footer"
};
foreach (var selector in selectors)
{
foreach (var el in htmlDoc.QuerySelectorAll(selector))
{
el.Remove();
}
}
foreach (var el in htmlDoc.QuerySelectorAll("*[style]"))
{
el.RemoveAttribute("style");
}
foreach (var el in htmlDoc.QuerySelectorAll("tr#infinite_scroll"))
{
el.RemoveAttribute("id");
}
htmlDoc.QuerySelector("body").ClearAttr();
foreach (IComment comment in htmlDoc.Descendents <IComment>())
{
comment.Remove();
}
string html = htmlDoc.DocumentElement.InnerHtml;
html = html.Replace(" ", " ");
html = Regex.Replace(html, @"^\s+$", "", RegexOptions.Multiline);
html = Regex.Replace(html, @"^\s+<", "<", RegexOptions.Multiline);
//html.Length.ToString("N0").Dump();
//File.WriteAllText(@"C:\Users\IEUser\Desktop\TRAINING\L11-S01-LINQ\Titanic\bin\Debug\TitanicData\titanic01-new.html", html);
File.WriteAllText(filePath, html);
}
