private static void Main(string[] args)
{
Console.WriteLine("Zeta Html Compressor Console. http://zeta.li");
Console.WriteLine();
Console.WriteLine("Syntax: hc.exe <HTML input file path>");
Console.WriteLine();
if (args.Length != 1)
{
Console.Error.WriteLine("No input file specified or invalid number of arguments ({0} found).", args.Length);
}
else
{
var inputFilePath = args[0];
var input = File.ReadAllText(inputFilePath);
var compressor = new HtmlContentCompressor();
var output = compressor.Compress(input);
var outputFilePath =
Path.Combine(
Path.GetDirectoryName(inputFilePath),
Path.GetFileNameWithoutExtension(inputFilePath) + // @"-output" +
Path.GetExtension(inputFilePath));
if (File.Exists(outputFilePath))
{
File.Delete(outputFilePath);
}
File.WriteAllText(outputFilePath, output, Encoding.UTF8);
Console.WriteLine("Successfully written output file '{0}'.", outputFilePath);
}
}
private static void Main(string[] args)
{
Console.WriteLine("Zeta Html Compressor Console. http://zeta.li");
Console.WriteLine();
Console.WriteLine("Syntax: hc.exe <HTML input file path>");
Console.WriteLine();
if (args.Length != 1)
{
Console.Error.WriteLine("No input file specified or invalid number of arguments ({0} found).", args.Length);
}
else
{
var inputFilePath = args[0];
var input = File.ReadAllText(inputFilePath);
var compressor = new HtmlContentCompressor();
var output = compressor.Compress(input);
var outputFilePath =
Path.Combine(
Path.GetDirectoryName(inputFilePath),
Path.GetFileNameWithoutExtension(inputFilePath) + // @"-output" +
Path.GetExtension(inputFilePath));
if (File.Exists(outputFilePath)) File.Delete(outputFilePath);
File.WriteAllText(outputFilePath, output, Encoding.UTF8);
Console.WriteLine("Successfully written output file '{0}'.", outputFilePath);
}
}
public void BenchHtmlCompressor()
{
var compressor = new HtmlContentCompressor();
var result = compressor.Compress(html);
var size = result.Length;
//File.WriteAllText(
// Path.Combine(Path.GetDirectoryName(typeof(Program).Assembly.Location), "HTMLStandard.min3.htm"),
// result);
}
public void Test02()
{
var comp = new HtmlContentCompressor();
comp.AddPreservePattern(new Regex(@"<!-- PROTECTED: .*? -->", RegexOptions.Singleline | RegexOptions.IgnoreCase));
var input = Resources.Html02a;
var output = comp.Compress(input);
Assert.IsNotEmpty(output);
var ratio = Math.Round((double)output.Length / input.Length * 100.0, 2);
Assert.IsTrue(ratio < 100.0);
}
public void MinifyHtml(string filepath)
{
HtmlContentCompressor contentCompressor = new HtmlContentCompressor();
//read
var html = File.ReadAllText(filepath);
//compress
var compressed = contentCompressor.Compress(html);
//write
File.WriteAllText(filepath, compressed);
}
public void Test02()
{
var comp = new HtmlContentCompressor();
comp.AddPreservePattern(new Regex(@"<!-- PROTECTED: .*? -->", RegexOptions.Singleline | RegexOptions.IgnoreCase));
var input = Resources.Html02a;
var output = comp.Compress(input);
Assert.IsNotEmpty(output);
var ratio = Math.Round((double)output.Length / input.Length * 100.0, 2);
Assert.IsTrue(ratio < 100.0);
}
public static string MinifyHTML(string html)
{
ZetaHtmlCompressor.HtmlContentCompressor compressor = new HtmlContentCompressor();
return(compressor.Compress(html));
}
static System.IO.FileInfo ProcessPage(string url, List <string> ankiCsvEntries, HashSet <string> visitedURLs)
{
var srcHtmlFile = new System.IO.FileInfo(UrlToTempDownloadPath(url) + ".html");
var dstHtmlFile = new System.IO.FileInfo(UrlToLocalPath(url) + ".html");
dstHtmlFile.Directory.Create();
if (visitedURLs.Contains(url))
{
return(dstHtmlFile);
}
visitedURLs.Add(url);
var doc = new HtmlDocument();
doc.Load(srcHtmlFile.FullName);
doc.DocumentNode.SelectSingleNode("html/head").Remove();
var appSiblings = doc.DocumentNode.SelectSingleNode("html/body").ChildNodes.Where(node => !(node.Name == "div" && node.Id == "app")).ToArray();
foreach (var appSibling in appSiblings)
{
appSibling.Remove();
}
var mainSiblings = doc.DocumentNode.SelectSingleNode("html/body/div[@id='app']/div").ChildNodes.Where(node => node.Name != "main").ToArray();
foreach (var mainSibling in mainSiblings)
{
mainSibling.Remove();
}
var images = doc.DocumentNode.SelectNodes("//img");
if (images != null)
{
foreach (var image in images)
{
string imgSrc = image.GetAttributeValue("src", null);
string imgSrcSet = image.GetAttributeValue("srcset", null);
if (imgSrcSet.Substring(0, imgSrc.Length) != imgSrc)
{
throw new Exception("Expected srcset to be the same link as src");
}
FileInfo srcImgFile = new FileInfo(UrlToLocalPath(imgSrc));
string ankiFriendlyName = "metal_" + new Uri(imgSrc).AbsolutePath.Substring(1).Replace('/', '_');
FileInfo dstImgFile = new FileInfo("anki_media/" + ankiFriendlyName);
dstImgFile.Directory.Create();
File.Copy(srcImgFile.FullName, dstImgFile.FullName, true);
image.SetAttributeValue("src", ankiFriendlyName);
image.SetAttributeValue("srcset", imgSrcSet.Replace(imgSrc, ankiFriendlyName));
}
}
// reserve an index so it properly shows up in pre-order traversal
int ankiCsvEntryIndex = -1;
if (ankiCsvEntries != null)
{
ankiCsvEntryIndex = ankiCsvEntries.Count();
ankiCsvEntries.Add("");
}
var sections = doc.DocumentNode.SelectNodes("html/body/div[@id='app']/div/main/section[@id='topics']/div/div/section");
foreach (var section in sections)
{
var topics = section.SelectNodes("div/div[@class='contenttable-section-content column large-9 medium-9 small-12']/div[@class='task-topics']/div");
foreach (var topic in topics)
{
var anchor = topic.SelectSingleNode("div/a");
var href = anchor.GetAttributeValue("href", null);
var entriesToPass = ankiCsvEntries;
var codeNode = anchor.SelectSingleNode("code");
if (codeNode != null)
{
if (new List <string> {
"func ", "var ", "case ", "static var ", "static let ", "static func ", "class func ", "init", "subscript"
}.Contains(codeNode.FirstChild.InnerText))
{
entriesToPass = null;
}
}
System.IO.FileInfo processedPage = ProcessPage("https://developer.apple.com" + href, entriesToPass, visitedURLs);
anchor.SetAttributeValue("href", "https://developer.apple.com" + href);
}
}
doc.DocumentNode.SelectSingleNode("html/body/div[@id='app']/div/main/div[@class='topic-title']/span[@class='eyebrow']")?.Remove();
doc.DocumentNode.SelectSingleNode("html/body/div[@id='app']/div/main/div[@class='topic-container section-content row']/div[@class='topic-summary column large-3 medium-3 small-12']")?.Remove();
doc.DocumentNode.SelectSingleNode("html/body/div[@id='app']/div/main/section[@id='see-also']")?.Remove();
doc.DocumentNode.SelectSingleNode("html/body/div[@id='app']/div/main/div[@class='betainfo section-alt row']")?.Remove();
var anchors = doc.DocumentNode.SelectNodes("//a");
if (anchors != null)
{
foreach (var anchor in anchors)
{
string href = anchor.GetAttributeValue("href", null);
if (href.StartsWith("/documentation/"))
{
anchor.SetAttributeValue("href", "https://developer.apple.com" + href);
}
}
}
doc.Save(dstHtmlFile.FullName);
if (ankiCsvEntryIndex != -1)
{
string docAsString = File.ReadAllText(dstHtmlFile.FullName);
if (docAsString.Contains('\t'))
{
throw new Exception("tab is not a good enough CSV separator!");
}
docAsString = new HtmlContentCompressor().Compress(docAsString);
docAsString = new System.Text.RegularExpressions.Regex("\r?\n").Replace(docAsString, "<br>");
var topicHeading = doc.DocumentNode.SelectSingleNode("html/body/div[@id='app']/div/main/div[@class='topic-title']/h1[@class='topic-heading']").InnerText;
var titleOnlyDoc = new HtmlDocument();
titleOnlyDoc.LoadHtml(docAsString);
titleOnlyDoc.DocumentNode.SelectSingleNode("html/body/div[@id='app']/div/main/div[@class='topic-container section-content row']/div[@id='topic-content']")?.Remove();
titleOnlyDoc.DocumentNode.SelectSingleNode("html/body/div[@id='app']/div/main/section[@id='topics']")?.Remove();
StringBuilder sb = new StringBuilder();
StringWriter sw = new StringWriter(sb);
titleOnlyDoc.Save(sw);
string titleOnlyString = sb.ToString();
string ankiCsvEntry =
topicHeading + "\t" +
url + "\t" +
titleOnlyString + "\t" +
docAsString;
ankiCsvEntries[ankiCsvEntryIndex] = ankiCsvEntry;
}
return(dstHtmlFile);
}
