C#:.NET:HTML

.NET Array Dictionary List String 2D Async DataTable Dates DateTime Enum File For Foreach Format IEnumerable If IndexOf Lambda LINQ Parse Path Process Property Regex Replace Row Sort Split Static StringBuilder Substring Switch Tuple

Scraping HTML extracts important page elements. It has many legal uses for webmasters and ASP.NET developers. With the Regex type and WebClient, we implement screen scraping for HTML.

WebClient

Example. First, for my demonstration I will scrape HTML links from Wikipedia.org. This is permitted by Wikipedia's GPL license, and this demonstration is fair use. Here we see code that downloads the English Wikipedia page.

Note: It opens Wikipedia and downloads the content at the specified URL. Part 2 uses my special code to loop over each link and its text.

C# program that scrapes HTML

using System.Diagnostics;
using System.Net;

class Program
{
    static void Main()
    {
	// Scrape links from wikipedia.org

	// 1.
	// URL: http://en.wikipedia.org/wiki/Main_Page
	WebClient w = new WebClient();
	string s = w.DownloadString("http://en.wikipedia.org/wiki/Main_Page");

	// 2.
	foreach (LinkItem i in LinkFinder.Find(s))
	{
	    Debug.WriteLine(i);
	}
    }
}

Example 2. Here I show a simple class that receives the HTML string and then extracts all the links and their text into structs. It is fairly fast, but I offer some optimization tips further down. It would be better to use a class.

Class
C# program that scrapes with Regex

using System.Collections.Generic;
using System.Text.RegularExpressions;

public struct LinkItem
{
    public string Href;
    public string Text;

    public override string ToString()
    {
	return Href + "\n\t" + Text;
    }
}

static class LinkFinder
{
    public static List<LinkItem> Find(string file)
    {
	List<LinkItem> list = new List<LinkItem>();

	// 1.
	// Find all matches in file.
	MatchCollection m1 = Regex.Matches(file, @"(<a.*?>.*?</a>)",
	    RegexOptions.Singleline);

	// 2.
	// Loop over each match.
	foreach (Match m in m1)
	{
	    string value = m.Groups[1].Value;
	    LinkItem i = new LinkItem();

	    // 3.
	    // Get href attribute.
	    Match m2 = Regex.Match(value, @"href=\""(.*?)\""",
		RegexOptions.Singleline);
	    if (m2.Success)
	    {
		i.Href = m2.Groups[1].Value;
	    }

	    // 4.
	    // Remove inner tags from text.
	    string t = Regex.Replace(value, @"\s*<.*?>\s*", "",
		RegexOptions.Singleline);
	    i.Text = t;

	    list.Add(i);
	}
	return list;
    }
}

This example first finds all hyperlink tags. We store all the complete A tags into a MatchCollection. These are objects that store the complete HTML strings. They used used later in the method.

In step 2 it loops over all hyperlink tag strings. In the algorithm, the next part examines all the text of the A tags. This is necessary for reading the parts of the A tags. For each A tag, it reads in the HREF attribute.

HREF: This attribute points to other web resources. This part is not failsafe, but almost always works.

Finally, the method returns the List of LinkItem objects it has built up. This list can then be used in the foreach-loop from the first C# example. The ToString method override above simply provides a standard way of printing the links.


Tests. My first two attempts at this code were incorrect and had unacceptable bugs, but the version shown here works. You need to use RegexOptions.SingleLine. The dot in a Regex matches all characters except a newline unless this is specified.

Tip: To match multiline links, we require RegexOptions.Singleline. This is an important option.

RegexOptions: MSDN

Test the program on your website. It prints out matches to the console. Here we see part of the current results for the Wikipedia home page. The original HTML shows where the links were extracted. They are contained in a LI tag.

Note: You will see my program successfully extracted the anchor text and also the HREF value.

Output

#column-one
    navigation
#searchInput
    search
/wiki/Wikipedia
    Wikipedia
/wiki/Free_content
    free
/wiki/Encyclopedia
    encyclopedia
/wiki/Wikipedia:Introduction
    anyone can edit
/wiki/Special:Statistics
    2,617,101
/wiki/English_language
    English
/wiki/Portal:Arts
    Arts
/wiki/Portal:Biography
    Biography
/wiki/Portal:Geography
    Geography
/wiki/Portal:History
    History
/wiki/Portal:Mathematics
    Mathematics
/wiki/Portal:Science
    Science
/wiki/Portal:Society
    Society
/wiki/Portal:Technology_and_applied_sciences
    Technology

Original website HTML

<ul>
<li><a href="/wiki/Portal:Arts" title="Portal:Arts">Arts</a></li>
<li><a href="/wiki/Portal:Biography" title="Portal:Biography">Biography</a></li>
<li><a href="/wiki/Portal:Geography" title="Portal:Geography">Geography</a></li>

</ul>

SingleLine is an important option. MSDN states that SingleLine "Specifies single-line mode. Changes the meaning of the dot so it matches every character (instead of every character except \n)."


Performance. You can improve performance of the regular expressions by specifying RegexOptions.Compiled, and also use instance Regex objects, not the static methods I show. Normally, your Internet connection will be the bottleneck.

Regex.Match

Summary. We scraped HTML content from the Internet. The code is more flexible than some other approaches. Using three regular expressions, you can extract HTML links into objects with a fair degree of accuracy.

Note: I have tested this code on several sites where it is legal. It is a valuable tool for webmasters.