public void TestSpecialCharacter()
{
const string html = "<>A";
var enc = new ASCIIEncoding();
var bis = new MemoryStream(enc.GetBytes(html));
var parse = new ReadHTML(bis);
Assert.IsTrue(parse.Read() == '<');
Assert.IsTrue(parse.Read() == '>');
Assert.IsTrue(parse.Read() == 'A');
}
public void TestComment()
{
const string html = "a<!-- Hello -->b";
var enc = new ASCIIEncoding();
var bis = new MemoryStream(enc.GetBytes(html));
var parse = new ReadHTML(bis);
Assert.IsTrue(parse.Read() == 'a');
Assert.IsTrue(parse.Read() == 0);
Assert.IsTrue(parse.Read() == 'b');
}
public void TestTagToString()
{
const string html = "<br/>";
var enc = new ASCIIEncoding();
var bis = new MemoryStream(enc.GetBytes(html));
var parse = new ReadHTML(bis);
Assert.IsTrue(parse.Read() == 0);
}
public void TestBothWithAttributes()
{
const string html = "<img src=\"picture.gif\" alt=\"A Picture\"/>";
var enc = new ASCIIEncoding();
var bis = new MemoryStream(enc.GetBytes(html));
var parse = new ReadHTML(bis);
Assert.IsTrue(parse.Read() == 0);
}
public void TestSimpleAttribute()
{
const string html = "<!DOCTYPE \"test\">";
var enc = new ASCIIEncoding();
var bis = new MemoryStream(enc.GetBytes(html));
var parse = new ReadHTML(bis);
Assert.IsTrue(parse.Read() == 0);
Tag tag = parse.LastTag;
Assert.AreEqual(tag.ToString(), html);
}
public void TestScript2()
{
const string html = "a<script>1<2</script>b<br>";
var enc = new ASCIIEncoding();
var bis = new MemoryStream(enc.GetBytes(html));
var parse = new ReadHTML(bis);
Assert.IsTrue(parse.Read() == 'a');
Assert.IsTrue(parse.Read() == 0);
Assert.IsTrue(parse.Read() == '1');
Assert.IsTrue(parse.Read() == '<');
Assert.IsTrue(parse.Read() == '2');
Assert.IsTrue(parse.Read() == 0);
Assert.IsTrue(parse.Read() == 'b');
Assert.IsTrue(parse.Read() == 0);
}
public void TestAttributesNoDelim()
{
const string html = "<img src=picture.gif alt=APicture>";
var enc = new ASCIIEncoding();
var bis = new MemoryStream(enc.GetBytes(html));
var parse = new ReadHTML(bis);
Assert.IsTrue(parse.Read() == 0);
Tag tag = parse.LastTag;
Assert.IsNotNull(tag);
Assert.IsTrue(tag.Name.Equals("img"));
Assert.IsTrue(tag.GetAttributeValue("src").Equals("picture.gif"));
Assert.IsTrue(tag.GetAttributeValue("alt").Equals("APicture"));
}
/// <summary>
/// Load the data units. Once the lower level data units have been
/// loaded, the contents can be loaded.
/// </summary>
/// <param name="istream">The input stream that the data units are loaded from.</param>
protected void LoadDataUnits(Stream istream)
{
var text = new StringBuilder();
int ch;
var parse = new ReadHTML(istream);
bool style = false;
bool script = false;
while ((ch = parse.Read()) != -1)
{
if (ch == 0)
{
if (style)
{
CreateCodeDataUnit(text.ToString());
}
else if (script)
{
CreateCodeDataUnit(text.ToString());
}
else
{
CreateTextDataUnit(text.ToString());
}
style = false;
script = false;
text.Length = 0;
CreateTagDataUnit(parse.LastTag);
if (String.Compare(parse.LastTag.Name, "style", true) == 0)
{
style = true;
}
else if (string.Compare(parse.LastTag.Name,
"script", true) == 0)
{
script = true;
}
}
else
{
text.Append((char)ch);
}
}
CreateTextDataUnit(text.ToString());
}
/// <summary>
/// Perform a Yahoo search.
/// </summary>
/// <param name="url">The REST URL.</param>
/// <returns>The search results.</returns>
private ICollection <Uri> DoSearch(Uri url)
{
ICollection <Uri> result = new List <Uri>();
// submit the search
WebRequest http = WebRequest.Create(url);
var response = (HttpWebResponse)http.GetResponse();
using (Stream istream = response.GetResponseStream())
{
var parse = new ReadHTML(istream);
var buffer = new StringBuilder();
bool capture = false;
// parse the results
int ch;
while ((ch = parse.Read()) != -1)
{
if (ch == 0)
{
Tag tag = parse.LastTag;
if (tag.Name.Equals("Url", StringComparison.CurrentCultureIgnoreCase))
{
buffer.Length = 0;
capture = true;
}
else if (tag.Name.Equals("/Url", StringComparison.CurrentCultureIgnoreCase))
{
result.Add(new Uri(buffer.ToString()));
buffer.Length = 0;
capture = false;
}
}
else
{
if (capture)
{
buffer.Append((char)ch);
}
}
}
}
response.Close();
return(result);
}
public void TestBoth()
{
const string html = "<br/>";
const string htmlName = "br";
var enc = new ASCIIEncoding();
var bis = new MemoryStream(enc.GetBytes(html));
var parse = new ReadHTML(bis);
Assert.IsTrue(parse.Read() == 0);
Tag tag = parse.LastTag;
Assert.IsNotNull(tag);
Assert.IsTrue(tag.TagType == Tag.Type.Begin);
Assert.IsTrue(tag.Name.Equals(htmlName));
parse.ReadToTag();
tag = parse.LastTag;
Assert.IsNotNull(tag);
Assert.IsTrue(tag.TagType == Tag.Type.End);
Assert.IsTrue(tag.Name.Equals(htmlName));
}
public void TestAttributeLess()
{
const string html = "12<b>12</b>1";
var enc = new ASCIIEncoding();
var bis = new MemoryStream(enc.GetBytes(html));
var parse = new ReadHTML(bis);
Assert.IsTrue(parse.Read() == '1');
Assert.IsTrue(parse.Read() == '2');
Assert.IsTrue(parse.Read() == 0);
Assert.IsTrue(parse.LastTag.Name.Equals("b"));
Assert.IsTrue(parse.LastTag.TagType == Tag.Type.Begin);
Assert.IsTrue(parse.Read() == '1');
Assert.IsTrue(parse.Read() == '2');
Assert.IsTrue(parse.Read() == 0);
Tag tag = parse.LastTag;
Assert.IsTrue(tag.Name.Equals("b"));
Assert.IsTrue(tag.TagType == Tag.Type.End);
Assert.AreEqual(tag.ToString(), "</b>");
Assert.IsTrue(parse.Read() == '1');
}