/// <summary>
/// Generates a preview from the given url
/// </summary>
/// <param name="strUrl">The url to generate a preview for</param>
/// <returns>A preview object, null if invalid url</returns>
public static URLPreview CreatePreview(string strUrl)
{
// Prepare a web request
WebRequest request = WebRequest.Create(strUrl);
// Get the web response
HttpWebResponse response = (HttpWebResponse)request.GetResponse();
// If the response type is not text/html we have no idea what to do with it
// TODO: Properly handle some other types? Display the type (video, audio, ..)?
if (response.ContentType.Contains("text/html"))
{
Stream responseStream = response.GetResponseStream();
URLPreview result = null;
/*
* Stream[] streams = new Stream[10000];
* for(int x = 0; x < 10000; x ++)
* {
* responseStream.CopyTo(streams[x]);
* }
*
* Stopwatch swWatch = Stopwatch.StartNew();
* for(int x = 0; x < 10000; x ++)
* {
* result = ParseResponseStream(streams[x]);
* }
*/
Stopwatch swWatch = Stopwatch.StartNew();
string str = "";
using (StreamReader sr = new StreamReader(responseStream))
str = sr.ReadToEnd();
swWatch.Stop();
Console.WriteLine("Read from stream: " + swWatch.ElapsedMilliseconds.ToString());
swWatch = Stopwatch.StartNew();
//for(int x = 0; x < 10000; x++)
{
result = ParseResponseString(str);
}
swWatch.Stop();
Console.WriteLine("Parse (average of 10000): " + (swWatch.ElapsedMilliseconds /*/ 10000*/).ToString());
return(result);
}
return(null);
}
private static void ContentParser(URLPreview destinationPreview, string tag, string strContent)
{
int tagLength = tag.Length;
// Title, Description and ImageUrl in meta og tags
if (tagLength == "meta".Length)
{
int indexOfOg = strContent.IndexOf("og:") + 3;
if (indexOfOg != -1)
{
int indexOfContentValueStart = strContent.IndexOf("content") + 9;
int contentValueLength = (strContent.IndexOf("\"", indexOfContentValueStart) - indexOfContentValueStart);
if (indexOfContentValueStart != -1)
{
int indexOfProperty = strContent.IndexOf("title", indexOfOg, 5);
if (indexOfProperty != -1)
{
destinationPreview.Title = strContent.Substring(indexOfContentValueStart, contentValueLength);
destinationPreview.PreferredTitle = true;
return;
}
indexOfProperty = strContent.IndexOf("description", indexOfOg, 11);
if (indexOfProperty != -1)
{
destinationPreview.Description = strContent.Substring(indexOfContentValueStart, contentValueLength);
destinationPreview.PreferredDescription = true;
return;
}
indexOfProperty = strContent.IndexOf("image", indexOfOg, 5);
if (indexOfProperty != -1)
{
destinationPreview.ImageUrl = strContent.Substring(indexOfContentValueStart, contentValueLength);
destinationPreview.PreferredImageUrl = true;
return;
}
}
}
}
// Title
else if (tagLength == "title".Length)
{
destinationPreview.Title = strContent;
}
// ImageUrl
else if (tagLength == "img".Length)
{
int indexOfSrc = strContent.IndexOf("src");
if (indexOfSrc != -1)
{
int indexOfContentValueStart = indexOfSrc + 5;
int contentValueLength = (strContent.IndexOf("\"", indexOfContentValueStart) - indexOfContentValueStart);
if (indexOfContentValueStart != -1)
{
destinationPreview.ImageUrl = strContent.Substring(indexOfContentValueStart, contentValueLength);
return;
}
}
}
}
/*
* /// <summary>
* /// Parses the given response stream, only handles whatever is needed for previews.
* /// The following code is not written for readability, but for speed. So nothing is split up into seperate functions, function calls are slow
* /// </summary>
* /// <param name="sResponse">The stream to read from</param>
* private static URLPreview ParseResponseStream(Stream oResponse)
* {
* bool conditionMetTitle = false;
* bool conditionMetImage = false;
* bool conditionMetDescription = false;
*
* string title = String.Empty;
* string imageUrl = String.Empty;
* string description = String.Empty;
*
* using(StreamReader sr = new StreamReader(oResponse))
* {
* //try
* //{
* bool ignoreTag = false;
* ParserState previousState = ParserState.None;
* ParserState currentState = ParserState.None;
* char[] previousChar = {'a', 'a'};
* char currentChar = 'a';
* string currentTag = "";
* string currentTagProperties = "";
*
* while(!sr.EndOfStream)
* {
* previousChar[1] = previousChar[0];
* previousChar[0] = currentChar;
* currentChar = (char)sr.Read();
*
* if(currentState == ParserState.Comment)
* {
* if(currentChar == '>' && previousChar[0] == '-' && previousChar[1] == '-')
* currentState = previousState;
* }
* else if(currentState == ParserState.String)
* {
* if(currentChar == '"')
* currentState = previousState;
* }
* else if(ignoreTag)
* {
* if(currentChar == '<')
* ignoreTag = false;
* }
*
* if(!ignoreTag)
* {
* switch(currentChar)
* {
* case '<':
* currentState = ParserState.TagOpen;
* currentTag = "";
* currentTagProperties = "";
* break;
* case '!':
* if(currentState == ParserState.TagOpen)
* previousState = currentState;
* currentState = ParserState.Comment;
* break;
* case '"':
* currentState = ParserState.String;
* break;
* case '/':
* if(currentState == ParserState.TagOpen)
* ignoreTag = true;
* break;
* case '>':
* currentState = ParserState.None;
* ignoreTag = !IsWantedTag(currentTag);
* if(currentTag.Length == 0)
* Console.WriteLine("Empty tag 2");
* break;
* default:
* if(currentState == ParserState.TagOpen)
* {
* if(!Char.IsWhiteSpace(currentChar))
* currentState = ParserState.TagName;
* }
*
* if(currentState == ParserState.TagName)
* {
* if(!Char.IsWhiteSpace(currentChar))
* {
* if(currentTag.Length == iLongestSearchTag)
* ignoreTag = true; // We no longer want this tag
* else
* currentTag += currentChar;
* }
* else
* {
* ignoreTag = !IsWantedTag(currentTag);
* }
* }
* break;
* }
* }
* }
* //}
* //catch
* //{
* // Woopsie daisy, looks like we won't have a preview here either
* // Add appropriate log message
* // return null;
* //}
* }
*
* return null;
* }
*/
#endregion
/// <summary>
/// Parses a given string, attempts to extract useful data for a URL Preview
/// </summary>
/// <param name="str">The string to extract data from</param>
/// <returns>A URLPreview or null</returns>
private static URLPreview ParseResponseString(String str)
{
URLPreview preview = new URLPreview();
bool conditionMetTitle = false;
bool conditionMetImage = false;
string title = String.Empty;
string imageUrl = String.Empty;
string description = String.Empty;
int strLength = str.Length;
//try
//{
bool ignoreTag = false;
int tagIndex = -1;
ParserState currentState = ParserState.None;
ParserLocation currentLocation = ParserLocation.HTML;
char previousChar = 'a';
char currentChar = 'a';
string currentTag = "";
string currentTagContent = "";
for (int x = 0; x < strLength; x++)
{
previousChar = currentChar;
currentChar = (char)str[x];
// There is a very specific order here
// 1. Check if we are in a comment (comments are always ignored)
// 2. Check if we are in a string, if we are check if the string doesn't end here
// 2.5 Check if we're parsing content, if we are we want this string data
// 3. Check if we are ignoring a tag, if we are we check if the tag wasn't closed
// 4. If we aren't ignoring a tag, we handle the rest
// Check if we are still in a comment or string
if (currentLocation == ParserLocation.Comment)
{
if (currentChar == '>' && previousChar == '-')
{
currentLocation = ParserLocation.HTML;
}
continue; // Current character is inside a command or the last character of a comment, do nothing
}
else if (currentLocation == ParserLocation.String)
{
// If we're parsing tag content we want the string data
if (currentState == ParserState.TagAttributes || currentState == ParserState.Content)
{
currentTagContent += currentChar;
}
// Check if the string ends here
if (currentChar == '"')
{
currentLocation = ParserLocation.HTML;
}
continue; // Current character is inside a string or the last character of a string, do nothing
}
else if (ignoreTag)
{
if (currentChar != '<')
{
continue; // We're ignoring this tag until we find a new opening char, do nothing
}
// New tag, let it flow through and continue with the code
ignoreTag = false;
}
// If we aren't ignoring this tag we continue whatever we want to continue
if (!ignoreTag)
{
switch (currentChar)
{
case '<':
if (currentState == ParserState.Content)
{
ContentParser(preview, currentTag, currentTagContent);
//Console.WriteLine("Tag: " + currentTag + ", Content: " + currentTagContent);
}
currentState = ParserState.TagOpen;
currentTag = "";
currentTagContent = "";
break;
case '>':
if (currentState == ParserState.TagAttributes)
{
ContentParser(preview, currentTag, currentTagContent);
currentState = ParserState.None;
//Console.WriteLine("Tag: " + currentTag + ", Attributes: " + currentTagContent);
}
else
{
tagIndex = IsWantedTag(currentTag);
if (tagIndex == -1)
{
currentState = ParserState.None;
ignoreTag = true;
}
else
{
currentState = searchTags[tagIndex].parseMethod;
}
}
break;
case '-':
// There are tags like <!DOCTYPE, so we can't just check for !
if (previousChar == '!')
{
if (currentState == ParserState.TagOpen)
{
currentState = ParserState.TagName;
currentLocation = ParserLocation.Comment;
}
}
break;
case '"':
currentLocation = ParserLocation.String;
if (currentState == ParserState.TagAttributes || currentState == ParserState.Content)
{
currentTagContent += '"';
}
break;
case '/':
if (currentState == ParserState.TagOpen)
{
ignoreTag = true;
}
break;
default:
// If we're parsing tag content we want these characters
if (currentState == ParserState.TagAttributes || currentState == ParserState.Content)
{
currentTagContent += currentChar;
}
else
{
// If the parser is at a tag open '<' and the next character is not a whitespace, then we're reading the tag name
if (currentState == ParserState.TagOpen)
{
if (!Char.IsWhiteSpace(currentChar))
{
currentState = ParserState.TagName;
}
}
// If we're at the tag name we want to store that in a variable
if (currentState == ParserState.TagName)
{
if (!Char.IsWhiteSpace(currentChar))
{
if (currentTag.Length == iLongestSearchTag)
{
ignoreTag = true; // We no longer want this tag
}
else
{
currentTag += currentChar;
}
}
else
{
tagIndex = IsWantedTag(currentTag);
if (tagIndex == -1)
{
ignoreTag = true;
}
else
{
currentState = searchTags[tagIndex].parseMethod;
}
}
}
}
break;
}
}
if (preview.PreferredTitle && preview.PreferredImageUrl && preview.PreferredDescription)
{
return(preview);
}
}
//}
//catch
//{
// Woopsie daisy, looks like we won't have a preview here either
// Add appropriate log message
// return null;
//}
return(preview);
}
