private static string GetSpecial(StringGetter r)
{
StringBuilder result = new StringBuilder();
r.Mark(1);//Mark the present position in the stream
int c = r.Read();
while (IsLetter((char)c))
{
result.Append((char)c);
r.Mark(1);
c = r.Read();
}
if (c == ';')
{
result.Append(';');
}
else
{
r.Reset();
}
return(result.ToString());
}
private static string GetTag(StringGetter r)
{
StringBuilder result = new StringBuilder();
int level = 1;
result.Append('<');
while (level > 0)
{
int c = r.Read();
if (c == -1)
{
break; // EOF
} // EOF
result.Append((char)c);
if (c == '<')
{
level++;
}
else if (c == '>')
{
level--;
}
}
return(result.ToString());
}
////////////////////////////////////////////////////////////////////////////
//--------------------------------- REVISIONS ------------------------------
// Date Name Tracking # Description
// --------- ------------------- ------------- ----------------------
// 04JAN2009 James Shen Initial Creation
////////////////////////////////////////////////////////////////////////////
/**
* Convert \\uxxxx to it's string format
* @param source string to Convert.
* @return result string.
*/
public static string ConvertUTF8(string source)
{
StringBuilder result2 = new StringBuilder();
StringGetter input = new StringGetter(source);
try
{
int c = input.Read();
while (c != -1) // Convert until EOF
{
string text;
if (c == '\\')
{
c = input.Read();
switch ((char)c)
{
case 'u':
text = "";
for (int i = 0; i < 4; i++)
{
text += ((char)input.Read()).ToString();
}
text = int.Parse(text, NumberStyles.HexNumber).ToString();
break;
case 'x':
text = "";
for (int i = 0; i < 2; i++)
{
text += ((char)input.Read());
}
text = int.Parse(text, NumberStyles.HexNumber).ToString();
break;
default:
text = "\\" + (char)c;
break;
}
}
else
{
text = "" + (char)c;
}
StringBuilder s = result2;
s.Append(text);
c = input.Read();
}
}
catch
{
input.Close();
}
return result2.ToString().Trim();
}
////////////////////////////////////////////////////////////////////////////
//--------------------------------- REVISIONS ------------------------------
// Date Name Tracking # Description
// --------- ------------------- ------------- ----------------------
// 04JAN2009 James Shen Initial Creation
////////////////////////////////////////////////////////////////////////////
/**
* Convert \\uxxxx to it's string format
* @param source string to Convert.
* @return result string.
*/
public static string ConvertUTF8(string source)
{
StringBuilder result2 = new StringBuilder();
StringGetter input = new StringGetter(source);
try
{
int c = input.Read();
while (c != -1) // Convert until EOF
{
string text;
if (c == '\\')
{
c = input.Read();
switch ((char)c)
{
case 'u':
text = "";
for (int i = 0; i < 4; i++)
{
text += ((char)input.Read()).ToString();
}
text = int.Parse(text, NumberStyles.HexNumber).ToString();
break;
case 'x':
text = "";
for (int i = 0; i < 2; i++)
{
text += ((char)input.Read());
}
text = int.Parse(text, NumberStyles.HexNumber).ToString();
break;
default:
text = "\\" + (char)c;
break;
}
}
else
{
text = "" + (char)c;
}
StringBuilder s = result2;
s.Append(text);
c = input.Read();
}
}
catch
{
input.Close();
}
return(result2.ToString().Trim());
}
////////////////////////////////////////////////////////////////////////////
//--------------------------------- REVISIONS ------------------------------
// Date Name Tracking # Description
// --------- ------------------- ------------- ----------------------
// 04JAN2009 James Shen Initial Creation
////////////////////////////////////////////////////////////////////////////
/**
* Remove double "\\" and change to one "\"
* @param source the string need to change
* @return the result string.
*/
public static string RemoveDoubleBackSlash(string source)
{
StringBuilder result2 = new StringBuilder();
StringGetter input = new StringGetter(source);
try
{
int c = input.Read();
while (c != -1) // Convert until EOF
{
string text;
if (c == '\\')
{
c = input.Read();
if (c == '\\')
{
text = "\\";
}
else
{
text = "\\" + (char)c;
}
}
else
{
text = "" + (char)c;
}
StringBuilder s = result2;
s.Append(text);
c = input.Read();
}
}
catch
{
input.Close();
}
return(result2.ToString().Trim());
}
private static string GetTag(StringGetter r)
{
StringBuilder result = new StringBuilder();
int level = 1;
result.Append('<');
while (level > 0)
{
int c = r.Read();
if (c == -1)
{
break; // EOF
} // EOF
result.Append((char)c);
if (c == '<')
{
level++;
}
else if (c == '>')
{
level--;
}
}
return result.ToString();
}
private static string GetSpecial(StringGetter r)
{
StringBuilder result = new StringBuilder();
r.Mark(1);//Mark the present position in the stream
int c = r.Read();
while (IsLetter((char)c))
{
result.Append((char)c);
r.Mark(1);
c = r.Read();
}
if (c == ';')
{
result.Append(';');
}
else
{
r.Reset();
}
return result.ToString();
}
//[------------------------------ PUBLIC METHODS --------------------------]
////////////////////////////////////////////////////////////////////////////
//--------------------------------- REVISIONS ------------------------------
// Date Name Tracking # Description
// --------- ------------------- ------------- ----------------------
// 04JAN2009 James Shen Initial Creation
////////////////////////////////////////////////////////////////////////////
/**
* Convert html to plain text
* @param source HTML string.
* @return plain text.
*/
public string Convert(string source)
{
StringBuilder result = new StringBuilder();
StringBuilder result2 = new StringBuilder();
StringGetter input = new StringGetter(source);
try
{
int c = input.Read();
while (c != -1) // Convert until EOF
{
string text;
if (c == '<') // It's a tag!!
{
string currentTag = GetTag(input); // Get the rest of the tag
text = ConvertTag(currentTag);
}
else if (c == '&')
{
string specialchar = GetSpecial(input);
if (specialchar.Equals("lt;") || specialchar.Equals("#60"))
{
text = "<";
}
else if (specialchar.Equals("gt;") || specialchar.Equals("#62"))
{
text = ">";
}
else if (specialchar.Equals("amp;") || specialchar.Equals("#38"))
{
text = "&";
}
else if (specialchar.Equals("nbsp;"))
{
text = " ";
}
else if (specialchar.Equals("quot;") || specialchar.Equals("#34"))
{
text = "\"";
}
else if (specialchar.Equals("copy;") || specialchar.Equals("#169"))
{
text = "[Copyright]";
}
else if (specialchar.Equals("reg;") || specialchar.Equals("#174"))
{
text = "[Registered]";
}
else if (specialchar.Equals("trade;") || specialchar.Equals("#153"))
{
text = "[Trademark]";
}
else
{
text = "&" + specialchar;
}
}
else if (!_pre && IsWhitespace((char)c))
{
StringBuilder s = _inBody ? result : result2;
if (s.Length > 0 && IsWhitespace(s[s.Length - 1]))
{
text = "";
}
else
{
text = " ";
}
}
else
{
text = "" + (char)c;
}
StringBuilder s2 = _inBody ? result : result2;
s2.Append(text);
c = input.Read();
}
}
catch
{
input.Close();
}
StringBuilder s1 = _bodyFound ? result : result2;
return s1.ToString().Trim();
}
////////////////////////////////////////////////////////////////////////////
//--------------------------------- REVISIONS ------------------------------
// Date Name Tracking # Description
// --------- ------------------- ------------- ----------------------
// 04JAN2009 James Shen Initial Creation
////////////////////////////////////////////////////////////////////////////
/**
* Remove double "\\" and change to one "\"
* @param source the string need to change
* @return the result string.
*/
public static string RemoveDoubleBackSlash(string source)
{
StringBuilder result2 = new StringBuilder();
StringGetter input = new StringGetter(source);
try
{
int c = input.Read();
while (c != -1) // Convert until EOF
{
string text;
if (c == '\\')
{
c = input.Read();
if (c == '\\')
{
text = "\\";
}
else
{
text = "\\" + (char)c;
}
}
else
{
text = "" + (char)c;
}
StringBuilder s = result2;
s.Append(text);
c = input.Read();
}
}
catch
{
input.Close();
}
return result2.ToString().Trim();
}
//[------------------------------ PUBLIC METHODS --------------------------]
////////////////////////////////////////////////////////////////////////////
//--------------------------------- REVISIONS ------------------------------
// Date Name Tracking # Description
// --------- ------------------- ------------- ----------------------
// 04JAN2009 James Shen Initial Creation
////////////////////////////////////////////////////////////////////////////
/**
* Convert html to plain text
* @param source HTML string.
* @return plain text.
*/
public string Convert(string source)
{
StringBuilder result = new StringBuilder();
StringBuilder result2 = new StringBuilder();
StringGetter input = new StringGetter(source);
try
{
int c = input.Read();
while (c != -1) // Convert until EOF
{
string text;
if (c == '<') // It's a tag!!
{
string currentTag = GetTag(input); // Get the rest of the tag
text = ConvertTag(currentTag);
}
else if (c == '&')
{
string specialchar = GetSpecial(input);
if (specialchar.Equals("lt;") || specialchar.Equals("#60"))
{
text = "<";
}
else if (specialchar.Equals("gt;") || specialchar.Equals("#62"))
{
text = ">";
}
else if (specialchar.Equals("amp;") || specialchar.Equals("#38"))
{
text = "&";
}
else if (specialchar.Equals("nbsp;"))
{
text = " ";
}
else if (specialchar.Equals("quot;") || specialchar.Equals("#34"))
{
text = "\"";
}
else if (specialchar.Equals("copy;") || specialchar.Equals("#169"))
{
text = "[Copyright]";
}
else if (specialchar.Equals("reg;") || specialchar.Equals("#174"))
{
text = "[Registered]";
}
else if (specialchar.Equals("trade;") || specialchar.Equals("#153"))
{
text = "[Trademark]";
}
else
{
text = "&" + specialchar;
}
}
else if (!_pre && IsWhitespace((char)c))
{
StringBuilder s = _inBody ? result : result2;
if (s.Length > 0 && IsWhitespace(s[s.Length - 1]))
{
text = "";
}
else
{
text = " ";
}
}
else
{
text = "" + (char)c;
}
StringBuilder s2 = _inBody ? result : result2;
s2.Append(text);
c = input.Read();
}
}
catch
{
input.Close();
}
StringBuilder s1 = _bodyFound ? result : result2;
return(s1.ToString().Trim());
}
