private static string GetCharacterEncoding(URLConnection urlConnection)
{
ParsedContentType pct = new ParsedContentType(urlConnection.GetContentType());
string encoding = pct.GetEncoding();
if (encoding != null)
{
return encoding;
}
string contentType = pct.GetContentType();
if (contentType != null && contentType.StartsWith("text/"))
{
return "8859_1";
}
else
{
return "utf-8";
}
}
/// <exception cref="System.IO.IOException"></exception>
public static object ReadFileOrUrl(string path, bool convertToString, string defaultEncoding)
{
Uri url = ToUrl(path);
Stream @is = null;
int capacityHint = 0;
string encoding;
string contentType;
byte[] data;
try
{
if (url == null)
{
FilePath file = new FilePath(path);
contentType = encoding = null;
capacityHint = (int)file.Length();
@is = new FileInputStream(file);
}
else
{
URLConnection uc = url.OpenConnection();
@is = uc.GetInputStream();
if (convertToString)
{
ParsedContentType pct = new ParsedContentType(uc.GetContentType());
contentType = pct.GetContentType();
encoding = pct.GetEncoding();
}
else
{
contentType = encoding = null;
}
capacityHint = uc.GetContentLength();
// Ignore insane values for Content-Length
if (capacityHint > (1 << 20))
{
capacityHint = -1;
}
}
if (capacityHint <= 0)
{
capacityHint = 4096;
}
data = Kit.ReadStream(@is, capacityHint);
}
finally
{
if (@is != null)
{
@is.Close();
}
}
object result;
if (!convertToString)
{
result = data;
}
else
{
if (encoding == null)
{
// None explicitly specified in Content-type header. Use RFC-4329
// 4.2.2 section to autodetect
if (data.Length > 3 && data[0] == -1 && data[1] == -2 && data[2] == 0 && data[3] == 0)
{
encoding = "UTF-32LE";
}
else
{
if (data.Length > 3 && data[0] == 0 && data[1] == 0 && data[2] == -2 && data[3] == -1)
{
encoding = "UTF-32BE";
}
else
{
if (data.Length > 2 && data[0] == -17 && data[1] == -69 && data[2] == -65)
{
encoding = "UTF-8";
}
else
{
if (data.Length > 1 && data[0] == -1 && data[1] == -2)
{
encoding = "UTF-16LE";
}
else
{
if (data.Length > 1 && data[0] == -2 && data[1] == -1)
{
encoding = "UTF-16BE";
}
else
{
// No autodetect. See if we have explicit value on command line
encoding = defaultEncoding;
if (encoding == null)
{
// No explicit encoding specification
if (url == null)
{
// Local files default to system encoding
encoding = Runtime.GetProperty("file.encoding");
}
else
{
if (contentType != null && contentType.StartsWith("application/"))
{
// application/* types default to UTF-8
encoding = "UTF-8";
}
else
{
// text/* MIME types default to US-ASCII
encoding = "US-ASCII";
}
}
}
}
}
}
}
}
}
string strResult = Sharpen.Runtime.GetStringForBytes(data, encoding);
// Skip BOM
if (strResult.Length > 0 && strResult[0] == '\uFEFF')
{
strResult = Sharpen.Runtime.Substring(strResult, 1);
}
result = strResult;
}
return result;
}
