private static string sniffEncoding(PeterO.Support.InputStream s)
{
byte[] data=new byte[4];
int count=0;
s.mark(data.Length+2);
try {
count=s.Read(data,0,data.Length);
} finally {
s.reset();
}
if(count>=2 && (data[0]&0xFF)==0xfe && (data[1]&0xFF)==0xff)
return "utf-16be";
if(count>=2 && (data[0]&0xFF)==0xff && (data[1]&0xFF)==0xfe)
return "utf-16le";
if(count>=3 && (data[0]&0xFF)==0xef && (data[1]&0xFF)==0xbb &&
(data[2]&0xFF)==0xbf)
return "utf-8";
if(count>=4 && (data[0]&0xFF)==0x00 && data[1]==0x3c &&
data[2]==0x00 && data[3]==0x3f)
return "utf-16be";
if(count>=4 && data[0]==0x3c && data[1]==0x00 &&
data[2]==0x3f && data[3]==0x00)
return "utf-16le";
if(count>=4 && data[0]==0x3c && data[1]==0x3f &&
data[2]==0x78 && data[3]==0x6d){ // <?xm
data=new byte[128];
s.mark(data.Length+2);
try {
count=s.Read(data,0,data.Length);
} finally {
s.reset();
}
int i=4;
if(i+1>count)return "utf-8";
if(data[i++]!='l')return "utf-8"; // l in <?xml
bool space=false;
while(i<count){
if(data[i]==0x09||data[i]==0x0a||data[i]==0x0d||data[i]==0x20)
{ space=true; i++; } else {
break;
}
}
if(!space || i+7>count)return "utf-8";
if(!(data[i]=='v' && data[i+1]=='e' && data[i+2]=='r' &&
data[i+3]=='s' && data[i+4]=='i' && data[i+5]=='o' &&
data[i+6]=='n'))return "utf-8";
i+=7;
while(i<count){
if(data[i]==0x09||data[i]==0x0a||data[i]==0x0d||data[i]==0x20) {
i++;
} else {
break;
}
}
if(i+1>count || data[i++]!='=')return "utf-8";
while(i<count){
if(data[i]==0x09||data[i]==0x0a||data[i]==0x0d||data[i]==0x20) {
i++;
} else {
break;
}
}
if(i+1>count)return "utf-8";
int ch=data[i++];
if(ch!='"' && ch!='\'')return "utf-8";
while(i<count){
if(data[i]==ch){ i++; break; }
i++;
}
space=false;
while(i<count){
if(data[i]==0x09||data[i]==0x0a||data[i]==0x0d||data[i]==0x20)
{ space=true; i++; } else {
break;
}
}
if(i+8>count)return "utf-8";
if(!(data[i]=='e' && data[i+1]=='n' && data[i+2]=='c' &&
data[i+3]=='o' && data[i+4]=='d' && data[i+5]=='i' &&
data[i+6]=='n' && data[i+7]=='g'))return "utf-8";
i+=8;
while(i<count){
if(data[i]==0x09||data[i]==0x0a||data[i]==0x0d||data[i]==0x20) {
i++;
} else {
break;
}
}
if(i+1>count || data[i++]!='=')return "utf-8";
while(i<count){
if(data[i]==0x09||data[i]==0x0a||data[i]==0x0d||data[i]==0x20) {
i++;
} else {
break;
}
}
if(i+1>count)return "utf-8";
ch=data[i++];
if(ch!='"' && ch!='\'')return "utf-8";
StringBuilder builder=new StringBuilder();
while(i<count){
if(data[i]==ch){
string encoding=TextEncoding.resolveEncoding(builder.ToString());
if(encoding==null)
return null;
if(encoding.Equals("utf-16le") || encoding.Equals("utf-16be"))
return null;
return builder.ToString();
}
builder.Append((char)data[i]);
i++;
}
return "utf-8";
}
return "utf-8";
}