private static void percentEncodeUtf8(IntList buffer, int cp)
{
if(cp<=0x7F){
buffer.appendInt('%');
buffer.appendInt(hex[(cp>>4)&0x0F]);
buffer.appendInt(hex[(cp)&0x0F]);
} else if(cp<=0x7FF){
percentEncode(buffer,(0xC0|((cp>>6)&0x1F)));
percentEncode(buffer,(0x80|(cp &0x3F)));
} else if(cp<=0xFFFF){
percentEncode(buffer,(0xE0|((cp>>12)&0x0F)));
percentEncode(buffer,(0x80|((cp>>6 )&0x3F)));
percentEncode(buffer,(0x80|(cp &0x3F)));
} else {
percentEncode(buffer,(0xF0|((cp>>18)&0x07)));
percentEncode(buffer,(0x80|((cp>>12)&0x3F)));
percentEncode(buffer,(0x80|((cp>>6 )&0x3F)));
percentEncode(buffer,(0x80|(cp &0x3F)));
}
}
public static URL parse(string s, URL baseurl, string encoding, bool strict)
{
if(s==null)
throw new ArgumentException();
int beginning=0;
int ending=s.Length-1;
bool relative=false;
URL url=new URL();
ITextEncoder encoder=null;
ParseState state=ParseState.SchemeStart;
if(encoding!=null){
encoder=TextEncoding.getEncoder(encoding);
}
if(s.IndexOf("http://",StringComparison.Ordinal)==0){
state=ParseState.AuthorityIgnoreSlashes;
url.scheme="http";
beginning=7;
relative=true;
} else {
while(beginning<s.Length){
char c=s[beginning];
if(c!=0x09 && c!=0x0a && c!=0x0c && c!=0x0d && c!=0x20){
break;
}
beginning++;
}
}
while(ending>=beginning){
char c=s[ending];
if(c!=0x09 && c!=0x0a && c!=0x0c && c!=0x0d && c!=0x20){
ending++;
break;
}
ending--;
}
if(ending<beginning) {
ending=beginning;
}
bool atflag=false;
bool bracketflag=false;
IntList buffer=new IntList();
IntList query=null;
IntList fragment=null;
IntList password=null;
IntList username=null;
IntList schemeData=null;
bool error=false;
IList<string> path=new List<string>();
int index=beginning;
int hostStart=-1;
int portstate=0;
while(index<=ending){
int oldindex=index;
int c=-1;
if(index>=ending){
c=-1;
index++;
} else {
c=s[index];
if(c>=0xD800 && c<=0xDBFF && index+1<ending &&
s[index+1]>=0xDC00 && s[index+1]<=0xDFFF){
// Get the Unicode code point for the surrogate pair
c=0x10000+(c-0xD800)*0x400+(s[index+1]-0xDC00);
index++;
} else if(c>=0xD800 && c<=0xDFFF)
// illegal surrogate
throw new ArgumentException();
index++;
}
switch(state){
case ParseState.SchemeStart:
if(c>='A' && c<='Z'){
buffer.appendInt(c+0x20);
state=ParseState.Scheme;
} else if(c>='a' && c<='z'){
buffer.appendInt(c);
state=ParseState.Scheme;
} else {
index=oldindex;
state=ParseState.NoScheme;
}
break;
case ParseState.Scheme:
if(c>='A' && c<='Z'){
buffer.appendInt(c+0x20);
} else if((c>='a' && c<='z') || c=='.' || c=='-' || c=='+'){
buffer.appendInt(c);
} else if(c==':'){
url.scheme=buffer.ToString();
buffer.clearAll();
if(url.scheme.Equals("http") ||
url.scheme.Equals("https") ||
url.scheme.Equals("ftp") ||
url.scheme.Equals("gopher") ||
url.scheme.Equals("ws") ||
url.scheme.Equals("wss") ||
url.scheme.Equals("file")){
relative=true;
}
if(url.scheme.Equals("file")){
state=ParseState.Relative;
relative=true;
} else if(relative && baseurl!=null && url.scheme.Equals(baseurl.scheme)){
state=ParseState.RelativeOrAuthority;
} else if(relative){
state=ParseState.AuthorityFirstSlash;
} else {
schemeData=new IntList();
state=ParseState.SchemeData;
}
} else {
buffer.clearAll();
index=beginning;
state=ParseState.NoScheme;
}
break;
case ParseState.SchemeData:
if(c=='?'){
query=new IntList();
state=ParseState.Query;
break;
} else if(c=='#'){
fragment=new IntList();
state=ParseState.Fragment;
break;
}
if((c>=0 && (!isUrlCodePoint(c) && c!='%') || (c=='%' &&
(index+2>ending ||
!isHexDigit(s[index]) ||
!isHexDigit(s[index+1]))))){
error=true;
}
if(c>=0 && c!=0x09 && c!=0x0a && c!=0x0d){
if(c<0x20 || c==0x7F){
percentEncode(schemeData,c);
} else if(c<0x7F){
schemeData.appendInt(c);
} else {
percentEncodeUtf8(schemeData,c);
}
}
break;
case ParseState.NoScheme:
if(baseurl==null)
return null;
//Console.WriteLine("no scheme: [%s] [%s]",s,baseurl);
if(!(baseurl.scheme.Equals("http") ||
baseurl.scheme.Equals("https") ||
baseurl.scheme.Equals("ftp") ||
baseurl.scheme.Equals("gopher") ||
baseurl.scheme.Equals("ws") ||
baseurl.scheme.Equals("wss") ||
baseurl.scheme.Equals("file")
))
return null;
state=ParseState.Relative;
index=oldindex;
break;
case ParseState.RelativeOrAuthority:
if(c=='/' && index<ending && s[index]=='/'){
index++;
state=ParseState.AuthorityIgnoreSlashes;
} else {
error=true;
state=ParseState.Relative;
index=oldindex;
}
break;
case ParseState.Relative:{
relative=true;
if(!"file".Equals(url.scheme)){
url.scheme=baseurl.scheme;
}
if(c<0){
url.host=baseurl.host;
url.port=baseurl.port;
path=pathList(baseurl.path);
url.query=baseurl.query;
} else if(c=='/' || c=='\\'){
if(c=='\\') {
error=true;
}
state=ParseState.RelativeSlash;
} else if(c=='?'){
url.host=baseurl.host;
url.port=baseurl.port;
path=pathList(baseurl.path);
query=new IntList();
state=ParseState.Query;
} else if(c=='#'){
url.host=baseurl.host;
url.port=baseurl.port;
path=pathList(baseurl.path);
url.query=baseurl.query;
fragment=new IntList();
state=ParseState.Fragment;
} else {
url.host=baseurl.host;
url.port=baseurl.port;
path=pathList(baseurl.path);
if(path.Count>0) { // Pop path
path.RemoveAt(path.Count-1);
}
state=ParseState.RelativePath;
index=oldindex;
}
break;
}
case ParseState.RelativeSlash:
if(c=='/' || c=='\\'){
if(c=='\\') {
error=true;
}
if("file".Equals(url.scheme)){
state=ParseState.FileHost;
} else {
state=ParseState.AuthorityIgnoreSlashes;
}
} else {
if(baseurl!=null){
url.host=baseurl.host;
url.port=baseurl.port;
}
state=ParseState.RelativePath;
index=oldindex;
}
break;
case ParseState.AuthorityFirstSlash:
if(c=='/'){
state=ParseState.AuthoritySecondSlash;
} else {
error=true;
state=ParseState.AuthorityIgnoreSlashes;
index=oldindex;
}
break;
case ParseState.AuthoritySecondSlash:
if(c=='/'){
state=ParseState.AuthorityIgnoreSlashes;
} else {
error=true;
state=ParseState.AuthorityIgnoreSlashes;
index=oldindex;
}
break;
case ParseState.AuthorityIgnoreSlashes:
if(c!='/' && c!='\\'){
username=new IntList();
index=oldindex;
hostStart=index;
state=ParseState.Authority;
} else {
error=true;
}
break;
case ParseState.Authority:
if(c=='@'){
if(atflag){
IntList result=(password==null) ? username : password;
error=true;
result.appendInt('%');
result.appendInt('4');
result.appendInt('0');
}
atflag=true;
int[] array=buffer.array();
for(int i=0;i<buffer.Count;i++){
int cp=array[i];
if(cp==0x9 || cp==0xa || cp==0xd){
error=true;
continue;
}
if((!isUrlCodePoint(c) && c!='%') || (cp=='%' &&
(i+3>buffer.Count ||
!isHexDigit(array[index+1]) ||
!isHexDigit(array[index+2])))){
error=true;
}
if(cp==':' && password==null){
password=new IntList();
continue;
}
IntList result=(password==null) ? username : password;
if(cp<=0x20 || cp>=0x7F || ((cp&0x7F)==cp && "#<>?`\"".IndexOf((char)cp)>=0)){
percentEncodeUtf8(result,cp);
} else {
result.appendInt(cp);
}
}
//Console.WriteLine("username=%s",username);
//Console.WriteLine("password=%s",password);
buffer.clearAll();
hostStart=index;
} else if(c<0 || ((c&0x7F)==c && "/\\?#".IndexOf((char)c)>=0)){
buffer.clearAll();
state=ParseState.Host;
index=hostStart;
} else {
buffer.appendInt(c);
}
break;
case ParseState.FileHost:
if(c<0 || ((c&0x7F)==c && "/\\?#".IndexOf((char)c)>=0)){
index=oldindex;
if(buffer.Count==2){
int c1=buffer[0];
int c2=buffer[1];
if((c2=='|' || c2==':') && ((c1>='A' && c1<='Z') || (c1>='a' && c1<='z'))){
state=ParseState.RelativePath;
break;
}
}
string host=hostParse(buffer.ToString());
if(host==null)
throw new ArgumentException();
url.host=host;
buffer.clearAll();
state=ParseState.RelativePathStart;
} else if(c==0x09 || c==0x0a || c==0x0d){
error=true;
} else {
buffer.appendInt(c);
}
break;
case ParseState.Host:
case ParseState.HostName:
if(c==':' && !bracketflag){
string host=hostParse(buffer.ToString());
if(host==null)
return null;
url.host=host;
buffer.clearAll();
state=ParseState.Port;
} else if(c<0 || ((c&0x7F)==c && "/\\?#".IndexOf((char)c)>=0)){
string host=hostParse(buffer.ToString());
if(host==null)
return null;
url.host=host;
buffer.clearAll();
index=oldindex;
state=ParseState.RelativePathStart;
} else if(c==0x09 || c==0x0a || c==0x0d){
error=true;
} else {
if(c=='[') {
bracketflag=true;
} else if(c==']') {
bracketflag=false;
}
buffer.appendInt(c);
}
break;
case ParseState.Port:
if(c>='0' && c<='9'){
if(c!='0') {
portstate=2; // first non-zero found
} else if(portstate==0){
portstate=1; // have a port number
}
if(portstate==2) {
buffer.appendInt(c);
}
} else if(c<0 || ((c&0x7F)==c && "/\\?#".IndexOf((char)c)>=0)){
string bufport="";
if(portstate==1) {
bufport="0";
} else if(portstate==2) {
bufport=buffer.ToString();
}
//Console.WriteLine("port: [%s]",buffer.ToString());
if((url.scheme.Equals("http") || url.scheme.Equals("ws"))
&& bufport.Equals("80")) {
bufport="";
}
if((url.scheme.Equals("https") || url.scheme.Equals("wss"))
&& bufport.Equals("443")) {
bufport="";
}
if((url.scheme.Equals("gopher"))
&& bufport.Equals("70")) {
bufport="";
}
if((url.scheme.Equals("ftp"))
&& bufport.Equals("21")) {
bufport="";
}
url.port=bufport;
buffer.clearAll();
state=ParseState.RelativePathStart;
index=oldindex;
} else if(c==0x09 || c==0x0a || c==0x0d){
error=true;
} else
return null;
break;
case ParseState.Query:
if(c<0 || c=='#'){
bool utf8=true;
if(relative){
utf8=true;
}
if(utf8 || encoder==null){
// NOTE: Encoder errors can never happen in
// this case
for(int i=0;i<buffer.Count;i++){
int ch=buffer[i];
if(ch<0x21 || ch>0x7e || ch==0x22 || ch==0x23 ||
ch==0x3c || ch==0x3e || ch==0x60){
percentEncodeUtf8(query,ch);
} else {
query.appendInt(ch);
}
}
} else {
try {
MemoryOutputStream baos=new MemoryOutputStream();
encoder.encode(baos,buffer.array(),0,buffer.Count,encodingError);
byte[] bytes=baos.toByteArray();
foreach(var ch in bytes) {
if(ch<0x21 || ch>0x7e || ch==0x22 || ch==0x23 ||
ch==0x3c || ch==0x3e || ch==0x60){
percentEncode(query,ch);
} else {
query.appendInt(ch);
}
}
baos.Close();
} catch (IOException e) {
throw e;
}
throw new InvalidOperationException();
}
buffer.clearAll();
if(c=='#'){
fragment=new IntList();
state=ParseState.Fragment;
}
} else if(c==0x09 || c==0x0a || c==0x0d){
error=true;
} else {
if((!isUrlCodePoint(c) && c!='%') || (c=='%' &&
(index+2>ending ||
!isHexDigit(s[index]) ||
!isHexDigit(s[index+1])))){
error=true;
}
buffer.appendInt(c);
}
break;
case ParseState.RelativePathStart:
if(c=='\\'){
error=true;
}
state=ParseState.RelativePath;
if((c!='/' && c!='\\')){
index=oldindex;
}
break;
case ParseState.RelativePath:
if((c<0 || c=='/' || c=='\\') ||
(c=='?' || c=='#')){
if(c=='\\') {
error=true;
}
if(buffer.Count==2 && buffer[0]=='.'
&& buffer[1]=='.'){
if(path.Count>0){
path.RemoveAt(path.Count-1);
}
if((c!='/' && c!='\\')){
path.Add("");
}
} else if(buffer.Count==1 && buffer[0]=='.'){
if((c!='/' && c!='\\')){
path.Add("");
}
} else {
if("file".Equals(url.scheme) && path.Count==0 &&
buffer.Count==2){
int c1=buffer[0];
int c2=buffer[1];
if((c2=='|' || c2==':') && ((c1>='A' && c1<='Z') || (c1>='a' && c1<='z'))){
buffer[1]=':';
}
}
path.Add(buffer.ToString());
}
buffer.clearAll();
if(c=='?'){
query=new IntList();
state=ParseState.Query;
}
if(c=='#'){
fragment=new IntList();
state=ParseState.Fragment;
}
} else if(c=='%' && index+2<=ending &&
s[index]=='2' &&
(s[index+1]=='e' || s[index+1]=='E')){
index+=2;
buffer.appendInt('.');
} else if(c==0x09 || c==0x0a || c==0x0d){
error=true;
} else {
if((!isUrlCodePoint(c) && c!='%') || (c=='%' &&
(index+2>ending ||
!isHexDigit(s[index]) ||
!isHexDigit(s[index+1])))){
error=true;
}
if(c<=0x20 || c>=0x7F || ((c&0x7F)==c && "#<>?`\"".IndexOf((char)c)>=0)){
percentEncodeUtf8(buffer,c);
} else {
buffer.appendInt(c);
}
}
break;
case ParseState.Fragment:
if(c<0) {
break;
}
if(c==0x09 || c==0x0a || c==0x0d) {
error=true;
} else {
if((!isUrlCodePoint(c) && c!='%') || (c=='%' &&
(index+2>ending ||
!isHexDigit(s[index]) ||
!isHexDigit(s[index+1])))){
error=true;
}
if(c<0x20 || c==0x7F){
percentEncode(fragment,c);
} else if(c<0x7F){
fragment.appendInt(c);
} else {
percentEncodeUtf8(fragment,c);
}
}
break;
default:
throw new InvalidOperationException();
}
}
if(error && strict)
return null;
if(schemeData!=null) {
url.schemeData=schemeData.ToString();
}
StringBuilder builder=new StringBuilder();
if(path.Count==0){
builder.Append('/');
} else {
foreach(var segment in path){
builder.Append('/');
builder.Append(segment);
}
}
url.path=builder.ToString();
if(query!=null) {
url.query=query.ToString();
}
if(fragment!=null) {
url.fragment=fragment.ToString();
}
if(password!=null) {
url.password=password.ToString();
}
if(username!=null) {
url.username=username.ToString();
}
return url;
}
private static void percentEncode(IntList buffer, int b)
{
buffer.appendInt('%');
buffer.appendInt(hex[(b>>4)&0x0F]);
buffer.appendInt(hex[(b)&0x0F]);
}