fix to rfc normalization

It kept prepending a '/' to the trailing segment even if the segment was past
[?#]

Signed-off-by: Michael Krelin <hacker@klever.net>

1 files changed, 2 insertions, 1 deletions

diff --git a/lib/util.cc b/lib/util.cc
index ee75d29..a9b9bed 100644
--- a/lib/util.cc
+++ b/lib/util.cc
@@ -212,104 +212,105 @@ namespace opkele {
 		 rv.append(uri,colon+1,ul-colon-1);
 		 return rv;
 	     }
 	     rv += "//";
 	     string::size_type interesting = uri.find_first_of(":/#?",colon+3);
 	     if(interesting==string::npos) {
 		 transform(
 			 uri.begin()+colon+3,uri.begin()+ul,
 			 back_inserter(rv), ::tolower );
 		 rv += '/'; return rv;
 	     }
 	     transform(
 		     uri.begin()+colon+3,uri.begin()+interesting,
 		     back_inserter(rv), ::tolower );
 	     bool qf = false;
 	     char ic = uri[interesting];
 	     if(ic==':') {
 		 string::size_type ni = uri.find_first_of("/#?%",interesting+1);
 		 const char *nptr = uri.data()+interesting+1;
 		 char *eptr = 0;
 		 long port = strtol(nptr,&eptr,10);
 		 if( (port>0) && (port<65535) && port!=(s?443:80) ) {
 		     char tmp[8];
 		     snprintf(tmp,sizeof(tmp),":%ld",port);
 		     rv += tmp;
 		 }
 		 if(ni==string::npos) {
 		     rv += '/'; return rv;
 		 }
 		 interesting = ni;
 	     }else if(ic!='/') {
 		 rv += '/'; rv += ic;
 		 qf = true;
 		 ++interesting;
 	     }
 	     string::size_type n = interesting;
 	     char tmp[3] = { 0,0,0 };
 	     stack<string::size_type> psegs; psegs.push(rv.length());
 	     string pseg;
 	     for(;n<ul;) {
 		 string::size_type unsafe = uri.find_first_of(qf?"%":"%/?#",n);
 		 if(unsafe==string::npos) {
 		     pseg.append(uri,n,ul-n-1); n = ul-1;
 		 }else{
 		     pseg.append(uri,n,unsafe-n);
 		     n = unsafe;
 		 }
 		 char c = uri[n++];
 		 if(c=='%') {
 		     if((n+1)>=ul)
 			 throw bad_input(OPKELE_CP_ "Unexpected end of URI encountered while parsing percent-encoded character");
 		     tmp[0] = uri[n++];
 		     tmp[1] = uri[n++];
 		     if(!( isxdigit(tmp[0]) && isxdigit(tmp[1]) ))
 			 throw bad_input(OPKELE_CP_ "Invalid percent-encoded character in URI being normalized");
 		     int cc = strtol(tmp,0,16);
 		     if( isalpha(cc) || isdigit(cc) || strchr("._~-",cc) )
 			 pseg += cc;
 		     else{
 			 pseg += '%';
 			 pseg += toupper(tmp[0]); pseg += toupper(tmp[1]);
 		     }
 		 }else if(qf) {
 		     rv += pseg; rv += c;
 		     pseg.clear();
 		 }else if(n>=ul || strchr("?/#",c)) {
 		     if(pseg.empty() || pseg==".") {
 		     }else if(pseg=="..") {
 			 if(psegs.size()>1) {
 			     rv.resize(psegs.top()); psegs.pop();
 			 }
 		     }else{
 			 psegs.push(rv.length());
 			 if(c!='/') {
 			     pseg += c;
 			     qf = true;
 			 }
 			 rv += '/'; rv += pseg;
 		     }
 		     if(c=='/' && (n>=ul || strchr("?#",uri[n])) ) {
 			 rv += '/';
 			 if(n<ul)
 			     qf = true;
 		     }else if(strchr("?#",c)) {
 			 if(psegs.size()==1 && psegs.top()==rv.length())
 			     rv += '/';
 			 if(pseg.empty())
 			     rv += c;
 			 qf = true;
 		     }
 		     pseg.clear();
 		 }else{
 		     pseg += c;
 		 }
 	     }
 	     if(!pseg.empty()) {
-		 rv += '/'; rv += pseg;
+		 if(!qf) rv += '/';
+		 rv += pseg;
 	     }
 	     return rv;
 	 }
 
     }
 
 }