1 files changed, 38 insertions, 17 deletions

diff --git a/lib/util.cc b/lib/util.cc
index 416e2cc..a9b9bed 100644
--- a/lib/util.cc
+++ b/lib/util.cc
@@ -114,22 +114,32 @@ namespace opkele {
 	}
 
 	time_t w3c_to_time(const string& w) {
+	    int fraction;
 	    struct tm tm_t;
 	    memset(&tm_t,0,sizeof(tm_t));
-	    if(
+	    if( (
+			sscanf(
+			    w.c_str(),
+			    "%04d-%02d-%02dT%02d:%02d:%02dZ",
+			    &tm_t.tm_year,&tm_t.tm_mon,&tm_t.tm_mday,
+			    &tm_t.tm_hour,&tm_t.tm_min,&tm_t.tm_sec
+			    ) != 6
+		) && (
 		    sscanf(
 			w.c_str(),
-			"%04d-%02d-%02dT%02d:%02d:%02dZ",
+			"%04d-%02d-%02dT%02d:%02d:%02d.%03dZ",
 			&tm_t.tm_year,&tm_t.tm_mon,&tm_t.tm_mday,
-			&tm_t.tm_hour,&tm_t.tm_min,&tm_t.tm_sec
-		    ) != 6 )
+			&tm_t.tm_hour,&tm_t.tm_min,&tm_t.tm_sec,
+			&fraction
+			) != 7
+		    ) )
 		throw failed_conversion(OPKELE_CP_ "failed to sscanf()");
 	    tm_t.tm_mon--;
 	    tm_t.tm_year-=1900;
 	    time_t rv = mktime(&tm_t);
 	    if(rv==(time_t)-1)
 		throw failed_conversion(OPKELE_CP_ "failed to mktime()");
-	    return rv;
+	    return rv-timezone;
 	}
 
 	/*
@@ -164,7 +174,7 @@ namespace opkele {
 	/*
 	 * Normalize URL according to the rules, described in rfc 3986, section 6
 	 *
-	 * - uppercase hext triplets (e.g. %ab -> %AB)
+	 * - uppercase hex triplets (e.g. %ab -> %AB)
 	 * - lowercase scheme and host
 	 * - decode %-encoded characters, specified as unreserved in rfc 3986, section 2.3,
 	 *   that is - [:alpha:][:digit:]._~-
@@ -173,30 +183,40 @@ namespace opkele {
 	 * - if there's no path component, add '/'
 	 */
 	 string rfc_3986_normalize_uri(const string& uri) {
+	     static const char *whitespace = " \t\r\n";
 	     string rv;
-	     string::size_type colon = uri.find(':');
+	     string::size_type ns = uri.find_first_not_of(whitespace);
+	     if(ns==string::npos)
+		 throw bad_input(OPKELE_CP_ "Can't normalize empty URI");
+	     string::size_type colon = uri.find(':',ns);
 	     if(colon==string::npos)
 		 throw bad_input(OPKELE_CP_ "No scheme specified in URI");
 	     transform(
-		     uri.begin(), uri.begin()+colon+1,
+		     uri.begin()+ns, uri.begin()+colon+1,
 		     back_inserter(rv), ::tolower );
 	     bool s;
-	     if(rv=="http:")
-		 s = false;
-	     else if(rv=="https:")
-		 s = true;
-	     else
-		 throw not_implemented(OPKELE_CP_ "Only http(s) URIs can be normalized here");
-	     string::size_type ul = uri.length();
+	     string::size_type ul = uri.find_last_not_of(whitespace)+1;
 	     if(ul <= (colon+3))
 		 throw bad_input(OPKELE_CP_ "Unexpected end of URI being normalized encountered");
 	     if(uri[colon+1]!='/' || uri[colon+2]!='/')
 		 throw bad_input(OPKELE_CP_ "Unexpected input in URI being normalized after scheme component");
+	     if(rv=="http:")
+		 s = false;
+	     else if(rv=="https:")
+		 s = true;
+	     else{
+		 /* TODO: support more schemes.
+		  * e.g. xri. How do we normalize
+		  * xri?
+		  */
+		 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.end(),
+			 uri.begin()+colon+3,uri.begin()+ul,
 			 back_inserter(rv), ::tolower );
 		 rv += '/'; return rv;
 	     }
@@ -285,7 +305,8 @@ namespace opkele {
 		 }
 	     }
 	     if(!pseg.empty()) {
-		 rv += '/'; rv += pseg;
+		 if(!qf) rv += '/';
+		 rv += pseg;
 	     }
 	     return rv;
 	 }