added an identifier normalization utility function

* moved iname leader characters and whitespace characters strings to
  opkele::data namespace
* added opkele::util::normalize_identifier() function

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

5 files changed, 53 insertions, 12 deletions

diff --git a/include/opkele/data.h b/include/opkele/data.h
index d0b0516..904b5ae 100644
--- a/include/opkele/data.h
+++ b/include/opkele/data.h
@@ -11,8 +11,10 @@ namespace opkele {
      */
     namespace data {
 	extern const char *_default_p;
 	extern const char *_default_g;
+	extern const char *_iname_leaders;
+	extern const char *_whitespace_chars;
     }
 }
 
 #endif /* __OPKELE_DATA_H */
diff --git a/include/opkele/util.h b/include/opkele/util.h
index 60955e1..fd974a1 100644
--- a/include/opkele/util.h
+++ b/include/opkele/util.h
@@ -83,8 +83,10 @@ namespace opkele {
 	 * @throw bad_input in case of malformed URI
 	 */
 	string rfc_3986_normalize_uri(const string& uri);
 
+	string normalize_identifier(const string& usi,bool strip_fragment);
+
 	/**
 	 * Match URI against realm
 	 * @param uri URI to match
 	 * @param realm realm to match against
diff --git a/lib/data.cc b/lib/data.cc
index c040430..f71788f 100644
--- a/lib/data.cc
+++ b/lib/data.cc
@@ -6,6 +6,9 @@ namespace opkele {
 
 	const char *_default_p = "155172898181473697471232257763715539915724801966915404479707795314057629378541917580651227423698188993727816152646631438561595825688188889951272158842675419950341258706556549803580104870537681476726513255747040765857479291291572334510643245094715007229621094194349783925984760375594985848253359305585439638443";
 	const char *_default_g = "2";
 
+	const char *_iname_leaders = "=@+$!(";
+	const char *_whitespace_chars = " \t\r\n";
+
     }
 }
diff --git a/lib/discovery.cc b/lib/discovery.cc
index b7f2db6..5913ad4 100644
--- a/lib/discovery.cc
+++ b/lib/discovery.cc
@@ -5,8 +5,9 @@
 #include <opkele/discovery.h>
 #include <opkele/exception.h>
 #include <opkele/util.h>
 #include <opkele/tidy.h>
+#include <opkele/data.h>
 #include <opkele/debug.h>
 
 #include "config.h"
 
@@ -20,10 +21,8 @@ namespace opkele {
 
     /* TODO: the whole discovery thing needs cleanup and optimization due to
      * many changes of concept. */
 
-    static const char *whitespace = " \t\r\n";
-    static const char *i_leaders = "=@+$!(";
     static const size_t max_html = 16384;
 
     static const struct service_type_t {
 	const char *uri;
@@ -127,20 +126,20 @@ namespace opkele {
 
 	    string discover(endpoint_discovery_iterator& oi,const string& identity) {
 		string rv;
 		idiscovery_t idis;
-		string::size_type fsc = identity.find_first_not_of(whitespace);
+		string::size_type fsc = identity.find_first_not_of(data::_whitespace_chars);
 		if(fsc==string::npos)
 		    throw bad_input(OPKELE_CP_ "whitespace-only identity");
-		string::size_type lsc = identity.find_last_not_of(whitespace);
+		string::size_type lsc = identity.find_last_not_of(data::_whitespace_chars);
 		assert(lsc!=string::npos);
 		if(!strncasecmp(identity.c_str()+fsc,"xri://",sizeof("xri://")-1))
 		    fsc += sizeof("xri://")-1;
 		if((fsc+1)>=lsc)
 		    throw bad_input(OPKELE_CP_ "not a character of importance in identity");
 		string id(identity,fsc,lsc-fsc+1);
 		idis.clear();
-		if(strchr(i_leaders,id[0])) {
+		if(strchr(data::_iname_leaders,id[0])) {
 		    /* TODO: further normalize xri identity? Like folding case
 		     * or whatever... */
 		    rv = id;
 		    set<string> cids;
@@ -494,15 +493,15 @@ namespace opkele {
 			}else if( !strcasecmp(a[0],"href") ) {
 			    const char *ns = a[1];
 			    for(;*ns && isspace(*ns);++ns);
 			    href.assign(ns);
-			    string::size_type lns=href.find_last_not_of(whitespace);
+			    string::size_type lns=href.find_last_not_of(data::_whitespace_chars);
 			    href.erase(lns+1);
 			}
 		    }
-		    for(string::size_type ns=rels.find_first_not_of(whitespace);
-			    ns!=string::npos; ns=rels.find_first_not_of(whitespace,ns)) {
-			string::size_type s = rels.find_first_of(whitespace,ns);
+		    for(string::size_type ns=rels.find_first_not_of(data::_whitespace_chars);
+			    ns!=string::npos; ns=rels.find_first_not_of(data::_whitespace_chars,ns)) {
+			string::size_type s = rels.find_first_of(data::_whitespace_chars,ns);
 			string rel;
 			if(s==string::npos) {
 			    rel.assign(rels,ns,string::npos);
 			    ns = string::npos;
diff --git a/lib/util.cc b/lib/util.cc
index bb8a2e8..29e6738 100644
--- a/lib/util.cc
+++ b/lib/util.cc
@@ -12,8 +12,9 @@
 #include <openssl/hmac.h>
 #include <curl/curl.h>
 #include <opkele/util.h>
 #include <opkele/exception.h>
+#include <opkele/data.h>
 #include <opkele/debug.h>
 
 #include <config.h>
 #ifdef HAVE_DEMANGLE
@@ -210,11 +211,10 @@ namespace opkele {
 	 * - remove empty and default ports
 	 * - 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 ns = uri.find_first_not_of(whitespace);
+	     string::size_type ns = uri.find_first_not_of(data::_whitespace_chars);
 	     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)
@@ -222,9 +222,9 @@ namespace opkele {
 	     transform(
 		     uri.begin()+ns, uri.begin()+colon+1,
 		     back_inserter(rv), ::tolower );
 	     bool s;
-	     string::size_type ul = uri.find_last_not_of(whitespace)+1;
+	     string::size_type ul = uri.find_last_not_of(data::_whitespace_chars)+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");
@@ -429,7 +429,42 @@ namespace opkele {
 		    md,&md_len);
 	    return encode_base64(md,md_len);
 	}
 
+	string normalize_identifier(const string& usi,bool strip_fragment) {
+	    if(usi.empty())
+		return usi;
+	    string rv;
+	    string::size_type fsc = usi.find_first_not_of(data::_whitespace_chars);
+	    if(fsc==string::npos)
+		return rv;
+	    string::size_type lsc = usi.find_last_not_of(data::_whitespace_chars);
+	    assert(lsc!=string::npos);
+	    if(!strncasecmp(usi.c_str()+fsc,"xri://",sizeof("xri://")-1))
+		fsc += sizeof("xri://")-1;
+	    if( (fsc+1) >= lsc )
+		return rv;
+	    rv.assign(usi,fsc,lsc-fsc+1);
+	    if(strchr(data::_iname_leaders,rv[0])) {
+		/* TODO: further normalize xri identity, fold case or
+		 * whatever... */
+	    }else{
+		if(rv.find("://")==string::npos)
+		    rv.insert(0,"http://");
+		if(strip_fragment) {
+		    string::size_type fp = rv.find('#');
+		    if(fp!=string::npos) {
+			string::size_type qp = rv.find('?');
+			if(qp==string::npos || qp<fp)
+			    rv.erase(fp);
+			else if(qp>fp)
+			    rv.erase(fp,qp-fp);
+		    }
+		}
+		rv = rfc_3986_normalize_uri(rv);
+	    }
+	    return rv;
+	}
+
     }
 
 }