Added ErrorTokenError, now properly raised on buildpattern() with all…

…owunknown=false, closes issue #25

include/common.h

@@ -46,4 +46,11 @@ class KeyError: public std::exception {
     return "Colibri KeyError";
   }
 };
+
+class UnknownTokenError: public std::exception {
+  virtual const char* what() const throw()
+  {
+    return "The input contained an unknown token";
+  }
+};
 #endif

src/classencoder.cpp

@@ -411,7 +411,7 @@ int ClassEncoder::encodestring(const string & line, unsigned char * outputbuffer
                         added[cls] = word;
           	    	} else if (!allowunknown) {	
                         //cerr << "ERROR: Unknown word '" << word << "', does not occur in model. You may want to pass either option -U or option -e to colibri-classencode to deal with unknown words." << endl;
-  	        			return 0;         
+                        throw UnknownTokenError();
 	  	        	} else {
 	  	        		//cerr << "WARNING: Unknown word '" << word << "', does not occur in model. Replacing with placeholder" << endl;
 	  	        		cls = unknownclass;	

src/test.cpp

@@ -496,13 +496,13 @@ int main( int argc, char *argv[] ) {
         bool caught = false;
         try {
             Pattern unkgram = encoder.buildpattern(unkstring);
-        } catch (const InternalError &e) {
+        } catch (const UnknownTokenError &e) {
             caught = true;
         }
         test(caught,true);
 
         Pattern unkgram2 = encoder.buildpattern(unkstring, true);
-        cerr << "Encode as unknown when allowunknown=true"; test(unkgram2.decode(classdecoder),"To {?} {?} or not");
+        cerr << "Encode as unknown when allowunknown=true"; test(unkgram2.decode(classdecoder),"to {?} {?} or not");
 
         cerr << endl << "************************** Skipgram tests  ***************************************" << endl << endl;