Merge pull request #56 from Omercanyt/dev

Added minimum and maximum generation length limits

Markopy/src/CLI/markopy_cli.py

@@ -24,6 +24,8 @@
 parser.add_argument("-d", "--dataset",  help="Dataset file to read input from for training. Will be ignored for generation mode.")
 parser.add_argument("-s", "--seperator",help="Seperator character to use with training data.(character between occurrence and value)")
 parser.add_argument("-w", "--wordlist", help="Wordlist file path to export generation results to. Will be ignored for training mode")
+parser.add_argument("--min", default=6, help="Minimum length that is allowed during generation")
+parser.add_argument("--max", default=12,help="Maximum length that is allowed during generation")
 parser.add_argument("-n", "--count",    help="Number of lines to generate. Ignored in training mode.")
 parser.add_argument("-v", "--verbosity",action="count", help="Output verbosity.")
 args = parser.parse_args() 
@@ -89,7 +91,7 @@ def cli_generate(model):
     if(os.path.isfile(args.wordlist)):
         logging.pprint(f"{args.wordlist} exists and will be overwritten.", 1)
 
-    model.Generate(int(args.count), args.wordlist)
+    model.Generate(int(args.count), args.wordlist, args.min, args.max)
 
 
 

MarkovModel/src/model.cpp

@@ -93,16 +93,30 @@ bool Markov::Model<NodeStorageType>::Export(const char* filename) {
 }
 
 template <typename NodeStorageType>
-NodeStorageType* Markov::Model<NodeStorageType>::RandomWalk() {
+NodeStorageType* Markov::Model<NodeStorageType>::RandomWalk(int minSetting, int maxSetting) {  
 	Markov::Node<NodeStorageType>* n = this->starterNode;
 	int len = 0;
 	NodeStorageType *ret = new NodeStorageType[64];
-
+	Markov::Node<NodeStorageType> *temp_node;
 	while (n != NULL) {
-		n = n->RandomNext();
+		//n = n->RandomNext();
+		temp_node = n->RandomNext();
 		//dirty cutoff, needs better solution
-		if (len == 60) break;
-		if (n == NULL) break;
+		if (len == 60) 
+			break;
+		if (len > maxSetting) {
+			//std::cout<<"MAX ->"<< "node*: " << temp_node << ", len: " << len << "\n";
+			break;
+		}
+
+		if ((temp_node == NULL) && (len < minSetting)) {
+			//std::cout << "node*: " << temp_node << ", len: " << len << "\n";
+			continue;
+		}	
+
+		if (temp_node == NULL)
+			break;
+		n = temp_node;
 
 		//std::cout << n->NodeValue();
 		ret[len++] = n->NodeValue();

MarkovModel/src/model.h

@@ -34,7 +34,7 @@ namespace Markov {
 		* Start from the starter node, invoke RandomNext on current node until terminator node is reached.
 		* @return Null terminated string that was generated.
 		*/
-		NodeStorageType* RandomWalk();
+		NodeStorageType* RandomWalk(int minSetting, int maxSetting);
 
 		/** @brief Adjust the model with a single string. 
 		* Start from the starter node, and for each character, AdjustEdge the edge EdgeWeight from current node to the next, until NULL character is reached.

MarkovPasswords/src/markovPasswords.cpp

@@ -70,15 +70,15 @@ std::ofstream* MarkovPasswords::Save(const char* filename) {
 }
 
 
-void MarkovPasswords::Generate(unsigned long int n, const char* wordlistFileName)  {
+void MarkovPasswords::Generate(unsigned long int n, const char* wordlistFileName, int minLen, int maxLen)  {
 	char* res;
 	char print[100];
 	std::ofstream wordlist;	
 
 
 	wordlist.open(wordlistFileName);
 	for (int i = 0; i < n; i++) {
-		res = this->RandomWalk();
+		res = this->RandomWalk(minLen, maxLen); 
 #ifdef _WIN32
 		strcpy_s(print, 100, (char*)res);
 #else

MarkovPasswords/src/markovPasswords.h

@@ -46,7 +46,7 @@ class MarkovPasswords : public Markov::Model<char>{
 	* @param n - Number of passwords to generate.
 	* @return std::ofstream* of the output file.
 	*/
-	void Generate(unsigned long int n, const char* wordlistFileName);
+	void Generate(unsigned long int n, const char* wordlistFileName, int minLen=6, int maxLen=12);
 
 private:
 	std::ifstream* datasetFile;