Merge pull request #5 from Arkorri/version-1.1

Version 1.1

MLdigits/src/DataHandler.cpp

@@ -74,7 +74,11 @@ bool dataHandler::read_feature_vector(std::string path){
 	if(file){
 		for(unsigned int i = 0; i < 4; i++){
 			if(fread(bytes, sizeof(bytes), 1, file)){
-				header[i] = convert_to_little_endian(bytes);
+				if(isIntel){
+					header[i] = convert_to_little_endian(bytes);
+				} else {
+					header[i] = (uint32_t)bytes;
+				}//if/else
 			}//if
 		}//for
 		std::cout << "Finished getting data file header" << std::endl;
@@ -87,7 +91,6 @@ bool dataHandler::read_feature_vector(std::string path){
 					image->append_to_vector(element[0]);
 				} else {
 					std::cerr << "Error reading from file in dataHandler.read_feature_vector()" << std::endl;
-					this->unload();
 					return false;
 				}//if/else
 			}//for
@@ -96,7 +99,6 @@ bool dataHandler::read_feature_vector(std::string path){
 		std::cout << "Successfully read file and stored data points. " << this->data_array->size() << std::endl;
 	} else {
 		std::cerr << "Unable to read file in dataHandler.read_feature_vector()" << std::endl;
-		this->unload();
 		return false;
 	}//if/else
 	return true;
@@ -109,7 +111,11 @@ uint32_t header[2];// |MAGIC|NUM IMAGES|
 	if(file){
 		for(unsigned int i = 0; i < 2; i++){
 			if(fread(bytes, sizeof(bytes), 1, file)){
-				header[i] = convert_to_little_endian(bytes);
+				if(isIntel){
+					header[i] = convert_to_little_endian(bytes);
+				} else {
+					header[i] = (uint32_t)bytes;
+				}//if/else
 			}//if
 		}//for
 		std::cout << "Finished getting label file header" << std::endl;
@@ -119,14 +125,12 @@ uint32_t header[2];// |MAGIC|NUM IMAGES|
 				this->data_array->at(i)->set_lable(element[0]);
 			} else {
 				std::cerr << "Error reading from file in dataHandler.read_feature_labels()" << std::endl;
-				this->unload();
 				return false;
 			}//if/else
 		}//for
 		std::cout << "Successfully read file and stored labels. " << std::endl;
 	} else {
 		std::cerr << "Unable to read file in dataHandler.read_feature_labels()" << std::endl;
-		this->unload();
 		return false;
 	}//if/else
 	return true;
@@ -220,14 +224,14 @@ bool dataHandler::setFeaturePath(void){
 	std::string extension = ".idx3-ubyte";
 	std::string path = "";
 	do{
-		path = getString(64);
+		path = getString(1,64);
 		if(path.length() > extension.length()){
 			if(0 == path.compare(path.length() - extension.length(), extension.length(), extension)){
 				this->featurePath = path;
 				return true;
 			} else {
 				std::cerr << "incorrect file type, please try again (must end in " << extension << ")" << std::endl;
-			}//if/e;se
+			}//if/else
 		}else {
 			std::cerr << "incorrect file type, please try again (must end in " << extension << ")" << std::endl;
 		}//if/else
@@ -244,7 +248,7 @@ bool dataHandler::setLabelPath(void){
 	std::string extension = ".idx1-ubyte";
 	std::string path = "";
 	do{
-		path = getString(64);
+		path = getString(1,64);
 		if(path.length() > extension.length()){
 			if(0 == path.compare(path.length() - extension.length(), extension.length(), extension)){
 				this->labelPath = path;
@@ -264,3 +268,7 @@ bool dataHandler::setLabelPath(std::string path){//overloaded for default
 	return true;
 }//setLabelPath
 
+void dataHandler::setIsIntel(bool x){
+	this->isIntel = x;
+}//setIsIntel
+

MLdigits/src/DataHandler.hpp

@@ -170,6 +170,12 @@ class dataHandler {
 	//					-returns true if successful, false if not
 	bool setLabelPath(std::string);
 
+	//setIsIntel
+	//Desciption:		sets weather the system is using intel or not
+	//Preconditions:	-none
+	//Postconditions:	-isIntel is set as true or false
+	void setIsIntel(bool);
+
 private:
 	std::vector<data *> *data_array;		//array holding all data from the file
 	std::vector<data *> *training_data;		//array holding a portion of data_array
@@ -181,6 +187,7 @@ class dataHandler {
 	std::string featurePath;				//directory path for the file holding class data
 	std::string labelPath;					//directory path for the file holding class names
 	bool loaded;							//keeps track of if the class is holding data
+	bool isIntel;							//keeps track of if the system is using an Intel processor
 
 	//used for data splits
 	//double represents a % of data to be used for each set of data

MLdigits/src/MLdigits.cpp

@@ -21,6 +21,8 @@ int main() {
 //	unsigned int best_k = 1;
 
 	do{
+		std::cout << "Is your system using an Intel processor?" << std::endl;
+		dh->setIsIntel(yn());
 		std::cout << "Do you want to use default data or use custom data?" << std::endl;
 		std::cout << "1. Default Data (recommended)" << std::endl;
 		std::cout << "2. Custom Data" << std::endl;

MLdigits/src/Utilities.cpp

@@ -7,26 +7,46 @@
 
 #include "Utilities.hpp"
 #include "climits"
+#include <cstring>
 
-std::string getString(unsigned int maxLength){
+//std::string getString(unsigned int maxLength){
+//	std::string str = "";
+//	do{
+//		std::cout << "Please enter a string less than " << std::to_string(maxLength) << " characters" << std::endl;
+//		std::getline(std::cin, str);
+//		if(str.length() <= maxLength){
+//			return str;
+//		} else {
+//			std::cout << "Invalid value, please try again" << std::endl;
+//		}//if/else
+//	}while(str.length() > maxLength);//do-while
+//	return "";
+//}//getString
+
+void clearCin(void){
+	std::cin.clear();
+	std::cin.ignore(INT_MAX, '\n');
+}//clearCin
+
+std::string getString(unsigned int min, unsigned int max) {
 	std::string str = "";
 	do{
-		std::cout << "Please enter a string less than " << std::to_string(maxLength) << " characters" << std::endl;
+		std::cout << "Please enter a string between " << std::to_string(min) <<
+				" and " << std::to_string(max) << std::endl;
 		std::getline(std::cin, str);
-		if(str.length() <= maxLength){
-			return str;
-		} else {
+		if (str.length() < min || str.length() > max) {
 			std::cout << "Invalid value, please try again" << std::endl;
+		} else {
+			const char *Whitespace = " \t\v\r\n";						//checks for, and removes leading/trailing whitespace
+			size_t start = str.find_first_not_of(Whitespace);			//..
+			size_t end = str.find_last_not_of(Whitespace);				//..
+			return start == end ?										//..
+					std::string() : str.substr(start, end - start + 1);	//..
 		}//if/else
-	}while(str.length() > maxLength);//do-while
+	}while(str.length() < min || str.length() > max);//do-while
 	return "";
 }//getString
 
-void clearCin(void){
-	std::cin.clear();
-	std::cin.ignore(INT_MAX, '\n');
-}//clearCin
-
 int getInt(unsigned int min, unsigned int max){
 	unsigned int num = 0;
 	do{
@@ -44,3 +64,32 @@ int getInt(unsigned int min, unsigned int max){
 	return -1;
 }//getInt
 
+char getChar(void) {
+	char c_str[32];
+	do{
+		std::cin.getline(c_str, INT_MAX);
+		if (strlen(c_str) > 1) {
+			std::cout << "Invalid value, please enter a single character" << std::endl;
+		} else {
+			return c_str[0];
+		}//if/else
+	}while(strlen(c_str) > 1);//do-while
+	return '0';
+}//getChar
+
+bool yn(void) {
+	char choice;
+	do{
+		std::cout << "(y/n)" << std::endl;
+		choice = getChar();
+		if (toupper(choice) == 'Y') {
+			return true;
+		} else if (toupper(choice) == 'N') {
+			return false;
+		} else {
+			std::cout << "Invalid input, please try again (y/n)" << std::endl;
+		}//if/elif/else
+	}while(toupper(choice) != 'Y' && toupper(choice) != 'N');//do-while
+	return false;//should never reach
+}//yn
+

MLdigits/src/Utilities.hpp

@@ -11,6 +11,11 @@
 #include <iostream>
 #include <string>
 
+	//function pre-fix template
+	//Desciption:
+	//Preconditions:
+	//Postconditions:
+
 	//error struct for try/catch blocks
 	struct error{
 		int errorCode;
@@ -25,7 +30,19 @@
 	//Preconditions:	-maxLength is positive
 	//Postconditions:	-a string is returned
 	//					-returns "" as default
-	std::string getString(unsigned int maxLength);
+	//std::string getString(unsigned int maxLength);
+
+	//getString
+	//Description:		gets a string from the user
+	//						that's length is between a min
+	//						and max value. Continues to ask
+	//						until value is valid
+	//Preconditions:	-max is positive
+	//					-min is positive
+	//Postconditions:	-a string is returned
+	//					-returns "" as default
+	//					-leading/trailing whitespace removed
+	std::string getString(unsigned int min, unsigned int max);
 
 	//getInt
 	//Description:		gets an integer from the user between
@@ -42,4 +59,20 @@
 	//Postconditions:	-cin is empty
 	void clearCin(void);
 
+	//getChar
+	//Desciption:		gets and returns a single character
+	//						input from the user
+	//Preconditions:	-none
+	//Postconditions:	-returns a char
+	//					-returns '0' as default
+	char getChar(void);
+
+	//function yn
+	//Desciption:		gets and returns a char of 'y' or 'n'
+	//						from the user
+	//Preconditions:	-none
+	//Postconditions:	-returns true if y
+	//					-false if n
+	bool yn(void);
+
 #endif /* UTILITIES_HPP_ */

README.md

@@ -13,9 +13,11 @@ Intel(R) Core(TM) i7-10710U CPU @ 1.10GHz, 1608 Mhz, 6 Core(s), 12 Logical Proce
 -If you are using eclipse, the zip file can be **imported** as an **existing project**
 
 ## How to use the provided main
-**1. On program start it will ask if you want to use custom data, or provided MNIST data**\
+**1. On program start it will ask if you are using an Intel processor, reply with a 'y' or 'n'**\
+\
+**2. It will then ask if you want to use custom data, or provided MNIST data**\
 -It is highly recommended to use the provided data\
--If you want to use custom data, when the prompts you for the path you must input the full path starting from the directory in which the .exe is called\
+-If you want to use custom data, when prompted for the file path you must input the full path starting from the directory in which the .exe is called\
 -Custom file(s) must also have the extension idx3-ubyte (for image files) or idx1-ubyte (for label files)\
 \
 **2. Input the desired K value when prompted**\
@@ -30,24 +32,26 @@ Intel(R) Core(TM) i7-10710U CPU @ 1.10GHz, 1608 Mhz, 6 Core(s), 12 Logical Proce
 \
 **1. Create a new DataHandler object**\
 \
-**2. Set the file paths using the DataHandler setLabelPath(string) and setFeaturePath(string) function**\
+**2. Use the DataHandler setIsIntel(bool) function to say whether or not an Intel processor is being used**\
+\
+**3. Set the file paths using the DataHandler setLabelPath(string) and setFeaturePath(string) function**\
 -Paths must be defined from the location in which the program was launched\
 -i.e. if launching from eclipse with the files in the project folder ("./filename")\
 \
-**3. Load the data with the DataHandler load() function**\
+**4. Load the data with the DataHandler load() function**\
 -This will return true if the load was successful, or false if it failed\
 -If the load failed, the class will automatically unload data\
 \
-**4. Create a new Knn object**\
+**5. Create a new Knn object**\
 -This can be done with the default constructor, or can be given a k value by passing an integer\
 -If the default constructor was called you must first call Knn->set_k(int) before step 5\
 \
-**5. Call the Knn .load(DataHandler) function by passing it a pointer to the previously created DataHandler object**\
+**6. Call the Knn .load(DataHandler) function by passing it a pointer to the previously created DataHandler object**\
 -This will return true if successfull, or false if the load failed\
 \
-**6. You can now call the Knn validate_performance() and test_performance() functions**\
+**7. You can now call the Knn validate_performance() and test_performance() functions**\
 -These will run the algorithm and print results to the console.\
 -After they finish they will return a double containing the overall performance as a percent\
 -These functions take a long time to complete\
 \
-**7. If you wish to input new data into the classes you must call the unload() function for each object first**
+**8. If you wish to input new data into the classes you must call the unload() function for each object first**