Computing system and method for automated program error repair

1. A computing system for an automated program error repair, the system comprising:

a storage storing a program code written in a specific programming language;

a preprocessing processor configured to acquire the program code from the storage and preprocess the program code by tokenizing the program code with tokens, converting the tokens into vectors, and adding location information for the tokens; and

an automated error repair processor configured to:

receive the tokenized program code as an input from the preprocessing processor,

perform a deep learning on the tokenized program code,

detect an error in a compilation or execution process of the program code,

determine a location of an error occurrence based on the location information provided by the preprocessing processor,

automatically correct the detected error in the program code, based on a result of the deep learning,

output an error-corrected program code, to the storage, in which all error codes are changed to restored codes, and

display the restored codes for errors on a display along with error locations.

2. The computing system of claim 1 , wherein the preprocessing processor includes:

a token processor configured to acquire the program code from the storage, to decompose the program code into lexemes as the tokens, and to add lexeme types as additional information for the tokens;

a vector converter configured to convert the tokens generated by the token processor into vectors in a form of numbers understandable to a computer; and

a location information encoder configured to identify a location of each lexeme in the program code and to add the identified locations as the location information for the tokens.

3. The computing system of claim 2 , wherein the lexeme types comprise a keyword, an identifier, or an operator.

4. The computing system of claim 2 , wherein the location information encoder is further configured to identify a location of each lexeme in the program code, and add identified locations as location information for the tokens.

5. The computing system of claim 1 , wherein the automated error repair processor is further configured to:

understand a sequence of codes;

detect errors in the sequence of the codes;

memorize the detected errors; and

memorize recovered codes for the detected errors through the deep learning.

6. A method for an automated program error repair implemented by a computing system, the method comprising:

acquiring a program code written in a specific programming language from a storage of the computing system;

preprocessing the program code, at a preprocessing processor of the computing system, wherein the preprocessing includes tokenizing the program code with tokens, converting the tokens into vectors, and adding location information for the tokens;

performing a deep learning on the tokenized program code;

detecting, at an automated error repair processor of the computing system, an error in a compilation or execution process of the program code;

determining, at the automated error repair processor, a location of an error occurrence based on the location information provided by the preprocessing processor;

automatically correcting, at the automated error repair processor, the detected error in the program code, based on a result of the deep learning;

outputting, at the automated error repair processor, the error-corrected program code, to the storage, in which all error codes are changed to restored codes; and

displaying the restored codes for errors on a display along with error locations.

7. The method of claim 6 , wherein the preprocessing includes:

decomposing the program code into lexemes as the tokens;

adding lexeme types as additional information for the tokens;

converting the tokens into vectors in a form of numbers understandable to a computer;

identifying a location of each lexeme in the program code; and

adding the identified locations as the location information for the tokens.