We use 4 DFA's but since DFA's are equivalent to NFA's it is basically the union of 4 DFA's converted to one DFA. This proves that our code works. This implementation uses standard switch/if statements to parse through the string, and simulate a DFA.
https://docs.oracle.com/javase/specs/jls/se17/html/jls-3.html#jls-DecimalFloatingPointLiteral
Type floating point literal or press q to exit (Interactive)
Ran and compiled using C++17.
Used g++ with g++ project1.cpp and run program.
Run project using any C++ compiler.
For this project, we decided that it was easier and safer to use CFG method for evaluating expressions. This is because we ran into lots of bugs while trying the PDA method. Initially, our CFG code was very slow as we converted original grammar to chomsky normal form and checked every splitted substring. The faster method doesn't use chomsky normal form and works on the original grammar.
Each of the main functions in project 2 correspond to rules in the CFG. The functions return true and the floating point values if they successfully splitted the string into at least one of the substitutions used for that rule in the CFG. To check for terminal a, we used project1 code and computed DFA to return possible floating point value.
Spaces: To handle spaces, spaces can be everywhere except inbetween digits. For digits, underscores are used instead. This is solved by trim function which trims only the left and right parts of string.
Grammar used is:
<EXPR> → <EXPR> + <TERM> | <TERM>
<TERM> → <TERM> * <FACTOR> | <FACTOR>
<FACTOR> → (<EXPR>) | a
where a is a valid floating point string.
This grammar is stripped straight from the book on context-free grammars on page 105 (Ch 2 Section 1).
Type floating point expression or press q to exit (Interactive)
Ran and compiled using C++17.
Used g++ with g++ project2.cpp and run program.
Run project using any C++ compiler.