前言

来源于设计计算器的小组作业（Github链接：https://github.com/Skywuuuu/LexAndYACC），运用了Yacc和Lex来设计一个简单的计算器。文章内容为英文，但通俗易懂（用翻译软件翻译一下应该没问题，暂时不校对中文版了（偷懒））。感谢小伙伴们的帮助，以下排名不分先后：️️dzj，lxz，lyhan，skywuuu. 文章有点长，Introduction可跳过，Preparation部分讲的是如何在windows和Ubuntu下安装Flex和Bison，之后是展示我们设计的功能（Demonstration fo functions）和代码解释（Code Explanation），遇到的一些困难和我们编译器的不足也可以跳过，最后一个部分是介绍ANTLR的。以下，enjoy：

Introduction

To deepen our understandingofthe working procedures of the compiler and learn how to design it practically,we construct acalculatoras a project.We will discuss how to develop this simple compiler using two basic tools which are Lex (Lexical Analyzer Generator) and Yacc (Yet Another Compiler-Compiler) in the Ubuntu(Different operating systems will support different versions of software). Specifically, the details about the implementation of the calculator and how to use it correctlywill be elaboratedstep by step as a manual.In the preparation stage, userswould knowhow to download and install the compulsory packages inwindowsand Ubuntuandhow to compile the source filesand runthem.Three output formats supported by our calculator andthe basicand advancedoperationswill beintroduced in the demonstration of calculator functions.In the code explanation, detailed information about sixdifferent source files and difficulties we met will be discussed. Then, somelimitationsofour calculator will be mentioned.Lastly, we will introduce a more advanced generator, ANTLR, and discuss why it is better.

Preparation

To construct a calculator, flex and GNU bison are necessary and powerful tools to use.

Download and Installation

Windows

Flex

If the operating system of your computer is Windows, you should download and install flex. The steps are as below:

Clickthis linkand you will see this page.

Click Setup (withthered underline in the picture) and thenit will automatically jump to another page and the download will begin in five seconds. After that, you will get an executable file (file ends with .exe). You can find it under the default download directory (C:\Users\Your_name\Downloads).

Double clicks and the installation begins. You will see the window below.

Press Next button to continue until you see the window below. Just use the default path and click Next. Copy the address path below for later use is better.

Just choose Full installation, which means all of them are ticked. And click next.

Click next until you complete the flex setup and clickFinish to exit the installation.

Finally, you need to set the environmental variable.
1. Right click this PC and choose the Properties.
2. Choose Advanced system settings.
3. Click Environment variables.
4. Under the System variables, click New…
5. Give the variable a name and paste the path, for example, my path is D:\GnuWin32\bin. Do remember to include directory bin.
6. Finally, click OK.
Test whether flex is successfully installed into your computer.
1. Click windows.
2. Type cmd, the abbreviation of command,and open the command window.
3. Type flex -V to check the version of flex. If it shows your version, that means your installation succeeds.

Bison

In this part, we will introduce you the way to download and install bison.

Clickthis linkand you will see this page.

Click Setup (with red underline) and thenit will automatically jump to another page and the download will begin. After that, you will get an executable file (file ends with .exe). You can find it under the directory C:\Users\Your_name\Downloads.

Step 3 to step 8 are similar to flex, so we just skip the introduction.

Ubuntu

Under Ubuntu, the download and installation are quite simple. Type the command below:

If ubuntu asks you to choose yes or no, just type Y and enter to continue.

How to compile all of the files

Compile lex file
- flex LA.l
Compile yacc file
- yacc -dtv SA.y
Compile C file by g++ to get the object file. (The reason for using g++ is that the source files apply some c++ characteristics.)
- g++ -c c_file.c
- g++ -c lex.yy.c
- g++ -c y.tab.c
Compile all object files to generate an executable file
- g++ -o a.out y.tab.o lex.yy.o c_file.o
For convenience, we write a shell script including all of the commandsabove to compile the file and generate an executable file.The shell script file is demonstrated below:

For different output formats, we just need to run the corresponding shell script to compile the source files. (Interpreter: makefile_i.sh, Compiler: makefile_c.sh, Graph: makefile_g.sh)

Demonstration of functions

Three types of outputs

Before demonstrating the functions of this calculator, we introduceyou3 different versions of outputs, which are interpreterversion, compilerversion, and syntax treeversion.

Interpreterversion

Fortheinterpreter, actually, all of the shown output in the function demonstration part is in this type. The interpreter will execute the input code line by line, therefore, we can see the output of print function on the terminal.

Compilerversion

The compilerversioncan output the corresponding intermediate code of the input code.

Syntax tree / Graphversion

The last type of output isthesyntax tree or graph. It can demonstrate the syntax tree of the input code.

Print

The calculator can output the value of a variable through “print”.

Input:

Output:

Noticethat if we print a variable having not been assigned value, it will output 0.

Input:

Output:

Basic arithmetic operations

Addition

Input:

Output:

Subtraction

Input:

Output:

Multiplication

Input:

Output:

Division

Input:

Output:

Other operations

Function fac is the abbreviation of factorial, which means users can use this function to calculate the mathematical expression like 5! and get the answer. However, ifanegative number or floating number is given, an error will occur.

Input:

Output:

Function B2D is the abbreviation of transferring binary number to decimal number, which means users can use this function to transfer, for example, 10000B to 16. Remember to add capital B at the end ofthebinary string.

Input:

Output:

Function sin means sin operation is supported in our compiler. We keep four decimal places.

Input:

Output:

Input:

Output:

Input:

Output:

Input:

Output:

Function pow is the abbreviation of calculating the power of a number. You can directly use ‘^’ symbol to get the power of a number.

Input:

Output:

Input:

Output:

Logic Judgement

if-else

This calculator accepts if-else statement. It can support condition statements including >, <, !=, ==, >=, <=. The if-else statement supports 2 types of format. The first type has braces enclosing the statement. For example:

Input :

Output:

It also acceptsstatementswithout braces, but only 1 statement can be writteninside the if and else respectively.

Input:

Output:

Iteration

while

Input:

Output:

Code Explanation

In this section,we will explainsix source files, which are LA.l, SA.y, calc.h, interpreter.c, compiler.c, graph.c. Notice that when we add a new operator, not only the lex and yacc file should be modified, the interpreter.c, compiler.c, and graph.c should also be changed. And due to the limitation of the articlelength,we decide to explain the code structure based on interpreter.c, andthe way to addanew operator based oncompiler.c andgraph.c, because the logic of these three files issimilar.

LA.l

The basic technology of Lex is the FSA (finite-state automation), which consists of one beginning state and one or many ending states. The Lex implements the FSA through the regular expressions, which will be simulated to execute the similar procedures in FSA to match the input contents and then return the tokens or execute other commands set previously. The LA.l file in our project consists ofthree parts:definitions,rules, andsubroutineswhich are separated by %%.

Definitions

In the partof definition, the header files of C or C++should be included.Contents between%{ and }%will be written into the .C file entirely withoutanychange after the compilation steps.Therefore,we can use some build-in functions or variables conveniently.

Rules

Variables and functions

In the rules part, there are some default variables and functions in lex file we need to useas follows.

"yytext"is the pointer to the matched string.
“yylval” isthe semantic value associated with the token. It will be passed to the yacc file as the parameters to execute the next operation.Thus,when further operation needs the content of the matched strings, yylval will be needed.Specifically, in our file, the yylval.s represents the stringdata typeand the yylval.iValue stands for floating numbers because floating numbersare more suitable for most of the operations (For the integer, we just store them as xxx.000000).
yyerroris used to output the error information whenthe input doesn’t match the rules.
atof() andatoi() are used to convert the input values to the float, int format.

Regular Expressions and Actions

Decimal Number

Since we need the token and the detailed value of them for someoperations, we keep the value in the yylval and pass them to the yacc file as semantic values. The two ‘?’ in [0-9]+(.[0-9]+)?([eE][0-9]+)? improve the robustness of the input, which can handle the cases in both integer and float formats since ‘?’ means those two items can exist or not.

Relation Operators

Symbols****

We only need the operators’ name to pass to yacc file so that we can use themtocombine with different expressions. So we just return their names as tokens using ‘*’.

Binary Number

We first use the malloc() to allocate a memory with string type for the yylval.s, and we use the strcpy() that is declared in the to copy the string content from yytext to yylval.s. So the whole bit string can be passed to yacc file as semantic value with token ‘BIN’ .

Mathematical Functions

We just return the tokens and the yacc file will execute the corresponding functions we write accordingtothe token it receives. We just execute these three commands like basic operands, ‘+’, ‘-‘, ‘*’, ‘/’, and the onlydifferentthing is that we write the details of these three functions by ourselves.

Keywords

Identifiers

The identifiers startwith ‘’ or letters and just can be constructed by letter, ‘’ or numbers. And it will pass the token and semantic value to SA.y file.

Ignore Whitespaces

Error Reporting Function

When tokens don’t satisfy the rules, we call the yyerror function to output the error information.

Subroutines

Theyywrap( )functionwill be executed in this part. Itis used to determine whether the program has read all the input streams. If it’s called at the end of the last input command, it will return 1 so that system can receive and stop the program, otherwise it will return 0.

SA.y

Yacc is a useful program used toproduce the source code of the syntax analyzer by the language based on the gramma we rule. The grammar of the Yacc file is similar to the lex one, whichisdivided into three parts:definitions,rules**,andsubroutines**.