Description
Program
Your next project will tie together many of the Operating System/Unix concepts that we have studied
so far this quarter. For the project, you will eventually write a Unix program called the Betterthan
Average Shell, bash, which is a simple Unix command interpreter. To simplify the design of your shell,
this first half of the project only deals with the input parsing aspects of the project. This semester,
we’re going to use the standard Unix tools Flex and Bison to build the input parser for the shell.
Flex generates a lexical analyzer that breaks an input stream into tokens. Bison understands language
grammars. Once you’ve specified the grammar for your bash, Bison will do most of the parsing for
you. We’ll talk about both of these programs extensively in class.
Shell Grammar
To use Lex and Bison, you need to understand the grammar that they need to read. For your shell,
the command input must be in the following format:
cmd (arg)* (<infile)? (2>(>)?errfile)?)* (| cmd (arg)* (2>(>)?errfile)?)* (>(>)? outfile)?
Input Tokens
Your shell understands the following input tokens:
|
The PIPE symbol, forward stdout to next cmd
< stdin redirection >
stdout redirection (create)
>>
stdout redirection (append)
2>
stderr redirection (create)
2>>
stderr redirection (append)
WORD
Any sequence of characters not already defined as tokens (separated by white space). Alternately,
a WORD can be any sequence of characters between double quotes
Example
In the github repository for this assignment, under ”examples”, there’s an example lex/yacc/C program
called arith that understands a grammar very similar to what you’ll need for your shell project. We’ll
cover that example in great detail in class.
Headstart
In the github repository for this assignment, under ”headstart”, I’ve placed a skeleton bash “headstart”
version with a Makefile stub header files, and etc. You should start from that. It should compile and
”run” a little, but you’ll need to finish a lot of it before it will do much.
Requirements
Your program must work for any reasonable number of arguments and must give meaningful error
messages. Your shell should terminate neatly and quietly when it reaches the end of file. You must
use Flex for lexical analysis and Bison for parsing.
Turnin
You must use the class github archive to submit all of your source files, along with a Makefile that
generates a program called bash that can read input from standard input.
Example Input and Output
For example, if the input to your program looks like this:
echo Hello World
/bin/cat < /etc/motd /bin/cat > outfile
echo “this is a test” | cat
gcc -c -Wall file.c 2>> ERRS
cat < infile | tr “a-z” “A-Z” | cat 2>caterrs | wc -l > out
then the output to your program should look something like this:
========== line 1 =================
Command name: ’echo’
argv[0]: ’echo’
argv[1]: ’Hello’
argv[2]: ’World’
stdin: ’’
stdout: ’’
stderr: ’’
========== line 2 =================
Command name: ’/bin/cat’
argv[0]: ’/bin/cat’
stdin: ’/etc/motd’
stdout: ’’
stderr: ’’
========== line 3 =================
Command name: ’/bin/cat’
argv[0]: ’/bin/cat’
stdin: ’’
stdout: ’outfile’
stderr: ’’
========== line 4 =================
Command name: ’echo’
argv[0]: ’echo’
argv[1]: ’this is a test’
stdin: ’’
stdout: PIPE1
stderr: ’’
Command name: ’cat’
argv[0]: ’cat’
stdin: PIPE1
stdout: ’’
stderr: ’’
========== line 5 =================
Command name: ’gcc’
argv[0]: ’gcc’
argv[1]: ’-c’
argv[2]: ’-Wall’
argv[3]: ’file.c’
stdin: ’’
stdout: ’’
stderr: ’ERRS’ (append)
========== line 6 =================
Command name: ’cat’
argv[0]: ’cat’
stdin: ’infile’
stdout: PIPE1
stderr: ’’
Command name: ’tr’
argv[0]: ’tr’
argv[1]: ’a-z’
argv[2]: ’A-Z’
stdin: PIPE1
stdout: PIPE2
stderr: ’’
Command name: ’cat’
argv[0]: ’cat’
stdin: PIPE2
stdout: PIPE3
stderr: ’caterrs’
Command name: ’wc’
argv[0]: ’wc’
argv[1]: ’-l’
stdin: PIPE3
stdout: ’out’
stderr: ’’
