Module oursh::program::posix

The shell language (often called sh) at the heart of the most popular shells.

While shells typically implement many extensions to the POSIX standard we’ll be implementing only the most basic set of functionality and offloading all extensions to the modern language.

Compatibility

Shell languages like bash or zsh are supersets of the POSIX sh language. This means two things:

• All sh programs are valid bash, zsh, etc programs
• Not all bash programs, for example, are valid sh programs.

This explains why some shell scripts will start with #!/bin/sh or #!/bin/bash, depending on what features of the language the script needs.

Examples

There are more than enough examples of sh scripts out there, but here is a collection of examples tested in this shell’s implementation of the POSIX standard. This section will not be a complete description of the syntax of the sh language, but will be updated with as many interesting cases as possible.

Running a command (like date) in a shell script is the simplest thing you can do.

date
date --iso-8601

./a.out

All variables start with a $ when referring to them, but when assigning you omit the $. It’s also worth mentioning that the lack of whitespace around the = in assignment is required. It’s often conventional to write your variable names in all caps, but this is not a limitation of the language.

NAME="Nathan Lilienthal"
i=0

echo $NAME
echo $i

Variables are loaded from the environment (often simply called ENV) as well. For more information on the enviroment read section 8.1.

echo $PATH
echo $TERM

In addition to variables beginning with a $ being expanded to the value they were set to, other syntax can perform expansion. See section 3§2.6 for a complete description of word expansion.

echo ${1}
echo ${1:-default}
echo ${1:=default}
echo ${1:?}
echo ${1:+new}
echo ${#1}
echo ${1%.*}
echo ${1%%.*}
echo ${1#prefix_}

In addition to running a program at the top level, programs can be run in a subshell with mechanisms to capture the output. This is called command substitution.

files=`ls`
files=$(ls)  # Same as above.

Conditionals in the wild are often written with the non-POSIX [[ syntax, traditional conditional checks use either test or [.

if test -z "$1"; then
    exit 1
fi

if [ "$1" -eq "foo" ]; then
    echo "bar"
fi

Specification

The syntax and semantics of this module are strictly defined by the POSIX (IEEE Std 1003.1) standard, in section 3§2 of “The Open Group Base Specifications” [1].

Re-exports

pub use self::ast::Program;

pub use self::ast::Command;

pub use self::builtin::Builtin;

Modules

ast

Abstract Syntax Tree for the POSIX language.

builtin

Commands that are run from the shell directly, without forking another process.

lex

Custom LALRPOP lexer for tokenizing the input stream.

parse

LALRPOP generated parser module.