Reactionary Software by fschmidt

Reactionary Bash Tutorial


Contents


Introduction

contents

I really don't want to write this tutorial, but all the existing Bash tutorials are so horrible that I have no choice. I looked at books, websites, and YouTube - all horrible. They don't start with the basics. They include all kinds of useless crap. And they don't explain core concepts. So I have no choice but to write this for my Learn Reactionary Programming Bash lesson.

Bash is a shell, one of many, but the one I prefer. I will focus on Mac and Windows. I don't have Linux, and I hate Linux, so I won't discuss it. Most of Bash is the same on Mac and Windows, but where they differ, I will discuss both.

Running Bash

contents

How you access Bash depends on your operating system. If you are on a Mac then you access Bash through the Mac Terminal which is found in "Applications > Utilities > Terminal.app". Be sure to set the default shell to Bash. If you are on Windows then install MSYS2. The default terminal isn't so good, so I suggest using the Windows Terminal.

Getting Started

contents

When I start Bash on my Mac I see:

Last login: Thu Jan 4 23:25:35 on ttys004 The default interactive shell is now zsh. To update your account to use zsh, please run `chsh -s /bin/zsh`. For more details, please visit https://support.apple.com/kb/HT208050. ~ $

On Windows - MSYS2 I just see:

~ $

The line with the $ is the command prompt. The cursor is at the end of it, and if I type, my text will go there. You may have different text before the $ which is okay, but the line should end with $. If it doesn't, something is wrong.

Now type "qqq". When I say type "whatever", you should type return/enter at the end. Only when you type return/enter will Bash process what you typed. Now you should see:

~ $ qqq -bash: qqq: command not found ~ $

Bash doesn't know what "qqq" means and says so. Now try the following... Note that you type what is after the $ and Bash should respond as shown.

~ $ echo hi hi ~ $ echo how are you how are you ~ $ echo bye bye

The echo command just echoes what comes after. Now press the up-arrow on your keyboard. This should put the previous command where your cursor is. Up-arrow again brings the command before that. Try down-arrow and left-arrow and right-arrow. You can use this to navigate through your command history. The delete key also works for editing lines. And of course you can type. When you press return/enter then Bash will get your edited command and process it.

When you enter echo how are you, echo is the command. This command has 3 arguments: how, are, and you. Commands and arguments are separated with spaces. It doesn't matter how many spaces, so:

~ $ echo how are you how are you

echo just returns the arguments separated by one space.

~ $ echo one; echo two one two

You can put multiple commands on one line separated by a ;.

The "man" Command

contents

Enter:

~ $ man echo

You should get something like:

ECHO(1) BSD General Commands Manual ECHO(1) NAME echo -- write arguments to the standard output SYNOPSIS echo [-n] [string ...] DESCRIPTION The echo utility writes any specified operands, separated by single blank (` ') characters and followed by a newline (`\n') character, to the stan- dard output. The following option is available: -n Do not print the trailing newline character. This may also be achieved by appending `\c' to the end of the string, as is done by iBCS2 compatible systems. Note that this option as well as the effect of `\c' are implementation-defined in IEEE Std 1003.1-2001 (``POSIX.1'') as amended by Cor. 1-2002. Applications aiming for maximum portability are strongly encouraged to use printf(1) to suppress the newline character. :

But if you are on Windows, you may not have man installed. In that case, do:

~ $ pacman -S man-db

to install man as described here and then try man echo again.

The man command shows documentation of commands. Unfortunately it has a silly user interface based on memorizing keys, so I will just tell you the few keys you need. Down-arrow and up-arrow move down and up by one line. The space key moves down by one page. And most importantly, typing "q" quits and takes you back to Bash. You just have to memorize this.

Now try entering man man. You don't need all the stuff shown, but you can see what a complicated man page looks like. You can use man to get the documentation of other commands as I discuss them.

Directories

contents

You should be familiar with Mac Finder or Windows File Explorer, and you should know from this that directories (also called "folders") are organized into a tree.

On Mac:

~ $ pwd /Users/fschmidt ~ $ open . ~ $

On Windows:

~ $ pwd /home/fschmidt ~ $ explorer . ~ $

When using Bash, you are always in some directory, called the current directory or the working directory. pwd shows you the full path to this directory. Do man pwd for details. open . should open the Mac Finder for the current directory, and explorer . should open the Windows File Explorer for the current directory.

Continuing on my Mac:

~ $ mkdir learn ~ $ cd learn ~/learn $ pwd /Users/fschmidt/learn

mkdir makes a directory in the current directory. You should be able to see the created directory in Mac Finder or Windows File Explorer. cd stands for "change directory". This changes the current directory. cd is a built-in command (built into Bash), and man isn't useful with built-in commands, so instead of man cd, try help cd. Continuing...

~/learn $ pwd /Users/fschmidt/learn ~/learn $ ls ~/learn $ touch file1 ~/learn $ ls file1 ~/learn $ touch file2 ~/learn $ touch file3 ~/learn $ ls file1 file2 file3 ~/learn $ mkdir dir1 ~/learn $ ls dir1 file1 file2 file3 ~/learn $ ls -F dir1/ file1 file2 file3 ~/learn $ ls -a . .. dir1 file1 file2 file3 ~/learn $ ls -a -F ./ ../ dir1/ file1 file2 file3 ~/learn $ ls -aF ./ ../ dir1/ file1 file2 file3

ls lists files and touch creates an empty file. Arguments that start with "-" are options. Do man ls to see what the options I used do. -F appends a "/" to directories, and -a shows files starting with "." which are usually hidden. Options can be combined.

~/learn $ ls file1 file1 ~/learn $ ls qqq ls: qqq: No such file or directory ~/learn $ ls file1 qqq file2 ls: qqq: No such file or directory file1 file2 ~/learn $ ls dir1 ~/learn $ touch dir1/d1file ~/learn $ ls dir1 d1file ~/learn $ ls -d dir1 dir1 ~/learn $ ls file1 file2 dir1 file1 file2 dir1: d1file ~/learn $ ls -d file1 file2 dir1 dir1 file1 file2 ~/learn $ ls -dF file1 file2 dir1 dir1/ file1 file2

Without file arguments, ls lists files in the current directory. With file arguments, it lists those files if they exist. If the file is a directory, it will list what is in the directory unless the -d option is used.

~/learn $ ls dir1 file1 file2 file3 ~/learn $ ls . dir1 file1 file2 file3 ~/learn $ ls -d . . ~/learn $ ls -dF . ./ ~/learn $ ls ./file1 ./file1 ~/learn $ ls dir1 d1file ~/learn $ ls ./dir1 d1file ~/learn $ pwd /Users/fschmidt/learn ~/learn $ cd . ~/learn $ pwd /Users/fschmidt/learn

. is the current directory.

~/learn $ pwd /Users/fschmidt/learn ~/learn $ cd dir1 ~/learn/dir1 $ pwd /Users/fschmidt/learn/dir1 ~/learn/dir1 $ ls . d1file ~/learn/dir1 $ ls .. dir1 file1 file2 file3 ~/learn/dir1 $ cd .. ~/learn $ pwd /Users/fschmidt/learn ~/learn $ cd dir1 ~/learn/dir1 $ pwd /Users/fschmidt/learn/dir1 ~/learn/dir1 $ cd ../.. ~ $ pwd /Users/fschmidt ~ $ cd learn ~/learn $ pwd /Users/fschmidt/learn

.. is the parent directory.

~/learn $ echo * dir1 file1 file2 file3 ~/learn $ echo d* dir1 ~/learn $ echo f* file1 file2 file3 ~/learn $ echo *1 dir1 file1 ~/learn $ echo dir1/* dir1/d1file ~/learn $ echo */* dir1/d1file ~/learn $ echo qqq* qqq*

* does wildcard matching of files. It is important to understand that Bash does the wildcard matching and then passes the resulting arguments to the command. echo never sees the "*" unless there is no match.

~/learn $ ls * file1 file2 file3 dir1: d1file ~/learn $ ls -dF * dir1/ file1 file2 file3 ~/learn $ ls -dF d* dir1/ ~/learn $ ls -dF f* file1 file2 file3 ~/learn $ ls -dF *1 dir1/ file1 ~/learn $ ls dir1/* dir1/d1file ~/learn $ ls */* dir1/d1file ~/learn $ ls -dF qqq* ls: qqq*: No such file or directory

Should be self-explanatory.

~/learn $ pwd /Users/fschmidt/learn ~/learn $ cd ~ ~ $ pwd /Users/fschmidt ~ $ cd learn/dir1 ~/learn/dir1 $ pwd /Users/fschmidt/learn/dir1 ~/learn/dir1 $ cd ~ $ pwd /Users/fschmidt ~ $ cd ~/learn ~/learn $ pwd /Users/fschmidt/learn ~/learn $ echo ~ /Users/fschmidt ~/learn $ echo . . ~/learn $ echo .. ..

~ means your home directory. cd without arguments is the same as cd ~. ~ is expanded into your home directory by Bash.

~/learn $ ls -ltF total 0 drwxr-xr-x 3 fschmidt staff 96 Jan 5 02:33 dir1/ -rw-r--r-- 1 fschmidt staff 0 Jan 5 02:21 file3 -rw-r--r-- 1 fschmidt staff 0 Jan 5 02:21 file2 -rw-r--r-- 1 fschmidt staff 0 Jan 5 02:21 file1

-l gives you this ugly techy format. You get the date that the file was last modified. Before the date is the file size. -t sorts by date descending.

Lastly I will describe autocompletion. I type echo d without enter/return but instead then press the tab key. It autocompletes to echo dir1/. I press tab again and it autocompletes to echo dir1/d1file. Pressing tab while entering a file or directory makes Bash try to autocomplete using matching file names. If I enter echo f and press tab, I get echo file. It doesn't know which to choose next. Another tab just beeps. And another tab shows me the options like this:

~/learn $ echo file file1 file2 file3 ~/learn $ echo file

In general, you can press tab anytime while entering a file name and see what happens. Autocompletion saves a lot of typing.

Working with Files

contents
~/learn $ ls -F dir1/ file1 file2 file3 ~/learn $ cp file1 copied ~/learn $ ls -F copied dir1/ file1 file2 file3 ~/learn $ mv copied moved ~/learn $ ls -F dir1/ file1 file2 file3 moved ~/learn $ rm moved ~/learn $ ls -F dir1/ file1 file2 file3

cp copies files or directories. mv moves files or directories. rm removes files or directories. See the man pages of these commands for details.

~/learn $ ls -F dir1/ file1 file2 file3 ~/learn $ mkdir dir2 ~/learn $ touch dir2/d2file ~/learn $ ls -F dir1/ dir2/ file1 file2 file3 ~/learn $ ls dir2 d2file ~/learn $ rm dir2 rm: dir2: is a directory ~/learn $ rm -d dir2 rm: dir2: Directory not empty ~/learn $ rm dir2/d2file ~/learn $ rm -d dir2 ~/learn $ ls -F dir1/ file1 file2 file3 ~/learn $ ls -F dir1/ file1 file2 file3 ~/learn $ mkdir dir2 ~/learn $ touch dir2/d2file ~/learn $ ls -F dir1/ dir2/ file1 file2 file3 ~/learn $ rm -r dir2 ~/learn $ ls -F dir1/ file1 file2 file3 ~/learn $ ls -F dir1/ file1 file2 file3 ~/learn $ cp dir1 dir2 cp: dir1 is a directory (not copied). ~/learn $ cp -r dir1 dir2 ~/learn $ ls -F dir1/ dir2/ file1 file2 file3 ~/learn $ ls dir2 d1file ~/learn $ cp f* dir2 ~/learn $ ls dir2 d1file file1 file2 file3 ~/learn $ rm -r dir2 ~/learn $ ls -F dir1/ file1 file2 file3 ~/learn $ ls -F dir1/ file1 file2 file3 ~/learn $ mkdir dir2 ~/learn $ cp -r dir1 dir2 ~/learn $ ls -F dir2 dir1/ ~/learn $ ls -F dir2/dir1 d1file ~/learn $ rm -r dir2 ~/learn $ ls -F dir1/ file1 file2 file3

I could explain all this, but I won't. You should learn to understand commands and their options using man and by playing with them. Don't continue until you completely understand the above.

Quoting

contents
~/learn $ echo a b a b ~/learn $ echo "a b" a b ~/learn $ echo 'a b' a b ~/learn $ echo "a b" c a b c

Bash treats text in quotes as one argument. So in echo a b, echo has two arguments: "a" and "b". In echo "a b", echo has one argument: "a b". In echo 'a b', echo has one argument: "a b". In echo "a b" c, echo has two arguments: "a b" and "c".

~/learn $ echo a\ \ \ b a b

Outside of quotes, \ is not treated as a separator, but rather is treated as a space character that is part of the argument.

Variables

contents
~/learn $ echo $X ~/learn $ X="some text" ~/learn $ echo $X some text ~/learn $ echo "X is: $X" X is: some text ~/learn $ echo 'X is: $X' X is: $X ~/learn $ X="$X and more" ~/learn $ echo $X some text and more

Here X is a variable. You get its value with $X. This also works inside double-quotes but not inside single-quotes.

There are special variables called environment variables that are used by Bash.

~/learn $ echo $PATH /usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin:/Users/fschmidt/Dropbox/bin:/Users/fschmidt/hg/luan/scripts:/usr/local/opt/postgresql@9.5/bin ~/learn $ which ls /bin/ls ~/learn $ cd /bin /bin $ pwd /bin /bin $ ls [ dd launchctl pwd test bash df link rm unlink cat echo ln rmdir wait4path chmod ed ls sh zsh cp expr mkdir sleep csh hostname mv stty dash kill pax sync date ksh ps tcsh /bin $ ls -F [* dd* launchctl* pwd* test* bash* df* link* rm* unlink* cat* echo* ln* rmdir* wait4path* chmod* ed* ls* sh* zsh* cp* expr* mkdir* sleep* csh* hostname* mv* stty* dash* kill* pax* sync* date* ksh* ps* tcsh* /bin $ cd ~/learn ~/learn $

PATH is an environment variable containing a list of directories separated by : that are searched for commands by Bash. The which command shows the full path to a command. ls -F appends a * to executable files.

~/learn $ subl file1 -bash: subl: command not found ~/learn $ "/Applications/Sublime Text.app/Contents/SharedSupport/bin/subl" file1 ~/learn $ PATH="$PATH:/Applications/Sublime Text.app/Contents/SharedSupport/bin" ~/learn $ echo $PATH /usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin:/Users/fschmidt/Dropbox/bin:/Users/fschmidt/hg/luan/scripts:/usr/local/opt/postgresql@9.5/bin:/Applications/Sublime Text.app/Contents/SharedSupport/bin ~/learn $ subl file1 ~/learn $

Here I edit the file file1 with Sublime Text, first by using the full path, and then by adding the directory to PATH so that Bash can find subl.

I have Microsoft Word on Windows. From the Windows Command Prompt (not Bash):

C:\Users\fschmidt>winword C:\Users\fschmidt>where winword C:\Program Files\Microsoft Office\root\Office16\WINWORD.EXE

winword runs Microsoft Word. The Command Prompt where command is like the Bash which command. So now on MSYS2:

~ $ winword bash: winword: command not found ~ $ echo $PATH /usr/local/bin:/usr/bin:/bin:/opt/bin:/c/Windows/System32:/c/Windows:/c/Windows/System32/Wbem:/c/Windows/System32/WindowsPowerShell/v1.0/:/usr/bin/site_perl:/usr/bin/vendor_perl:/usr/bin/core_perl:/c/Program Files/TortoiseHg:/c/Program Files/Java/jdk1.8.0_202/bin ~ $ PATH="$PATH:/c/Program Files/Microsoft Office/root/Office16" ~ $ echo $PATH /usr/local/bin:/usr/bin:/bin:/opt/bin:/c/Windows/System32:/c/Windows:/c/Windows/System32/Wbem:/c/Windows/System32/WindowsPowerShell/v1.0/:/usr/bin/site_perl:/usr/bin/vendor_perl:/usr/bin/core_perl:/c/Program Files/TortoiseHg:/c/Program Files/Java/jdk1.8.0_202/bin:/c/Program Files/Microsoft Office/root/Office16 ~ $ winword ~ $

Returning to the Mac, there is another way to run applications found in Finder's "Applications" simply as applications instead of as commands.

~/learn $ open -a 'Sublime Text' file1

Another useful environment variable is PS1 which controls the command prompt. I already have this set up, but if I didn't:

Franklins-MacBook-Pro:learn fschmidt$ echo $PS1 \h:\W \u\$ Franklins-MacBook-Pro:learn fschmidt$ PS1="\w $ " ~/learn $ echo $PS1 \w $ ~/learn $

Google "bash PS1" for more info.

.bash_profile

contents
~/learn $ cd ~ $ ls .bash_profile .bash_profile

If .bash_profile isn't found then do touch .bash_profile to create it. This file contains Bash commands that are run when Bash starts. If you already have this file, it is likely to contain comments that start with #. Comments are ignored like this:

~ $ # comment line, does nothing ~ $ echo whatever # end of line comment whatever ~ $

To edit .bash_profile on a Mac, you can do:

~ $ open -a 'Sublime Text' .bash_profile

To edit .bash_profile on Windows, you can do:

~ $ notepad .bash_profile

Now try adding this line to .bash_profile:

echo hello there

Now when you open a new Bash terminal, you should see "hello there". .bash_profile runs when Bash is started by opening a new Bash terminal.

I set PS1 and PATH in .bash_profile to have the command prompt I want, and access to the commands that I want. I suggest that you make the Sublime Text command subl available in PATH.

The "find" Command

contents
~/learn $ find . . ./file3 ./file2 ./file1 ./dir1 ./dir1/d1file ~/learn $ find . -name 'file*' ./file3 ./file2 ./file1 ~/learn $ find . -name '*file' ./dir1/d1file ~/learn $ find . -name 'd*' ./dir1 ./dir1/d1file ~/learn $ find . -name '*1' -or -name '*2' ./file2 ./file1 ./dir1

find recursively searches for files in a directory tree. Note that in this case the * wildcard matching is not being done by Bash, it is being done by find. find has many options for searching for files and acting on them, see man find.

Input and Output

contents
~/learn $ echo 'this is a test' >test.txt ~/learn $ ls -F dir1/ file1 file2 file3 test.txt ~/learn $ cat test.txt this is a test ~/learn $ echo 'this is another test' >test.txt ~/learn $ cat test.txt this is another test ~/learn $ echo 'another line' >>test.txt ~/learn $ cat test.txt this is another test another line ~/learn $ cat <test.txt this is another test another line ~/learn $ cat <<End >test.txt > I am typing this > and this > End ~/learn $ cat test.txt I am typing this and this ~/learn $ (echo one; echo two) >test.txt ~/learn $ cat test.txt one two

All programs have standard input, standard output, and standard error. Programs write normal output to standard output and error messages to standard error. By default, standard input comes from the terminal, and standard output and standard error go to the terminal, but this can be changed. >file sends standard output to file. >>file appends standard output to file. <file reads standard input from file. <<whatever reads standard input from the text that follows until a line with just whatever. Commands can be combined between ( and ). Be sure to man cat to understand how cat works.

~/learn $ ls >ls.txt ~/learn $ cat ls.txt dir1 file1 file2 file3 ls.txt test.txt ~/learn $ ls -d f* q* >ls.txt ls: q*: No such file or directory ~/learn $ cat ls.txt file1 file2 file3 ~/learn $ ls -d f* q* 2>ls.txt file1 file2 file3 ~/learn $ cat ls.txt ls: q*: No such file or directory ~/learn $ ls -d f* q* | tee ls.txt ls: q*: No such file or directory file1 file2 file3 ~/learn $ cat ls.txt file1 file2 file3 ~/learn $ ls -d f* q* 2>&1 | tee ls.txt ls: q*: No such file or directory file1 file2 file3 ~/learn $ cat ls.txt ls: q*: No such file or directory file1 file2 file3

2>file sends standard error to file. | sends standard output of the previous command to standard input of the following command. 2>&1 sends standard error to standard output. tee file reads standard input and then writes it to both standard output and to file.

~/learn $ find . -type f | wc -l 6

There are 6 files in learn. Use man to figure out how this works.

Control Keys

contents
~/learn $ sleep 3 ~/learn $ sleep 30 ^C ~/learn $

sleep 3 sleeps for 3 seconds, meaning it does nothing for 3 seconds. I waited 3 seconds for this command to finish. Then I ran sleep 30 which would sleep for 30 seconds, but I lost my patience and pressed control+c which interrupts the program and breaks out of it. You can try control+c if you ever get stuck waiting for a command to finish.

~/learn $ wc I am typing this and this now I will end my input with control+d 3 14 65 ~/learn $ wc this time I will use control+c to break out ^C ~/learn $

Control+d means end of input.

Command Substitution

contents
~/learn $ echo I am in $(pwd) I am in /Users/fschmidt/learn ~/learn $ echo this directory contains: $(ls) this directory contains: dir1 file1 file2 file3 ls.txt test.txt ~/learn $ echo this directory contains $(ls | wc -l) files this directory contains 6 files

cmd $(commands) will use the output of commands as argument text for cmd.

~/learn $ cat $(find . -type f) | wc -c 86

The files in learn contain a total of 86 bytes. Use man to figure out how this works.

Ampersand

contents
~/learn $ (sleep 5; echo done) & [1] 10080 ~/learn $ echo waiting waiting ~/learn $ done [1]+ Done ( sleep 5; echo done ) ~/learn $

Normally Bash waits for a command to complete before showing the command prompt and allowing input. But ending a command line with & tells bash not to wait, but instead to run the command in a separate process. Above in ~/learn $ echo waiting, I typed in echo waiting. But in ~/learn $ done, I did not type done. Instead this was produced by echo done after 5 seconds. [1] 10080 tells me that a process was started and [1]+ Done ( sleep 5; echo done ) tells me that the process finished.

This is useful where you do not want to wait for a command to finish. Consider this on Windows:

~ $ notepad

Here you will not get a command prompt again until you quit Notepad because Bash is waiting for this command to finish. So instead do: ~ $ notepad & [1] 2010 ~ $

Now Notepad will run and you can continue using Bash.

Shell Scripts

contents

Make a file called test.sh containing the following:

echo this is a shell script

Now from Bash:

~/learn $ cat test.sh echo this is a shell script ~/learn $ ./test.sh -bash: ./test.sh: Permission denied ~/learn $ ls -F test.sh test.sh ~/learn $ chmod +x test.sh ~/learn $ ls -F test.sh test.sh* ~/learn $ ./test.sh this is a shell script ~/learn $

chmod +x file makes file into an executable that can be run. Now I will edit test.sh

~/learn $ # edit test.sh ~/learn $ cat test.sh nonsense echo this is a shell script ~/learn $ ./test.sh ./test.sh: line 1: nonsense: command not found this is a shell script ~/learn $ # edit test.sh ~/learn $ cat test.sh set -e nonsense echo this is a shell script ~/learn $ ./test.sh ./test.sh: line 2: nonsense: command not found ~/learn $

By default, scripts continue running after an error. In longer scripts, we want the script to exit after an error. set -e does this, see help set.

~/learn $ X=some ~/learn $ echo $X some ~/learn $ echo $Xthing ~/learn $ echo ${X}thing something ~/learn $ # edit test.sh ~/learn $ cat test.sh echo "\$* = $*" echo "\$# = $#" echo "\$0 = $0" echo "\$1 = $1" echo "\$2 = $2" echo "\$3 = $3" echo "\$4 = $4" echo "\$14 = $14" echo "\${14} = ${14}" echo "\$@ = $@" ./count.sh "$*" ./count.sh "$@" ~/learn $ ./test.sh a b "c d" $* = a b c d $# = 3 $0 = ./test.sh $1 = a $2 = b $3 = c d $4 = $14 = a4 ${14} = $@ = a b c d 1 3 ~/learn $ cat count.sh echo $# ~/learn $

Bash scripts have special defined variables. The difference between $* and $@ is subtle, and you will usually just use $*. $* returns all arguments as one string while $@ returns the arguments separately, but this distinction rarely makes any difference.

Variables and Scripts

contents
~/learn $ X=value ~/learn $ echo $X value ~/learn $ # edit test.sh ~/learn $ cat test.sh echo "\$X = $X" ~/learn $ ./test.sh $X = ~/learn $ export X ~/learn $ ./test.sh $X = value

Variables are defined in the current shell. Shell scripts are run in their own shell. So by default, they don't see variables defined in the terminal/parent shell. export var makes var available in descendant processes, meaning available in shell scripts. It is a good idea to do export PATH in .bash_profile so that your PATH is available to your scripts.

~/learn $ X=terminal ~/learn $ echo $X terminal ~/learn $ # edit test.sh ~/learn $ cat test.sh X=script export X ~/learn $ ./test.sh ~/learn $ echo $X terminal ~/learn $ . test.sh ~/learn $ echo $X script

You can export a variable from parent to children but not from children to parent. . script includes the text in the file script in the current shell. In this case, it is not run in a separate shell. This is the only way to have a script set variables in your terminal shell.

~/learn $ pwd /Users/fschmidt/learn ~/learn $ # edit test.sh ~/learn $ cat test.sh cd ~ ~/learn $ ./test.sh ~/learn $ pwd /Users/fschmidt/learn ~/learn $ . test.sh ~ $ pwd /Users/fschmidt ~ $ cd learn ~/learn $

This illustrates the difference between ./script and . script.

Your Scripts

contents
~/learn $ echo $PATH /usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin:/Users/fschmidt/Dropbox/bin:/Users/fschmidt/hg/luan/scripts:/usr/local/opt/postgresql@9.5/bin:/Applications/Sublime Text.app/Contents/SharedSupport/bin ~/learn $ echo ~/Dropbox/bin /Users/fschmidt/Dropbox/bin ~/learn $ ls -F ~/Dropbox/bin/e /Users/fschmidt/Dropbox/bin/e* ~/learn $ cat ~/Dropbox/bin/e open -a 'Sublime Text' $* ~/learn $ e test.sh ~/learn $

When you write useful scripts, put them in a directory and add that directory to your PATH. I use ~/Dropbox/bin and I have a script named e in that directory for editing files. So e test.sh lets me edit test.sh from the command line.

Advanced Scripting

contents

Here is a more advanced script called undocx.sh that unpacks a Word DOCX file.

#!/bin/bash set -e if [ $# -ne 1 ]; then echo "usage: $0 filename" exit 1 fi FILE="$1" NEWDIR=$(basename $FILE .docx) mkdir $NEWDIR unzip $FILE -d $NEWDIR export XMLLINT_INDENT=$'\t' for file in $(find $NEWDIR -name "*.xml" -o -name "*.rels"); do mv "$file" temp.xml xmllint --format temp.xml >"$file" done rm temp.xml

Bash is a full programming language containing all the usual features. Some commands in my script are well explained by man, but some are not. In particular, the documentation for if and for are poor. In cases like this, I suggest asking ChatGPT like this: Please explain the Bash "if" statement. Please explain the Bash "for" statement.

ChatGPT knows Bash well. I trust ChatGPT to explain details but not to explain core concepts. You can also try Google, but ChatGPT is better than modern programmers.

Conclusion

contents

At least 90% of your usage of Bash will be simple commands that you enter in the terminal. Try to use Bash as much as possible instead of using the GUI so that you get practice using it. Unless you become system administrator, you won't use advanced scripting much. But with a solid understanding of the core basics, you should be able to figure out how to read or write advanced scripts when needed.