Shell Programming—an Introduction Copyright Conditions: Open Publication License (see http://www.opencontent.org/openpub/) Nick Urbanik nicku@vtc.edu.hk Department of Information and Communications Technology OSSI — ver. 1.12 Shell Programming - p. 1/86 Aim After successfully working through this exercise, You will: I write simple shell scripts using for, if, while, case, getopts statements; I I Intro Aim Why Shell Scripting? Where to get more information The Shell is an Interpreter The Shebang Making Executable Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find write shell script functions, and be able to handle parameters; understand basic regular expressions, and be able to create your own regular expressions; understand how to execute and debug these scripts; understand some simple shell scripts written by others. I I OSSI — ver. 1.12 Shell Programming - p. 2/86 Why Shell Scripting? I I I I Basic startup, shutdown of Linux, Unix systems uses large number of shell scripts N understanding shell scripting important to understand and perhaps modify behaviour of system Very high level: powerful script can be very short Can build, test script incrementally Useful on the command line: “one liners” Intro Aim Why Shell Scripting? Where to get more information The Shell is an Interpreter The Shebang Making Executable Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 3/86 Where to get more information I I I the Libarary has two copies of the book, Learning the Bash Shell, second edition, by Cameron Newham & Bill Rosenblatt, O’Reilly, 1998. There is a free on-line book about shell programming at: http://tldp.org/LDP/abs/html/index.html and http://tldp.org/LDP/abs/abs-guide.pdf. It has hundreds of pages, and is packed with examples. The handy reference to shell programming is: $ pinfo bash or $ man bash all built-in commands, e.g., $ help let Intro Aim Why Shell Scripting? Where to get more information The Shell is an Interpreter The Shebang Making Executable Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find I IMPORTANT: bash provides simple on-line help for OSSI — ver. 1.12 Shell Programming - p. 4/86 The Shell is an Interpreter I I I Some languages are compiled: C, C++, Java,. . . Some languages are interpreted: Java bytecode, Shell Shell is an interpreter: kernel does not run shell program directly: N kernel runs the shell program /bin/sh with script file name as a parameter N the kernel cannot execute the shell script directly, as it can a binary executable file that results from compiling a C program Intro Aim Why Shell Scripting? Where to get more information The Shell is an Interpreter The Shebang Making Executable Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 5/86 The Shebang I I I I I You ask the Linux kernel to execute the shell script kernel reads first two characters of the executable file N If first 2 chars are “#!” then N kernel executes the name that follows, with the file name of the script as a parameter Example: a file called find.sh has this as the first line: #! /bin/sh then kernel executes this: /bin/sh find.sh What will happen in each case if an executable file begins with: N #! /bin/rm N #! /bin/ls Intro Aim Why Shell Scripting? Where to get more information The Shell is an Interpreter The Shebang Making Executable Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 6/86 Making the script executable To easily execute a script, it should: I be on the PATH I have execute permission. How to do each of these? I Red Hat Linux by default, includes the directory ∼/bin on the PATH, so create this directory, and put your scripts there: $ mkdir ∼/bin I If your script is called script, then this command will make it executable: $ chmod +x script Intro Aim Why Shell Scripting? Where to get more information The Shell is an Interpreter The Shebang Making Executable Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 7/86 Special Characters I Many characters are special to the shell, and have a particular meaning to the shell. Character ∼ ‘ # $ & * | Intro Quoting and Funny Chars Special Characters Special Chars—2 Special Chars—3 Quoting Quoting—2 Quoting—When to use it? Variables Command Substitution Meaning Home directory Command substitution. Better: $(...) Comment Variable expression Background Job File name matching wildcard Pipe § 15 §7 § 24 See slide Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find 2.10 on page 43 2.18 on page 51 2.9 on page 42 OSSI — ver. 1.12 Shell Programming - p. 8/86 Special Characters—continued: 2 Character ( ) [ ] { ; \ ’ " Meaning Start subshell End subshell Start character set file name matching End character set file name matching Start command block Command separator Quote next character Strong quote Weak quote See slide § 45, 17, 39 § 45, 17, 39 2.9 on page 42 2.9 on page 42 § 39 § 40 § 23 § 23 § 23 Intro Quoting and Funny Chars Special Characters Special Chars—2 Special Chars—3 Quoting Quoting—2 Quoting—When to use it? Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 9/86 Special Characters—continued: 3 Character < > / ? ! space or tab Meaning Input redirect Output redirect Pathname directory separator Single-character match in filenames Pipline logical NOT shell normally splits at white space See slide 2.7 on page 40 2.6 on page 39 2.18 on page 51 § 28 §44 Intro Quoting and Funny Chars Special Characters Special Chars—2 Special Chars—3 Quoting Quoting—2 Quoting—When to use it? Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find I Note that references to pages in the tables above refer to the modules in the workshop notes OSSI — ver. 1.12 Shell Programming - p. 10/86 Quoting I I I I Sometimes you want to use a special character literally ; i.e., without its special meaning. Called quoting Suppose you want to print the string: 2 * 3 > 5 is a valid inequality? If you did this: $ echo 2 * 3 > 5 is a valid inequality the new file ‘5’ is created, containing the character ‘2’, then the names of all the files in the current directory, then the string “3 is a valid inequality”. Intro Quoting and Funny Chars Special Characters Special Chars—2 Special Chars—3 Quoting Quoting—2 Quoting—When to use it? Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 11/86 Quoting—2 I I To make it work, you need to protect the special characters ‘*’ and ‘>’ from the shell by quoting them. There are three methods of quoting: N Using double quotes (“weak quotes”) N Using single quotes (“strong quotes”) N Using a backslash in front of each special character you want to quote This example shows all three: $ echo "2 * 3 > 5 is a valid inequality" $ echo ’2 * 3 > 5 is a valid inequality’ $ echo 2 \* 3 \> 5 is a valid inequality Intro Quoting and Funny Chars Special Characters Special Chars—2 Special Chars—3 Quoting Quoting—2 Quoting—When to use it? Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 12/86 Quoting—When to use it? I I I Use quoting when you want to pass special characters to another program. Examples of programs that often use special characters: N find, locate, grep, expr, sed and echo Here are examples where quoting is required for the program to work properly: $ find . -name \*.jpg $ locate ’/usr/bin/c*’ $ grep ’main.*(’ *.c $ i=$(expr i \* 5) Intro Quoting and Funny Chars Special Characters Special Chars—2 Special Chars—3 Quoting Quoting—2 Quoting—When to use it? Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 13/86 True and False I I I I I Shell programs depend on executing external programs When any external program execution is successful, the exit status is zero, 0 An error results in a non-zero error code To match this, in shell programming: N The value 0 is true N any non-zero value is false This is opposite from other programming languages Intro Quoting and Funny Chars Variables True and False Variables—1 Variables—Assignments Variables—Local to Script Variables—Unsetting Them Command-line Parameters Special Built-in Variables Variables: use Braces ${...} After $9 More about Quoting Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find Shell Programming - p. 14/86 OSSI — ver. 1.12 Variables—1 I I I Variables not declared; they just appear when assigned to Assignment: N no dollar sign N no space around equals sign N examples: $ x=10 # correct $ x = 10 # wrong: try to execute program called ‘‘x’’ Read value of variable: N put a ‘$’ in front of variable name N example: $ echo "The value of x is $x" Intro Quoting and Funny Chars Variables True and False Variables—1 Variables—Assignments Variables—Local to Script Variables—Unsetting Them Command-line Parameters Special Built-in Variables Variables: use Braces ${...} After $9 More about Quoting Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find Shell Programming - p. 15/86 OSSI — ver. 1.12 Variables—Assignments I I You can put multiple assignments on one line: i=0 j=10 k=100 You can set a variable temporarily while executing a program: $ echo $EDITOR emacsclient $ EDITOR=gedit crontab -e $ echo $EDITOR emacsclient Intro Quoting and Funny Chars Variables True and False Variables—1 Variables—Assignments Variables—Local to Script Variables—Unsetting Them Command-line Parameters Special Built-in Variables Variables: use Braces ${...} After $9 More about Quoting Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find Shell Programming - p. 16/86 OSSI — ver. 1.12 Variables—Local to Script I I Variables disappear after a script finishes Variables created in a sub shell disappear N parent shell cannot read variables in a sub shell N example: $ cat variables #! /bin/sh echo $HOME HOME=happy echo $HOME $ ./variables /home/nicku happy $ echo $HOME /home/nicku Intro Quoting and Funny Chars Variables True and False Variables—1 Variables—Assignments Variables—Local to Script Variables—Unsetting Them Command-line Parameters Special Built-in Variables Variables: use Braces ${...} After $9 More about Quoting Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find Shell Programming - p. 17/86 OSSI — ver. 1.12 Variables—Unsetting Them I I You can make a variable hold the null string by assigning it to nothing, but it does not disappear totally: $ VAR= $ set | grep ’ˆVAR’ VAR= You can make it disappear totally using unset: $ unset VAR $ set | grep ’ˆVAR’ Intro Quoting and Funny Chars Variables True and False Variables—1 Variables—Assignments Variables—Local to Script Variables—Unsetting Them Command-line Parameters Special Built-in Variables Variables: use Braces ${...} After $9 More about Quoting Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find Shell Programming - p. 18/86 OSSI — ver. 1.12 Command-line Parameters I I Command-line parameters are called $0, $1, $2, . . . Example: when call a shell script called “shell-script” like this: $ shell-script param1 param2 param3 param4 variable $0 $1 $2 $3 $4 $# N Intro Quoting and Funny Chars Variables True and False Variables—1 Variables—Assignments Variables—Local to Script Variables—Unsetting Them Command-line Parameters Special Built-in Variables Variables: use Braces ${...} After $9 More about Quoting Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions value shell-script param1 param2 param3 param4 number of parameters to the program, e.g., 4 Note: these variables are read-only. Debugging Regular Expressions awk and sed find Shell Programming - p. 19/86 OSSI — ver. 1.12 Special Built-in Variables I I I I I Both $@ and $* are a list of all the parameters. The only difference between them is when they are quoted in quotes—see manual page for bash $? is exit status of last command $$ is the process ID of the current shell Example shell script: #! /bin/sh echo $0 is the full name of this shell script echo first parameter is $1 echo first parameter is $2 echo first parameter is $3 echo total number of parameters is $# echo process ID is $$ Intro Quoting and Funny Chars Variables True and False Variables—1 Variables—Assignments Variables—Local to Script Variables—Unsetting Them Command-line Parameters Special Built-in Variables Variables: use Braces ${...} After $9 More about Quoting Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find Shell Programming - p. 20/86 OSSI — ver. 1.12 Variables: use Braces ${...} I I It’s good to put braces round a variable name when getting its value Then no problem to join its value with other text: $ test=123 $ echo ${test} 123 # No good, variable $test456 is undefined: $ echo $test456 $ echo ${test}456 123456 Intro Quoting and Funny Chars Variables True and False Variables—1 Variables—Assignments Variables—Local to Script Variables—Unsetting Them Command-line Parameters Special Built-in Variables Variables: use Braces ${...} After $9 More about Quoting Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find Shell Programming - p. 21/86 OSSI — ver. 1.12 Braces and Parameters after $9 I I Need braces to access parameters after $9: $ cat paramten #! /bin/sh echo $10 echo ${10} $ ./paramten a b c d e f g h i j a0 j Notice that $10 is the same as ${1}0, i.e., the first parameter “a” then the literal character zero “0” Intro Quoting and Funny Chars Variables True and False Variables—1 Variables—Assignments Variables—Local to Script Variables—Unsetting Them Command-line Parameters Special Built-in Variables Variables: use Braces ${...} After $9 More about Quoting Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find Shell Programming - p. 22/86 OSSI — ver. 1.12 More about Quoting I I I Double quotes: "..." stop the special behaviour of all special characters, except for: N variable interpretation ($) N backticks (‘) — see slide 24 N the backslash (\) Single quotes ’...’: N stop the special behaviour of all special characters Backslash: N preserves literal behaviour of character, except for newline; see slides §29, §31, §35 N Putting “\” at the end of the line lets you continue a long line on more than one physical line, but the shell will treat it as if it were all on one line. Intro Quoting and Funny Chars Variables True and False Variables—1 Variables—Assignments Variables—Local to Script Variables—Unsetting Them Command-line Parameters Special Built-in Variables Variables: use Braces ${...} After $9 More about Quoting Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find Shell Programming - p. 23/86 OSSI — ver. 1.12 Command Substitution — $(...) or ‘...‘ I I I Enclose command in $(...) or backticks:‘...‘ Means, “Execute the command in the $(...) and put the output back here.” Here is an example using expr: $ expr 3 + 2 5 $ i=expr 3 + 2 # error: try execute command ‘3’ $ i=$(expr 3 + 2) # correct $ i=‘expr 3 + 2‘ # also correct Intro Quoting and Funny Chars Variables Command Substitution Command Substitution Example of Cmd Subst Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 24/86 Command Substitution—Example I I I I We want to put the output of the command hostname into a variable: $ hostname nickpc.tyict.vtc.edu.hk $ h=hostname $ echo $h hostname Oh dear, we only stored the name of the command, not the output of the command! Command substitution solves the problem: $ h=$(hostname) $ echo $h nickpc.tyict.vtc.edu.hk We put $(...) around the command. You can then assign the output of the command. Intro Quoting and Funny Chars Variables Command Substitution Command Substitution Example of Cmd Subst Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 25/86 Conditions—String Comparisons I I I All programming languages depend on conditions for if statements and for while loops Shell programming uses a built-in command which is either test or [...] Examples of string comparisons: [ [ [ [ [ Intro Quoting and Funny Chars Variables Command Substitution Conditions Comparing Strings Comparing Integers "$USER" = root ] # true if the value of $USER is "root" File Tests & NOT AND OR Conditions "$USER" != root ] # true if the value of $USER is not "root" -z "$USER" ] # true if the string "$USER" has zero length Arithmetic string1 \< string2 ] # true if string1 sorts less than string2 Statements string1 \> string2 ] # true if string1 sorts greater than string2 Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find I I I Note that we need to quote the ‘>’ and the ‘<’ to avoid interpreting them as file redirection. Note: the spaces after the “[“ and before the “]” are essential. Also spaces are essential around operators OSSI — ver. 1.12 Shell Programming - p. 26/86 Conditions—Integer Comparisons I Examples of numeric integer comparisons: [ [ [ [ [ [ "$x" "$x" "$x" "$x" "$x" "$x" -eq -ne -lt -gt -le -ge 5 5 5 5 5 5 ] ] ] ] ] ] # # # # # # true true true true true true if if if if if if the value of $x is 5 integer $x is not 5 integer $x is < 5 integer $x is > 5 integer $x is ≤ 5 integer $x is ≥ 5 Intro Quoting and Funny Chars Variables Command Substitution Conditions Comparing Strings Comparing Integers File Tests & NOT AND OR Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find I I Note again that the spaces after the “[“ and before the “]” are essential. Also spaces are essential around operators OSSI — ver. 1.12 Shell Programming - p. 27/86 Conditions—File Tests, NOT Operator I I I The shell provides many tests of information about files. Do man test to see the complete list. Some examples: [ [ [ [ [ [ [ [ [ -f file ] # true if file is an ordinary file ! -f file ] # true if file is NOT an ordinary file -d file ] # true if file is a directory -u file ] # true if file has SUID permission -g file ] # true if file has SGID permission -x file ] # true if file exists and is executable -r file ] # true if file exists and is readable -w file ] # true if file exists and is writeable file1 -nt file2 ] # true if file1 is newer than file2 Intro Quoting and Funny Chars Variables Command Substitution Conditions Comparing Strings Comparing Integers File Tests & NOT AND OR Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find $ $ $ $ $ $ $ $ $ I I Note again: the spaces after the “[“ and before the “]” are essential. Also spaces are essential around operators OSSI — ver. 1.12 Shell Programming - p. 28/86 Conditions—Combining Comparisons I Examples of combining comparisons with AND: -a and OR: -o, and grouping with \(...\) # true if the value of $x is 5 AND $USER is not equal to root: [ "$x" -eq 5 -a "$USER" != root ] # true if the value of $x is 5 OR $USER is not equal to root: [ "$x" -eq 5 -o "$USER" != root ] # true if ( the value of $x is 5 OR $USER is not equal to root ) AND # ( $y > 7 OR $HOME has the value happy ) [ \( "$x" -eq 5 -o "$USER" != root \) -a \ \( "$y" -gt 7 -o "$HOME" = happy \) ] Intro Quoting and Funny Chars Variables Command Substitution Conditions Comparing Strings Comparing Integers File Tests & NOT AND OR Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find I I Note again that the spaces after the “[“ and before the “]” are essential. Do man test to see the information about all the operators. OSSI — ver. 1.12 Shell Programming - p. 29/86 Arithmetic Assignments I Can do with the external program expr N . . . but expr is not so easy to use, although it is very standard and Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Arithmetic Assignments $((...)) portable: see man expr N Easier is to use the built in let command I see help let N Examples: $ $ $ $ $ $ $ $ let let let let let let let let x=1+4 ++x x=’1 + 4’ ’x = 1 + 4’ x="(2 + 3) * 5" "x = 2 + 3 * 5" "x += 5" "x = x + 5" # Now x is 6 ((...)) Statements Input & Output Command-line Parameters # # # # now now now now x x x x is is is is 25 17 22 27; NOTE NO $ Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find N Notice that you do not need to quote the special characters with let. N Quote if you want to use white space. N Do not put a dollar in front of variable, even on right side of assignment; see last example. OSSI — ver. 1.12 Shell Programming - p. 30/86 Arithmetic Expressions with $((...)) I I I The shell interprets anything inside $((...)) as an arithmetic expression You could calculate the number of days left in the year like this: $ echo "There are \ $(( (365-$(date +%j)) / 7 )) weeks \ left till December 31" No dollar sign in front of variables in these arithmetic expressions. Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Arithmetic Assignments $((...)) ((...)) Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 31/86 Arithmetic Conditions with ((...)) I I I A (less portable) alternative to the arithmetic conditions in slide 27 is putting the expression in ((...)) So you can do (( (3>2) && (4<=1) )) instead of [ \( 3 -gt 2 \) -a \( 4 -le 1 \) ] Operators that work with let, $((...)) and ((...)) include: ++ -- ** + - * / % << >> & | ˜ ! ˆ < > <= >= == != ?: which have exactly the same effect as in the C programming language N except exponentiation operator **, i.e., echo $((2**20)) prints the value of 220 , i.e., 1048576 N For details, see $ help let Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Arithmetic Assignments $((...)) ((...)) Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 32/86 if Statement I Syntax: if test-commands then statements-if-test-commands-1-true Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements elif then else test-commands-2 statements-if-test-commands-2-true statements-if-all-test-commands-false if Statement while Statement for Statement for Loops: Another Example for (( ; ; )) break and continue fi I Blocks: {...} Flow Control: || && Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find Example: if grep nick /etc/passwd > /dev/null 2>&1 then echo Nick has a local account here else echo Nick has no local account here fi OSSI — ver. 1.12 Shell Programming - p. 33/86 while Statement I Syntax: while do done test-commands Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements if Statement while Statement for Statement for Loops: Another Example for (( ; ; )) break and continue loop-body-statements I Example: i=0 while [ "$i" -lt 10 ] do echo -n "$i " # -n suppresses newline. let "i = i + 1" # i=$(expr $i + 1) also works done Blocks: {...} Flow Control: || && Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 34/86 for Statement I Syntax: for name in words do loop-body-statements Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements if Statement while Statement for Statement for Loops: Another Example for (( ; ; )) break and continue I done Example: for planet in Mercury Venus Earth Mars \ Jupiter Saturn Uranus Neptune Pluto do echo $planet done N The backslash “\” quotes the newline. It’s just a way of folding a long line in a shell script over two or more lines. Blocks: {...} Flow Control: || && Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 35/86 for Loops: Another Example I Here the shell turns *.txt into a list of file names ending in “.txt”: for i in *.txt do echo $i grep ’lost treasure’ $i done You can leave the in words out; in that case, name is set to each parameter in turn: i=0 for parameter do let ’i = i + 1’ echo "parameter $i is $parameter" done Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements if Statement while Statement for Statement for Loops: Another Example for (( ; ; )) break and continue I Blocks: {...} Flow Control: || && Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 36/86 for Loops: second, C-like syntax I There is a second (less frequently used, and less portable) C-like for loop syntax: for (( do done expr1 Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements if Statement while Statement for Statement for Loops: Another Example for (( ; ; )) break and continue ; expr2 ; expr3 )) loop-body-statements I I Rules: same as for arithmetic conditions—see slide 32 Example: for (( i = 0; i < 10; ++i )) do echo $i done Blocks: {...} Flow Control: || && Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 37/86 break and continue I I I I Use inside a loop Work like they do in C break terminates the innermost loop; execution goes on after the loop continue will skip the rest of the body of the loop, and resume execution on the next itteration of the loop. Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements if Statement while Statement for Statement for Loops: Another Example for (( ; ; )) break and continue Blocks: {...} Flow Control: || && Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 38/86 Blocks: {...} I I I I A subshell is one way of grouping commands together, but it starts a new process, and any variable changes are localised An alternative is to group commands into a block, enclosing a set of commands in braces: {...} Useful for grouping commands for file input or output N . . . though variables are not localised See next slide for another application. Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements if Statement while Statement for Statement for Loops: Another Example for (( ; ; )) break and continue Blocks: {...} Flow Control: || && Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 39/86 Error Handling: ||, && and exit I I Suppose we want the user to provide exactly two parameters, and exit otherwise A common method of handling this is something like: [ $# -eq 2 ] || { echo "Need two parameters"; exit 1; } Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements if Statement while Statement for Statement for Loops: Another Example for (( ; ; )) break and continue I I I I Read this as “the number of parameters is two OR exit” Works because this logical OR uses short-circuit Boolean evaluation; the second statement is executed only if the first fails (is false) Logical AND “&&” can be used in the same way; the second statement will be executed only if the first is successful (true) A note about blocks: must have semicolon “;” or newline at end of last statement before closing brace Blocks: {...} Flow Control: || && Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 40/86 Output: echo and printf I I I I I To perform output, use echo, or for more formatting, printf. Use echo -n to print no newline at end. Just echo by itself prints a newline printf works the same as in the C programming language, except no parentheses or commas: $ printf "%16s\t%8d\n" $my_string $my_number Do man printf (or look it up in the bash manual page) to read all about it. Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Output: echo and printf Input: the read Command Split with set More about set, and IFS Example: Changing IFS Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 41/86 Input: the read Command I I I I I I I For input, use the built-in shell command read read reads standard input and puts the result into one or more variables If use one variable, variable holds the whole line Syntax: read var1 ... Often used with a while loop like this: while read var1 var2 do # do something with $var1 and $var2 done Loop terminates when reach end of file To prompt and read a value from a user, you could do: while [ -z "$value" ]; do echo -n "Enter a value: " read value done # Now do something with $value Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Output: echo and printf Input: the read Command Split with set More about set, and IFS Example: Changing IFS Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 42/86 set: Splitting a Multi-Word Variable I I I Sometimes may want to split a multi-word variable into single-word variables read won’t work like this: MY_FILE_INFO=$(ls -lR | grep $file) # ... echo $MY_FILE_INFO | read perms links \ user group size month day time filename Use the builtin command set instead: MY_FILE_INFO=$(ls -lR | grep $file) # ... set $MY_FILE_INFO perms=$1 links=$2 user=$3 group=$4 size=$5 month=$6 day=$7 time=$8 filename=$9 Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Output: echo and printf Input: the read Command Split with set More about set, and IFS Example: Changing IFS Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 43/86 More about set, and IFS I I I I I I set splits its arguments into pieces (usually) at whitespace It sets the first value as $1, the second as $2, and so on. Note that you can change how set and the shell splits things up by changing the value of a special variable called IFS IFS stands for Internal Field Separator Normally the value of IFS is the string “ space tab newline ” Next slide shows how changing IFS to a colon let us easily next slide split the PATH into separate directories: Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Output: echo and printf Input: the read Command Split with set More about set, and IFS Example: Changing IFS Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 44/86 Example: Changing IFS I I Notice that here, I make the change to IFS in a subshell. I have simply typed the loop at the prompt. As I said in slide 17, changes in a subshell are local to the subshell: $ echo $PATH /usr/bin:/bin:/usr/X11R6/bin:/home/nicku/bin $ (IFS=: > for dir in $PATH > do > echo $dir > done >) /usr/bin /bin /usr/X11R6/bin /home/nicku/bin Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Output: echo and printf Input: the read Command Split with set More about set, and IFS Example: Changing IFS Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 45/86 case Statement I I I Similar to the switch statement in C, but more useful and more general Uses pattern matching against a string to decide on an action to take Syntax: case expression in pattern1 ) statements ;; pattern2 ) statements ;; ... esac Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters case Statement case Statement: Example shift Up shift: Many Places Command-Line Options—1 Command-Line Options—2 Command-Line Options—3 getopts—4 getopts—5 Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 46/86 case Statement: Example This example code runs the appropriate program on a graphics file, depending on the file extension, to convert the file to another format: case $filename in *.tif) tifftopnm $filename > $ppmfile ;; *.jpg) tjpeg $filename > $ppmfile ;; *) echo -n "Sorry, cannot handle this " echo "graphics format" ;; esac I Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters case Statement case Statement: Example shift Up shift: Many Places Command-Line Options—1 Command-Line Options—2 Command-Line Options—3 getopts—4 getopts—5 Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 47/86 shift: Move all Parameters Up I I I Sometimes we want to process command-line parameters in a loop The shift statement is made for this Say that we have four parameters: parameter value parameter value Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters case Statement case Statement: Example shift Up shift: Many Places $1 $2 I one two $3 $4 three four Then after executing the shift statement, the values are now: parameter value parameter value Command-Line Options—1 Command-Line Options—2 Command-Line Options—3 getopts—4 getopts—5 Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find $1 $2 two three $3 $4 four no longer exists OSSI — ver. 1.12 Shell Programming - p. 48/86 shift: Many Places I You can give a number argument to shift: N If before, we have four parameters: parameter value parameter value Intro Quoting and Funny Chars Variables Command Substitution Conditions $1 $2 N one two $3 $4 three four Arithmetic Statements Input & Output Command-line Parameters case Statement case Statement: Example shift Up shift: Many Places After executing the statement: $ shift 2 we have two parameters left: parameter value parameter value no longer exists no longer exists Command-Line Options—1 Command-Line Options—2 Command-Line Options—3 getopts—4 getopts—5 $1 $2 three four $3 $4 Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 49/86 Command-Line Options—1 I Sometimes we want to modify the behaviour of a shell script N For example, want an option to show more information on request N could use an option “-v” (for “verbose”) to tell the shell script that we want it to tell us more information about what it is doing N If script is called showme, then we could use our -v option like this: $ showme -v N the script then shows more information. Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters case Statement case Statement: Example shift Up shift: Many Places Command-Line Options—1 Command-Line Options—2 Command-Line Options—3 getopts—4 getopts—5 Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 50/86 Command-Line Options—2 I I I For example, We might provide an option to give a starting point for a script to search for SUID programs Could make the option -d directory If script is called findsuid, could call it like this: $ findsuid -d /usr to tell the script to start searching in the directory /usr instead of the current directory Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters case Statement case Statement: Example shift Up shift: Many Places Command-Line Options—1 Command-Line Options—2 Command-Line Options—3 getopts—4 getopts—5 Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 51/86 Command-Line Options—3 I We could do this using shift, a while loop, and a case statement, like this: while [ -n "$(echo $1 | grep ’-’)" ] do case $1 in -v) VERBOSE=1 ;; -d) shift DIRECTORY=$1 ;; *) echo "usage: $0 [-v] [-d dir]" exit 1 ;; esac shift done Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters case Statement case Statement: Example shift Up shift: Many Places Command-Line Options—1 Command-Line Options—2 Command-Line Options—3 getopts—4 getopts—5 Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 52/86 getopts: Command-Line Options—4 I Problems with above solution: inflexibility: N Does not allow options to be “bundled” together like -abc instead of -a -b -c N Requires a space between option and its argument, i.e., doesn’t let you do -d/etc as well as -d /etc N Better method: use the built-in command getopts: while getopts ":vd:" opt do case opt in v) VERBOSE=1 ;; d) DIRECTORY=$OPTARG ;; *) echo "usage: $0 [-v] [-d dir]" exit 1 ;; esac done shift $((OPTIND - 1)) Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters case Statement case Statement: Example shift Up shift: Many Places Command-Line Options—1 Command-Line Options—2 Command-Line Options—3 getopts—4 getopts—5 Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 53/86 getopts: Command-Line Options—5 I I I getopts takes two arguments: N first comes the string that can contain letters and colons. I Each letter represents one option I A colon comes after a letter to indicate that option takes an arguement, like -d directory I A colon at the beginning makes getopts less noisy, so you can provide your own error message, as shown in the example. N The second is a variable that will hold the option (without the hyphen “-”) Shift out all processed options using the variable OPTIND, leaving any other arguments accessible Search for getopts in the bash man page Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters case Statement case Statement: Example shift Up shift: Many Places Command-Line Options—1 Command-Line Options—2 Command-Line Options—3 getopts—4 getopts—5 Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 54/86 Temporary Files: mktemp I I I I I Sometimes it is convenient to store temporary data in a temporary file The mktemp program is designed for this We use it something like this: TMPFILE=$(mktemp /tmp/temp.XXXXXX) || exit 1 mktemp will create a new file, replacing the “XXXXXX” with a random string Do man mktemp for the complete manual. Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Temporary Files: mktemp Signals that Kill Signals: trap Signals: trap Example Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 55/86 Signals that may Terminate your Script I I Many key strokes will send a signal to a process Examples: ¨  N Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Temporary Files: mktemp Signals that Kill Signals: trap Signals: trap Example Functions Debugging Regular Expressions awk and sed find N I I I When you log out, all your processes are sent a SIGHUP (hangup) signal If your script is connected to another process that terminates unexpectedly, it will receive a SIGPIPE signal If anyone terminates the program with the kill program, the default signal is SIGTERM Control-C © sends a SIGINT signal to the current  process running in the foreground  ¨ Control-\ © sends a SIGQUIT signal  OSSI — ver. 1.12 Shell Programming - p. 56/86 Signals: trap I I Sometimes you want your script to clean up after itself nicely, and remove temporary files Do this using trap Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Temporary Files: mktemp Signals that Kill Signals: trap Signals: trap Example Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 57/86 Signals: trap Example I I I I I Supose your script creates some temporary files, and you want to remove them if your script recieves any of these signals You can “catch” the signal, and remove the files when the signals are received before the program terminates Suppose the temporary files have names stored in the variables TEMP1 and TEMP2 Then you would trap these signals like this: trap "rm $TEMP1 $TEMP2" HUP INT QUIT PIPE TERM Conveniently, (but not very portably), bash provides a “pretend” signal called EXIT; can add this to the list of signals you trap, so that the temporary files will be removed when the program exits normally. Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Temporary Files: mktemp Signals that Kill Signals: trap Signals: trap Example Functions Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 58/86 Functions I I I The shell supports functions and function calls A function works like an external command, except that it does not start another process Syntax: function functname { shell commands Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters } Or: functname () Temporary Files, Signals Functions Functions Parameters in Functions Example, Calling a Function Debugging Regular Expressions awk and sed find { shell commands } OSSI — ver. 1.12 Shell Programming - p. 59/86 Parameters in Functions I I I I I Work the same as parameters to entire shell script First parameter is $1, second is $2,. . . , the tenth parameter is ${10}, and so on. $# is the number of parameters passed to the function As with command line parameters, they are read-only Assign to meaningful names to make your program more understandable Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Functions Parameters in Functions Example, Calling a Function Debugging Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 60/86 Example, Calling a Function I This is a simple example program: #! /bin/sh function cube { echo $(($1 * $1 * $1)) } j=$(cube 5) echo $j # Output is 125 Note the use of command substitution to get a return value The function prints result to standard output. Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Functions Parameters in Functions Example, Calling a Function Debugging Regular Expressions awk and sed find I I OSSI — ver. 1.12 Shell Programming - p. 61/86 Debugging Shell Scripts—1 I I If you run the script with: $ sh -v script then each statement will be printed as it is executed If you run the script with: $ sh -x script then an execution trace will show the value of all variables as the script executes. Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Debugging Shell Scripts—1 Debugging Shell Scripts—2 Writing Shell Scripts Useful External Programs—1 Useful External Programs—2 Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 62/86 Debugging Shell Scripts—2 I I I Use echo to display the value of variables as the program executes You can turn the -x shell option on in any part of your script with the line: set -x and turn it off with: set +x N No, that’s not a typo: +x turns it off, -x turns it on. The book Learning the bash Shell includes a bash shell debugger if you get desperate Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Debugging Shell Scripts—1 Debugging Shell Scripts—2 Writing Shell Scripts Useful External Programs—1 Useful External Programs—2 Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 63/86 Writing Shell Scripts I I Build your shell script incrementally: N Open the editor in one window (and leave it open), have a terminal window open in which to run your program as you write it N Test as you implement: this makes shell script development easy N Do not write a very complex script, and then begin testing it! Use the standard software engineering practice you know: N Use meaningful variable names, function names N Make your program self-documenting N Add comment blocks to explain obscure or difficult parts of your program Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Debugging Shell Scripts—1 Debugging Shell Scripts—2 Writing Shell Scripts Useful External Programs—1 Useful External Programs—2 Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 64/86 Useful External Programs—1 Each of these has a manual page, and many have info manuals. Read their online documentation for more information. I awk — powerful tool for processing columns of data I basename — remove directory and (optionally) extension from file name I cat — copy to standard output I cut — process columns of data I du — show disk space used by directories and files I egrep, grep — find lines containing patterns in files I find — find files using many criteria Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Debugging Shell Scripts—1 Debugging Shell Scripts—2 Writing Shell Scripts Useful External Programs—1 Useful External Programs—2 Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 65/86 Useful External Programs—2 I I I I I I I last — show the last time a user was logged in lastb — show last bad log in attempt by a user rpm — RPM package manager: manage software package database sed — stream editor: edit files automatically sort — sort lines of files by many different criteria tr — translate one set of characters to another set uniq — replace repeated lines with just one line, optionally with a count of the number of repeated lines Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Debugging Shell Scripts—1 Debugging Shell Scripts—2 Writing Shell Scripts Useful External Programs—1 Useful External Programs—2 Regular Expressions awk and sed find OSSI — ver. 1.12 Shell Programming - p. 66/86 Regular Expressions I I I I Many programs and programming languages use regular expressions, including Java 1.4 and later, Perl, VB.NET, C# (and any language using the .NET Framework), PHP, Python, Ruby, Tcl and MySQL (plus many others; even MS Word uses regular expressions under Edit → Find → More → Use wildcards) These programs use regular expressions: N grep, egrep, sed, awk All programmer’s editors support regular expressions (Emacs, vi, . . . ) Regular expressions provide a powerful language for manipulating data and extracting important information from masses of data Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions Regular Expressions What is In a RegEx? Literal characters Character Classes: [...] [^...] Match Any Character Match Start or End Repetitions Matching Alternatives: “|” Examples awk and sed find Shell Programming - p. 67/86 OSSI — ver. 1.12 What is In a Regular Expression? I There are two types of character in a regular expression: N Metacharacters I These include: I *\.+?^()[{| N Ordinary, literal characters: I i.e., all the other characters that are not metacharacters Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions Regular Expressions What is In a RegEx? Literal characters Character Classes: [...] [^...] Match Any Character Match Start or End Repetitions Matching Alternatives: “|” Examples awk and sed find Shell Programming - p. 68/86 OSSI — ver. 1.12 Literal characters I I I I I Find all lines containing "chan" in the password file: $ grep chan /etc/passwd The regular expression is "chan" It is made entirely of literal characters It matches only lines that contain the exact string It will match lines containing the words chan, changed, merchant, mechanism,. . . Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions Regular Expressions What is In a RegEx? Literal characters Character Classes: [...] [^...] Match Any Character Match Start or End Repetitions Matching Alternatives: “|” Examples awk and sed find Shell Programming - p. 69/86 OSSI — ver. 1.12 Character Classes: [...] I I A character class represents one character Examples: # $ # $ # $ # $ Find grep Find grep Find grep Find grep all words in the dictionary that contain a vowel: "[aeiou]" /usr/share/dict/words all lines that contain a digit: "[0123456789]" /usr/share/dict/words all lines that contain a digit: "[0-9]" /usr/share/dict/words all lines that contain a capital letter: "[A-Z]" /usr/share/dict/words Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions Regular Expressions What is In a RegEx? Literal characters Character Classes: [...] [^...] Match Any Character Match Start or End Repetitions Matching Alternatives: “|” Examples awk and sed find Shell Programming - p. 70/86 OSSI — ver. 1.12 Negated Character Classes: [^...] I Examples of negated character classes: # # $ # # $ $ # # $ Find all words in the dictionary that contain a character that is not a vowel: grep "[ˆaeiou]" /usr/share/dict/words Two ways of finding all lines that contain a character that is not a digit: grep "[ˆ0123456789]" /usr/share/dict/words grep "[ˆ0-9]" /usr/share/dict/words Find all lines that contain a character that is not a digit, or a letter grep "[ˆ0-9a-zA-Z]" /usr/share/dict/words Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions Regular Expressions What is In a RegEx? Literal characters Character Classes: [...] [^...] I Remember: each set of square brackets represents exactly one character. Match Any Character Match Start or End Repetitions Matching Alternatives: “|” Examples awk and sed find Shell Programming - p. 71/86 OSSI — ver. 1.12 Match Any Character I I The dot “.” matches any single character, except a newline. The pattern ‘.....’ matches all lines that contain at least five characters Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions Regular Expressions What is In a RegEx? Literal characters Character Classes: [...] [^...] Match Any Character Match Start or End Repetitions Matching Alternatives: “|” Examples awk and sed find Shell Programming - p. 72/86 OSSI — ver. 1.12 Matching the Beginning or End of Line I I I I I I To match a line that contains exactly five characters: $ grep ’ˆ.....$’ /usr/share/dict/words The hat, ˆ represents the position right at the start of the line The dollar $ represents the position right at the end of the line. Neither ˆ nor $ represents a character They represent a position Sometimes called anchors, since they anchor the other characters to a specific part of the string Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions Regular Expressions What is In a RegEx? Literal characters Character Classes: [...] [^...] Match Any Character Match Start or End Repetitions Matching Alternatives: “|” Examples awk and sed find Shell Programming - p. 73/86 OSSI — ver. 1.12 Match Repetitions: *, ?, +, {n}, {n,m} I I I I I I I To match zero or more: a* represents zero or more of the lower case letter a, so the pattern will match "" (the empty string), “a”, “aa”, “aaaaaaaaaaaaaaa”, “qwewtrryu” or the “nothing” in front of any string! To match one or more: ‘a+’ matches one or more “a”s ‘a?’ matches zero or one “a” ‘a{10}’ matches exactly 10 “a”s ‘a{5,10}’ matches between 5 and 10 (inclusive) “a”s Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions Regular Expressions What is In a RegEx? Literal characters Character Classes: [...] [^...] Match Any Character Match Start or End Repetitions Matching Alternatives: “|” Examples awk and sed find Shell Programming - p. 74/86 OSSI — ver. 1.12 Matching Alternatives: “|” I I I I the vertical bar represents alternatives: The regular expresssion ‘nick|albert|alex’ will match either the string “nick” or the string “albert” or the string “alex” Note that the vertical bar has very low precedence: the pattern ‘^fred|nurk’ matches “fred” only if it occurs at the start of the line, while it will match “nurk” at any position in the line Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions Regular Expressions What is In a RegEx? Literal characters Character Classes: [...] [^...] Match Any Character Match Start or End Repetitions Matching Alternatives: “|” Examples awk and sed find Shell Programming - p. 75/86 OSSI — ver. 1.12 Putting it All Together: Examples I I I Find all words that contain at least three ‘a’s: $ egrep ’a.*a.*a’ /usr/share/dict/words N Why is this different from $ egrep ’aaa’ /usr/share/dict/words Find all words that begin in ‘a’ and finish in ‘z’, ignoring case: $ egrep -i ’ˆa.*z$’ /usr/share/dict/words N How is this different from: $ egrep -i ’ˆa.*z’ /usr/share/dict/words Find all words that contain at least two vowels: $ grep ’[aeiou].*[aeiou]’ /usr/share/dict/words Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions Regular Expressions What is In a RegEx? Literal characters Character Classes: [...] I I Find all words that contain exactly two vowels: $ egrep \ ’ˆ[ˆaeiou]*[aeiou][ˆaeiou]*[aeiou][ˆaeiou]*$’ \ /usr/share/dict/words Find all lines that are empty, or contain only spaces: $ grep ’ˆ *$’ file [^...] Match Any Character Match Start or End Repetitions Matching Alternatives: “|” Examples awk and sed find Shell Programming - p. 76/86 OSSI — ver. 1.12 Basic awk I I I I awk is a complete programming language Mostly used for one-line solutions to problems of extracting columns of data from text, and processing it A complete book is available on awk; you can buy it here: http://www.oreilly.com/catalog/awkprog3/ or read it on your computer, as it is the official manual for gawk (GNU awk); do $ info gawk or read it in Emacs. N A printable postscript file of the book (353 pages) is on my computer at /usr/share/doc/gawk-3.1.3/gawk.ps Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed Basic awk What Does awk Do? awk Examples sed—the Stream Editor sed—Backreferencees Backrefs Example find OSSI — ver. 1.12 Shell Programming - p. 77/86 What Does awk Do? I I I I I I awk reads file(s) or standard input one line at a time, and automatically splits the line into fields, and calls them $1, $2,. . . , $NF NF is equal to the number of fields the line was split into $0 contains the whole line awk has an option -F that allows you to select another pattern as the field separator N Normally awk splits columns by white space To execute code after all lines are processed, create an END block. Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed Basic awk What Does awk Do? awk Examples sed—the Stream Editor sed—Backreferencees Backrefs Example find OSSI — ver. 1.12 Shell Programming - p. 78/86 awk Examples I I I Print the sizes of all files in current directory: ls -l | awk ’{print $5}’ Add the sizes of all files in current directory: ls -l | awk ’{sum += $5} END{print sum}’ Print only the permissions, user, group and file names of files in current directory: ls -l | awk ’{print $1, $3, $4, $NF}’ Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed Basic awk What Does awk Do? awk Examples sed—the Stream Editor sed—Backreferencees Backrefs Example find OSSI — ver. 1.12 Shell Programming - p. 79/86 sed—the Stream Editor I I I I I sed provides many facilities for editing files The substitute command, s///, is the most important The syntax (using sed as an editor of standard input), is: $ sed ’s/ original / replacement /’ Example: replace the first instance of Windows with Linux on each line of the input: sed ’s/Windows/Linux/’ Example: replace all instances of Windows with Linux on each line of the input: sed ’s/Windows/Linux/g’ N Note: by default, sed uses “basic regular expressions”, which require a backslash ‘\’ in front of the metacharacters ’{’, ‘(’, ‘)’, ‘|’, ‘+’ and ‘?’. N To use “extended regular expressions” (which we covered here), call sed with the option -r, as in this example: $ sed -r s/a+// Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed Basic awk What Does awk Do? awk Examples sed—the Stream Editor sed—Backreferencees Backrefs Example find OSSI — ver. 1.12 Shell Programming - p. 80/86 sed—Backreferencees I I I You can match part of the original in a sed -r substitute command, and put that part back into the replacement part. You enclose the part you want to refer to later in (...) You can get the first value in the replacement part by \1, the second opening parenthesis of (...) by \2, and so on. Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed Basic awk What Does awk Do? awk Examples sed—the Stream Editor sed—Backreferencees Backrefs Example find OSSI — ver. 1.12 Shell Programming - p. 81/86 sed—Backreferencees: Example I I I If you do $ find /etc | xargs file -b you will get a lot of output like this: symbolic link to bg5ps.conf.zh_TW.Big5 symbolic link to rc.d/rc.local symbolic link to rc.d/rc symbolic link to rc.d/rc.sysinit symbolic link to ../../X11/xdm/Xservers If you want to edit each line to remove everything after “symbolic link”, then you could pipe the data through sed like this: $ find /etc | xargs file -b \ | sed -r ’s/(symbolic link).*/\1/’ See slide 83 for an application Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed Basic awk What Does awk Do? awk Examples sed—the Stream Editor sed—Backreferencees Backrefs Example find OSSI — ver. 1.12 Shell Programming - p. 82/86 find Examples I I Count the number of unique manual pages on the computer: $ find /usr/share/man -type f | wc -l Print a table of types of file under the /etc directory, with the most common file type down at the bottom: $ find /etc | xargs file -b \ | sed -r ’s/(symbolic link).*/\1/’ \ | sort \ | uniq -c \ | sort -n Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find find Examples Finding SUID Programs Long find Example rpm Queries OSSI — ver. 1.12 Shell Programming - p. 83/86 Finding SUID Programs I I Finding SUID or SGID files: $ sudo find / -type f \ \( perm -2000 -o -perm -4000 \) \ > files.secure Let’s compare with a list of SUID and SGID files to see if there are any changes, since SUID and SGID programs can be a security risk: $ sudo find / -type f \ \( perm -2000 -o -perm -4000 \) \ | diff - files.secure Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find find Examples Finding SUID Programs Long find Example rpm Queries OSSI — ver. 1.12 Shell Programming - p. 84/86 A find Example with Many Options I Set all directories to have the access mode 771, set all backup files (*.BAK) to mode 600, all shell scripts (*.sh) to mode 755, and all text files (*.txt) to mode 644: -type -name -name -name d "*.BAK" "*.sh" "*.txt" -a -a -a -a exec exec exec exec chmod chmod chmod chmod 771 600 755 644 Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find find Examples $ find . \( \( \( \( {} {} {} {} \; \; \; \; \) -o \ \) -o \ \) -o \ \) Finding SUID Programs Long find Example rpm Queries OSSI — ver. 1.12 Shell Programming - p. 85/86 rpm Database Query Commands I I I I I The rpm software package management system includes a database with very detailed information about every file of every software package that is installed on the computer. You can query this database using the rpm command. The manual page does not give the complete picture, but there is a book called Maximum RPM that comes on the Red Hat documentation CD This package is installed on ictlab You can see the appropriate section at this URL: http://ictlab.tyict.vtc.edu.hk/doc/maximum- rpm- 1.0/html/s1- rpm- query- parts.html Intro Quoting and Funny Chars Variables Command Substitution Conditions Arithmetic Statements Input & Output Command-line Parameters Temporary Files, Signals Functions Debugging Regular Expressions awk and sed find find Examples Finding SUID Programs Long find Example rpm Queries OSSI — ver. 1.12 Shell Programming - p. 86/86