最近在学习shell时发现虽然我经常使用test和case，但是它们仍然有许多不为人知的秘密。
首先来看test
两个test
在linux中有两个test命令，一个是作为shell的内置命令，它是一个程序；另一个一般是磁盘的/usr/bin/test命令，但他一般不会被调用，因为首先找到的是shell的内置命令。
此外，test的另一个名称是[ 。 当调用[ 时，它需要一个] 作为其参数，但是如果不使用]，[ 也能起到同样的作用。
[root@localhost ~]# type test; type [ test is a shell builtin [ is a shell builtin [root@localhost ~]# which test; which [ /usr/bin/test /usr/bin/[
从中可以看到，在shell中，[是程序，而]只是[所需要的参数。
常用的测试标志
-e 是否存在；
-s 是否存在且不为空；
-f 是否为常规文件；
-b 是否为块设备（块设备的驱动对象是像磁盘这样的按块进行操作的设备）；
-c 是否为字符设备（字符设备的驱动对象可以向其读写字符的设备，如终端 ）；
-L(-h) 是否为符号链接；
-(rwx) 是否具有相应的权限，
8. -O(-G) 是否属于当前用户和/或组ID;
9. -u(-g) 是否具有suid或sgid(suid位与sgid位允许程序以文件所属用户(或组)的身份运行，而不一定是运行程序的用户(组) )；
[root@localhost ~]# cat suid.sh #!/bin/bash while read -p "What file do you want to test? :" filename do if [ ! -e "$filename" ]; then echo "The file does not exists." continue fi #Okey, the file exists ls -ld $filename if [ -u $filename ]; then echo " $filename will run as user \"`stat --printf=%U $filename`\"" fi if [ -g $filename ]; then echo "$filename will run as group \"`stat --printf=%G $filename`\"" fi done
10. -N标志检测文件自从上次读取之后是否被修改。
[root@localhost ~]# echo hello > /tmp/myfile.log
[root@localhost ~]# echo hello world >> /tmp/myfile.log
[root@localhost ~]# cat watchfile.sh
#!/bin/bash
GAP=10 # how long to wait
LOGFILE=$1 # file to log to
# Get the current length of the file.
len=`wc -l $LOGFILE | awk '{ print $1 }'`
echo "Current size is $len lines"
while :
do
if [ -N $LOGFILE ]; then
echo "`date`: New entries in $LOGFILE:"
newlen=`wc -l $LOGFILE | awk '{ print $1 }'`
newlines=`expr $newlen - $len`
tail -$newlines $LOGFILE
len=$newlen
fi
sleep $GAP
done文件比较测试
-ef比较两个文件是否为同一文件系统中相同inode节点的硬链接。
[root@localhost ~]# cat equalfile.sh #!/bin/bash file1=$1 file2=$2 ls -il $file1 $file2 if [ $file1 -ef $file2 ]; then echo "$file1 is the same file as $file2" else echo "$file1 is not the same file as $file2" diff -q $file1 $file2 if [ "$?" -eq "0" ]; then echo "However, their contents are identical." fi fi [root@localhost ~]# echo testing > file1 [root@localhost ~]# ln file1 file2 [root@localhost ~]# echo testing > file3 [root@localhost ~]# ls -il file? 131679 -rw-r--r--. 2 root root 8 Nov 28 05:42 file1 131679 -rw-r--r--. 2 root root 8 Nov 28 05:42 file2 131995 -rw-r--r--. 1 root root 8 Nov 28 05:42 file3 [root@localhost ~]# sh equalfile.sh file1 file2 131679 -rw-r--r--. 2 root root 8 Nov 28 05:42 file1 131679 -rw-r--r--. 2 root root 8 Nov 28 05:42 file2 file1 is the same file as file2 [root@localhost ~]# sh equalfile.sh file1 file3 131679 -rw-r--r--. 2 root root 8 Nov 28 05:42 file1 131995 -rw-r--r--. 1 root root 8 Nov 28 05:42 file3 file1 is not the same file as file3 However, their contents are identical. [root@localhost ~]# echo something > file4 [root@localhost ~]# sh equalfile.sh file1 file4 131679 -rw-r--r--. 2 root root 8 Nov 28 05:42 file1 132003 -rw-r--r--. 1 root root 10 Nov 28 05:45 file4 file1 is not the same file as file4 Files file1 and file4 differ
字符串比较测试
可以测试字符串是否相等，以及按照字母表测试某个字符串是否排在另一个前面。
需要注意的是： <和>只能在复合命令[[ ... ]]中起作用；
==和=！可以在单方括号和双括号中起作用；
=只能在单括号中起作用；
导致这种复杂性的原因是为了在添加更加强大的[[命令的同时，保持与Bourne shell的兼容性。
参数:
-z 测试在字符串长度为0时返回真，-n 在字符串长度非零时返回真。
因为测试的字符串可能为空，所以需要用引号将变量名引用起来：否则[ -z $input ]就会成为[ -z ],这是错的，只有[ -z "$input" ]才是合法的。
[root@localhost ~]# cat alnum.sh #!/bin/bash if [ "$1" = "$2" ]; then echo "$1 is the same as $2" else echo "$1 is not the same as $2" if [[ "$1" < "$2" ]]; then echo "$1 comes before $2" else echo "$1 comes after $2" fi fi [root@localhost ~]# sh alnum.sh apples bananas apples is not the same as bananas apples comes before bananas [root@localhost ~]# sh alnum.sh oranges oranges oranges is the same as oranges
正则表达式测试
参数： =～ 它将右边当作扩展的正则表达式，故可以通过[[ $pkgname =~ .*\.deb ]] 来查找匹配模式*.deb的文件名。注意要使用[[ ... ]]
[root@localhost ~]# cat identify_pkgs.sh
#!/bin/bash
for pkg in pkgs/*
do
pkgname=`basename $pkg`
echo $pkgname
if [[ $pkgname =~ (.+)_(.*)_(.*)\.deb ]]; then
echo "Package ${BASH_REMATCH[1]} Version ${BASH_REMATCH[2]} is a"
echo " Debian package for the ${BASH_REMATCH[3]} architecture."
echo
elif [[ $pkgname =~ (.+)-(.+)\.(.*)\.rpm ]]; then
echo "Package ${BASH_REMATCH[1]} Version ${BASH_REMATCH[2]} is an"
echo " RPM for the ${BASH_REMATCH[3]} architecture."
echo
else
echo "File \"$pkgname\" does not appear to match the"
echo " standard .deb or .rpm naming conventions."
fi
done
[root@localhost ~]# ls pkgs/
dbus-x11_1.2.24-4_and64.deb gnome-desktop-1.023.x86_64.rpm
firmware-linux-free_2.6.32-29_all.deb libgnomeprint-2.18.6-2.6.x86_64.rpm
[root@localhost ~]# sh identify_pkgs.sh
dbus-x11_1.2.24-4_and64.deb
Package dbus-x11 Version 1.2.24-4 is a
Debian package for the and64 architecture.
firmware-linux-free_2.6.32-29_all.deb
Package firmware-linux-free Version 2.6.32-29 is a
Debian package for the all architecture.
gnome-desktop-1.023.x86_64.rpm
Package gnome-desktop Version 1.023 is an
RPM for the x86_64 architecture.
libgnomeprint-2.18.6-2.6.x86_64.rpm
Package libgnomeprint-2.18.6 Version 2.6 is an
RPM for the x86_64 architecture.数值测试
常用参数: -eq 相等返回真； -ne 不相等返回真；
-lt 小于返回真； -gt 大于返回真；
-le 小于等于返回真； -ge 大于等于返回真；
[root@localhost ~]# cat numberguess.sh #!/bin/bash MAX=50 guess=-1 let answer=($RANDOM %$MAX) let answer+=1 ceiling=$MAX floor=0 guesses=0 while [ "$guess" -ne "$answer" ] do echo "The magic number is between $floor and $ceiling." echo -en " Make your guess:" read guess guesses=`expr $guesses + 1` if [ "$guess" -lt "$answer" ]; then echo "$guess is too low" if [ "$guess" -ge "$floor" ]; then floor=`expr $guess + 1` fi fi if [ "$guess" -gt "$answer" ]; then echo "$guess is too high" if [ "$guess" -lt "$ceiling" ]; then ceiling=`expr $guess - 1` fi fi done echo "You got it in $guesses guesses!" [root@localhost ~]# sh numberguess.sh The magic number is between 0 and 50. Make your guess:25 25 is too high The magic number is between 0 and 24. Make your guess:12 12 is too low The magic number is between 13 and 24. Make your guess:18 18 is too high The magic number is between 13 and 17. Make your guess:15 15 is too low The magic number is between 16 and 17. Make your guess:16 16 is too low The magic number is between 17 and 17. Make your guess:17 You got it in 6 guesses!
组合测试
使用&&和||操作符来将测试组合
[root@localhost ~]# wc -l /etc/hosts || echo wc failed to read /etc/hosts 2 /etc/hosts [root@localhost ~]# wc -l /etc/hosts.bak || echo wc failed to read /etc/hosts wc: /etc/hosts.bak: No such file or directory wc failed to read /etc/hosts [root@localhost ~]# wc -l /etc/hosts | grep "^3" || echo /etc/hosts is 3 lines file /etc/hosts is 3 lines file [root@localhost ~]# wc -l /etc/hosts | grep "^2" && echo /etc/hosts is 2 lines file 2 /etc/hosts /etc/hosts is 2 lines file
再来看case
对于case很多人都了解，但是关于用 ;; , ;;& 和 ;&来终止语句，你了解吗？
;; 表示不再执行其他语句；
;;& 表示还要匹配接下来的所有模式；
;& 表示接下来的模式已经匹配；
[root@localhost ~]# cat case2.sh #!/bin/bash read -p "Which city are you closest to?:" city case $city in "New York"|London|Paris|Tokyo) echo "That is a capital city" ;;& Chicago|Detroit|"New York"|Washington) echo "you are in the USA" ;; London|Edinburgh|Cardiff|Dublin) echo "you are in the United Kingdom" ;; "Ramsey Street") echo "G'Day Netghbour!" ;& Melbourne|Canberra|Sydney) echo "you are in australai" ;; Paris) echo "you are in France" ;; Tokyo) echo "you are in Japan" ;; N*) echo "your word begins with N but is not New York!" ;; *) echo "I'm sorry, I don't konw anything about $city" ;; esac [root@localhost ~]# sh case2.sh Which city are you closest to?:London That is a capital city you are in the United Kingdom [root@localhost ~]# sh case2.sh Which city are you closest to?:Ramsey street I'm sorry, I don't konw anything about Ramsey street [root@localhost ~]# sh case2.sh Which city are you closest to?:Ramsey Street G'Day Netghbour! you are in australai
