最新消息:

BASH的保护性编程技巧

bash admin 2553浏览 0评论

这是我写BASH程序的招式。这里本没有什么新的内容,但是从我的经验来看,人们爱滥用BASH。他们忽略了计算机科学,而从他们的程序中创造的是“大泥球”(译注:指架构不清晰的软件系统)。

在此我告诉你方法,以保护你的程序免于障碍,并保持代码的整洁。

不可改变的全局变量

  • 尽量少用全局变量
  • 以大写命名
  • 只读声明
  • 用全局变量来代替隐晦的$0,$1等
  • 在我的程序中常使用的全局变量:
1
2
3
readonly PROGNAME=$(basename $0)
readonly PROGDIR=$(readlink -m $(dirname $0))
readonly ARGS="$@"

一切皆是局部的

所有变量都应为局部的。

1
2
3
4
5
6
7
change_owner_of_file() {
    local filename=$1
    local user=$2
    local group=$3
    chown $user:$group $filename
}
1
2
3
4
5
6
7
8
9
10
11
change_owner_of_files() {
    local user=$1; shift
    local group=$1; shift
    local files=$@
    local i
    for i in $files
    do
        chown $user:$group $i
    done
}
  • 自注释(self documenting)的参数
  • 通常作为循环用的变量i,把它声明为局部变量是很重要的。
  • 局部变量不作用于全局域。
1
2
kfir@goofy ~ $ local a
bash: local: can only be used in a function

main()

  • 有助于保持所有变量的局部性
  • 直观的函数式编程
  • 代码中唯一的全局命令是:main
1
2
3
4
5
6
7
8
9
10
main() {
    local files="/tmp/a /tmp/b"
    local i
    for i in $files
    do
        change_owner_of_file kfir users $i
    done
}
main

一切皆是函数

  • 唯一全局性运行的代码是:

– 不可变的全局变量声明

– main()函数

  • 保持代码整洁
  • 过程变得清晰
1
2
3
main() {
    local files=$(ls /tmp | grep pid | grep -v daemon)
}
1
2
3
4
5
6
7
8
9
10
11
temporary_files() {
    local dir=$1
    ls $dir
        | grep pid
        | grep -v daemon
}
main() {
    local files=$(temporary_files /tmp)
}
  • 第二个例子好得多。查找文件是temporary_files()的问题而非main()的。这段代码用temporary_files()的单元测试也是可测试的。
  • 如果你一定要尝试第一个例子,你会得到查找临时文件以和main算法的大杂烩。
1
2
3
4
5
6
7
8
9
10
11
12
test_temporary_files() {
    local dir=/tmp
    touch $dir/a-pid1232.tmp
    touch $dir/a-pid1232-daemon.tmp
    returns "$dir/a-pid1232.tmp" temporary_files $dir
    touch $dir/b-pid1534.tmp
    returns "$dir/a-pid1232.tmp $dir/b-pid1534.tmp" temporary_files $dir
}
如你所见,这个测试不关心main()。

调试函数

  • 带-x标志运行程序:
1
bash -x my_prog.sh
只调试一小段代码,使用set-x和set+x,会只对被set -x和set +x包含的当前代码打印调试信息。
1
2
3
4
5
6
7
8
9
temporary_files() {
    local dir=$1
    set -x
    ls $dir
        | grep pid
        | grep -v daemon
    set +x
}
打印函数名和它的参数:
1
2
3
4
5
6
7
8
temporary_files() {
    echo $FUNCNAME $@
    local dir=$1
    ls $dir
        | grep pid
        | grep -v daemon
}

调用函数:

1
temporary_files /tmp

会打印到标准输出:

1
temporary_files /tmp

代码的清晰度

这段代码做了什么?

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
main() {
    local dir=/tmp
    [[ -z $dir ]]
        && do_something...
    [[ -n $dir ]]
        && do_something...
    [[ -f $dir ]]
        && do_something...
    [[ -d $dir ]]
        && do_something...
}
main

让你的代码说话:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
is_empty() {
    local var=$1
    [[ -z $var ]]
}
is_not_empty() {
    local var=$1
    [[ -n $var ]]
}
is_file() {
    local file=$1
    [[ -f $file ]]
}
is_dir() {
    local dir=$1
    [[ -d $dir ]]
}
main() {
    local dir=/tmp
    is_empty $dir
        && do_something...
    is_not_empty $dir
        && do_something...
    is_file $dir
        && do_something...
    is_dir $dir
        && do_something...
}
main

每一行只做一件事

  • 用反斜杠来作分隔符。例如:
1
2
3
4
5
temporary_files() {
    local dir=$1
    ls $dir | grep pid | grep -v daemon
}

可以写得简洁得多:

1
2
3
4
5
6
7
temporary_files() {
    local dir=$1
    ls $dir
        | grep pid
        | grep -v daemon
}
  • 符号在缩进行的开始

符号在行末的坏例子:(译注:原文在此例中用了temporary_files()代码段,疑似是贴错了。结合上下文,应为print_dir_if_not_empty())

1
2
3
4
5
6
7
print_dir_if_not_empty() {
    local dir=$1
    is_empty $dir &&
        echo "dir is empty" ||
        echo "dir=$dir"
}

好的例子:我们可以清晰看到行和连接符号之间的联系。

1
2
3
4
5
6
7
print_dir_if_not_empty() {
    local dir=$1
    is_empty $dir
        && echo "dir is empty"
        || echo "dir=$dir"
}

打印用法

不要这样做:

1
2
3
echo "this prog does:..."
echo "flags:"
echo "-h print help"

它应该是个函数:

1
2
3
4
5
usage() {
    echo "this prog does:..."
    echo "flags:"
    echo "-h print help"
}

echo在每一行重复。因此我们得到了这个文档:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
usage() {
    cat <<- EOF
    usage: $PROGNAME options
    Program deletes files from filesystems to release space.
    It gets config file that define fileystem paths to work on, and whitelist rules to
    keep certain files.
    OPTIONS:
       -c --config              configuration file containing the rules. use --help-config to see the syntax.
       -n --pretend             do not really delete, just how what you are going to do.
       -t --test                run unit test to check the program
       -v --verbose             Verbose. You can specify more then one -v to have more verbose
       -x --debug               debug
       -h --help                show this help
          --help-config         configuration help
    Examples:
       Run all tests:
       $PROGNAME --test all
       Run specific test:
       $PROGNAME --test test_string.sh
       Run:
       $PROGNAME --config /path/to/config/$PROGNAME.conf
       Just show what you are going to do:
       $PROGNAME -vn -c /path/to/config/$PROGNAME.conf
    EOF
}

注意在每一行的行首应该有一个真正的制表符‘t’。

在vim里,如果你的tab是4个空格,你可以用这个替换命令:

1
:s/^    /t/

命令行参数

这里是一个例子,完成了上面usage函数的用法。我从Kirk’s blog post – bash shell script to use getopts with gnu style long positional parameters得到这段代码

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
cmdline() {
    # got this idea from here:
    # http://kirk.webfinish.com/2009/10/bash-shell-script-to-use-getopts-with-gnu-style-long-positional-parameters/
    local arg=
    for arg
    do
        local delim=""
        case "$arg" in
            #translate --gnu-long-options to -g (short options)
            --config)         args="${args}-c ";;
            --pretend)        args="${args}-n ";;
            --test)           args="${args}-t ";;
            --help-config)    usage_config &amp;&amp; exit 0;;
            --help)           args="${args}-h ";;
            --verbose)        args="${args}-v ";;
            --debug)          args="${args}-x ";;
            #pass through anything else
            *) [[ "${arg:0:1}" == "-" ]] || delim="""
                args="${args}${delim}${arg}${delim} ";;
        esac
    done
    #Reset the positional parameters to the short options
    eval set -- $args
    while getopts "nvhxt:c:" OPTION
    do
         case $OPTION in
         v)
             readonly VERBOSE=1
             ;;
         h)
             usage
             exit 0
             ;;
         x)
             readonly DEBUG='-x'
             set -x
             ;;
         t)
             RUN_TESTS=$OPTARG
             verbose VINFO "Running tests"
             ;;
         c)
             readonly CONFIG_FILE=$OPTARG
             ;;
         n)
             readonly PRETEND=1
             ;;
        esac
    done
    if [[ $recursive_testing || -z $RUN_TESTS ]]; then
        [[ ! -f $CONFIG_FILE ]]
            &amp;&amp; eexit "You must provide --config file"
    fi
    return 0
}

你像这样,使用我们在头上定义的不可变的ARGS变量:

1
2
3
4
main() {
    cmdline $ARGS
}
main

单元测试

  • 在更高级的语言中很重要。
  • 使用shunit2做单元测试
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
test_config_line_paths() {
    local s='partition cpm-all, 80-90,'
    returns "/a" "config_line_paths '$s /a, '"
    returns "/a /b/c" "config_line_paths '$s /a:/b/c, '"
    returns "/a /b /c" "config_line_paths '$s   /a  :    /b : /c, '"
}
config_line_paths() {
    local partition_line="$@"
    echo $partition_line
        | csv_column 3
        | delete_spaces
        | column 1
        | colons_to_spaces
}
source /usr/bin/shunit2

这里是另一个使用df命令的例子:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
DF=df
mock_df_with_eols() {
    cat &lt;&lt;- EOF
    Filesystem           1K-blocks      Used Available Use% Mounted on
    /very/long/device/path
                         124628916  23063572 100299192  19% /
    EOF
}
test_disk_size() {
    returns 1000 "disk_size /dev/sda1"
    DF=mock_df_with_eols
    returns 124628916 "disk_size /very/long/device/path"
}
df_column() {
    local disk_device=$1
    local column=$2
    $DF $disk_device
        | grep -v 'Use%'
        | tr 'n' ' '
        | awk "{print $$column}"
}
disk_size() {
    local disk_device=$1
    df_column $disk_device 2
}

这里我有个例外,为了测试,我在全局域中声明了DF为非只读。这是因为shunit2不允许改变全局域函数。

转载请注明:爱开源 » BASH的保护性编程技巧

您必须 登录 才能发表评论!