我有一个shell脚本与这段代码:

var=`hg st -R "$path"`
if [ -n "$var" ]; then
    echo $var
fi

但是条件代码总是执行,因为hg st总是打印至少一个换行符。

是否有一个简单的方法从$var中剥离空白(如PHP中的trim())?

or

有没有处理这个问题的标准方法?

我可以使用sed或AWK,但我认为有一个更优雅的解决方案来解决这个问题。


当前回答

从Bash指南的通配符部分

在参数展开中使用extglob

 #Turn on extended globbing  
shopt -s extglob  
 #Trim leading and trailing whitespace from a variable  
x=${x##+([[:space:]])}; x=${x%%+([[:space:]])}  
 #Turn off extended globbing  
shopt -u extglob  

下面是相同的函数封装在函数中(注意:需要引用传递给函数的输入字符串):

trim() {
    # Determine if 'extglob' is currently on.
    local extglobWasOff=1
    shopt extglob >/dev/null && extglobWasOff=0 
    (( extglobWasOff )) && shopt -s extglob # Turn 'extglob' on, if currently turned off.
    # Trim leading and trailing whitespace
    local var=$1
    var=${var##+([[:space:]])}
    var=${var%%+([[:space:]])}
    (( extglobWasOff )) && shopt -u extglob # If 'extglob' was off before, turn it back off.
    echo -n "$var"  # Output trimmed string.
}

用法:

string="   abc def ghi  ";
#need to quote input-string to preserve internal white-space if any
trimmed=$(trim "$string");  
echo "$trimmed";

如果我们将函数更改为在subshell中执行,我们不必担心检查extglob的当前shell选项,我们可以只设置它而不影响当前shell。这极大地简化了函数。我还更新了位置参数“就地”,所以我甚至不需要一个局部变量

trim() {
    shopt -s extglob
    set -- "${1##+([[:space:]])}"
    printf "%s" "${1%%+([[:space:]])}" 
}

so:

$ s=$'\t\n \r\tfoo  '
$ shopt -u extglob
$ shopt extglob
extglob         off
$ printf ">%q<\n" "$s" "$(trim "$s")"
>$'\t\n \r\tfoo  '<
>foo<
$ shopt extglob
extglob         off

其他回答

# Trim whitespace from both ends of specified parameter

trim () {
    read -rd '' $1 <<<"${!1}"
}

# Unit test for trim()

test_trim () {
    local foo="$1"
    trim foo
    test "$foo" = "$2"
}

test_trim hey hey &&
test_trim '  hey' hey &&
test_trim 'ho  ' ho &&
test_trim 'hey ho' 'hey ho' &&
test_trim '  hey  ho  ' 'hey  ho' &&
test_trim $'\n\n\t hey\n\t ho \t\n' $'hey\n\t ho' &&
test_trim $'\n' '' &&
test_trim '\n' '\n' &&
echo passed

数组赋值在内部字段分隔符(默认为空格/制表符/换行符)上展开参数拆分。

words=($var)
var="${words[@]}"
# Strip leading and trailing white space (new line inclusive).
trim(){
    [[ "$1" =~ [^[:space:]](.*[^[:space:]])? ]]
    printf "%s" "$BASH_REMATCH"
}

OR

# Strip leading white space (new line inclusive).
ltrim(){
    [[ "$1" =~ [^[:space:]].* ]]
    printf "%s" "$BASH_REMATCH"
}

# Strip trailing white space (new line inclusive).
rtrim(){
    [[ "$1" =~ .*[^[:space:]] ]]
    printf "%s" "$BASH_REMATCH"
}

# Strip leading and trailing white space (new line inclusive).
trim(){
    printf "%s" "$(rtrim "$(ltrim "$1")")"
}

OR

# Strip leading and trailing specified characters.  ex: str=$(trim "$str" $'\n a')
trim(){
    if [ "$2" ]; then
        trim_chrs="$2"
    else
        trim_chrs="[:space:]"
    fi

    [[ "$1" =~ ^["$trim_chrs"]*(.*[^"$trim_chrs"])["$trim_chrs"]*$ ]]
    printf "%s" "${BASH_REMATCH[1]}"
}

OR

# Strip leading specified characters.  ex: str=$(ltrim "$str" $'\n a')
ltrim(){
    if [ "$2" ]; then
        trim_chrs="$2"
    else
        trim_chrs="[:space:]"
    fi

    [[ "$1" =~ ^["$trim_chrs"]*(.*[^"$trim_chrs"]) ]]
    printf "%s" "${BASH_REMATCH[1]}"
}

# Strip trailing specified characters.  ex: str=$(rtrim "$str" $'\n a')
rtrim(){
    if [ "$2" ]; then
        trim_chrs="$2"
    else
        trim_chrs="[:space:]"
    fi

    [[ "$1" =~ ^(.*[^"$trim_chrs"])["$trim_chrs"]*$ ]]
    printf "%s" "${BASH_REMATCH[1]}"
}

# Strip leading and trailing specified characters.  ex: str=$(trim "$str" $'\n a')
trim(){
    printf "%s" "$(rtrim "$(ltrim "$1" "$2")" "$2")"
}

OR

建立在moskit的expr soulution…

# Strip leading and trailing white space (new line inclusive).
trim(){
    printf "%s" "`expr "$1" : "^[[:space:]]*\(.*[^[:space:]]\)[[:space:]]*$"`"
}

OR

# Strip leading white space (new line inclusive).
ltrim(){
    printf "%s" "`expr "$1" : "^[[:space:]]*\(.*[^[:space:]]\)"`"
}

# Strip trailing white space (new line inclusive).
rtrim(){
    printf "%s" "`expr "$1" : "^\(.*[^[:space:]]\)[[:space:]]*$"`"
}

# Strip leading and trailing white space (new line inclusive).
trim(){
    printf "%s" "$(rtrim "$(ltrim "$1")")"
}

我发现我需要从混乱的sdiff输出中添加一些代码来清理它:

sdiff -s column1.txt column2.txt | grep -F '<' | cut -f1 -d"<" > c12diff.txt 
sed -n 1'p' c12diff.txt | sed 's/ *$//g' | tr -d '\n' | tr -d '\t'

这将删除尾随空格和其他不可见字符。

我所知道的单行用例最简单的方法是:

echo "  ABC  " | sed -e 's# \+\(.\+\) \+#\1#'

工作原理:

-e启用高级正则表达式 我使用# sed,因为我不喜欢“凌乱的库”模式,如/\////\/\\\/\/ Sed希望转义大多数正则表达式控制字符,因此所有\ 否则就是^ +(.+)+$,也就是开头的空格,一组no。1,空格在最后。 所有这些都被“第一组”所取代。

因此,ABC变成了ABC。

这应该在使用sed的最新系统上得到支持。


对于标签来说,就是这样

echo "  ABC  " | sed -e 's#[\t ]\+\(.\+\)[\t ]\+#\1#'

对于多行内容,已经需要其他答案中描述的[:space:]这样的字符类,并且可能不是所有sed实现都支持。

参考资料:Sed手册