在POSIX shell/ bin / sh中处理变量作用域在while循环中使用适当的管道？

如何解决在POSIX shell/ bin / sh中处理变量作用域在while循环中使用适当的管道？

我偶然发现了这个问题：

• Bash Script Variable Scope Issue

管道的右侧在子外壳内部运行。子外壳内部设置的任何变量都不会在父外壳中设置。要解决此问题，请使用重定向而不是管道：

因此，这是一个示例脚本：

NUMCOUNT=0

testlines="Lorem ipsum dolor sit amet,consectetur adipiscing elit.
Duis sem sapien,congue sed lacus nec,scelerisque bibendum arcu.
Pellentesque placerat dapibus erat eu iaculis.
Maecenas vitae blandit quam.
Suspendisse eu aliquet nunc."

## this does not work - because NUMCOUNT is not preserved in a subshell - so the final line echoes $NUMCOUNT as 0,which it should not be
#echo "$testlines" | while read line; do
#  if echo "$line" | grep -m 1 -q 'que'; then
#    NUMCOUNT=$((NUMCOUNT+1))
#    echo "found match nr. $NUMCOUNT for substring 'que' in line: $line"
#  fi
#done

##this works in bash - but for dash (/bin/sh): "Syntax error: redirection unexpected"
#while read line; do
#  if echo "$line" | grep -m 1 -q 'que'; then
#    NUMCOUNT=$((NUMCOUNT+1))
#    echo "found match nr. $NUMCOUNT for substring 'que' in line: $line"
#  fi
#done <<< "$testlines"

# this works in bash - but for dash (/bin/sh): "Syntax error: redirection unexpected"
while read line; do
  if echo "$line" | grep -m 1 -q 'que'; then
    NUMCOUNT=$((NUMCOUNT+1))
    echo "found match nr. $NUMCOUNT for substring 'que' in line: $line"
  fi
done < <(echo "$testlines")

echo "Script done - found $NUMCOUNT matches"

如果我在bash中运行，一切正常：

$ bash test.sh
found match nr. 1 for substring 'que' in line: Duis sem sapien,scelerisque bibendum arcu.
found match nr. 2 for substring 'que' in line: Pellentesque placerat dapibus erat eu iaculis.
found match nr. 3 for substring 'que' in line: Suspendisse eu aliquet nunc.
Script done - found 3 matches

如果我在dash中运行它，它将失败：

$ /bin/sh test.sh
test.sh: 31: test.sh: Syntax error: redirection unexpected

如何在POSIX /bin/sh shell中正确实现此循环，以便在脚本的最后一行保留$ NUMCOUNT？

解决方法

如何在POSIX / bin / sh shell中正确实现此循环，以便在脚本的最后一行保留$ NUMCOUNT？

在posix中，将fifo与用于传输数据的后台进程一起使用。但实际上，就您而言，只需将数据保存到文件中即可。

tmp=$(mktemp)
printf "%s\n" "$testlines" > "$tmp" 
while IFS= read -r line; do
    if printf "%s\n" "$line" | grep -m 1 -q 'que'; then
        ...
    fi
done < "$tmp"
rm "$tmp"

或在后台进程中使用fifo： f

ifo=$(mktemp -u)
mkfifo "$fifo"
printf "%s\n" "$testlines" > "$fifo" &
while IFS= read -r line; do
  if echo "$line" | grep -m 1 -q 'que'; then
    NUMCOUNT=$((NUMCOUNT+1))
    echo "found match nr. $NUMCOUNT for substring 'que' in line: $line"
  fi
done < "$fifo"
rm "$fifo"

echo "Script done - found $NUMCOUNT matches"

您也可以将其保存在变量中，进行预处理以提取NUMCOUNT并打印其余内容，而不是在当前shell中运行循环。

tmp=$(
    printf "%s\n" "$testlines" |
    awk -v q="'" '
    /que/{
        ++NUMCOUNT
        print "found match nr. " NUMCOUNT " for substring "q"que"q" in line: "$0
    }
    END{
        # extra last line to output NUMCOUNT
        print NUMCOUNT
    }'
)
# extract NUMCOUNT - last line
NUMCOUNT=$(printf "%s\n" "$tmp" | tail -n 1)
# extract everything except NUMCOUNT - ie. remove last line
tmp=$(printf "%s\n" "$tmp" | head -n +1)
# output the actual output
printf "%s\n" "$tmp"
# output your trailing line
printf "Script done - found %d matches\n" "$NUMCOUNT"

或者您可以继续在子Shell中执行。

printf "%s\n" "$testlines" | {
   while IFS= read -r line; do
        ...
   done
   echo "Script done - found $NUMCOUNT matches"
   # go on
}

注意：bashfaq How can I read a file (data stream,variable) line-by-line (and/or field-by-field)和unix.stackexchange Why is printf better than echo?