使用cmp命令结合sed处理后文件比对不一致问题排查

问题描述 #

我写了一个Bash脚本，用来检查今日和昨日的文件差异：先用sed把文件里的during_schedule=YES;替换成during_schedule=NO;，再用cmp命令比对处理后的内容。但明明确认处理后两者内容一致，脚本却一直显示文件不同。原脚本如下：

#!/bin/bash

todayfile=$(date +%d%m%Y)
yesterdayfile=$(date -d "yesterday" +%d%m%Y)


echo "----- POLICY -----"
file1_in=$(sed 's|during_schedule=YES;|during_schedule=NO;|' /opt/custom_scripts/arch/policy_$todayfile.txt)
file2_in=$(sed 's|during_schedule=YES;|during_schedule=NO;|' /opt/custom_scripts/arch/policy_$yesterdayfile.txt)

file1="/opt/custom_scripts/arch/policy_$todayfile.txt"
file2="/opt/custom_scripts/arch/policy_$yesterdayfile.txt"

if cmp -s "$file1_in" "$file2_in"; then
    printf 'INFO: The file "%s" is the same as "%s"\n' "$file1" "$file2"
else
    printf 'ERROR: The file "%s" is different from "%s"\n' "$file1" "$file2"
fi

问题原因 #

核心错误在于：cmp命令的作用是比对文件，但你传给它的$file1_in和$file2_in是sed处理后输出的文本内容，而非文件名。当cmp尝试把这些文本内容当作文件名去打开时，必然找不到对应文件，因此直接返回“内容不同”的结果。

解决方案 #

下面提供三种可行的修正方法：

方法1：用diff结合进程比对（无需临时文件）

利用Bash的进程替换<(command)，将sed的输出当作临时文件提供给diff，通过退出码判断内容是否一致：

#!/bin/bash

todayfile=$(date +%d%m%Y)
yesterdayfile=$(date -d "yesterday" +%d%m%Y)

echo "----- POLICY -----"
file1="/opt/custom_scripts/arch/policy_$todayfile.txt"
file2="/opt/custom_scripts/arch/policy_$yesterdayfile.txt"

if diff -q <(sed 's|during_schedule=YES;|during_schedule=NO;|' "$file1") <(sed 's|during_schedule=YES;|during_schedule=NO;|' "$file2") >/dev/null; then
    printf 'INFO: The file "%s" is the same as "%s"\n' "$file1" "$file2"
else
    printf 'ERROR: The file "%s" is different from "%s"\n' "$file1" "$file2"
fi

方法2：写入临时文件后用cmp比对

如果坚持使用cmp，可以先把sed处理后的内容写入临时文件，再比对临时文件：

#!/bin/bash

todayfile=$(date +%d%m%Y)
yesterdayfile=$(date -d "yesterday" +%d%m%Y)

echo "----- POLICY -----"
file1="/opt/custom_scripts/arch/policy_$todayfile.txt"
file2="/opt/custom_scripts/arch/policy_$yesterdayfile.txt"

# 创建临时文件
temp1=$(mktemp)
temp2=$(mktemp)

# 写入处理后的内容
sed 's|during_schedule=YES;|during_schedule=NO;|' "$file1" > "$temp1"
sed 's|during_schedule=YES;|during_schedule=NO;|' "$file2" > "$temp2"

# 比对临时文件
if cmp -s "$temp1" "$temp2"; then
    printf 'INFO: The file "%s" is the same as "%s"\n' "$file1" "$file2"
else
    printf 'ERROR: The file "%s" is different from "%s"\n' "$file1" "$file2"
fi

# 清理临时文件
rm -f "$temp1" "$temp2"

方法3：直接字符串比较（适合小文件）

如果文件内容不大，可以直接用Bash的字符串相等判断来比对变量内容：

#!/bin/bash

todayfile=$(date +%d%m%Y)
yesterdayfile=$(date -d "yesterday" +%d%m%Y)

echo "----- POLICY -----"
file1="/opt/custom_scripts/arch/policy_$todayfile.txt"
file2="/opt/custom_scripts/arch/policy_$yesterdayfile.txt"

file1_in=$(sed 's|during_schedule=YES;|during_schedule=NO;|' "$file1")
file2_in=$(sed 's|during_schedule=YES;|during_schedule=NO;|' "$file2")

if [ "$file1_in" = "$file2_in" ]; then
    printf 'INFO: The file "%s" is the same as "%s"\n' "$file1" "$file2"
else
    printf 'ERROR: The file "%s" is different from "%s"\n' "$file1" "$file2"
fi

内容的提问来源于stack exchange，提问作者majan