有没有一种方法可以执行两个目录的递归比较，但仅比较（在各自位置）匹配特定文件名或文件类型谓词的文件？

例如，我想这样做的东西像

diff -r dir-a dir-b -filenames *.java, ivy.xml, build.xml

...甚至更好：

diff -r dir-a dir-b -filetype text

显然，并不是强制性使用，diff因为我认为有一个咒语，find并且-exec diff也可能会成功（我只是不知道在后一种情况下如何生成互补文件路径）。

Shell脚本 differ-r

该shellscript可以执行两个目录的递归diff，但只能（在它们各自的位置）比较与特定文件名或文件类型模式匹配的文件。

#!/bin/bash

greenvid="\0033[32m"
resetvid="\0033[0m"

if [ $# -ne 3 ]
then
 echo "Usage: compare files in two directories including subdirectories"
 echo "         $0 <source-dir> <target-dir> <pattern>"
 echo "Example: $0  subdir-1     subdir-2     \"*.txt\""
 exit
fi

cmd='for pathname do
        greenvid="\0033[32m"
        resetvid="\0033[0m"
        echo -e "${greenvid}diff \"$pathname\" \"${pathname/'\"$1\"'/'\"$2\"'}\"${resetvid}"
        diff "$pathname" "${pathname/'\"$1\"'/'\"$2\"'}"
    done'
#echo "$cmd"

find "$1" -type f -name "$3" -exec bash -c "$cmd" bash {} +

演示版

档案：

$ find -type f
./1/ett.txt
./1/two.doc
./1/t r e.txt
./1/sub/only-one.doc
./1/sub/hello.doc
./1/sub/hejsan.doc
./differ-r2
./differ-r1
./differ-r
./2/ett.txt
./2/two.doc
./2/t r e.txt
./2/sub/hello.doc
./2/sub/hejsan.doc

用法：

$ ./differ-r
Usage: compare files in two directories including subdirectories
         ./differ-r <source-dir> <target-dir> <pattern>
Example: ./differ-r  subdir-1     subdir-2     "*.txt"

正在运行differ-r：

diff当不匹配时，执行的命令行将以绿色文本打印，而输出将以默认文本（以下屏幕截图中的黑底白字）进行打印。

$ ./differ-r 1 2 "*.doc"
diff "1/two.doc" "2/two.doc"
diff "1/sub/only-one.doc" "2/sub/only-one.doc"
diff: 2/sub/only-one.doc: No such file or directory
diff "1/sub/hello.doc" "2/sub/hello.doc"
2d1
< world
diff "1/sub/hejsan.doc" "2/sub/hejsan.doc"

$ ./differ-r 1 2 "*.txt"
diff "1/ett.txt" "2/ett.txt"
2c2
< stabben
---
> farsan
diff "1/t r e.txt" "2/t r e.txt"
1c1
< t r e
---
> 3
$ 

$ ./differ-r 1 2 "*"
diff "1/ett.txt" "2/ett.txt"
2c2
< stabben
---
> farsan
diff "1/two.doc" "2/two.doc"
diff "1/t r e.txt" "2/t r e.txt"
1c1
< t r e
---
> 3
diff "1/sub/only-one.doc" "2/sub/only-one.doc"
diff: 2/sub/only-one.doc: No such file or directory
diff "1/sub/hello.doc" "2/sub/hello.doc"
2d1
< world
diff "1/sub/hejsan.doc" "2/sub/hejsan.doc"

$ ./differ-r 2 1 "*"
diff "2/ett.txt" "1/ett.txt"
2c2
< farsan
---
> stabben
diff "2/two.doc" "1/two.doc"
diff "2/t r e.txt" "1/t r e.txt"
1c1
< 3
---
> t r e
diff "2/sub/hello.doc" "1/sub/hello.doc"
1a2
> world
diff "2/sub/hejsan.doc" "1/sub/hejsan.doc"

rsync 带过滤器

如果您不需要获得描述差异的任何输出，而只知道哪些文件不同或丢失（以便rsync复制它们），则可以使用以下命令行。

rsync --filter="+ <pattern>" --filter="+ */" --filter="- *"--filter="- */"  -avcn <source directory>/ <target directory>

演示版

$ rsync --filter="+ *.doc" --filter="+ */" --filter="- *"  -avcn 1/ 2
sending incremental file list
./
sub/
sub/hello.doc
sub/only-one.doc

sent 276 bytes  received 35 bytes  622.00 bytes/sec
total size is 40  speedup is 0.13 (DRY RUN)

sent 360 bytes  received 41 bytes  802.00 bytes/sec
total size is 61  speedup is 0.15 (DRY RUN)
olle@bionic64 /media/multimed-2/test/test0/temp $ rsync --filter="+ *.txt" --filter="+ */" --filter="- *" -avcn 1/ 2
sending incremental file list
./
ett.txt
t r e.txt
sub/

sent 184 bytes  received 29 bytes  426.00 bytes/sec
total size is 21  speedup is 0.10 (DRY RUN)

如果您想要干净的输出而没有注释行和目录，则可以grep这样输出：

$ pattern="*.doc"; rsync --filter="+ $pattern" --filter="+ */" --filter="- *"  -avcn 1/ 2 | grep "${pattern/\*/.\*}"
sub/hello.doc
sub/only-one.doc

Shell脚本 rsync-diff

可以将此单行代码编写为shellscript的核心命令rsync-diff。

#!/bin/bash

LANG=C

if [ $# -ne 3 ]
then
 echo "Usage: compare files in two directories including subdirectories"
 echo "         $0 <source-dir> <target-dir> <pattern>"
 echo "Example: $0  subdir-1     subdir-2     \"*.txt\""
 exit
fi

pattern="$3"; rsync --filter="+ $pattern" --filter="+ */" --filter="- *" \
 -avcn "$1"/ "$2" | grep "${pattern//\*/.\*}" | grep -v \
  -e '/$' \
  -e '^sending incremental file list$' \
  -e '^sent.*received.*sec$' \
  -e '^total size is.*speedup.*(DRY RUN)$'

由于您提到“显然，使用diff不是强制性的”，

你这个应该做的工作合并 进行什么样的文件类型的忽视很容易配置：

此外，另一种选择是编写一个简单的脚本，该脚本将从白名单转移到黑名单，然后将黑名单与该--exclude选项一起传递给diff 。

随着外壳支持命令替换，你可以使用下面的一行代码（如已被@JammingThebBits说明）：

diff -r dir-a dir-b --exclude-from=<( \
find dir-a dir-b -type f -not \( -name '*.xml'  -or -name '*.java' \) \
| sed 's:^.*/\([^/]*\)$:\1:' \
)

它的工作原理是这样的：find搜索不感兴趣的文件，sed提取基本名（basename如果有很多文件，运行速度非常慢），然后将它们放在一个临时文件中；然后将此类文件传递给diff它，告诉它从比较中排除它们（双重排除=包含）。

如果没有命令替换，请将sed输出放入文件中，并将其显式传递给diff。

在该示例中，我仅搜索XML和JAVA文件，并通过使用OR进行分隔来根据需要进行更改。