我应该如何使用sed来删除特定的字符串并允许重复使用更多的字符？

Question

我生成了一个文件列表，其中有17417行，如下所示：

./usr
./usr/share
./usr/share/mime-info
./usr/share/mime-info/libreoffice7.0.mime
./usr/share/mime-info/libreoffice7.0.keys
./usr/share/appdata
./usr/share/appdata/libreoffice7.0-writer.appdata.xml
./usr/share/appdata/org.libreoffice7.0.kde.metainfo.xml
./usr/share/appdata/libreoffice7.0-draw.appdata.xml
./usr/share/appdata/libreoffice7.0-impress.appdata.xml
./usr/share/appdata/libreoffice7.0-base.appdata.xml
./usr/share/appdata/libreoffice7.0-calc.appdata.xml
./usr/share/applications
./usr/share/applications/libreoffice7.0-xsltfilter.desktop
./usr/share/applications/libreoffice7.0-writer.desktop
./usr/share/applications/libreoffice7.0-base.desktop
./usr/share/applications/libreoffice7.0-math.desktop
./usr/share/applications/libreoffice7.0-startcenter.desktop
./usr/share/applications/libreoffice7.0-calc.desktop
./usr/share/applications/libreoffice7.0-draw.desktop
./usr/share/applications/libreoffice7.0-impress.desktop
./usr/share/icons
./usr/share/icons/gnome
./usr/share/icons/gnome/16x16
./usr/share/icons/gnome/16x16/mimetypes
./usr/share/icons/gnome/16x16/mimetypes/libreoffice7.0-oasis-formula.png

问题是，我想删除以下几行：

./usr
./usr/share
./usr/share/mime-info
./usr/share/appdata
./usr/share/applications
./usr/share/icons
./usr/share/icons/gnome
./usr/share/icons/gnome/16x16
./usr/share/icons/gnome/16x16/mimetypes

还有“。”一开始，结果必须是：

/usr/share/mime-info/libreoffice7.0.mime
/usr/share/mime-info/libreoffice7.0.keys
/usr/share/appdata/libreoffice7.0-writer.appdata.xml
/usr/share/appdata/org.libreoffice7.0.kde.metainfo.xml
/usr/share/appdata/libreoffice7.0-draw.appdata.xml
/usr/share/appdata/libreoffice7.0-impress.appdata.xml
/usr/share/appdata/libreoffice7.0-base.appdata.xml
/usr/share/appdata/libreoffice7.0-calc.appdata.xml
/usr/share/applications/libreoffice7.0-xsltfilter.desktop
/usr/share/applications/libreoffice7.0-writer.desktop
/usr/share/applications/libreoffice7.0-base.desktop
/usr/share/applications/libreoffice7.0-math.desktop
/usr/share/applications/libreoffice7.0-startcenter.desktop
/usr/share/applications/libreoffice7.0-calc.desktop
/usr/share/applications/libreoffice7.0-draw.desktop
/usr/share/applications/libreoffice7.0-impress.desktop
/usr/share/icons/gnome/16x16/mimetypes/libreoffice7.0-oasis-formula.png

这有可能使用sed吗？或者使用另一种工具更实用

Answer 1

假设您想删除包含其他路径名的行，请尝试：

sort -r list.txt | awk '                # sort the list in the reverse order
{
    sub("^\\.", "")                     # remove leading dot
    s = prev; sub("/[^/]+$", "", s)     # remove the rightmost slash and following characters
    if (s != $0) print                  # if s != $0, it means $0 is not a substring of the previous line
    prev = $0                           # keep $0 for the next line
}'

结果：

/usr/share/mime-info/libreoffice7.0.mime
/usr/share/mime-info/libreoffice7.0.keys
/usr/share/icons/gnome/16x16/mimetypes/libreoffice7.0-oasis-formula.png
/usr/share/applications/libreoffice7.0-xsltfilter.desktop
/usr/share/applications/libreoffice7.0-writer.desktop
/usr/share/applications/libreoffice7.0-startcenter.desktop
/usr/share/applications/libreoffice7.0-math.desktop
/usr/share/applications/libreoffice7.0-impress.desktop
/usr/share/applications/libreoffice7.0-draw.desktop
/usr/share/applications/libreoffice7.0-calc.desktop
/usr/share/applications/libreoffice7.0-base.desktop
/usr/share/appdata/org.libreoffice7.0.kde.metainfo.xml
/usr/share/appdata/libreoffice7.0-writer.appdata.xml
/usr/share/appdata/libreoffice7.0-impress.appdata.xml
/usr/share/appdata/libreoffice7.0-draw.appdata.xml
/usr/share/appdata/libreoffice7.0-calc.appdata.xml
/usr/share/appdata/libreoffice7.0-base.appdata.xml

Answer 2

使用文件名list中的列表，您可以：

sed -n 's/^[.]//;/\/.*[._].*$/p' list

其中：

• sed -n禁止打印模式空间；then
• s/^[.]//是替换形式，它简单地从每一行中删除第一个字符'.'；then
• /\/.*[._].*$/p匹配在最后一个'/'之后包含'.'或'_' (可选的)的行，p会导致该行被打印。

示例使用/输出

$ sed -n 's/^[.]//;/\/.*[._].*$/p' list
/usr/share/mime-info/libreoffice7.0.mime
/usr/share/mime-info/libreoffice7.0.keys
/usr/share/appdata/libreoffice7.0-writer.appdata.xml
/usr/share/appdata/org.libreoffice7.0.kde.metainfo.xml
/usr/share/appdata/libreoffice7.0-draw.appdata.xml
/usr/share/appdata/libreoffice7.0-impress.appdata.xml
/usr/share/appdata/libreoffice7.0-base.appdata.xml
/usr/share/appdata/libreoffice7.0-calc.appdata.xml
/usr/share/applications/libreoffice7.0-xsltfilter.desktop
/usr/share/applications/libreoffice7.0-writer.desktop
/usr/share/applications/libreoffice7.0-base.desktop
/usr/share/applications/libreoffice7.0-math.desktop
/usr/share/applications/libreoffice7.0-startcenter.desktop
/usr/share/applications/libreoffice7.0-calc.desktop
/usr/share/applications/libreoffice7.0-draw.desktop
/usr/share/applications/libreoffice7.0-impress.desktop
/usr/share/icons/gnome/16x16/mimetypes/libreoffice7.0-oasis-formula.png

注意，如果没有GNU sed，就可以使用';'链接表达式：

sed -n -e 's/^[.]//' -e '/\/.*[._].*$/p' list