如何用通用正则从字符串中提取Linux绝对文件路径？

问题背景 #

之前用两个针对性的grep正则提取绝对路径：

grep -o '/media/ismail/.*'
grep -o 'file:\S*'

但这类写法依赖/media/ismail/这类固定前缀，没法通用匹配所有Linux绝对路径，需要一个能适配各种场景的正则表达式。

示例输入 #

evince /media/ismail/SSDWorking/book-collection/_Books/kids/Coping Skills.pdf
/usr/lib/libreoffice/program/soffice.bin --impress file:///home/ismail/Desktop/LibreOffice%20Impress.odp --splash-pipe=5
/usr/bin/gjs /usr/bin/com.github.johnfactotum.Foliate /media/ismail/SSDWorking/book-collection/_Books/kids/Coping Skills.epub
mpv /media/ismail/8TBRaid0/_IslamNaseed/_JunaidJamshed/Mera Dil Badal De.mp3

通用解决方案 #

用支持扩展正则的grep命令，同时兼容带空格的路径和file://前缀：

grep -Eo '(file:///)?/[^[:space:]]*(\s[^[:space:]]*)*'

正则说明

• (file:///)?：匹配可选的file:///协议前缀
• /：锚定Linux绝对路径的起始斜杠
• [^[:space:]]*：匹配连续的非空格字符
• (\s[^[:space:]]*)*：匹配路径中出现的空格+非空格字符组合，兼容带空格的文件名

执行结果 #

/media/ismail/SSDWorking/book-collection/_Books/kids/Coping Skills.pdf
file:///home/ismail/Desktop/LibreOffice%20Impress.odp
/media/ismail/SSDWorking/book-collection/_Books/kids/Coping Skills.epub
/media/ismail/8TBRaid0/_IslamNaseed/_JunaidJamshed/Mera Dil Badal De.mp3

内容的提问来源于stack exchange，提问作者Ahmad Ismail