正则表达的中的存储器

wgqjjq

在原版的《linux shell s by example》中

$s/\([0o]ccur\)ence/\1rence/

解释：The editor searches for the entire string occurence or Occurence (note: the words are misspelled), and if found, the pattern portion enclosed in parentheses is tagged (i.e., either occur or Occur is tagged). Since this is the first pattern tagged, it is called tag 1. The pattern is stored in a memory register called register 1. On the replacement side, the contents of the register are replaced for \1 and the rest of the word rence, is appended to it. We started with "occurence" and ended up with "occurrence."

说是标签1中被保存在寄存器1中，把\1中的内容替换，occurence开始，以occurrence结束，谁能详细解释下各部分的用法？
还有这个

s/\(square\) and \(fair\)/\2 and \1/

标签2的寄存器存了fair，标签1寄存器存了square，\2为fair，\1为fair，匹配square and fair，不知道这样了解是否对？

galford433

上面\1 = square ; \2 = fair。寄存器最多可以有9个，依次对应相应的\(*\)里面的*。
这些在shell下面跑一次就有印象了

wgqjjq

谢谢，发现这种正则表达式都有一个规律以s/开头，中间用/连接，比如s/\(square\) and \(fair\)/\2 and \1/，有谁能详细总结下用法。

jinl

s/part/repl/
是sed里的命令,正则表达式可以在其中使用而已

寂寞烈火

看置顶十二篇之sed

walkerxk

s///是sed的应用，可以用正则，也可以不用，正则也可以是/re/d，正则和s///没有关系。建议去看看man sed和info sed。