sed(awk?)删除几乎重复的行

sed (awk?) to remove nearly duplicate lines

我有一个文件,该文件将 HTML 风格的评论与其真实文本交替显示:

<!-- Here's a first line -->
Here's a first line
<!-- Here's a second line -->
Here's a third line

如果评论除标签本身外与以下行相同,我想将其删除,否则保留:

Here's a first line
<!-- Here's a second line -->
Here's a third line

我在这里读过类似的问题,但无法推断出适合我的情况的解决方案。

你可以使用这个awk:

awk '/<!--.*?-->/{h=[=10=]; gsub(/ *(<!--|-->) */, ""); s=[=10=]; next}
      [=10=]!=s{[=10=]=h ORS [=10=]} 1' file.html
Here's a first line
<!-- Here's a second line -->
Here's a third line
sed '/^<!-- \(.*\) -->$/N;s/^<!-- \(.*\) -->\n$//'
#
#    /^<!-- \(.*\) -->$/   match an HTML comment as its own line, in which case
#                       N; add the next line to the pattern space and keep going
# 
#                         s/^<!-- \(.*\) -->\n$/     detect a comment as you
#                                                 /  described and replace it
#                                                      appropriately

如图:

$ sed '/^<!-- \(.*\) -->$/N;s/^<!-- \(.*\) -->\n$//' <<EOF
> <!-- Foo -->
> Foo
> <!-- Bar -->
> Baz
> <!-- Quux -->
> Quux
> 
> Something
> Something
> Another something
> EOF

给出:

Foo
<!-- Bar -->
Baz
Quux

Something
Something
Another something

您可能需要对此进行调整以处理缩进,但这不足为奇。您可能还想切换到 sed -r,这将要求不对括号进行转义。

这可能适合您 (GNU sed):

sed -r '$!N;/<!-- (.*) -->\n$/!P;D' file

这会比较整个文件中所有连续的行是否符合请求的条件,如果找到则不打印该对的第一行。

N.B。这迎合了连续的评论行