行内的sed替换模式

Question

How can I substitute characters only within a specific pattern, preferably in sed but awk or otherwise if there's an easier option? I would like to replace spaces in my html h3 ids with hyphens (-), but I don't want it to hyphen the entire line.

Eg, in my foo.html:

<p>This is a paragraph which likes its spaces.</p>

<h3 id="No spaces in this id please">Keep spaces in this title</h3>

<p>Here's another paragraph with spaces.</p>

<h3 id="Another id that should be spaceless">Spaces please!</h3>

<p>Yes I would like extra noodles in my soup.</p>

What I want are h3s like this:

<h3 id="Another-id-that-should-be-spaceless">Spaces please!</h3>

I've tried

sed -e "/^<h3 id=\"/,/\">/s/ /-/g;" <foo.html >bar.html

But this greedily adds hyphens to lines (2nd p) and parts (h3 content) which shouldn't have hyphens! Bar.html:

<p>This is a paragraph which likes its spaces.</p>

<h3-id="No-spaces-in-this-id-please">Keep-spaces-in-this-title</h3>

<p>Here's-another-paragraph-with-spaces.</p>

<h3-id="Another-id-that-should-be-spaceless">Spaces-please!</h3>

<p>Yes I would like extra noodles in my soup.</p>

Note I'm using GNU sed. Thanks!

Answer 1

This sed replace one space at a time in the id value of h3 tags. When substitution succeeds, the t command loops to :a label to search for remaining spaces to replace:

sed -e ':a;s/\(<h3[^>]*id="[^"> ]*\) \(.*\)/\1-\2/;ta;' < foo.html > bar.html