Re: Query replace regex with 2 alternatives

[Jambunathan K]

Dan Espen <despen@verizon.net> writes:

> Could use some help on query/replace/regex.
>
> I have an html file full of &lt; and &gt;.
> I want to replace only some of the pairs with "[" and "]".
>
> I figured out the match string:
>
> "\\(&lt;\\|&gt;\\)
>
> (typed as)
>
> "\(&lt;\|&gt;\)
>
> but when it comes to the replacement, I'm not clear on how to say,
> first match gets [ and second match gets ].
> I believe emacs can do it but I don't see it documented.
> I see references to \1 \2 but not in the replace string.

rx-to-string is the easiest way to build such an regexp.

C-h f rx

Do this 

1. M-x ielm RET
2. Copy the below regexp to the prompt

    (rx-to-string '(and (group-n 1 "&lt;") 
                        (group-n 2 (minimal-match
                                    (zero-or-more anything)))
                        (group-n 3 "&gt;")))

   Here is a sample session.

    ,----
    | ELISP> (rx-to-string '(and (group-n 1 "&lt;") 
    |               (group-n 2 (minimal-match
    |                           (zero-or-more anything)))
    |               (group-n 3 "&gt;")))
    | "\\(?:\\(?1:&lt;\\)\\(?2:\\(?:.\\|\n\\)*?\\)\\(?3:&gt;\\)\\)"
    `----

3. C-x b file.html
4. M-x reb-change-syntax RET read RET
5. M-x re-builder RET
6. Copy paste the above regexp in to *RE-Builder* buffer
7. You will see the various components highlighted in HTML buffer
8. M-x reb-change-syntax RET string RET
9. You will see the above regexp changed from read syntax to string
   syntax.  Something like.  (Yes, the regexp is on two lines)

"\(?:\(?1:&lt;\)\(?2:\(?:.\|
\)*?\)\(?3:&gt;\)\)"

10. C-M-% 
    Copy the above regexp without surrounding double quotes RET
    <\2> RET

You are done.



--