More fun with regex / wildcards

[zelda_pinwheel]

I'm trying to automate a find and replace routine to change chapter numbers from paragraphs to headings. in other words, currently, they look like this : <p>1</p> and i want them to look like this : <h3>1</h3>

i first tried to do this with wildcards and i have no troubles finding the numbers, but unfortunately if i put <h3>?</h3> in the "replace" box, it replaces the number with "?". not helpful.

so now i'm trying regex search. i looked on the regex page here to find the code : http://doc.trolltech.com/4.6/qregexp.html#introduction

i should say, i know nothing about regex, but i'm trying to learn a bit.

this expression <p>[0-9]{1,3}</p> finds what i want, but when i replace, i have the same problem as with the wildcards...

can someone tell me, what is the regex (or wildcard !) code for "don't touch the stuff in the middle, change only the tags" ??? there must be a way to do this !

[zelda_pinwheel]

oh, also, if anyone has a good (simple) tutorial for regex, i am interested !

[HarryT]

In most regex implementations, the "matches" for expressions are represented in the "replace" string by "\1", "\2", etc, so to replace:

<p>nnn<p>

by

<h3>nnn</h3>

you'd use a search string of:

<p>[0-9]*</p>

and a replace string of:

<h3>\1</h3>

where the "\1" will contain whatever was matched by the "[0-9]*".

[zelda_pinwheel]

Originally Posted by HarryT

In most regex implementations, the "matches" for expressions are represented in the "replace" string by "\1", "\2", etc, so to replace:

<p>nnn<p>

by

<h3>nnn</h3>

you'd use a search string of:

<p>[0-9]*</p>

and a replace string of:

<h3>\1</h3>

where the "\1" will contain whatever was matched by the "[0-9]*".

no joy. i tried putting "<h3>\1</h3>" in the replace box and it put \1 in place of my chapter number.

[HarryT]

Try using a search string of:

<p>([0-9]*)</p>

and a replace string of:

<h3>\1</h3>

ie, with parentheses around the "search string" for the numbers.

[Valloric]

Originally Posted by HarryT

In most regex implementations, the "matches" for expressions are represented in the "replace" string by "\1", "\2", etc, so to replace:

<p>nnn<p>

by

<h3>nnn</h3>

you'd use a search string of:

<p>[0-9]*</p>

and a replace string of:

<h3>\1</h3>

where the "\1" will contain whatever was matched by the "[0-9]*".

Harry, I don't know what regex flavor you are using, but all the engines that I know of require parentheses for capture groups.

So like this:

<p>([0-9]*)</p>

And then:

<h3>\1</h3>

For what Zelda wants, I'd use this:

<p>(\d+)</p>

With this:

<h3>\1</h3>

[HarryT]

Originally Posted by Valloric

Harry, I don't know what regex flavor you are using, but all the engines that I know of require parentheses for capture groups.

I know. See my previous post!

[Valloric]

A good regex info site is http://www.regular-expressions.info/, and a great regex test site is Regexr. But do note that Regexr use the Perl notation for backreferences ("$#", e.g "$1", "$2"...) while Sigil uses the sed notation ("\#", e.g. "\1", "\2" ).

[Valloric]

Originally Posted by HarryT

I know. See my previous post!

That post was not there when I pressed the "Reply" button.

[HarryT]

No, I realised after I'd written it that I'd missed out the parentheses. A "senior moment" on my part.

[zelda_pinwheel]

Originally Posted by Valloric

Harry, I don't know what regex flavor you are using, but all the engines that I know of require parentheses for capture groups.

So like this:

<p>([0-9]*)</p>

And then:

<h3>\1</h3>

For what Zelda wants, I'd use this:

<p>(\d+)</p>

With this:

<h3>\1</h3>

woohoo !! thanks valloric ! that worked perfectly. now when i'm done with all my other corrections i can copy all my html into previous sigil and add in my chapter break divs with a simple find and replace on <h3>. joy.

Originally Posted by Valloric

A good regex info site is http://www.regular-expressions.info/, and a great regex test site is Regexr. But do note that Regexr use the Perl notation for backreferences ("$#", e.g "$1", "$2"...) while Sigil uses the sed notation ("\#", e.g. "\1", "\2" ).

and thank you for that, too. i'm sure i'll have plenty of other expressions to find.

[gwynevans]

Originally Posted by zelda_pinwheel

i'm sure i'll have plenty of other expressions to find.

If you're running Windows, the Regex Coach (donationware) can be a useful utility to have around.

[zelda_pinwheel]

Originally Posted by gwynevans

If you're running Windows, the Regex Coach (donationware) can be a useful utility to have around.

thank you ! i am indeed running windows, and a coach sounds like just what i need. i'll take a look.

[charleski]

Be sure to keep the 'minimal matching' box checked whenever you use regexes (unless you're sure you don't want it). Sigil's support for non-greedy matching is very admirable and makes all the difference.

If you're experimenting with more complex regexes it's always a good idea to address the problem in stages, and save the book under a new name after each replace operation so you can backtrack if something goes wrong.

[zelda_pinwheel]

thanks charleski, that is very sound advice, especially as i have already gotten myself into trouble doing a "replace all" with the wrong replacement... (and that was *before* i started to tackle regex searches).