REGEX can't seem to figure it out

[JLius]

Hi all

Came about this old post:

Originally Posted by Perkin

If all of the chapters contain exactly the same

<p class="calibre10"><span class="calibre8 bold calibre11 calibre8">

and then the chapter number followed by

</span></p>

then search for (regex, minimal)

<p class="calibre10"><span class="calibre8 bold calibre11 calibre8">(\d+)</span></p>

replace

<h2>Chapter \1</h2>

I thought this could save me a lot of time cleaning up an epub document. Didn't know about regex, so I gave it a try.

I want to replace <i class="calibre2">words</i> by <em>words</em)
So I searched for <i class="calibre2">(\w+)</i> and replaced it with <em>\1</em>

That worked fine when there was only one word between > <. But when there are more: <i class="calibre2">two words</i>, my seach doesn't find them.

Does w+ only search 1 word? How do I get it to search for any number of words? Do I change the \1 in the replace with * or something else?

Thanks in advance!

JLius

[Toxaris]

That would go wrong if there are digits in the original tags. Just a question, why do you want to replace <i> by <em>? Both work perfectly fine and although <i> is officially deprecated, I don't think there is a change it will be gone for long. Not only that, there is a distinct difference between <i> and <em>. The first is a stylish indication, the second one semantic. It might be that emphasis is the wrong semantic value for the <i>.
Anyway, back to your question.

If you want to retain the <i> you can just replace <i class="calibre2"> by <i>. If not, you can try to do the following:
Search: <i class="calibre2">(.*?)</i>
Replace: <em>\1</em>

Explanation: the . selects any character. The * says 0 or more and the question mark says 0 or 1.

Regex can be confusing in the beginning, but is worth the time you need to invest. There are various helpers out there. Some here have good results with RegEx Buddy. You can also use a cheatsheet like this one: http://www.cheatography.com/davechil...r-expressions/

[DiapDealer]

\w does not mean match a "word." So if follows that \w+ does not mean "one or more words."

\w is a shortcut for a word character. Singular. Which (most commonly) translates to [A-Za-z0-9_]. A space is not a word character--which is why your regex stopped matching when it encountered the space between words. There really is no regex shortcut for matching a "word." You can find the boundaries between words, though. Some regex flavors (calibre's for example) can match the beginnings or ends of words.

Toxaris's suggestion should do what you want (unless there are any complicated, nested i tags--which isn't very likely).

You could also search for:

<(/?)i( class=".*?")?>

And replace with:

<\1em>

Which will match/alter the tags while ignoring their contents. You may have to uncheck/check any "minimal matching/greedy" option your particular search engine might have.

[JLius]

Super, many thanks you both!

[JLius]

By the way, for others that might find their way to this tread, Toxaris: your suggestion didn't work, for reasons I don't understand the <em> tags got all over my paragraphs in places they don't belong.
Diapdealer: as far as I can tell, your suggestion worked.

[Toxaris]

That should not have happened, unless the structure of the document was not in order or that there were nested tags which is very unlikely. With what program did you execute the RegEx?

[DiapDealer]

I agree. Your regex should have worked, Toxaris. I guess it could get a bit greedy depending on the flags/checkboxes that were set by default ... but then so could mine, for that matter.

[mrmikel]

It can be helpful sometimes to do things in stages, taking out an element or two at a time. It can make the search expression less complicated.

It is also helpful to set up the search expression and repeatedly find without replacing to make sure it will not go berserk. In Sigil finding what you didn't intend can be catastrophic unless you do a save just before your replace all.
It can be hard for any normal human to envision all possible things any search will grab, even for just plain characters.

In my current book, I have an ATIS. In fiddling with it, I turned satisfaction into s ATIS faction.

[JLius]

I used calibre.
Don't know what nested tags are, but I'll give an example:

<p class="calibre3"> text here <i class="calibre2"> text here </i> text here </p>

Toxaris' regex did something like this to various paragraphs:
<p></em>some text that didn't need em <em>text that needed the em</em>regular text <em> </p>

Didn't make sense to me. Maybe my first <i class="calibre2">(\w+)</i> had something to do with it.

[dgatwood]

For a gentle introduction to regular expressions:

https://developer.apple.com/library/...nfettered.html

[eschwartz]

Originally Posted by dgatwood

For a gentle introduction to regular expressions:

https://developer.apple.com/library/...nfettered.html

I would go with http://www.regular-expressions.info/

It is more general (as opposed to being focused on usage within perl/sed/grep/awk) as well as being heavily geared towards the beginner, with color-coded examples.

[Hitch]

Originally Posted by eschwartz

I would go with http://www.regular-expressions.info/

It is more general (as opposed to being focused on usage within perl/sed/grep/awk) as well as being heavily geared towards the beginner, with color-coded examples.

And don't forget the Hitch Bible: Regex Buddy. ;-)

Hitch

[Toxaris]

Originally Posted by Hitch

And don't forget the Hitch Bible: Regex Buddy. ;-)

Hitch

Already covered you in the beginning Hitch!

[Hitch]

Originally Posted by Toxaris

Already covered you in the beginning Hitch!

Sorry, Tox, should have caught that. Well, there you go: recommended twice now. ;-)

Hitch