Regex to search at beginning of line

[kakkalla]

HI all.

I search for this type of code:

<p class="list1"><span class="list1">2.</span>Mathematics disorder (315.1)</p>

In the Find panel, I have:

<p class="(list.+)"><span class="(list.*)">(.+)</span>(.+)</p>

and the replace field I have:

<div class="\1"><p class="\1"><span class="\1">\3</span></p><p class="\1">\4</p></div>

As a test, I went back before the first replacement and ran the "find" again. It found the same paragraph inside the div tag, which is not what I want.

I tried using the "^" in front of my search string, but this didn't stop it from finding the same p tag inside the div tag.

Any ideas? Thanks in advance.

[RbnJrg]

Originally Posted by kakkalla

HI all.

I search for this type of code:

<p class="list1"><span class="list1">2.</span>Mathematics disorder (315.1)</p>

In the Find panel, I have:

<p class="(list.+)"><span class="(list.*)">(.+)</span>(.+)</p>

and the replace field I have:

<div class="\1"><p class="\1"><span class="\1">\3</span></p><p class="\1">\4</p></div>

As a test, I went back before the first replacement and ran the "find" again. It found the same paragraph inside the div tag, which is not what I want.

I tried using the "^" in front of my search string, but this didn't stop it from finding the same p tag inside the div tag.

Any ideas? Thanks in advance.

Hi kakkalla;

In the "Find" field, try using the following chain:

<p class="list(.+)"><span class="list(.*)">(.+)</span>(.+)</p>

And in the "Replace" field, use:

<div class="list\1"><p class="list\1"><span class="list\1">\3</span></p><p class="list\1">\4</p></div>

But it makes little sense to use the same class for the <p> and <span> tags because both of them will apply the same format. In that case, the <span> code is not necessary (is redundant). You could use in the "Replace" field the following chain:

<div class="list\1"><p class="list\1">\3</p><p class="list\1">\4</p></div>

Regards
Rubén

[kakkalla]

Thanks for the info.

I understand what you are telling me to do but I am trying to understand the Regex syntax. Why doesn't the ^ tell the find to begin at the beginning of the line?

[DiapDealer]

Originally Posted by kakkalla

Why doesn't the ^ tell the find to begin at the beginning of the line?

By default, PCRE treats the search subject as one string. Meaning that for all intents and purposes, the contents of your html file is one, big string. Hence the ^ and $ metacharacters will only match the very beginning and the very end of that one, big string respectively. Putting PCRE in multi-line mode by prefacing your expression with (?m) will then use ^ and $ to match the beginning and the ends of lines in the manner you're expecting it to.

(?m)^<p class="(list.+)"><span class="(list.*)">(.+)</span>(.+)</p>

You could also forget multi-line mode (and ^ or $ entirely) and use a negative lookbehind to achieve a similar effect:

(?<!\>)<p class="(list.+)"><span class="(list.*)">(.+)</span>(.+)</p>

Meaning don't match the pattern if it's preceded by the closing '>' of another tag.

I highly recommend using something similar to the second method, since the common practice of indenting html code--for readability--could easily mess with your idea of what might constitute "the beginning" of a line.

[kakkalla]

THANK you so much!

I'll give this a try tomorrow. It is now 1.30 am so I have to sleep!

Thanks again for the details. I really appreciate your time in helping me.

Keep safe,

[theducks]

Originally Posted by kakkalla

THANK you so much!

I'll give this a try tomorrow. It is now 1.30 am so I have to sleep!

Thanks again for the details. I really appreciate your time in helping me.

Keep safe,

My 2 cents:
Try to use the least 'greedy' pattern match, rather than trap with a big net.

(list\d+)

You want 'list'+digits only. Not 'listing' or some other selector that starts with 'list'

[DiapDealer]

An excellent point.

[st_albert]

It could also be that "pretty print" is adding whitespace to the beginning of the line (e.g. tab?).

I did a quick experiment, and while

find:
^<p class=

did not find what I wanted, this one did:

find:
^\W*<p class=

( it also includes the previous end-of-line, though)

[kakkalla]

Thank you all for the information.

I did the (?m)^ and it worked. It didn't find the p tag inside the div tag. When I did remove the div tag from the front and the end of the para, I searched again but it did not find the p tag in its original format.

I took note of what st_albert said and Voila! it worked.

I assume the \W means white space?

Is there a nice succinct cheat sheet somewhere I can download that has all the codes used by Sigil.

I have seen this page: http://docs.python.org/2/library/re.html

but it is a bit daunting for a newbie!

Thanks again to all of you for your help!

[DiapDealer]

The whitespace preceding indented lines of html is exactly why I suggested using a negative lookbehind assertion instead of the ^ metacharacter.

[theducks]

'cheatography regex'

look for the download PDF button
The site has other useful cheat sheets (css2, html)

[kakkalla]

Thanks so much!

I remember seeing this some time back but never kept a bookmark for it. Thank you!

[Leonatus]

I had put this question into an old thread, so nobody read it. Please don't mind me repeating it here:
"I have some ebooks where, in the current of the text, appear page numbers (probably referring to the original printed version), sometimes even with hyperlink referring to the original TOC. I would like to delete them, but have no clue on regex matters. In one particular book, the numbers appear in squared brackets, such as [Pg 4]. Those numbers have up to three digits. The tags are like this: <span class="pagenum"><a class="pcalibre pcalibre1" id="Page_4">[Pg 4]</a></span>. Is there a way of removing them by one single regex command in Sigil or Calibre?"

By the way: It would be very nice, if there were a feature in Sigil to do such operations without necessity of beeing a Regex crack (I tried a lot of things, but nothing that worked only approximately). No idea if there are technical possibilities, but I think many users would be happy to be able to manipulate theír books in an easier way.

[Perkin]

If each of the pagenums have the <span class="pagenum"> and have no further embedded spans, then a simple regex would be

<span class="pagenum".+/span>

replace with nothing

you may need to insert a questionmark after the plus sign, to make it less greedy

[Leonatus]

Ah, thanks! I always thought that it must me quite simple. I had inserted a d+ after the "pagenum". That lead - I don't know why - to nothing. And I thought that I would have to include - anyhow - the [Pg 4] expression.
Well, I'll try it.