Regex search and replace in bulk metadata help

[slantybard]

I am trying to remove all html formatting from the comments in my library metadata so that when I make catalogs, the html span, div, font, stuff doesn't make the resulting catalog epub look messy.

I tried the "remove" formatting button individually, but that doesn't remove the html code from the comments section.

Then, after searching the forum, I found a pretty good, but not perfect solution, where using the S&R under the bulk metadata edit function, I used the search regex:

<[^<>]*>

and replaced it with nothing. This successfully removed 99% of the formatting html code littering my comments metadata. Unfortunately, the <br> tags were not removed and since these often show up in the comments metadata within sentences, the comments still get messed up. I have tried S&R with <br> and <\s*?[^>(br)]+\s*?>, but the <br> tags still show up.

My regex skills are not the best nor the worst, but I was wondering if anyone has any ideas that could help me remove the <br> tags in bulk?

[eschwartz]

I think your regex is fine...

calibre reformats the comments, putting them into a final div IIRC, and if you have multi-paragraph comments those need to be divided up somehow.

Paragraph == anytime there is a line-break.
In HTML mode line-breaks are collapsed into whitespace and paragraphs respect the <p> tags, but after nuking the tags if you have line-breaks calibre will do it's best to recreate a basic structure, and probably inserted hard line-breaks where it thought () you wanted them.

[slantybard]

eschwartz, your reply makes total sense. However, the "line breaks" not having a <br> tag in the html comments are not <br>'d by calibre during the S&R process. This only occurs with the comments that already contain <br> tags interspersed within the sentences from the downloaded metadata. This means that calibre must be ignoring the <br> tags when searched from the edit metadata dialog.

I'm super happy that the S&R removes all the other html junk and will live with manually editing out the <br> as necessary, I just thought that I was doing something wrong or there was an easy answer I was missing.

[theducks]

Originally Posted by slantybard

eschwartz, your reply makes total sense. However, the "line breaks" not having a <br> tag in the html comments are not <br>'d by calibre during the S&R process. This only occurs with the comments that already contain <br> tags interspersed within the sentences from the downloaded metadata. This means that calibre must be ignoring the <br> tags when searched from the edit metadata dialog.

I'm super happy that the S&R removes all the other html junk and will live with manually editing out the <br> as necessary, I just thought that I was doing something wrong or there was an easy answer I was missing.

<br /> removal needs care
choices are replace with:
a space (long line)
replace with the parent tag pair reversed: </p> <p>, </h3> <h3> ... (multiple similar styled lines)

[theducks]

You might avoid this in the future by making a metadat download setting change

[eschwartz]

First:: I *thought* calibre would turn hard line breaks into <br>'s, but now I am not so sure...


Anyway. So I entered this into the comments column in an empty book (test library ):

<p>I think your regex is fine...
<br><br>
calibre reformats the comments, putting them into a final div IIRC, and if you have multi-paragraph comments those need to be divided up somehow.
<br><br>
:bulb2: Paragraph == anytime there is a line-break.
<br><br>
In HTML mode line-breaks are collapsed into whitespace and paragraphs respect the &lt;p&gt; tags, but after nuking the tags if you have line-breaks calibre will do it's best to recreate a basic structure, and probably inserted hard line-breaks where it thought (:rolleyes:) you wanted them.</p>

Entered Bulk S&R
Search:

<br>

Replace: one standard space

Result:

<p>I think your regex is fine...
  
calibre reformats the comments, putting them into a final div IIRC, and if you have multi-paragraph comments those need to be divided up somehow.
  
:bulb2: Paragraph == anytime there is a line-break.
  
In HTML mode line-breaks are collapsed into whitespace and paragraphs respect the &lt;p&gt; tags, but after nuking the tags if you have line-breaks calibre will do it's best to recreate a basic structure, and probably inserted hard line-breaks where it thought (:rolleyes:) you wanted them.</p>

No <br>'s in sight.

[slantybard]

Originally Posted by theducks

You might avoid this in the future by making a metadat download setting change

Yes, I have indeed made that change to prevent this in the future. I may end up re-downloading the comments as needed to fix this issue. Thanks.

[slantybard]

Originally Posted by eschwartz

No <br>'s in sight.

Interesting since when I try the same S&R, the <br>'s are left alone. I am going to update my calibre version from 2.24 to the latest and see if the behaviour changes.