how are nested+contradictory CSS dealt with ?

cybmole · 10-20-2011, 03:09 AM

a line from a calibre conversion with an overkill of styles

Code:

<h1 class="calibre11" id="calibre_pb_34"><span class="calibre12 bold"><strong class="calibre13"><span class="calibre14 calibre15">16</span> KARMIC KILTER</strong></span></h1>

in the css, calibre11 thru 15 each have different font size settings. which ones actually get applied to the text ?

or to put it another way - how much of that code could be deleted with no side effects ?

and do strong and font-weight: bold both do the same thing ?
.calibre11 {
display: block;
font-size: 1.2em;
font-weight: bold;
line-height: 1.2;
margin-bottom: 0.67em;
margin-left: 0;
margin-right: 0;
margin-top: 0.67em;
text-align: right
}
.calibre12 {
font-size: 0.8m;
line-height: 1.2
}
.calibre13 {
font-weight: bolder;
line-height: 1.2
}
.calibre14 {
line-height: 1.2
}
.calibre15 {
color: #BDBDBD;
font-size: 1.2em;
line-height: 1.2
}

final challenge question - a regex which would remove all the unnecessary stuff from that line & from all similar <h1 lines ?

jackie_w · 10-20-2011, 01:38 PM

At the risk of woefully exposing my limited css knowledge, I believe the following to be true:

With nested html elements, the various css attributes are not all treated the same.

The font-size attribute is multiplicative, at least when using em units.
Therefore, in your example, I would expect the text 'KARMIC KILTER' to render at 1.2 x 0.8em, i.e 0.96em.
Similarly, I would expect the text '16' to render at 1.2 x 0.8 x 1.2em, i.e. 1.152em
I believe nested margins are additive not multiplicative.
Nested attributes like color, font-family are neither.
Using your example, if .calibre11 had color: red and .calibre15 has color: #BDBDBD
'KARMIC KILTER' would be in red, '16' would be in #BDBDBD (some shade of grey).

Quote:

do strong and font-weight: bold both do the same thing?

They appear to to the same thing on an ereader but whether they necessarily do in a web browser, I don't know.
 and seem to do the same thing.

Quote:

how much of that code could be deleted with no side effects

In your example I don't think is doing anything useful. I think line-height only works on block elements e.g. <body>, <h?>, , <div>, <blockquote>.

, , , , are inline elements, i.e. they work on a selected portion of a block element.

Time to read up on html and css, I think, cybmole

It will help a lot. You might find this site useful, it lets you play around typing in sample html/css and immediately viewing the results.

Re: the Regex, I'll leave that for someone else

cybmole · 10-20-2011, 02:00 PM

i did a little experimentation in sigil & + a little googling-
I got the impression that with most attributes the outer one takes preference . so font size is taken from calibre 11 only.

could well be though that different programs / readers interpret differently

i would strip or comment out colors anyway for e-reader as they are not supported.

thanks for the tutorial link.

Serpentine · 10-20-2011, 11:33 PM

Amusing you want to catch all h1/2 etc:

find : <(h\d)[^<>]*>(.+)</\1>
replace : <\1>\2</\1>

If you are cleaning up books or something, it's often best to work from an HTMLZ export - In the format options you can select how CSS is handled, the 'tag' option will not add any extra stuff, tho I cant remember if it preserves bold/italic etc formatting - i.e placing tags..

cybmole · 10-21-2011, 01:16 AM

Quote:

Originally Posted by Serpentine

Amusing you want to catch all h1/2 etc:

find : <(h\d)[^<>]*>(.+)</\1>
replace : <\1>\2</\1>

If you are cleaning up books or something, it's often best to work from an HTMLZ export - In the format options you can select how CSS is handled, the 'tag' option will not add any extra stuff, tho I cant remember if it preserves bold/italic etc formatting - i.e placing tags..

impressive, I wimped out & did multiple simpler reductions.

but I think your code misses the fact that there's both a chapter number and a chapter title within the dross, both of which should ideally be salvaged.

DoctorOhh · 10-21-2011, 01:21 AM

Quote:

Originally Posted by cybmole

final challenge question - a regex which would remove all the unnecessary stuff from that line & from all similar <h1 lines ?

I do simple search & replace in Sigil, no complex regex usually needed.

Agama · 10-21-2011, 02:19 AM

Quote:

Originally Posted by cybmole

I got the impression that with most attributes the outer one takes preference . so font size is taken from calibre 11 only.

As jackie_w said, font sizes in em are multiplicative. I have attached an example to show this in action. Change the font-size in style "calibre", (e.g. to font-size:1.5em), and you will see that all the other styles get mulitpliied. Similarly you can see that the within the is multiplying the "calibre4" font-sizes and the "calibre" size.

(Save/Rename the attached css.txt file as css.html to try it out).

Dopedangel · 10-21-2011, 06:42 AM

What kind of ebook needs so many styles. I mostly read fiction so I just convert to text with textile. Then convert to epub from that you can add your custom styles for headings paragraphs italics while converting from text to epub if you are so inclined.

cybmole · 10-21-2011, 07:38 AM

Quote:

Originally Posted by dwanthny

I do simple search & replace in Sigil, no complex regex usually needed.

so do I, I take stuff out a bit at a time & check for unexpected damage; but sometimes I like to ponder how/if things can be done in one pass.

I see going to text & back to epub as an absolute last resort.

if the source is epub I prefer to retain the original styling, except for tweaking font sizes & centering headers. Sometimes that leads to wanting to clean up the code.

I prefer to have everything in the 0.8 - 1.2 em font size range. anything smaller is hard to read & anything larger looks like overkill on the actual e-reader.

theducks · 10-21-2011, 08:41 AM

nested spans (or divs) are problematic to replace/remove with REGEX (unless I have been missing a trick

) in that they must match the correct Closing Tag, not the first occurrence of a closing tag.

BTW a simple What If I took it out test.
/* just comment out the selector in the style sheet */
and see what gets ugly

you can always remove the comment marks.

Unlike CSS errors, undefined styles do not seem to cause problems with ADE on my PEz

Serpentine · 10-21-2011, 11:15 AM

Quote:

Originally Posted by cybmole

but I think your code misses the fact that there's both a chapter number and a chapter title within the dross, both of which should ideally be salvaged.

cybmole · 10-21-2011, 11:39 AM

Quote:

Originally Posted by Serpentine

If you know the book you're working with also does not contain any 'useful' formatting in the spans,

but if I've not already read the book, I don't know that. so I limit the amount of advance tweaking that I do, to improving header layouts, paragraph spacings & indents.

I used to always regenerate TOC, but with latest sigil I sometimes leave it mostly as-is, just edit as needed.

I find that many novels slip in occasional format changes e,g, for signs, diary/letter extracts, newspaper clippings ... to make than stand out from the general text, & I quite like those.

If, when reading, I see anything else that bugs me, I make a note of it using the reader notes facility then do an additional post-read edit. I do that also for annoying typos.

JSWolf · 10-21-2011, 11:41 AM

Quote:

Originally Posted by Dopedangel

What kind of ebook needs so many styles. I mostly read fiction so I just convert to text with textile. Then convert to epub from that you can add your custom styles for headings paragraphs italics while converting from text to epub if you are so inclined.

Maybe one built from an MS Word document.

Serpentine · 10-21-2011, 11:55 AM

Quote:

Originally Posted by cybmole

but if I've not already read the book, I don't know that. so I limit the amount of advance tweaking that I do, to improving header layouts, paragraph spacings & indents.

Yeah, I glance through the file first, looking for problems. If there are any issues I just go and clean it up beforehand. You can usually make a good guestimate of the classes from just reading the CSS and looking for font styling.

Quote:

Originally Posted by cybmole

If, when reading, I see anything else that bugs me, I make a note of it using the reader notes facility then do an additional post-read edit. I do that also for annoying typos.

Yeah, I've just finished making 84 edits on an obviously pdf converted retail book - sent the changes back to the publisher :/.

Scifi conversions nearly always have their letters/signs etc messed up during conversion, especially when print books use a teletype style for them. So I spend a lot of time putting in <tt>'s. Yay.

10-20-2011, 03:09 AM	#1
cybmole Wizard Posts: 3,720 Karma: 1759970 Join Date: Sep 2010 Device: none	how are nested+contradictory CSS dealt with ? a line from a calibre conversion with an overkill of styles Code: <h1 class="calibre11" id="calibre_pb_34"><span class="calibre12 bold"><strong class="calibre13"><span class="calibre14 calibre15">16</span> KARMIC KILTER</strong></span></h1> in the css, calibre11 thru 15 each have different font size settings. which ones actually get applied to the text ? or to put it another way - how much of that code could be deleted with no side effects ? and do strong and font-weight: bold both do the same thing ? .calibre11 { display: block; font-size: 1.2em; font-weight: bold; line-height: 1.2; margin-bottom: 0.67em; margin-left: 0; margin-right: 0; margin-top: 0.67em; text-align: right } .calibre12 { font-size: 0.8m; line-height: 1.2 } .calibre13 { font-weight: bolder; line-height: 1.2 } .calibre14 { line-height: 1.2 } .calibre15 { color: #BDBDBD; font-size: 1.2em; line-height: 1.2 } final challenge question - a regex which would remove all the unnecessary stuff from that line & from all similar <h1 lines ? Last edited by cybmole; 10-20-2011 at 03:13 AM.

10-20-2011, 11:33 PM	#4
Serpentine Evangelist Posts: 416 Karma: 1045911 Join Date: Sep 2011 Location: Cape Town, South Africa Device: Kindle 3	Amusing you want to catch all h1/2 etc: find : <(h\d)[^<>]*>(.+)</\1> replace : <\1>\2</\1> If you are cleaning up books or something, it's often best to work from an HTMLZ export - In the format options you can select how CSS is handled, the 'tag' option will not add any extra stuff, tho I cant remember if it preserves bold/italic etc formatting - i.e placing <i> tags..

10-21-2011, 08:41 AM	#10
theducks Well trained by Cats Posts: 30,914 Karma: 60358908 Join Date: Aug 2009 Location: The Central Coast of California Device: Kobo Libra2,Kobo Aura2v1, K4NT(Fixed: New Bat.), Galaxy Tab A	nested spans (or divs) are problematic to replace/remove with REGEX (unless I have been missing a trick ) in that they must match the correct Closing Tag, not the first occurrence of a closing tag. BTW a simple What If I took it out test. /* just comment out the selector in the style sheet / and see what gets ugly you can always remove the comment marks. Unlike CSS errors, undefined styles do not seem to cause problems with ADE on my PEz Last edited by theducks; 10-21-2011 at 08:44 AM.*

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Has any dealt with this company?	kenm	Android Devices	3	10-14-2010 09:05 AM
NCX creation -- nested or not, css classes	illustrata	ePub	3	08-25-2010 08:56 AM
Has anybody ever dealt with this web store?	pathfinderca	Alternative Devices	5	12-03-2009 04:34 PM
Contradictory error messages in ConvertLit	Fledchen	Workshop	6	02-21-2009 12:38 PM

10-20-2011, 02:00 PM	#3
cybmole Wizard Posts: 3,720 Karma: 1759970 Join Date: Sep 2010 Device: none	i did a little experimentation in sigil & + a little googling- I got the impression that with most attributes the outer one takes preference . so font size is taken from calibre 11 only. could well be though that different programs / readers interpret differently i would strip or comment out colors anyway for e-reader as they are not supported. thanks for the tutorial link.

10-21-2011, 06:42 AM	#8
Dopedangel Wizard Posts: 1,791 Karma: 30548723 Join Date: Dec 2006 Location: Singapore Device: Boyue	What kind of ebook needs so many styles. I mostly read fiction so I just convert to text with textile. Then convert to epub from that you can add your custom styles for headings paragraphs italics while converting from text to epub if you are so inclined.