Regex: CSS font-size keyword to ems

[Brett Merkey]

I have found that the Kindle PaperWhite does not handle CSS font-size keywords well. If device font-size settings are not at the default, the Kindle will display fonts at uncomfortably extreme sizes. I have created a Calibre regex-function which maps keywords to ems.

1) Restrict search to style files. Apply "Beautify files" tool to make sure code is ready.

2) Set Mode to Regex-function.

3) Search: font-size: \K([^\d.;]+)

4) Click "Create/Edit" button and paste the following into the code box:

def replace(match, number, file_name, metadata, dictionaries, data, functions, *args, **kwargs):
    # Match from list of size keywords
    i = ["medium", "small", "large", "x-small", "xx-small", "x-large", "xx-large", "larger", "smaller"].index(match.group(1))
    # Map of corresponding sizes. Adjust as needed.
    size = ["1em", ".85em", "1.2em", ".75em", ".75em", "1.3em", "1.5em", "1.1em", ".9em"]
    
    return size[i]

5) Name the function and run.

You can see from the code snippet that you can adjust the mappings to meet your own requirements.

[Thom*]

Nice!
Alternatively, you could have written 9 simple Search and Replace statements.

[Brett Merkey]

...9 simple Search and Replace statements

There is something not quite "simple" in the number 9 — and something not quite simple in remembering, from book to book, which keyword maps to which em value. I let Calibre remember all this for me by making a regex-function.

Thank you, Calibre!

[retiredbiker]

Originally Posted by Brett Merkey

There is something not quite "simple" in the number 9 — and something not quite simple in remembering, from book to book, which keyword maps to which em value. I let Calibre remember all this for me by making a regex-function.

Thank you, Calibre!

I had one search, that I would edit and run up to 9 times for a given book!

Thank you, Brett, this works like a charm and will same me a ton of work. I have been just gritting my teeth whenever I get a book from Open Road (among others); now it will be easy.

[Brett Merkey]

Thanks for the encouragement, RetiredBiker. I've been using this timesaver for a while. Perhaps I will review my collection of regexes and see if there any others that might be of general use.

[Phssthpok]

Many thanks for this -- another nice addition to my Saved Searches collection!

[Phssthpok]

...although I've just discovered that if I specify "All style files" it applies to ALL my saved searches, and I have to switch from "All text files" to "All styles files" and back again when I run it.

Wouldn't it be nice if each saved search included a specification of which files it applied to?

[Brett Merkey]

Wouldn't it be nice if each saved search included a specification of which files it applied to?

I've wondered about that. The Saved Search feature saves every other relevant parameter but that one. I have to watch that and remember to switch back and forth since my regex searches apply, each in turn, to "all text" or "all selected" or "all style" files.

[retiredbiker]

I groused about having to change file sets at first, but now I have searches that can run on either text or style files, depending on the book. One that gets rid of hundreds of "@page" lines, for example.

The list of saved searches gets so long it it becomes unwieldy after a while. Maybe a way to save different sets of searches--like text searches, style searches, and the infamous Other--that might be nice. But then, having to choose a set each time could be cumbersome.

[Phssthpok]

Originally Posted by retiredbiker

The list of saved searches gets so long it it becomes unwieldy after a while. Maybe a way to save different sets of searches--like text searches, style searches, and the infamous Other--that might be nice. But then, having to choose a set each time could be cumbersome.

I've asked about this before, but Kovid isn't keen on the idea. Oh well.

[theducks]

Originally Posted by Phssthpok

I've asked about this before, but Kovid isn't keen on the idea. Oh well.

One of the features of Sigil I make use of:
Search Groups (like folders). I have (also) USAGE type groups.
CSS S&R, Cleanup (messy code, de-wordify), Touchup (font size rename pt->em)
You can treat the group as a folder or you can RUN the group with a couple of clicks

Neither Sigil or Calibre lets you set the file type or S&R mode as part of the saved search.

[Brett Merkey]

Neither Sigil or Calibre lets you set the file type or S&R mode as part of the saved search

Calibre does not let you set the S&R mode but that is because the S&R mode is saved as part of the definition.

Perhaps there is some technical or user case reason why the type of file the search acts upon can not be saved. No problem, really. Just kind of anomalous.

[Thasaidon]

Originally Posted by theducks

One of the features of Sigil I make use of:
Search Groups (like folders). I have (also) USAGE type groups.
CSS S&R, Cleanup (messy code, de-wordify), Touchup (font size rename pt->em)
You can treat the group as a folder or you can RUN the group with a couple of clicks

These techniques sound very useful. I will be sorting mine into usage groups.

I have tried to run an existing group but cannot find the right couple of clicks. Could you give me a hint please.

[theducks]

Originally Posted by Thasaidon

These techniques sound very useful. I will be sorting mine into usage groups.

I have tried to run an existing group but cannot find the right couple of clicks. Could you give me a hint please.

This really is a Sigil forum question.
The Group only works from the Saved Search Window (not the context menu)
SELECT the group title: Click the Load Search button: <your choice on the how to run>

[Thasaidon]

Originally Posted by theducks

This really is a Sigil forum question.
The Group only works from the Saved Search Window (not the context menu)
SELECT the group title: Click the Load Search button: <your choice on the how to run>

You are quite right I should have asked this in the Sigil forum. It just never occurred to me after I saw your post.

Thank you very much for the info.