regex class search

[rjwse@aol.com]

Sometimes the simplest things really give me a kick. I use classes almost exclusively instead of styles. My CSS is a long list of classes in the main consisting of the first letter of each word in every which kind of style definition. In this way I can quickly hop and chop from this to that. I got stumped on how to use calibre's 'find' for any kind of class name consisting of one or more letters. It got lost due to being surrounded by spaces, other classes, or quotation marks. So I asked AI and it gave me: class="[^"]*X[^"]*"
I remove X and put in whatever I am looking for, then make sure the search says 'regex' instead of 'fuzzy.' I also put it in 'saved searches' in case I forget.

[dunhill]

Originally Posted by rjwse@aol.com

Sometimes the simplest things really give me a kick. I use classes almost exclusively instead of styles. My CSS is a long list of classes in the main consisting of the first letter of each word in every which kind of style definition. In this way I can quickly hop and chop from this to that. I got stumped on how to use calibre's 'find' for any kind of class name consisting of one or more letters. It got lost due to being surrounded by spaces, other classes, or quotation marks. So I asked AI and it gave me: class="[^"]*X[^"]*"
I remove X and put in whatever I am looking for, then make sure the search says 'regex' instead of 'fuzzy.' I also put it in 'saved searches' in case I forget.

You could try this:
Exact Search (Without False Positives)
If you want to find the exact class `bc` (for example, `background-color`) and avoid finding `abc` or `bcc`, use this variation:

`class="[^"]*(^|\s)bc(\s|$)[^"]*"`

`(^|\s)`: Ensures that your class name is preceded by a quote or a space.

`(\s|$)`: Ensures that it is followed by a space or a closing quote.

Searching for Classes Starting with...`
If, on the other hand, you want to see all the classes that define something related to the border (assuming they all start with `b`), you can use:

`class="[^"]*(^|\s)b[a-z]*(\s|$)[^"]*"`

[Karellen]

Curious to know how many errors are reported when using epubCheck and checkbook.

[lomkiri]

Originally Posted by dunhill

You could try this:
`class="[^"]*(^|\s)bc(\s|$)[^"]*"`

`(^|\s)`: Ensures that your class name is preceded by a quote or a space.

`(\s|$)`: Ensures that it is followed by a space or a closing quote.

I don't agree with you:
(^|\s)bc : Ensures that bc is at the beginning of the line (which never happens since you put class="[^"]* before this expression) or after \s

bc(\s|$) : same problem, ensures that bc is at the EoL (never happens) or before \s

So it would select class="ab bc de" but neither class="ab bc" nor class="bc cd ef"

A working regexp would be :
class="[^"]*?\s*\Kbc(?=[\s"])
it will select only bc in the class (only the first one if there is several of it)
\K asks for resetting the chain (forget what is before \K)
(?=[\s"]) is a lookahead, it selects bc only if it's followed by \s or "

Originally Posted by dunhill

If, on the other hand, you want to see all the classes that define something related to the border (assuming they all start with `b`), you can use:

`class="[^"]*(^|\s)b[a-z]*(\s|$)[^"]*"`

There is also _ and - to be considered.
In that case, this one does the job (finds only the 1st one):
class="[^"]*?\s*\Kbc[\w-]*(?=[\s"])

or, if you want bc and up to 3 more char (e.g. bc, bcd, bcdef, bc-d, bc_de)
class="[^"]*?\s*\Kbc[\w-]{0,3}(?=[\s"])
(change {0,3} for whatever number you need)

[dunhill]

You're right, Lomkiri. My approach with (^|\s) and (\s|$) aimed to emulate word boundaries, but indeed, start and end line anchors can cause problems depending on the HTML formatting. That's why I suggested trying it, but your use of \K is much more precise, especially since it allows you to select only the class without dragging the class=" attribute. Great contribution!

[lomkiri]

You're right, with word boundaries (\b), my 2nd and 3rd regex (for words starting with bc) are more elegant:
class="[^"]*?\K\bbc[\w-]*\b
class="[^"]*?\K\bbc[\w-]{0,3}\b

In the 1st one, we have to keep the lookahead to not have a match with "bc" in "bc-de"
class="[^"]*?\K\bbc(?=[\s"])

[rjwse@aol.com]

Curious to know how many errors are reported when using epubCheck and checkbook.

Do you mean errors reported due to using classes instead of styles? Zero errors. Check book does not report errors when using classes, neither does amazon.

[Karellen]

Originally Posted by rjwse@aol.com

Do you mean errors reported due to using classes instead of styles? Zero errors. Check book does not report errors when using classes, neither does amazon.

No, I just mean errors in general when you run epubCheck (which is an plugin) and Calibre's own Checkbook.
A couple of things don't look quite right in your screenshot.