Regex examples

[roger64]

I am sorry but I am not sure to understand. Look at the screenshot for a big quote. What's missing?

Oh, you mean having this:

«

at the opening of each paragraph and

»

at the end of the last one?

If this is what you meant, problem is that the original book has none of them. Also, present or not, these quotes would not have changed a thing in the regex.

[gipsy]

Hi,
i use the following regex to replace some hyphenation of words.

Find: (\p{Greek})-(\p{Greek})
Replace: \1\2

There is an way to ignore some of the regex results?
for example to ignore: πλάι-πλάι, ίσα-ίσα, μισό-μισό and all the - with the same word as \1 and \2

Thanks

[Doitsu]

Originally Posted by gipsy

There is an way to ignore some of the regex results

Yes, but this requires some serious "Regex Fu" (e.g. negative lookbehinds).

I'd simply search for repeated Greek words with a hyphen between them and replace the hyphens with a substitute character (@):

Find:(\p{Greek}+)-(\1)
Replace:\1@\2

Then you can use your regex and at the end you can globally replace all at signs (@) with hyphens.

@Jellby: Can you optimize this simple regex by creating a regex that will find a Greek word not followed by a hyphen and the same Greek word using backreferences and negative lookbehinds?

[Jellby]

Originally Posted by Doitsu

@Jellby: Can you optimize this simple regex by creating a regex that will find a Greek word not followed by a hyphen and the same Greek word using backreferences and negative lookbehinds?

I could try, but I can only test it in vim, which uses a different regex dialect, so I don't think it would be very useful. Besides, I would rather do as you suggested: first replace repeated Greek words.

[mzmm]

you could try this:

find:
(?<=\P{Greek})(\p{Greek}+)-(?!\1)

replace:
\1

it's looking one or more greek characters in a capturing group:

(\p{Greek}+)

that are preceded by anything other than a greek character:

(?<=\P{Greek})

then a hyphen:

-

that is not followed by the group it matched previously:

(?!\1)

replacing it with \1 just removes the hyphen

** edit **

i was trying to get this to work with unicode ranges so that it could be simplified further (no need for the look-behind), but couldn't seem to get it working in sigil, or my other text editor which has PCRE, for that matter.

i was trying to match [\u0370-\u03FF] and (?-u)[\u0370-\u03FF] with no success. anyone have tips on this?

** edit 2 **

i was hoping to get rid of the look-behind by starting the expression with a word boundary, but turns out \b is only useful for ASCII characters, i.e. [a-zA-Z0-9_], so looks like the look-behind may be necessary in these cases.

here's an updated version based on Doitsu's comment below that includes Greek_Extended in the search pattern:

(?<![\x{0370}-\x{03FF}\x{1F00}-\x{1FFF}])([\x{0370}-\x{03FF}\x{1F00}-\x{1FFF}]+)-(?!\1)

[mzmm]

Originally Posted by Doitsu

I'd simply search for repeated Greek words with a hyphen between them and replace the hyphens with a substitute character (@):

Find:(\p{Greek}+)-(\1)
Replace:\1@\2

Then you can use your regex and at the end you can globally replace all at signs (@) with hyphens.

because of how common the @ sign has become, i'd suggest using a more obscure character like ¬ if you're going to do the search replace in steps, though.

[Doitsu]

Originally Posted by mzmm

i was trying to match [\u0370-\u03FF] and (?-u)[\u0370-\u03FF] with no success. anyone have tips on this?

AFAIK, Sigil uses a PCRE engine that doesn't support \u; you'll have to use \x{xxxx} instead: [\x{0370}-\x{03FF}]

However, to be on the safe side, you may want to include the precomposed characters from the "Greek Extended" block (U+1F00 to U+1FFF):

[\x{0370}-\x{03FF}\x{1F00}-\x{1FFF}]

[mzmm]

Originally Posted by Doitsu

AFAIK, Sigil uses a PCRE engine that doesn't support \u; you'll have to use \x{xxxx} instead: [\x{0370}-\x{03FF}]

However, to be on the safe side, you may want to include the precomposed characters from the "Greek Extended" block (U+1F00 to U+1FFF):

[\x{0370}-\x{03FF}\x{1F00}-\x{1FFF}]

Nice! thanks @Doitsu

[gipsy]

Thanks Doitsu, thanks mzmm!
It works better now! (the edit #2)

[gipsy]

I try to make a step forward. But i failed :P

to exclude some results (for example the γερο- from γερο-Κομπ) i try the

Find: (?<![\x{0370}-\x{03FF}\x{1F00}-\x{1FFF}])([^(γερο)\-][\x{0370}-\x{03FF}\x{1F00}-\x{1FFF}]+)-(?!\1)
Replace: \1

but it also exclude words like γκρεμοτσακι-ζόταν and γρα-τζουνιές

Any thoughts?

Thanks again

[mzmm]

mm, i think you're going to run into problems pretty quickly if you start trying to get regex to understand words (as it sounds like you've already found out). i'd really recommend against trying to write a single regex to rule them all.

that said, i'd probably come at it the other way, so you'd include Κομπ in the look-ahead, rather than γερο in the capturing group:

...<same as before>...(?!Κομπ|\1)

the pipe | separating them means 'or'.

i have no idea if this makes syntactical sense to do this in the greek language, but it matches the examples you've provided.

[John Doe]

I hope this question hasn't been asked before, but here goes:

I want to make epub3 files. With notes (epub:type="noteref" and so on.)

I do know how to make the files, but it isn't automated in Sigil, sadly!

But when I make the files and create the links, Sigil makes this:

<a href="#id1">This text will have a link to a note</a>

<a id="id1">This will be the note </a>

I want to find all the

<a href="id***">

(where the id has different numbers), and replace them with

<a epub:type="noteref" href="#id***" xmlns:epub="http://www.idpf.org/2007/ops">

where I can replace the old code with new code leaving the ID numbers back.

Is that possible?

[mzmm]

you should be able to use

find:
<a href="#id(\d+)"

replace:
<a epub:type="noteref" href="#id\1" xmlns:epub="http://www.idpf.org/2007/ops"

[John Doe]

Originally Posted by mzmm

you should be able to use

find:
<a href="#id(\d+)"

replace:
<a epub:type="noteref" href="#id\1" xmlns:epub="http://www.idpf.org/2007/ops"

Yes, that Works! Thank a lot!!

Could you please explain how it works?

[mzmm]

yep:

find:

<a href="#id

followed by a capturing group `()` that contains one or more integers `\d+`

(\d+)

followed by

"

replace:

<a epub:type="noteref" href="#id

followed by a back-reference to the captured group above

\1

followed by

" xmlns:epub="http://www.idpf.org/2007/ops"

this site is an invaluable reference for anything regex, basic to advanced:
http://www.regular-expressions.info