Find straight quotes in the text

[mikapanja]

Originally Posted by BetterRed

Two passes on closing quotes are only required if punctuation is sometimes inside e.g ‘This is the end of the sentence.’ and other times outside e.g. ‘This is the end of the sentence’.; having both in the same text would be an exception, probably an error.

Not necessarily. Let me use your examples.

- ‘This is the end of the sentence.’ - most cases, covered by pass one

- He played Billy Holiday’s ‘I Cover the Waterfront’ all night.... - various names (books, songs, movies) and expressions in the middle of the sentence are not covered by pass one, but by pass two (no punctuation before the closing quote, and a space after the closing quote). Other punctuation marks in pass two are probably overkill, but will do no harm. If one is averse to using them, the pass two could be shortened to:

Search: ’([ ])
Replace: ”\1

[JSWolf]

Originally Posted by kovidgoyal

Smarten punctuation does preserve straight quotes on measurements. https://bugs.launchpad.net/calibre/+bug/1285351

To be precise it replaces them with the correct typographic prime

Is there any way Smarten punctuation can be programmed to do a single quote to double quote conversion for when we have UK type eBooks that have the single quote? I prefer double quotes as that is what I am used to.

[JSWolf]

Originally Posted by mikapanja

@ JSWolf

I believe this should cover it, but it will replace all the quotes, both in dialogues and outside them, except those that are part of a contraction. Of course, the book must have proper punctuation to begin with.

1) Replace opening single curly quotes with opening double curly quotes
Search: ‘
Replace: “

2) Replace closing single curly quotes with closing double curly quotes (two passes)

Search: ([.|…|,|:|;|?|!|\-|–|—])’
Replace: \1”

Search: ’([ |.|…|,|:|;|?|!|\-|–|—])
Replace: ”\1

Some of the combinations are unlikely to be found, but I guess better an overkill than missed instances

Thanks. I'll have to give it a go and see how well it works.

[arspr]

Be really careful when detecting straight single quotes BEFORE words as in English there are at least two very different uses:
'em >>> Contractions, so you should substitute by a RIGHT/CLOSING curly quote.
'Start of dialogue' >>> so you should substitute by a LEFT/OPENING curly quote.

[Tex2002ans]

Originally Posted by arspr

Be really careful when detecting straight single quotes BEFORE words as in English there are at least two very different uses:
'em >>> Contractions, so you should substitute by a RIGHT/CLOSING curly quote.

I use these Regex in Sigil after running any sort of "Smarten Punctuation":

Note: That is a LEFT single quote in the Search.

Search: ‘([Tt]was\b)
Replace: ’\1

Search: ‘([Tt]is\b)
Replace: ’\1

Search: ‘([Ee]m\b)
Replace: ’\1

This works on your common words: 'twas, 'tis, 'em. (I am not too sure how many others there are, but those three are the only ones I run across in the books I work on).

I also do a Regex for this one and replace on a case-by-case basis for this one:

Search: ‘([0-9])
Replace: ’\1

Which is typically used when shortening decades: ’90s, ’00s, ’60s, ...

[GrannyGrump]

Another frequent failure for "Smarten" routines happens when they hit a quoted phrase followed by an em-dash. Often the quote marks preceding the em-dash get reversed.
So instead of seeing
John said: “I told you,“’twas brillig’”—a hush fell.

you see

... “‘twas brillig‘“—a hush fell.

Ah, "smartened" curly quotes do make life interesting.

[arspr]

Originally Posted by Tex2002ans

(I am not too sure how many others there are, but those three are the only ones I run across in the books I work on).

Not very representative but in "The Sot-Weed Factor" there are lots of them:
'twixt, 'tween, 'sheart, 'twill, 'twere, 'twould, 'swounds, ...

And of course 'n' as in Guns 'n' Roses.

[davidfor]

Originally Posted by arspr

Not very representative but in "The Sot-Weed Factor" there are lots of them:
'twixt, 'tween, 'sheart, 'twill, 'twere, 'twould, 'swounds, ...

And of course 'n' as in Guns 'n' Roses.

My saved search is:

 ‘(em|cause|Cause|cos|til|bout|er|um|tis)

That gets most of 'em. The spelling check finds most of the others.

[Tex2002ans]

Originally Posted by arspr

And of course 'n' as in Guns 'n' Roses.

Rock ’n’ Roll as well.

Originally Posted by davidfor

My saved search is:

 ‘(em|cause|Cause|cos|til|bout|er|um|tis)

That gets most of 'em. The spelling check finds most of the others.

This is actually much more efficient than what I use, I will be commandeering this for my own usage.

And speaking of the decades rule I mentioned.... The book I am converting just this second uses the WORD FORM of it:

’eighties
’nineties

Never ran across that before. Coincidence? I think not!

[ShellShock]

I want to widen the original question if I may: "Is there a way to limit the scope of a search/replace to the actual text that is displayed to the reader?" E.g., we can currently set the scope to "Current file", "All text files" etc. I was thinking of an additional option to set the scope, e.g., "All inner text". I believe this is what smarty pants does (could be wrong).

This option would enable us to roll our own smarty pants using the Saved searches, without worrying about false positive matches on the xml tags. I currently have a regex to match punctuation that is also matching <> in the xml elements. I guess I could try to exclude these from the regex, but it would be safer if the match scope was limited to inner text.

[Sabardeyn]

Add 'cuz to the list for a slang/jargon version of 'cause, et al.

Although this might lead to confusion with cuz (cousin), as in "this is my cuz". Or would it be spelled cous? Anyway, something to consider.