Recurvise Grep puzzle

[vinny38]

Hi Grep enthusiasts.

Here's a puzzle I'm struggling to solve.

I want to set up my FindChangeList.txt file in order to find and replace typographic quotation marks when wrapped by others. (Double quotation)

What I have

Je vous cite, pour mémoire, le procès-verbal de la dernière réunion : « L’Assemblée juge nécessaire de proposer la modification suivante à l’article 8 du règlement : « Le conseil se compose au maximum de neuf membres. » » Jacques Godbout dépeint le climat politique qui régnait dans les années quarante : « Dans le village de Lanoraie, où nous passions les vacances d’été, un curé affublé d’une perruque carrée, le teint blême, terminait ses sermons par la célèbre formule : « L’enfer est rouge, le ciel est bleu. » Nous étions, enfants libéraux, condamnés à l’enfer. » Comme on dit : « Et voilà ! »

What I want

Je vous cite, pour mémoire, le procès-verbal de la dernière réunion : « L’Assemblée juge nécessaire de proposer la modification suivante à l’article 8 du règlement : “Le conseil se compose au maximum de neuf membres.” » Jacques Godbout dépeint le climat politique qui régnait dans les années quarante : « Dans le village de Lanoraie, où nous passions les vacances d’été, un curé affublé d’une perruque carrée, le teint blême, terminait ses sermons par la célèbre formule : “L’enfer est rouge, le ciel est bleu.” Nous étions, enfants libéraux, condamnés à l’enfer. » Comme on dit : « Et voilà ! »

The colored version for a better understanding:

Of course, the regex must also work if the text is already "clean".

I have the feeling the solution lies in the recursion operator (?R) but it's very undocumented and I can't figure it out.

See : Recursive Regex—Tutorial and (?R) grep recursive pattern …

Any suggestion welcome.

Vinny

[vladan saveljic]

Try this grep Vinny

find:

«.*\K«(?=.*)|»(?=.*»)

replace with:

"

[vinny38]

Hi Vladan.

Thanks a lot for your help.

Unfortunately, that won't do.

First thing is that because of Grep greed, «.*\K« will catch the last « of the paragraph.

We could workaround this by using the shortest match operator «.*?\K«

Problem is that if you replay the query, it will now catch the first « of the second quotation, which should be left untouched.

[Peter Kahrel · Accepted solution]

You don't really need recursion, you're after a single embedding. If the inner quote has no formatting at all, you could do this:

Find what: «[^»]+\K«(.+?)»

Change to: “$1”

If the inner quote does have formatting or index markers, or anchors, whatever, you cant use it, then you need some script.

P.

[vinny38]

Hi Peter,

thanks a lot for your answer.

You're right, I wrongly focused on the recursion while a negative class was definitely what I needed.

Just added the missing thin spaces and your regex works like a charm.

The only thing that slightly bothers me is that, yes, a variable, or most probably a footnote could sometimes be included in the inner quote.

And the idea is to set up the FindChangeList.txt file, so no scripting can be involved.

Anyhow, it still is a good start.

Thanks a lot

Vinny

[Peter Kahrel]

. . . which is (or could be) very simple:

app.findGrepPreferences = null;
app.findGrepPreferences.findWhat = '«[^»]+\\K«(.+?)»';
f = app.documents[0].findGrep();
for (i = 0; i < f.length; i++) {
  f.characters[0].contents = '“';
  f.characters[-1].contents = '”';
}

Tested on only a tiny sample.

P.