R devel - Apr 2009 - should sub(perl=TRUE) also handle \E in replacement, to complement \U and \L?

Currently sub(perl=TRUE) allows you to specify \U and \L
in the replacement argument so that the rest of the subpatterns
in the line (the \\<digit> things) will be converted to upper
or lower case, respectively.  perl also also has a \E operator
to end these case conversions for the rest of the subpatterns
(so they retain whatever case they had in the original text).
For symmetry's sake I think it would be nice if R supported that
also.  E.g., to capitalize the first and last letters of every
word, leaving the case of the interior letters alone, could be
done with:
> gsub("(\\w)(\\w*)(\\w)", "\\U\\1\\E\\2\\U\\3",
"useRs may fly into JFK
or laGuardia", perl=TRUE)
[1] "UseRS MaY FlY IntO JFK OR LaGuardiA"
> sub("(\\w)(\\w*)(\\w)", "\\U\\1\\E\\2\\U\\3",
"useRs may fly into JFK
or laGuardia", perl=TRUE)
[1] "UseRS may fly into JFK or laGuardia"

A question regarding this came up in r-help today.

Bill Dunlap
TIBCO Software Inc - Spotfire Division
wdunlap tibco.com 

Index: src/library/base/man/grep.Rd
==================================================================---
src/library/base/man/grep.Rd	(revision 48319)
+++ src/library/base/man/grep.Rd	(working copy)
@@ -73,7 +73,7 @@
     \code{"\\9"} to parenthesized subexpressions of \code{pattern}.
For
     \code{perl = TRUE} only, it can also contain \code{"\\U"} or
     \code{"\\L"} to convert the rest of the replacement to upper or
-    lower case.
+    lower case, or \code{"\\E"} to end such case conversion.
   }
 }
 \details{


Index: src/main/pcre.c
==================================================================---
src/main/pcre.c	(revision 48319)
+++ src/main/pcre.c	(working copy)
@@ -90,6 +90,9 @@
 	    } else if (p[1] == 'L') {
 		p++; n -= 2;
 		upper = FALSE; lower = TRUE;
+	    } else if (p[1] == 'E') { /* end case modification */
+		p++; n -= 2;
+		upper = FALSE; lower = FALSE;
 	    } else if (p[1] == 0) {
 		/* can't escape the final '\0' */
 		n--;
@@ -168,6 +171,9 @@
 	    } else if (p[1] == 'L') {
 		p += 2;
 		upper = FALSE; lower = TRUE;
+	    } else if (p[1] == 'E') { /* end case modification */
+		p += 2;
+		upper = FALSE; lower = FALSE;
 	    } else if (p[1] == 0) {
 		p += 1;
 	    } else {
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: pcre.diff.txt
URL:
<https://stat.ethz.ch/pipermail/r-devel/attachments/20090413/db64170b/attachment.txt>

>>>>> "WD" == William Dunlap <wdunlap at
tibco.com>
>>>>>     on Mon, 13 Apr 2009 11:56:51 -0700 writes:
    WD> Currently sub(perl=TRUE) allows you to specify \U and \L
    WD> in the replacement argument so that the rest of the
    WD> subpatterns in the line (the \\<digit> things) will be
    WD> converted to upper or lower case, respectively.  perl
    WD> also also has a \E operator to end these case
    WD> conversions for the rest of the subpatterns (so they
    WD> retain whatever case they had in the original text).
    WD> For symmetry's sake I think it would be nice if R
    WD> supported that also.  E.g., to capitalize the first and
    WD> last letters of every word, leaving the case of the
    WD> interior letters alone, could be done with:

    >> gsub("(\\w)(\\w*)(\\w)", "\\U\\1\\E\\2\\U\\3",
"useRs may
    >> fly into JFK
    WD> or laGuardia", perl=TRUE) [1] "UseRS MaY FlY IntO JFK OR
    WD> LaGuardiA"
    >> sub("(\\w)(\\w*)(\\w)", "\\U\\1\\E\\2\\U\\3",
"useRs may
    >> fly into JFK
    WD> or laGuardia", perl=TRUE) [1] "UseRS may fly into JFK or
    WD> laGuardia"

    WD> A question regarding this came up in r-help today.

    WD> Bill Dunlap TIBCO Software Inc - Spotfire Division
    WD> wdunlap tibco.com

Thanks a lot, Bill, for your patch!

I have applied and committed it (after testing) to R-devel
[rev. 48321].

Best regards,
Martin Maechler, ETH Zurich