From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from imr1.bain.com.au ([203.0.62.7]) by hawkwind.utcs.utoronto.ca with SMTP id <45564>; Wed, 30 May 2001 03:33:12 -0500
Received: from bmr1-e1.aus.deuba.com by imr1.bain.com.au 
          id f4U5YY826129; Wed, 30 May 2001 15:34:34 +1000 (EST)
Received: from mailhost.aus.deuba.com [10.179.4.6] by bmr1-e1.aus.deuba.com 
          id f4U5YX027913; Wed, 30 May 2001 15:34:33 +1000 (EST)
Received: from merton.aus.deuba.com (merton.bain.com.au [10.179.23.56])
	by mailhost.aus.deuba.com (8.9.1/8.9.1) with ESMTP id PAA12433
	for <rc@hawkwind.utcs.utoronto.ca>; Wed, 30 May 2001 15:34:33 +1000 (EST)
Received: (from callum@localhost)
	by merton.aus.deuba.com (8.11.2/8.11.2/SuSE Linux 8.11.1-0.5) id f4U5YWh18448
	for rc@hawkwind.utcs.utoronto.ca; Wed, 30 May 2001 15:34:32 +1000
From:	Callum Gibson <callum.gibson@db.com>
Message-Id: <200105300534.f4U5YWh18448@merton.aus.deuba.com>
Subject: history and multiple substitutions
To:	rc@hawkwind.utcs.utoronto.ca (rc Shell)
Date:	Wed, 30 May 2001 01:34:32 -0500
X-Mailer: ELM [version 2.5 PL2]
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

G'day.

I'm a recovering *csh user. I hope it's not too presumptuous to post
a patch in my first post, but I figure source code is always welcome.
A colleague at work has put me on to rc and he thought this patch nifty
enough to pass on to you all.

One of the annoyances of csh's ^ (and now rc's :) history substitution
for me has been that it only replaces the first occurrence on the line. I
frequently redirect output into a file based upon the args of the program
I'm running.

e.g.
; prog abc > output.abc
; prog def > output.def
etc.
So ^abc^def (or "- abc:def") will clobber your first output file.

The patch below alters history.c such that consecutive colons (:) in the
- and -- history subsitition function, will subsitute your pattern according
to the number of colons you supply. (A more general approach is to change
the existing syntax to mean replace until there are no more matches, but I
decided to apply POLA here.) The original semantics don't escape unscathed
but I think this is a reasonable compromise.

Thus:
; prog abc > output.abc
; - abc::def
prog def > output.def
; - def::ghi
prog ghi > output.ghi
etc.

I hope someone finds it useful. If there's enough interest I'll update the
manpage accordingly too. Oh, this diff is against rc-1.6 release, but it
looks like history.c hasn't changed in some time in any case.

regards,
    C

(c)2001 Callum Gibson                                  callum.gibson@db.com
Fixed Income IT, Deutsche Bank, Australia                    61 2 9258 1620
### The opinions presented herein do not represent those of my employer ###

--- history.c.orig	Sat Dec  5 03:32:40 1998
+++ history.c	Wed May 30 11:48:07 2001
@@ -26,6 +26,7 @@
 
 static struct {
 	char *old, *new;
+	int reps;	/* no. of repetitions. i.e. 1 means sub twice. */
 } *replace;
 
 static char **search, *progname, *history;
@@ -296,8 +297,12 @@
 			search[nsearch++] = argv[i];
 		else {
 			*(char *)s = '\0';	/* do we confuse ps too much? */
+			replace[nreplace].reps = 0;
+			while(*(++s) == ':') {
+			      replace[nreplace].reps++;
+			}
 			replace[nreplace].old = argv[i];
-			replace[nreplace].new = s + 1;
+			replace[nreplace].new = s;
 			nreplace++;
 		}
 
@@ -312,8 +317,11 @@
 	for (i = 0; i < nreplace; i++)
 		if (!isin(s, replace[i].old))
 			goto next;
-		else
-			s = sub(s, replace[i].old, replace[i].new);
+		else {
+		    	int j;
+			for (j = 0; j <= replace[i].reps; j++)
+			      s = sub(s, replace[i].old, replace[i].new);
+		}
 	if (editit) {
 		s = edit(s);
 		if (s == NULL)