From f326cd09b24c2cd30794330ee374eef0c83b3cb5 Mon Sep 17 00:00:00 2001
From: Kaz Kylheku <kaz@kylheku.com>
Date: Fri, 16 Sep 2016 20:14:29 -0700
Subject: Bugfix in split-str: empty-match regexes.

* lib.c (split_str_keep): In the regex case, changing
to an infinite loop. The do/while is no longer needed because
the if statement includes a test of the position having
reached the end of the string. This is done before it is
incremented by len, so we avoid wrongly keeping a separator.

* txr.1: Clarified that an empty regex match
behaves like an sep which is an empty string,
---
 lib.c | 9 +++++----
 txr.1 | 9 +++++++--
 2 files changed, 12 insertions(+), 6 deletions(-)

diff --git a/lib.c b/lib.c
index 13982b69..56d4cf88 100644
--- a/lib.c
+++ b/lib.c
@@ -3814,24 +3814,25 @@ val split_str_keep(val str, val sep, val keep_sep)
   if (regexp(sep)) {
     list_collect_decl (out, iter);
     val pos = zero;
+    val slen = length(str);
 
-    do {
+    for (;;) {
       cons_bind (new_pos, len, search_regex(str, sep, pos, nil));
 
-      if (eql(pos, new_pos) && len == zero)
+      if (len == zero && new_pos != slen)
         new_pos = plus(new_pos, one);
 
       iter = list_collect(iter, sub_str(str, pos, new_pos));
       pos = new_pos;
 
-      if (len) {
+      if (len && pos != slen) {
         pos = plus(pos, len);
         if (keep_sep)
           iter = list_collect(iter, sub_str(str, new_pos, pos));
         continue;
       }
       break;
-    } while (le(pos, length_str(str)));
+    }
 
     return out;
   } else {
diff --git a/txr.1 b/txr.1
index 1cbaff45..e0f0b17c 100644
--- a/txr.1
+++ b/txr.1
@@ -18469,11 +18469,16 @@ characters. In this case, if
 .meta string
 is of length one or zero, then it is considered to have no such pieces, and a
 list of one element is returned containing the original string.
+These remarks also apply to the situation when
+.meta sep
+is a regular expression which matches only an empty
+substring of
+.metn string .
 
 If a match for
 .meta sep
-is not found in the string at all, then the string is not
-split at all: a list of one element is returned containing the original
+is not found in the string at all (not even an empty match), then the string is
+not split at all: a list of one element is returned containing the original
 string.
 
 If
-- 
cgit v1.2.3