[gawk-diffs] [SCM] gawk branch, feature/wasted-byte, updated. gawk-4.1.0-1310-gb88e0a7

[Arnold Robbins]

This is an automated email from the git hooks/post-receive script. It was
generated because a ref change was pushed to the repository containing
the project "gawk".

The branch, feature/wasted-byte has been updated
       via  b88e0a71eaae6efb9bf4164ae095bccc07965115 (commit)
      from  b57630628a5c6d8d8d16f05b357f61b34c861359 (commit)

Those revisions listed above that are new to this repository have
not appeared on any other notification email; so we list those
revisions in full, below.

- Log -----------------------------------------------------------------
http://git.sv.gnu.org/cgit/gawk.git/commit/?id=b88e0a71eaae6efb9bf4164ae095bccc07965115

commit b88e0a71eaae6efb9bf4164ae095bccc07965115
Author: Arnold D. Robbins <address@hidden>
Date:   Mon Apr 13 18:31:23 2015 +0300

    Improve computations in do_sub.

diff --git a/ChangeLog b/ChangeLog
index eaab207..cff7389 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,10 @@
+2015-04-14         Andrew J. Schorr      <address@hidden>
+                  Arnold D. Robbins     <address@hidden>
+
+       * builtin.c (do_sub): Make computations smarter; initial len
+       to malloc, test for final amount after all matches done and
+       need to copy in the final part of the original string.
+
 2015-04-13         Arnold D. Robbins     <address@hidden>
 
        * regcomp.c (analyze): Prevent malloc(0).
diff --git a/builtin.c b/builtin.c
index b8e2fce..3d3c1c4 100644
--- a/builtin.c
+++ b/builtin.c
@@ -2920,8 +2920,16 @@ set_how_many:
                 * create the result, copying in parts of the original
                 * string 
                 */
-               len = matchstart - text + repllen
-                     + ampersands * (matchend - matchstart);
+
+               /*
+                * add 1 to len to handle "empty" case where
+                * matchend == matchstart and we force a match on a single
+                * char.  Use 'matchend - text' instead of 'matchstart - text'
+                * because we may not actually make any substitution depending
+                * on the 'global' and 'how_many' values.
+                */
+               len = matchend - text + repllen
+                     + ampersands * (matchend - matchstart) + 1;
                sofar = bp - buf;
                while (buflen < (sofar + len + 1)) {
                        buflen *= 2;
@@ -3028,6 +3036,11 @@ set_how_many:
                textlen = text + textlen - matchend;
                text = matchend;
 
+#if 0
+               if (bp - buf > sofar + len)
+                       fprintf(stderr, "debug: len = %zu, but used %ld\n", 
len, (long)((bp - buf) - (long)sofar));
+#endif
+
                if ((current >= how_many && ! global)
                    || ((long) textlen <= 0 && matchstart == matchend)
                    || research(rp, t->stptr, text - t->stptr, textlen, 
RE_NEED_START) == -1)
@@ -3035,12 +3048,17 @@ set_how_many:
 
        }
        sofar = bp - buf;
-       if (buflen - sofar - textlen - 1) {
+       if (buflen < (sofar + textlen + 1)) {
                buflen = sofar + textlen + 1;
                erealloc(buf, char *, buflen, "do_sub");
                bp = buf + sofar;
        }
-       for (scan = matchend; scan < text + textlen; scan++)
+       /*
+        * Note that text == matchend, since that assignment is made before
+        * exiting the 'for' loop above. Thus we copy in the rest of the
+        * original string.
+        */
+       for (scan = text; scan < text + textlen; scan++)
                *bp++ = *scan;
        *bp = '\0';
        textlen = bp - buf;

-----------------------------------------------------------------------

Summary of changes:
 ChangeLog |    7 +++++++
 builtin.c |   26 ++++++++++++++++++++++----
 2 files changed, 29 insertions(+), 4 deletions(-)


hooks/post-receive
-- 
gawk