From: Lars Ingebrigtsen <larsi@gnus.org>
Date: Wed, 22 Sep 2021 05:22:17 +0000 (+0200)
Subject: Speed up `replace-match' slightly
X-Git-Tag: emacs-28.0.90~747
X-Git-Url: http://git.eshelyaron.com/gitweb/?a=commitdiff_plain;h=f2bd2386a79fed311361fd9ca809ab1b2a0362a6;p=emacs.git

Speed up `replace-match' slightly

* src/search.c (Freplace_match): Speed up non-literal (but
actually literal) common case.

This makes

(benchmark-run 1000000
  (replace-regexp-in-string
  "a+" "foo"
  "ogihdipofdhookfdohkfdpokhpokhfdpokfdhpokfdhkdfkhgoadfphokfkhpofdkhkdpokf"))

about 10% faster.
---

diff --git a/src/search.c b/src/search.c
index 14adeb58e96..08f1e9474f1 100644
--- a/src/search.c
+++ b/src/search.c
@@ -2387,6 +2387,13 @@ since only regular expressions have distinguished subexpressions.  */)
   if (! NILP (string))
     CHECK_STRING (string);
 
+  /* Most replacement texts don't contain any backslash directives in
+     the replacements.  Check whether that's the case, which will
+     enable us to take the fast path later.  */
+  if (NILP (literal)
+      && !memchr (SSDATA (newtext), '\\', SBYTES (newtext)))
+    literal = Qt;
+
   case_action = nochange;	/* We tried an initialization */
 				/* but some C compilers blew it */