From f2ef00adde5e3c3f0bae06574282e3ccfbed05f3 Mon Sep 17 00:00:00 2001
From: Karl Williamson <khw@cpan.org>
Date: Sun, 30 May 2021 16:50:26 -0600
Subject: [PATCH] Avoid some conditionals in is...UTF8_CHAR()

These three functions to determine if the next bit of a string is UTF-8
(constrained in three different ways) have basically the same short
loop.

One of the initial conditions in the while() is always true the first
time around.  By moving that condition to the middle of the loop, we
avoid it for the common case where the loop is executed just once.  This
is when the input is a UTF-8 invariant character (ASCII on ASCII
platforms).

If the functions were constrained to require the first byte pointed to
by the input to exist, the while() could be a do {} while(), and there
would be no extra conditional in calling this vs checking if the next
character is invariant, and if not calling this.  And there would be
fewer conditionals for the case of 2 or more bytes in the character.
---
 inline.h | 43 ++++++++++++++++++++++++++-----------------
 1 file changed, 26 insertions(+), 17 deletions(-)

diff --git a/inline.h b/inline.h
index 045f88066636..65b2d9df0c64 100644
--- a/inline.h
+++ b/inline.h
@@ -1127,16 +1127,19 @@ Perl_isUTF8_CHAR(const U8 * const s0, const U8 * const e)
      * on 32-bit ASCII platforms where it trivially is an error).  Call a
      * helper function for the other platforms. */
 
-    while (s < e && LIKELY(state != 1)) {
-        state = PL_extended_utf8_dfa_tab[256
+    while (s < e) {
+        state = PL_extended_utf8_dfa_tab[  256
                                          + state
                                          + PL_extended_utf8_dfa_tab[*s]];
-        if (state != 0) {
-            s++;
-            continue;
+        s++;
+
+        if (state == 0) {
+            return s - s0;
         }
 
-        return s - s0 + 1;
+        if (UNLIKELY(state == 1)) {
+            break;
+        }
     }
 
 #if defined(UV_IS_QUAD) || defined(EBCDIC)
@@ -1195,15 +1198,19 @@ Perl_isSTRICT_UTF8_CHAR(const U8 * const s0, const U8 * const e)
 
     PERL_ARGS_ASSERT_ISSTRICT_UTF8_CHAR;
 
-    while (s < e && LIKELY(state != 1)) {
-        state = PL_strict_utf8_dfa_tab[256 + state + PL_strict_utf8_dfa_tab[*s]];
+    while (s < e) {
+        state = PL_strict_utf8_dfa_tab[  256
+                                       + state
+                                       + PL_strict_utf8_dfa_tab[*s]];
+        s++;
 
-        if (state != 0) {
-            s++;
-            continue;
+        if (state == 0) {
+            return s - s0;
         }
 
-        return s - s0 + 1;
+        if (UNLIKELY(state == 1)) {
+            break;
+        }
     }
 
 #ifndef EBCDIC
@@ -1261,15 +1268,17 @@ Perl_isC9_STRICT_UTF8_CHAR(const U8 * const s0, const U8 * const e)
 
     PERL_ARGS_ASSERT_ISC9_STRICT_UTF8_CHAR;
 
-    while (s < e && LIKELY(state != 1)) {
+    while (s < e) {
         state = PL_c9_utf8_dfa_tab[256 + state + PL_c9_utf8_dfa_tab[*s]];
+        s++;
 
-        if (state != 0) {
-            s++;
-            continue;
+        if (state == 0) {
+            return s - s0;
         }
 
-        return s - s0 + 1;
+        if (UNLIKELY(state == 1)) {
+            break;
+        }
     }
 
     return 0;