From: Mattias Engdegård <mattiase@acm.org>
Date: Wed, 9 Aug 2023 10:34:06 +0000 (+0200)
Subject: Faster NSString to Lisp string conversion
X-Git-Url: http://git.eshelyaron.com/gitweb/?a=commitdiff_plain;h=722b1ebc6e0ca23c018d5264e5c70b8f37bd9150;p=emacs.git

Faster NSString to Lisp string conversion

Since we know that the value from [NSString UTF8String] is valid
UTF-8, using make_string is wastefully slow.

* src/nsfns.m (count_utf8_chars): New function, at least twice as fast
as parse_str_as_multibyte used by make_string for this purpose.
([NSString lispString]): Use count_utf8_chars.  We now always make a
multibyte string because there is no reason not to.
---

diff --git a/src/nsfns.m b/src/nsfns.m
index f962de02cb9..508568d90c3 100644
--- a/src/nsfns.m
+++ b/src/nsfns.m
@@ -3796,6 +3796,27 @@ all_nonzero_ascii (unsigned char *str, ptrdiff_t n)
   return true;
 }
 
+/* Count the number of characters in STR, NBYTES long.
+   The string is valid UTF-8 except that it may contain unpaired surrogates.  */
+static ptrdiff_t
+count_utf8_chars (const char *str, ptrdiff_t nbytes)
+{
+  /* This is faster than parse_str_as_multibyte, and much faster than
+     [NSString lengthOfBytesUsingEncoding: NSUTF32StringEncoding].  */
+  const char *end = str + nbytes;
+  ptrdiff_t nc = 0;
+  while (str < end)
+    {
+      nc++;
+      unsigned char c = *str;
+      str += (  c <= 0x7f ? 1    // 0xxxxxxx
+              : c <= 0xdf ? 2    // 110xxxxx 10xxxxxx
+              : c <= 0xef ? 3    // 1110xxxx 10xxxxxx 10xxxxxx
+              :             4);  // 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
+    }
+  return nc;
+}
+
 @implementation NSString (EmacsString)
 /* Make an NSString from a Lisp string.  STRING must not be in an
    encoded form (e.g. UTF-8).  */
@@ -3840,9 +3861,9 @@ all_nonzero_ascii (unsigned char *str, ptrdiff_t n)
 /* Make a Lisp string from an NSString.  */
 - (Lisp_Object)lispString
 {
-  // make_string behaves predictably and correctly with UTF-8 input.
-  return make_string ([self UTF8String],
-                      [self lengthOfBytesUsingEncoding: NSUTF8StringEncoding]);
+  const char *utf8 = [self UTF8String];
+  ptrdiff_t bytes = [self lengthOfBytesUsingEncoding: NSUTF8StringEncoding];
+  return make_multibyte_string (utf8, count_utf8_chars (utf8, bytes), bytes);
 }
 @end