Re: [PATCH] Allow inserting non-BMP characters

[Philipp Stephani <p.stephani2@gmail.com>]

[-- Attachment #1.1: Type: text/plain, Size: 1301 bytes --]

Eli Zaretskii <eliz@gnu.org> schrieb am Fr., 29. Dez. 2017 um 21:27 Uhr:

> > From: Philipp Stephani <p.stephani2@gmail.com>
> > Date: Fri, 29 Dec 2017 20:14:41 +0000
> > Cc: alan@idiocy.org, phst@google.com, emacs-devel@gnu.org
> >
> > diff --git a/src/coding.c b/src/coding.c
> > index 1705838ffa..9903d87b92 100644
> > --- a/src/coding.c
> > +++ b/src/coding.c
> > @@ -1515,13 +1515,6 @@ encode_coding_utf_8 (struct coding_system *coding)
> >  /* See the above "GENERAL NOTES on `detect_coding_XXX ()' functions".
> >     Return true if a text is encoded in one of UTF-16 based coding
> systems.  */
> >
> > -#define UTF_16_HIGH_SURROGATE_P(val) \
> > -  (((val) & 0xFC00) == 0xD800)
> > -
> > -#define UTF_16_LOW_SURROGATE_P(val) \
> > -  (((val) & 0xFC00) == 0xDC00)
> > -
> > -
> >  static bool
> >  detect_coding_utf_16 (struct coding_system *coding,
> >                     struct coding_detection_info *detect_info)
> > @@ -1686,7 +1679,7 @@ decode_coding_utf_16 (struct coding_system
> *coding)
> >
> >        if (surrogate)
> >       {
> > -       if (! UTF_16_LOW_SURROGATE_P (c))
> > +       if (! char_low_surrogate_p (c))
>
> Please don't.  This makes decoding UTF-16 a tad slower for no good
> reason (more than a tad in unoptimized builds).
>
>
OK, here's a new version of the patch.

[-- Attachment #1.2: Type: text/html, Size: 2093 bytes --]

[-- Attachment #2: 0001-Allow-inserting-non-BMP-characters.txt --]
[-- Type: text/plain, Size: 4030 bytes --]

From aa1d5b2600c5ef0a76c8525462ea88db2a2fa25c Mon Sep 17 00:00:00 2001
From: Philipp Stephani <phst@google.com>
Date: Mon, 25 Dec 2017 22:00:00 +0100
Subject: [PATCH] Allow inserting non-BMP characters

* src/coding.h (UTF_16_HIGH_SURROGATE_P, UTF_16_LOW_SURROGATE_P): Move
from coding.c and document.
(surrogates_to_codepoint): New function.

* src/nsterm.m (insertText:): Properly handle surrogate pairs.
---
 src/coding.c |  7 -------
 src/coding.h | 24 ++++++++++++++++++++++++
 src/nsterm.m | 25 +++++++++++++++++++------
 3 files changed, 43 insertions(+), 13 deletions(-)

diff --git a/src/coding.c b/src/coding.c
index d8bc525026..da62540344 100644
--- a/src/coding.c
+++ b/src/coding.c
@@ -1515,13 +1515,6 @@ encode_coding_utf_8 (struct coding_system *coding)
 /* See the above "GENERAL NOTES on `detect_coding_XXX ()' functions".
    Return true if a text is encoded in one of UTF-16 based coding systems.  */
 
-#define UTF_16_HIGH_SURROGATE_P(val) \
-  (((val) & 0xFC00) == 0xD800)
-
-#define UTF_16_LOW_SURROGATE_P(val) \
-  (((val) & 0xFC00) == 0xDC00)
-
-
 static bool
 detect_coding_utf_16 (struct coding_system *coding,
 		      struct coding_detection_info *detect_info)
diff --git a/src/coding.h b/src/coding.h
index 54100ccd31..d90b799d76 100644
--- a/src/coding.h
+++ b/src/coding.h
@@ -662,6 +662,30 @@ struct coding_system
 /* Note that this encodes utf-8, not utf-8-emacs, so it's not a no-op.  */
 #define ENCODE_UTF_8(str) code_convert_string_norecord (str, Qutf_8, true)
 
+/* Return true if VAL is a high surrogate.  VAL must be a 16-bit code
+   unit.  */
+
+#define UTF_16_HIGH_SURROGATE_P(val) \
+  (((val) & 0xFC00) == 0xD800)
+
+/* Return true if VAL is a low surrogate.  VAL must be a 16-bit code
+   unit.  */
+
+#define UTF_16_LOW_SURROGATE_P(val) \
+  (((val) & 0xFC00) == 0xDC00)
+
+/* Return the Unicode code point for the given UTF-16 surrogates.  */
+
+INLINE int
+surrogates_to_codepoint (int low, int high)
+{
+  eassert (0 <= low && low <= 0xFFFF);
+  eassert (0 <= high && high <= 0xFFFF);
+  eassert (UTF_16_LOW_SURROGATE_P (low));
+  eassert (UTF_16_HIGH_SURROGATE_P (high));
+  return 0x10000 + (low - 0xDC00) + ((high - 0xD800) * 0x400);
+}
+
 /* Extern declarations.  */
 extern Lisp_Object code_conversion_save (bool, bool);
 extern bool encode_coding_utf_8 (struct coding_system *);
diff --git a/src/nsterm.m b/src/nsterm.m
index dd5c7d91ea..3f59e33c7b 100644
--- a/src/nsterm.m
+++ b/src/nsterm.m
@@ -6283,14 +6283,13 @@ flag set (this is probably a bug in the OS).
          by doCommandBySelector: deleteBackward: */
 - (void)insertText: (id)aString
 {
-  int code;
-  int len = [(NSString *)aString length];
-  int i;
+  NSString *s = aString;
+  NSUInteger len = [s length];
 
   NSTRACE ("[EmacsView insertText:]");
 
   if (NS_KEYLOG)
-    NSLog (@"insertText '%@'\tlen = %d", aString, len);
+    NSLog (@"insertText '%@'\tlen = %lu", aString, (unsigned long) len);
   processingCompose = NO;
 
   if (!emacs_event)
@@ -6300,10 +6299,24 @@ - (void)insertText: (id)aString
   if (workingText != nil)
     [self deleteWorkingText];
 
+  /* It might be preferable to use getCharacters:range: below,
+     cf. https://developer.apple.com/library/content/documentation/Cocoa/Conceptual/CocoaPerformance/Articles/StringDrawing.html#//apple_ref/doc/uid/TP40001445-112378.
+     However, we probably can't use SAFE_NALLOCA here because it might
+     exit nonlocally.  */
+
   /* now insert the string as keystrokes */
-  for (i =0; i<len; i++)
+  for (NSUInteger i = 0; i < len; i++)
     {
-      code = [aString characterAtIndex: i];
+      NSUInteger code = [s characterAtIndex:i];
+      if (UTF_16_HIGH_SURROGATE_P (code) && i < len - 1)
+        {
+          unichar low = [s characterAtIndex:i + 1];
+          if (UTF_16_LOW_SURROGATE_P (low))
+            {
+              code = surrogates_to_codepoint (low, code);
+              ++i;
+            }
+        }
       /* TODO: still need this? */
       if (code == 0x2DC)
         code = '~'; /* 0x7E */
-- 
2.15.1