From 8526ae3d0d0636e92bcc206a715b0b53546e2674 Mon Sep 17 00:00:00 2001
From: leitner <leitner>
Date: Tue, 25 Jul 2017 19:50:33 +0000
Subject: [PATCH] fix utf-8 json encoding for outside basic multilingual plane

---
 textcode/fmt_jsonescape.c  | 4 ++--
 textcode/scan_jsonescape.c | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/textcode/fmt_jsonescape.c b/textcode/fmt_jsonescape.c
index dde2731..4c78046 100644
--- a/textcode/fmt_jsonescape.c
+++ b/textcode/fmt_jsonescape.c
@@ -47,7 +47,7 @@ escape:
 	  if (dest) {
 	    dest[written  ]='\\';
 	    dest[written+1]='u';
-	    fmt_xlong(dest+written+2,0xd800 + ((u>>10) & 0x3ff));
+	    fmt_xlong(dest+written+2,0xd800 + ((u>>10) & 0x3bf));
 	    dest[written+6]='\\';
 	    dest[written+7]='u';
 	    fmt_xlong(dest+written+8,0xdc00 + (u & 0x3ff));
@@ -82,6 +82,6 @@ int main() {
   /* test escaping of unprintable characters */
   assert(fmt_jsonescape(buf,"\001x",2)==7 && !memcmp(buf,"\\u0001x",7));
   /* test conversion of large UTF-8 chars to UTF-16 surrogate pairs (poop emoji) */
-  assert(fmt_jsonescape(buf,"\xf0\x9f\x92\xa9x",5)==13 && !memcmp(buf,"\\ud87d\\udca9x",13));
+  assert(fmt_jsonescape(buf,"\xf0\x9f\x92\xa9x",5)==13 && !memcmp(buf,"\\ud83d\\udca9x",13));
 }
 #endif
diff --git a/textcode/scan_jsonescape.c b/textcode/scan_jsonescape.c
index d28350d..30ac6fb 100644
--- a/textcode/scan_jsonescape.c
+++ b/textcode/scan_jsonescape.c
@@ -99,7 +99,7 @@ int main() {
   assert(scan_jsonescape("a\\udafd0",buf,&l)==1);
   /* correct surrogate pair */
   assert(scan_jsonescape("a\\ud834\\udd1eb",buf,&l)==14 && l==6 && !memcmp(buf,"a\xf0\x9d\x84\x9e""b",6));
-  assert(scan_jsonescape("\\ud87d\\udca9x",buf,&l)==13 && l==5 && !memcmp(buf,"\xf0\x9f\x92\xa9x",5));
+  assert(scan_jsonescape("\\ud83d\\udca9x",buf,&l)==13 && l==5 && !memcmp(buf,"\xf0\x9f\x92\xa9x",5));
   /* how about some incorrect UTF-8? */
   assert(scan_jsonescape("a\xc0\xaf",buf,&l)==1 && l==1 && !memcmp(buf,"a",1));
   return 0;