Web.JSON: Convert UTF-16 surrogate pairs.

Author: Milly

autoload/vital/__vital__/Web/JSON.vim

@@ -63,7 +63,6 @@ function! s:_resolve(val, prefix) abort
   return a:val
 endfunction
 
-
 function! s:_vital_created(module) abort
   " define constant variables
   if !exists('s:const')
@@ -96,6 +95,10 @@ function! s:decode(json, ...) abort
   let json = join(split(json, "\n"), '')
   let json = substitute(json, '\\u34;', '\\"', 'g')
   let json = substitute(json, '\\u\(\x\x\x\x\)', '\=s:string.nr2enc_char("0x".submatch(1))', 'g')
+  " convert surrogate pair
+  let json = substitute(json, '\([\uD800-\uDBFF]\)\([\uDC00-\uDFFF]\)',
+        \ '\=nr2char(0x10000+and(0x7ff,char2nr(submatch(1)))*0x400+and(0x3ff,char2nr(submatch(2))))',
+        \ 'g')
   if settings.use_token
     let prefix = '__Web.JSON__'
     while stridx(json, prefix) != -1

test/Web/JSON.vimspec

@@ -34,7 +34,12 @@ Describe Web.JSON
       Assert Equals(JSON.decode('"\b\t\n\u000b\f\r\u000e\u000f"'), "\x08\x09\x0a\x0b\x0c\x0d\x0e\x0f")
       Assert Equals(JSON.decode('"\u0010\u0011\u0012\u0013\u0014\u0015\u0016\u0017"'), "\x10\x11\x12\x13\x14\x15\x16\x17")
       Assert Equals(JSON.decode('"\u0018\u0019\u001a\u001b\u001c\u001d\u001e\u001f"'), "\x18\x19\x1a\x1b\x1c\x1d\x1e\x1f")
-      " there should be iconv tests as well
+      " multibyte
+      Assert Equals(JSON.decode('"s¢cĴgё"'), "s¢cĴgё")
+      " UTF-16 surrogate pair
+      Assert Equals(JSON.decode('"\ud83c\udf63"'), "\xf0\x9f\x8d\xa3")
+      " unpaired UTF-16 surrogate
+      Assert Equals(JSON.decode('"\ud83c\u00a0"'), "\ud83c\u00a0")
     End
 
     It decodes lists
@@ -101,7 +106,10 @@ Describe Web.JSON
       Assert Equals(JSON.encode("\x08\x09\x0a\x0b\x0c\x0d\x0e\x0f"), '"\b\t\n\u000b\f\r\u000e\u000f"')
       Assert Equals(JSON.encode("\x10\x11\x12\x13\x14\x15\x16\x17"), '"\u0010\u0011\u0012\u0013\u0014\u0015\u0016\u0017"')
       Assert Equals(JSON.encode("\x18\x19\x1a\x1b\x1c\x1d\x1e\x1f"), '"\u0018\u0019\u001a\u001b\u001c\u001d\u001e\u001f"')
-      " there should be iconv tests as well
+      " multibyte
+      Assert Equals(JSON.encode("s¢cĴgё"), '"s¢cĴgё"')
+      " UTF-16 surrogate pair
+      Assert Equals(JSON.encode("\xf0\x9f\x8d\xa3"), "\"\xf0\x9f\x8d\xa3\"")
     End
 
     It encodes lists