From 80fe0ae75835556fc41daef16ad0dffd4a0563fc Mon Sep 17 00:00:00 2001
From: Steve Hoeksema <steve.h@buildkite.com>
Date: Tue, 9 Apr 2024 12:18:50 +1200
Subject: [PATCH] Add support for escaped UTF-16 surrogate pairs

I have followed the convention above that error handling is TODO, but I can start implementing it if you like.

References:

* https://russellcottrell.com/greek/utilities/SurrogatePairCalculator.htm
* https://mathiasbynens.be/notes/javascript-unicode
* https://en.wikipedia.org/wiki/UTF-16#U+D800_to_U+DFFF_(surrogates)
---
 decode_test.go     | 12 ++++++++++++
 scanner/scanner.go | 28 ++++++++++++++++++++++++++++
 2 files changed, 40 insertions(+)

diff --git a/decode_test.go b/decode_test.go
index cabfd33c..23229ef5 100644
--- a/decode_test.go
+++ b/decode_test.go
@@ -437,6 +437,18 @@ func TestDecoder(t *testing.T) {
 			`"1": "a\x2Fb\u002Fc\U0000002Fd"`,
 			map[interface{}]interface{}{"1": `a/b/c/d`},
 		},
+		{
+			`"\ud83e\udd23"`,
+			"🤣",
+		},
+		{
+			`"\uD83D\uDE00\uD83D\uDE01"`,
+			"😀😁",
+		},
+		{
+			`"\uD83D\uDE00a\uD83D\uDE01"`,
+			"😀a😁",
+		},
 		{
 			"'1': \"2\\n3\"",
 			map[interface{}]interface{}{"1": "2\n3"},
diff --git a/scanner/scanner.go b/scanner/scanner.go
index b0eac48d..865cfb84 100644
--- a/scanner/scanner.go
+++ b/scanner/scanner.go
@@ -391,6 +391,34 @@ func (s *Scanner) scanDoubleQuote(ctx *Context) (tk *token.Token, pos int) {
 						return
 					}
 					codeNum := hexRunesToInt(src[idx+2 : idx+6])
+
+					// Handle surrogate pairs
+					if codeNum >= 0xD800 && codeNum <= 0xDBFF {
+						high := codeNum
+
+						if idx+11 >= size {
+							// TODO: need to return error
+							//err = xerrors.New("not enough characters for surrogate pair")
+							return
+						}
+
+						if src[idx+6] != '\\' || src[idx+7] != 'u' {
+							// TODO: need to return error
+							//err = xerrors.New("expected escape code after high surrogate")
+							return
+						}
+
+						low := hexRunesToInt(src[idx+8 : idx+12])
+						if low < 0xDC00 || low > 0xDFFF {
+							// TODO: need to return error
+							//err = xerrors.New("expected low surrogate after high surrogate")
+							return
+						}
+
+						codeNum = ((high - 0xD800) * 0x400) + (low - 0xDC00) + 0x10000
+						idx += 6
+					}
+
 					value = append(value, rune(codeNum))
 					idx += 5
 					continue