From: Pavel Charvat <pchar@ucw.cz>
Date: Sat, 9 Mar 2024 22:49:44 +0000 (+0000)
Subject: JSON: Don't generate corrupted output for string with surrogates.
X-Git-Url: http://mj.ucw.cz/gitweb/?a=commitdiff_plain;h=refs%2Fheads%2Fdev-json;p=libucw.git

JSON: Don't generate corrupted output for string with surrogates.
---

diff --git a/ucw-json/format.c b/ucw-json/format.c
index 227a26b5..790fa15a 100644
--- a/ucw-json/format.c
+++ b/ucw-json/format.c
@@ -50,14 +50,16 @@ static void write_string(struct json_context *js, const char *p)
 	      bprintf(fb, "\\u%04x", u);
 	    }
 	}
-      else if (u >= 0x007f && (js->format_options & JSON_FORMAT_ESCAPE_NONASCII))
+      else if (u < 0x7f)
+	bputc(fb, u);
+      else if (u >= 0x110000 || u >= 0xd800 && u <= 0xdfff)
+	bprintf(fb, "\\u%04x", UNI_REPLACEMENT);
+      else if (js->format_options & JSON_FORMAT_ESCAPE_NONASCII)
 	{
 	  if (u < 0x10000)
 	    bprintf(fb, "\\u%04x", u);
-	  else if (u < 0x110000)
-	    bprintf(fb, "\\u%04x\\u%04x", 0xd800 + ((u - 0x10000) >> 10), 0xdc00 + (u & 0x3ff));
 	  else
-	    bprintf(fb, "\\u%04x", UNI_REPLACEMENT);
+	    bprintf(fb, "\\u%04x\\u%04x", 0xd800 + ((u - 0x10000) >> 10), 0xdc00 + (u & 0x3ff));
 	}
       else
 	bput_utf8_32(fb, u);