[JSONX] Fix unicode handling when unescaping strings (#405)

Previously `JSONX.unescape_string()` would insert single codeunits at a time
into the unescaped output (if the current character is not an escape
sequence). This caused characters composed of multiple codeunits to be unescaped
as individual characters.

Author: JamesWrigley

vendor/README.md

@@ -96,3 +96,9 @@ Compared to the full JSON.jl package, JSONX is intentionally simplified:
 - **Error robust**: Comprehensive error checking and reporting
 
 Note: Functions are not exported, so use `JSONX.parse` and `JSONX.json` with the module prefix.
+
+## Changelog
+
+- Fixed unicode handling when unescaping strings (https://github.com/JuliaIO/JSON.jl/pull/405).
+- **Breaking**: Added support for parsing Int64's (https://github.com/JuliaIO/JSON.jl/pull/396).
+- Added `JSONText` (https://github.com/JuliaIO/JSON.jl/pull/394).

vendor/jsonx.jl

@@ -138,10 +138,14 @@ function unescape_string(str::String, start_pos::Int, end_pos::Int)
                 b == 0x00 && throw(ArgumentError("Invalid escape sequence \\$(Char(esc_c))"))
                 print(io, Char(b))
             end
+
+            pos += 1
         else
-            print(io, Char(c))
+            # Regular character
+            print(io, str[pos])
+            pos = nextind(str, pos)
         end
-        pos += 1
+
     end
     return String(take!(io))
 end

vendor/test.jl

@@ -264,6 +264,8 @@ include("jsonx.jl")
         @test JSONX.parse("\"\\uD83C\\uDF55\"") == "🍕"  # Pizza emoji
         # Mixed content
         @test JSONX.parse("\"Hello \\u0041\\u006E\\u0064\\u0072\\u0065\\u0077!\"") == "Hello Andrew!"
+        # String unescaping
+        @test JSONX.parse(raw"\"𝔸\\a\"") == "𝔸\\a"
         # Writing Unicode
         @test JSONX.json("A") == "\"A\""
         @test JSONX.json("😀") == "\"😀\""