unicode support improvements

[jangko]

• add more test cases for unicode string
  see clarify that utf-8 is just a possible encoding of strings graphql/graphql-spec#684 and RFC: Allow full unicode range graphql/graphql-spec#849

[jangko]

Summary of Unicode important points:

• GraphQL does not have an official encoding, it stays at the abstraction of a "sequence of code points"

• Graphql document:

  • UTF16-BE BOM: 0xFE, 0xFF
  • UTF-8 BOM: 0xEF,0xBB,0xBF
  • string encoding: UTF-16 or UTF-8
  • escaped unicode in quoted string take the form of UTF-16 BE:
    • 4 digit hex: \u000A
    • variable length: \u{1F4A9}
    • range (>= 0x0000 and <= 0xD7FF or >= 0xE000 and <= 0x10FFFF)
    • Escape sequences are only meaningful within a single-quoted string
    • SurrogatePair: "\uD83D\uDCA9" is equal to "\u{1F4A9}"
  • Document: ASCII, UTF-16, UTF-8
    • graphql will use ASCII for the language expression, but strings and comments can use Unicode encoding

• JSON Serialization Format:
  • https://www.ietf.org/rfc/rfc4627.txt
  • document: UTF-8, UTF-16, or UTF-32
  • string encoding: default to UTF-8
  • escaped unicode: UTF-16 BE with surrogate pair