lib/core: Renamed `Text::bytelen` to `Text::byte_length`
[nit.git] / lib / json / static.nit
index fdb9f65..dd1cc85 100644 (file)
@@ -31,6 +31,9 @@ private import json_lexer
 interface Jsonable
        # Encode `self` in JSON.
        #
+       # This is a recursive method which can be refined by any subclasses.
+       # To write any `Serializable` object to JSON, see `serialize_to_json`.
+       #
        # SEE: `append_json`
        fun to_json: String is abstract
 
@@ -47,9 +50,9 @@ interface Jsonable
        # avoid cyclic references between `append_json` and `to_json` when none are
        # implemented.
        protected fun to_json_by_append: String do
-               var buffer = new RopeBuffer
+               var buffer = new FlatBuffer
                append_json(buffer)
-               return buffer.write_to_string
+               return buffer.to_s
        end
 
        # Append the JSON representation of `self` to the specified buffer.
@@ -80,7 +83,7 @@ interface Jsonable
                var res = new FlatBuffer
                pretty_json_visit(res, 0)
                res.add '\n'
-               return res.write_to_string
+               return res.to_s
        end
 
        private fun pretty_json_visit(buffer: FlatBuffer, indent: Int) is abstract
@@ -89,32 +92,40 @@ end
 redef class Text
        super Jsonable
 
+       # Removes JSON-escaping if necessary in a JSON string
+       #
+       #     assert "\\\"string\\uD83D\\uDE02\\\"".unescape_json == "\"string😂\""
+       fun unescape_json: Text do
+               if not json_need_escape then return self
+               return self.json_to_nit_string
+       end
+
+       # Does `self` need treatment from JSON to Nit ?
+       #
+       # i.e. is there at least one `\` character in it ?
+       #
+       #     assert not "string".json_need_escape
+       #     assert "\\\"string\\\"".json_need_escape
+       protected fun json_need_escape: Bool do return has('\\')
+
        redef fun append_json(buffer) do
                buffer.add '\"'
-               for i in [0..self.length[ do
+               for i in [0 .. self.length[ do
                        var char = self[i]
                        if char == '\\' then
                                buffer.append "\\\\"
                        else if char == '\"' then
                                buffer.append "\\\""
-                       else if char == '\/' then
-                               buffer.append "\\/"
-                       else if char < 16.ascii then
+                       else if char < ' ' then
                                if char == '\n' then
                                        buffer.append "\\n"
                                else if char == '\r' then
                                        buffer.append "\\r"
                                else if char == '\t' then
                                        buffer.append "\\t"
-                               else if char == 0x0C.ascii then
-                                       buffer.append "\\f"
-                               else if char == 0x08.ascii then
-                                       buffer.append "\\b"
                                else
-                                       buffer.append "\\u000{char.ascii.to_hex}"
+                                       buffer.append char.escape_to_utf16
                                end
-                       else if char < ' ' then
-                               buffer.append "\\u00{char.ascii.to_hex}"
                        else
                                buffer.add char
                        end
@@ -122,13 +133,65 @@ redef class Text
                buffer.add '\"'
        end
 
+       # Escapes `self` from a JSON string to a Nit string
+       #
+       #     assert "\\\"string\\\"".json_to_nit_string == "\"string\""
+       #     assert "\\nEscape\\t\\n".json_to_nit_string == "\nEscape\t\n"
+       #     assert "\\u0041zu\\uD800\\uDFD3".json_to_nit_string == "Azu𐏓"
+       protected fun json_to_nit_string: String do
+               var res = new FlatBuffer.with_capacity(byte_length)
+               var i = 0
+               var ln = self.length
+               while i < ln do
+                       var char = self[i]
+                       if char == '\\' then
+                               i += 1
+                               char = self[i]
+                               if char == 'b' then
+                                       char = 0x08.code_point
+                               else if char == 'f' then
+                                       char = 0x0C.code_point
+                               else if char == 'n' then
+                                       char = '\n'
+                               else if char == 'r' then
+                                       char = '\r'
+                               else if char == 't' then
+                                       char = '\t'
+                               else if char == 'u' then
+                                       var u16_esc = from_utf16_digit(i + 1)
+                                       char = u16_esc.code_point
+                                       if char.is_surrogate and i + 10 < ln then
+                                               if self[i + 5] == '\\' and self[i + 6] == 'u' then
+                                                       u16_esc <<= 16
+                                                       u16_esc += from_utf16_digit(i + 7)
+                                                       char = u16_esc.from_utf16_surr.code_point
+                                                       i += 6
+                                               else
+                                                       char = 0xFFFD.code_point
+                                               end
+                                       end
+                                       i += 4
+                               end
+                               # `"`, `/` or `\` => Keep `char` as-is.
+                       end
+                       res.add char
+                       i += 1
+               end
+               return res.to_s
+       end
+
+
        # Encode `self` in JSON.
        #
        # ~~~
        # assert "\t\"http://example.com\"\r\n\0\\".to_json ==
-       #     "\"\\t\\\"http:\\/\\/example.com\\\"\\r\\n\\u0000\\\\\""
+       #     "\"\\t\\\"http://example.com\\\"\\r\\n\\u0000\\\\\""
        # ~~~
-       redef fun to_json do return to_json_by_append
+       redef fun to_json do
+               var b = new FlatBuffer.with_capacity(byte_length)
+               append_json(b)
+               return b.to_s
+       end
 
        # Parse `self` as JSON.
        #
@@ -175,6 +238,16 @@ redef class Text
        end
 end
 
+redef class FlatText
+       redef fun json_need_escape do
+               var its = items
+               for i in [first_byte .. last_byte] do
+                       if its[i] == 0x5Cu8 then return true
+               end
+               return false
+       end
+end
+
 redef class Buffer
 
        # Append the JSON representation of `jsonable` to `self`.
@@ -369,6 +442,11 @@ redef class JsonParseError
                                "\"position\":{position.to_json}," +
                                "\"message\":{message.to_json}\}"
        end
+
+       redef fun pretty_json_visit(buf, indents) do
+               buf.clear
+               buf.append(to_json)
+       end
 end
 
 redef class Position
@@ -426,41 +504,7 @@ end
 
 redef class Nstring
        # The represented string.
-       private fun to_nit_string: String do
-               var res = new FlatBuffer
-               var i = 1
-               while i < text.length - 1 do
-                       var char = text[i]
-                       if char == '\\' then
-                               i += 1
-                               char = text[i]
-                               if char == 'b' then
-                                       char = 0x08.ascii
-                               else if char == 'f' then
-                                       char = 0x0C.ascii
-                               else if char == 'n' then
-                                       char = '\n'
-                               else if char == 'r' then
-                                       char = '\r'
-                               else if char == 't' then
-                                       char = '\t'
-                               else if char == 'u' then
-                                       var code = text.substring(i + 1, 4).to_hex
-                                       # TODO UTF-16 escaping is not supported yet.
-                                       if code >= 128 then
-                                               char = '?'
-                                       else
-                                               char = code.ascii
-                                       end
-                                       i += 4
-                               end
-                               # `"`, `/` or `\` => Keep `char` as-is.
-                       end
-                       res.add char
-                       i += 1
-               end
-               return res.write_to_string
-       end
+       private fun to_nit_string: String do return text.substring(1, text.length - 2).unescape_json.to_s
 end
 
 redef class Nvalue_object