text: use UInt32 to manipulate chars
[nit.git] / lib / json / static.nit
index 4841020..f5eef1d 100644 (file)
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-# Static interface to get Nit objects from a Json string.
+# Static interface to read Nit objects from JSON strings
 #
-# `String::json_to_nit_object` returns an equivalent Nit object from
-# the Json source. This object can then be type checked by the usual
-# languages features (`isa` and `as`).
+# `Text::parse_json` returns a simple Nit object from the JSON source.
+# This object can then be type checked as usual with `isa` and `as`.
 module static
 
 import error
@@ -29,53 +28,107 @@ private import json_lexer
 
 # Something that can be translated to JSON.
 interface Jsonable
-       # Encode `self` in JSON.
-       fun to_json: String is abstract
+       super Serializable
 end
 
 redef class Text
        super Jsonable
 
-       # Encode `self` in JSON.
+       # Removes JSON-escaping if necessary in a JSON string
        #
-       #     assert "\t\"http://example.com\"\r\n\0\\".to_json ==
-       #               "\"\\t\\\"http:\\/\\/example.com\\\"\\r\\n\\u0000\\\\\""
-       redef fun to_json do
-               var buffer = new FlatBuffer
-               buffer.add '\"'
-               for i in [0..self.length[ do
+       #     assert "\\\"string\\uD83D\\uDE02\\\"".unescape_json == "\"stringšŸ˜‚\""
+       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('\\')
+
+       # 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 == "Azuš“"
+       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
-                               buffer.append "\\\\"
-                       else if char == '\"' then
-                               buffer.append "\\\""
-                       else if char == '\/' then
-                               buffer.append "\\/"
-                       else if char < 16.ascii 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}"
+                               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.to_u32.from_utf16_surr.code_point
+                                                       i += 6
+                                               else
+                                                       char = 0xFFFD.code_point
+                                               end
+                                       end
+                                       i += 4
                                end
-                       else if char < ' ' then
-                               buffer.append "\\u00{char.ascii.to_hex}"
-                       else
-                               buffer.add char
+                               # `"`, `/` or `\` => Keep `char` as-is.
                        end
+                       res.add char
+                       i += 1
                end
-               buffer.add '\"'
-               return buffer.write_to_string
+               return res.to_s
        end
 
-       fun json_to_nit_object: nullable Jsonable do
+       # Parse `self` as JSON.
+       #
+       # If `self` is not a valid JSON document or contains an unsupported escape
+       # sequence, return a `JSONParseError`.
+       #
+       # Example with `JsonObject`:
+       #
+       #     var obj = "\{\"foo\": \{\"bar\": true, \"goo\": [1, 2, 3]\}\}".parse_json
+       #     assert obj isa JsonObject
+       #     assert obj["foo"] isa JsonObject
+       #     assert obj["foo"].as(JsonObject)["bar"] == true
+       #
+       # Example with `JsonArray`:
+       #
+       #     var arr = "[1, 2, 3]".parse_json
+       #     assert arr isa JsonArray
+       #     assert arr.length == 3
+       #     assert arr.first == 1
+       #     assert arr.last == 3
+       #
+       # Example with `String`:
+       #
+       #     var str = "\"foo, bar, baz\"".parse_json
+       #     assert str isa String
+       #     assert str == "foo, bar, baz"
+       #
+       # Example of a syntaxic error:
+       #
+       #     var bad = "\{foo: \"bar\"\}".parse_json
+       #     assert bad isa JsonParseError
+       #     assert bad.position.col_start == 2
+       fun parse_json: nullable Jsonable do
                var lexer = new Lexer_json(to_s)
                var parser = new Parser_json
                var tokens = lexer.lex
@@ -89,93 +142,32 @@ redef class Text
        end
 end
 
-redef class Buffer
-
-       # Append the JSON representation of `jsonable` to `self`.
-       #
-       # Append `"null"` for `null`.
-       private fun append_json_of(jsonable: nullable Jsonable) do
-               if jsonable isa Jsonable then
-                       append jsonable.to_json
-               else
-                       append "null"
+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 Int
        super Jsonable
-
-       # Encode `self` in JSON.
-       #
-       #     assert 0.to_json == "0"
-       #     assert (-42).to_json == "-42"
-       redef fun to_json do return self.to_s
 end
 
 redef class Float
        super Jsonable
-
-       # Encode `self` in JSON.
-       #
-       # Note: Because this method use `to_s`, it may lose precision.
-       #
-       # ~~~
-       # # Will not work as expected.
-       # # assert (-0.0).to_json == "-0.0"
-       #
-       # assert (.5).to_json == "0.5"
-       # assert (0.0).to_json == "0.0"
-       # ~~~
-       redef fun to_json do return self.to_s
 end
 
 redef class Bool
        super Jsonable
-
-       # Encode `self` in JSON.
-       #
-       #     assert true.to_json == "true"
-       #     assert false.to_json == "false"
-       redef fun to_json do return self.to_s
 end
 
 # A map that can be translated into a JSON object.
 interface JsonMapRead[K: String, V: nullable Jsonable]
        super MapRead[K, V]
        super Jsonable
-
-       # Encode `self` in JSON.
-       #
-       #     var obj = new JsonObject
-       #     obj["foo"] = "bar"
-       #     assert obj.to_json == "\{\"foo\":\"bar\"\}"
-       #     obj = new JsonObject
-       #     obj["baz"] = null
-       #     assert obj.to_json == "\{\"baz\":null\}"
-       redef fun to_json do
-               var buffer = new FlatBuffer
-               buffer.append "\{"
-               var it = iterator
-               if it.is_ok then
-                       append_json_entry(it, buffer)
-                       while it.is_ok do
-                               buffer.append ","
-                               append_json_entry(it, buffer)
-                       end
-               end
-               it.finish
-               buffer.append "\}"
-               return buffer.write_to_string
-       end
-
-       private fun append_json_entry(iterator: MapIterator[String, nullable Jsonable],
-                       buffer: Buffer) do
-               buffer.append iterator.key.to_json
-               buffer.append ":"
-               buffer.append_json_of(iterator.item)
-               iterator.next
-       end
 end
 
 # A JSON Object.
@@ -188,36 +180,6 @@ end
 class JsonSequenceRead[E: nullable Jsonable]
        super Jsonable
        super SequenceRead[E]
-
-       # Encode `self` in JSON.
-       #
-       #     var arr = new JsonArray.with_items("foo", null)
-       #     assert arr.to_json == "[\"foo\",null]"
-       #     arr.pop
-       #     assert arr.to_json =="[\"foo\"]"
-       #     arr.pop
-       #     assert arr.to_json =="[]"
-       redef fun to_json do
-               var buffer = new FlatBuffer
-               buffer.append "["
-               var it = iterator
-               if it.is_ok then
-                       append_json_entry(it, buffer)
-                       while it.is_ok do
-                               buffer.append ","
-                               append_json_entry(it, buffer)
-                       end
-               end
-               it.finish
-               buffer.append "]"
-               return buffer.write_to_string
-       end
-
-       private fun append_json_entry(iterator: Iterator[nullable Jsonable],
-                       buffer: Buffer) do
-               buffer.append_json_of(iterator.item)
-               iterator.next
-       end
 end
 
 # A JSON array.
@@ -228,46 +190,18 @@ end
 
 redef class JsonParseError
        super Jsonable
-
-       # Get the JSON representation of `self`.
-       #
-       #     var err = new JsonParseError("foo", new Position(1, 2, 3, 4, 5, 6))
-       #     assert err.to_json == "\{\"error\":\"JsonParseError\"," +
-       #               "\"position\":\{" +
-       #                       "\"pos_start\":1,\"pos_end\":2," +
-       #                       "\"line_start\":3,\"line_end\":4," +
-       #                       "\"col_start\":5,\"col_end\":6" +
-       #               "\},\"message\":\"foo\"\}"
-       redef fun to_json do
-               return "\{\"error\":\"JsonParseError\"," +
-                               "\"position\":{position.to_json}," +
-                               "\"message\":{message.to_json}\}"
-       end
 end
 
 redef class Position
        super Jsonable
-
-       # Get the JSON representation of `self`.
-       #
-       #     var pos = new Position(1, 2, 3, 4, 5, 6)
-       #     assert pos.to_json == "\{" +
-       #                       "\"pos_start\":1,\"pos_end\":2," +
-       #                       "\"line_start\":3,\"line_end\":4," +
-       #                       "\"col_start\":5,\"col_end\":6" +
-       #               "\}"
-       redef fun to_json do
-               return "\{\"pos_start\":{pos_start},\"pos_end\":{pos_end}," +
-                               "\"line_start\":{line_start},\"line_end\":{line_end}," +
-                               "\"col_start\":{col_start},\"col_end\":{col_end}\}"
-       end
 end
 
 ################################################################################
 # Redef parser
 
 redef class Nvalue
-       fun to_nit_object: nullable Jsonable is abstract
+       # The represented value.
+       private fun to_nit_object: nullable Jsonable is abstract
 end
 
 redef class Nvalue_number
@@ -296,41 +230,8 @@ redef class Nvalue_null
 end
 
 redef class Nstring
-       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
+       # The represented string.
+       private fun to_nit_string: String do return text.substring(1, text.length - 2).unescape_json.to_s
 end
 
 redef class Nvalue_object
@@ -346,7 +247,8 @@ redef class Nvalue_object
 end
 
 redef class Nmembers
-       fun pairs: Array[Npair] is abstract
+       #Ā All the key-value pairs.
+       private fun pairs: Array[Npair] is abstract
 end
 
 redef class Nmembers_tail
@@ -363,8 +265,11 @@ redef class Nmembers_head
 end
 
 redef class Npair
-       fun name: String do return n_string.to_nit_string
-       fun value: nullable Jsonable do return n_value.to_nit_object
+       # The represented key.
+       private fun name: String do return n_string.to_nit_string
+
+       # The represented value.
+       private fun value: nullable Jsonable do return n_value.to_nit_object
 end
 
 redef class Nvalue_array
@@ -381,7 +286,8 @@ redef class Nvalue_array
 end
 
 redef class Nelements
-       fun items: Array[Nvalue] is abstract
+       # All the items.
+       private fun items: Array[Nvalue] is abstract
 end
 
 redef class Nelements_tail