163
|
1 using System.Linq;
|
|
2 using Implab.Automaton.RegularExpressions;
|
|
3
|
|
4 namespace Implab.Formats.JSON {
|
|
5 class JSONGrammar : Grammar<JSONGrammar> {
|
|
6 public enum TokenType {
|
|
7 None,
|
|
8 BeginObject,
|
|
9 EndObject,
|
|
10 BeginArray,
|
|
11 EndArray,
|
|
12 String,
|
|
13 Number,
|
|
14 Literal,
|
|
15 NameSeparator,
|
|
16 ValueSeparator,
|
|
17
|
|
18 StringBound,
|
|
19 EscapedChar,
|
|
20 UnescapedChar,
|
|
21 EscapedUnicode,
|
|
22
|
|
23 Minus,
|
|
24 Plus,
|
|
25 Sign,
|
|
26 Integer,
|
|
27 Dot,
|
|
28 Exp
|
|
29 }
|
|
30
|
|
31 readonly CDFADefinition m_jsonDFA;
|
|
32 readonly CDFADefinition m_stringDFA;
|
|
33
|
|
34 public JSONGrammar() {
|
|
35 DefineAlphabet(Enumerable.Range(0, 0x20).Select(x => (char)x));
|
|
36 var hexDigit = SymbolRangeToken('a','f').Or(SymbolRangeToken('A','F')).Or(SymbolRangeToken('0','9'));
|
|
37 var digit9 = SymbolRangeToken('1', '9');
|
|
38 var zero = SymbolToken('0');
|
|
39 var digit = zero.Or(digit9);
|
|
40 var dot = SymbolToken('.');
|
|
41 var minus = SymbolToken('-');
|
|
42 var sign = SymbolSetToken('-', '+');
|
|
43 var expSign = SymbolSetToken('e', 'E');
|
|
44 var letters = SymbolRangeToken('a', 'z');
|
|
45 var integer = zero.Or(digit9.Cat(digit.EClosure()));
|
|
46 var frac = dot.Cat(digit.Closure());
|
|
47 var exp = expSign.Cat(sign.Optional()).Cat(digit.Closure());
|
|
48 var quote = SymbolToken('"');
|
|
49 var backSlash = SymbolToken('\\');
|
|
50 var specialEscapeChars = SymbolSetToken('\\', '"', '/', 'b', 'f', 't', 'n', 'r');
|
|
51 var unicodeEspace = SymbolToken('u').Cat(hexDigit.Repeat(4));
|
|
52 var whitespace = SymbolSetToken('\n', '\r', '\t', ' ').EClosure();
|
|
53 var beginObject = whitespace.Cat(SymbolToken('{')).Cat(whitespace);
|
|
54 var endObject = whitespace.Cat(SymbolToken('}')).Cat(whitespace);
|
|
55 var beginArray = whitespace.Cat(SymbolToken('[')).Cat(whitespace);
|
|
56 var endArray = whitespace.Cat(SymbolToken(']')).Cat(whitespace);
|
|
57 var nameSep = whitespace.Cat(SymbolToken(':')).Cat(whitespace);
|
|
58 var valueSep = whitespace.Cat(SymbolToken(',')).Cat(whitespace);
|
|
59
|
|
60 var number = minus.Optional().Cat(integer).Cat(frac.Optional()).Cat(exp.Optional());
|
|
61 var literal = letters.Closure();
|
|
62 var unescaped = SymbolTokenExcept(Enumerable.Range(0, 0x20).Union(new int[] { '\\', '"' }).Select(x => (char)x));
|
|
63
|
|
64 var jsonExpression =
|
|
65 number.Tag(TokenType.Number)
|
|
66 .Or(literal.Tag(TokenType.Literal))
|
|
67 .Or(quote.Tag(TokenType.StringBound))
|
|
68 .Or(beginObject.Tag(TokenType.BeginObject))
|
|
69 .Or(endObject.Tag(TokenType.EndObject))
|
|
70 .Or(beginArray.Tag(TokenType.BeginArray))
|
|
71 .Or(endArray.Tag(TokenType.EndArray))
|
|
72 .Or(nameSep.Tag(TokenType.NameSeparator))
|
|
73 .Or(valueSep.Tag(TokenType.ValueSeparator));
|
|
74
|
|
75
|
|
76 var jsonStringExpression =
|
|
77 quote.Tag(TokenType.StringBound)
|
|
78 .Or(backSlash.Cat(specialEscapeChars).Tag(TokenType.EscapedChar))
|
|
79 .Or(backSlash.Cat(unicodeEspace).Tag(TokenType.EscapedUnicode))
|
|
80 .Or(unescaped.Closure().Tag(TokenType.UnescapedChar));
|
|
81
|
|
82
|
|
83 m_jsonDFA = BuildDFA(jsonExpression);
|
|
84 m_stringDFA = BuildDFA(jsonStringExpression);
|
|
85 }
|
|
86
|
|
87 public CDFADefinition JsonDFA {
|
|
88 get {
|
|
89 return m_jsonDFA;
|
|
90 }
|
|
91 }
|
|
92
|
|
93 public CDFADefinition JsonStringDFA {
|
|
94 get {
|
|
95 return m_stringDFA;
|
|
96 }
|
|
97 }
|
|
98 }
|
|
99 }
|