|
163
|
1 using System.Linq;
|
|
|
2 using Implab.Automaton.RegularExpressions;
|
|
165
|
3 using System;
|
|
172
|
4 using Implab.Automaton;
|
|
180
|
5 using Implab.Components;
|
|
163
|
6
|
|
|
7 namespace Implab.Formats.JSON {
|
|
178
|
8 class JSONGrammar : Grammar<char> {
|
|
163
|
9 public enum TokenType {
|
|
|
10 None,
|
|
|
11 BeginObject,
|
|
|
12 EndObject,
|
|
|
13 BeginArray,
|
|
|
14 EndArray,
|
|
|
15 String,
|
|
|
16 Number,
|
|
|
17 Literal,
|
|
|
18 NameSeparator,
|
|
|
19 ValueSeparator,
|
|
|
20
|
|
|
21 StringBound,
|
|
|
22 EscapedChar,
|
|
|
23 UnescapedChar,
|
|
176
|
24 EscapedUnicode
|
|
163
|
25 }
|
|
|
26
|
|
180
|
27 static LazyAndWeak<JSONGrammar> _instance = new LazyAndWeak<JSONGrammar>(() => new JSONGrammar());
|
|
165
|
28
|
|
|
29 public static JSONGrammar Instance {
|
|
|
30 get { return _instance.Value; }
|
|
|
31 }
|
|
|
32
|
|
178
|
33 readonly ScannerContext<TokenType> m_jsonExpression;
|
|
|
34 readonly ScannerContext<TokenType> m_stringExpression;
|
|
180
|
35 readonly CharAlphabet m_defaultAlphabet = new CharAlphabet();
|
|
163
|
36
|
|
|
37 public JSONGrammar() {
|
|
|
38 DefineAlphabet(Enumerable.Range(0, 0x20).Select(x => (char)x));
|
|
|
39 var hexDigit = SymbolRangeToken('a','f').Or(SymbolRangeToken('A','F')).Or(SymbolRangeToken('0','9'));
|
|
|
40 var digit9 = SymbolRangeToken('1', '9');
|
|
|
41 var zero = SymbolToken('0');
|
|
|
42 var digit = zero.Or(digit9);
|
|
|
43 var dot = SymbolToken('.');
|
|
|
44 var minus = SymbolToken('-');
|
|
|
45 var sign = SymbolSetToken('-', '+');
|
|
|
46 var expSign = SymbolSetToken('e', 'E');
|
|
|
47 var letters = SymbolRangeToken('a', 'z');
|
|
|
48 var integer = zero.Or(digit9.Cat(digit.EClosure()));
|
|
|
49 var frac = dot.Cat(digit.Closure());
|
|
|
50 var exp = expSign.Cat(sign.Optional()).Cat(digit.Closure());
|
|
|
51 var quote = SymbolToken('"');
|
|
|
52 var backSlash = SymbolToken('\\');
|
|
|
53 var specialEscapeChars = SymbolSetToken('\\', '"', '/', 'b', 'f', 't', 'n', 'r');
|
|
|
54 var unicodeEspace = SymbolToken('u').Cat(hexDigit.Repeat(4));
|
|
|
55 var whitespace = SymbolSetToken('\n', '\r', '\t', ' ').EClosure();
|
|
|
56 var beginObject = whitespace.Cat(SymbolToken('{')).Cat(whitespace);
|
|
|
57 var endObject = whitespace.Cat(SymbolToken('}')).Cat(whitespace);
|
|
|
58 var beginArray = whitespace.Cat(SymbolToken('[')).Cat(whitespace);
|
|
|
59 var endArray = whitespace.Cat(SymbolToken(']')).Cat(whitespace);
|
|
|
60 var nameSep = whitespace.Cat(SymbolToken(':')).Cat(whitespace);
|
|
|
61 var valueSep = whitespace.Cat(SymbolToken(',')).Cat(whitespace);
|
|
|
62
|
|
|
63 var number = minus.Optional().Cat(integer).Cat(frac.Optional()).Cat(exp.Optional());
|
|
|
64 var literal = letters.Closure();
|
|
|
65 var unescaped = SymbolTokenExcept(Enumerable.Range(0, 0x20).Union(new int[] { '\\', '"' }).Select(x => (char)x));
|
|
|
66
|
|
|
67 var jsonExpression =
|
|
|
68 number.Tag(TokenType.Number)
|
|
|
69 .Or(literal.Tag(TokenType.Literal))
|
|
|
70 .Or(quote.Tag(TokenType.StringBound))
|
|
|
71 .Or(beginObject.Tag(TokenType.BeginObject))
|
|
|
72 .Or(endObject.Tag(TokenType.EndObject))
|
|
|
73 .Or(beginArray.Tag(TokenType.BeginArray))
|
|
|
74 .Or(endArray.Tag(TokenType.EndArray))
|
|
|
75 .Or(nameSep.Tag(TokenType.NameSeparator))
|
|
|
76 .Or(valueSep.Tag(TokenType.ValueSeparator));
|
|
|
77
|
|
|
78
|
|
|
79 var jsonStringExpression =
|
|
|
80 quote.Tag(TokenType.StringBound)
|
|
|
81 .Or(backSlash.Cat(specialEscapeChars).Tag(TokenType.EscapedChar))
|
|
|
82 .Or(backSlash.Cat(unicodeEspace).Tag(TokenType.EscapedUnicode))
|
|
|
83 .Or(unescaped.Closure().Tag(TokenType.UnescapedChar));
|
|
|
84
|
|
|
85
|
|
178
|
86 m_jsonExpression = BuildScannerContext<TokenType>(jsonExpression);
|
|
|
87 m_stringExpression = BuildScannerContext<TokenType>(jsonStringExpression);
|
|
|
88
|
|
|
89
|
|
163
|
90 }
|
|
|
91
|
|
180
|
92 protected override IAlphabetBuilder<char> AlphabetBuilder {
|
|
|
93 get {
|
|
|
94 return m_defaultAlphabet;
|
|
|
95 }
|
|
|
96 }
|
|
|
97
|
|
178
|
98 public ScannerContext<TokenType> JsonExpression {
|
|
163
|
99 get {
|
|
178
|
100 return m_jsonExpression;
|
|
163
|
101 }
|
|
|
102 }
|
|
|
103
|
|
178
|
104 public ScannerContext<TokenType> JsonStringExpression {
|
|
163
|
105 get {
|
|
178
|
106 return m_stringExpression;
|
|
163
|
107 }
|
|
|
108 }
|
|
165
|
109
|
|
178
|
110 Token SymbolRangeToken(char start, char stop) {
|
|
165
|
111 return SymbolToken(Enumerable.Range(start,stop - start).Cast<char>());
|
|
|
112 }
|
|
172
|
113
|
|
180
|
114 protected override IndexedAlphabetBase<char> CreateAlphabet() {
|
|
172
|
115 return new CharAlphabet();
|
|
|
116 }
|
|
165
|
117
|
|
163
|
118 }
|
|
|
119 }
|
