| @@ -1,264 +1,280 | |||||
| 1 | using Implab; |
|
1 | using Implab; | |
| 2 | using Implab.Parsing; |
|
2 | using Implab.Parsing; | |
| 3 | using System; |
|
3 | using System; | |
| 4 | using System.Collections.Generic; |
|
4 | using System.Collections.Generic; | |
| 5 | using System.Diagnostics; |
|
5 | using System.Diagnostics; | |
| 6 | using System.IO; |
|
6 | using System.IO; | |
| 7 | using System.Linq; |
|
7 | using System.Linq; | |
| 8 | using System.Text; |
|
8 | using System.Text; | |
| 9 | using System.Threading.Tasks; |
|
9 | using System.Threading.Tasks; | |
| 10 |
|
10 | |||
| 11 | namespace Implab.JSON { |
|
11 | namespace Implab.JSON { | |
| 12 | /// <summary> |
|
12 | /// <summary> | |
| 13 | /// internal |
|
13 | /// internal | |
| 14 | /// </summary> |
|
14 | /// </summary> | |
| 15 | public struct JSONParserContext { |
|
15 | public struct JSONParserContext { | |
| 16 | public string memberName; |
|
16 | public string memberName; | |
| 17 | public JSONElementContext elementContext; |
|
17 | public JSONElementContext elementContext; | |
| 18 | } |
|
18 | } | |
| 19 |
|
19 | |||
| 20 | /// <summary> |
|
20 | /// <summary> | |
| 21 | /// Pull парсер JSON данных. |
|
21 | /// Pull парсер JSON данных. | |
| 22 | /// </summary> |
|
22 | /// </summary> | |
|
|
23 | /// <remarks> | |||
|
|
24 | /// Следует отметить отдельную интерпретацию свойства <see cref="Level"/>, | |||
|
|
25 | /// оно означает текущий уровень вложенности объектов, однако закрывающий | |||
|
|
26 | /// элемент объекта и массива имеет уровень меньше, чем сам объект. | |||
|
|
27 | /// <code> | |||
|
|
28 | /// { // Level = 1 | |||
|
|
29 | /// "name" : "Peter", // Level = 1 | |||
|
|
30 | /// "address" : { // Level = 2 | |||
|
|
31 | /// city : "Stern" // Level = 2 | |||
|
|
32 | /// } // Level = 1 | |||
|
|
33 | /// } // Level = 0 | |||
|
|
34 | /// </code> | |||
|
|
35 | /// </remarks> | |||
| 23 | public class JSONParser : DFAutomaton<JSONParserContext>, IDisposable { |
|
36 | public class JSONParser : DFAutomaton<JSONParserContext>, IDisposable { | |
| 24 |
|
37 | |||
| 25 | enum MemberContext { |
|
38 | enum MemberContext { | |
| 26 | MemberName, |
|
39 | MemberName, | |
| 27 | MemberValue |
|
40 | MemberValue | |
| 28 | } |
|
41 | } | |
| 29 |
|
42 | |||
| 30 | static readonly EnumAlphabet<JsonTokenType> _alphabet = EnumAlphabet<JsonTokenType>.FullAlphabet; |
|
43 | static readonly EnumAlphabet<JsonTokenType> _alphabet = EnumAlphabet<JsonTokenType>.FullAlphabet; | |
| 31 | static readonly DFAStateDescriptior[] _jsonDFA; |
|
44 | static readonly DFAStateDescriptior[] _jsonDFA; | |
| 32 | static readonly DFAStateDescriptior[] _objectDFA; |
|
45 | static readonly DFAStateDescriptior[] _objectDFA; | |
| 33 | static readonly DFAStateDescriptior[] _arrayDFA; |
|
46 | static readonly DFAStateDescriptior[] _arrayDFA; | |
| 34 |
|
47 | |||
| 35 | static JSONParser() { |
|
48 | static JSONParser() { | |
| 36 | var jsonExpression = Token.New(JsonTokenType.BeginObject, JsonTokenType.BeginArray).Tag(0); |
|
49 | var jsonExpression = Token.New(JsonTokenType.BeginObject, JsonTokenType.BeginArray).Tag(0); | |
| 37 |
|
50 | |||
| 38 | var valueExpression = Token.New(JsonTokenType.BeginArray, JsonTokenType.BeginObject, JsonTokenType.Literal, JsonTokenType.Number, JsonTokenType.String); |
|
51 | var valueExpression = Token.New(JsonTokenType.BeginArray, JsonTokenType.BeginObject, JsonTokenType.Literal, JsonTokenType.Number, JsonTokenType.String); | |
| 39 | var memberExpression = Token.New(JsonTokenType.String).Cat(Token.New(JsonTokenType.NameSeparator)).Cat(valueExpression); |
|
52 | var memberExpression = Token.New(JsonTokenType.String).Cat(Token.New(JsonTokenType.NameSeparator)).Cat(valueExpression); | |
| 40 | var objectExpression = memberExpression |
|
53 | var objectExpression = memberExpression | |
| 41 | .Cat( |
|
54 | .Cat( | |
| 42 | Token.New(JsonTokenType.ValueSeparator) |
|
55 | Token.New(JsonTokenType.ValueSeparator) | |
| 43 | .Cat(memberExpression) |
|
56 | .Cat(memberExpression) | |
| 44 | .EClosure() |
|
57 | .EClosure() | |
| 45 | ) |
|
58 | ) | |
| 46 | .Optional() |
|
59 | .Optional() | |
| 47 | .Cat(Token.New(JsonTokenType.EndObject)) |
|
60 | .Cat(Token.New(JsonTokenType.EndObject)) | |
| 48 | .Tag(0); |
|
61 | .Tag(0); | |
| 49 | var arrayExpression = valueExpression |
|
62 | var arrayExpression = valueExpression | |
| 50 | .Cat( |
|
63 | .Cat( | |
| 51 | Token.New(JsonTokenType.ValueSeparator) |
|
64 | Token.New(JsonTokenType.ValueSeparator) | |
| 52 | .Cat(valueExpression) |
|
65 | .Cat(valueExpression) | |
| 53 | .EClosure() |
|
66 | .EClosure() | |
| 54 | ) |
|
67 | ) | |
| 55 | .Optional() |
|
68 | .Optional() | |
| 56 | .Cat(Token.New(JsonTokenType.EndArray)) |
|
69 | .Cat(Token.New(JsonTokenType.EndArray)) | |
| 57 | .Tag(0); |
|
70 | .Tag(0); | |
| 58 |
|
71 | |||
| 59 | _jsonDFA = BuildDFA(jsonExpression).States; |
|
72 | _jsonDFA = BuildDFA(jsonExpression).States; | |
| 60 | _objectDFA = BuildDFA(objectExpression).States; |
|
73 | _objectDFA = BuildDFA(objectExpression).States; | |
| 61 | _arrayDFA = BuildDFA(arrayExpression).States; |
|
74 | _arrayDFA = BuildDFA(arrayExpression).States; | |
| 62 | } |
|
75 | } | |
| 63 |
|
76 | |||
| 64 | static EDFADefinition<JsonTokenType> BuildDFA(Token expr) { |
|
77 | static EDFADefinition<JsonTokenType> BuildDFA(Token expr) { | |
| 65 | var builder = new DFABuilder(); |
|
78 | var builder = new DFABuilder(); | |
| 66 | var dfa = new EDFADefinition<JsonTokenType>(_alphabet); |
|
79 | var dfa = new EDFADefinition<JsonTokenType>(_alphabet); | |
| 67 | expr.Accept(builder); |
|
80 | expr.Accept(builder); | |
| 68 |
|
81 | |||
| 69 | builder.BuildDFA(dfa); |
|
82 | builder.BuildDFA(dfa); | |
| 70 | return dfa; |
|
83 | return dfa; | |
| 71 | } |
|
84 | } | |
| 72 |
|
85 | |||
| 73 | JSONScanner m_scanner; |
|
86 | JSONScanner m_scanner; | |
| 74 | MemberContext m_memberContext; |
|
87 | MemberContext m_memberContext; | |
| 75 |
|
88 | |||
| 76 | JSONElementType m_elementType; |
|
89 | JSONElementType m_elementType; | |
| 77 | object m_elementValue; |
|
90 | object m_elementValue; | |
| 78 |
|
91 | |||
| 79 | /// <summary> |
|
92 | /// <summary> | |
| 80 | /// Создает новый парсер на основе строки, содержащей JSON |
|
93 | /// Создает новый парсер на основе строки, содержащей JSON | |
| 81 | /// </summary> |
|
94 | /// </summary> | |
| 82 | /// <param name="text"></param> |
|
95 | /// <param name="text"></param> | |
| 83 | public JSONParser(string text) |
|
96 | public JSONParser(string text) | |
| 84 | : base(_jsonDFA, INITIAL_STATE, new JSONParserContext { elementContext = JSONElementContext.None, memberName = String.Empty }) { |
|
97 | : base(_jsonDFA, INITIAL_STATE, new JSONParserContext { elementContext = JSONElementContext.None, memberName = String.Empty }) { | |
| 85 | Safe.ArgumentNotEmpty(text, "text"); |
|
98 | Safe.ArgumentNotEmpty(text, "text"); | |
| 86 | m_scanner = new JSONScanner(); |
|
99 | m_scanner = new JSONScanner(); | |
| 87 | m_scanner.Feed(text.ToCharArray()); |
|
100 | m_scanner.Feed(text.ToCharArray()); | |
| 88 | } |
|
101 | } | |
| 89 |
|
102 | |||
| 90 | /// <summary> |
|
103 | /// <summary> | |
| 91 | /// Создает новый экземпляр парсера, на основе текстового потока. |
|
104 | /// Создает новый экземпляр парсера, на основе текстового потока. | |
| 92 | /// </summary> |
|
105 | /// </summary> | |
| 93 | /// <param name="reader">Текстовый поток.</param> |
|
106 | /// <param name="reader">Текстовый поток.</param> | |
| 94 | /// <param name="dispose">Признак того, что парсер должен конролировать время жизни входного потока.</param> |
|
107 | /// <param name="dispose">Признак того, что парсер должен конролировать время жизни входного потока.</param> | |
| 95 | public JSONParser(TextReader reader, bool dispose) |
|
108 | public JSONParser(TextReader reader, bool dispose) | |
| 96 | : base(_jsonDFA, INITIAL_STATE, new JSONParserContext { elementContext = JSONElementContext.None, memberName = String.Empty }) { |
|
109 | : base(_jsonDFA, INITIAL_STATE, new JSONParserContext { elementContext = JSONElementContext.None, memberName = String.Empty }) { | |
| 97 | Safe.ArgumentNotNull(reader, "reader"); |
|
110 | Safe.ArgumentNotNull(reader, "reader"); | |
| 98 | m_scanner = new JSONScanner(); |
|
111 | m_scanner = new JSONScanner(); | |
| 99 | m_scanner.Feed(reader, dispose); |
|
112 | m_scanner.Feed(reader, dispose); | |
| 100 | } |
|
113 | } | |
| 101 |
|
114 | |||
| 102 | /// <summary> |
|
115 | /// <summary> | |
| 103 | /// Тип текущего элемента на котором стоит парсер. |
|
116 | /// Тип текущего элемента на котором стоит парсер. | |
| 104 | /// </summary> |
|
117 | /// </summary> | |
| 105 | public JSONElementType ElementType { |
|
118 | public JSONElementType ElementType { | |
| 106 | get { return m_elementType; } |
|
119 | get { return m_elementType; } | |
| 107 | } |
|
120 | } | |
| 108 |
|
121 | |||
| 109 | /// <summary> |
|
122 | /// <summary> | |
| 110 | /// Имя элемента - имя свойства родительского контейнера. Для элементов массивов и корневого всегда |
|
123 | /// Имя элемента - имя свойства родительского контейнера. Для элементов массивов и корневого всегда | |
| 111 | /// пустая строка. |
|
124 | /// пустая строка. | |
| 112 | /// </summary> |
|
125 | /// </summary> | |
| 113 | public string ElementName { |
|
126 | public string ElementName { | |
| 114 | get { return m_context.info.memberName; } |
|
127 | get { return m_context.info.memberName; } | |
| 115 | } |
|
128 | } | |
| 116 |
|
129 | |||
| 117 | /// <summary> |
|
130 | /// <summary> | |
| 118 | /// Значение элемента. Только для элементов типа <see cref="JSONElementType.Value"/>, для остальных <c>null</c> |
|
131 | /// Значение элемента. Только для элементов типа <see cref="JSONElementType.Value"/>, для остальных <c>null</c> | |
| 119 | /// </summary> |
|
132 | /// </summary> | |
| 120 | public object ElementValue { |
|
133 | public object ElementValue { | |
| 121 | get { return m_elementValue; } |
|
134 | get { return m_elementValue; } | |
| 122 | } |
|
135 | } | |
| 123 |
|
136 | |||
| 124 | /// <summary> |
|
137 | /// <summary> | |
| 125 | /// Читает слеюудущий объект из потока |
|
138 | /// Читает слеюудущий объект из потока | |
| 126 | /// </summary> |
|
139 | /// </summary> | |
| 127 | /// <returns><c>true</c> - операция чтения прошла успешно, <c>false</c> - конец данных</returns> |
|
140 | /// <returns><c>true</c> - операция чтения прошла успешно, <c>false</c> - конец данных</returns> | |
| 128 | public bool Read() { |
|
141 | public bool Read() { | |
| 129 | if (m_context.current == UNREACHEBLE_STATE) |
|
142 | if (m_context.current == UNREACHEBLE_STATE) | |
| 130 | throw new InvalidOperationException("The parser is in invalid state"); |
|
143 | throw new InvalidOperationException("The parser is in invalid state"); | |
| 131 | object tokenValue; |
|
144 | object tokenValue; | |
| 132 | JsonTokenType tokenType; |
|
145 | JsonTokenType tokenType; | |
| 133 | m_context.info.memberName = String.Empty; |
|
146 | m_context.info.memberName = String.Empty; | |
| 134 | while (m_scanner.ReadToken(out tokenValue, out tokenType)) { |
|
147 | while (m_scanner.ReadToken(out tokenValue, out tokenType)) { | |
| 135 | Move((int)tokenType); |
|
148 | Move((int)tokenType); | |
| 136 | if (m_context.current == UNREACHEBLE_STATE) |
|
149 | if (m_context.current == UNREACHEBLE_STATE) | |
| 137 | UnexpectedToken(tokenValue, tokenType); |
|
150 | UnexpectedToken(tokenValue, tokenType); | |
| 138 | switch (tokenType) { |
|
151 | switch (tokenType) { | |
| 139 | case JsonTokenType.BeginObject: |
|
152 | case JsonTokenType.BeginObject: | |
| 140 | Switch( |
|
153 | Switch( | |
| 141 | _objectDFA, |
|
154 | _objectDFA, | |
| 142 | INITIAL_STATE, |
|
155 | INITIAL_STATE, | |
| 143 | new JSONParserContext { |
|
156 | new JSONParserContext { | |
| 144 | memberName = m_context.info.memberName, |
|
157 | memberName = m_context.info.memberName, | |
| 145 | elementContext = JSONElementContext.Object |
|
158 | elementContext = JSONElementContext.Object | |
| 146 | } |
|
159 | } | |
| 147 | ); |
|
160 | ); | |
| 148 | m_elementValue = null; |
|
161 | m_elementValue = null; | |
| 149 | m_memberContext = MemberContext.MemberName; |
|
162 | m_memberContext = MemberContext.MemberName; | |
| 150 | m_elementType = JSONElementType.BeginObject; |
|
163 | m_elementType = JSONElementType.BeginObject; | |
| 151 | return true; |
|
164 | return true; | |
| 152 | case JsonTokenType.EndObject: |
|
165 | case JsonTokenType.EndObject: | |
| 153 | Restore(); |
|
166 | Restore(); | |
| 154 | m_elementValue = null; |
|
167 | m_elementValue = null; | |
| 155 | m_elementType = JSONElementType.EndObject; |
|
168 | m_elementType = JSONElementType.EndObject; | |
| 156 | return true; |
|
169 | return true; | |
| 157 | case JsonTokenType.BeginArray: |
|
170 | case JsonTokenType.BeginArray: | |
| 158 | Switch( |
|
171 | Switch( | |
| 159 | _arrayDFA, |
|
172 | _arrayDFA, | |
| 160 | INITIAL_STATE, |
|
173 | INITIAL_STATE, | |
| 161 | new JSONParserContext { |
|
174 | new JSONParserContext { | |
| 162 | memberName = m_context.info.memberName, |
|
175 | memberName = m_context.info.memberName, | |
| 163 | elementContext = JSONElementContext.Array |
|
176 | elementContext = JSONElementContext.Array | |
| 164 | } |
|
177 | } | |
| 165 | ); |
|
178 | ); | |
| 166 | m_elementValue = null; |
|
179 | m_elementValue = null; | |
| 167 | m_memberContext = MemberContext.MemberValue; |
|
180 | m_memberContext = MemberContext.MemberValue; | |
| 168 | m_elementType = JSONElementType.BeginArray; |
|
181 | m_elementType = JSONElementType.BeginArray; | |
| 169 | return true; |
|
182 | return true; | |
| 170 | case JsonTokenType.EndArray: |
|
183 | case JsonTokenType.EndArray: | |
| 171 | Restore(); |
|
184 | Restore(); | |
| 172 | m_elementValue = null; |
|
185 | m_elementValue = null; | |
| 173 | m_elementType = JSONElementType.EndArray; |
|
186 | m_elementType = JSONElementType.EndArray; | |
| 174 | return true; |
|
187 | return true; | |
| 175 | case JsonTokenType.String: |
|
188 | case JsonTokenType.String: | |
| 176 | if (m_memberContext == MemberContext.MemberName) { |
|
189 | if (m_memberContext == MemberContext.MemberName) { | |
| 177 | m_context.info.memberName = (string)tokenValue; |
|
190 | m_context.info.memberName = (string)tokenValue; | |
| 178 | break; |
|
191 | break; | |
| 179 | } else { |
|
192 | } else { | |
| 180 | m_elementType = JSONElementType.Value; |
|
193 | m_elementType = JSONElementType.Value; | |
| 181 | m_elementValue = tokenValue; |
|
194 | m_elementValue = tokenValue; | |
| 182 | return true; |
|
195 | return true; | |
| 183 | } |
|
196 | } | |
| 184 | case JsonTokenType.Number: |
|
197 | case JsonTokenType.Number: | |
| 185 | m_elementType = JSONElementType.Value; |
|
198 | m_elementType = JSONElementType.Value; | |
| 186 | m_elementValue = tokenValue; |
|
199 | m_elementValue = tokenValue; | |
| 187 | return true; |
|
200 | return true; | |
| 188 | case JsonTokenType.Literal: |
|
201 | case JsonTokenType.Literal: | |
| 189 | m_elementType = JSONElementType.Value; |
|
202 | m_elementType = JSONElementType.Value; | |
| 190 | m_elementValue = ParseLiteral((string)tokenValue); |
|
203 | m_elementValue = ParseLiteral((string)tokenValue); | |
| 191 | return true; |
|
204 | return true; | |
| 192 | case JsonTokenType.NameSeparator: |
|
205 | case JsonTokenType.NameSeparator: | |
| 193 | m_memberContext = MemberContext.MemberValue; |
|
206 | m_memberContext = MemberContext.MemberValue; | |
| 194 | break; |
|
207 | break; | |
| 195 | case JsonTokenType.ValueSeparator: |
|
208 | case JsonTokenType.ValueSeparator: | |
| 196 | m_memberContext = m_context.info.elementContext == JSONElementContext.Object ? MemberContext.MemberName : MemberContext.MemberValue; |
|
209 | m_memberContext = m_context.info.elementContext == JSONElementContext.Object ? MemberContext.MemberName : MemberContext.MemberValue; | |
| 197 | break; |
|
210 | break; | |
| 198 | default: |
|
211 | default: | |
| 199 | UnexpectedToken(tokenValue, tokenType); |
|
212 | UnexpectedToken(tokenValue, tokenType); | |
| 200 | break; |
|
213 | break; | |
| 201 | } |
|
214 | } | |
| 202 | } |
|
215 | } | |
| 203 | if (m_context.info.elementContext != JSONElementContext.None) |
|
216 | if (m_context.info.elementContext != JSONElementContext.None) | |
| 204 | throw new ParserException("Unexpedted end of data"); |
|
217 | throw new ParserException("Unexpedted end of data"); | |
| 205 | return false; |
|
218 | return false; | |
| 206 | } |
|
219 | } | |
| 207 |
|
220 | |||
| 208 | object ParseLiteral(string literal) { |
|
221 | object ParseLiteral(string literal) { | |
| 209 | switch (literal) { |
|
222 | switch (literal) { | |
| 210 | case "null": |
|
223 | case "null": | |
| 211 | return null; |
|
224 | return null; | |
| 212 | case "false": |
|
225 | case "false": | |
| 213 | return false; |
|
226 | return false; | |
| 214 | case "true": |
|
227 | case "true": | |
| 215 | return true; |
|
228 | return true; | |
| 216 | default: |
|
229 | default: | |
| 217 | UnexpectedToken(literal, JsonTokenType.Literal); |
|
230 | UnexpectedToken(literal, JsonTokenType.Literal); | |
| 218 | return null; // avoid compliler error |
|
231 | return null; // avoid compliler error | |
| 219 | } |
|
232 | } | |
| 220 | } |
|
233 | } | |
| 221 |
|
234 | |||
| 222 | void UnexpectedToken(object value, JsonTokenType tokenType) { |
|
235 | void UnexpectedToken(object value, JsonTokenType tokenType) { | |
| 223 | throw new ParserException(String.Format("Unexpected token {0}: '{1}'", tokenType, value)); |
|
236 | throw new ParserException(String.Format("Unexpected token {0}: '{1}'", tokenType, value)); | |
| 224 | } |
|
237 | } | |
| 225 |
|
238 | |||
| 226 |
|
239 | |||
| 227 | /// <summary> |
|
240 | /// <summary> | |
| 228 | /// Признак конца потока |
|
241 | /// Признак конца потока | |
| 229 | /// </summary> |
|
242 | /// </summary> | |
| 230 | public bool EOF { |
|
243 | public bool EOF { | |
| 231 | get { |
|
244 | get { | |
| 232 | return m_scanner.EOF; |
|
245 | return m_scanner.EOF; | |
| 233 | } |
|
246 | } | |
| 234 | } |
|
247 | } | |
| 235 |
|
248 | |||
| 236 | protected virtual void Dispose(bool disposing) { |
|
249 | protected virtual void Dispose(bool disposing) { | |
| 237 | if (disposing) { |
|
250 | if (disposing) { | |
| 238 | m_scanner.Dispose(); |
|
251 | m_scanner.Dispose(); | |
| 239 | } |
|
252 | } | |
| 240 | } |
|
253 | } | |
| 241 |
|
254 | |||
| 242 | /// <summary> |
|
255 | /// <summary> | |
| 243 | /// Освобождает парсер и связанный с ним сканнер. |
|
256 | /// Освобождает парсер и связанный с ним сканнер. | |
| 244 | /// </summary> |
|
257 | /// </summary> | |
| 245 | public void Dispose() { |
|
258 | public void Dispose() { | |
| 246 | Dispose(true); |
|
259 | Dispose(true); | |
| 247 | GC.SuppressFinalize(this); |
|
260 | GC.SuppressFinalize(this); | |
| 248 | } |
|
261 | } | |
| 249 |
|
262 | |||
| 250 | ~JSONParser() { |
|
263 | ~JSONParser() { | |
| 251 | Dispose(false); |
|
264 | Dispose(false); | |
| 252 | } |
|
265 | } | |
| 253 |
|
266 | |||
| 254 | public void Skip() { |
|
267 | /// <summary> | |
| 255 | var level = Level-1; |
|
268 | /// Переходит в конец текущего объекта. | |
|
|
269 | /// </summary> | |||
|
|
270 | public void SeekElementEnd() { | |||
|
|
271 | var level = Level - 1; | |||
| 256 |
|
272 | |||
| 257 | Debug.Assert(level >= 0); |
|
273 | Debug.Assert(level >= 0); | |
| 258 |
|
274 | |||
| 259 | while (Level != level) |
|
275 | while (Level != level) | |
| 260 | Read(); |
|
276 | Read(); | |
| 261 | } |
|
277 | } | |
| 262 | } |
|
278 | } | |
| 263 |
|
279 | |||
| 264 | } |
|
280 | } | |
General Comments 0
You need to be logged in to leave comments.
Login now
