Mercurial > pub > ImplabNet
annotate Implab/JSON/JSONParser.cs @ 89:ce0171cacec4 v2
improved performance of a chained map operation
| author | cin |
|---|---|
| date | Wed, 08 Oct 2014 02:19:45 +0400 |
| parents | 0349189d2564 |
| children | f973c5df9972 |
| rev | line source |
|---|---|
| 55 | 1 using Implab; |
| 2 using Implab.Parsing; | |
| 3 using System; | |
| 4 using System.Collections.Generic; | |
| 5 using System.Diagnostics; | |
| 59 | 6 using System.IO; |
| 55 | 7 using System.Linq; |
| 8 using System.Text; | |
| 9 using System.Threading.Tasks; | |
| 10 | |
| 11 namespace Implab.JSON { | |
| 12 /// <summary> | |
| 13 /// internal | |
| 14 /// </summary> | |
| 15 public struct JSONParserContext { | |
| 16 public string memberName; | |
| 17 public JSONElementContext elementContext; | |
| 18 } | |
| 19 | |
| 20 /// <summary> | |
| 21 /// Pull парсер JSON данных. | |
| 22 /// </summary> | |
| 70 | 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> | |
| 59 | 36 public class JSONParser : DFAutomaton<JSONParserContext>, IDisposable { |
| 55 | 37 |
| 38 enum MemberContext { | |
| 39 MemberName, | |
| 40 MemberValue | |
| 59 | 41 } |
| 55 | 42 |
| 43 static readonly EnumAlphabet<JsonTokenType> _alphabet = EnumAlphabet<JsonTokenType>.FullAlphabet; | |
| 44 static readonly DFAStateDescriptior[] _jsonDFA; | |
| 45 static readonly DFAStateDescriptior[] _objectDFA; | |
| 46 static readonly DFAStateDescriptior[] _arrayDFA; | |
| 47 | |
| 48 static JSONParser() { | |
| 49 var jsonExpression = Token.New(JsonTokenType.BeginObject, JsonTokenType.BeginArray).Tag(0); | |
| 50 | |
| 51 var valueExpression = Token.New(JsonTokenType.BeginArray, JsonTokenType.BeginObject, JsonTokenType.Literal, JsonTokenType.Number, JsonTokenType.String); | |
| 52 var memberExpression = Token.New(JsonTokenType.String).Cat(Token.New(JsonTokenType.NameSeparator)).Cat(valueExpression); | |
| 53 var objectExpression = memberExpression | |
| 54 .Cat( | |
| 55 Token.New(JsonTokenType.ValueSeparator) | |
| 56 .Cat(memberExpression) | |
| 57 .EClosure() | |
| 58 ) | |
| 59 .Optional() | |
| 60 .Cat(Token.New(JsonTokenType.EndObject)) | |
| 61 .Tag(0); | |
| 62 var arrayExpression = valueExpression | |
| 63 .Cat( | |
| 64 Token.New(JsonTokenType.ValueSeparator) | |
| 65 .Cat(valueExpression) | |
| 66 .EClosure() | |
| 67 ) | |
| 68 .Optional() | |
| 69 .Cat(Token.New(JsonTokenType.EndArray)) | |
| 70 .Tag(0); | |
| 71 | |
| 72 _jsonDFA = BuildDFA(jsonExpression).States; | |
| 73 _objectDFA = BuildDFA(objectExpression).States; | |
| 74 _arrayDFA = BuildDFA(arrayExpression).States; | |
| 75 } | |
| 76 | |
| 77 static EDFADefinition<JsonTokenType> BuildDFA(Token expr) { | |
| 78 var builder = new DFABuilder(); | |
| 79 var dfa = new EDFADefinition<JsonTokenType>(_alphabet); | |
| 80 expr.Accept(builder); | |
| 81 | |
| 82 builder.BuildDFA(dfa); | |
| 83 return dfa; | |
| 84 } | |
| 85 | |
| 86 JSONScanner m_scanner; | |
| 87 MemberContext m_memberContext; | |
| 88 | |
| 89 JSONElementType m_elementType; | |
| 90 object m_elementValue; | |
| 91 | |
| 59 | 92 /// <summary> |
| 93 /// Создает новый парсер на основе строки, содержащей JSON | |
| 94 /// </summary> | |
| 95 /// <param name="text"></param> | |
| 55 | 96 public JSONParser(string text) |
| 59 | 97 : base(_jsonDFA, INITIAL_STATE, new JSONParserContext { elementContext = JSONElementContext.None, memberName = String.Empty }) { |
| 55 | 98 Safe.ArgumentNotEmpty(text, "text"); |
| 99 m_scanner = new JSONScanner(); | |
| 100 m_scanner.Feed(text.ToCharArray()); | |
| 101 } | |
| 102 | |
| 59 | 103 /// <summary> |
| 104 /// Создает новый экземпляр парсера, на основе текстового потока. | |
| 105 /// </summary> | |
| 106 /// <param name="reader">Текстовый поток.</param> | |
| 107 /// <param name="dispose">Признак того, что парсер должен конролировать время жизни входного потока.</param> | |
| 108 public JSONParser(TextReader reader, bool dispose) | |
| 109 : base(_jsonDFA, INITIAL_STATE, new JSONParserContext { elementContext = JSONElementContext.None, memberName = String.Empty }) { | |
| 110 Safe.ArgumentNotNull(reader, "reader"); | |
| 111 m_scanner = new JSONScanner(); | |
| 112 m_scanner.Feed(reader, dispose); | |
| 113 } | |
| 114 | |
| 115 /// <summary> | |
| 116 /// Тип текущего элемента на котором стоит парсер. | |
| 117 /// </summary> | |
| 55 | 118 public JSONElementType ElementType { |
| 119 get { return m_elementType; } | |
| 120 } | |
| 121 | |
| 59 | 122 /// <summary> |
| 123 /// Имя элемента - имя свойства родительского контейнера. Для элементов массивов и корневого всегда | |
| 124 /// пустая строка. | |
| 125 /// </summary> | |
| 55 | 126 public string ElementName { |
| 127 get { return m_context.info.memberName; } | |
| 128 } | |
| 129 | |
| 59 | 130 /// <summary> |
| 131 /// Значение элемента. Только для элементов типа <see cref="JSONElementType.Value"/>, для остальных <c>null</c> | |
| 132 /// </summary> | |
| 55 | 133 public object ElementValue { |
| 134 get { return m_elementValue; } | |
| 135 } | |
| 136 | |
| 59 | 137 /// <summary> |
| 138 /// Читает слеюудущий объект из потока | |
| 139 /// </summary> | |
| 140 /// <returns><c>true</c> - операция чтения прошла успешно, <c>false</c> - конец данных</returns> | |
| 55 | 141 public bool Read() { |
| 142 if (m_context.current == UNREACHEBLE_STATE) | |
| 143 throw new InvalidOperationException("The parser is in invalid state"); | |
| 144 object tokenValue; | |
| 145 JsonTokenType tokenType; | |
| 146 m_context.info.memberName = String.Empty; | |
| 147 while (m_scanner.ReadToken(out tokenValue, out tokenType)) { | |
| 148 Move((int)tokenType); | |
| 149 if (m_context.current == UNREACHEBLE_STATE) | |
| 150 UnexpectedToken(tokenValue, tokenType); | |
| 151 switch (tokenType) { | |
| 152 case JsonTokenType.BeginObject: | |
| 153 Switch( | |
| 154 _objectDFA, | |
| 155 INITIAL_STATE, | |
| 59 | 156 new JSONParserContext { |
| 55 | 157 memberName = m_context.info.memberName, |
| 158 elementContext = JSONElementContext.Object | |
| 159 } | |
| 160 ); | |
| 161 m_elementValue = null; | |
| 162 m_memberContext = MemberContext.MemberName; | |
| 163 m_elementType = JSONElementType.BeginObject; | |
| 164 return true; | |
| 165 case JsonTokenType.EndObject: | |
| 166 Restore(); | |
| 167 m_elementValue = null; | |
| 168 m_elementType = JSONElementType.EndObject; | |
| 169 return true; | |
| 170 case JsonTokenType.BeginArray: | |
| 171 Switch( | |
| 172 _arrayDFA, | |
| 173 INITIAL_STATE, | |
| 174 new JSONParserContext { | |
| 175 memberName = m_context.info.memberName, | |
| 176 elementContext = JSONElementContext.Array | |
| 177 } | |
| 178 ); | |
| 179 m_elementValue = null; | |
| 180 m_memberContext = MemberContext.MemberValue; | |
| 181 m_elementType = JSONElementType.BeginArray; | |
| 182 return true; | |
| 183 case JsonTokenType.EndArray: | |
| 184 Restore(); | |
| 185 m_elementValue = null; | |
| 186 m_elementType = JSONElementType.EndArray; | |
| 187 return true; | |
| 188 case JsonTokenType.String: | |
| 189 if (m_memberContext == MemberContext.MemberName) { | |
| 190 m_context.info.memberName = (string)tokenValue; | |
| 191 break; | |
| 192 } else { | |
| 193 m_elementType = JSONElementType.Value; | |
| 194 m_elementValue = tokenValue; | |
| 195 return true; | |
| 196 } | |
| 197 case JsonTokenType.Number: | |
| 198 m_elementType = JSONElementType.Value; | |
| 199 m_elementValue = tokenValue; | |
| 200 return true; | |
| 201 case JsonTokenType.Literal: | |
| 202 m_elementType = JSONElementType.Value; | |
| 203 m_elementValue = ParseLiteral((string)tokenValue); | |
| 204 return true; | |
| 205 case JsonTokenType.NameSeparator: | |
| 206 m_memberContext = MemberContext.MemberValue; | |
| 207 break; | |
| 208 case JsonTokenType.ValueSeparator: | |
| 209 m_memberContext = m_context.info.elementContext == JSONElementContext.Object ? MemberContext.MemberName : MemberContext.MemberValue; | |
| 210 break; | |
| 211 default: | |
| 212 UnexpectedToken(tokenValue, tokenType); | |
| 213 break; | |
| 214 } | |
| 215 } | |
| 216 if (m_context.info.elementContext != JSONElementContext.None) | |
| 217 throw new ParserException("Unexpedted end of data"); | |
| 218 return false; | |
| 219 } | |
| 220 | |
| 221 object ParseLiteral(string literal) { | |
| 222 switch (literal) { | |
| 223 case "null": | |
| 224 return null; | |
| 59 | 225 case "false": |
| 55 | 226 return false; |
| 227 case "true": | |
| 228 return true; | |
| 229 default: | |
| 230 UnexpectedToken(literal, JsonTokenType.Literal); | |
| 231 return null; // avoid compliler error | |
| 232 } | |
| 233 } | |
| 234 | |
| 235 void UnexpectedToken(object value, JsonTokenType tokenType) { | |
| 236 throw new ParserException(String.Format("Unexpected token {0}: '{1}'", tokenType, value)); | |
| 237 } | |
| 238 | |
| 57 | 239 |
| 59 | 240 /// <summary> |
| 241 /// Признак конца потока | |
| 242 /// </summary> | |
| 57 | 243 public bool EOF { |
| 244 get { | |
| 245 return m_scanner.EOF; | |
| 246 } | |
| 247 } | |
| 59 | 248 |
| 249 protected virtual void Dispose(bool disposing) { | |
| 250 if (disposing) { | |
| 251 m_scanner.Dispose(); | |
| 252 } | |
| 253 } | |
| 254 | |
| 255 /// <summary> | |
| 256 /// Освобождает парсер и связанный с ним сканнер. | |
| 257 /// </summary> | |
| 258 public void Dispose() { | |
| 259 Dispose(true); | |
| 260 GC.SuppressFinalize(this); | |
| 261 } | |
| 262 | |
| 263 ~JSONParser() { | |
| 264 Dispose(false); | |
| 265 } | |
|
62
62b440d46313
Added Skip method to JSON parser to skip contents of the current node
cin
parents:
59
diff
changeset
|
266 |
| 70 | 267 /// <summary> |
| 268 /// Переходит в конец текущего объекта. | |
| 269 /// </summary> | |
| 270 public void SeekElementEnd() { | |
| 271 var level = Level - 1; | |
|
62
62b440d46313
Added Skip method to JSON parser to skip contents of the current node
cin
parents:
59
diff
changeset
|
272 |
|
62b440d46313
Added Skip method to JSON parser to skip contents of the current node
cin
parents:
59
diff
changeset
|
273 Debug.Assert(level >= 0); |
|
62b440d46313
Added Skip method to JSON parser to skip contents of the current node
cin
parents:
59
diff
changeset
|
274 |
|
62b440d46313
Added Skip method to JSON parser to skip contents of the current node
cin
parents:
59
diff
changeset
|
275 while (Level != level) |
|
62b440d46313
Added Skip method to JSON parser to skip contents of the current node
cin
parents:
59
diff
changeset
|
276 Read(); |
|
62b440d46313
Added Skip method to JSON parser to skip contents of the current node
cin
parents:
59
diff
changeset
|
277 } |
| 55 | 278 } |
| 279 | |
| 280 } |