Mercurial > pub > ImplabNet
comparison Implab/Formats/JSON/JsonScanner.cs @ 228:6fa235c5a760 v2
Rewritten JsonScanner, JsonParser, fixed naming style
| author | cin |
|---|---|
| date | Tue, 12 Sep 2017 01:19:12 +0300 |
| parents | Implab/Formats/JSON/JSONScanner.cs@7d07503621fe |
| children | 5f7a3e1d32b9 |
comparison
equal
deleted
inserted
replaced
| 227:8d5de4eb9c2c | 228:6fa235c5a760 |
|---|---|
| 1 using System; | |
| 2 using System.Globalization; | |
| 3 using Implab.Automaton; | |
| 4 using System.Text; | |
| 5 using Implab.Components; | |
| 6 using System.IO; | |
| 7 | |
| 8 namespace Implab.Formats.Json { | |
| 9 /// <summary> | |
| 10 /// Сканнер (лексер), разбивающий поток символов на токены JSON. | |
| 11 /// </summary> | |
| 12 public abstract class JsonScanner : Disposable { | |
| 13 readonly InputScanner<JsonGrammar.TokenType> m_jsonContext = JsonGrammar.CreateJsonExpressionScanner(); | |
| 14 readonly InputScanner<JsonGrammar.TokenType> m_stringContext = JsonGrammar.CreateStringExpressionScanner(); | |
| 15 | |
| 16 readonly char[] m_unescapeBuf = new char[4]; | |
| 17 readonly char[] m_buffer; | |
| 18 int m_length; | |
| 19 int m_pos; | |
| 20 readonly StringBuilder m_tokenBuilder = new StringBuilder(); | |
| 21 | |
| 22 protected JsonScanner(char[] buffer, int pos, int length) { | |
| 23 m_buffer = buffer; | |
| 24 m_pos = pos; | |
| 25 m_length = length; | |
| 26 } | |
| 27 | |
| 28 bool Read(InputScanner<JsonGrammar.TokenType> scanner, out JsonGrammar.TokenType tokenType) { | |
| 29 scanner.Reset(); | |
| 30 | |
| 31 if (m_pos == m_length) { | |
| 32 m_pos = 0; | |
| 33 m_length = Read(m_buffer, 0, m_buffer.Length); | |
| 34 if (m_length == 0) { | |
| 35 tokenType = JsonGrammar.TokenType.None; | |
| 36 return false; // EOF | |
| 37 } | |
| 38 } | |
| 39 | |
| 40 while(scanner.Scan(m_buffer, m_pos, m_length - m_pos)) { | |
| 41 m_tokenBuilder.Append(m_buffer, m_pos, m_length - m_pos); | |
| 42 m_pos = 0; | |
| 43 m_length = Read(m_buffer, 0, m_buffer.Length); | |
| 44 } | |
| 45 var scannerPos = scanner.Position; | |
| 46 if (scannerPos != m_pos) { | |
| 47 m_tokenBuilder.Append(m_buffer, m_pos, scannerPos - m_pos); | |
| 48 m_pos = scannerPos; | |
| 49 } | |
| 50 | |
| 51 if (!scanner.IsFinal) { | |
| 52 if (m_length == 0) { | |
| 53 // unexpected EOF | |
| 54 throw new ParserException("Unexpected EOF"); | |
| 55 } else { | |
| 56 // unecpected character | |
| 57 throw new ParserException($"Unexpected character '{m_buffer[m_pos + 1]}'"); | |
| 58 } | |
| 59 } | |
| 60 tokenType = scanner.Tag; | |
| 61 return true; | |
| 62 } | |
| 63 | |
| 64 protected abstract int Read(char[] buffer, int offset, int size); | |
| 65 | |
| 66 | |
| 67 /// <summary> | |
| 68 /// Читает следующий лексический элемент из входных данных. | |
| 69 /// </summary> | |
| 70 /// <param name="tokenValue">Возвращает значение прочитанного токена.</param> | |
| 71 /// <param name="tokenType">Возвращает тип прочитанного токена.</param> | |
| 72 /// <returns><c>true</c> - чтение произведено успешно. <c>false</c> - достигнут конец входных данных</returns> | |
| 73 /// <remarks>В случе если токен не распознается, возникает исключение. Значения токенов обрабатываются, т.е. | |
| 74 /// в строках обрабатываются экранированные символы, числа становтся типа double.</remarks> | |
| 75 public bool ReadToken(out object tokenValue, out JsonTokenType tokenType) { | |
| 76 JsonGrammar.TokenType tag; | |
| 77 m_tokenBuilder.Clear(); | |
| 78 while (Read(m_jsonContext, out tag)) { | |
| 79 switch (tag) { | |
| 80 case JsonGrammar.TokenType.StringBound: | |
| 81 tokenValue = ReadString(); | |
| 82 tokenType = JsonTokenType.String; | |
| 83 break; | |
| 84 case JsonGrammar.TokenType.Number: | |
| 85 tokenValue = Double.Parse(m_tokenBuilder.ToString(), CultureInfo.InvariantCulture); | |
| 86 tokenType = JsonTokenType.Number; | |
| 87 break; | |
| 88 case JsonGrammar.TokenType.Literal: | |
| 89 tokenType = JsonTokenType.Literal; | |
| 90 tokenValue = m_tokenBuilder.ToString(); | |
| 91 break; | |
| 92 case JsonGrammar.TokenType.Whitespace: | |
| 93 m_tokenBuilder.Clear(); | |
| 94 continue; | |
| 95 default: | |
| 96 tokenType = (JsonTokenType)tag; | |
| 97 tokenValue = null; | |
| 98 break; | |
| 99 } | |
| 100 return true; | |
| 101 } | |
| 102 tokenValue = null; | |
| 103 tokenType = JsonTokenType.None; | |
| 104 return false; | |
| 105 } | |
| 106 | |
| 107 string ReadString() { | |
| 108 JsonGrammar.TokenType tag; | |
| 109 m_tokenBuilder.Clear(); | |
| 110 | |
| 111 while (Read(m_stringContext, out tag)) { | |
| 112 switch (tag) { | |
| 113 case JsonGrammar.TokenType.StringBound: | |
| 114 m_tokenBuilder.Length--; | |
| 115 return m_tokenBuilder.ToString(); | |
| 116 case JsonGrammar.TokenType.UnescapedChar: | |
| 117 break; | |
| 118 case JsonGrammar.TokenType.EscapedUnicode: // \xXXXX - unicode escape sequence | |
| 119 m_tokenBuilder.CopyTo(m_tokenBuilder.Length - 4, m_unescapeBuf, 0, 4); | |
| 120 m_tokenBuilder.Length -= 6; | |
| 121 m_tokenBuilder.Append(StringTranslator.TranslateHexUnicode(m_unescapeBuf, 0)); | |
| 122 break; | |
| 123 case JsonGrammar.TokenType.EscapedChar: // \t - escape sequence | |
| 124 var ch = m_tokenBuilder[m_tokenBuilder.Length-1]; | |
| 125 m_tokenBuilder.Length -= 2; | |
| 126 m_tokenBuilder.Append(StringTranslator.TranslateEscapedChar(ch)); | |
| 127 break; | |
| 128 } | |
| 129 } | |
| 130 | |
| 131 throw new ParserException("Unexpected end of data"); | |
| 132 } | |
| 133 } | |
| 134 } |
