|
173
|
1 using System;
|
|
|
2 using Implab.Components;
|
|
175
|
3 using Implab.Automaton.RegularExpressions;
|
|
|
4 using System.Diagnostics;
|
|
|
5 using Implab.Automaton;
|
|
176
|
6 using System.IO;
|
|
|
7 using System.Text;
|
|
173
|
8
|
|
|
9 namespace Implab.Formats {
|
|
176
|
10 public abstract class TextScanner : Disposable {
|
|
|
11 readonly int m_bufferMax;
|
|
|
12 readonly int m_chunkSize;
|
|
173
|
13
|
|
176
|
14 char[] m_buffer;
|
|
174
|
15 int m_bufferOffset;
|
|
175
|
16 int m_bufferSize;
|
|
176
|
17 int m_tokenOffset;
|
|
173
|
18 int m_tokenLength;
|
|
174
|
19
|
|
176
|
20 /// <summary>
|
|
|
21 /// Initializes a new instance of the <see cref="Implab.Formats.TextScanner`1"/> class.
|
|
|
22 /// </summary>
|
|
|
23 /// <param name="bufferMax">Buffer max.</param>
|
|
|
24 /// <param name="chunkSize">Chunk size.</param>
|
|
|
25 protected TextScanner(int bufferMax, int chunkSize) {
|
|
|
26 Debug.Assert(m_chunkSize <= m_bufferMax);
|
|
|
27
|
|
|
28 m_bufferMax = bufferMax;
|
|
|
29 m_chunkSize = chunkSize;
|
|
|
30 }
|
|
173
|
31
|
|
176
|
32 /// <summary>
|
|
|
33 /// Initializes a new instance of the <see cref="Implab.Formats.TextScanner`1"/> class.
|
|
|
34 /// </summary>
|
|
|
35 /// <param name="buffer">Buffer.</param>
|
|
|
36 protected TextScanner(char[] buffer) {
|
|
|
37 if (buffer != null) {
|
|
|
38 m_buffer = buffer;
|
|
|
39 m_bufferSize = buffer.Length;
|
|
|
40 }
|
|
|
41 }
|
|
|
42
|
|
|
43 /// <summary>
|
|
|
44 /// (hungry) Reads the next token.
|
|
|
45 /// </summary>
|
|
|
46 /// <returns><c>true</c>, if token internal was read, <c>false</c> if there is no more tokens in the stream.</returns>
|
|
|
47 /// <param name="dfa">The transition map for the automaton</param>
|
|
|
48 /// <param name="final">Final states of the automaton.</param>
|
|
|
49 /// <param name="tags">Tags.</param>
|
|
|
50 /// <param name="state">The initial state for the automaton.</param>
|
|
|
51 internal bool ReadToken<TTag>(int[,] dfa, int[] final, TTag[][] tags, int state, int[] alphabet, out TTag[] tag) {
|
|
|
52 Safe.ArgumentNotNull();
|
|
|
53 m_tokenLength = 0;
|
|
|
54
|
|
|
55 var maxSymbol = alphabet.Length - 1;
|
|
174
|
56
|
|
175
|
57 do {
|
|
176
|
58 // after the next chunk is read the offset in the buffer may change
|
|
|
59 int pos = m_bufferOffset + m_tokenLength;
|
|
|
60
|
|
|
61 while(pos < m_bufferSize) {
|
|
175
|
62 var ch = m_buffer[pos];
|
|
176
|
63
|
|
|
64 state = dfa[state,ch > maxSymbol ? DFAConst.UNCLASSIFIED_INPUT : alphabet[ch]];
|
|
175
|
65 if (state == DFAConst.UNREACHABLE_STATE)
|
|
|
66 break;
|
|
176
|
67
|
|
|
68 pos++;
|
|
175
|
69 }
|
|
176
|
70
|
|
|
71 m_tokenLength = pos - m_bufferOffset;
|
|
|
72 } while (state != DFAConst.UNREACHABLE_STATE && Feed());
|
|
|
73
|
|
|
74 m_tokenOffset = m_bufferOffset;
|
|
|
75 m_bufferOffset += m_tokenLength;
|
|
174
|
76
|
|
176
|
77 if (final[state]) {
|
|
|
78 tag = tags[state];
|
|
|
79 return true;
|
|
|
80 } else {
|
|
|
81 if (m_bufferOffset == m_bufferSize) {
|
|
|
82 if (m_tokenLength == 0) //EOF
|
|
|
83 return false;
|
|
|
84
|
|
|
85 throw new ParserException();
|
|
|
86 }
|
|
|
87 throw new ParserException(String.Format("Unexpected symbol '{0}'", m_buffer[m_bufferOffset]));
|
|
|
88
|
|
|
89 }
|
|
|
90 }
|
|
173
|
91
|
|
176
|
92 protected void Feed(char[] buffer, int offset, int length) {
|
|
|
93 m_buffer = buffer;
|
|
|
94 m_bufferOffset = offset;
|
|
|
95 m_bufferSize = offset + length;
|
|
173
|
96 }
|
|
|
97
|
|
176
|
98 protected bool Feed() {
|
|
|
99 if (m_chunkSize <= 0)
|
|
|
100 return false;
|
|
|
101
|
|
|
102 if (m_buffer != null) {
|
|
|
103 var free = m_buffer.Length - m_bufferSize;
|
|
|
104
|
|
|
105 if (free < m_chunkSize) {
|
|
|
106 free += m_chunkSize;
|
|
|
107 var used = m_bufferSize - m_bufferOffset;
|
|
|
108 var size = used + free;
|
|
|
109
|
|
|
110 if (size > m_bufferMax)
|
|
|
111 throw new ParserException(String.Format("The buffer limit ({0} Kb) is reached"), m_bufferMax/1024);
|
|
|
112
|
|
|
113 var temp = new char[size];
|
|
175
|
114
|
|
176
|
115 var read = Read(temp, used, m_chunkSize);
|
|
|
116 if (read == 0)
|
|
|
117 return false;
|
|
|
118
|
|
|
119 Array.Copy(m_buffer, m_bufferOffset, temp, 0, used);
|
|
|
120
|
|
|
121 m_bufferOffset = 0;
|
|
|
122 m_bufferSize = used + read;
|
|
|
123 m_buffer = temp;
|
|
|
124 }
|
|
|
125 } else {
|
|
|
126 Debug.Assert(m_bufferOffset == 0);
|
|
|
127 m_buffer = new char[m_chunkSize];
|
|
|
128 m_bufferSize = Read(m_buffer, 0, m_chunkSize);
|
|
|
129 return (m_bufferSize != 0);
|
|
|
130 }
|
|
175
|
131 }
|
|
|
132
|
|
|
133 protected abstract int Read(char[] buffer, int offset, int size);
|
|
173
|
134
|
|
176
|
135 public string GetTokenValue() {
|
|
|
136 return new String(m_buffer, m_tokenOffset, m_tokenLength);
|
|
173
|
137 }
|
|
|
138
|
|
176
|
139 public void CopyTokenTo(char[] buffer, int offset) {
|
|
|
140 m_buffer.CopyTo(buffer, offset);
|
|
|
141 }
|
|
|
142
|
|
|
143 public void CopyTokenTo(StringBuilder sb) {
|
|
|
144 sb.Append(m_buffer, m_tokenOffset, m_tokenLength);
|
|
|
145 }
|
|
175
|
146
|
|
173
|
147 }
|
|
|
148 }
|
|
|
149
|
