|
173
|
1 using System;
|
|
|
2 using Implab.Components;
|
|
175
|
3 using System.Diagnostics;
|
|
|
4 using Implab.Automaton;
|
|
176
|
5 using System.Text;
|
|
173
|
6
|
|
|
7 namespace Implab.Formats {
|
|
176
|
8 public abstract class TextScanner : Disposable {
|
|
|
9 readonly int m_bufferMax;
|
|
|
10 readonly int m_chunkSize;
|
|
173
|
11
|
|
176
|
12 char[] m_buffer;
|
|
174
|
13 int m_bufferOffset;
|
|
175
|
14 int m_bufferSize;
|
|
176
|
15 int m_tokenOffset;
|
|
173
|
16 int m_tokenLength;
|
|
174
|
17
|
|
176
|
18 /// <summary>
|
|
177
|
19 /// Initializes a new instance of the <see cref="Implab.Formats.TextScanner"/> class.
|
|
176
|
20 /// </summary>
|
|
|
21 /// <param name="bufferMax">Buffer max.</param>
|
|
|
22 /// <param name="chunkSize">Chunk size.</param>
|
|
|
23 protected TextScanner(int bufferMax, int chunkSize) {
|
|
|
24 Debug.Assert(m_chunkSize <= m_bufferMax);
|
|
|
25
|
|
|
26 m_bufferMax = bufferMax;
|
|
|
27 m_chunkSize = chunkSize;
|
|
|
28 }
|
|
173
|
29
|
|
176
|
30 /// <summary>
|
|
177
|
31 /// Initializes a new instance of the <see cref="Implab.Formats.TextScanner"/> class.
|
|
176
|
32 /// </summary>
|
|
|
33 /// <param name="buffer">Buffer.</param>
|
|
|
34 protected TextScanner(char[] buffer) {
|
|
|
35 if (buffer != null) {
|
|
|
36 m_buffer = buffer;
|
|
|
37 m_bufferSize = buffer.Length;
|
|
|
38 }
|
|
|
39 }
|
|
|
40
|
|
|
41 /// <summary>
|
|
|
42 /// (hungry) Reads the next token.
|
|
|
43 /// </summary>
|
|
|
44 /// <returns><c>true</c>, if token internal was read, <c>false</c> if there is no more tokens in the stream.</returns>
|
|
|
45 /// <param name="dfa">The transition map for the automaton</param>
|
|
|
46 /// <param name="final">Final states of the automaton.</param>
|
|
|
47 /// <param name="tags">Tags.</param>
|
|
|
48 /// <param name="state">The initial state for the automaton.</param>
|
|
177
|
49 /// <param name="alphabet"></param>
|
|
|
50 /// <param name = "tag"></param>
|
|
|
51 internal bool ReadToken<TTag>(int[,] dfa, bool[] final, TTag[][] tags, int state, int[] alphabet, out TTag[] tag) {
|
|
176
|
52 m_tokenLength = 0;
|
|
180
|
53 tag = null;
|
|
176
|
54
|
|
|
55 var maxSymbol = alphabet.Length - 1;
|
|
182
|
56 int next;
|
|
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;
|
|
182
|
60 next = state;
|
|
177
|
61 while (pos < m_bufferSize) {
|
|
175
|
62 var ch = m_buffer[pos];
|
|
176
|
63
|
|
182
|
64 next = dfa[next, ch > maxSymbol ? AutomatonConst.UNCLASSIFIED_INPUT : alphabet[ch]];
|
|
181
|
65
|
|
|
66 if (next == AutomatonConst.UNREACHABLE_STATE)
|
|
175
|
67 break;
|
|
182
|
68
|
|
181
|
69 state = next;
|
|
176
|
70 pos++;
|
|
175
|
71 }
|
|
176
|
72 m_tokenLength = pos - m_bufferOffset;
|
|
182
|
73 } while (next != AutomatonConst.UNREACHABLE_STATE && Feed());
|
|
176
|
74
|
|
|
75 m_tokenOffset = m_bufferOffset;
|
|
|
76 m_bufferOffset += m_tokenLength;
|
|
174
|
77
|
|
176
|
78 if (final[state]) {
|
|
|
79 tag = tags[state];
|
|
|
80 return true;
|
|
177
|
81 }
|
|
|
82
|
|
|
83 if (m_bufferOffset == m_bufferSize) {
|
|
|
84 if (m_tokenLength == 0) //EOF
|
|
176
|
85 return false;
|
|
|
86
|
|
177
|
87 throw new ParserException();
|
|
|
88 }
|
|
|
89
|
|
|
90 throw new ParserException(String.Format("Unexpected symbol '{0}'", m_buffer[m_bufferOffset]));
|
|
176
|
91
|
|
|
92 }
|
|
173
|
93
|
|
176
|
94 protected void Feed(char[] buffer, int offset, int length) {
|
|
|
95 m_buffer = buffer;
|
|
|
96 m_bufferOffset = offset;
|
|
|
97 m_bufferSize = offset + length;
|
|
173
|
98 }
|
|
|
99
|
|
176
|
100 protected bool Feed() {
|
|
|
101 if (m_chunkSize <= 0)
|
|
|
102 return false;
|
|
|
103
|
|
|
104 if (m_buffer != null) {
|
|
|
105 var free = m_buffer.Length - m_bufferSize;
|
|
|
106
|
|
|
107 if (free < m_chunkSize) {
|
|
|
108 free += m_chunkSize;
|
|
|
109 var used = m_bufferSize - m_bufferOffset;
|
|
|
110 var size = used + free;
|
|
|
111
|
|
|
112 if (size > m_bufferMax)
|
|
180
|
113 throw new ParserException(String.Format("The buffer limit ({0} Kb) is reached", m_bufferMax / 1024));
|
|
176
|
114
|
|
|
115 var temp = new char[size];
|
|
175
|
116
|
|
176
|
117 var read = Read(temp, used, m_chunkSize);
|
|
|
118 if (read == 0)
|
|
|
119 return false;
|
|
|
120
|
|
|
121 Array.Copy(m_buffer, m_bufferOffset, temp, 0, used);
|
|
|
122
|
|
|
123 m_bufferOffset = 0;
|
|
|
124 m_bufferSize = used + read;
|
|
|
125 m_buffer = temp;
|
|
180
|
126 } else {
|
|
|
127 var read = Read(m_buffer, m_bufferSize, m_chunkSize);
|
|
|
128 if (read == 0)
|
|
|
129 return false;
|
|
|
130 m_bufferSize += m_chunkSize;
|
|
176
|
131 }
|
|
180
|
132 return true;
|
|
176
|
133 } else {
|
|
|
134 Debug.Assert(m_bufferOffset == 0);
|
|
|
135 m_buffer = new char[m_chunkSize];
|
|
|
136 m_bufferSize = Read(m_buffer, 0, m_chunkSize);
|
|
|
137 return (m_bufferSize != 0);
|
|
|
138 }
|
|
175
|
139 }
|
|
|
140
|
|
|
141 protected abstract int Read(char[] buffer, int offset, int size);
|
|
173
|
142
|
|
176
|
143 public string GetTokenValue() {
|
|
|
144 return new String(m_buffer, m_tokenOffset, m_tokenLength);
|
|
173
|
145 }
|
|
|
146
|
|
176
|
147 public void CopyTokenTo(char[] buffer, int offset) {
|
|
182
|
148 Array.Copy(m_buffer, m_tokenOffset,buffer, offset, m_tokenLength);
|
|
176
|
149 }
|
|
|
150
|
|
|
151 public void CopyTokenTo(StringBuilder sb) {
|
|
|
152 sb.Append(m_buffer, m_tokenOffset, m_tokenLength);
|
|
|
153 }
|
|
175
|
154
|
|
173
|
155 }
|
|
|
156 }
|
|
|
157
|
