##// END OF EJS Templates
sync
cin -
r174:983df35b3ca1 ref20160224
parent child
Show More
This diff has been collapsed as it changes many lines, (510 lines changed) Show them Hide them
@@ -1,255 +1,255
1 using Implab;
2 using System;
3 using System.Collections.Generic;
4 using System.IO;
5 using Implab.Components;
6 using Implab.Automaton.RegularExpressions;
7
8 namespace Implab.Automaton {
9 /// <summary>
10 /// Π‘Π°Π·ΠΎΠ²Ρ‹ΠΉ класс для Ρ€Π°Π·Π±ΠΎΡ€Π° ΠΏΠΎΡ‚ΠΎΠΊΠ° Π²Ρ…ΠΎΠ΄Π½Ρ‹Ρ… символов Π½Π° Ρ‚ΠΎΠΊΠ΅Π½Ρ‹.
11 /// </summary>
12 /// <remarks>
13 /// Π‘ΠΊΠ°Π½Π½Π΅Ρ€ ΠΈΠΌΠ΅Π΅Ρ‚ Π²Π½ΡƒΡ‚Ρ€ΠΈ Π±ΡƒΡ„Ρ„Π΅Ρ€ с симолами Π²Ρ…ΠΎΠ΄Π½ΠΎΠ³ΠΎ тСкста, ΠΏΠΎ ΠΊΠΎΡ‚ΠΎΡ€ΠΎΠΌΡƒ ΠΏΠ΅Ρ€Π΅ΠΌΠ΅Ρ‰Π°ΡŽΡ‚ΡΡ Π΄Π²Π°
14 /// указатСля, Π½Π°Ρ‡Π°Π»Π° ΠΈ ΠΊΠΎΠ½Ρ†Π° Ρ‚ΠΎΠΊΠ΅Π½Π°, ΠΏΡ€ΠΈ ΠΏΠ΅Ρ€Π΅ΠΌΠ΅Ρ‰Π΅Π½ΠΈΠΈ ΠΈΡΠΊΠΎΠ»ΡŒΠ·ΡƒΠ΅Ρ‚ΡΡ Π”ΠšΠ для опрСдСлСния
15 /// ΠΊΠΎΠ½Ρ†Π° Ρ‚ΠΎΠΊΠ΅Π½Π° ΠΈ допустимости Ρ‚Π΅ΠΊΡƒΡ‰Π΅Π³ΠΎ символа.
16 /// </remarks>
17 public abstract class Scanner<TTag> : Disposable {
18 protected struct ScannerConfig {
19 public readonly DFAStateDescriptor<TTag>[] states;
20 public readonly int[] alphabet;
21 public readonly int initialState;
22
23 public ScannerConfig(DFAStateDescriptor<TTag>[] states, int[] alphabet, int initialState) {
24 this.initialState = initialState;
25 this.alphabet = alphabet;
26 this.states = states;
27 }
28 }
29
30 Stack<ScannerConfig> m_defs = new Stack<ScannerConfig>();
31
32 ScannerConfig m_config;
33
34 protected DFAStateDescriptor<TTag> m_currentState;
35 int m_previewCode;
36
37 protected int m_tokenLen;
38 protected int m_tokenOffset;
39
40 protected char[] m_buffer;
41 protected int m_bufferSize;
42 protected int m_pointer;
43
44 TextReader m_reader;
45 bool m_disposeReader;
46 int m_chunkSize = 1024; // 1k
47 int m_limit = 10 * 1024 * 1024; // 10Mb
48
49 protected Scanner(ScannerConfig config) {
50 Safe.ArgumentNotEmpty(config.states, "config.states");
51 Safe.ArgumentNotNull(config.alphabet, "config.alphabet");
52
53 m_config = config;
54 }
55
56 /// <summary>
57 /// ЗаполняСт Π²Ρ…ΠΎΠ΄Π½Ρ‹ΠΌΠΈ Π΄Π°Π½Π½Ρ‹ΠΌΠΈ Π±ΡƒΡ„Ρ„Π΅Ρ€.
58 /// </summary>
59 /// <param name="data">Π”Π°Π½Π½Ρ‹Π΅ для ΠΎΠ±Ρ€Π°Π±ΠΎΡ‚ΠΊΠΈ.</param>
60 /// <remarks>ΠšΠΎΠΏΠΈΡ€ΠΎΠ²Π°Π½ΠΈΠ΅ Π΄Π°Π½Π½Ρ‹Ρ… Π½Π΅ происходит, ΠΏΠ΅Ρ€Π΅Π΄Π°Π½Π½Ρ‹ΠΉ массив ΠΈΡΠΏΠΎΠ»ΡŒΠ·ΡƒΠ΅Ρ‚ΡΡ Π²
61 /// качСствС Π²Ρ…ΠΎΠ΄Π½ΠΎΠ³ΠΎ Π±ΡƒΡ„Ρ„Π΅Ρ€Π°.</remarks>
62 public void Feed(char[] data) {
63 Safe.ArgumentNotNull(data, "data");
64
65 Feed(data, data.Length);
66 }
67
68 /// <summary>
69 /// ЗаполняСт Π±ΡƒΡ„Ρ„ΡƒΡ€ чтСния Π²Ρ…ΠΎΠ΄Π½Ρ‹ΠΌΠΈ Π΄Π°Π½Π½Ρ‹ΠΌΠΈ.
70 /// </summary>
71 /// <param name="data">Π”Π°Π½Π½Ρ‹Π΅ для ΠΎΠ±Ρ€Π°Π±ΠΎΡ‚ΠΊΠΈ.</param>
72 /// <param name="length">Π”Π»ΠΈΠ½Π° Π΄Π°Π½Π½Ρ‹Ρ… для ΠΎΠ±Ρ€Π°Π±ΠΎΡ‚ΠΊΠΈ.</param>
73 /// <remarks>ΠšΠΎΠΏΠΈΡ€ΠΎΠ²Π°Π½ΠΈΠ΅ Π΄Π°Π½Π½Ρ‹Ρ… Π½Π΅ происходит, ΠΏΠ΅Ρ€Π΅Π΄Π°Π½Π½Ρ‹ΠΉ массив ΠΈΡΠΏΠΎΠ»ΡŒΠ·ΡƒΠ΅Ρ‚ΡΡ Π²
74 /// качСствС Π²Ρ…ΠΎΠ΄Π½ΠΎΠ³ΠΎ Π±ΡƒΡ„Ρ„Π΅Ρ€Π°.</remarks>
75 public void Feed(char[] data, int length) {
76 Safe.ArgumentNotNull(data, "data");
77 Safe.ArgumentInRange(length, 0, data.Length, "length");
78 AssertNotDisposed();
79
80 m_pointer = -1;
81 m_buffer = data;
82 m_bufferSize = length;
83 Shift();
84 }
85
86 public void Feed(TextReader reader, bool dispose) {
87 Safe.ArgumentNotNull(reader, "reader");
88 AssertNotDisposed();
89
90 if (m_reader != null && m_disposeReader)
91 m_reader.Dispose();
92
93 m_reader = reader;
94 m_disposeReader = dispose;
95 m_pointer = -1;
96 m_buffer = new char[m_chunkSize];
97 m_bufferSize = 0;
98 Shift();
99 }
100
101 /// <summary>
102 /// ΠŸΠΎΠ»ΡƒΡ‡Π°Π΅Ρ‚ Ρ‚Π΅ΠΊΡƒΡ‰ΠΈΠΉ Ρ‚ΠΎΠΊΠ΅Π½ Π² Π²ΠΈΠ΄Π΅ строки.
103 /// </summary>
104 /// <returns></returns>
105 protected string GetTokenValue() {
106 return new String(m_buffer, m_tokenOffset, m_tokenLen);
107 }
108
109 /// <summary>
110 /// ΠœΠ΅Ρ‚ΠΊΠΈ Ρ‚Π΅ΠΊΡƒΡ‰Π΅Π³ΠΎ Ρ‚ΠΎΠΊΠ΅Π½Π°, ΠΊΠΎΡ‚ΠΎΡ€Ρ‹Π΅ Π±Ρ‹Π»ΠΈ Π½Π°Π·Π½Π°Ρ‡Π΅Π½Ρ‹ Π² рСгулярном Π²Ρ‹Ρ€Π°ΠΆΠ΅Π½ΠΈΠΈ.
111 /// </summary>
112 protected TTag[] TokenTags {
113 get {
114 return m_currentState.tags;
115 }
116 }
117
118 /// <summary>
119 /// ΠŸΡ€ΠΈΠ·Π½Π°ΠΊ ΠΊΠΎΠ½Ρ†Π° Π΄Π°Π½Π½Ρ‹Ρ…
120 /// </summary>
121 public bool EOF {
122 get {
123 return m_pointer >= m_bufferSize;
124 }
125 }
126
127 /// <summary>
128 /// Π§ΠΈΡ‚Π°Π΅Ρ‚ ΡΠ»Π΅Π΄ΡƒΡŽΡ‰ΠΈΠΉ Ρ‚ΠΎΠΊΠ΅Π½, ΠΏΡ€ΠΈ этом <see cref="m_tokenOffset"/> ΡƒΠΊΠ°Π·Ρ‹Π²Π°Π΅Ρ‚ Π½Π° Π½Π°Ρ‡Π°Π»ΠΎ Ρ‚ΠΎΠΊΠ΅Π½Π°,
129 /// <see cref="m_tokenLen"/> Π½Π° Π΄Π»ΠΈΠ½Ρƒ Ρ‚ΠΎΠΊΠ΅Π½Π°, <see cref="m_buffer"/> - массив символов, Π²
130 /// ΠΊΠΎΡ‚ΠΎΡ€ΠΎΠΌ находится Ρ‚ΠΎΠΊΠ΅Π½.
131 /// </summary>
132 /// <returns><c>false</c> - достигнут ΠΊΠΎΠ½Π΅Ρ† Π΄Π°Π½Π½Ρ‹Ρ…, Ρ‚ΠΎΠΊΠ΅Π½ Π½Π΅ ΠΏΡ€ΠΎΡ‡ΠΈΡ‚Π°Π½.</returns>
133 protected bool ReadTokenInternal() {
134 if (m_pointer >= m_bufferSize)
135 return false;
136
137 m_currentState = m_config.states[m_config.initialState];
138 m_tokenLen = 0;
139 m_tokenOffset = m_pointer;
140 int nextState;
141 do {
142 nextState = m_currentState.transitions[m_previewCode];
143 if (nextState == DFAConst.UNREACHABLE_STATE) {
144 if (m_currentState.final)
145 return true;
146
147 throw new ParserException(
148 String.Format(
149 "Unexpected symbol '{0}', at pos {1}",
150 m_buffer[m_pointer],
151 Position
152 )
153 );
154 }
155 m_currentState = m_config.states[nextState];
156 m_tokenLen++;
157
158 } while (Shift());
159
160 // END OF DATA
161 if (!m_currentState.final)
162 throw new ParserException("Unexpected end of data");
163
164 return true;
165 }
166
167
168 bool Shift() {
169 m_pointer++;
170
171 if (m_pointer >= m_bufferSize) {
172 if (!ReadNextChunk())
173 return false;
174 }
175
176 m_previewCode = m_config.alphabet[m_buffer[m_pointer]];
177
178 return true;
179 }
180
181 bool ReadNextChunk() {
182 if (m_reader == null)
183 return false;
184
185 // extend buffer if nesessary
186 if (m_pointer + m_chunkSize > m_buffer.Length) {
187 // trim unused buffer head
188 var size = m_tokenLen + m_chunkSize;
189 if (size >= m_limit)
190 throw new ParserException(String.Format("Input buffer {0} bytes limit exceeded", m_limit));
191 var temp = new char[size];
192 Array.Copy(m_buffer, m_tokenOffset, temp, 0, m_tokenLen);
193 m_pointer -= m_tokenOffset;
194 m_bufferSize -= m_tokenOffset;
195 m_tokenOffset = 0;
196 m_buffer = temp;
197 }
198
199 var read = m_reader.Read(m_buffer, m_tokenLen, m_chunkSize);
200 if (read == 0)
201 return false;
202
203 m_bufferSize += read;
204
205 return true;
206 }
207
208 /// <summary>
209 /// ΠŸΠΎΠ·ΠΈΡ†ΠΈΡ сканнСра Π²ΠΎ Π²Ρ…ΠΎΠ΄Π½ΠΎΠΌ Π±ΡƒΡ„Π΅Ρ€Π΅
210 /// </summary>
211 public int Position {
212 get {
213 return m_pointer + 1;
214 }
215 }
216
217 /// <summary>
218 /// ΠŸΡ€Π΅ΠΊΠ»ΡŽΡ‡Π°Π΅Ρ‚ Π²Π½ΡƒΡ‚Ρ€Π΅Π½Π½ΠΈΠΉ Π”ΠšΠ Π½Π° ΡƒΠΊΠ°Π·Π°Π½Π½Ρ‹ΠΉ, позволяСт Ρ€Π΅Π°Π»ΠΈΠ·ΠΎΠ²Π°Ρ‚ΡŒ ΠΏΠΎΠ΄ΠΎΠ±ΠΈΠ΅ Π·Π°Ρ…Π²Π°Ρ‚Ρ‹Π²Π°ΡŽΡ‰Π΅ΠΉ
219 /// Π³Ρ€ΡƒΠΏΠΏΠΈΡ€ΠΎΠ²ΠΊΠΈ.
220 /// </summary>
221 /// <param name = "config"></param>
222 protected void Switch(ScannerConfig config) {
223 Safe.ArgumentNotNull(config.states, "config.states");
224
225 m_defs.Push(m_config);
226 m_config = config;
227
228 m_previewCode = m_config.alphabet[m_buffer[m_pointer]];
229 }
230
231 /// <summary>
232 /// ВосстанавливаСт ΠΏΡ€Π΅Π΄Ρ‹Π΄ΡƒΡ‰Π΅ΠΉ Π”ΠšΠ сканнСра.
233 /// </summary>
234 protected void Restore() {
235 if (m_defs.Count == 0)
236 throw new InvalidOperationException();
237 m_config = m_defs.Pop();
238
239 m_previewCode = m_config.alphabet[m_buffer[m_pointer]];
240 }
241
242 protected override void Dispose(bool disposing) {
243 if (disposing) {
244 if (m_reader != null && m_disposeReader)
245 m_reader.Dispose();
246 m_buffer = null;
247 m_bufferSize = 0;
248 m_pointer = 0;
249 m_tokenLen = 0;
250 m_tokenOffset = 0;
251 }
252 base.Dispose(disposing);
253 }
254 }
255 }
1 using Implab;
2 using System;
3 using System.Collections.Generic;
4 using System.IO;
5 using Implab.Components;
6 using Implab.Automaton.RegularExpressions;
7
8 namespace Implab.Automaton {
9 /// <summary>
10 /// Π‘Π°Π·ΠΎΠ²Ρ‹ΠΉ класс для Ρ€Π°Π·Π±ΠΎΡ€Π° ΠΏΠΎΡ‚ΠΎΠΊΠ° Π²Ρ…ΠΎΠ΄Π½Ρ‹Ρ… символов Π½Π° Ρ‚ΠΎΠΊΠ΅Π½Ρ‹.
11 /// </summary>
12 /// <remarks>
13 /// Π‘ΠΊΠ°Π½Π½Π΅Ρ€ ΠΈΠΌΠ΅Π΅Ρ‚ Π²Π½ΡƒΡ‚Ρ€ΠΈ Π±ΡƒΡ„Ρ„Π΅Ρ€ с симолами Π²Ρ…ΠΎΠ΄Π½ΠΎΠ³ΠΎ тСкста, ΠΏΠΎ ΠΊΠΎΡ‚ΠΎΡ€ΠΎΠΌΡƒ ΠΏΠ΅Ρ€Π΅ΠΌΠ΅Ρ‰Π°ΡŽΡ‚ΡΡ Π΄Π²Π°
14 /// указатСля, Π½Π°Ρ‡Π°Π»Π° ΠΈ ΠΊΠΎΠ½Ρ†Π° Ρ‚ΠΎΠΊΠ΅Π½Π°, ΠΏΡ€ΠΈ ΠΏΠ΅Ρ€Π΅ΠΌΠ΅Ρ‰Π΅Π½ΠΈΠΈ ΠΈΡΠΊΠΎΠ»ΡŒΠ·ΡƒΠ΅Ρ‚ΡΡ Π”ΠšΠ для опрСдСлСния
15 /// ΠΊΠΎΠ½Ρ†Π° Ρ‚ΠΎΠΊΠ΅Π½Π° ΠΈ допустимости Ρ‚Π΅ΠΊΡƒΡ‰Π΅Π³ΠΎ символа.
16 /// </remarks>
17 public abstract class Scanner<TTag> : Disposable {
18 protected struct ScannerConfig {
19 public readonly DFAStateDescriptor<TTag>[] states;
20 public readonly int[] alphabet;
21 public readonly int initialState;
22
23 public ScannerConfig(DFAStateDescriptor<TTag>[] states, int[] alphabet, int initialState) {
24 this.initialState = initialState;
25 this.alphabet = alphabet;
26 this.states = states;
27 }
28 }
29
30 Stack<ScannerConfig> m_defs = new Stack<ScannerConfig>();
31
32 ScannerConfig m_config;
33
34 protected DFAStateDescriptor<TTag> m_currentState;
35 int m_previewCode;
36
37 protected int m_tokenLen;
38 protected int m_tokenOffset;
39
40 protected char[] m_buffer;
41 protected int m_bufferSize;
42 protected int m_pointer;
43
44 TextReader m_reader;
45 bool m_disposeReader;
46 int m_chunkSize = 1024; // 1k
47 int m_limit = 10 * 1024 * 1024; // 10Mb
48
49 protected Scanner(ScannerConfig config) {
50 Safe.ArgumentNotEmpty(config.states, "config.states");
51 Safe.ArgumentNotNull(config.alphabet, "config.alphabet");
52
53 m_config = config;
54 }
55
56 /// <summary>
57 /// ЗаполняСт Π²Ρ…ΠΎΠ΄Π½Ρ‹ΠΌΠΈ Π΄Π°Π½Π½Ρ‹ΠΌΠΈ Π±ΡƒΡ„Ρ„Π΅Ρ€.
58 /// </summary>
59 /// <param name="data">Π”Π°Π½Π½Ρ‹Π΅ для ΠΎΠ±Ρ€Π°Π±ΠΎΡ‚ΠΊΠΈ.</param>
60 /// <remarks>ΠšΠΎΠΏΠΈΡ€ΠΎΠ²Π°Π½ΠΈΠ΅ Π΄Π°Π½Π½Ρ‹Ρ… Π½Π΅ происходит, ΠΏΠ΅Ρ€Π΅Π΄Π°Π½Π½Ρ‹ΠΉ массив ΠΈΡΠΏΠΎΠ»ΡŒΠ·ΡƒΠ΅Ρ‚ΡΡ Π²
61 /// качСствС Π²Ρ…ΠΎΠ΄Π½ΠΎΠ³ΠΎ Π±ΡƒΡ„Ρ„Π΅Ρ€Π°.</remarks>
62 public void Feed(char[] data) {
63 Safe.ArgumentNotNull(data, "data");
64
65 Feed(data, data.Length);
66 }
67
68 /// <summary>
69 /// ЗаполняСт Π±ΡƒΡ„Ρ„ΡƒΡ€ чтСния Π²Ρ…ΠΎΠ΄Π½Ρ‹ΠΌΠΈ Π΄Π°Π½Π½Ρ‹ΠΌΠΈ.
70 /// </summary>
71 /// <param name="data">Π”Π°Π½Π½Ρ‹Π΅ для ΠΎΠ±Ρ€Π°Π±ΠΎΡ‚ΠΊΠΈ.</param>
72 /// <param name="length">Π”Π»ΠΈΠ½Π° Π΄Π°Π½Π½Ρ‹Ρ… для ΠΎΠ±Ρ€Π°Π±ΠΎΡ‚ΠΊΠΈ.</param>
73 /// <remarks>ΠšΠΎΠΏΠΈΡ€ΠΎΠ²Π°Π½ΠΈΠ΅ Π΄Π°Π½Π½Ρ‹Ρ… Π½Π΅ происходит, ΠΏΠ΅Ρ€Π΅Π΄Π°Π½Π½Ρ‹ΠΉ массив ΠΈΡΠΏΠΎΠ»ΡŒΠ·ΡƒΠ΅Ρ‚ΡΡ Π²
74 /// качСствС Π²Ρ…ΠΎΠ΄Π½ΠΎΠ³ΠΎ Π±ΡƒΡ„Ρ„Π΅Ρ€Π°.</remarks>
75 public void Feed(char[] data, int length) {
76 Safe.ArgumentNotNull(data, "data");
77 Safe.ArgumentInRange(length, 0, data.Length, "length");
78 AssertNotDisposed();
79
80 m_pointer = -1;
81 m_buffer = data;
82 m_bufferSize = length;
83 Shift();
84 }
85
86 public void Feed(TextReader reader, bool dispose) {
87 Safe.ArgumentNotNull(reader, "reader");
88 AssertNotDisposed();
89
90 if (m_reader != null && m_disposeReader)
91 m_reader.Dispose();
92
93 m_reader = reader;
94 m_disposeReader = dispose;
95 m_pointer = -1;
96 m_buffer = new char[m_chunkSize];
97 m_bufferSize = 0;
98 Shift();
99 }
100
101 /// <summary>
102 /// ΠŸΠΎΠ»ΡƒΡ‡Π°Π΅Ρ‚ Ρ‚Π΅ΠΊΡƒΡ‰ΠΈΠΉ Ρ‚ΠΎΠΊΠ΅Π½ Π² Π²ΠΈΠ΄Π΅ строки.
103 /// </summary>
104 /// <returns></returns>
105 protected string GetTokenValue() {
106 return new String(m_buffer, m_tokenOffset, m_tokenLen);
107 }
108
109 /// <summary>
110 /// ΠœΠ΅Ρ‚ΠΊΠΈ Ρ‚Π΅ΠΊΡƒΡ‰Π΅Π³ΠΎ Ρ‚ΠΎΠΊΠ΅Π½Π°, ΠΊΠΎΡ‚ΠΎΡ€Ρ‹Π΅ Π±Ρ‹Π»ΠΈ Π½Π°Π·Π½Π°Ρ‡Π΅Π½Ρ‹ Π² рСгулярном Π²Ρ‹Ρ€Π°ΠΆΠ΅Π½ΠΈΠΈ.
111 /// </summary>
112 protected TTag[] TokenTags {
113 get {
114 return m_currentState.tags;
115 }
116 }
117
118 /// <summary>
119 /// ΠŸΡ€ΠΈΠ·Π½Π°ΠΊ ΠΊΠΎΠ½Ρ†Π° Π΄Π°Π½Π½Ρ‹Ρ…
120 /// </summary>
121 public bool EOF {
122 get {
123 return m_pointer >= m_bufferSize;
124 }
125 }
126
127 /// <summary>
128 /// Π§ΠΈΡ‚Π°Π΅Ρ‚ ΡΠ»Π΅Π΄ΡƒΡŽΡ‰ΠΈΠΉ Ρ‚ΠΎΠΊΠ΅Π½, ΠΏΡ€ΠΈ этом <see cref="m_tokenOffset"/> ΡƒΠΊΠ°Π·Ρ‹Π²Π°Π΅Ρ‚ Π½Π° Π½Π°Ρ‡Π°Π»ΠΎ Ρ‚ΠΎΠΊΠ΅Π½Π°,
129 /// <see cref="m_tokenLen"/> Π½Π° Π΄Π»ΠΈΠ½Ρƒ Ρ‚ΠΎΠΊΠ΅Π½Π°, <see cref="m_buffer"/> - массив символов, Π²
130 /// ΠΊΠΎΡ‚ΠΎΡ€ΠΎΠΌ находится Ρ‚ΠΎΠΊΠ΅Π½.
131 /// </summary>
132 /// <returns><c>false</c> - достигнут ΠΊΠΎΠ½Π΅Ρ† Π΄Π°Π½Π½Ρ‹Ρ…, Ρ‚ΠΎΠΊΠ΅Π½ Π½Π΅ ΠΏΡ€ΠΎΡ‡ΠΈΡ‚Π°Π½.</returns>
133 protected bool ReadTokenInternal() {
134 if (m_pointer >= m_bufferSize)
135 return false;
136
137 m_currentState = m_config.states[m_config.initialState];
138 m_tokenLen = 0;
139 m_tokenOffset = m_pointer;
140 int nextState;
141 do {
142 nextState = m_currentState.transitions[m_previewCode];
143 if (nextState == DFAConst.UNREACHABLE_STATE) {
144 if (m_currentState.final)
145 return true;
146
147 throw new ParserException(
148 String.Format(
149 "Unexpected symbol '{0}', at pos {1}",
150 m_buffer[m_pointer],
151 Position
152 )
153 );
154 }
155 m_currentState = m_config.states[nextState];
156 m_tokenLen++;
157
158 } while (Shift());
159
160 // END OF DATA
161 if (!m_currentState.final)
162 throw new ParserException("Unexpected end of data");
163
164 return true;
165 }
166
167
168 bool Shift() {
169 m_pointer++;
170
171 if (m_pointer >= m_bufferSize) {
172 if (!ReadNextChunk())
173 return false;
174 }
175
176 m_previewCode = m_config.alphabet[m_buffer[m_pointer]];
177
178 return true;
179 }
180
181 bool ReadNextChunk() {
182 if (m_reader == null)
183 return false;
184
185 // extend buffer if nesessary
186 if (m_pointer + m_chunkSize > m_buffer.Length) {
187 // trim unused buffer head
188 var size = m_tokenLen + m_chunkSize;
189 if (size >= m_limit)
190 throw new ParserException(String.Format("Input buffer {0} bytes limit exceeded", m_limit));
191 var temp = new char[size];
192 Array.Copy(m_buffer, m_tokenOffset, temp, 0, m_tokenLen);
193 m_pointer -= m_tokenOffset;
194 m_bufferSize -= m_tokenOffset;
195 m_tokenOffset = 0;
196 m_buffer = temp;
197 }
198
199 var read = m_reader.Read(m_buffer, m_tokenLen, m_chunkSize);
200 if (read == 0)
201 return false;
202
203 m_bufferSize += read;
204
205 return true;
206 }
207
208 /// <summary>
209 /// ΠŸΠΎΠ·ΠΈΡ†ΠΈΡ сканнСра Π²ΠΎ Π²Ρ…ΠΎΠ΄Π½ΠΎΠΌ Π±ΡƒΡ„Π΅Ρ€Π΅
210 /// </summary>
211 public int Position {
212 get {
213 return m_pointer + 1;
214 }
215 }
216
217 /// <summary>
218 /// ΠŸΡ€Π΅ΠΊΠ»ΡŽΡ‡Π°Π΅Ρ‚ Π²Π½ΡƒΡ‚Ρ€Π΅Π½Π½ΠΈΠΉ Π”ΠšΠ Π½Π° ΡƒΠΊΠ°Π·Π°Π½Π½Ρ‹ΠΉ, позволяСт Ρ€Π΅Π°Π»ΠΈΠ·ΠΎΠ²Π°Ρ‚ΡŒ ΠΏΠΎΠ΄ΠΎΠ±ΠΈΠ΅ Π·Π°Ρ…Π²Π°Ρ‚Ρ‹Π²Π°ΡŽΡ‰Π΅ΠΉ
219 /// Π³Ρ€ΡƒΠΏΠΏΠΈΡ€ΠΎΠ²ΠΊΠΈ.
220 /// </summary>
221 /// <param name = "config"></param>
222 protected void Switch(ScannerConfig config) {
223 Safe.ArgumentNotNull(config.states, "config.states");
224
225 m_defs.Push(m_config);
226 m_config = config;
227
228 m_previewCode = m_config.alphabet[m_buffer[m_pointer]];
229 }
230
231 /// <summary>
232 /// ВосстанавливаСт ΠΏΡ€Π΅Π΄Ρ‹Π΄ΡƒΡ‰Π΅ΠΉ Π”ΠšΠ сканнСра.
233 /// </summary>
234 protected void Restore() {
235 if (m_defs.Count == 0)
236 throw new InvalidOperationException();
237 m_config = m_defs.Pop();
238
239 m_previewCode = m_config.alphabet[m_buffer[m_pointer]];
240 }
241
242 protected override void Dispose(bool disposing) {
243 if (disposing) {
244 if (m_reader != null && m_disposeReader)
245 m_reader.Dispose();
246 m_buffer = null;
247 m_bufferSize = 0;
248 m_pointer = 0;
249 m_tokenLen = 0;
250 m_tokenOffset = 0;
251 }
252 base.Dispose(disposing);
253 }
254 }
255 }
@@ -1,143 +1,62
1 1 using System;
2 2 using Implab.Automaton.RegularExpressions;
3 3 using Implab.Automaton;
4 using System.Diagnostics;
4 5
5 6 namespace Implab.Formats {
6 7 public struct BufferScanner<TTag> {
7 char[] m_buffer;
8 int m_offset;
9 int m_position;
10 int m_hi;
11
12 readonly int m_chunk;
13 readonly int m_limit;
14
15 8 readonly DFAStateDescriptor<TTag>[] m_dfa;
16 9 int m_state;
10 int m_pos;
17 11
18 public BufferScanner(DFAStateDescriptor<TTag>[] dfa, int initialState, int chunk, int limit) {
12 public BufferScanner(DFAStateDescriptor<TTag>[] dfa, int initialState) {
19 13 m_dfa = dfa;
20 14 m_state = initialState;
21 m_chunk = chunk;
22 m_limit = limit;
23 m_buffer = null;
24 m_offset = 0;
25 m_position = 0;
26 m_hi = 0;
27 }
28
29 public char[] Buffer {
30 get {
31 return m_buffer;
32 }
33 }
34
35 public int HiMark {
36 get {
37 return m_hi;
38 }
39 15 }
40 16
41 17 public int Position {
42 get {
43 return m_position;
44 }
45 }
46
47 public int Length {
48 get {
49 return m_hi - m_position;
50 }
51 }
52
53 public int TokenOffset {
54 get {
55 return m_offset;
56 }
57 }
58
59 public int TokenLength {
60 get {
61 return m_position - m_offset;
62 }
63 }
64
65 public void Init(char[] buffer, int position, int length) {
66 m_buffer = buffer;
67 m_position = position;
68 m_offset = position;
69 m_hi = position + length;
70 }
71
72 public int Extend() {
73 // free space
74 var free = m_buffer.Length - m_hi;
75
76 // if the buffer have enough free space
77 if (free > 0)
78 return free;
79
80 // effective size of the buffer
81 var size = m_buffer.Length - m_offset;
82
83 // calculate the new size
84 int grow = Math.Min(m_limit - size, m_chunk);
85 if (grow <= 0)
86 throw new ParserException(String.Format("Input buffer {0} bytes limit exceeded", m_limit));
87
88 var temp = new char[size + grow];
89 Array.Copy(m_buffer, m_offset, temp, 0, m_hi - m_offset);
90 m_position -= m_offset;
91 m_hi -= m_offset;
92 m_offset = 0;
93 m_buffer = temp;
94
95 return free + grow;
96 }
97
98 public void RaiseMark(int size) {
99 m_hi += size;
18 get { return m_pos; }
100 19 }
101 20
102 21 /// <summary>
103 22 /// Scan this instance.
104 23 /// </summary>
105 24 /// <returns><c>true</c> - additional data required</returns>
106 public bool Scan() {
107 while (m_position < m_hi) {
108 var ch = m_buffer[m_position];
109 var next = m_dfa[m_state].transitions[(int)ch];
25 public bool Scan(int[] buffer, int position, int length) {
26 var hi = position + length;
27 m_pos = position;
28
29 while (position < hi) {
30 var next = m_dfa[m_state].transitions[buffer[position]];
110 31 if (next == DFAConst.UNREACHABLE_STATE) {
111 32 if (m_dfa[m_state].final)
112 33 return false;
113 34
114 35 throw new ParserException(
115 36 String.Format(
116 "Unexpected token '{0}'",
117 new string(m_buffer, m_offset, m_position - m_offset)
37 "Unexpected symbol"
118 38 )
119 39 );
120 40 }
41 m_pos++;
121 42 m_state = next;
122 m_position++;
123 43 }
124 44
125 45 return true;
126 46 }
127 47
128 48 public void Eof() {
129 49 if (!m_dfa[m_state].final)
130 50 throw new ParserException(
131 51 String.Format(
132 "Unexpected token '{0}'",
133 new string(m_buffer, m_offset, m_position - m_offset)
52 "Unexpected EOF"
134 53 )
135 54 );
136 55 }
137 56
138 57 public TTag[] GetTokenTags() {
139 58 return m_dfa[m_state].tags;
140 59 }
141 60 }
142 61 }
143 62
@@ -1,72 +1,61
1 1 using System;
2 2 using Implab.Components;
3 3
4 4 namespace Implab.Formats {
5 5 public abstract class TextScanner<TTag> : Disposable {
6 6
7 char[] m_buffer;
8 int m_offset;
9 int m_length;
10 int m_tokenOffset;
7 readonly int[] m_buffer;
8 int m_bufferOffset;
9 int m_dataLength;
11 10 int m_tokenLength;
11
12 12 TTag[] m_tags;
13 13
14 14 BufferScanner<TTag> m_scanner;
15 15
16 16 protected bool ReadTokenInternal() {
17 17 if (EOF)
18 18 return false;
19 19
20 // create a new scanner from template (scanners are structs)
20 // create a new scanner from template (scanners are value types)
21 21 var inst = m_scanner;
22 22
23 // initialize the scanner
24 inst.Init(m_buffer, m_offset, m_length);
23 m_tokenLength = 0;
24
25 while (inst.Scan(m_buffer, m_bufferOffset, m_dataLength)) {
26 m_tokenLength += m_dataLength;
27
28 var actual = Read(m_buffer, 0, m_buffer.Length);
29
30 m_bufferOffset = 0;
31 m_dataLength = actual;
25 32
26 // do work
27 while (inst.Scan())
28 Feed(ref inst);
33 if (actual == 0) {
34 inst.Eof();
35 break;
36 }
37 }
38
39 var len = inst.Position - m_bufferOffset;
40 m_tokenLength += len;
41 m_dataLength -= len;
42 m_bufferOffset = inst.Position;
29 43
30 44 // save result;
31 m_buffer = inst.Buffer;
32 m_length = inst.Length;
33 m_offset = inst.Position;
34 m_tokenOffset = inst.TokenOffset;
35 m_tokenLength = inst.TokenLength;
36
45
37 46 m_tags = inst.GetTokenTags();
38 47 }
39 48
40 protected string GetToken() {
41 return new String(m_buffer, m_tokenOffset, m_tokenLength);
42 }
49 protected abstract int Read(int[] buffer, int offset, int size);
43 50
44 51 protected TTag[] Tags {
45 52 get {
46 53 return m_tags;
47 54 }
48 55 }
49 56
50 /// <summary>
51 /// Feed the specified scanner.
52 /// </summary>
53 /// <param name="scanner">Scanner.</param>
54 /// <example>
55 /// protected override void Feed(ref BufferScanner<TTag> scanner) {
56 /// var size = scanner.Extend();
57 /// var actual = m_reader.Read(scanner.Buffer, scanner.HiMark, size);
58 /// if (actual == 0) {
59 /// m_eof = true;
60 /// scanner.Eof();
61 /// } else {
62 /// scanner.RaiseHiMark(actual);
63 /// }
64 /// }
65 /// </example>
66 protected abstract void Feed(ref BufferScanner<TTag> scanner);
67
68 57 public abstract bool EOF { get; }
69
58
70 59 }
71 60 }
72 61
General Comments 0
You need to be logged in to leave comments. Login now