Commit | Line | Data |
---|---|---|
bcb8c2cb PT |
1 | /******************************************************************************* |
2 | * Copyright (c) 2012, 2014 Ericsson | |
3 | * | |
4 | * All rights reserved. This program and the accompanying materials are | |
5 | * made available under the terms of the Eclipse Public License v1.0 which | |
6 | * accompanies this distribution, and is available at | |
7 | * http://www.eclipse.org/legal/epl-v10.html | |
8 | * | |
9 | * Contributors: | |
10 | * Patrick Tasse - Initial API and implementation | |
738a2796 | 11 | * Marc-Andre Laperle - Add persistent index support |
bcb8c2cb PT |
12 | *******************************************************************************/ |
13 | ||
2bdf0193 | 14 | package org.eclipse.tracecompass.tmf.core.trace.text; |
bcb8c2cb PT |
15 | |
16 | import java.io.File; | |
17 | import java.io.IOException; | |
738a2796 | 18 | import java.nio.ByteBuffer; |
543e2b1e BH |
19 | import java.util.Collections; |
20 | import java.util.List; | |
bcb8c2cb PT |
21 | import java.util.regex.Matcher; |
22 | import java.util.regex.Pattern; | |
23 | ||
24 | import org.eclipse.core.resources.IProject; | |
25 | import org.eclipse.core.resources.IResource; | |
26 | import org.eclipse.core.runtime.IStatus; | |
27 | import org.eclipse.core.runtime.Status; | |
2bdf0193 AM |
28 | import org.eclipse.tracecompass.internal.tmf.core.Activator; |
29 | import org.eclipse.tracecompass.tmf.core.event.ITmfEvent; | |
30 | import org.eclipse.tracecompass.tmf.core.exceptions.TmfTraceException; | |
31 | import org.eclipse.tracecompass.tmf.core.io.BufferedRandomAccessFile; | |
32 | import org.eclipse.tracecompass.tmf.core.timestamp.TmfTimestamp; | |
33 | import org.eclipse.tracecompass.tmf.core.trace.ITmfContext; | |
2bdf0193 AM |
34 | import org.eclipse.tracecompass.tmf.core.trace.TmfTrace; |
35 | import org.eclipse.tracecompass.tmf.core.trace.TraceValidationStatus; | |
36 | import org.eclipse.tracecompass.tmf.core.trace.indexer.ITmfPersistentlyIndexable; | |
37 | import org.eclipse.tracecompass.tmf.core.trace.indexer.ITmfTraceIndexer; | |
38 | import org.eclipse.tracecompass.tmf.core.trace.indexer.TmfBTreeTraceIndexer; | |
39 | import org.eclipse.tracecompass.tmf.core.trace.indexer.checkpoint.ITmfCheckpoint; | |
40 | import org.eclipse.tracecompass.tmf.core.trace.indexer.checkpoint.TmfCheckpoint; | |
41 | import org.eclipse.tracecompass.tmf.core.trace.location.ITmfLocation; | |
42 | import org.eclipse.tracecompass.tmf.core.trace.location.TmfLongLocation; | |
bcb8c2cb PT |
43 | |
44 | /** | |
45 | * Extension of TmfTrace for handling of line-based text traces parsed using | |
46 | * regular expressions. Each line that matches the first line pattern indicates | |
47 | * the start of a new event. The subsequent lines can contain additional | |
48 | * information that is added to the current event. | |
49 | * | |
50 | * @param <T> | |
51 | * TmfEvent class returned by this trace | |
52 | * | |
53 | * @since 3.0 | |
54 | */ | |
5733be39 | 55 | public abstract class TextTrace<T extends TextTraceEvent> extends TmfTrace implements ITmfPersistentlyIndexable { |
bcb8c2cb | 56 | |
661becf8 | 57 | private static final TmfLongLocation NULL_LOCATION = new TmfLongLocation(-1L); |
bcb8c2cb PT |
58 | private static final int MAX_LINES = 100; |
59 | private static final int MAX_CONFIDENCE = 100; | |
60 | ||
61 | /** The default separator used for multi-line fields */ | |
62 | protected static final String SEPARATOR = " | "; //$NON-NLS-1$ | |
63 | ||
64 | /** The text file */ | |
65 | protected BufferedRandomAccessFile fFile; | |
66 | ||
67 | /** | |
68 | * Constructor | |
69 | */ | |
70 | public TextTrace() { | |
71 | } | |
72 | ||
73 | /** | |
74 | * {@inheritDoc} | |
75 | * <p> | |
543e2b1e BH |
76 | * The default implementation computes the confidence as the sum of weighted |
77 | * values of the first 100 lines of the file which match any of the provided | |
78 | * validation patterns. For each matching line a weighted value between 1.5 | |
79 | * and 2.0 is assigned based on the group count of the matching patterns. | |
80 | * The higher the group count, the closer the weighted value will be to 2.0. | |
bcb8c2cb PT |
81 | */ |
82 | @Override | |
83 | public IStatus validate(IProject project, String path) { | |
84 | File file = new File(path); | |
85 | if (!file.exists()) { | |
86 | return new Status(IStatus.ERROR, Activator.PLUGIN_ID, "File not found: " + path); //$NON-NLS-1$ | |
87 | } | |
88 | if (!file.isFile()) { | |
89 | return new Status(IStatus.ERROR, Activator.PLUGIN_ID, "Not a file. It's a directory: " + path); //$NON-NLS-1$ | |
90 | } | |
91 | int confidence = 0; | |
92 | try (BufferedRandomAccessFile rafile = new BufferedRandomAccessFile(path, "r")) { //$NON-NLS-1$ | |
93 | int lineCount = 0; | |
918551aa | 94 | double matches = 0.0; |
bcb8c2cb | 95 | String line = rafile.getNextLine(); |
543e2b1e | 96 | List<Pattern> validationPatterns = getValidationPatterns(); |
bcb8c2cb | 97 | while ((line != null) && (lineCount++ < MAX_LINES)) { |
543e2b1e BH |
98 | for(Pattern pattern : validationPatterns) { |
99 | Matcher matcher = pattern.matcher(line); | |
100 | if (matcher.matches()) { | |
101 | int groupCount = matcher.groupCount(); | |
102 | matches += (1.0 + groupCount / ((double) groupCount + 1)); | |
103 | } | |
bcb8c2cb | 104 | } |
918551aa | 105 | confidence = (int) (MAX_CONFIDENCE * matches / lineCount); |
bcb8c2cb PT |
106 | line = rafile.getNextLine(); |
107 | } | |
108 | } catch (IOException e) { | |
109 | Activator.logError("Error validating file: " + path, e); //$NON-NLS-1$ | |
110 | return new Status(IStatus.ERROR, Activator.PLUGIN_ID, "IOException validating file: " + path, e); //$NON-NLS-1$ | |
111 | } | |
112 | ||
113 | return new TraceValidationStatus(confidence, Activator.PLUGIN_ID); | |
114 | ||
115 | } | |
bcb8c2cb PT |
116 | @Override |
117 | public void initTrace(IResource resource, String path, Class<? extends ITmfEvent> type) throws TmfTraceException { | |
118 | super.initTrace(resource, path, type); | |
119 | try { | |
120 | fFile = new BufferedRandomAccessFile(getPath(), "r"); //$NON-NLS-1$ | |
121 | } catch (IOException e) { | |
122 | throw new TmfTraceException(e.getMessage(), e); | |
123 | } | |
124 | } | |
125 | ||
126 | @Override | |
127 | public synchronized void dispose() { | |
128 | super.dispose(); | |
129 | if (fFile != null) { | |
130 | try { | |
131 | fFile.close(); | |
132 | } catch (IOException e) { | |
133 | } finally { | |
134 | fFile = null; | |
135 | } | |
136 | } | |
137 | } | |
138 | ||
139 | @Override | |
5ece050b | 140 | public synchronized TextTraceContext seekEvent(ITmfLocation location) { |
bcb8c2cb PT |
141 | TextTraceContext context = new TextTraceContext(NULL_LOCATION, ITmfContext.UNKNOWN_RANK); |
142 | if (NULL_LOCATION.equals(location) || fFile == null) { | |
143 | return context; | |
144 | } | |
145 | try { | |
146 | if (location == null) { | |
147 | fFile.seek(0); | |
148 | } else if (location.getLocationInfo() instanceof Long) { | |
149 | fFile.seek((Long) location.getLocationInfo()); | |
150 | } | |
151 | long rawPos = fFile.getFilePointer(); | |
152 | String line = fFile.getNextLine(); | |
153 | while (line != null) { | |
154 | Matcher matcher = getFirstLinePattern().matcher(line); | |
13797536 MK |
155 | if (matcher.matches()) { |
156 | setupContext(context, rawPos, line, matcher); | |
bcb8c2cb PT |
157 | return context; |
158 | } | |
159 | rawPos = fFile.getFilePointer(); | |
160 | line = fFile.getNextLine(); | |
161 | } | |
162 | return context; | |
163 | } catch (IOException e) { | |
164 | Activator.logError("Error seeking file: " + getPath(), e); //$NON-NLS-1$ | |
165 | return context; | |
166 | } | |
167 | } | |
168 | ||
13797536 MK |
169 | private void setupContext(TextTraceContext context, long rawPos, String line, Matcher matcher) throws IOException { |
170 | context.setLocation(new TmfLongLocation(rawPos)); | |
171 | context.firstLineMatcher = matcher; | |
172 | context.firstLine = line; | |
173 | context.nextLineLocation = fFile.getFilePointer(); | |
174 | } | |
175 | ||
bcb8c2cb | 176 | @Override |
5ece050b | 177 | public synchronized TextTraceContext seekEvent(double ratio) { |
bcb8c2cb PT |
178 | if (fFile == null) { |
179 | return new TextTraceContext(NULL_LOCATION, ITmfContext.UNKNOWN_RANK); | |
180 | } | |
181 | try { | |
182 | long pos = (long) (ratio * fFile.length()); | |
183 | while (pos > 0) { | |
184 | fFile.seek(pos - 1); | |
185 | if (fFile.read() == '\n') { | |
186 | break; | |
187 | } | |
188 | pos--; | |
189 | } | |
190 | ITmfLocation location = new TmfLongLocation(Long.valueOf(pos)); | |
5ece050b | 191 | TextTraceContext context = seekEvent(location); |
bcb8c2cb PT |
192 | context.setRank(ITmfContext.UNKNOWN_RANK); |
193 | return context; | |
194 | } catch (IOException e) { | |
195 | Activator.logError("Error seeking file: " + getPath(), e); //$NON-NLS-1$ | |
196 | return new TextTraceContext(NULL_LOCATION, ITmfContext.UNKNOWN_RANK); | |
197 | } | |
198 | } | |
199 | ||
200 | @Override | |
201 | public double getLocationRatio(ITmfLocation location) { | |
202 | if (fFile == null) { | |
203 | return 0; | |
204 | } | |
205 | try { | |
206 | long length = fFile.length(); | |
207 | if (length == 0) { | |
208 | return 0; | |
209 | } | |
210 | if (location.getLocationInfo() instanceof Long) { | |
211 | return (double) ((Long) location.getLocationInfo()) / length; | |
212 | } | |
213 | } catch (IOException e) { | |
214 | Activator.logError("Error reading file: " + getPath(), e); //$NON-NLS-1$ | |
215 | } | |
216 | return 0; | |
217 | } | |
218 | ||
219 | @Override | |
220 | public ITmfLocation getCurrentLocation() { | |
221 | return null; | |
222 | } | |
223 | ||
224 | @Override | |
5ece050b AM |
225 | public TextTraceEvent parseEvent(ITmfContext tmfContext) { |
226 | TextTraceContext context = seekEvent(tmfContext.getLocation()); | |
bcb8c2cb PT |
227 | return parse(context); |
228 | } | |
229 | ||
230 | @Override | |
231 | public synchronized T getNext(ITmfContext context) { | |
5ece050b AM |
232 | if (!(context instanceof TextTraceContext)) { |
233 | throw new IllegalArgumentException(); | |
234 | } | |
235 | TextTraceContext savedContext = new TextTraceContext(context.getLocation(), context.getRank()); | |
236 | T event = parse((TextTraceContext) context); | |
bcb8c2cb PT |
237 | if (event != null) { |
238 | updateAttributes(savedContext, event.getTimestamp()); | |
239 | context.increaseRank(); | |
240 | } | |
241 | return event; | |
242 | } | |
243 | ||
244 | /** | |
245 | * Parse the next event. The context is advanced. | |
246 | * | |
247 | * @param tmfContext | |
248 | * the context | |
249 | * @return the next event or null | |
250 | */ | |
5ece050b | 251 | protected synchronized T parse(TextTraceContext tmfContext) { |
bcb8c2cb PT |
252 | if (fFile == null) { |
253 | return null; | |
254 | } | |
5ece050b | 255 | TextTraceContext context = tmfContext; |
bcb8c2cb PT |
256 | if (context.getLocation() == null || !(context.getLocation().getLocationInfo() instanceof Long) || NULL_LOCATION.equals(context.getLocation())) { |
257 | return null; | |
258 | } | |
259 | ||
260 | T event = parseFirstLine(context.firstLineMatcher, context.firstLine); | |
261 | ||
262 | try { | |
263 | if (fFile.getFilePointer() != context.nextLineLocation) { | |
264 | fFile.seek(context.nextLineLocation); | |
265 | } | |
266 | long rawPos = fFile.getFilePointer(); | |
267 | String line = fFile.getNextLine(); | |
268 | while (line != null) { | |
269 | Matcher matcher = getFirstLinePattern().matcher(line); | |
13797536 MK |
270 | if (matcher.matches()) { |
271 | setupContext(context, rawPos, line, matcher); | |
bcb8c2cb PT |
272 | return event; |
273 | } | |
274 | parseNextLine(event, line); | |
275 | rawPos = fFile.getFilePointer(); | |
276 | line = fFile.getNextLine(); | |
277 | } | |
278 | } catch (IOException e) { | |
279 | Activator.logError("Error reading file: " + getPath(), e); //$NON-NLS-1$ | |
280 | } | |
281 | ||
282 | context.setLocation(NULL_LOCATION); | |
283 | return event; | |
284 | } | |
285 | ||
286 | /** | |
287 | * Gets the first line pattern. | |
288 | * | |
289 | * @return The first line pattern | |
290 | */ | |
291 | protected abstract Pattern getFirstLinePattern(); | |
292 | ||
293 | /** | |
6b44794a MK |
294 | * Parses the first line data and returns a new event. When constructing the |
295 | * event, the concrete trace should use the trace's timestamp transform to | |
296 | * create the timestamp, by either transforming the parsed time value | |
297 | * directly or by using the method {@link #createTimestamp(long)}. | |
bcb8c2cb PT |
298 | * |
299 | * @param matcher | |
300 | * The matcher | |
301 | * @param line | |
302 | * The line to parse | |
303 | * @return The parsed event | |
304 | */ | |
305 | protected abstract T parseFirstLine(Matcher matcher, String line); | |
306 | ||
307 | /** | |
308 | * Parses the next line data for the current event. | |
309 | * | |
310 | * @param event | |
311 | * The current event being parsed | |
312 | * @param line | |
313 | * The line to parse | |
314 | */ | |
315 | protected abstract void parseNextLine(T event, String line); | |
316 | ||
543e2b1e BH |
317 | /** |
318 | * Returns a ordered list of validation patterns that will be used in | |
319 | * the default {@link #validate(IProject, String)} method to match | |
320 | * the first 100 to compute the confidence level | |
321 | * | |
322 | * @return collection of patterns to validate against | |
a465519a | 323 | * @since 3.2 |
543e2b1e BH |
324 | */ |
325 | protected List<Pattern> getValidationPatterns() { | |
326 | return Collections.singletonList(getFirstLinePattern()); | |
327 | } | |
328 | ||
bcb8c2cb PT |
329 | // ------------------------------------------------------------------------ |
330 | // Helper methods | |
331 | // ------------------------------------------------------------------------ | |
332 | ||
333 | /** | |
334 | * Strip quotes surrounding a string | |
335 | * | |
336 | * @param input | |
337 | * The input string | |
338 | * @return The string without quotes | |
339 | */ | |
340 | protected static String replaceQuotes(String input) { | |
341 | String out = input.replaceAll("^\"|(\"\\s*)$", ""); //$NON-NLS-1$//$NON-NLS-2$ | |
342 | return out; | |
343 | } | |
344 | ||
345 | /** | |
346 | * Strip brackets surrounding a string | |
347 | * | |
348 | * @param input | |
349 | * The input string | |
350 | * @return The string without brackets | |
351 | */ | |
352 | protected static String replaceBrackets(String input) { | |
353 | String out = input.replaceAll("^\\{|(\\}\\s*)$", ""); //$NON-NLS-1$//$NON-NLS-2$ | |
354 | return out; | |
355 | } | |
356 | ||
738a2796 MAL |
357 | private static int fCheckpointSize = -1; |
358 | ||
359 | @Override | |
360 | public synchronized int getCheckpointSize() { | |
361 | if (fCheckpointSize == -1) { | |
362 | TmfCheckpoint c = new TmfCheckpoint(TmfTimestamp.ZERO, new TmfLongLocation(0L), 0); | |
363 | ByteBuffer b = ByteBuffer.allocate(ITmfCheckpoint.MAX_SERIALIZE_SIZE); | |
364 | b.clear(); | |
365 | c.serialize(b); | |
366 | fCheckpointSize = b.position(); | |
367 | } | |
368 | ||
369 | return fCheckpointSize; | |
370 | } | |
371 | ||
372 | @Override | |
373 | protected ITmfTraceIndexer createIndexer(int interval) { | |
374 | return new TmfBTreeTraceIndexer(this, interval); | |
375 | } | |
376 | ||
377 | @Override | |
378 | public ITmfLocation restoreLocation(ByteBuffer bufferIn) { | |
379 | return new TmfLongLocation(bufferIn); | |
380 | } | |
bcb8c2cb | 381 | } |