Commit | Line | Data |
---|---|---|
bcb8c2cb | 1 | /******************************************************************************* |
6c230e50 | 2 | * Copyright (c) 2012, 2015 Ericsson |
bcb8c2cb PT |
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; | |
6c230e50 | 28 | import org.eclipse.jdt.annotation.NonNull; |
2bdf0193 AM |
29 | import org.eclipse.tracecompass.internal.tmf.core.Activator; |
30 | import org.eclipse.tracecompass.tmf.core.event.ITmfEvent; | |
31 | import org.eclipse.tracecompass.tmf.core.exceptions.TmfTraceException; | |
32 | import org.eclipse.tracecompass.tmf.core.io.BufferedRandomAccessFile; | |
33 | import org.eclipse.tracecompass.tmf.core.timestamp.TmfTimestamp; | |
34 | import org.eclipse.tracecompass.tmf.core.trace.ITmfContext; | |
2bdf0193 | 35 | import org.eclipse.tracecompass.tmf.core.trace.TmfTrace; |
d26d67f5 | 36 | import org.eclipse.tracecompass.tmf.core.trace.TmfTraceUtils; |
2bdf0193 AM |
37 | import org.eclipse.tracecompass.tmf.core.trace.TraceValidationStatus; |
38 | import org.eclipse.tracecompass.tmf.core.trace.indexer.ITmfPersistentlyIndexable; | |
39 | import org.eclipse.tracecompass.tmf.core.trace.indexer.ITmfTraceIndexer; | |
40 | import org.eclipse.tracecompass.tmf.core.trace.indexer.TmfBTreeTraceIndexer; | |
41 | import org.eclipse.tracecompass.tmf.core.trace.indexer.checkpoint.ITmfCheckpoint; | |
42 | import org.eclipse.tracecompass.tmf.core.trace.indexer.checkpoint.TmfCheckpoint; | |
43 | import org.eclipse.tracecompass.tmf.core.trace.location.ITmfLocation; | |
44 | import org.eclipse.tracecompass.tmf.core.trace.location.TmfLongLocation; | |
bcb8c2cb PT |
45 | |
46 | /** | |
47 | * Extension of TmfTrace for handling of line-based text traces parsed using | |
48 | * regular expressions. Each line that matches the first line pattern indicates | |
49 | * the start of a new event. The subsequent lines can contain additional | |
50 | * information that is added to the current event. | |
51 | * | |
52 | * @param <T> | |
53 | * TmfEvent class returned by this trace | |
bcb8c2cb | 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 | } | |
d26d67f5 BH |
91 | try { |
92 | if (!TmfTraceUtils.isText(file)) { | |
93 | return new TraceValidationStatus(0, Activator.PLUGIN_ID); | |
94 | } | |
95 | } catch (IOException e) { | |
96 | Activator.logError("Error validating file: " + path, e); //$NON-NLS-1$ | |
97 | return new Status(IStatus.ERROR, Activator.PLUGIN_ID, "IOException validating file: " + path, e); //$NON-NLS-1$ | |
98 | } | |
bcb8c2cb PT |
99 | int confidence = 0; |
100 | try (BufferedRandomAccessFile rafile = new BufferedRandomAccessFile(path, "r")) { //$NON-NLS-1$ | |
101 | int lineCount = 0; | |
918551aa | 102 | double matches = 0.0; |
bcb8c2cb | 103 | String line = rafile.getNextLine(); |
543e2b1e | 104 | List<Pattern> validationPatterns = getValidationPatterns(); |
bcb8c2cb | 105 | while ((line != null) && (lineCount++ < MAX_LINES)) { |
6c230e50 | 106 | line = preProcessLine(line); |
543e2b1e BH |
107 | for(Pattern pattern : validationPatterns) { |
108 | Matcher matcher = pattern.matcher(line); | |
109 | if (matcher.matches()) { | |
110 | int groupCount = matcher.groupCount(); | |
111 | matches += (1.0 + groupCount / ((double) groupCount + 1)); | |
112 | } | |
bcb8c2cb | 113 | } |
918551aa | 114 | confidence = (int) (MAX_CONFIDENCE * matches / lineCount); |
bcb8c2cb PT |
115 | line = rafile.getNextLine(); |
116 | } | |
117 | } catch (IOException e) { | |
118 | Activator.logError("Error validating file: " + path, e); //$NON-NLS-1$ | |
119 | return new Status(IStatus.ERROR, Activator.PLUGIN_ID, "IOException validating file: " + path, e); //$NON-NLS-1$ | |
120 | } | |
121 | ||
122 | return new TraceValidationStatus(confidence, Activator.PLUGIN_ID); | |
123 | ||
124 | } | |
bcb8c2cb PT |
125 | @Override |
126 | public void initTrace(IResource resource, String path, Class<? extends ITmfEvent> type) throws TmfTraceException { | |
127 | super.initTrace(resource, path, type); | |
128 | try { | |
129 | fFile = new BufferedRandomAccessFile(getPath(), "r"); //$NON-NLS-1$ | |
130 | } catch (IOException e) { | |
131 | throw new TmfTraceException(e.getMessage(), e); | |
132 | } | |
133 | } | |
134 | ||
135 | @Override | |
136 | public synchronized void dispose() { | |
137 | super.dispose(); | |
138 | if (fFile != null) { | |
139 | try { | |
140 | fFile.close(); | |
141 | } catch (IOException e) { | |
142 | } finally { | |
143 | fFile = null; | |
144 | } | |
145 | } | |
146 | } | |
147 | ||
148 | @Override | |
5ece050b | 149 | public synchronized TextTraceContext seekEvent(ITmfLocation location) { |
bcb8c2cb PT |
150 | TextTraceContext context = new TextTraceContext(NULL_LOCATION, ITmfContext.UNKNOWN_RANK); |
151 | if (NULL_LOCATION.equals(location) || fFile == null) { | |
152 | return context; | |
153 | } | |
154 | try { | |
155 | if (location == null) { | |
156 | fFile.seek(0); | |
157 | } else if (location.getLocationInfo() instanceof Long) { | |
158 | fFile.seek((Long) location.getLocationInfo()); | |
159 | } | |
160 | long rawPos = fFile.getFilePointer(); | |
161 | String line = fFile.getNextLine(); | |
162 | while (line != null) { | |
6c230e50 | 163 | line = preProcessLine(line); |
bcb8c2cb | 164 | Matcher matcher = getFirstLinePattern().matcher(line); |
13797536 MK |
165 | if (matcher.matches()) { |
166 | setupContext(context, rawPos, line, matcher); | |
bcb8c2cb PT |
167 | return context; |
168 | } | |
169 | rawPos = fFile.getFilePointer(); | |
170 | line = fFile.getNextLine(); | |
171 | } | |
172 | return context; | |
173 | } catch (IOException e) { | |
174 | Activator.logError("Error seeking file: " + getPath(), e); //$NON-NLS-1$ | |
175 | return context; | |
176 | } | |
177 | } | |
178 | ||
13797536 MK |
179 | private void setupContext(TextTraceContext context, long rawPos, String line, Matcher matcher) throws IOException { |
180 | context.setLocation(new TmfLongLocation(rawPos)); | |
181 | context.firstLineMatcher = matcher; | |
182 | context.firstLine = line; | |
183 | context.nextLineLocation = fFile.getFilePointer(); | |
184 | } | |
185 | ||
bcb8c2cb | 186 | @Override |
5ece050b | 187 | public synchronized TextTraceContext seekEvent(double ratio) { |
bcb8c2cb PT |
188 | if (fFile == null) { |
189 | return new TextTraceContext(NULL_LOCATION, ITmfContext.UNKNOWN_RANK); | |
190 | } | |
191 | try { | |
a0789537 | 192 | long pos = Math.round(ratio * fFile.length()); |
bcb8c2cb PT |
193 | while (pos > 0) { |
194 | fFile.seek(pos - 1); | |
195 | if (fFile.read() == '\n') { | |
196 | break; | |
197 | } | |
198 | pos--; | |
199 | } | |
200 | ITmfLocation location = new TmfLongLocation(Long.valueOf(pos)); | |
5ece050b | 201 | TextTraceContext context = seekEvent(location); |
bcb8c2cb PT |
202 | context.setRank(ITmfContext.UNKNOWN_RANK); |
203 | return context; | |
204 | } catch (IOException e) { | |
205 | Activator.logError("Error seeking file: " + getPath(), e); //$NON-NLS-1$ | |
206 | return new TextTraceContext(NULL_LOCATION, ITmfContext.UNKNOWN_RANK); | |
207 | } | |
208 | } | |
209 | ||
210 | @Override | |
211 | public double getLocationRatio(ITmfLocation location) { | |
212 | if (fFile == null) { | |
213 | return 0; | |
214 | } | |
215 | try { | |
216 | long length = fFile.length(); | |
217 | if (length == 0) { | |
218 | return 0; | |
219 | } | |
220 | if (location.getLocationInfo() instanceof Long) { | |
221 | return (double) ((Long) location.getLocationInfo()) / length; | |
222 | } | |
223 | } catch (IOException e) { | |
224 | Activator.logError("Error reading file: " + getPath(), e); //$NON-NLS-1$ | |
225 | } | |
226 | return 0; | |
227 | } | |
228 | ||
229 | @Override | |
230 | public ITmfLocation getCurrentLocation() { | |
231 | return null; | |
232 | } | |
233 | ||
234 | @Override | |
5ece050b AM |
235 | public TextTraceEvent parseEvent(ITmfContext tmfContext) { |
236 | TextTraceContext context = seekEvent(tmfContext.getLocation()); | |
bcb8c2cb PT |
237 | return parse(context); |
238 | } | |
239 | ||
240 | @Override | |
241 | public synchronized T getNext(ITmfContext context) { | |
5ece050b AM |
242 | if (!(context instanceof TextTraceContext)) { |
243 | throw new IllegalArgumentException(); | |
244 | } | |
245 | TextTraceContext savedContext = new TextTraceContext(context.getLocation(), context.getRank()); | |
246 | T event = parse((TextTraceContext) context); | |
bcb8c2cb | 247 | if (event != null) { |
5904c11e | 248 | updateAttributes(savedContext, event); |
bcb8c2cb PT |
249 | context.increaseRank(); |
250 | } | |
251 | return event; | |
252 | } | |
253 | ||
254 | /** | |
255 | * Parse the next event. The context is advanced. | |
256 | * | |
257 | * @param tmfContext | |
258 | * the context | |
259 | * @return the next event or null | |
260 | */ | |
5ece050b | 261 | protected synchronized T parse(TextTraceContext tmfContext) { |
bcb8c2cb PT |
262 | if (fFile == null) { |
263 | return null; | |
264 | } | |
5ece050b | 265 | TextTraceContext context = tmfContext; |
bcb8c2cb PT |
266 | if (context.getLocation() == null || !(context.getLocation().getLocationInfo() instanceof Long) || NULL_LOCATION.equals(context.getLocation())) { |
267 | return null; | |
268 | } | |
269 | ||
270 | T event = parseFirstLine(context.firstLineMatcher, context.firstLine); | |
271 | ||
272 | try { | |
273 | if (fFile.getFilePointer() != context.nextLineLocation) { | |
274 | fFile.seek(context.nextLineLocation); | |
275 | } | |
276 | long rawPos = fFile.getFilePointer(); | |
277 | String line = fFile.getNextLine(); | |
278 | while (line != null) { | |
6c230e50 | 279 | line = preProcessLine(line); |
bcb8c2cb | 280 | Matcher matcher = getFirstLinePattern().matcher(line); |
13797536 MK |
281 | if (matcher.matches()) { |
282 | setupContext(context, rawPos, line, matcher); | |
bcb8c2cb PT |
283 | return event; |
284 | } | |
285 | parseNextLine(event, line); | |
286 | rawPos = fFile.getFilePointer(); | |
287 | line = fFile.getNextLine(); | |
288 | } | |
289 | } catch (IOException e) { | |
290 | Activator.logError("Error reading file: " + getPath(), e); //$NON-NLS-1$ | |
291 | } | |
292 | ||
293 | context.setLocation(NULL_LOCATION); | |
294 | return event; | |
295 | } | |
296 | ||
6c230e50 BH |
297 | /** |
298 | * Pre-processes the input line. The default implementation returns the | |
299 | * input line. | |
300 | * | |
301 | * @param line | |
302 | * non-null input string | |
303 | * @return the pre-processed input line | |
304 | */ | |
305 | @NonNull | |
306 | protected String preProcessLine(@NonNull String line) { | |
307 | return line; | |
308 | } | |
309 | ||
bcb8c2cb PT |
310 | /** |
311 | * Gets the first line pattern. | |
312 | * | |
313 | * @return The first line pattern | |
314 | */ | |
315 | protected abstract Pattern getFirstLinePattern(); | |
316 | ||
317 | /** | |
6b44794a MK |
318 | * Parses the first line data and returns a new event. When constructing the |
319 | * event, the concrete trace should use the trace's timestamp transform to | |
320 | * create the timestamp, by either transforming the parsed time value | |
321 | * directly or by using the method {@link #createTimestamp(long)}. | |
bcb8c2cb PT |
322 | * |
323 | * @param matcher | |
324 | * The matcher | |
325 | * @param line | |
326 | * The line to parse | |
327 | * @return The parsed event | |
328 | */ | |
329 | protected abstract T parseFirstLine(Matcher matcher, String line); | |
330 | ||
331 | /** | |
332 | * Parses the next line data for the current event. | |
333 | * | |
334 | * @param event | |
335 | * The current event being parsed | |
336 | * @param line | |
337 | * The line to parse | |
338 | */ | |
339 | protected abstract void parseNextLine(T event, String line); | |
340 | ||
543e2b1e BH |
341 | /** |
342 | * Returns a ordered list of validation patterns that will be used in | |
343 | * the default {@link #validate(IProject, String)} method to match | |
344 | * the first 100 to compute the confidence level | |
345 | * | |
346 | * @return collection of patterns to validate against | |
543e2b1e BH |
347 | */ |
348 | protected List<Pattern> getValidationPatterns() { | |
349 | return Collections.singletonList(getFirstLinePattern()); | |
350 | } | |
351 | ||
bcb8c2cb PT |
352 | // ------------------------------------------------------------------------ |
353 | // Helper methods | |
354 | // ------------------------------------------------------------------------ | |
355 | ||
356 | /** | |
357 | * Strip quotes surrounding a string | |
358 | * | |
359 | * @param input | |
360 | * The input string | |
361 | * @return The string without quotes | |
362 | */ | |
363 | protected static String replaceQuotes(String input) { | |
364 | String out = input.replaceAll("^\"|(\"\\s*)$", ""); //$NON-NLS-1$//$NON-NLS-2$ | |
365 | return out; | |
366 | } | |
367 | ||
368 | /** | |
369 | * Strip brackets surrounding a string | |
370 | * | |
371 | * @param input | |
372 | * The input string | |
373 | * @return The string without brackets | |
374 | */ | |
375 | protected static String replaceBrackets(String input) { | |
376 | String out = input.replaceAll("^\\{|(\\}\\s*)$", ""); //$NON-NLS-1$//$NON-NLS-2$ | |
377 | return out; | |
378 | } | |
379 | ||
738a2796 MAL |
380 | private static int fCheckpointSize = -1; |
381 | ||
382 | @Override | |
383 | public synchronized int getCheckpointSize() { | |
384 | if (fCheckpointSize == -1) { | |
385 | TmfCheckpoint c = new TmfCheckpoint(TmfTimestamp.ZERO, new TmfLongLocation(0L), 0); | |
386 | ByteBuffer b = ByteBuffer.allocate(ITmfCheckpoint.MAX_SERIALIZE_SIZE); | |
387 | b.clear(); | |
388 | c.serialize(b); | |
389 | fCheckpointSize = b.position(); | |
390 | } | |
391 | ||
392 | return fCheckpointSize; | |
393 | } | |
394 | ||
395 | @Override | |
396 | protected ITmfTraceIndexer createIndexer(int interval) { | |
397 | return new TmfBTreeTraceIndexer(this, interval); | |
398 | } | |
399 | ||
400 | @Override | |
401 | public ITmfLocation restoreLocation(ByteBuffer bufferIn) { | |
402 | return new TmfLongLocation(bufferIn); | |
403 | } | |
bcb8c2cb | 404 | } |