|
1
16
17 package xni;
18
19 import java.io.PrintWriter  ;
20
21 import org.apache.xerces.parsers.XMLDocumentParser;
22 import org.apache.xerces.xni.Augmentations;
23 import org.apache.xerces.xni.QName;
24 import org.apache.xerces.xni.XMLAttributes;
25 import org.apache.xerces.xni.XMLLocator;
26 import org.apache.xerces.xni.XMLString;
27 import org.apache.xerces.xni.XNIException;
28 import org.apache.xerces.xni.NamespaceContext;
29 import org.apache.xerces.xni.parser.XMLConfigurationException;
30 import org.apache.xerces.xni.parser.XMLErrorHandler;
31 import org.apache.xerces.xni.parser.XMLInputSource;
32 import org.apache.xerces.xni.parser.XMLParseException;
33 import org.apache.xerces.xni.parser.XMLParserConfiguration;
34
35
54 public class Counter
55 extends XMLDocumentParser
56 implements XMLErrorHandler {
57
58
62
64
65 protected static final String NAMESPACES_FEATURE_ID =
66 "http://xml.org/sax/features/namespaces";
67
68
69 protected static final String NAMESPACE_PREFIXES_FEATURE_ID =
70 "http://xml.org/sax/features/namespace-prefixes";
71
72
73 protected static final String VALIDATION_FEATURE_ID =
74 "http://xml.org/sax/features/validation";
75
76
77 protected static final String SCHEMA_VALIDATION_FEATURE_ID =
78 "http://apache.org/xml/features/validation/schema";
79
80
81 protected static final String SCHEMA_FULL_CHECKING_FEATURE_ID =
82 "http://apache.org/xml/features/validation/schema-full-checking";
83
84
86
87 protected static final String DEFAULT_PARSER_CONFIG =
88 "org.apache.xerces.parsers.XIncludeAwareParserConfiguration";
89
90
91 protected static final int DEFAULT_REPETITION = 1;
92
93
94 protected static final boolean DEFAULT_NAMESPACES = true;
95
96
97 protected static final boolean DEFAULT_NAMESPACE_PREFIXES = false;
98
99
100 protected static final boolean DEFAULT_VALIDATION = false;
101
102
103 protected static final boolean DEFAULT_SCHEMA_VALIDATION = false;
104
105
106 protected static final boolean DEFAULT_SCHEMA_FULL_CHECKING = false;
107
108
109 protected static final boolean DEFAULT_MEMORY_USAGE = false;
110
111
112 protected static final boolean DEFAULT_TAGGINESS = false;
113
114
118
119 protected long fElements;
120
121
122 protected long fAttributes;
123
124
125 protected long fCharacters;
126
127
128 protected long fIgnorableWhitespace;
129
130
131 protected long fTagCharacters;
132
133
134 protected long fOtherCharacters;
135
136
140
141 public Counter(XMLParserConfiguration configuration) {
142 super(configuration);
143 fConfiguration.setErrorHandler(this);
144 }
146
150
151 public void printResults(PrintWriter out, String uri, long time,
152 long memory, boolean tagginess,
153 int repetition) {
154
155 out.print(uri);
157 out.print(": ");
158 if (repetition == 1) {
159 out.print(time);
160 }
161 else {
162 out.print(time);
163 out.print('/');
164 out.print(repetition);
165 out.print('=');
166 out.print(time/repetition);
167 }
168 out.print(" ms");
169 if (memory != Long.MIN_VALUE) {
170 out.print(", ");
171 out.print(memory);
172 out.print(" bytes");
173 }
174 out.print(" (");
175 out.print(fElements);
176 out.print(" elems, ");
177 out.print(fAttributes);
178 out.print(" attrs, ");
179 out.print(fIgnorableWhitespace);
180 out.print(" spaces, ");
181 out.print(fCharacters);
182 out.print(" chars)");
183 if (tagginess) {
184 out.print(' ');
185 long totalCharacters = fTagCharacters + fOtherCharacters
186 + fCharacters + fIgnorableWhitespace;
187 long tagValue = fTagCharacters * 100 / totalCharacters;
188 out.print(tagValue);
189 out.print("% tagginess");
190 }
191 out.println();
192 out.flush();
193
194 }
196
200
201 public void startDocument(XMLLocator locator, String encoding, NamespaceContext namespaceContext, Augmentations augs)
202 throws XNIException {
203
204 fElements = 0;
205 fAttributes = 0;
206 fCharacters = 0;
207 fIgnorableWhitespace = 0;
208 fTagCharacters = 0;
209 fOtherCharacters = 0;
210
211 }
213
214 public void startElement(QName element, XMLAttributes attrs, Augmentations augs)
215 throws XNIException {
216
217 fElements++;
218 fTagCharacters++; fTagCharacters += element.rawname.length();
220 if (attrs != null) {
221 int attrCount = attrs.getLength();
222 fAttributes += attrCount;
223 for (int i = 0; i < attrCount; i++) {
224 fTagCharacters++; fTagCharacters += attrs.getQName(i).length();
226 fTagCharacters++; fTagCharacters++; fOtherCharacters += attrs.getValue(i).length();
229 fTagCharacters++; }
231 }
232 fTagCharacters++;
234 }
236
237 public void emptyElement(QName element, XMLAttributes attrs, Augmentations augs)
238 throws XNIException {
239
240 fElements++;
241 fTagCharacters++; fTagCharacters += element.rawname.length();
243 if (attrs != null) {
244 int attrCount = attrs.getLength();
245 fAttributes += attrCount;
246 for (int i = 0; i < attrCount; i++) {
247 fTagCharacters++; fTagCharacters += attrs.getQName(i).length();
249 fTagCharacters++; fTagCharacters++; fOtherCharacters += attrs.getValue(i).length();
252 fTagCharacters++; }
254 }
255 fTagCharacters++; fTagCharacters++;
258 }
260
261 public void characters(XMLString text, Augmentations augs) throws XNIException {
262
263 fCharacters += text.length;
264
265 }
267
268 public void ignorableWhitespace(XMLString text, Augmentations augs) throws XNIException {
269
270 fIgnorableWhitespace += text.length;
271
272 }
274
275 public void processingInstruction(String target, XMLString data, Augmentations augs)
276 throws XNIException {
277 fTagCharacters += 2; fTagCharacters += target.length();
279 if (data.length > 0) {
280 fTagCharacters++; fOtherCharacters += data.length;
282 }
283 fTagCharacters += 2; }
286
290
291 public void warning(String domain, String key, XMLParseException ex)
292 throws XNIException {
293 printError("Warning", ex);
294 }
296
297 public void error(String domain, String key, XMLParseException ex)
298 throws XNIException {
299 printError("Error", ex);
300 }
302
303 public void fatalError(String domain, String key, XMLParseException ex)
304 throws XNIException {
305 printError("Fatal Error", ex);
306 throw ex;
307 }
309
313
314 protected void printError(String type, XMLParseException ex) {
315
316 System.err.print("[");
317 System.err.print(type);
318 System.err.print("] ");
319 String systemId = ex.getExpandedSystemId();
320 if (systemId != null) {
321 int index = systemId.lastIndexOf('/');
322 if (index != -1)
323 systemId = systemId.substring(index + 1);
324 System.err.print(systemId);
325 }
326 System.err.print(':');
327 System.err.print(ex.getLineNumber());
328 System.err.print(':');
329 System.err.print(ex.getColumnNumber());
330 System.err.print(": ");
331 System.err.print(ex.getMessage());
332 System.err.println();
333 System.err.flush();
334
335 }
337
341
342 public static void main(String argv[]) {
343
344 if (argv.length == 0) {
346 printUsage();
347 System.exit(1);
348 }
349
350 PrintWriter out = new PrintWriter (System.out);
352 XMLDocumentParser parser = null;
353 XMLParserConfiguration parserConfig = null;
354 int repetition = DEFAULT_REPETITION;
355 boolean namespaces = DEFAULT_NAMESPACES;
356 boolean namespacePrefixes = DEFAULT_NAMESPACE_PREFIXES;
357 boolean validation = DEFAULT_VALIDATION;
358 boolean schemaValidation = DEFAULT_SCHEMA_VALIDATION;
359 boolean schemaFullChecking = DEFAULT_SCHEMA_FULL_CHECKING;
360 boolean memoryUsage = DEFAULT_MEMORY_USAGE;
361 boolean tagginess = DEFAULT_TAGGINESS;
362
363 for (int i = 0; i < argv.length; i++) {
365 String arg = argv[i];
366 if (arg.startsWith("-")) {
367 String option = arg.substring(1);
368 if (option.equals("p")) {
369 if (++i == argv.length) {
371 System.err.println("error: Missing argument to -p option.");
372 continue;
373 }
374 String parserName = argv[i];
375
376 try {
378 parserConfig = (XMLParserConfiguration)ObjectFactory.newInstance(parserName,
379 ObjectFactory.findClassLoader(), true);
380 parserConfig.addRecognizedFeatures(new String [] {
381 NAMESPACE_PREFIXES_FEATURE_ID,
382 });
383 parser = null;
384 }
385 catch (Exception e) {
386 parserConfig = null;
387 System.err.println("error: Unable to instantiate parser configuration ("+parserName+")");
388 }
389 continue;
390 }
391 if (option.equals("x")) {
392 if (++i == argv.length) {
393 System.err.println("error: Missing argument to -x option.");
394 continue;
395 }
396 String number = argv[i];
397 try {
398 int value = Integer.parseInt(number);
399 if (value < 1) {
400 System.err.println("error: Repetition must be at least 1.");
401 continue;
402 }
403 repetition = value;
404 }
405 catch (NumberFormatException e) {
406 System.err.println("error: invalid number ("+number+").");
407 }
408 continue;
409 }
410 if (option.equalsIgnoreCase("n")) {
411 namespaces = option.equals("n");
412 continue;
413 }
414 if (option.equalsIgnoreCase("np")) {
415 namespacePrefixes = option.equals("np");
416 continue;
417 }
418 if (option.equalsIgnoreCase("v")) {
419 validation = option.equals("v");
420 continue;
421 }
422 if (option.equalsIgnoreCase("s")) {
423 schemaValidation = option.equals("s");
424 continue;
425 }
426 if (option.equalsIgnoreCase("f")) {
427 schemaFullChecking = option.equals("f");
428 continue;
429 }
430 if (option.equalsIgnoreCase("m")) {
431 memoryUsage = option.equals("m");
432 continue;
433 }
434 if (option.equalsIgnoreCase("t")) {
435 tagginess = option.equals("t");
436 continue;
437 }
438 if (option.equals("-rem")) {
439 if (++i == argv.length) {
440 System.err.println("error: Missing argument to -# option.");
441 continue;
442 }
443 System.out.print("# ");
444 System.out.println(argv[i]);
445 continue;
446 }
447 if (option.equals("h")) {
448 printUsage();
449 continue;
450 }
451 System.err.println("error: unknown option ("+option+").");
452 continue;
453 }
454
455 if (parserConfig == null) {
457
458 try {
460 parserConfig = (XMLParserConfiguration)ObjectFactory.newInstance(DEFAULT_PARSER_CONFIG,
461 ObjectFactory.findClassLoader(), true);
462 parserConfig.addRecognizedFeatures(new String [] {
463 NAMESPACE_PREFIXES_FEATURE_ID,
464 });
465 }
466 catch (Exception e) {
467 System.err.println("error: Unable to instantiate parser configuration ("+DEFAULT_PARSER_CONFIG+")");
468 continue;
469 }
470 }
471
472 if (parser == null) {
474 parser = new Counter(parserConfig);
475 }
476 try {
477 parserConfig.setFeature(NAMESPACES_FEATURE_ID, namespaces);
478 }
479 catch (XMLConfigurationException e) {
480 System.err.println("warning: Parser does not support feature ("+NAMESPACES_FEATURE_ID+")");
481 }
482 try {
483 parserConfig.setFeature(VALIDATION_FEATURE_ID, validation);
484 }
485 catch (XMLConfigurationException e) {
486 System.err.println("warning: Parser does not support feature ("+VALIDATION_FEATURE_ID+")");
487 }
488 try {
489 parserConfig.setFeature(SCHEMA_VALIDATION_FEATURE_ID, schemaValidation);
490 }
491 catch (XMLConfigurationException e) {
492 if (e.getType() == XMLConfigurationException.NOT_SUPPORTED) {
493 System.err.println("warning: Parser does not support feature ("+SCHEMA_VALIDATION_FEATURE_ID+")");
494 }
495 }
496 try {
497 parserConfig.setFeature(SCHEMA_FULL_CHECKING_FEATURE_ID, schemaFullChecking);
498 }
499 catch (XMLConfigurationException e) {
500 if (e.getType() == XMLConfigurationException.NOT_SUPPORTED) {
501 System.err.println("warning: Parser does not support feature ("+SCHEMA_FULL_CHECKING_FEATURE_ID+")");
502 }
503 }
504
505 try {
507 long timeBefore = System.currentTimeMillis();
508 long memoryBefore = 0;
509 if(memoryUsage) {
510 System.gc();
511 memoryBefore = Runtime.getRuntime().freeMemory();
512 }
513 for (int j = 0; j < repetition; j++) {
514 parser.parse(new XMLInputSource(null, arg, null));
515 }
516 long memory = Long.MIN_VALUE;
517 if(memoryUsage) {
518 long memoryAfter = Runtime.getRuntime().freeMemory();
519 memory = memoryBefore - memoryAfter;
520 }
521
522 long timeAfter = System.currentTimeMillis();
523 long time = timeAfter - timeBefore;
524 ((Counter)parser).printResults(out, arg, time,
525 memory, tagginess,
526 repetition);
527 }
528 catch (XMLParseException e) {
529 }
531 catch (Exception e) {
532 System.err.println("error: Parse error occurred - "+e.getMessage());
533 if (e instanceof XNIException) {
534 e = ((XNIException)e).getException();
535 }
536 e.printStackTrace(System.err);
537 }
538 }
539
540 }
542
546
547 private static void printUsage() {
548
549 System.err.println("usage: java xni.Counter (options) uri ...");
550 System.err.println();
551
552 System.err.println("options:");
553 System.err.println(" -p name Select parser configuration by name.");
554 System.err.println(" -x number Select number of repetitions.");
555 System.err.println(" -n | -N Turn on/off namespace processing.");
556 System.err.println(" -np | -NP Turn on/off namespace prefixes.");
557 System.err.println(" NOTE: Requires use of -n.");
558 System.err.println(" -v | -V Turn on/off validation.");
559 System.err.println(" -s | -S Turn on/off Schema validation support.");
560 System.err.println(" NOTE: Not supported by all parser configurations.");
561 System.err.println(" -f | -F Turn on/off Schema full checking.");
562 System.err.println(" NOTE: Requires use of -s and not supported by all parsers.");
563 System.err.println(" -m | -M Turn on/off memory usage report.");
564 System.err.println(" -t | -T Turn on/off \"tagginess\" report.");
565 System.err.println(" --rem text Output user defined comment before next parse.");
566 System.err.println(" -h This help screen.");
567
568 System.err.println();
569 System.err.println("defaults:");
570 System.err.println(" Config: "+DEFAULT_PARSER_CONFIG);
571 System.err.println(" Repetition: "+DEFAULT_REPETITION);
572 System.err.print(" Namespaces: ");
573 System.err.println(DEFAULT_NAMESPACES ? "on" : "off");
574 System.err.print(" Prefixes: ");
575 System.err.println(DEFAULT_NAMESPACE_PREFIXES ? "on" : "off");
576 System.err.print(" Validation: ");
577 System.err.println(DEFAULT_VALIDATION ? "on" : "off");
578 System.err.print(" Schema: ");
579 System.err.println(DEFAULT_SCHEMA_VALIDATION ? "on" : "off");
580 System.err.print(" Schema full checking: ");
581 System.err.println(DEFAULT_SCHEMA_FULL_CHECKING ? "on" : "off");
582 System.err.print(" Memory: ");
583 System.err.println(DEFAULT_MEMORY_USAGE ? "on" : "off");
584 System.err.print(" Tagginess: ");
585 System.err.println(DEFAULT_TAGGINESS ? "on" : "off");
586
587 System.err.println();
588 System.err.println("notes:");
589 System.err.println(" The speed and memory results from this program should NOT be used as the");
590 System.err.println(" basis of parser performance comparison! Real analytical methods should be");
591 System.err.println(" used. For better results, perform multiple document parses within the same");
592 System.err.println(" virtual machine to remove class loading from parse time and memory usage.");
593 System.err.println();
594 System.err.println(" The \"tagginess\" measurement gives a rough estimate of the percentage of");
595 System.err.println(" markup versus content in the XML document. The percent tagginess of a ");
596 System.err.println(" document is equal to the minimum amount of tag characters required for ");
597 System.err.println(" elements, attributes, and processing instructions divided by the total");
598 System.err.println(" amount of characters (characters, ignorable whitespace, and tag characters)");
599 System.err.println(" in the document.");
600 System.err.println();
601 System.err.println(" Not all features are supported by different parser configurations.");
602
603 }
605 } |
Popular Tags |
Java API By Example, From Geeks To Geeks.