Subversion Repositories SmartDukaan

Rev

Go to most recent revision | Details | Last modification | View Log | RSS feed

Rev Author Line No. Line
17 naveen 1
/**
2
 * 
3
 */
4
package in.shop2020.metamodel.util;
5
 
20 naveen 6
import in.shop2020.metamodel.core.Bullet;
7
import in.shop2020.metamodel.core.CompositeDataObject;
21 naveen 8
import in.shop2020.metamodel.core.Entity;
20 naveen 9
import in.shop2020.metamodel.core.EnumDataObject;
21 naveen 10
import in.shop2020.metamodel.core.Feature;
34 naveen 11
import in.shop2020.metamodel.core.FreeformContent;
20 naveen 12
import in.shop2020.metamodel.core.PrimitiveDataObject;
21 naveen 13
import in.shop2020.metamodel.core.Slide;
19 naveen 14
import in.shop2020.metamodel.definitions.BulletDefinition;
17 naveen 15
import in.shop2020.metamodel.definitions.Catalog;
16
import in.shop2020.metamodel.definitions.Category;
19 naveen 17
import in.shop2020.metamodel.definitions.CompositeDefinition;
20 naveen 18
import in.shop2020.metamodel.definitions.CompositePartDefinition;
18 naveen 19
import in.shop2020.metamodel.definitions.DatatypeDefinition;
17 naveen 20
import in.shop2020.metamodel.definitions.DefinitionsContainer;
18 naveen 21
import in.shop2020.metamodel.definitions.EditorialImportance;
21 naveen 22
import in.shop2020.metamodel.definitions.EntityContainer;
19 naveen 23
import in.shop2020.metamodel.definitions.EnumDefinition;
18 naveen 24
import in.shop2020.metamodel.definitions.FeatureDefinition;
25
import in.shop2020.metamodel.definitions.SlideDefinition;
34 naveen 26
import in.shop2020.metamodel.definitions.SlideFeatureDefinition;
18 naveen 27
import in.shop2020.metamodel.definitions.Unit;
64 naveen 28
import in.shop2020.util.DBUtils;
18 naveen 29
import in.shop2020.util.Utils;
17 naveen 30
 
22 naveen 31
import java.io.BufferedReader;
32
import java.io.FileReader;
33
import java.io.LineNumberReader;
34
import java.util.ArrayList;
35
import java.util.Collection;
36
import java.util.HashMap;
37
import java.util.List;
38
import java.util.Map;
39
import java.util.TreeMap;
40
 
17 naveen 41
import org.apache.commons.lang.ArrayUtils;
42
import org.apache.commons.lang.StringUtils;
43
 
44
/**
49 naveen 45
 * NO LONGER USED - POI BASED DATA EXTRACTION FROM MS PPT FILE HAS REPLACED IT
46
 * 
17 naveen 47
 * CN - Content Tool
49 naveen 48
 * Imports content from text and exports content. Validates against content model
17 naveen 49
 * 
50
 * @author naveen
51
 *
52
 */
53
public class CN {
54
	public static final String CONTENT_SRC_TXT_PATH = 
26 naveen 55
		"/home/naveen/workspace/eclipse/content/txt/";
17 naveen 56
	public static final String CONTENT_SRC_HTML_PATH = 
26 naveen 57
		"/home/naveen/workspace/eclipse/content/html/";
17 naveen 58
	public static final String CONTENT_SRC_XML_PATH = 
26 naveen 59
		"/home/naveen/workspace/eclipse/content/xml/";
34 naveen 60
 
61
	//** LOCAL
17 naveen 62
	public static final String CONTENT_DB_PATH =  
26 naveen 63
		"/home/naveen/workspace/eclipse/db/entities/";
18 naveen 64
 
34 naveen 65
	/** WEB
66
	public static final String CONTENT_DB_PATH =  
67
		"/var/lib/tomcat6/webapps/shop2020/db/entities/";
68
	*/
69
 
70
 
17 naveen 71
	// Known patterns
72
	public static final String PATTERN_LEVEL_1 = "* ";
73
	public static final String PATTERN_LEVEL_2 = "   * ";
74
	public static final String PATTERN_LEVEL_3 = "      * ";
75
	public static final String PATTERN_LEVEL_4 = "         * ";
76
 
77
	private long categoryID;
78
	private String srcFile;
79
	private String dbFile;
80
 
22 naveen 81
	private String introductionTitle;
21 naveen 82
 
28 naveen 83
	// Slide names in sequence
84
	private List<String> slideNames = new ArrayList<String>();
85
 
22 naveen 86
	private Map<String, Collection<String>> slideNameToFeatureLabels = 
87
		new HashMap<String, Collection<String>>();
88
 
19 naveen 89
	private Map<String, Long> slideNameToSlideDefID = 
90
		new HashMap<String, Long>();
91
 
92
	private Map<String, Long> slideNameFeatureNameToFeatureDefID =
93
		new HashMap<String, Long>();
94
 
20 naveen 95
	private Map<String, List<Bullet>> slideNameFeatureNameToBullets =
96
		new HashMap<String, List<Bullet>>();
19 naveen 97
 
34 naveen 98
	private Map<String, String> slideNameToFreeformContent = 
99
		new HashMap<String, String>();
100
 
17 naveen 101
	/**
102
	 * @param args
103
	 */
104
	public static void main(String[] args) throws Exception {
79 naveen 105
		String[] commands = new String[] {"show", "import", "learn"};
17 naveen 106
 
107
		String usage = "Usage: CN ["+ StringUtils.join(commands, "|") +
25 naveen 108
			"] [{Entity ID}|{Category ID} {Content file name}]\n";
17 naveen 109
 
18 naveen 110
 
17 naveen 111
		String inputCommand = args[0];
25 naveen 112
 
17 naveen 113
		if(!ArrayUtils.contains(commands, inputCommand)) {
114
			System.out.println(usage);
115
			System.exit(-1);
116
		}
25 naveen 117
 
118
		if (inputCommand.equals("show")) {
79 naveen 119
			if(args.length < 2) {
120
				System.out.println(usage);
121
				System.exit(-1);
122
			}
123
 
25 naveen 124
			String entityID = args[1];
125
			CN cn = new CN();
126
			cn.showEntity(new Long(entityID).longValue());
127
			System.exit(0);
128
		}
17 naveen 129
 
25 naveen 130
		if (inputCommand.equals("import")) {
79 naveen 131
			if(args.length < 2) {
132
				System.out.println(usage);
133
				System.exit(-1);
134
			}
135
 
25 naveen 136
			String inputCategoryID = args[1];
137
			if(args.length < 3) {
138
				System.out.println(usage);
139
				System.exit(-1);
140
			}
141
 
142
			String inputFilename = args[2];
143
			long categoryID = new Long(inputCategoryID).longValue();
144
			CN cn = new CN(categoryID, inputFilename);
17 naveen 145
			cn.importEntity();
25 naveen 146
			System.exit(0);
17 naveen 147
		}
79 naveen 148
 
149
		if (inputCommand.equals("learn")) {
150
			CN cn = new CN();
151
			cn.learn();
152
		}
17 naveen 153
	}
154
 
155
	/**
156
	 * 
25 naveen 157
	 */
158
	public CN() {
159
	}
160
 
161
	/**
162
	 * 
17 naveen 163
	 * @param categoryID
164
	 * @param fileName
165
	 */
166
	public CN(long categoryID, String fileName) {
167
		this.categoryID = categoryID;
168
 
169
		this.srcFile = CONTENT_SRC_TXT_PATH + fileName + ".txt";
170
		this.dbFile = CONTENT_DB_PATH + "entities" + ".ser";
171
	}
172
 
173
	/**
79 naveen 174
	 * Learns all features where definition is marked "learned=true"
175
	 * @throws Exception 
176
	 */
177
	public void learn() throws Exception {
178
		Map<Long, List<Bullet>> learnedBullets = 
179
			new HashMap<Long, List<Bullet>>();
180
 
181
		EntityContainer entContainer = 
182
			Catalog.getInstance().getEntityContainer();
183
 
184
		Map<Long, Entity> entities = entContainer.getEntities();
185
 
186
		for(Entity entity : entities.values()) {
187
			List<Slide> slides = entity.getSlides();
188
 
189
			for(Slide slide : slides) {
190
				this.learn(slide, learnedBullets);
191
			}
192
		}
193
 
194
		Utils.info("learnedBullets=" + learnedBullets);
195
 
196
		String learnedBulletsDBFile = CN.CONTENT_DB_PATH + "learnedbullets.ser";
197
		DBUtils.store(learnedBullets, learnedBulletsDBFile);
198
	}
199
 
200
	/**
17 naveen 201
	 * 
79 naveen 202
	 * @param slide
203
	 * @param learnedBullets
204
	 * @throws Exception 
205
	 */
206
	private void learn(Slide slide, Map<Long, List<Bullet>> learnedBullets) 
207
		throws Exception {
208
 
209
		DefinitionsContainer defs = 
210
			Catalog.getInstance().getDefinitionsContainer();
211
 
212
		// Features
213
		List<Feature> features = slide.getFeatures();
214
 
215
		if(features != null) {
216
			for(Feature feature : features) {
217
				Long featureDefID = feature.getFeatureDefinitionID();
218
 
219
				FeatureDefinition featureDef = 
220
					defs.getFeatureDefinition(featureDefID);
221
 
222
				BulletDefinition bulletDef = featureDef.getBulletDefinition();
223
 
224
				if (!bulletDef.isLearned()) {
225
					continue;
226
				}
227
 
228
				List<Bullet> bullets = feature.getBullets();
229
				if(bullets != null) {
230
 
231
					List<Bullet> collectedBullets = 
232
						learnedBullets.get(new Long(featureDefID));
233
 
234
					if(collectedBullets == null) {
235
						collectedBullets = new ArrayList<Bullet>();
236
					}
237
 
238
					for(Bullet bullet : bullets) {
239
						if(!collectedBullets.contains(bullet)) {
240
							collectedBullets.add(bullet);
241
						}
242
					}
243
 
244
					learnedBullets.put(new Long(featureDefID), 
245
							collectedBullets);
246
				}
247
 
248
			}
249
		}
250
 
251
		// Children slides
252
		List<Slide> childrenSlides = slide.getChildrenSlides();
253
 
254
		if(childrenSlides != null) {
255
			for(Slide childSlide : childrenSlides) {
256
				this.learn(childSlide, learnedBullets);
257
			}
258
		}
259
	}
260
 
261
	/**
262
	 * 
17 naveen 263
	 * @throws Exception
264
	 */
25 naveen 265
	public void showEntity(long entityID) throws Exception {
266
		EntityContainer entContainer = 
267
			Catalog.getInstance().getEntityContainer();
268
 
33 naveen 269
		ExpandedEntity expandedEntity = 
270
			entContainer.getExpandedEntity(entityID);
271
 
272
		Utils.logger.info("expandedEntity=" + expandedEntity);
40 naveen 273
 
274
		/*
275
		Entity entity = entContainer.getEntity(entityID);
276
		Utils.logger.info("Entity=" + entity);
277
		*/
278
		List<ExpandedSlide> expslides = expandedEntity.getExpandedSlides();
279
 
280
		for(ExpandedSlide expslide : expslides) {
281
			String slidelabel = expslide.getSlideDefinition().getLabel();
282
			System.out.println(slidelabel);
283
 
284
			List<ExpandedFeature> expfeatures = expslide.getExpandedFeatures();
285
 
286
			for(ExpandedFeature expfeature : expfeatures ) {
287
				String featurelabel = expfeature.getFeatureDefinition().getLabel();
288
				System.out.println("\t"+featurelabel);
289
 
290
				List<ExpandedBullet> expbullets = 
291
						expfeature.getExpandedBullets();
292
 
293
				if(expbullets == null) continue;
294
 
295
				for (ExpandedBullet expbullet : expbullets) {
296
					System.out.println("\t\t" + expbullet.displayText());
297
				}
298
			}
299
		}
25 naveen 300
	}
301
 
302
	/**
303
	 * 
304
	 * @throws Exception
305
	 */
18 naveen 306
	public void importEntity() throws Exception {		
307
		//this.test();
308
		Utils.logger.info(this.srcFile);
17 naveen 309
 
18 naveen 310
		FileReader fileReader = new FileReader(this.srcFile);
311
		BufferedReader lineReader = new BufferedReader(fileReader);
312
		LineNumberReader reader = new LineNumberReader(lineReader);
17 naveen 313
 
19 naveen 314
		List<String> allLines = new ArrayList<String>();
315
 
18 naveen 316
		// 1 Collect all candidate slide names
19 naveen 317
		// Order of keys need to be restored
318
		Map<Integer, String> candidateSlideNames = 
319
			new TreeMap<Integer, String>();
320
 
321
		int lineCount = 0;
22 naveen 322
		boolean first = true;
18 naveen 323
		while(true) {
19 naveen 324
			String line = reader.readLine();			
18 naveen 325
			if(line == null) {
326
				break;
327
			}
328
 
19 naveen 329
			allLines.add(line);
330
			lineCount++;
331
 
18 naveen 332
			if(line.isEmpty()) {
333
				continue;
334
			}
22 naveen 335
 
18 naveen 336
 
337
			int lineNumber = reader.getLineNumber();
338
 
22 naveen 339
			Utils.logger.info(line);
18 naveen 340
			//Utils.logger.info("lineNumber="+lineNumber);
22 naveen 341
 
342
			// First non-empty line is introduction title - brand, model
343
			if(first) {
344
				this.introductionTitle = StringUtils.strip(line);
345
				first = false;
346
			}
18 naveen 347
 
348
			// RE-VISIT
349
			// 1.1 Slide names starting with Alpha
350
			//Utils.logger.info(line.substring(0, 1));
351
			if(StringUtils.isAlpha((line.substring(0, 1)))) {
352
				//Utils.logger.info(line);
353
				//Utils.logger.info("Level 0");
28 naveen 354
				candidateSlideNames.put(new Integer(lineNumber), 
355
						StringUtils.strip(line));
18 naveen 356
			}
357
		}
22 naveen 358
		Utils.logger.info("introductionTitle=" + this.introductionTitle);
359
		//System.exit(0);
19 naveen 360
 
361
		lineReader.close();
362
 
363
		Utils.logger.info("lineCount=" + lineCount);
18 naveen 364
		Utils.logger.info(candidateSlideNames.toString());
17 naveen 365
 
18 naveen 366
		// 2 Filter and validate parent slide names
367
		Map<Integer, String> slideNames = 
368
			this.processSlides(candidateSlideNames);
369
 
370
		Utils.logger.info(slideNames.toString());
371
 
19 naveen 372
		// 3 Collect all children slides
373
		// 4 Filter and validate children slide names
18 naveen 374
		// TODO
375
 
19 naveen 376
		// 5 Collect all Features for a slide
377
		// 5.1 Find line number range between which content for feature is found
378
		Map<String, int[]> featureLineNumberRangePerSlide = 
379
			new HashMap<String, int[]>();
380
 
381
		Utils.logger.info("slideNames.keySet=" + 
382
				slideNames.keySet().toString());
18 naveen 383
 
19 naveen 384
		Integer[] slideLineNumbers = slideNames.keySet().toArray(
385
				new Integer[0]);
18 naveen 386
 
19 naveen 387
		for(int i=0; i<slideLineNumbers.length-1; i++) {
388
			int from = slideLineNumbers[i]+1;
389
			int to = slideLineNumbers[i+1]-1;
390
			Utils.logger.info("from=" + from + " to=" + to);
391
 
392
			String tSlideName = slideNames.get(
393
					new Integer(slideLineNumbers[i]));
394
 
395
			Utils.logger.info("tSlideName=" + tSlideName);
396
 
397
			featureLineNumberRangePerSlide.put(tSlideName, 
398
					new int[] {from, to});
399
		}
400
 
401
		// For last slide
402
		Utils.logger.info("tSlideName=" + 
403
				slideNames.get(new Integer(
404
						slideLineNumbers[slideLineNumbers.length-1])));
405
 
406
		Utils.logger.info("lineNumbers[lineNumbers.length-1]=" + 
407
				slideLineNumbers[slideLineNumbers.length-1]);
408
 
409
		Utils.logger.info(
410
			"from=" + (slideLineNumbers[slideLineNumbers.length-1]+1) + " to=" +
411
			(lineCount-1));
412
 
413
		featureLineNumberRangePerSlide.put(
414
			slideNames.get(
415
				new Integer(slideLineNumbers[slideLineNumbers.length-1])), 
416
				new int[] {(slideLineNumbers[slideLineNumbers.length-1] + 1), 
417
				(lineCount-1)});
418
 
419
		Utils.logger.info("featureLineNumberRangePerSlide=" + 
420
				featureLineNumberRangePerSlide.toString());
421
 
422
		// 6 Filter and validate features
423
		DefinitionsContainer defs = 
424
			Catalog.getInstance().getDefinitionsContainer();
425
 
22 naveen 426
		Map<String, Map<Integer, String>> slideFeatureLabels = 
427
			new HashMap<String, Map<Integer, String>>();
19 naveen 428
 
429
		for(String slideName : slideNames.values()) {
430
			//Utils.logger.info("slideName=" + slideName);
431
			List<SlideDefinition> slideDefs = 
32 naveen 432
				defs.getSlideDefinitions(this.categoryID, slideName);
19 naveen 433
 
434
			if(slideDefs == null) {
435
				continue;
436
			}
437
 
438
			if (slideDefs.isEmpty()) {
439
				continue;
440
			} 
441
 
442
			// RE-VISIT
443
			// Pick the first
34 naveen 444
			SlideDefinition slideDef = slideDefs.get(0);
445
			Utils.logger.info("slideDef=" + slideDef);
446
 
447
			Long slideID = slideDef.getID();
19 naveen 448
			this.slideNameToSlideDefID.put(slideName, slideID);
449
 
34 naveen 450
			// If there are NO children slides or features defined for the slide
451
			// Every under it is free-form content
452
			List<Long> childrenSlideDefIDs = 
453
				slideDef.getChildrenSlideDefinitionIDs();
454
 
455
			List<SlideFeatureDefinition> slideFeatureDef = 
456
				slideDef.getSlideFeatureDefinitions();
457
 
458
			if((childrenSlideDefIDs == null || childrenSlideDefIDs.size() == 0) 
459
				&& (slideFeatureDef == null || slideFeatureDef.size() == 0)){
460
 
461
				int[] slideFFCRange = 
462
					featureLineNumberRangePerSlide.get(slideName);
463
 
464
				StringBuffer ffcSB = new StringBuffer();
465
				Utils.logger.info("slideFFCRange[0]=" + slideFFCRange[0]);
466
				Utils.logger.info("slideFFCRange[1]=" + slideFFCRange[1]);
467
 
468
				for(int i=slideFFCRange[0]; i<slideFFCRange[1]; i++) {
469
					ffcSB.append(allLines.get(i));
470
				}
471
 
472
				Utils.logger.info(slideName + " " + " ffcSB.toString()=" + 
473
						ffcSB.toString());
474
 
475
				this.slideNameToFreeformContent.put(slideName, 
476
						ffcSB.toString());
477
 
478
				continue;
479
			}
480
 
19 naveen 481
			Map<Integer, String> featureLabels = 
34 naveen 482
				this.processFeature(allLines, slideDef, 
19 naveen 483
					featureLineNumberRangePerSlide.get(slideName));
484
 
485
			// Find feature definition IDs
486
			for(String featureLabel : featureLabels.values()) {
487
				FeatureDefinition featureDef = defs.getFeatureDefinition(
488
						slideID.longValue(), featureLabel);
489
 
490
				this.slideNameFeatureNameToFeatureDefID.put(
491
						(slideName + "_" + featureLabel), 
492
						new Long(featureDef.getID()));
493
			}
494
 
495
			slideFeatureLabels.put(slideName, featureLabels);
21 naveen 496
			this.slideNameToFeatureLabels.put(slideName, 
497
					featureLabels.values());
19 naveen 498
		}
499
 
500
		Utils.logger.info("slideFeatureLabels=" + slideFeatureLabels);
21 naveen 501
		Utils.logger.info("slideNameToFeatureLabels=" + 
502
				this.slideNameToFeatureLabels);
19 naveen 503
 
504
		Utils.logger.info("slideNameToSlideDefID=" + 
505
				this.slideNameToSlideDefID);
506
 
507
		Utils.logger.info("slideNameFeatureNameToFeatureDefID=" + 
508
				this.slideNameFeatureNameToFeatureDefID);
509
 
510
		// 7 Bullets
511
		// Collect all bullets
512
		for(String slideName : slideFeatureLabels.keySet()) {
513
			Map<Integer, String> featureLabels = 
514
				slideFeatureLabels.get(slideName);
29 naveen 515
 
516
			if(featureLabels == null || featureLabels.size() == 0) {
517
				continue;
518
			}
519
 
19 naveen 520
			// 7.1 Find line number range between which content for bullets 
521
			// is found (per feature on this slide)
522
			Map<String, int[]> bulletLineNumberRangePerFeature = 
523
				this.getBulletLineRange(allLines, featureLabels);
524
 
525
			for(String featureLabel : bulletLineNumberRangePerFeature.keySet()){
526
 
527
				String key = slideName + "_" + featureLabel;
528
				Long featureDefID = 
529
					this.slideNameFeatureNameToFeatureDefID.get(key);
530
				FeatureDefinition featureDef = 
531
					defs.getFeatureDefinition(featureDefID.longValue());
532
 
533
				Long slideDefID = this.slideNameToSlideDefID.get(slideName);
534
				SlideDefinition slideDef = defs.getSlideDefinition(slideDefID);
535
 
536
				this.processBullets(allLines, 
537
						bulletLineNumberRangePerFeature.get(featureLabel), 
538
						featureDef, slideDef);
539
			}
540
		}
541
 
21 naveen 542
		// Construct Content Model
543
		Entity entity = this.contructEntityObject();
544
		Utils.logger.info("entity="+entity);
545
 
546
		EntityContainer entContainer = 
547
			Catalog.getInstance().getEntityContainer();
548
 
549
		entContainer.addEntity(entity);
550
 
551
		// RE-VISIT
552
		// Store it back
553
		DBUtils.store(entContainer.getEntities(), this.dbFile);
22 naveen 554
 
555
		// Store the index separately
556
		String entitiesbycategoryDBFile = CN.CONTENT_DB_PATH + 
557
			"entitiesbycategory" + ".ser";
34 naveen 558
 
559
		DBUtils.store(entContainer.getEntitiesbyCategory(), 
560
				entitiesbycategoryDBFile);
17 naveen 561
	}
562
 
563
	/**
564
	 * 
21 naveen 565
	 * @return Entity 
566
	 * @throws Exception 
567
	 */
568
	private Entity contructEntityObject() throws Exception {
569
		SequenceGenerator sg = SequenceGenerator.getInstance();
570
		long entityID = sg.getNextSequence(SequenceGenerator.ENTITY);
571
 
572
		Entity entity = new Entity(entityID, this.categoryID);
22 naveen 573
		String brandModel[] = StringUtils.split(this.introductionTitle, " ");
21 naveen 574
 
22 naveen 575
		if(brandModel.length == 1) {
576
			entity.setBrand(StringUtils.strip(brandModel[0]));
577
		}
578
		else if(brandModel.length == 2) {
579
			entity.setBrand(StringUtils.strip(brandModel[0]));
580
			entity.setModelName(StringUtils.strip(brandModel[1]));
581
		}
582
		else if(brandModel.length == 3) {
583
			entity.setBrand(StringUtils.strip(brandModel[0]));
584
			entity.setModelNumber(StringUtils.strip(brandModel[1]));
585
			entity.setModelName(StringUtils.strip(brandModel[2]));
586
		}
35 naveen 587
		else if(brandModel.length == 4) {
588
			entity.setBrand(StringUtils.strip(brandModel[0]) + " " + 
589
					StringUtils.strip(brandModel[1]));
22 naveen 590
 
35 naveen 591
			entity.setModelNumber(StringUtils.strip(brandModel[2]));
592
			entity.setModelName(StringUtils.strip(brandModel[3]));
593
		}
594
 
28 naveen 595
		for(String slideName : this.slideNames) {
21 naveen 596
			Collection<String> featureNames = 
597
				this.slideNameToFeatureLabels.get(slideName);
598
 
599
			Long slideDefID = this.slideNameToSlideDefID.get(slideName);
600
			Slide slide = new Slide(slideDefID.longValue());
601
 
34 naveen 602
			// Only if there are features
603
			if(featureNames != null) {
604
				List<Feature> features = new ArrayList<Feature>();
21 naveen 605
 
34 naveen 606
	 			for(String featureName : featureNames) {
607
					Long featureDefID = this.slideNameFeatureNameToFeatureDefID.get(
608
							slideName + "_" + featureName);
609
 
610
					List<Bullet> bullets = this.slideNameFeatureNameToBullets.get(
611
							slideName + "_" + featureName);
612
 
613
					Feature feature = new Feature(featureDefID.longValue());
614
					feature.setBullets(bullets);
615
 
616
					features.add(feature);
617
				}
21 naveen 618
 
34 naveen 619
				slide.setFeatures(features);
21 naveen 620
			}
621
 
34 naveen 622
			// Add free-form content is collect above
623
			String ffc = this.slideNameToFreeformContent.get(slideName);
624
			Utils.logger.info("slideName=" + slideName + " ffc=" + ffc);
625
 
626
			if(ffc != null) {
627
				slide.setFreeformContent(new FreeformContent(ffc));
628
			}
629
 
21 naveen 630
			entity.addSlide(slide);
631
		}
632
 
633
		return entity;
634
	}
635
 
636
	/**
637
	 * 
19 naveen 638
	 * @param allLines
639
	 * @param bulletLineNumberRange
640
	 * @param featureDef
641
	 * @throws Exception 
642
	 */
643
	private void processBullets(List<String> allLines, 
644
			int[] bulletLineNumberRange, FeatureDefinition featureDef, 
645
			SlideDefinition slideDef) throws Exception {
646
 
647
		int from = bulletLineNumberRange[0];
648
		int to = bulletLineNumberRange[1];
649
 
650
		BulletDefinition bulletDef = featureDef.getBulletDefinition();
651
		Utils.logger.info("bulletDefinition=" + bulletDef);
652
 
653
		DefinitionsContainer defs = 
654
			Catalog.getInstance().getDefinitionsContainer();
655
 
656
		String crumb = "\"" + slideDef.getLabel() + "\" > \"" + 
657
			featureDef.getLabel() + "\"";
658
 
659
		// Enforce isMultivalue
32 naveen 660
		// RE-VISIT - This validation should be inside for loop below
661
		// Skipping for now this may not be a common error
662
		/*
19 naveen 663
		if(!bulletDef.isMultivalue() && (from != to)) {
664
			Utils.logger.severe("Only one bullet is allowed for " + crumb);
665
			return;
666
		}
32 naveen 667
		*/
18 naveen 668
 
19 naveen 669
		// If unit is defined
670
		Unit unitDef =  null;
671
		if(bulletDef.getUnitID() != 0L) {
672
			long unitID = bulletDef.getUnitID();
673
			unitDef = defs.getUnit(unitID);
674
			Utils.logger.info("unitDef=" + unitDef);
675
		}
676
 
677
		long datatypeDefID = bulletDef.getDatatypeDefinitionID();
678
		DatatypeDefinition datatypeDef = 
679
			defs.getDatatypeDefinition(datatypeDefID);
680
		Utils.logger.info("datatypeDef=" + datatypeDef);
681
 
20 naveen 682
 
28 naveen 683
		String key = StringUtils.lowerCase(slideDef.getLabel()) + "_" + 
684
			StringUtils.lowerCase(featureDef.getLabel());
685
 
20 naveen 686
		List<Bullet> bullets = 
687
			this.slideNameFeatureNameToBullets.get(key);
688
 
689
		if(bullets == null) {
690
			bullets = new ArrayList<Bullet>();
691
			this.slideNameFeatureNameToBullets.put(key, bullets);
692
		}
693
 
19 naveen 694
		// If primitive
695
		boolean isEnum = false;
696
		boolean isComposite = false;
697
		boolean isPrimitive = false;
698
		if(datatypeDef instanceof EnumDefinition) {
699
			isEnum = true;
700
		}
701
		else if(datatypeDef instanceof CompositeDefinition) {
702
			isComposite = true;
703
		}
704
		else {
705
			isPrimitive = true;
706
		}
707
 
708
		Utils.logger.info("isEnum=" + isEnum + " isComposite=" + isComposite +
709
				" isPrimitive=" + isPrimitive);
710
 
29 naveen 711
		Utils.logger.info("from=" + from + " to=" + to);
19 naveen 712
		for(int i=from; i<to+1; i++) {
713
			String line = allLines.get(i);
29 naveen 714
			Utils.logger.info("Line " + i + ": " + line);
715
 
716
			// VALIDATE
32 naveen 717
			// It means free-form content formatted at feature level
29 naveen 718
			// Assume there are no more bullets
719
			if(StringUtils.startsWith(line, CN.PATTERN_LEVEL_1)) {
720
				break;
721
			}
19 naveen 722
 
723
			if(!StringUtils.startsWith(line, CN.PATTERN_LEVEL_2)) {
724
				continue;
725
			}
726
 
727
			String bulletText = StringUtils.strip(StringUtils.stripStart(line, 
728
					CN.PATTERN_LEVEL_2));
729
			Utils.logger.info("bulletText=" + bulletText);
730
 
731
			// if no unit is defined for this bullet whole is treated as value
732
			String bulletValue = bulletText;
733
 
734
			// If unit is defined
735
			if(bulletDef.getUnitID() != 0L) {
736
 
737
				// Validate unit
738
				String[] parts = StringUtils.split(bulletText, " ");
739
				if(parts.length < 2) {
740
					Utils.logger.severe("Unit is missing, " + crumb + " = " + 
741
							bulletText);
742
 
743
					continue;
744
				}
745
 
746
				if(parts.length > 2) {
747
					Utils.logger.severe("Invalid value, " + crumb + " = " + 
748
							bulletText);
749
 
750
					continue;
751
				}
752
 
753
				bulletValue = parts[0];
754
				String unitValue = parts[1];
755
				Utils.logger.info("unitValue="+unitValue);
756
 
757
				if(!(unitValue.equalsIgnoreCase(unitDef.getShortForm()) ||
758
						unitValue.equalsIgnoreCase(unitDef.getFullForm()))) {
759
					Utils.logger.severe("Invalid unit, " + crumb + " = " + 
760
							bulletText);
761
 
762
					continue;
763
				}
764
			}
765
 
766
			// Validate bullet value
767
			Utils.logger.info("bulletValue=" + bulletValue);
768
			if(isPrimitive) {
20 naveen 769
				if(!this.validatePrimitive(bulletValue, datatypeDef, crumb)) {
770
					continue;
19 naveen 771
				}
772
 
20 naveen 773
				bullets.add(new Bullet(new PrimitiveDataObject(bulletValue)));
19 naveen 774
			}
775
 
776
			// Enum and fixed
777
			else if(isEnum && !bulletDef.isLearned()) {
778
				long enumValueID = defs.getEnumValueID(datatypeDef.getID(), 
779
						bulletValue);
33 naveen 780
				Utils.logger.info("enumValueID=" + enumValueID);
19 naveen 781
 
34 naveen 782
				// Treat it to be free-form
19 naveen 783
				if(enumValueID == -1L) {
34 naveen 784
					Utils.logger.warning("Not one of the valid enum values, " +
785
							"treating it like free-form content - " + crumb + 
786
							" = " + bulletValue);
19 naveen 787
 
788
					continue;
789
				}
790
 
20 naveen 791
				EnumDataObject enumDataObject = new EnumDataObject(enumValueID);
792
				bullets.add(new Bullet(enumDataObject));
19 naveen 793
			}
794
 
795
			// Composite
796
			else if(isComposite) {
20 naveen 797
				CompositeDefinition compositeDef = 
798
					(CompositeDefinition)datatypeDef;
19 naveen 799
 
20 naveen 800
				String separator = compositeDef.getSeparator();
801
				String[] compositeParts = 
802
					StringUtils.split(bulletValue, separator);
803
 
804
				List<CompositePartDefinition> compositePartDefs = 
805
					compositeDef.getCompositePartDefinitions();
806
 
807
				// Validate number of parts
808
				if(compositeParts.length != compositePartDefs.size()) {
809
					Utils.logger.severe("Invalid value, " + crumb + " = " + 
810
							bulletValue);
811
 
812
					continue;
813
				}
814
 
815
				// Remove spurious whitespaces
816
				boolean validPart = true;
817
				for(int j=0;j<compositeParts.length;j++) {
818
					compositeParts[j] = StringUtils.strip(compositeParts[j]);
819
					Utils.logger.info("compositeParts[j]=" + compositeParts[j]);
820
 
821
					// Validate each part
822
					// Each part can be enum or composite in itself
823
					// We will stick to primitive for now
824
					long partDatatypeDefID = 
825
						compositePartDefs.get(j).getDatatypeDefinitionID();
826
 
827
					DatatypeDefinition partDatatypeDef = 
828
						defs.getDatatypeDefinition(partDatatypeDefID);
829
					Utils.logger.info("partDatatypeDef=" + partDatatypeDef);
830
 
831
					if(!this.validatePrimitive(compositeParts[j], 
832
							partDatatypeDef, crumb)) {
833
						validPart = false;
834
						break;
835
					}
836
				}
837
 
838
				if(!validPart) {
839
					continue;
840
				}
841
 
842
				CompositeDataObject compositeDataObject = 
843
					new CompositeDataObject();
844
 
845
				for(int j=0;j<compositeParts.length;j++) {
846
					compositeDataObject.addPrimitiveDataObject(
847
							new PrimitiveDataObject(compositeParts[j]));
848
				}
849
 
850
				bullets.add(new Bullet(compositeDataObject));
19 naveen 851
			}
852
		}
853
 
20 naveen 854
		Utils.logger.info("slideNameFeatureNameToBullets=" + 
855
				this.slideNameFeatureNameToBullets);
19 naveen 856
	}
857
 
18 naveen 858
	/**
19 naveen 859
	 * 
20 naveen 860
	 * @param bulletValue
861
	 * @param datatypeDef
862
	 * @param crumb
863
	 */
864
	private boolean validatePrimitive(String bulletValue, 
865
			DatatypeDefinition datatypeDef, String crumb) {
866
		String dt = datatypeDef.getName();
867
		boolean valid = true;
868
 
869
		// integer
870
		if(dt.equals("integer")) {
871
			try {
872
				Integer.parseInt(bulletValue);
873
			}
874
			catch(NumberFormatException nfe) {
875
				Utils.logger.severe("Invalid integer value, " + crumb +
876
						" = " + bulletValue);
877
				valid = false;
878
			}
879
		}
880
 
881
		// decimal
882
		else if(dt.endsWith("decimal")) {
883
			try {
884
				Float.parseFloat(bulletValue);
885
			}
886
			catch(NumberFormatException nfe) {
887
				Utils.logger.severe("Invalid decimal value, " + crumb +
888
						" = " + bulletValue);
889
				valid = false;
890
			}
891
		}
892
 
893
		return valid;
894
	}
895
 
896
	// string
897
	// any thing goes
898
	/**
899
	 * 
19 naveen 900
	 * @param allLines
901
	 * @param featureLabels
902
	 * @return
903
	 */
904
	private Map<String, int[]> getBulletLineRange(List<String> allLines, 
905
			Map<Integer, String> featureLabels) {
906
		Utils.logger.info("allLines=" + allLines);
907
		Utils.logger.info("featureLabels=" + featureLabels);
908
 
909
		// 5.1 Find line number range between which content for feature is found
910
		Map<String, int[]> bulletLineNumberRangePerFeature = 
911
			new HashMap<String, int[]>();
912
 
913
		Utils.logger.info("featureLabels.keySet=" + 
914
				featureLabels.keySet().toString());
915
 
916
		Integer[] featureLineNumbers = featureLabels.keySet().toArray(
917
				new Integer[0]);
918
 
919
		for(int i=0; i<featureLineNumbers.length-1; i++) {
920
			String tFeatureLabel = featureLabels.get(
921
					new Integer(featureLineNumbers[i]));
922
 
923
			Utils.logger.info("tFeatureLabel=" + tFeatureLabel);
924
 
925
			int from = featureLineNumbers[i]+1;
926
			int to = featureLineNumbers[i+1]-1;
927
			Utils.logger.info("from=" + from + " to=" + to);
928
 
929
			bulletLineNumberRangePerFeature.put(tFeatureLabel, 
930
					new int[] {from, to});
931
		}
932
 
933
		// For last feature
934
		String lastFeatureLabel = featureLabels.get(
935
			new Integer(featureLineNumbers[featureLineNumbers.length-1]));
936
		Utils.logger.info("lastFeatureLabel=" + lastFeatureLabel);
937
 
938
		int lastFrom = featureLineNumbers[featureLineNumbers.length-1] + 1;
939
		Utils.logger.info("lastFrom=" + lastFrom);
940
 
941
		// First empty line from "lastFrom"
942
		int lastTo = lastFrom;
943
		while(true) {
944
			if(lastTo == (allLines.size() - 1)) {
945
				break;
946
			}
947
 
948
			String tLastLine = allLines.get(lastTo + 1);
949
			Utils.logger.info("tLastLine=" + tLastLine);
950
			if(tLastLine == null) {
951
				break;
952
			}
953
 
954
			if(!StringUtils.startsWith(tLastLine, CN.PATTERN_LEVEL_2)) {
955
				break;
956
			}
957
			lastTo++;
958
			Utils.logger.info("running lastTo=" + lastTo);
959
		}
960
 
961
		Utils.logger.info("lastFrom=" + lastFrom + " lastTo=" + lastTo);
962
 
963
		bulletLineNumberRangePerFeature.put(lastFeatureLabel, 
964
				new int[] {lastFrom, lastTo});
965
 
966
		Utils.logger.info("featureLineNumberRangePerSlide=" + 
967
				bulletLineNumberRangePerFeature.toString());		
968
 
969
		return bulletLineNumberRangePerFeature;
970
	}
971
 
972
 
973
	/**
974
	 * 
975
	 * @param allLines
976
	 * @param slideDef
977
	 * @param lineNumberRange
978
	 * @throws Exception
979
	 */
980
	private Map<Integer, String> processFeature(List<String> allLines, 
981
			SlideDefinition slideDef, 
982
			int[] lineNumberRange) throws Exception {
983
 
984
		int from = lineNumberRange[0];
985
		int to = lineNumberRange[1];
986
 
987
		Utils.logger.info("from=" + from + " to=" + to);
988
 
989
		int numberOfLines = (to - from);
990
		Utils.logger.info("numberOfLines=" + numberOfLines);
991
 
992
		Map<Integer, String> candidateFeatures = new TreeMap<Integer, String>();
993
		for(int i=from; i<(to+1); i++) {
994
			String line = allLines.get(i);
995
 
996
			if(StringUtils.startsWith(line, CN.PATTERN_LEVEL_1)) {
997
				//Utils.logger.info(line);
998
				//Utils.logger.info("Level 1");
999
				candidateFeatures.put(new Integer(i), 
1000
						StringUtils.strip(StringUtils.strip(line, 
1001
								CN.PATTERN_LEVEL_1)));
1002
			}
1003
		}
1004
 
1005
		Utils.logger.info("candidateFeatures=" + candidateFeatures);
1006
 
1007
		// Get all feature definitions for the slide
1008
		DefinitionsContainer defs = 
1009
			Catalog.getInstance().getDefinitionsContainer();
1010
 
32 naveen 1011
		Utils.logger.info("slideDef.getID=" + slideDef.getID());
1012
 
19 naveen 1013
		List<FeatureDefinition> featureDefs = 
1014
			defs.getFeatureDefinitions(slideDef.getID());
1015
		Utils.logger.info("featureDefs=" + featureDefs);
1016
 
1017
		List<String> validFeatureLabels = new ArrayList<String>();
1018
		for(int i=0; i<featureDefs.size(); i++) {
28 naveen 1019
			validFeatureLabels.add(
1020
					StringUtils.lowerCase(featureDefs.get(i).getLabel()));
19 naveen 1021
		}
1022
		Utils.logger.info("validFeatureLabels=" + validFeatureLabels);
1023
 
1024
		// Filter valid features
1025
		Map<Integer, String> featureLabels = new TreeMap<Integer, String>();
1026
 
1027
		// Discard if feature is not one of the valids
1028
		// RE-VISIT
1029
		for(Integer lineNumber : candidateFeatures.keySet()) {
28 naveen 1030
			if(validFeatureLabels.contains(StringUtils.lowerCase(
1031
					candidateFeatures.get(lineNumber)))) {
1032
				featureLabels.put(lineNumber, StringUtils.lowerCase(
1033
						candidateFeatures.get(lineNumber)));
19 naveen 1034
			}
1035
		}
1036
		Utils.logger.info("featureLabels=" + featureLabels);
1037
 
1038
		// Fetch all mandatory features
1039
		List<FeatureDefinition> mandatoryFeatureDefs = 
1040
			defs.getFeatureDefinitions(slideDef.getID(),
1041
			EditorialImportance.MANDATORY);
1042
 
1043
		Utils.logger.info("mandatoryFeatureDefs=" + mandatoryFeatureDefs);
1044
 
1045
		// Severe error if mandatory features are not included
1046
		for(FeatureDefinition featureDef : mandatoryFeatureDefs) {
28 naveen 1047
			if(!featureLabels.values().contains(StringUtils.lowerCase(
1048
					featureDef.getLabel()))) {
1049
				Utils.logger.severe("Mandatory feature \"" + slideDef.getLabel() 
1050
						+ " > " + featureDef.getLabel() + "\" is missing");
19 naveen 1051
			}
1052
		}
1053
 
1054
		// Fetch all recommended features
1055
		List<FeatureDefinition> recommendedFeatureDefs = 
1056
			defs.getFeatureDefinitions(slideDef.getID(),
1057
			EditorialImportance.RECOMMENDED);
1058
 
1059
		Utils.logger.info("recommendedFeatureDefs=" + recommendedFeatureDefs);
1060
 
1061
		// Warn if recommended features are not included
1062
		for(FeatureDefinition featureDef : recommendedFeatureDefs) {
28 naveen 1063
			if(!featureLabels.values().contains(StringUtils.lowerCase(
1064
					featureDef.getLabel()))) {
19 naveen 1065
				Utils.logger.warning("Recommended feature \"" + 
1066
						featureDef.getLabel() + "\" is missing");
1067
			}
1068
		}
1069
 
1070
		return featureLabels;
1071
	}
1072
 
1073
	/**
18 naveen 1074
	 * Filter and validate parent slide names
1075
	 * 
1076
	 * @param candidateSlideNames
1077
	 * @return List<LineInfo>
1078
	 * @throws Exception 
1079
	 */
1080
	private Map<Integer, String> processSlides(
1081
			Map<Integer, String> candidateSlideLines) throws Exception {
1082
 
19 naveen 1083
		// Order of keys need to be restored
18 naveen 1084
		Map<Integer, String> processedSlideLines = 
19 naveen 1085
			new TreeMap<Integer, String>();
18 naveen 1086
 
1087
		// 1 Retrieve meta-data
1088
		// 1.1 Retrieve all valid slide names in the content model
1089
		DefinitionsContainer defs = 
1090
			Catalog.getInstance().getDefinitionsContainer();
1091
 
32 naveen 1092
		List<SlideDefinition> slideDefs = defs.getSlideDefinitions(
1093
				this.categoryID);
18 naveen 1094
 
1095
		List<String> validSlideNames = new ArrayList<String>();
32 naveen 1096
		for (SlideDefinition slideDef : slideDefs) {
18 naveen 1097
 
1098
			// To avoid Introduction slide
1099
			if(!slideDef.getLabel().isEmpty()) {
28 naveen 1100
				validSlideNames.add(StringUtils.lowerCase(slideDef.getLabel()));
18 naveen 1101
			}
1102
		}
1103
		Utils.logger.info("validSlideNames=" + validSlideNames.toString());
1104
 
1105
		// 2 Rules
1106
		// 2.1 Discard if slide is not one of the valids
1107
		// REVISIT
1108
		for(Integer lineNumber : candidateSlideLines.keySet()) {
28 naveen 1109
			if(validSlideNames.contains(StringUtils.lowerCase(
1110
					candidateSlideLines.get(lineNumber)))) {
1111
 
1112
				processedSlideLines.put(lineNumber, StringUtils.lowerCase(
1113
								candidateSlideLines.get(lineNumber)));
1114
 
1115
				this.slideNames.add(StringUtils.lowerCase(
1116
						candidateSlideLines.get(lineNumber)));
18 naveen 1117
			}
1118
		}
1119
 
1120
		Utils.logger.info("processedSlideLines=" + 
1121
				processedSlideLines.toString());
1122
 
1123
		// 3 Retrieve "Mandatory" slide names for the category
1124
		List<SlideDefinition> mandatorySlideDefs = 
1125
			defs.getSlides(this.categoryID, EditorialImportance.MANDATORY);
1126
 
1127
		Utils.logger.info("mandatorySlideDefs=" + 
1128
				mandatorySlideDefs.toString());
1129
 
1130
		// 3.1 All mandatory slides exist - Severe
1131
		for(SlideDefinition mandatorySlideDef : mandatorySlideDefs) {
1132
 
1133
			// Avoid introduction slide
1134
			if(mandatorySlideDef.getLabel().isEmpty()) {
1135
				continue;
1136
			}
1137
 
1138
			if(!processedSlideLines.values().contains(
28 naveen 1139
					StringUtils.lowerCase(mandatorySlideDef.getLabel()))) {
18 naveen 1140
				Utils.logger.severe("Mandatory slide \"" + 
1141
						mandatorySlideDef.getLabel() + "\" is missing");
1142
			}
1143
		}
1144
 
1145
		// 4 Retrieve "Recommended" slide names for the category
1146
		List<SlideDefinition> recommendedSlideDefs = 
1147
			defs.getSlides(this.categoryID, EditorialImportance.RECOMMENDED);
1148
 
1149
		Utils.logger.info("recommendedSlideDefs=" + 
1150
				recommendedSlideDefs.toString());
1151
 
1152
		// 4.1 All recommended slides exist - Warn	
1153
		for(SlideDefinition recommendedSlideDef : recommendedSlideDefs) {
1154
			if(!processedSlideLines.values().contains(
28 naveen 1155
					StringUtils.lowerCase(recommendedSlideDef.getLabel()))) {
18 naveen 1156
				Utils.logger.warning("Recommended slide \"" + 
1157
						recommendedSlideDef.getLabel() + "\" is missing");
1158
			}
1159
		}
1160
 
1161
		return processedSlideLines;
1162
	}
1163
 
1164
	/**
1165
	 * @throws Exception 
1166
	 * 
1167
	 */
1168
	@SuppressWarnings("unused")
1169
	private void test() throws Exception {
1170
		// test category
1171
		DefinitionsContainer allDefs = 
1172
			Catalog.getInstance().getDefinitionsContainer();
1173
 
1174
		Category category = allDefs.getCategory(this.categoryID);
1175
		Utils.logger.info(category.toString());
1176
 
1177
		// test data type defs
1178
		DatatypeDefinition datatypeDef = allDefs.getDatatypeDefinition(70004L);
1179
		Utils.logger.info(datatypeDef.toString());
1180
 
1181
		// test units
1182
		Unit unit = allDefs.getUnit(50004L);
1183
		Utils.logger.info(unit.toString());
1184
 
1185
		// test slide defs
1186
		SlideDefinition slideDef = allDefs.getSlideDefinition(130002L);
1187
		Utils.logger.info(slideDef.toString());
1188
 
1189
		// test feature defs
1190
		FeatureDefinition featureDef = allDefs.getFeatureDefinition(120002L);
1191
		Utils.logger.info(featureDef.toString());
1192
	}
17 naveen 1193
}