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