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