Subversion Repositories SmartDukaan

Rev

Rev 2435 | Rev 2946 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
317 ashish 1 
/**
2 
 *
3 
 */
4 
package in.shop2020.serving.services;
5 
 
6 
 
1698 chandransh 7 
import in.shop2020.config.ConfigException;
8 
import in.shop2020.thrift.clients.config.ConfigClient;
2070 rajveer 9 
import in.shop2020.utils.CategoryManager;
1698 chandransh 10 
 
354 rajveer 11 
import java.util.Arrays;
12 
import java.util.HashMap;
790 vikas 13 
import java.util.LinkedList;
14 
import java.util.List;
354 rajveer 15 
import java.util.TreeMap;
16 
 
17 
 
317 ashish 18 
import javax.xml.xpath.XPath;
19 
import javax.xml.xpath.XPathConstants;
20 
import javax.xml.xpath.XPathExpressionException;
21 
import javax.xml.xpath.XPathFactory;
22 
 
832 rajveer 23 
import org.apache.log4j.Logger;
317 ashish 24 
import org.w3c.dom.Node;
25 
import org.w3c.dom.NodeList;
26 
import org.xml.sax.InputSource;
27 
 
354 rajveer 28 
 
317 ashish 29 
/**
545 rajveer 30 
 * @author rajveer
317 ashish 31 
 *
32 
 */
33 
public class SolrSearchService {
34 
	/**
35 
	 *
36 
	 */
2147 chandransh 37 
	private static Logger log = Logger.getLogger(Class.class);
545 rajveer 38 
 
39 
	/**
40 
	 *
41 
	 */
1698 chandransh 42 
	public static final String SOLR_URL;
317 ashish 43 
 
1698 chandransh 44 
	static {
45 
		String solr_url = null;
46 
		try {
47 
			solr_url = ConfigClient.getClient().get("solr_url");
48 
		}catch(ConfigException cex){
49 
			cex.printStackTrace();
50 
			solr_url = "http://localhost:8983/solr/select/";
51 
		}
52 
		SOLR_URL = solr_url;
53 
	}
54 
 
317 ashish 55 
	/**
56 
	 *
57 
	 */
58 
	private XPath xpath;
59 
 
60 
	/**
61 
	 *
62 
	 */
63 
	private InputSource inputSource;
64 
 
354 rajveer 65 
	TreeMap<String,HashMap<String,Integer>> facetMap;
66 
 
790 vikas 67 
	List<String> resultMap;
545 rajveer 68 
 
69 
	long numberOfResults=0;
317 ashish 70 
	/**
71 
	 *
72 
	 * @param query
73 
	 * @param facetDefinitionIDs
74 
	 */
569 rajveer 75 
	public SolrSearchService(String query, String[] facetqueries, String[] facetDefinitionIDs, long start, long rows,  Double minPrice, Double maxPrice, long categoryId, String sortOrder) {
354 rajveer 76 
 
317 ashish 77 
		this.xpath = XPathFactory.newInstance().newXPath();
545 rajveer 78 
 
79 
    	// if multiple words are given, it should do AND for all of them
2606 rajveer 80 
		query = query.trim().replaceAll("\\s+", " ");
545 rajveer 81 
    	if(query.contains(" ")){
82 
    		String[] tokens = query.split("\\s+");
83 
    		query = tokens[0];
84 
    		for(int i = 1; i < tokens.length; i++){
85 
    			query = query + " AND " + tokens[i];
86 
    		}
87 
    	}
88 
    	log.info("query=" + query);
89 
 
317 ashish 90 
		String uri = SOLR_URL + "?wt=xml&q=" + query;
91 
 
545 rajveer 92 
		uri += "&stats=on&stats.field=F_50002";
93 
 
94 
 
95 
		if(minPrice != null || maxPrice != null){
96 
			String minString = "0";
97 
			String maxString = "*";
98 
			if(minPrice != null){
99 
				minString = minPrice.toString();
100 
			}
101 
			if(maxPrice != null){
102 
				maxString = maxPrice.toString();
103 
			}
104 
			uri += "&fq=F_50002:["+  minString + " " + maxString + "]";
354 rajveer 105 
		}
106 
 
569 rajveer 107 
		if(categoryId != 10000){
108 
			uri += "&fq=F_50010:\"" + CategoryManager.getCategoryManager().getCategoryLabel(categoryId) + "\"";
109 
		}
354 rajveer 110 
 
569 rajveer 111 
		if(sortOrder != null){
112 
			uri += "&sort=" + sortOrder;
113 
		}
545 rajveer 114 
 
317 ashish 115 
		if(facetqueries != null) {
116 
			for(int i=0; i<facetqueries.length; i++) {
536 rajveer 117 
				if(facetqueries[i].contains(" ") && !facetqueries[i].contains("F_50002")){
118 
					String[] tokens = facetqueries[i].split(":");
545 rajveer 119 
					uri += "&fq=" + tokens[0] + ":\"" + tokens[1] + "\"";
536 rajveer 120 
				}else{
121 
					uri += "&fq=" + facetqueries[i] + "";
122 
				}
317 ashish 123 
			}
124 
		}
545 rajveer 125 
		uri += "&fl=ID,Name&facet=true&start=" + start + "&rows=" + rows;
2435 rajveer 126 
		if(facetDefinitionIDs != null){
127 
			for(int i=0; i<facetDefinitionIDs.length; i++) {
128 
				uri += "&facet.field=" + facetDefinitionIDs[i];
129 
			}
317 ashish 130 
		}
2435 rajveer 131 
		log.debug("uri=" + uri);
317 ashish 132 
 
133 
		this.inputSource = new InputSource(uri);
517 rajveer 134 
 
545 rajveer 135 
		this.facetMap = getFacetMap();
354 rajveer 136 
	}
137 
 
545 rajveer 138 
	public TreeMap<String,HashMap<String,Integer>> removeUnwantedFacets(TreeMap<String,HashMap<String,Integer>> facetMap, long numberOfResults){
139 
		TreeMap<String,HashMap<String,Integer>> tempFacets = new TreeMap<String, HashMap<String,Integer>>();
354 rajveer 140 
		for(String facet : facetMap.keySet()){
545 rajveer 141 
			if(facetMap.get(facet).size() > 1){
142 
				HashMap<String,Integer> tempMap = new HashMap<String, Integer>();
143 
 
354 rajveer 144 
				for(String facetValueName : facetMap.get(facet).keySet()){
545 rajveer 145 
					if(facetMap.get(facet).get(facetValueName) != 0 && facetMap.get(facet).get(facetValueName) != numberOfResults){
146 
						tempMap.put(facetValueName, facetMap.get(facet).get(facetValueName));
354 rajveer 147 
					}
148 
				}
545 rajveer 149 
				if(!tempMap.isEmpty()){
150 
					tempFacets.put(facet, tempMap);
354 rajveer 151 
				}
545 rajveer 152 
			}
354 rajveer 153 
		}
550 rajveer 154 
		if(tempFacets.containsKey("F_50010")){
155 
			tempFacets.remove("F_50011");
156 
		}
354 rajveer 157 
 
158 
		return tempFacets;
159 
	}
160 
 
545 rajveer 161 
/*
162 
	int sumOfFacetCounts = 0;
163 
	for(String facetValueName : facetMap.get(facet).keySet()){
164 
		if(facetMap.get(facet).get(facetValueName) == 0 || facetMap.get(facet).get(facetValueName) == numberOfResults){
165 
			tempFacets.get(facet).remove(facetValueName);
166 
			// FIXME
167 
			/
168 
			HashMap<String, Integer> tmp = tempFacets.get(facet);
169 
			tmp.remove(facetValueName);
170 
			tempFacets.put(facet, tmp);
171 
			/
172 
		}else{
173 
			sumOfFacetCounts += facetMap.get(facet).get(facetValueName);
174 
		}
175 
	}
176 
	if(sumOfFacetCounts < numberOfResults*PERCENTAGE_OF_TOTAL_RESULTS/100){
177 
		tempFacets.remove(facet);
178 
	}
179 
}
180 
 
181 
*/
354 rajveer 182 
	public HashMap<String,Integer> getFacetDetails(String facetName){
2606 rajveer 183 
		if(facetMap != null){
184 
			return facetMap.get(facetName);
185 
		}else{
186 
			return null;
187 
		}
354 rajveer 188 
	}
189 
 
190 
	public TreeMap<String,HashMap<String,Integer>> getFacetMap() {
191 
		facetMap = new TreeMap<String,HashMap<String,Integer>>();
192 
 
193 
		String facetNamePath = "/response/lst/lst[@name = 'facet_fields']/lst";
545 rajveer 194 
 
354 rajveer 195 
		NodeList nodes = null;
196 
		try {
197 
			nodes = (NodeList) this.xpath.evaluate(facetNamePath, this.inputSource, XPathConstants.NODESET);
198 
		}
199 
		catch (XPathExpressionException xpee) {
200 
			return null;
201 
		}
202 
 
203 
		if(nodes.getLength() == 0) {
204 
			return null;
205 
		}
206 
 
207 
		NodeList subNodes = null;
208 
		String facetName = new String();
209 
 
210 
		for(int i=0; i<nodes.getLength(); i++) {
211 
			Node node = nodes.item(i);
212 
			facetName = node.getAttributes().getNamedItem("name").getNodeValue();
213 
			subNodes = node.getChildNodes();
214 
			HashMap<String,Integer> facetValueCountMap = new HashMap<String,Integer>();
215 
			for(int j=0; j<subNodes.getLength(); j++) {
216 
				Node subNode = subNodes.item(j);
217 
				if(Integer.parseInt(subNode.getTextContent())==0)
218 
						continue;
219 
				facetValueCountMap.put(subNode.getAttributes().getNamedItem("name").getNodeValue(), Integer.parseInt(subNode.getTextContent()));
220 
			}
221 
			facetMap.put(facetName, facetValueCountMap);
222 
			}
223 
		System.out.println(facetMap);
517 rajveer 224 
 
545 rajveer 225 
		this.numberOfResults  = this.getTotalResults();
517 rajveer 226 
 
354 rajveer 227 
		facetMap = removeUnwantedFacets(facetMap, numberOfResults);
228 
		System.out.println(facetMap);
229 
		return facetMap;
230 
		}
231 
 
790 vikas 232 
	public List<String> getResultMap() {
233 
		resultMap = new LinkedList<String>();
354 rajveer 234 
 
235 
		String resultDocsPath = "/response/result/doc";
236 
 
237 
 
238 
		NodeList nodes = null;
239 
		try {
240 
			nodes = (NodeList) this.xpath.evaluate(resultDocsPath, this.inputSource, XPathConstants.NODESET);
241 
		}
242 
		catch (XPathExpressionException xpee) {
243 
			return null;
244 
		}
245 
 
246 
		if(nodes.getLength() == 0) {
247 
			return null;
248 
		}
249 
 
250 
		for(int i=0; i<nodes.getLength(); i++) {
251 
			Node node = nodes.item(i);
252 
			String docID = node.getFirstChild().getTextContent();
790 vikas 253 
			resultMap.add(docID);
354 rajveer 254 
 		}
255 
		System.out.println("resultMap is " + resultMap);
256 
		return resultMap;
257 
	}
258 
 
259 
	public HashMap<String, Double> getPriceStatsMap() {
260 
		HashMap<String, Double> priceStatsMap = new HashMap<String, Double>();
261 
 
262 
		String resultDocsPath = "/response/lst[@name = 'stats']/lst[@name = 'stats_fields']/lst[@name = 'F_50002']";
263 
 
264 
 
265 
		NodeList nodes = null;
266 
		try {
267 
			nodes = (NodeList) this.xpath.evaluate(resultDocsPath, this.inputSource, XPathConstants.NODESET);
268 
		}
269 
		catch (XPathExpressionException xpee) {
270 
			return null;
271 
		}
272 
 
273 
		if(nodes.getLength() == 0) {
274 
			return null;
275 
		}
276 
 
277 
		NodeList subNodes = nodes.item(0).getChildNodes();
278 
 
279 
		for(int i=0; i<subNodes.getLength(); i++) {
280 
			Node node = subNodes.item(i);
281 
 
282 
			String parameter = node.getAttributes().getNamedItem("name").getNodeValue();
283 
			String value = node.getTextContent();
284 
			priceStatsMap.put(parameter, Double.parseDouble(value));
285 
 		}
286 
		System.out.println("priceStatsMap is " + priceStatsMap);
287 
		return priceStatsMap;
288 
	}
289 
 
290 
	public HashMap<String,Integer> getRangeQueryResultMap() {
291 
		HashMap<String, Integer> rangeQueryResultMap = new HashMap<String,Integer>();
292 
 
293 
		String resultDocsPath = "/response/lst[@name = 'facet_counts']/lst[@name = 'facet_queries']/int";
294 
 
295 
 
296 
		NodeList nodes = null;
297 
		try {
298 
			nodes = (NodeList) this.xpath.evaluate(resultDocsPath, this.inputSource, XPathConstants.NODESET);
299 
		}
300 
		catch (XPathExpressionException xpee) {
301 
			return null;
302 
		}
303 
 
304 
		if(nodes.getLength() == 0) {
305 
			return null;
306 
		}
307 
 
308 
 
309 
		for(int i=0; i<nodes.getLength(); i++) {
310 
			Node node = nodes.item(i);
311 
 
312 
			String query = node.getAttributes().getNamedItem("name").getNodeValue();
313 
			String docCount = node.getTextContent();
314 
 
315 
			rangeQueryResultMap.put(query,Integer.parseInt(docCount));
316 
 		}
317 
		System.out.println("rangeQueryResultMap is " + rangeQueryResultMap);
318 
		return rangeQueryResultMap;
319 
 
320 
	}
321 
 
545 rajveer 322 
	/**
323 
	 *
324 
	 */
325 
	public long getTotalResults(){
326 
		String resultDocsPath = "/response/result";
327 
		NodeList nodes = null;
328 
		try {
329 
			nodes = (NodeList) this.xpath.evaluate(resultDocsPath, this.inputSource, XPathConstants.NODESET);
330 
		}
331 
		catch (XPathExpressionException xpee) {
332 
			return 0;
333 
		}
334 
 
335 
		Node node = nodes.item(0);
336 
 
337 
		return Long.parseLong(node.getAttributes().getNamedItem("numFound").getNodeValue());
338 
 
339 
	}
354 rajveer 340 
		/**
317 ashish 341 
	 *
342 
	 * @return
343 
	 */
344 
	public long[] getResultEntityIDs() {
345 
		String expression = "/response/result/doc/long";
346 
 
347 
		NodeList nodes = null;
348 
		try {
349 
			nodes = (NodeList) this.xpath.evaluate(expression, this.inputSource,
350 
					XPathConstants.NODESET);
351 
		}
352 
		catch(XPathExpressionException xpee) {
353 
			return null;
354 
		}
355 
 
356 
		if(nodes.getLength() == 0) {
357 
			return null;
358 
		}
359 
 
360 
		long[] values = new long[nodes.getLength()];
361 
		for(int i=0; i<nodes.getLength(); i++) {
362 
			Node node = nodes.item(i);
363 
			String value = node.getTextContent();
364 
			values[i] = Long.parseLong(value);
365 
 		}
366 
 
367 
		return values;
368 
	}
369 
 
370 
	/**
371 
	 *
372 
	 * @return
373 
	 */
374 
	public String[] getResultCategoryNames() {
375 
		String expression = "/response/lst/lst[@name = 'facet_fields']/";
376 
		expression += "lst[@name = 'Category']/int/@name";
377 
 
378 
		NodeList nodes = null;
379 
		try {
380 
			nodes = (NodeList) this.xpath.evaluate(expression,
381 
				this.inputSource, XPathConstants.NODESET);
382 
		}
383 
		catch (XPathExpressionException xpee) {
384 
			return null;
385 
		}
386 
 
387 
		if(nodes.getLength() == 0) {
388 
			return null;
389 
		}
390 
 
391 
		String[] values = new String[nodes.getLength()];
392 
		for(int i=0; i<nodes.getLength(); i++) {
393 
			Node node = nodes.item(i);
394 
			values[i] = node.getTextContent();
395 
 		}
396 
 
397 
		return values;
398 
	}
399 
 
400 
	/**
401 
	 *
402 
	 * @return
403 
	 */
404 
	public int[] getResultCategoryCounts() {
405 
		String expression = "/response/lst/lst[@name = 'facet_fields']/";
406 
		expression += "lst[@name = 'Category']/int";
407 
 
408 
		NodeList nodes = null;
409 
		try {
410 
			nodes = (NodeList) this.xpath.evaluate(expression,
411 
				this.inputSource, XPathConstants.NODESET);
412 
		}
413 
		catch (XPathExpressionException xpee) {
414 
			return null;
415 
		}
416 
 
417 
		if(nodes.getLength() == 0) {
418 
			return null;
419 
		}
420 
 
421 
		int[] values = new int[nodes.getLength()];
422 
		for(int i=0; i<nodes.getLength(); i++) {
423 
			Node node = nodes.item(i);
424 
			values[i] = Integer.parseInt(node.getTextContent());
425 
 		}
426 
 
427 
		return values;
428 
	}
429 
 
430 
	/**
431 
	 *
432 
	 * @return
433 
	 */
434 
	public String[]  getResultEntityNames() {
435 
		String expression = "/response/result/doc/str";
436 
 
437 
		NodeList nodes = null;
438 
		try {
439 
			nodes = (NodeList) this.xpath.evaluate(expression, this.inputSource,
440 
					XPathConstants.NODESET);
441 
		}
442 
		catch(XPathExpressionException xpee) {
443 
			return null;
444 
		}
445 
 
446 
		if(nodes.getLength() == 0) {
447 
			return null;
448 
		}
449 
 
450 
		String[] values = new String[nodes.getLength()];
451 
		for(int i=0; i<nodes.getLength(); i++) {
452 
			Node node = nodes.item(i);
453 
			String value = node.getTextContent();
454 
			values[i] = value;
455 
 		}
456 
 
457 
		return values;
458 
	}
459 
 
460 
	/**
461 
	 *
462 
	 * @param facetDefinitionID
463 
	 * @return
464 
	 */
354 rajveer 465 
	public String[] getFacetValues(String facetDefinitionID) {
317 ashish 466 
		String expression = "/response/lst/lst[@name = 'facet_fields']/";
354 rajveer 467 
		expression += "lst[@name = '"+ facetDefinitionID +"']/int/@name";
317 ashish 468 
 
469 
		NodeList nodes = null;
470 
		try {
471 
			nodes = (NodeList) this.xpath.evaluate(expression,
472 
				this.inputSource, XPathConstants.NODESET);
473 
		}
474 
		catch (XPathExpressionException xpee) {
475 
			return null;
476 
		}
477 
 
478 
		if(nodes.getLength() == 0) {
479 
			return null;
480 
		}
481 
 
482 
		String[] values = new String[nodes.getLength()];
483 
		for(int i=0; i<nodes.getLength(); i++) {
484 
			Node node = nodes.item(i);
485 
			values[i] = node.getTextContent();
545 rajveer 486 
		}
317 ashish 487 
 
488 
		return values;
489 
	}
490 
 
491 
	/**
492 
	 *
493 
	 * @param facetDefinitionID
494 
	 * @return
495 
	 */
354 rajveer 496 
	public String[] getFacetCounts(String facetDefinitionID) {
317 ashish 497 
		String expression = "/response/lst/lst[@name = 'facet_fields']/";
354 rajveer 498 
		expression += "lst[@name = '" + facetDefinitionID + "']/int";
317 ashish 499 
 
500 
		NodeList nodes = null;
501 
		try {
502 
			nodes = (NodeList) this.xpath.evaluate(expression,
503 
				this.inputSource, XPathConstants.NODESET);
504 
		}
505 
		catch (XPathExpressionException xpee) {
506 
			return null;
507 
		}
508 
 
509 
		if(nodes.getLength() == 0) {
510 
			return null;
511 
		}
512 
 
513 
		String[] values = new String[nodes.getLength()];
514 
		for(int i=0; i<nodes.getLength(); i++) {
515 
			Node node = nodes.item(i);
516 
			values[i] = node.getTextContent();
354 rajveer 517 
//			log.info("Facets Counts  " + values[i]);
317 ashish 518 
 		}
519 
 
520 
		return values;
521 
	}
545 rajveer 522 
 
523 
	public static void main(String[] args){
524 
		/*
525 
    	// Hard coded for now
526 
    	String[] facetDefIDs = new String[] {"F_50001", "F_50002", "F_50003", "F_50004", "F_50005", "F_50006", "F_50007", "F_50008", "F_50009"};
527 
 
528 
    	// Hard-coded for now
529 
    	String[] facetLabels = new String[] {
530 
	    	"Brand", "Price","Form Factor", "Carry In Pocket", "Cellular Technologies",
531 
	    	"Data Connectivity", "Camera Resolution", "Built-in Memory",
532 
	    	"Talk time"
533 
    	};
534 
 
535 
		 */
536 
    	String[] facetDefIDs = new String[] {"Category","F_50002","F_50001",  "F_50006", "F_50007" };
2147 chandransh 537 
    	//String[] facetLabels = new String[] {"Category","Price", "Brand", "Data Connectivity", "Camera Resolution"	};
545 rajveer 538 
 
539 
 
540 
    	String[] fqrys = {};
569 rajveer 541 
		SolrSearchService search = new SolrSearchService("nokia", fqrys, facetDefIDs, 0 , 20, null, null, 10000, null);
545 rajveer 542 
 
543 
    	long[] entityIDs = search.getResultEntityIDs();
544 
    	log.info("entityIDs=" + Arrays.toString(entityIDs));
545 
 
546 
    	String[] entityNames = search.getResultEntityNames();
547 
    	log.info("entityNames=" + Arrays.toString(entityNames));
548 
    	search.getFacetMap();
549 
 
550 
    	search.getResultMap();
551 
    	search.getRangeQueryResultMap();
552 
    	search.getPriceStatsMap();
553 
    	search.getTotalResults();
554 
       	for (int i=0; i<facetDefIDs.length; i++) {
555 
       		search.getFacetCounts(facetDefIDs[i]);
556 
       		search.getFacetValues(facetDefIDs[i]);
557 
       	}
558 
 
559 
	}
317 ashish 560 
}