Subversion Repositories SmartDukaan

Rev

Rev 2949 | Rev 3246 | 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){
2949 chandransh 49 
		    log.error("Unable to get the solr URL from the config server. Setting the default value.", cex);
1698 chandransh 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 
 
209 
		for(int i=0; i<nodes.getLength(); i++) {
210 
			Node node = nodes.item(i);
2946 chandransh 211 
			String facetName = node.getAttributes().getNamedItem("name").getNodeValue();
354 rajveer 212 
			subNodes = node.getChildNodes();
213 
			HashMap<String,Integer> facetValueCountMap = new HashMap<String,Integer>();
214 
			for(int j=0; j<subNodes.getLength(); j++) {
215 
				Node subNode = subNodes.item(j);
216 
				if(Integer.parseInt(subNode.getTextContent())==0)
217 
						continue;
218 
				facetValueCountMap.put(subNode.getAttributes().getNamedItem("name").getNodeValue(), Integer.parseInt(subNode.getTextContent()));
219 
			}
220 
			facetMap.put(facetName, facetValueCountMap);
221 
			}
545 rajveer 222 
		this.numberOfResults  = this.getTotalResults();
517 rajveer 223 
 
354 rajveer 224 
		facetMap = removeUnwantedFacets(facetMap, numberOfResults);
225 
		return facetMap;
226 
		}
227 
 
790 vikas 228 
	public List<String> getResultMap() {
229 
		resultMap = new LinkedList<String>();
354 rajveer 230 
 
231 
		String resultDocsPath = "/response/result/doc";
232 
 
233 
 
234 
		NodeList nodes = null;
235 
		try {
236 
			nodes = (NodeList) this.xpath.evaluate(resultDocsPath, this.inputSource, XPathConstants.NODESET);
237 
		}
238 
		catch (XPathExpressionException xpee) {
239 
			return null;
240 
		}
241 
 
242 
		if(nodes.getLength() == 0) {
243 
			return null;
244 
		}
245 
 
246 
		for(int i=0; i<nodes.getLength(); i++) {
247 
			Node node = nodes.item(i);
248 
			String docID = node.getFirstChild().getTextContent();
790 vikas 249 
			resultMap.add(docID);
354 rajveer 250 
 		}
251 
		return resultMap;
252 
	}
253 
 
254 
	public HashMap<String, Double> getPriceStatsMap() {
255 
		HashMap<String, Double> priceStatsMap = new HashMap<String, Double>();
256 
 
257 
		String resultDocsPath = "/response/lst[@name = 'stats']/lst[@name = 'stats_fields']/lst[@name = 'F_50002']";
258 
 
259 
 
260 
		NodeList nodes = null;
261 
		try {
262 
			nodes = (NodeList) this.xpath.evaluate(resultDocsPath, this.inputSource, XPathConstants.NODESET);
263 
		}
264 
		catch (XPathExpressionException xpee) {
265 
			return null;
266 
		}
267 
 
268 
		if(nodes.getLength() == 0) {
269 
			return null;
270 
		}
271 
 
272 
		NodeList subNodes = nodes.item(0).getChildNodes();
273 
 
274 
		for(int i=0; i<subNodes.getLength(); i++) {
275 
			Node node = subNodes.item(i);
276 
 
277 
			String parameter = node.getAttributes().getNamedItem("name").getNodeValue();
278 
			String value = node.getTextContent();
279 
			priceStatsMap.put(parameter, Double.parseDouble(value));
280 
 		}
281 
		return priceStatsMap;
282 
	}
283 
 
284 
	public HashMap<String,Integer> getRangeQueryResultMap() {
285 
		HashMap<String, Integer> rangeQueryResultMap = new HashMap<String,Integer>();
286 
 
287 
		String resultDocsPath = "/response/lst[@name = 'facet_counts']/lst[@name = 'facet_queries']/int";
288 
 
289 
 
290 
		NodeList nodes = null;
291 
		try {
292 
			nodes = (NodeList) this.xpath.evaluate(resultDocsPath, this.inputSource, XPathConstants.NODESET);
293 
		}
294 
		catch (XPathExpressionException xpee) {
295 
			return null;
296 
		}
297 
 
298 
		if(nodes.getLength() == 0) {
299 
			return null;
300 
		}
301 
 
302 
 
303 
		for(int i=0; i<nodes.getLength(); i++) {
304 
			Node node = nodes.item(i);
305 
 
306 
			String query = node.getAttributes().getNamedItem("name").getNodeValue();
307 
			String docCount = node.getTextContent();
308 
 
309 
			rangeQueryResultMap.put(query,Integer.parseInt(docCount));
310 
 		}
311 
		return rangeQueryResultMap;
312 
 
313 
	}
314 
 
545 rajveer 315 
	/**
316 
	 *
317 
	 */
318 
	public long getTotalResults(){
319 
		String resultDocsPath = "/response/result";
320 
		NodeList nodes = null;
321 
		try {
322 
			nodes = (NodeList) this.xpath.evaluate(resultDocsPath, this.inputSource, XPathConstants.NODESET);
323 
		}
324 
		catch (XPathExpressionException xpee) {
325 
			return 0;
326 
		}
327 
 
328 
		Node node = nodes.item(0);
329 
 
330 
		return Long.parseLong(node.getAttributes().getNamedItem("numFound").getNodeValue());
331 
 
332 
	}
354 rajveer 333 
		/**
317 ashish 334 
	 *
335 
	 * @return
336 
	 */
337 
	public long[] getResultEntityIDs() {
338 
		String expression = "/response/result/doc/long";
339 
 
340 
		NodeList nodes = null;
341 
		try {
342 
			nodes = (NodeList) this.xpath.evaluate(expression, this.inputSource,
343 
					XPathConstants.NODESET);
344 
		}
345 
		catch(XPathExpressionException xpee) {
346 
			return null;
347 
		}
348 
 
349 
		if(nodes.getLength() == 0) {
350 
			return null;
351 
		}
352 
 
353 
		long[] values = new long[nodes.getLength()];
354 
		for(int i=0; i<nodes.getLength(); i++) {
355 
			Node node = nodes.item(i);
356 
			String value = node.getTextContent();
357 
			values[i] = Long.parseLong(value);
358 
 		}
359 
 
360 
		return values;
361 
	}
362 
 
363 
	/**
364 
	 *
365 
	 * @return
366 
	 */
367 
	public String[] getResultCategoryNames() {
368 
		String expression = "/response/lst/lst[@name = 'facet_fields']/";
369 
		expression += "lst[@name = 'Category']/int/@name";
370 
 
371 
		NodeList nodes = null;
372 
		try {
373 
			nodes = (NodeList) this.xpath.evaluate(expression,
374 
				this.inputSource, XPathConstants.NODESET);
375 
		}
376 
		catch (XPathExpressionException xpee) {
377 
			return null;
378 
		}
379 
 
380 
		if(nodes.getLength() == 0) {
381 
			return null;
382 
		}
383 
 
384 
		String[] values = new String[nodes.getLength()];
385 
		for(int i=0; i<nodes.getLength(); i++) {
386 
			Node node = nodes.item(i);
387 
			values[i] = node.getTextContent();
388 
 		}
389 
 
390 
		return values;
391 
	}
392 
 
393 
	/**
394 
	 *
395 
	 * @return
396 
	 */
397 
	public int[] getResultCategoryCounts() {
398 
		String expression = "/response/lst/lst[@name = 'facet_fields']/";
399 
		expression += "lst[@name = 'Category']/int";
400 
 
401 
		NodeList nodes = null;
402 
		try {
403 
			nodes = (NodeList) this.xpath.evaluate(expression,
404 
				this.inputSource, XPathConstants.NODESET);
405 
		}
406 
		catch (XPathExpressionException xpee) {
407 
			return null;
408 
		}
409 
 
410 
		if(nodes.getLength() == 0) {
411 
			return null;
412 
		}
413 
 
414 
		int[] values = new int[nodes.getLength()];
415 
		for(int i=0; i<nodes.getLength(); i++) {
416 
			Node node = nodes.item(i);
417 
			values[i] = Integer.parseInt(node.getTextContent());
418 
 		}
419 
 
420 
		return values;
421 
	}
422 
 
423 
	/**
424 
	 *
425 
	 * @return
426 
	 */
427 
	public String[]  getResultEntityNames() {
428 
		String expression = "/response/result/doc/str";
429 
 
430 
		NodeList nodes = null;
431 
		try {
432 
			nodes = (NodeList) this.xpath.evaluate(expression, this.inputSource,
433 
					XPathConstants.NODESET);
434 
		}
435 
		catch(XPathExpressionException xpee) {
436 
			return null;
437 
		}
438 
 
439 
		if(nodes.getLength() == 0) {
440 
			return null;
441 
		}
442 
 
443 
		String[] values = new String[nodes.getLength()];
444 
		for(int i=0; i<nodes.getLength(); i++) {
445 
			Node node = nodes.item(i);
446 
			String value = node.getTextContent();
447 
			values[i] = value;
448 
 		}
449 
 
450 
		return values;
451 
	}
452 
 
453 
	/**
454 
	 *
455 
	 * @param facetDefinitionID
456 
	 * @return
457 
	 */
354 rajveer 458 
	public String[] getFacetValues(String facetDefinitionID) {
317 ashish 459 
		String expression = "/response/lst/lst[@name = 'facet_fields']/";
354 rajveer 460 
		expression += "lst[@name = '"+ facetDefinitionID +"']/int/@name";
317 ashish 461 
 
462 
		NodeList nodes = null;
463 
		try {
464 
			nodes = (NodeList) this.xpath.evaluate(expression,
465 
				this.inputSource, XPathConstants.NODESET);
466 
		}
467 
		catch (XPathExpressionException xpee) {
468 
			return null;
469 
		}
470 
 
471 
		if(nodes.getLength() == 0) {
472 
			return null;
473 
		}
474 
 
475 
		String[] values = new String[nodes.getLength()];
476 
		for(int i=0; i<nodes.getLength(); i++) {
477 
			Node node = nodes.item(i);
478 
			values[i] = node.getTextContent();
545 rajveer 479 
		}
317 ashish 480 
 
481 
		return values;
482 
	}
483 
 
484 
	/**
485 
	 *
486 
	 * @param facetDefinitionID
487 
	 * @return
488 
	 */
354 rajveer 489 
	public String[] getFacetCounts(String facetDefinitionID) {
317 ashish 490 
		String expression = "/response/lst/lst[@name = 'facet_fields']/";
354 rajveer 491 
		expression += "lst[@name = '" + facetDefinitionID + "']/int";
317 ashish 492 
 
493 
		NodeList nodes = null;
494 
		try {
495 
			nodes = (NodeList) this.xpath.evaluate(expression,
496 
				this.inputSource, XPathConstants.NODESET);
497 
		}
498 
		catch (XPathExpressionException xpee) {
499 
			return null;
500 
		}
501 
 
502 
		if(nodes.getLength() == 0) {
503 
			return null;
504 
		}
505 
 
506 
		String[] values = new String[nodes.getLength()];
507 
		for(int i=0; i<nodes.getLength(); i++) {
508 
			Node node = nodes.item(i);
509 
			values[i] = node.getTextContent();
354 rajveer 510 
//			log.info("Facets Counts  " + values[i]);
317 ashish 511 
 		}
512 
 
513 
		return values;
514 
	}
545 rajveer 515 
 
516 
	public static void main(String[] args){
517 
		/*
518 
    	// Hard coded for now
519 
    	String[] facetDefIDs = new String[] {"F_50001", "F_50002", "F_50003", "F_50004", "F_50005", "F_50006", "F_50007", "F_50008", "F_50009"};
520 
 
521 
    	// Hard-coded for now
522 
    	String[] facetLabels = new String[] {
523 
	    	"Brand", "Price","Form Factor", "Carry In Pocket", "Cellular Technologies",
524 
	    	"Data Connectivity", "Camera Resolution", "Built-in Memory",
525 
	    	"Talk time"
526 
    	};
527 
 
528 
		 */
529 
    	String[] facetDefIDs = new String[] {"Category","F_50002","F_50001",  "F_50006", "F_50007" };
2147 chandransh 530 
    	//String[] facetLabels = new String[] {"Category","Price", "Brand", "Data Connectivity", "Camera Resolution"	};
545 rajveer 531 
 
532 
 
533 
    	String[] fqrys = {};
569 rajveer 534 
		SolrSearchService search = new SolrSearchService("nokia", fqrys, facetDefIDs, 0 , 20, null, null, 10000, null);
545 rajveer 535 
 
536 
    	long[] entityIDs = search.getResultEntityIDs();
537 
    	log.info("entityIDs=" + Arrays.toString(entityIDs));
538 
 
539 
    	String[] entityNames = search.getResultEntityNames();
540 
    	log.info("entityNames=" + Arrays.toString(entityNames));
541 
    	search.getFacetMap();
542 
 
543 
    	search.getResultMap();
544 
    	search.getRangeQueryResultMap();
545 
    	search.getPriceStatsMap();
546 
    	search.getTotalResults();
547 
       	for (int i=0; i<facetDefIDs.length; i++) {
548 
       		search.getFacetCounts(facetDefIDs[i]);
549 
       		search.getFacetValues(facetDefIDs[i]);
550 
       	}
551 
 
552 
	}
317 ashish 553 
}