| 9103 |
anupam.sin |
1 |
/**
|
|
|
2 |
*
|
|
|
3 |
*/
|
| 9269 |
amit.gupta |
4 |
package in.shop2020.mobileapi.serving.services;
|
| 9103 |
anupam.sin |
5 |
|
|
|
6 |
|
|
|
7 |
import in.shop2020.config.ConfigException;
|
| 9269 |
amit.gupta |
8 |
import in.shop2020.mobileapi.serving.controllers.SearchController;
|
|
|
9 |
import in.shop2020.mobileapi.serving.utils.Utils;
|
| 9103 |
anupam.sin |
10 |
import in.shop2020.thrift.clients.config.ConfigClient;
|
|
|
11 |
|
|
|
12 |
import java.io.UnsupportedEncodingException;
|
|
|
13 |
import java.net.URLEncoder;
|
|
|
14 |
import java.util.ArrayList;
|
|
|
15 |
import java.util.Arrays;
|
|
|
16 |
import java.util.Collections;
|
|
|
17 |
import java.util.HashMap;
|
|
|
18 |
import java.util.HashSet;
|
|
|
19 |
import java.util.Iterator;
|
|
|
20 |
import java.util.LinkedHashMap;
|
|
|
21 |
import java.util.LinkedList;
|
|
|
22 |
import java.util.List;
|
|
|
23 |
import java.util.Map;
|
|
|
24 |
import java.util.Set;
|
|
|
25 |
import java.util.TreeMap;
|
|
|
26 |
import java.util.regex.Matcher;
|
|
|
27 |
import java.util.regex.Pattern;
|
|
|
28 |
|
|
|
29 |
import javax.xml.xpath.XPath;
|
|
|
30 |
import javax.xml.xpath.XPathConstants;
|
|
|
31 |
import javax.xml.xpath.XPathExpressionException;
|
|
|
32 |
import javax.xml.xpath.XPathFactory;
|
|
|
33 |
|
|
|
34 |
import org.apache.log4j.Logger;
|
|
|
35 |
import org.w3c.dom.Node;
|
|
|
36 |
import org.w3c.dom.NodeList;
|
|
|
37 |
import org.xml.sax.InputSource;
|
|
|
38 |
|
|
|
39 |
|
|
|
40 |
/**
|
|
|
41 |
* @author rajveer
|
|
|
42 |
*
|
|
|
43 |
*/
|
|
|
44 |
public class SolrSearchService {
|
|
|
45 |
/**
|
|
|
46 |
*
|
|
|
47 |
*/
|
|
|
48 |
private static Logger log = Logger.getLogger(Class.class);
|
|
|
49 |
private HashMap<String, Double> dynamicPriceMap = null;
|
|
|
50 |
|
|
|
51 |
/**
|
|
|
52 |
*
|
|
|
53 |
*/
|
|
|
54 |
public static final String SOLR_URL;
|
|
|
55 |
|
|
|
56 |
private static final Pattern FACET_PATTERN = Pattern.compile("(?=(F_\\d{5}))");
|
|
|
57 |
|
|
|
58 |
private static final Map<String, List<String>> SORTED_FACET_VALUE_MAP = Collections.unmodifiableMap(
|
|
|
59 |
new HashMap<String, List<String>>(){
|
|
|
60 |
/**
|
|
|
61 |
*
|
|
|
62 |
*/
|
|
|
63 |
private static final long serialVersionUID = 1L;
|
|
|
64 |
{
|
| 9406 |
amit.gupta |
65 |
put("F_50007", Arrays.asList("Upto 2 Mpx", "2 - 5 Mpx", "5 - 10 Mpx", "Above 10 Mpx"));
|
|
|
66 |
put("F_50024", Arrays.asList("Upto 10 Mpx", "10 - 12 Mpx", "12 - 15 Mpx", "15 - 18 Mpx", "Above 18 Mpx"));
|
|
|
67 |
put("F_50025", Arrays.asList("Upto 4x", "4 - 6x", "6 - 10x", "10 - 14x", "14 - 18x", "Above 18x"));
|
|
|
68 |
put("F_50026", Arrays.asList("Below 2 in.", "2 to 2.9 in.", "3 to 3.9 in.", "Above 4 in."));
|
|
|
69 |
put("F_50032", Arrays.asList("Below 3 in.", "3 to 3.9 in.", "4 to 4.9 in.", "Above 5 in."));
|
|
|
70 |
put("F_50027", Arrays.asList("Upto 10 Mpx", "10 - 15 Mpx", "15 - 20 Mpx", "Above 20 Mpx"));
|
| 9103 |
anupam.sin |
71 |
}
|
|
|
72 |
});
|
|
|
73 |
public static final Map<String, List<String>> CATEGORY_FACET_MAP = Collections.unmodifiableMap(
|
|
|
74 |
new HashMap<String, List<String>>(){
|
|
|
75 |
/**
|
|
|
76 |
*
|
|
|
77 |
*/
|
|
|
78 |
private static final long serialVersionUID = 1L;
|
|
|
79 |
//Data Connectivity, Camera Resolution,Operating System,Screen Size
|
| 9406 |
amit.gupta |
80 |
List<String> mobileFacets = Arrays.asList("F_50038", "F_50006", "F_50007", "F_50031", "F_50032");
|
| 9103 |
anupam.sin |
81 |
//Camera Resolution,Operating System,Screen Size
|
| 9406 |
amit.gupta |
82 |
List<String> tabletFacets = Arrays.asList("F_50037","F_50036", "F_50035", "F_50034");
|
| 9103 |
anupam.sin |
83 |
//Operating System, Processor, Storage, RAM, Screen Size
|
|
|
84 |
List<String> laptopFacets = Arrays.asList("F_50013", "F_50014", "F_50015","F_50017", "F_50033");
|
|
|
85 |
//Resolution, Optical Zoon, Display Size
|
|
|
86 |
List<String> compactCameras = Arrays.asList("F_50024", "F_50025", "F_50026");
|
|
|
87 |
//Display Size
|
|
|
88 |
List<String> dslrCameras = Arrays.asList("F_50026");
|
|
|
89 |
//Capacity, Class
|
|
|
90 |
List<String> memoryCards = Arrays.asList("F_50018", "F_50019");
|
|
|
91 |
//Capacity
|
|
|
92 |
List<String> penDrives = Arrays.asList("F_50020");
|
|
|
93 |
//Capacity, Type, Interface
|
|
|
94 |
List<String> externalHardDisks = Arrays.asList("F_50021", "F_50022", "F_50023");
|
|
|
95 |
{
|
|
|
96 |
put(SearchController.getCategoryLabel(10001l), mobileFacets);
|
|
|
97 |
put(SearchController.getCategoryLabel(10006l), mobileFacets);
|
|
|
98 |
put(SearchController.getCategoryLabel(10009l), tabletFacets);
|
|
|
99 |
put(SearchController.getCategoryLabel(10010l), tabletFacets);
|
|
|
100 |
put(SearchController.getCategoryLabel(10013l), memoryCards);
|
|
|
101 |
put(SearchController.getCategoryLabel(10017l), penDrives);
|
|
|
102 |
put(SearchController.getCategoryLabel(10049l), laptopFacets);
|
|
|
103 |
put(SearchController.getCategoryLabel(10050l), laptopFacets);
|
|
|
104 |
put(SearchController.getCategoryLabel(10073l), externalHardDisks);
|
|
|
105 |
put(SearchController.getCategoryLabel(11002l), compactCameras);
|
|
|
106 |
put(SearchController.getCategoryLabel(11003l), dslrCameras);
|
|
|
107 |
}
|
|
|
108 |
});
|
|
|
109 |
static {
|
|
|
110 |
String solr_url = null;
|
|
|
111 |
try {
|
|
|
112 |
solr_url = ConfigClient.getClient().get("solr_url");
|
|
|
113 |
}catch(ConfigException cex){
|
|
|
114 |
log.error("Unable to get the solr URL from the config server. Setting the default value.", cex);
|
|
|
115 |
solr_url = "http://localhost:8983/solr/select/";
|
|
|
116 |
}
|
|
|
117 |
SOLR_URL = solr_url;
|
|
|
118 |
}
|
|
|
119 |
|
|
|
120 |
/**
|
|
|
121 |
*
|
|
|
122 |
*/
|
|
|
123 |
private XPath xpath;
|
|
|
124 |
|
|
|
125 |
/**
|
|
|
126 |
*
|
|
|
127 |
*/
|
|
|
128 |
private InputSource inputSource;
|
|
|
129 |
|
|
|
130 |
Map<String,Map<String,Integer>> facetMap;
|
|
|
131 |
|
|
|
132 |
private String query;
|
|
|
133 |
|
|
|
134 |
List<String> resultMap;
|
|
|
135 |
|
|
|
136 |
long numberOfResults=0;
|
|
|
137 |
|
|
|
138 |
String priceFacetName = "F_50002";
|
|
|
139 |
|
| 9406 |
amit.gupta |
140 |
List<String> filtrableFacets = new ArrayList<String>(Utils.rootfacetDefIDs);
|
| 9103 |
anupam.sin |
141 |
|
|
|
142 |
/**
|
|
|
143 |
*
|
|
|
144 |
* @param query
|
|
|
145 |
* @param facetDefinitionIDs
|
|
|
146 |
*/
|
|
|
147 |
public SolrSearchService(String query, String[] facetqueries, long start, long rows, Double minPrice, Double maxPrice, String sortOrder, long sourceId) {
|
|
|
148 |
|
|
|
149 |
this.query = query;
|
|
|
150 |
|
|
|
151 |
List<String> facetsQueried = new ArrayList<String>();
|
|
|
152 |
if(sourceId != -1){
|
|
|
153 |
priceFacetName = priceFacetName + "_" + sourceId;
|
|
|
154 |
}
|
|
|
155 |
|
|
|
156 |
setFilterableFacets(facetqueries);
|
| 9406 |
amit.gupta |
157 |
this.filtrableFacets.remove(Utils.availabilityFacet);
|
|
|
158 |
this.filtrableFacets.add(Utils.availabilityFacet);
|
| 9103 |
anupam.sin |
159 |
|
|
|
160 |
|
|
|
161 |
this.xpath = XPathFactory.newInstance().newXPath();
|
|
|
162 |
|
|
|
163 |
query = query.trim().replaceAll("\\s+", " ");
|
|
|
164 |
log.info("query=" + query);
|
|
|
165 |
|
|
|
166 |
String uri = SOLR_URL + "?wt=xml&q=" + this.query;
|
|
|
167 |
|
|
|
168 |
uri += "&stats=on&stats.field=" + priceFacetName;
|
|
|
169 |
|
|
|
170 |
if(sortOrder != null){
|
|
|
171 |
//replace the price facet name, so that it can pick price for the source.
|
|
|
172 |
sortOrder = sortOrder.replace("F_50002", priceFacetName);
|
|
|
173 |
uri += "&sort=" + sortOrder;
|
|
|
174 |
}
|
|
|
175 |
|
|
|
176 |
if(facetqueries != null) {
|
|
|
177 |
//sorting will guarantee all similar facets together so that we can assume or between all similar items without fail.
|
|
|
178 |
Arrays.sort(facetqueries);
|
|
|
179 |
String fq="";
|
|
|
180 |
for(int i=0; i<facetqueries.length; i++) {
|
|
|
181 |
String value = "";
|
|
|
182 |
String[] tokens = facetqueries[i].split(":");
|
|
|
183 |
try {
|
|
|
184 |
value = URLEncoder.encode(tokens[1], "UTF-8");
|
|
|
185 |
} catch (UnsupportedEncodingException e) {
|
|
|
186 |
// TODO Auto-generated catch block
|
|
|
187 |
e.printStackTrace();
|
|
|
188 |
}
|
|
|
189 |
if(facetsQueried.contains(tokens[0])) {
|
|
|
190 |
uri += " OR " + tokens[0] + ":\"" + value + "\"";
|
|
|
191 |
|
|
|
192 |
} else {
|
|
|
193 |
fq = "{!tag=dt" + facetsQueried.size() + "}";
|
|
|
194 |
facetsQueried.add(tokens[0]);
|
|
|
195 |
fq += tokens[0] + ":\"" + value + "\"";
|
|
|
196 |
uri += "&fq=" + fq;
|
|
|
197 |
}
|
|
|
198 |
}
|
|
|
199 |
}
|
|
|
200 |
String minString = "0";
|
|
|
201 |
String maxString = "*";
|
|
|
202 |
if(minPrice != null || maxPrice != null){
|
|
|
203 |
try {
|
| 9886 |
amit.gupta |
204 |
log.info("Uri:" + uri);
|
| 9103 |
anupam.sin |
205 |
dynamicPriceMap = getPriceStatsMap(new InputSource(uri));
|
|
|
206 |
} catch (Exception e){
|
|
|
207 |
e.printStackTrace();
|
|
|
208 |
}
|
|
|
209 |
if(minPrice != null){
|
|
|
210 |
minString = minPrice.toString();
|
|
|
211 |
}
|
|
|
212 |
if(maxPrice != null){
|
|
|
213 |
maxString = maxPrice.toString();
|
|
|
214 |
}
|
|
|
215 |
}
|
| 9421 |
amit.gupta |
216 |
uri = uri.replaceAll(" ", "%20");
|
| 9269 |
amit.gupta |
217 |
uri += "&fq=" + priceFacetName + ":["+ minString + "%20" + maxString + "]";
|
| 9103 |
anupam.sin |
218 |
uri += "&fl=ID,Name&facet=true&start=" + start + "&rows=" + rows + "&facet.mincount=1";
|
|
|
219 |
for(String facetDefinitionID : filtrableFacets) {
|
|
|
220 |
if(facetsQueried.contains(facetDefinitionID)){
|
|
|
221 |
uri += "&facet.field={!ex=dt" + facetsQueried.indexOf(facetDefinitionID)+ "}"+ facetDefinitionID;
|
|
|
222 |
} else {
|
|
|
223 |
uri += "&facet.field=" + facetDefinitionID;
|
|
|
224 |
}
|
|
|
225 |
}
|
|
|
226 |
log.info("uri=" + uri);
|
|
|
227 |
|
|
|
228 |
this.inputSource = new InputSource(uri);
|
|
|
229 |
|
|
|
230 |
this.facetMap = getFacetMap();
|
|
|
231 |
}
|
|
|
232 |
|
|
|
233 |
@SuppressWarnings("unchecked")
|
|
|
234 |
private void setFilterableFacets(String[] facetqueries) {
|
|
|
235 |
List<String> queriedFacets = getAllMatches(this.query);
|
| 9406 |
amit.gupta |
236 |
if(facetqueries != null && facetqueries.length!=0) {
|
| 9103 |
anupam.sin |
237 |
String facetString = Arrays.toString(facetqueries);
|
|
|
238 |
List<String> filteredFacets = getAllMatches(facetString);
|
|
|
239 |
if(filteredFacets.contains("F_50011")){
|
|
|
240 |
for(String facetQuery : facetqueries) {
|
|
|
241 |
if(facetQuery.contains("F_50011")){
|
|
|
242 |
String facetVal = facetQuery.split(":")[1];
|
|
|
243 |
if(CATEGORY_FACET_MAP.containsKey(facetVal)){
|
| 9406 |
amit.gupta |
244 |
this.filtrableFacets.addAll(CATEGORY_FACET_MAP.get(facetVal));
|
| 9103 |
anupam.sin |
245 |
return;
|
|
|
246 |
} else {
|
|
|
247 |
break;
|
|
|
248 |
}
|
|
|
249 |
}
|
|
|
250 |
}
|
|
|
251 |
}
|
|
|
252 |
if(filteredFacets.contains("F_50010")){
|
|
|
253 |
for(String facetQuery : facetqueries) {
|
|
|
254 |
if(facetQuery.contains("F_50010")){
|
|
|
255 |
String facetVal = facetQuery.split(":")[1];
|
|
|
256 |
if(CATEGORY_FACET_MAP.containsKey(facetVal)){
|
|
|
257 |
return;
|
|
|
258 |
} else {
|
|
|
259 |
break;
|
|
|
260 |
}
|
|
|
261 |
}
|
|
|
262 |
}
|
|
|
263 |
}
|
|
|
264 |
}
|
|
|
265 |
if(queriedFacets.contains("F_50011")) {
|
| 9406 |
amit.gupta |
266 |
String facetVal = this.query.split("F_50011:")[1].split("&")[0].replaceAll("[\"()]", "");
|
| 9103 |
anupam.sin |
267 |
if (facetVal.contains(" OR ")) {
|
|
|
268 |
return;
|
|
|
269 |
} else if(CATEGORY_FACET_MAP.containsKey(facetVal)){
|
| 9406 |
amit.gupta |
270 |
//facetVal = facetVal.split("&")[0].replaceAll("[\"()]", "");
|
|
|
271 |
this.filtrableFacets.addAll(CATEGORY_FACET_MAP.get(facetVal));
|
| 9103 |
anupam.sin |
272 |
return;
|
| 9406 |
amit.gupta |
273 |
} else {
|
|
|
274 |
return;
|
|
|
275 |
}
|
| 9103 |
anupam.sin |
276 |
}
|
|
|
277 |
if(queriedFacets.contains("F_50010")){
|
| 9406 |
amit.gupta |
278 |
String facetVal = this.query.split("F_50010:")[1].split("&")[0].replaceAll("[\"()]", "");
|
| 9103 |
anupam.sin |
279 |
if (facetVal.contains(" OR ")) {
|
|
|
280 |
return;
|
|
|
281 |
} else if(CATEGORY_FACET_MAP.containsKey(facetVal)){
|
| 9406 |
amit.gupta |
282 |
this.filtrableFacets.addAll(CATEGORY_FACET_MAP.get(facetVal));
|
| 9103 |
anupam.sin |
283 |
return;
|
| 9406 |
amit.gupta |
284 |
} else {
|
|
|
285 |
return;
|
| 9103 |
anupam.sin |
286 |
}
|
|
|
287 |
}
|
|
|
288 |
}
|
|
|
289 |
|
|
|
290 |
public List<String> getFilterableFacets() {
|
|
|
291 |
return this.filtrableFacets;
|
|
|
292 |
}
|
|
|
293 |
|
|
|
294 |
public Map<String,Map<String,Integer>> removeUnwantedFacets(Map<String,Map<String,Integer>> facetMap, long numberOfResults){
|
|
|
295 |
|
|
|
296 |
Set<String> facetsInQuery = new HashSet<String>(getAllMatches(this.query));
|
|
|
297 |
Map<String,Map<String,Integer>> tempFacets = new TreeMap<String, Map<String,Integer>>();
|
|
|
298 |
for(String facet : facetMap.keySet()){
|
|
|
299 |
if(facetMap.get(facet).size() > 0 && !facetsInQuery.contains(facet)){
|
|
|
300 |
Map<String,Integer> tempMap = new LinkedHashMap<String, Integer>();
|
|
|
301 |
|
|
|
302 |
for(String facetValueName : facetMap.get(facet).keySet()){
|
|
|
303 |
//if(facetMap.get(facet).get(facetValueName) != 0 && facetMap.get(facet).get(facetValueName) != numberOfResults){
|
|
|
304 |
tempMap.put(facetValueName, facetMap.get(facet).get(facetValueName));
|
|
|
305 |
//}
|
|
|
306 |
}
|
|
|
307 |
if(!tempMap.isEmpty()){
|
|
|
308 |
tempFacets.put(facet, tempMap);
|
|
|
309 |
}
|
|
|
310 |
}
|
|
|
311 |
}
|
|
|
312 |
/*if(tempFacets.containsKey("F_50010")){
|
|
|
313 |
tempFacets.remove("F_50011");
|
|
|
314 |
}*/
|
|
|
315 |
|
|
|
316 |
return tempFacets;
|
|
|
317 |
}
|
|
|
318 |
|
|
|
319 |
public Map<String,Integer> getFacetDetails(String facetName){
|
|
|
320 |
if(facetMap != null){
|
|
|
321 |
return facetMap.get(facetName);
|
|
|
322 |
}else{
|
|
|
323 |
return null;
|
|
|
324 |
}
|
|
|
325 |
}
|
|
|
326 |
|
|
|
327 |
public Map<String,Map<String,Integer>> getFacetMap() {
|
|
|
328 |
facetMap = new TreeMap<String,Map<String,Integer>>();
|
|
|
329 |
|
|
|
330 |
String facetNamePath = "/response/lst/lst[@name = 'facet_fields']/lst";
|
|
|
331 |
|
|
|
332 |
NodeList nodes = null;
|
|
|
333 |
try {
|
|
|
334 |
nodes = (NodeList) this.xpath.evaluate(facetNamePath, this.inputSource, XPathConstants.NODESET);
|
|
|
335 |
}
|
|
|
336 |
catch (XPathExpressionException xpee) {
|
|
|
337 |
return null;
|
|
|
338 |
}
|
|
|
339 |
|
|
|
340 |
if(nodes.getLength() == 0) {
|
|
|
341 |
return null;
|
|
|
342 |
}
|
|
|
343 |
|
|
|
344 |
NodeList subNodes = null;
|
|
|
345 |
|
|
|
346 |
for(int i=0; i<nodes.getLength(); i++) {
|
|
|
347 |
Node node = nodes.item(i);
|
|
|
348 |
String facetName = node.getAttributes().getNamedItem("name").getNodeValue();
|
|
|
349 |
subNodes = node.getChildNodes();
|
|
|
350 |
Map<String,Integer> facetValueCountMap = new LinkedHashMap<String,Integer>();
|
|
|
351 |
for(int j=0; j<subNodes.getLength(); j++) {
|
|
|
352 |
Node subNode = subNodes.item(j);
|
|
|
353 |
facetValueCountMap.put(subNode.getAttributes().getNamedItem("name").getNodeValue(), Integer.parseInt(subNode.getTextContent()));
|
|
|
354 |
}
|
|
|
355 |
if(SORTED_FACET_VALUE_MAP.containsKey(facetName)){
|
|
|
356 |
List<String> orderedValues = SORTED_FACET_VALUE_MAP.get(facetName);
|
|
|
357 |
Map<String, Integer> sortedMap = new LinkedHashMap<String, Integer>();
|
|
|
358 |
for (Iterator<String> it = orderedValues.iterator(); it.hasNext();) {
|
|
|
359 |
String val = it.next();
|
|
|
360 |
if(facetValueCountMap.containsKey(val)) {
|
|
|
361 |
sortedMap.put(val, facetValueCountMap.get(val));
|
|
|
362 |
}
|
|
|
363 |
}
|
|
|
364 |
facetMap.put(facetName, sortedMap);
|
|
|
365 |
} else {
|
|
|
366 |
facetMap.put(facetName, facetValueCountMap);
|
|
|
367 |
}
|
|
|
368 |
}
|
|
|
369 |
this.numberOfResults = this.getTotalResults();
|
|
|
370 |
|
|
|
371 |
facetMap = removeUnwantedFacets(facetMap, numberOfResults);
|
|
|
372 |
return facetMap;
|
|
|
373 |
}
|
|
|
374 |
|
|
|
375 |
public List<String> getResultMap() {
|
|
|
376 |
resultMap = new LinkedList<String>();
|
|
|
377 |
|
|
|
378 |
String resultDocsPath = "/response/result/doc";
|
|
|
379 |
|
|
|
380 |
|
|
|
381 |
NodeList nodes = null;
|
|
|
382 |
try {
|
|
|
383 |
nodes = (NodeList) this.xpath.evaluate(resultDocsPath, this.inputSource, XPathConstants.NODESET);
|
|
|
384 |
}
|
|
|
385 |
catch (XPathExpressionException xpee) {
|
|
|
386 |
return null;
|
|
|
387 |
}
|
|
|
388 |
|
|
|
389 |
if(nodes.getLength() == 0) {
|
|
|
390 |
return null;
|
|
|
391 |
}
|
|
|
392 |
|
|
|
393 |
for(int i=0; i<nodes.getLength(); i++) {
|
|
|
394 |
Node node = nodes.item(i);
|
|
|
395 |
String docID = node.getFirstChild().getTextContent();
|
|
|
396 |
resultMap.add(docID);
|
|
|
397 |
}
|
|
|
398 |
return resultMap;
|
|
|
399 |
}
|
|
|
400 |
|
|
|
401 |
public HashMap<String, Double> getPriceStatsMap() {
|
|
|
402 |
return this.getPriceStatsMap(this.inputSource);
|
|
|
403 |
}
|
|
|
404 |
|
|
|
405 |
public HashMap<String, Double> getPriceStatsMap(InputSource inputSource) {
|
|
|
406 |
HashMap<String, Double> priceStatsMap = new HashMap<String, Double>();
|
|
|
407 |
|
|
|
408 |
String resultDocsPath = "/response/lst[@name = 'stats']/lst[@name = 'stats_fields']/lst[@name = '" + priceFacetName + "']";
|
|
|
409 |
|
|
|
410 |
|
|
|
411 |
NodeList nodes = null;
|
|
|
412 |
try {
|
|
|
413 |
nodes = (NodeList) this.xpath.evaluate(resultDocsPath, inputSource, XPathConstants.NODESET);
|
|
|
414 |
}
|
|
|
415 |
catch (XPathExpressionException xpee) {
|
|
|
416 |
return null;
|
|
|
417 |
}
|
|
|
418 |
|
|
|
419 |
if(nodes.getLength() == 0) {
|
|
|
420 |
return null;
|
|
|
421 |
}
|
|
|
422 |
|
|
|
423 |
NodeList subNodes = nodes.item(0).getChildNodes();
|
|
|
424 |
|
|
|
425 |
for(int i=0; i<subNodes.getLength(); i++) {
|
|
|
426 |
Node node = subNodes.item(i);
|
|
|
427 |
|
|
|
428 |
String parameter = node.getAttributes().getNamedItem("name").getNodeValue();
|
|
|
429 |
String value = node.getTextContent();
|
|
|
430 |
priceStatsMap.put(parameter, Double.parseDouble(value));
|
|
|
431 |
}
|
|
|
432 |
return priceStatsMap;
|
|
|
433 |
}
|
|
|
434 |
|
|
|
435 |
public HashMap<String,Integer> getRangeQueryResultMap() {
|
|
|
436 |
HashMap<String, Integer> rangeQueryResultMap = new HashMap<String,Integer>();
|
|
|
437 |
|
|
|
438 |
String resultDocsPath = "/response/lst[@name = 'facet_counts']/lst[@name = 'facet_queries']/int";
|
|
|
439 |
|
|
|
440 |
|
|
|
441 |
NodeList nodes = null;
|
|
|
442 |
try {
|
|
|
443 |
nodes = (NodeList) this.xpath.evaluate(resultDocsPath, this.inputSource, XPathConstants.NODESET);
|
|
|
444 |
}
|
|
|
445 |
catch (XPathExpressionException xpee) {
|
|
|
446 |
return null;
|
|
|
447 |
}
|
|
|
448 |
|
|
|
449 |
if(nodes.getLength() == 0) {
|
|
|
450 |
return null;
|
|
|
451 |
}
|
|
|
452 |
|
|
|
453 |
|
|
|
454 |
for(int i=0; i<nodes.getLength(); i++) {
|
|
|
455 |
Node node = nodes.item(i);
|
|
|
456 |
|
|
|
457 |
String query = node.getAttributes().getNamedItem("name").getNodeValue();
|
|
|
458 |
String docCount = node.getTextContent();
|
|
|
459 |
|
|
|
460 |
rangeQueryResultMap.put(query,Integer.parseInt(docCount));
|
|
|
461 |
}
|
|
|
462 |
return rangeQueryResultMap;
|
|
|
463 |
|
|
|
464 |
}
|
|
|
465 |
|
|
|
466 |
/**
|
|
|
467 |
*
|
|
|
468 |
*/
|
|
|
469 |
public long getTotalResults(){
|
|
|
470 |
String resultDocsPath = "/response/result";
|
|
|
471 |
NodeList nodes = null;
|
|
|
472 |
try {
|
|
|
473 |
nodes = (NodeList) this.xpath.evaluate(resultDocsPath, this.inputSource, XPathConstants.NODESET);
|
|
|
474 |
}
|
|
|
475 |
catch (XPathExpressionException xpee) {
|
|
|
476 |
return 0;
|
|
|
477 |
}
|
|
|
478 |
|
|
|
479 |
Node node = nodes.item(0);
|
|
|
480 |
|
|
|
481 |
return Long.parseLong(node.getAttributes().getNamedItem("numFound").getNodeValue());
|
|
|
482 |
|
|
|
483 |
}
|
|
|
484 |
/**
|
|
|
485 |
*
|
|
|
486 |
* @return
|
|
|
487 |
*/
|
|
|
488 |
public long[] getResultEntityIDs() {
|
|
|
489 |
String expression = "/response/result/doc/long";
|
|
|
490 |
|
|
|
491 |
NodeList nodes = null;
|
|
|
492 |
try {
|
|
|
493 |
nodes = (NodeList) this.xpath.evaluate(expression, this.inputSource,
|
|
|
494 |
XPathConstants.NODESET);
|
|
|
495 |
}
|
|
|
496 |
catch(XPathExpressionException xpee) {
|
|
|
497 |
return null;
|
|
|
498 |
}
|
|
|
499 |
|
|
|
500 |
if(nodes.getLength() == 0) {
|
|
|
501 |
return null;
|
|
|
502 |
}
|
|
|
503 |
|
|
|
504 |
long[] values = new long[nodes.getLength()];
|
|
|
505 |
for(int i=0; i<nodes.getLength(); i++) {
|
|
|
506 |
Node node = nodes.item(i);
|
|
|
507 |
String value = node.getTextContent();
|
|
|
508 |
values[i] = Long.parseLong(value);
|
|
|
509 |
}
|
|
|
510 |
|
|
|
511 |
return values;
|
|
|
512 |
}
|
|
|
513 |
|
|
|
514 |
/**
|
|
|
515 |
*
|
|
|
516 |
* @return
|
|
|
517 |
*/
|
|
|
518 |
public String[] getResultCategoryNames() {
|
|
|
519 |
String expression = "/response/lst/lst[@name = 'facet_fields']/";
|
|
|
520 |
expression += "lst[@name = 'Category']/int/@name";
|
|
|
521 |
|
|
|
522 |
NodeList nodes = null;
|
|
|
523 |
try {
|
|
|
524 |
nodes = (NodeList) this.xpath.evaluate(expression,
|
|
|
525 |
this.inputSource, XPathConstants.NODESET);
|
|
|
526 |
}
|
|
|
527 |
catch (XPathExpressionException xpee) {
|
|
|
528 |
return null;
|
|
|
529 |
}
|
|
|
530 |
|
|
|
531 |
if(nodes.getLength() == 0) {
|
|
|
532 |
return null;
|
|
|
533 |
}
|
|
|
534 |
|
|
|
535 |
String[] values = new String[nodes.getLength()];
|
|
|
536 |
for(int i=0; i<nodes.getLength(); i++) {
|
|
|
537 |
Node node = nodes.item(i);
|
|
|
538 |
values[i] = node.getTextContent();
|
|
|
539 |
}
|
|
|
540 |
|
|
|
541 |
return values;
|
|
|
542 |
}
|
|
|
543 |
|
|
|
544 |
/**
|
|
|
545 |
*
|
|
|
546 |
* @return
|
|
|
547 |
*/
|
|
|
548 |
public int[] getResultCategoryCounts() {
|
|
|
549 |
String expression = "/response/lst/lst[@name = 'facet_fields']/";
|
|
|
550 |
expression += "lst[@name = 'Category']/int";
|
|
|
551 |
|
|
|
552 |
NodeList nodes = null;
|
|
|
553 |
try {
|
|
|
554 |
nodes = (NodeList) this.xpath.evaluate(expression,
|
|
|
555 |
this.inputSource, XPathConstants.NODESET);
|
|
|
556 |
}
|
|
|
557 |
catch (XPathExpressionException xpee) {
|
|
|
558 |
return null;
|
|
|
559 |
}
|
|
|
560 |
|
|
|
561 |
if(nodes.getLength() == 0) {
|
|
|
562 |
return null;
|
|
|
563 |
}
|
|
|
564 |
|
|
|
565 |
int[] values = new int[nodes.getLength()];
|
|
|
566 |
for(int i=0; i<nodes.getLength(); i++) {
|
|
|
567 |
Node node = nodes.item(i);
|
|
|
568 |
values[i] = Integer.parseInt(node.getTextContent());
|
|
|
569 |
}
|
|
|
570 |
|
|
|
571 |
return values;
|
|
|
572 |
}
|
|
|
573 |
|
|
|
574 |
/**
|
|
|
575 |
*
|
|
|
576 |
* @return
|
|
|
577 |
*/
|
|
|
578 |
public String[] getResultEntityNames() {
|
|
|
579 |
String expression = "/response/result/doc/str";
|
|
|
580 |
|
|
|
581 |
NodeList nodes = null;
|
|
|
582 |
try {
|
|
|
583 |
nodes = (NodeList) this.xpath.evaluate(expression, this.inputSource,
|
|
|
584 |
XPathConstants.NODESET);
|
|
|
585 |
}
|
|
|
586 |
catch(XPathExpressionException xpee) {
|
|
|
587 |
return null;
|
|
|
588 |
}
|
|
|
589 |
|
|
|
590 |
if(nodes.getLength() == 0) {
|
|
|
591 |
return null;
|
|
|
592 |
}
|
|
|
593 |
|
|
|
594 |
String[] values = new String[nodes.getLength()];
|
|
|
595 |
for(int i=0; i<nodes.getLength(); i++) {
|
|
|
596 |
Node node = nodes.item(i);
|
|
|
597 |
String value = node.getTextContent();
|
|
|
598 |
values[i] = value;
|
|
|
599 |
}
|
|
|
600 |
|
|
|
601 |
return values;
|
|
|
602 |
}
|
|
|
603 |
|
|
|
604 |
/**
|
|
|
605 |
*
|
|
|
606 |
* @param facetDefinitionID
|
|
|
607 |
* @return
|
|
|
608 |
*/
|
|
|
609 |
public String[] getFacetValues(String facetDefinitionID) {
|
|
|
610 |
String expression = "/response/lst/lst[@name = 'facet_fields']/";
|
|
|
611 |
expression += "lst[@name = '"+ facetDefinitionID +"']/int/@name";
|
|
|
612 |
|
|
|
613 |
NodeList nodes = null;
|
|
|
614 |
try {
|
|
|
615 |
nodes = (NodeList) this.xpath.evaluate(expression,
|
|
|
616 |
this.inputSource, XPathConstants.NODESET);
|
|
|
617 |
}
|
|
|
618 |
catch (XPathExpressionException xpee) {
|
|
|
619 |
return null;
|
|
|
620 |
}
|
|
|
621 |
|
|
|
622 |
if(nodes.getLength() == 0) {
|
|
|
623 |
return null;
|
|
|
624 |
}
|
|
|
625 |
|
|
|
626 |
String[] values = new String[nodes.getLength()];
|
|
|
627 |
for(int i=0; i<nodes.getLength(); i++) {
|
|
|
628 |
Node node = nodes.item(i);
|
|
|
629 |
values[i] = node.getTextContent();
|
|
|
630 |
}
|
|
|
631 |
|
|
|
632 |
return values;
|
|
|
633 |
}
|
|
|
634 |
|
|
|
635 |
/**
|
|
|
636 |
*
|
|
|
637 |
* @param facetDefinitionID
|
|
|
638 |
* @return
|
|
|
639 |
*/
|
|
|
640 |
public String[] getFacetCounts(String facetDefinitionID) {
|
|
|
641 |
String expression = "/response/lst/lst[@name = 'facet_fields']/";
|
|
|
642 |
expression += "lst[@name = '" + facetDefinitionID + "']/int";
|
|
|
643 |
|
|
|
644 |
NodeList nodes = null;
|
|
|
645 |
try {
|
|
|
646 |
nodes = (NodeList) this.xpath.evaluate(expression,
|
|
|
647 |
this.inputSource, XPathConstants.NODESET);
|
|
|
648 |
}
|
|
|
649 |
catch (XPathExpressionException xpee) {
|
|
|
650 |
return null;
|
|
|
651 |
}
|
|
|
652 |
|
|
|
653 |
if(nodes.getLength() == 0) {
|
|
|
654 |
return null;
|
|
|
655 |
}
|
|
|
656 |
|
|
|
657 |
String[] values = new String[nodes.getLength()];
|
|
|
658 |
for(int i=0; i<nodes.getLength(); i++) {
|
|
|
659 |
Node node = nodes.item(i);
|
|
|
660 |
values[i] = node.getTextContent();
|
|
|
661 |
}
|
|
|
662 |
|
|
|
663 |
return values;
|
|
|
664 |
}
|
|
|
665 |
|
|
|
666 |
public static void main(String[] args){
|
|
|
667 |
/*
|
|
|
668 |
// Hard coded for now
|
|
|
669 |
String[] facetDefIDs = new String[] {"F_50001", "F_50002", "F_50003", "F_50004", "F_50005", "F_50006", "F_50007", "F_50008", "F_50009"};
|
|
|
670 |
|
|
|
671 |
// Hard-coded for now
|
|
|
672 |
String[] facetLabels = new String[] {
|
|
|
673 |
"Brand", "Price","Form Factor", "Carry In Pocket", "Cellular Technologies",
|
|
|
674 |
"Data Connectivity", "Camera Resolution", "Built-in Memory",
|
|
|
675 |
"Talk time"
|
|
|
676 |
};
|
|
|
677 |
|
|
|
678 |
*/
|
|
|
679 |
String[] facetDefIDs = new String[] {"Category","F_50002","F_50001", "F_50006", "F_50007" };
|
|
|
680 |
//String[] facetLabels = new String[] {"Category","Price", "Brand", "Data Connectivity", "Camera Resolution" };
|
|
|
681 |
|
|
|
682 |
|
|
|
683 |
String[] fqrys = {};
|
|
|
684 |
SolrSearchService search = new SolrSearchService("nokia", fqrys, 0 , 20, null, null, null, -1);
|
|
|
685 |
|
|
|
686 |
long[] entityIDs = search.getResultEntityIDs();
|
|
|
687 |
log.info("entityIDs=" + Arrays.toString(entityIDs));
|
|
|
688 |
|
|
|
689 |
String[] entityNames = search.getResultEntityNames();
|
|
|
690 |
log.info("entityNames=" + Arrays.toString(entityNames));
|
|
|
691 |
search.getFacetMap();
|
|
|
692 |
|
|
|
693 |
search.getResultMap();
|
|
|
694 |
search.getRangeQueryResultMap();
|
|
|
695 |
search.getPriceStatsMap(new InputSource());
|
|
|
696 |
search.getTotalResults();
|
|
|
697 |
for (int i=0; i<facetDefIDs.length; i++) {
|
|
|
698 |
search.getFacetCounts(facetDefIDs[i]);
|
|
|
699 |
search.getFacetValues(facetDefIDs[i]);
|
|
|
700 |
}
|
|
|
701 |
|
|
|
702 |
}
|
|
|
703 |
|
|
|
704 |
|
|
|
705 |
public static List<String> getAllMatches(String text) {
|
|
|
706 |
List<String> matches = new ArrayList<String>();
|
|
|
707 |
Matcher m = FACET_PATTERN.matcher(text);
|
|
|
708 |
while(m.find()) {
|
|
|
709 |
matches.add(m.group(1));
|
|
|
710 |
}
|
|
|
711 |
return matches;
|
|
|
712 |
}
|
|
|
713 |
|
|
|
714 |
public Map<String, Double> getDynamicPriceMap() {
|
|
|
715 |
return this.dynamicPriceMap;
|
|
|
716 |
}
|
|
|
717 |
|
|
|
718 |
}
|