Revision 33185
Added by Eri Katsari almost 10 years ago
modules/dnet-openaire-stats/trunk/src/test/java/eu/dnetlib/data/mapreduce/hbase/statsExport/utils/ContextExporterTest.java | ||
---|---|---|
1 |
package eu.dnetlib.data.mapreduce.hbase.statsExport.utils; |
|
2 |
|
|
3 |
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException; |
|
4 |
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService; |
|
5 |
import org.apache.cxf.jaxws.JaxWsProxyFactoryBean; |
|
6 |
import org.apache.hadoop.conf.Configuration; |
|
7 |
import org.apache.hadoop.fs.FSDataOutputStream; |
|
8 |
import org.apache.hadoop.fs.FileSystem; |
|
9 |
import org.apache.hadoop.fs.Path; |
|
10 |
import org.apache.log4j.Logger; |
|
11 |
import org.junit.Before; |
|
12 |
import org.junit.Test; |
|
13 |
|
|
14 |
import java.util.ArrayList; |
|
15 |
import java.util.List; |
|
16 |
|
|
17 |
public class ContextExporterTest { |
|
18 |
private ContextTransformer contextTransformer = new ContextTransformer(); |
|
19 |
private String outputPath; |
|
20 |
private String url; |
|
21 |
|
|
22 |
private Logger log = Logger.getLogger(this.getClass()); |
|
23 |
|
|
24 |
private ArrayList<String> context = new ArrayList<String>(); |
|
25 |
private ArrayList<String> category = new ArrayList<String>(); |
|
26 |
private ArrayList<String> concept = new ArrayList<String>(); |
|
27 |
|
|
28 |
public ContextExporterTest() throws Exception { |
|
29 |
|
|
30 |
} |
|
31 |
|
|
32 |
@Before |
|
33 |
public void init() { |
|
34 |
url = "http://beta.services.openaire.eu:8280/is/services/isLookUp"; |
|
35 |
} |
|
36 |
|
|
37 |
@Test |
|
38 |
public void readFromUrl() throws Exception { |
|
39 |
|
|
40 |
List<String> concepts = getContextResouces(url); |
|
41 |
log.info("Returned concept " + concepts.size() |
|
42 |
); |
|
43 |
|
|
44 |
for (String data : concepts) { |
|
45 |
log.info("++++++++++++++ Transforming concept data "); |
|
46 |
String res = contextTransformer.transformXSL(data); |
|
47 |
|
|
48 |
// log.info("++++++++++++++"); |
|
49 |
// log.info(res); |
|
50 |
|
|
51 |
// log.info("++++++++++++++"); |
|
52 |
processData(res); |
|
53 |
|
|
54 |
} |
|
55 |
|
|
56 |
log.info(" ++++++++++++++concept ++++++++++++++"); |
|
57 |
|
|
58 |
// log.info(concept); |
|
59 |
|
|
60 |
|
|
61 |
log.info("Context size " + context.size() +" and data " +context); |
|
62 |
|
|
63 |
|
|
64 |
log.info("category list size " + category.size() +" and data "+ category); |
|
65 |
|
|
66 |
// writeData(context, outputPath + "context"); |
|
67 |
|
|
68 |
|
|
69 |
// writeData(category, outputPath + "category"); |
|
70 |
|
|
71 |
// writeData(concept, outputPath + "concept"); |
|
72 |
|
|
73 |
} |
|
74 |
|
|
75 |
|
|
76 |
private List<String> getContextResouces(String url) throws ISLookUpException { |
|
77 |
ISLookUpService lookUpService; |
|
78 |
|
|
79 |
JaxWsProxyFactoryBean factory = new JaxWsProxyFactoryBean(); |
|
80 |
factory.setServiceClass(ISLookUpService.class); |
|
81 |
factory.setAddress(url); |
|
82 |
|
|
83 |
lookUpService = (ISLookUpService) factory.create(); |
|
84 |
// for $x in //RESOURCE_PROFILE[.//RESOURCE_TYPE/@value='ContextDSResourceType'] |
|
85 |
// [.//RESOURCE_KIND/@value='ContextDSResources'] return $x |
|
86 |
return lookUpService.quickSearchProfile("//RESOURCE_PROFILE[.//RESOURCE_TYPE/@value='ContextDSResourceType'][.//RESOURCE_KIND/@value='ContextDSResources']"); |
|
87 |
} |
|
88 |
|
|
89 |
|
|
90 |
private void processData(String data) throws Exception { |
|
91 |
try { |
|
92 |
|
|
93 |
String[] split = data.split("COPY"); |
|
94 |
|
|
95 |
split[0] = split[0].replaceFirst("\n", ""); |
|
96 |
context.add(split[0]); |
|
97 |
split[1] = split[1].replaceFirst("\n", ""); |
|
98 |
if (split[1].endsWith("\n")) { |
|
99 |
split[1] = split[1].substring(0, split[1].lastIndexOf("\n")); |
|
100 |
log.info("split 1 : " + split[1] + "+++++++"); |
|
101 |
} |
|
102 |
split[1] = split[1].replaceFirst("\n", ""); |
|
103 |
category.add(split[1]); |
|
104 |
|
|
105 |
|
|
106 |
split[2] = split[2].replaceFirst("\n", ""); |
|
107 |
|
|
108 |
concept.add(split[2]); |
|
109 |
|
|
110 |
} catch (Exception e) { |
|
111 |
String msg = " Unable to create file with context, " + "concept and category values in output path " + outputPath + ". Reason: "; |
|
112 |
log.error(msg); |
|
113 |
throw new Exception(msg, e); |
|
114 |
} |
|
115 |
|
|
116 |
} |
|
117 |
|
|
118 |
private void writeData(String data, String destination) throws Exception { |
|
119 |
|
|
120 |
FSDataOutputStream fin = null; |
|
121 |
try { |
|
122 |
|
|
123 |
data = data.replaceAll("\n\n", ""); |
|
124 |
data = data.substring(0, data.lastIndexOf("\n")); |
|
125 |
if (data.endsWith("\n")) { |
|
126 |
data = data.substring(0, data.lastIndexOf("\n")); |
|
127 |
} |
|
128 |
log.info("***********************Writing data:***********************\n" + data); |
|
129 |
log.info("*********************** data:***********************\n"); |
|
130 |
FileSystem fs = FileSystem.get(new Configuration()); |
|
131 |
fin = fs.create(new Path(destination), true); |
|
132 |
|
|
133 |
fin.writeUTF(data); |
|
134 |
|
|
135 |
} catch (Exception e) { |
|
136 |
log.error("Failed to write exported data to a file : ", e); |
|
137 |
throw new Exception("Failed to write exported data to a file : " + e.toString(), e); |
|
138 |
|
|
139 |
} finally { |
|
140 |
|
|
141 |
fin.close(); |
|
142 |
|
|
143 |
} |
|
144 |
} |
|
145 |
|
|
146 |
public String getOutputPath() { |
|
147 |
return outputPath; |
|
148 |
} |
|
149 |
|
|
150 |
public void setOutputPath(String outputPath) { |
|
151 |
this.outputPath = outputPath; |
|
152 |
} |
|
153 |
|
|
154 |
} |
Also available in: Unified diff
updated xslt for concept to ommit newlines that were causing an issue with FET concept entries