1 |
1 |
package eu.dnetlib.msro.openaireplus.workflows.nodes.dhp;
|
|
2 |
|
2 |
3 |
import java.text.SimpleDateFormat;
|
3 |
4 |
import java.util.Date;
|
4 |
5 |
import java.util.Iterator;
|
... | ... | |
7 |
8 |
import java.util.stream.Collectors;
|
8 |
9 |
|
9 |
10 |
import eu.dnetlib.common.logging.DnetLogger;
|
10 |
|
import eu.dnetlib.dhp.collector.worker.model.ApiDescriptor;
|
|
11 |
import eu.dnetlib.dhp.collection.ApiDescriptor;
|
11 |
12 |
import eu.dnetlib.dhp.model.mdstore.Provenance;
|
12 |
13 |
import eu.dnetlib.enabling.datasources.common.Datasource;
|
13 |
14 |
import eu.dnetlib.enabling.datasources.common.DsmNotFoundException;
|
... | ... | |
30 |
31 |
|
31 |
32 |
import javax.annotation.Resource;
|
32 |
33 |
|
|
34 |
import static eu.dnetlib.dhp.common.Constants.*;
|
|
35 |
|
33 |
36 |
public class PrepareEnvCollectHadoopJobNode extends SimpleJobNode {
|
34 |
37 |
|
35 |
38 |
private static final Log log = LogFactory.getLog(PrepareEnvCollectHadoopJobNode.class);
|
... | ... | |
65 |
68 |
*/
|
66 |
69 |
private String metadataEncoding = "XML";
|
67 |
70 |
|
|
71 |
/**
|
|
72 |
* Maximum number of allowed retires before failing
|
|
73 |
*/
|
|
74 |
private int maxNumberOfRetry = 5;
|
|
75 |
|
|
76 |
/**
|
|
77 |
* Delay between request (Milliseconds)
|
|
78 |
*/
|
|
79 |
private int requestDelay = 0;
|
|
80 |
|
|
81 |
/**
|
|
82 |
* Time to wait after a failure before retrying (Seconds)
|
|
83 |
*/
|
|
84 |
private int retryDelay = 60;
|
|
85 |
|
|
86 |
/**
|
|
87 |
* Connect timeout (Seconds)
|
|
88 |
*/
|
|
89 |
private int connectTimeOut = 30;
|
|
90 |
|
|
91 |
/**
|
|
92 |
* Read timeout (Seconds)
|
|
93 |
*/
|
|
94 |
private int readTimeOut = 60;
|
|
95 |
|
68 |
96 |
@Value("${dnet.openaire.dhp.collection.app.path}")
|
69 |
97 |
private String oozieWfPath;
|
70 |
98 |
|
... | ... | |
98 |
126 |
|
99 |
127 |
if (opt.isPresent()) {
|
100 |
128 |
token.getEnv().setAttribute("mdId", getMdId());
|
101 |
|
token.getEnv().setAttribute("collectionMode", getCollectionMode());
|
102 |
|
token.getEnv().setAttribute("metadataEncoding", getMetadataEncoding());
|
103 |
|
token.getEnv().setAttribute("oozieWfPath", getOozieWfPath());
|
104 |
|
token.getEnv().setAttribute("dnetMessageManagerURL", getDnetMessageManagerURL());
|
|
129 |
token.getEnv().setAttribute(COLLECTION_MODE, getCollectionMode());
|
|
130 |
token.getEnv().setAttribute(METADATA_ENCODING, getMetadataEncoding());
|
|
131 |
token.getEnv().setAttribute(OOZIE_WF_PATH, getOozieWfPath());
|
|
132 |
token.getEnv().setAttribute(DNET_MESSAGE_MGR_URL, getDnetMessageManagerURL());
|
105 |
133 |
|
|
134 |
token.getEnv().setAttribute(MAX_NUMBER_OF_RETRY, getMaxNumberOfRetry());
|
|
135 |
token.getEnv().setAttribute(REQUEST_DELAY, getRequestDelay());
|
|
136 |
token.getEnv().setAttribute(CONNECT_TIMEOUT, getConnectTimeOut());
|
|
137 |
token.getEnv().setAttribute(READ_TIMEOUT, getReadTimeOut());
|
|
138 |
|
106 |
139 |
final ApiDescriptor api = opt.get();
|
107 |
140 |
if ("INCREMENTAL".equals(getCollectionMode())) {
|
108 |
141 |
final String fromDate = calculateFromDate(token);
|
... | ... | |
229 |
262 |
this.fromDateOverride = fromDateOverride;
|
230 |
263 |
}
|
231 |
264 |
|
|
265 |
public int getMaxNumberOfRetry() {
|
|
266 |
return maxNumberOfRetry;
|
|
267 |
}
|
|
268 |
|
|
269 |
public void setMaxNumberOfRetry(int maxNumberOfRetry) {
|
|
270 |
this.maxNumberOfRetry = maxNumberOfRetry;
|
|
271 |
}
|
|
272 |
|
|
273 |
public int getRequestDelay() {
|
|
274 |
return requestDelay;
|
|
275 |
}
|
|
276 |
|
|
277 |
public void setRequestDelay(int requestDelay) {
|
|
278 |
this.requestDelay = requestDelay;
|
|
279 |
}
|
|
280 |
|
|
281 |
public int getRetryDelay() {
|
|
282 |
return retryDelay;
|
|
283 |
}
|
|
284 |
|
|
285 |
public void setRetryDelay(int retryDelay) {
|
|
286 |
this.retryDelay = retryDelay;
|
|
287 |
}
|
|
288 |
|
|
289 |
public int getConnectTimeOut() {
|
|
290 |
return connectTimeOut;
|
|
291 |
}
|
|
292 |
|
|
293 |
public void setConnectTimeOut(int connectTimeOut) {
|
|
294 |
this.connectTimeOut = connectTimeOut;
|
|
295 |
}
|
|
296 |
|
|
297 |
public int getReadTimeOut() {
|
|
298 |
return readTimeOut;
|
|
299 |
}
|
|
300 |
|
|
301 |
public void setReadTimeOut(int readTimeOut) {
|
|
302 |
this.readTimeOut = readTimeOut;
|
|
303 |
}
|
|
304 |
|
232 |
305 |
public String getDnetMessageManagerURL() {
|
233 |
306 |
return dnetMessageManagerURL;
|
234 |
307 |
}
|
... | ... | |
236 |
309 |
public void setDnetMessageManagerURL(String dnetMessageManagerURL) {
|
237 |
310 |
this.dnetMessageManagerURL = dnetMessageManagerURL;
|
238 |
311 |
}
|
|
312 |
|
239 |
313 |
}
|
updated hadoop-specific collection workflow parameters: allow to configure the http client behaviour