| 1 |
27525
|
claudio.at
|
package eu.dnetlib.msro.openaireplus.workflows.nodes;
|
| 2 |
|
|
|
| 3 |
27568
|
claudio.at
|
import java.util.List;
|
| 4 |
|
|
import java.util.NoSuchElementException;
|
| 5 |
27525
|
claudio.at
|
import javax.annotation.Resource;
|
| 6 |
|
|
|
| 7 |
|
|
import com.google.common.base.Joiner;
|
| 8 |
|
|
import com.google.common.base.Splitter;
|
| 9 |
|
|
import com.google.common.collect.Iterables;
|
| 10 |
31423
|
claudio.at
|
import com.google.common.collect.Lists;
|
| 11 |
27525
|
claudio.at
|
import com.googlecode.sarasvati.NodeToken;
|
| 12 |
|
|
import eu.dnetlib.data.hadoop.config.ClusterName;
|
| 13 |
|
|
import eu.dnetlib.data.hadoop.config.ConfigurationEnumerator;
|
| 14 |
27568
|
claudio.at
|
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
|
| 15 |
27525
|
claudio.at
|
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;
|
| 16 |
32798
|
michele.ar
|
import eu.dnetlib.enabling.locators.UniqueServiceLocator;
|
| 17 |
27568
|
claudio.at
|
import eu.dnetlib.msro.rmi.MSROException;
|
| 18 |
27525
|
claudio.at
|
import eu.dnetlib.msro.workflows.nodes.SimpleJobNode;
|
| 19 |
43601
|
claudio.at
|
import org.apache.commons.logging.Log;
|
| 20 |
|
|
import org.apache.commons.logging.LogFactory;
|
| 21 |
|
|
import org.apache.hadoop.conf.Configuration;
|
| 22 |
|
|
import org.springframework.beans.factory.annotation.Required;
|
| 23 |
27525
|
claudio.at
|
|
| 24 |
43601
|
claudio.at
|
public abstract class PrepareIISParamsV2 extends SimpleJobNode {
|
| 25 |
27525
|
claudio.at
|
|
| 26 |
43601
|
claudio.at
|
private static final Log log = LogFactory.getLog(PrepareIISParamsV2.class);
|
| 27 |
32863
|
claudio.at
|
|
| 28 |
27525
|
claudio.at
|
@Resource
|
| 29 |
|
|
protected ConfigurationEnumerator configurationEnumerator;
|
| 30 |
|
|
|
| 31 |
32798
|
michele.ar
|
@Resource
|
| 32 |
|
|
private UniqueServiceLocator serviceLocator;
|
| 33 |
27525
|
claudio.at
|
|
| 34 |
|
|
private String clusterName;
|
| 35 |
|
|
|
| 36 |
|
|
private String clusterParam = "cluster";
|
| 37 |
|
|
|
| 38 |
|
|
private String oozieWfAppPath;
|
| 39 |
|
|
|
| 40 |
|
|
private String oozieWfAppPathParam = "oozie.wf.application.path";
|
| 41 |
|
|
|
| 42 |
|
|
private String xqueryMdStoreService;
|
| 43 |
|
|
|
| 44 |
|
|
private String mdStoreStoreLocationParam = "import_mdstore_service_location";
|
| 45 |
|
|
|
| 46 |
27568
|
claudio.at
|
private String xqueryObjectStoreService;
|
| 47 |
|
|
|
| 48 |
|
|
private String objectStoreLocationParam = "import_content_object_store_location";
|
| 49 |
|
|
|
| 50 |
33390
|
claudio.at
|
private String xqueryIsLookupService;
|
| 51 |
|
|
|
| 52 |
|
|
private String islookupLocationParam = "import_islookup_service_location";
|
| 53 |
|
|
|
| 54 |
48139
|
alessia.ba
|
private String importProjectConceptsContextCSVParam = "import_project_concepts_context_ids_csv";
|
| 55 |
33390
|
claudio.at
|
|
| 56 |
48139
|
alessia.ba
|
private String importProjectConceptsContextCSV;
|
| 57 |
33390
|
claudio.at
|
|
| 58 |
27568
|
claudio.at
|
private String xqueryDatasetStore;
|
| 59 |
|
|
|
| 60 |
29945
|
claudio.at
|
private String mdStoreDatasetParam = "import_dataset_mdstore_ids_csv";
|
| 61 |
27568
|
claudio.at
|
|
| 62 |
31423
|
claudio.at
|
private String objectStoreBlacklistCSV = "";
|
| 63 |
|
|
|
| 64 |
28824
|
claudio.at
|
protected void prepare(final NodeToken token) throws Exception {
|
| 65 |
27525
|
claudio.at
|
|
| 66 |
|
|
token.getEnv().setAttribute(getClusterParam(), getClusterName());
|
| 67 |
|
|
|
| 68 |
|
|
// Assumes we only have one mdStore service instance
|
| 69 |
27568
|
claudio.at
|
token.getEnv().setAttribute(getMdStoreStoreLocationParam(), getServiceEndpoint(getXqueryMdStoreService()));
|
| 70 |
|
|
// Assumes we only have one objectStore service instance
|
| 71 |
|
|
token.getEnv().setAttribute(getObjectStoreLocationParam(), getServiceEndpoint(getXqueryObjectStoreService()));
|
| 72 |
27525
|
claudio.at
|
|
| 73 |
33390
|
claudio.at
|
token.getEnv().setAttribute(getIslookupLocationParam(), getServiceEndpoint(getXqueryIsLookupService()));
|
| 74 |
48139
|
alessia.ba
|
token.getEnv().setAttribute(getImportProjectConceptsContextCSVParam(), getImportProjectConceptsContextCSV());
|
| 75 |
33390
|
claudio.at
|
|
| 76 |
27525
|
claudio.at
|
Configuration conf = configurationEnumerator.get(ClusterName.valueOf(getClusterName()));
|
| 77 |
|
|
String nameNode = conf.get("fs.defaultFS");
|
| 78 |
|
|
|
| 79 |
|
|
token.getEnv().setAttribute(getOozieWfAppPathParam(), getURI(nameNode, getOozieWfAppPath()));
|
| 80 |
31423
|
claudio.at
|
token.getEnv().setAttribute(getMdStoreDatasetParam(), asCSV(getProfileIds(getXqueryDatasetStore())));
|
| 81 |
27568
|
claudio.at
|
}
|
| 82 |
27525
|
claudio.at
|
|
| 83 |
28824
|
claudio.at
|
protected String getServiceEndpoint(final String xquery) throws MSROException {
|
| 84 |
27568
|
claudio.at
|
try {
|
| 85 |
32798
|
michele.ar
|
return Iterables.getOnlyElement(serviceLocator.getService(ISLookUpService.class).quickSearchProfile(xquery));
|
| 86 |
27568
|
claudio.at
|
} catch (ISLookUpException e) {
|
| 87 |
|
|
throw new MSROException("unable to fetch service endpoint", e);
|
| 88 |
|
|
} catch (NoSuchElementException e) {
|
| 89 |
|
|
throw new MSROException("unable to find service endpoint, xquery: " + getXqueryMdStoreService(), e);
|
| 90 |
|
|
} catch (IllegalArgumentException e) {
|
| 91 |
|
|
throw new MSROException("more than one services found, we assume to have only one available", e);
|
| 92 |
|
|
}
|
| 93 |
27525
|
claudio.at
|
}
|
| 94 |
|
|
|
| 95 |
28824
|
claudio.at
|
protected String getProfileId(final String xquery) throws MSROException {
|
| 96 |
27568
|
claudio.at
|
try {
|
| 97 |
32798
|
michele.ar
|
return Iterables.getOnlyElement(serviceLocator.getService(ISLookUpService.class).quickSearchProfile(xquery));
|
| 98 |
27568
|
claudio.at
|
} catch (ISLookUpException e) {
|
| 99 |
|
|
throw new MSROException("unable to fetch profile id", e);
|
| 100 |
|
|
} catch (NoSuchElementException e) {
|
| 101 |
|
|
throw new MSROException("unable to find profile profile, xquery: " + xquery, e);
|
| 102 |
|
|
} catch (IllegalArgumentException e) {
|
| 103 |
|
|
throw new MSROException("more than one profile profiles was found, we assume to have only one available, xquery: " + xquery, e);
|
| 104 |
|
|
}
|
| 105 |
|
|
}
|
| 106 |
|
|
|
| 107 |
31423
|
claudio.at
|
protected List<String> getProfileIds(final String xquery) throws MSROException {
|
| 108 |
27568
|
claudio.at
|
try {
|
| 109 |
32798
|
michele.ar
|
List<String> ids = serviceLocator.getService(ISLookUpService.class).quickSearchProfile(xquery);
|
| 110 |
27568
|
claudio.at
|
|
| 111 |
32863
|
claudio.at
|
if (ids.isEmpty()) {
|
| 112 |
|
|
log.warn("couldn't find any profile, xquery: " + xquery);
|
| 113 |
|
|
}
|
| 114 |
27568
|
claudio.at
|
|
| 115 |
31423
|
claudio.at
|
return ids;
|
| 116 |
27568
|
claudio.at
|
} catch (ISLookUpException e) {
|
| 117 |
|
|
throw new MSROException("unable to fetch profile ids, x query: " + xquery, e);
|
| 118 |
|
|
}
|
| 119 |
|
|
}
|
| 120 |
|
|
|
| 121 |
31423
|
claudio.at
|
protected String getFilteredObjectStoreCSV(final String xquery) throws MSROException {
|
| 122 |
|
|
return asCSV(filter(getProfileIds(xquery), asList(getObjectStoreBlacklistCSV())));
|
| 123 |
|
|
}
|
| 124 |
|
|
|
| 125 |
|
|
protected List<String> filter(final List<String> list, final List<String> filter) {
|
| 126 |
32798
|
michele.ar
|
if (filter == null || filter.isEmpty()) { return list; }
|
| 127 |
31423
|
claudio.at
|
list.removeAll(filter);
|
| 128 |
|
|
return list;
|
| 129 |
|
|
}
|
| 130 |
|
|
|
| 131 |
|
|
protected String asCSV(final List<String> list) {
|
| 132 |
|
|
return Joiner.on(",").skipNulls().join(list);
|
| 133 |
|
|
}
|
| 134 |
|
|
|
| 135 |
|
|
protected List<String> asList(final String csv) {
|
| 136 |
|
|
return Lists.newArrayList(Splitter.on(",").trimResults().omitEmptyStrings().split(csv));
|
| 137 |
|
|
}
|
| 138 |
|
|
|
| 139 |
28824
|
claudio.at
|
private String getURI(final String nameNode, final String relative) {
|
| 140 |
27525
|
claudio.at
|
// TODO ensure to return a valid URI
|
| 141 |
|
|
return nameNode + relative;
|
| 142 |
|
|
}
|
| 143 |
|
|
|
| 144 |
28824
|
claudio.at
|
private String getZkQuorumCSV(final Configuration conf, final String zkPort) {
|
| 145 |
27525
|
claudio.at
|
return Joiner.on(":" + zkPort + ",").join(Splitter.on(",").omitEmptyStrings().split(conf.get("hbase.zookeeper.quorum")));
|
| 146 |
|
|
}
|
| 147 |
|
|
|
| 148 |
|
|
@Required
|
| 149 |
28824
|
claudio.at
|
public void setXqueryMdStoreService(final String xqueryMdStoreService) {
|
| 150 |
27525
|
claudio.at
|
this.xqueryMdStoreService = xqueryMdStoreService;
|
| 151 |
|
|
}
|
| 152 |
|
|
|
| 153 |
|
|
public String getXqueryMdStoreService() {
|
| 154 |
|
|
return xqueryMdStoreService;
|
| 155 |
|
|
}
|
| 156 |
|
|
|
| 157 |
|
|
public String getMdStoreStoreLocationParam() {
|
| 158 |
|
|
return mdStoreStoreLocationParam;
|
| 159 |
|
|
}
|
| 160 |
|
|
|
| 161 |
28824
|
claudio.at
|
public void setMdStoreStoreLocationParam(final String mdStoreStoreLocationParam) {
|
| 162 |
27525
|
claudio.at
|
this.mdStoreStoreLocationParam = mdStoreStoreLocationParam;
|
| 163 |
|
|
}
|
| 164 |
|
|
|
| 165 |
|
|
public String getClusterName() {
|
| 166 |
|
|
return clusterName;
|
| 167 |
|
|
}
|
| 168 |
|
|
|
| 169 |
28824
|
claudio.at
|
public void setClusterName(final String clusterName) {
|
| 170 |
27525
|
claudio.at
|
this.clusterName = clusterName;
|
| 171 |
|
|
}
|
| 172 |
|
|
|
| 173 |
|
|
public String getClusterParam() {
|
| 174 |
|
|
return clusterParam;
|
| 175 |
|
|
}
|
| 176 |
|
|
|
| 177 |
28824
|
claudio.at
|
public void setClusterParam(final String clusterParam) {
|
| 178 |
27525
|
claudio.at
|
this.clusterParam = clusterParam;
|
| 179 |
|
|
}
|
| 180 |
|
|
|
| 181 |
|
|
public String getOozieWfAppPathParam() {
|
| 182 |
|
|
return oozieWfAppPathParam;
|
| 183 |
|
|
}
|
| 184 |
|
|
|
| 185 |
28824
|
claudio.at
|
public void setOozieWfAppPathParam(final String oozieWfAppPathParam) {
|
| 186 |
27525
|
claudio.at
|
this.oozieWfAppPathParam = oozieWfAppPathParam;
|
| 187 |
|
|
}
|
| 188 |
|
|
|
| 189 |
|
|
public String getOozieWfAppPath() {
|
| 190 |
|
|
return oozieWfAppPath;
|
| 191 |
|
|
}
|
| 192 |
|
|
|
| 193 |
28824
|
claudio.at
|
public void setOozieWfAppPath(final String oozieWfAppPath) {
|
| 194 |
27525
|
claudio.at
|
this.oozieWfAppPath = oozieWfAppPath;
|
| 195 |
|
|
}
|
| 196 |
|
|
|
| 197 |
27568
|
claudio.at
|
@Required
|
| 198 |
|
|
public String getXqueryDatasetStore() {
|
| 199 |
|
|
return xqueryDatasetStore;
|
| 200 |
|
|
}
|
| 201 |
|
|
|
| 202 |
28824
|
claudio.at
|
public void setXqueryDatasetStore(final String xqueryDatasetStore) {
|
| 203 |
27568
|
claudio.at
|
this.xqueryDatasetStore = xqueryDatasetStore;
|
| 204 |
|
|
}
|
| 205 |
|
|
|
| 206 |
|
|
public String getMdStoreDatasetParam() {
|
| 207 |
|
|
return mdStoreDatasetParam;
|
| 208 |
|
|
}
|
| 209 |
|
|
|
| 210 |
28824
|
claudio.at
|
public void setMdStoreDatasetParam(final String mdStoreDatasetParam) {
|
| 211 |
27568
|
claudio.at
|
this.mdStoreDatasetParam = mdStoreDatasetParam;
|
| 212 |
|
|
}
|
| 213 |
|
|
|
| 214 |
|
|
public String getXqueryObjectStoreService() {
|
| 215 |
|
|
return xqueryObjectStoreService;
|
| 216 |
|
|
}
|
| 217 |
|
|
|
| 218 |
|
|
@Required
|
| 219 |
28824
|
claudio.at
|
public void setXqueryObjectStoreService(final String xqueryObjectStoreService) {
|
| 220 |
27568
|
claudio.at
|
this.xqueryObjectStoreService = xqueryObjectStoreService;
|
| 221 |
|
|
}
|
| 222 |
|
|
|
| 223 |
|
|
public String getObjectStoreLocationParam() {
|
| 224 |
|
|
return objectStoreLocationParam;
|
| 225 |
|
|
}
|
| 226 |
|
|
|
| 227 |
28824
|
claudio.at
|
public void setObjectStoreLocationParam(final String objectStoreLocationParam) {
|
| 228 |
27568
|
claudio.at
|
this.objectStoreLocationParam = objectStoreLocationParam;
|
| 229 |
|
|
}
|
| 230 |
|
|
|
| 231 |
31423
|
claudio.at
|
public String getObjectStoreBlacklistCSV() {
|
| 232 |
|
|
return objectStoreBlacklistCSV;
|
| 233 |
|
|
}
|
| 234 |
|
|
|
| 235 |
|
|
public void setObjectStoreBlacklistCSV(final String objectStoreBlacklistCSV) {
|
| 236 |
|
|
this.objectStoreBlacklistCSV = objectStoreBlacklistCSV;
|
| 237 |
|
|
}
|
| 238 |
|
|
|
| 239 |
33390
|
claudio.at
|
public String getXqueryIsLookupService() {
|
| 240 |
|
|
return xqueryIsLookupService;
|
| 241 |
|
|
}
|
| 242 |
|
|
|
| 243 |
|
|
@Required
|
| 244 |
|
|
public void setXqueryIsLookupService(final String xqueryIsLookupService) {
|
| 245 |
|
|
this.xqueryIsLookupService = xqueryIsLookupService;
|
| 246 |
|
|
}
|
| 247 |
|
|
|
| 248 |
|
|
public String getIslookupLocationParam() {
|
| 249 |
|
|
return islookupLocationParam;
|
| 250 |
|
|
}
|
| 251 |
|
|
|
| 252 |
|
|
public void setIslookupLocationParam(final String islookupLocationParam) {
|
| 253 |
|
|
this.islookupLocationParam = islookupLocationParam;
|
| 254 |
|
|
}
|
| 255 |
|
|
|
| 256 |
48139
|
alessia.ba
|
public String getImportProjectConceptsContextCSVParam() {
|
| 257 |
|
|
return importProjectConceptsContextCSVParam;
|
| 258 |
33390
|
claudio.at
|
}
|
| 259 |
|
|
|
| 260 |
48139
|
alessia.ba
|
public void setImportProjectConceptsContextCSVParam(final String importProjectConceptsContextCSVParam) {
|
| 261 |
|
|
this.importProjectConceptsContextCSVParam = importProjectConceptsContextCSVParam;
|
| 262 |
33390
|
claudio.at
|
}
|
| 263 |
|
|
|
| 264 |
48139
|
alessia.ba
|
public String getImportProjectConceptsContextCSV() {
|
| 265 |
|
|
return importProjectConceptsContextCSV;
|
| 266 |
33390
|
claudio.at
|
}
|
| 267 |
|
|
|
| 268 |
48139
|
alessia.ba
|
public void setImportProjectConceptsContextCSV(final String importProjectConceptsContextCSV) {
|
| 269 |
|
|
this.importProjectConceptsContextCSV = importProjectConceptsContextCSV;
|
| 270 |
33390
|
claudio.at
|
}
|
| 271 |
27525
|
claudio.at
|
}
|