1
|
package eu.dnetlib.openaire.directindex.objects;
|
2
|
|
3
|
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
|
4
|
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;
|
5
|
import eu.dnetlib.miscutils.datetime.DateUtils;
|
6
|
import eu.dnetlib.miscutils.functional.hash.Hashing;
|
7
|
import eu.dnetlib.miscutils.functional.string.EscapeXml;
|
8
|
import eu.dnetlib.openaire.directindex.api.DirecIndexApiException;
|
9
|
import eu.dnetlib.openaire.directindex.api.OpenAIRESubmitterUtils;
|
10
|
import org.apache.commons.lang.StringUtils;
|
11
|
import org.apache.commons.logging.Log;
|
12
|
import org.apache.commons.logging.LogFactory;
|
13
|
import org.apache.velocity.app.VelocityEngine;
|
14
|
import org.springframework.ui.velocity.VelocityEngineUtils;
|
15
|
|
16
|
import java.text.SimpleDateFormat;
|
17
|
import java.util.Date;
|
18
|
import java.util.HashMap;
|
19
|
import java.util.List;
|
20
|
import java.util.Map;
|
21
|
import java.util.concurrent.TimeUnit;
|
22
|
|
23
|
public class ResultEntryToOaf {
|
24
|
|
25
|
private static final Log log = LogFactory.getLog(ResultEntryToOaf.class);
|
26
|
|
27
|
private static long last_cache_update = 0;
|
28
|
private static final Map<String, Map<String, String>> cached_vocabularies = new HashMap<>();
|
29
|
private static final Map<String, DatasourceEntry> cached_datasources = new HashMap<>();
|
30
|
private static final Map<String, String> cached_contexts = new HashMap<>();
|
31
|
|
32
|
private OpenAIRESubmitterUtils utils;
|
33
|
|
34
|
public String asOafRecord(final ResultEntry entry, final VelocityEngine ve,
|
35
|
final ISLookUpService lookupService,
|
36
|
final String oafSchemaLocation,
|
37
|
final String community_api) throws Exception {
|
38
|
|
39
|
if (StringUtils.isBlank(entry.getOriginalId())
|
40
|
&& StringUtils
|
41
|
.isBlank(entry.getOpenaireId())) { throw new DirecIndexApiException("One of the following fields is required: originalId or openaireId"); }
|
42
|
if (StringUtils.isBlank(entry.getTitle())) { throw new DirecIndexApiException("A required field is missing: title"); }
|
43
|
if (StringUtils.isBlank(entry.getUrl())) { throw new DirecIndexApiException("A required field is missing: url"); }
|
44
|
if (StringUtils.isBlank(entry.getLicenseCode()) && StringUtils.isBlank(entry.getAccessRightCode())) { throw new DirecIndexApiException("A required field is missing: accessRightCode"); }
|
45
|
if (StringUtils.isBlank(entry.getResourceType())) { throw new DirecIndexApiException("A required field is missing: resourceType"); }
|
46
|
if (StringUtils.isBlank(entry.getCollectedFromId())) { throw new DirecIndexApiException("A required field is missing: collectedFromId"); }
|
47
|
if (StringUtils.isBlank(entry.getType())) { throw new DirecIndexApiException("A required field is missing: type"); }
|
48
|
|
49
|
final DatasourceEntry collectedFromEntry = getDatasourceInfo(entry.getCollectedFromId(), lookupService);
|
50
|
final DatasourceEntry hostedByEntry = getDatasourceInfo(entry.getHostedById(), lookupService);
|
51
|
|
52
|
if (StringUtils.isBlank(entry.getOpenaireId())) {
|
53
|
entry.setOpenaireId(calculateOpenaireId(entry.getOriginalId(), collectedFromEntry));
|
54
|
}
|
55
|
|
56
|
if (!entry.getOpenaireId()
|
57
|
.matches("^\\w{12}::\\w{32}$")) { throw new DirecIndexApiException(
|
58
|
"Invalid openaireId: " + entry.getOpenaireId() + " - regex ^\\w{12}::\\w{32}$ not matched"); }
|
59
|
|
60
|
|
61
|
final List<OpenAIRESubmitterUtils.ContextInfo> contextInfos = utils.processContexts(entry.getContexts(), getContexts(lookupService));
|
62
|
|
63
|
final Map<String, Object> model = new HashMap<>();
|
64
|
model.put("esc", new EscapeXml());
|
65
|
model.put("util",utils);
|
66
|
model.put("pub", this);
|
67
|
model.put("objIdentifier", entry.getOpenaireId());
|
68
|
model.put("oafSchemaLocation", oafSchemaLocation);
|
69
|
model.put("resultTypes", getVocabulary("dnet:result_typologies", lookupService));
|
70
|
model.put("rights", getVocabulary("dnet:access_modes", lookupService));
|
71
|
model.put("resourceTypes", getVocabulary("dnet:publication_resource", lookupService));
|
72
|
model.put("pidTypes", getVocabulary("dnet:pid_types", lookupService));
|
73
|
model.put("languages", getVocabulary("dnet:languages", lookupService));
|
74
|
model.put("contexts", getContexts(lookupService));
|
75
|
model.put("contextInfo",contextInfos);
|
76
|
model.put("dateOfCollection", new SimpleDateFormat("yyyy-MM-dd\'T\'hh:mm:ss\'Z\'").format(new Date()));
|
77
|
model.put("collectedFrom", collectedFromEntry);
|
78
|
model.put("hostedBy", hostedByEntry);
|
79
|
|
80
|
return VelocityEngineUtils.mergeTemplateIntoString(ve, "/eu/dnetlib/openaire/directindex/indexRecord.xml.vm", "UTF-8", model);
|
81
|
}
|
82
|
|
83
|
private synchronized static DatasourceEntry getDatasourceInfo(final String dsId, final ISLookUpService lookupService) throws ISLookUpException {
|
84
|
if (StringUtils
|
85
|
.isBlank(dsId)) { return new DatasourceEntry("openaire____::1256f046-bf1f-4afc-8b47-d0b147148b18", "Unknown Repository", "unknown_____"); }
|
86
|
|
87
|
if (!cached_datasources.containsKey(dsId)) {
|
88
|
final String query =
|
89
|
"collection('/db/DRIVER/RepositoryServiceResources/RepositoryServiceResourceType')//CONFIGURATION[./DATASOURCE_ORIGINAL_ID='" + dsId
|
90
|
+ "']/concat(./OFFICIAL_NAME, ' @@@ ', .//FIELD/value[../key='NamespacePrefix'])";
|
91
|
final String s = lookupService.getResourceProfileByQuery(query);
|
92
|
final String[] arr = s.split("@@@");
|
93
|
|
94
|
final DatasourceEntry ds = new DatasourceEntry(dsId, arr[0].trim(), arr[1].trim());
|
95
|
|
96
|
if (StringUtils.isBlank(ds.getName()) || StringUtils.isBlank(ds.getPrefix())) {
|
97
|
log.error("Invalid datasource id: " + dsId);
|
98
|
throw new ISLookUpException("Invalid datasource id: " + dsId);
|
99
|
} else {
|
100
|
cached_datasources.put(dsId, ds);
|
101
|
}
|
102
|
}
|
103
|
|
104
|
return cached_datasources.get(dsId);
|
105
|
|
106
|
}
|
107
|
|
108
|
private synchronized static Map<String, String> getVocabulary(final String voc, final ISLookUpService lookupService) throws ISLookUpException {
|
109
|
|
110
|
if (DateUtils.now() - last_cache_update < TimeUnit.MINUTES.toMillis(15) && cached_vocabularies.containsKey(voc)) {
|
111
|
return cached_vocabularies.get(voc);
|
112
|
} else {
|
113
|
final String query = "collection('/db/DRIVER/VocabularyDSResources/VocabularyDSResourceType')[.//VOCABULARY_NAME/@code='" + voc
|
114
|
+ "']//TERM/concat(@code, ' @@@ ', @english_name)";
|
115
|
|
116
|
final Map<String, String> map = new HashMap<>();
|
117
|
for (final String s : lookupService.quickSearchProfile(query)) {
|
118
|
final String[] arr = s.split("@@@");
|
119
|
map.put(arr[0].trim(), arr[1].trim());
|
120
|
}
|
121
|
|
122
|
cached_vocabularies.put(voc, map);
|
123
|
|
124
|
last_cache_update = DateUtils.now();
|
125
|
|
126
|
return map;
|
127
|
}
|
128
|
}
|
129
|
|
130
|
private synchronized static Map<String, String> getContexts(final ISLookUpService lookupService) throws ISLookUpException {
|
131
|
if (DateUtils.now() - last_cache_update > TimeUnit.MINUTES.toMillis(15) || cached_contexts.isEmpty()) {
|
132
|
final String query =
|
133
|
"collection('/db/DRIVER/ContextDSResources/ContextDSResourceType')[.//context/@type='community' or .//context/@type='ri']//*[name()='context' or name()='category' or name()='concept']/concat(@id, ' @@@ ', @label)";
|
134
|
|
135
|
cached_contexts.clear();
|
136
|
for (final String s : lookupService.quickSearchProfile(query)) {
|
137
|
final String[] arr = s.split("@@@");
|
138
|
cached_contexts.put(arr[0].trim(), arr[1].trim());
|
139
|
}
|
140
|
last_cache_update = DateUtils.now();
|
141
|
}
|
142
|
return cached_contexts;
|
143
|
}
|
144
|
|
145
|
private static String calculateOpenaireId(final String originalId, final DatasourceEntry collectedFromEntry) {
|
146
|
return collectedFromEntry.getPrefix() + "::" + Hashing.md5(originalId);
|
147
|
}
|
148
|
|
149
|
public static String calculateOpenaireId(final String originalId, final String collectedFromId, final ISLookUpService lookupService)
|
150
|
throws ISLookUpException {
|
151
|
return calculateOpenaireId(originalId, getDatasourceInfo(collectedFromId, lookupService));
|
152
|
}
|
153
|
public OpenAIRESubmitterUtils getUtils() {
|
154
|
return utils;
|
155
|
}
|
156
|
|
157
|
public void setUtils(OpenAIRESubmitterUtils utils) {
|
158
|
this.utils = utils;
|
159
|
}
|
160
|
}
|