|
1
|
package eu.dnetlib.msro.openaireplus.workflows.nodes.contexts;
|
|
2
|
|
|
3
|
import java.io.StringReader;
|
|
4
|
import java.lang.reflect.Type;
|
|
5
|
import java.util.Map;
|
|
6
|
import java.util.Map.Entry;
|
|
7
|
|
|
8
|
import com.google.common.base.Function;
|
|
9
|
import com.google.common.collect.Maps;
|
|
10
|
import com.google.common.reflect.TypeToken;
|
|
11
|
import com.google.gson.Gson;
|
|
12
|
import org.apache.commons.lang.StringUtils;
|
|
13
|
import org.apache.commons.logging.Log;
|
|
14
|
import org.apache.commons.logging.LogFactory;
|
|
15
|
import org.dom4j.*;
|
|
16
|
import org.dom4j.io.SAXReader;
|
|
17
|
|
|
18
|
/**
|
|
19
|
* Created by claudio on 01/03/16.
|
|
20
|
*/
|
|
21
|
public class ContextUtils {
|
|
22
|
|
|
23
|
private static final Log log = LogFactory.getLog(ContextUtils.class);
|
|
24
|
|
|
25
|
public static ContextDesc getContext(final Iterable<String> it,
|
|
26
|
final String contextId,
|
|
27
|
final String contextLabel,
|
|
28
|
final String contextType,
|
|
29
|
final String params) throws
|
|
30
|
DocumentException {
|
|
31
|
|
|
32
|
return new ContextUtils().getContextDesc(it, contextId, contextLabel, contextType, getParamMap(params));
|
|
33
|
}
|
|
34
|
|
|
35
|
private static Map<String, String> getParamMap(String jsonMap) {
|
|
36
|
Map<String, String> paramMap = Maps.newHashMap();
|
|
37
|
if (StringUtils.isNotBlank(jsonMap)) {
|
|
38
|
Type mapType = new TypeToken<Map<String, String>>() {
|
|
39
|
}.getType();
|
|
40
|
paramMap = new Gson().fromJson(jsonMap, mapType);
|
|
41
|
}
|
|
42
|
return paramMap;
|
|
43
|
}
|
|
44
|
|
|
45
|
private ContextDesc getContextDesc(final Iterable<String> it,
|
|
46
|
final String contextId,
|
|
47
|
final String contextLabel,
|
|
48
|
final String contextType,
|
|
49
|
final Map<String, String> params) throws
|
|
50
|
DocumentException {
|
|
51
|
final ContextDesc context = new ContextDesc(contextId, contextLabel, contextType, params);
|
|
52
|
final SAXReader reader = new SAXReader();
|
|
53
|
|
|
54
|
for (String s : it) {
|
|
55
|
populateContext(context, reader, s);
|
|
56
|
}
|
|
57
|
return context;
|
|
58
|
}
|
|
59
|
|
|
60
|
private void populateContext(final ContextDesc context, final SAXReader reader, final String s) throws DocumentException {
|
|
61
|
final Document doc = reader.read(new StringReader(s));
|
|
62
|
|
|
63
|
for (Object o : doc.selectNodes("//fundingtree")) {
|
|
64
|
final Element treeNode = (Element) o;
|
|
65
|
|
|
66
|
final String funder = treeNode.valueOf("./funder/id");
|
|
67
|
if (StringUtils.isBlank(funder)) {
|
|
68
|
log.debug("No funder found, skipping population from the following XML: \n" + s);
|
|
69
|
return;
|
|
70
|
}
|
|
71
|
if (!context.getDbEntries().containsKey(funder)) {
|
|
72
|
log.info("Found funder: " + funder);
|
|
73
|
context.getDbEntries().put(funder, "<fundingtree>" + treeNode.selectSingleNode("./funder").asXML() + "</fundingtree>");
|
|
74
|
log.debug("db entry: " + context.getDbEntries().get(funder));
|
|
75
|
}
|
|
76
|
|
|
77
|
final String openaireId = treeNode.valueOf("./*[starts-with(local-name(),'funding_level_')]/id");
|
|
78
|
if (!context.getDbEntries().containsKey(openaireId) && StringUtils.isNotBlank(openaireId)) {
|
|
79
|
log.info("Found funding: " + openaireId);
|
|
80
|
context.getDbEntries().put(openaireId, treeNode.asXML());
|
|
81
|
log.debug("db entry: " + context.getDbEntries().get(openaireId));
|
|
82
|
final Node node0 = treeNode.selectSingleNode(".//funding_level_0");
|
|
83
|
if (node0 != null) {
|
|
84
|
final ContextPart part = calculatePart(node0);
|
|
85
|
if (context.getCategories().containsKey(part.getId())) {
|
|
86
|
for (ContextPart p : part.getParts().values()) {
|
|
87
|
context.getCategories().get(part.getId()).addPart(p);
|
|
88
|
}
|
|
89
|
} else {
|
|
90
|
context.getCategories().put(part.getId(), part);
|
|
91
|
}
|
|
92
|
}
|
|
93
|
}
|
|
94
|
}
|
|
95
|
}
|
|
96
|
|
|
97
|
private ContextPart calculatePart(final Node node) {
|
|
98
|
|
|
99
|
// final String newId = contextId + "::" + StringUtils.substringAfter(node.valueOf("./id"), "::");
|
|
100
|
// ids are built as: nsPrefix :: funderID :: fundingLevel0ID :: etc etc, hence it seems we might not need the contextId parameter.
|
|
101
|
final String newId = StringUtils.substringAfter(node.valueOf("./id"), "::");
|
|
102
|
|
|
103
|
final ContextPart part = new ContextPart(newId, node.valueOf("./description"));
|
|
104
|
|
|
105
|
part.getParams().put("name", node.valueOf("./name"));
|
|
106
|
part.getParams().put("openaireId", node.valueOf("./id"));
|
|
107
|
part.getParams().put("class", node.valueOf("./class"));
|
|
108
|
|
|
109
|
final Element parent = node.getParent() != null ? node.getParent().getParent() : null;
|
|
110
|
|
|
111
|
if (parent != null && parent.getName().startsWith("funding_level_")) {
|
|
112
|
final ContextPart p = calculatePart(parent);
|
|
113
|
part.getParts().put(p.getId(), p);
|
|
114
|
}
|
|
115
|
return part;
|
|
116
|
}
|
|
117
|
|
|
118
|
public static Function<Entry<String, String>, String> getContextRowTransformer() {
|
|
119
|
|
|
120
|
return new Function<Entry<String, String>, String>() {
|
|
121
|
|
|
122
|
private final SAXReader reader = new SAXReader();
|
|
123
|
|
|
124
|
@Override
|
|
125
|
public String apply(final Entry<String, String> e) {
|
|
126
|
try {
|
|
127
|
final Document docFundingPath = reader.read(new StringReader(e.getValue()));
|
|
128
|
|
|
129
|
final Element root = DocumentHelper.createElement("ROWS");
|
|
130
|
|
|
131
|
final Map<String, String> fundingFields = Maps.newLinkedHashMap();
|
|
132
|
fundingFields.put("_dnet_resource_identifier_", e.getKey());
|
|
133
|
fundingFields.put("id", e.getKey());
|
|
134
|
fundingFields.put("path", e.getValue());
|
|
135
|
fundingFields.put("jurisdiction", docFundingPath.valueOf("//funder/jurisdiction"));
|
|
136
|
final String desc = docFundingPath.valueOf("(//*[starts-with(local-name(),'funding_level_')])[1]/description");
|
|
137
|
if (StringUtils.isNotBlank(desc)) {
|
|
138
|
fundingFields.put("description", desc);
|
|
139
|
}
|
|
140
|
final Map<String, String> orgFields = findFunderInfo(docFundingPath);
|
|
141
|
final String orgId = orgFields.get("id");
|
|
142
|
if (StringUtils.isNotBlank(orgId)) {
|
|
143
|
addRow(root, "organizations", orgFields);
|
|
144
|
fundingFields.put("funder", orgId);
|
|
145
|
}
|
|
146
|
|
|
147
|
addRow(root, "fundingpaths", fundingFields);
|
|
148
|
|
|
149
|
if (log.isDebugEnabled()) {
|
|
150
|
log.debug("Db entries: " + root.asXML());
|
|
151
|
}
|
|
152
|
|
|
153
|
return root.asXML();
|
|
154
|
} catch (DocumentException e1) {
|
|
155
|
log.error("Error parsing xml", e1);
|
|
156
|
throw new RuntimeException("Error parsing xml", e1);
|
|
157
|
}
|
|
158
|
}
|
|
159
|
|
|
160
|
private Map<String, String> findFunderInfo(final Document doc) {
|
|
161
|
final Map<String, String> res = Maps.newLinkedHashMap();
|
|
162
|
res.put("_dnet_resource_identifier_", doc.valueOf("//funder/id"));
|
|
163
|
res.put("id", doc.valueOf("//funder/id"));
|
|
164
|
res.put("legalshortname", doc.valueOf("//funder/shortname"));
|
|
165
|
res.put("legalname", doc.valueOf("//funder/name"));
|
|
166
|
return res;
|
|
167
|
}
|
|
168
|
|
|
169
|
private void addRow(final Element root, final String table, final Map<String, String> fields) {
|
|
170
|
final Element row = root.addElement("ROW");
|
|
171
|
row.addAttribute("table", table);
|
|
172
|
for (Map.Entry<String, String> e : fields.entrySet()) {
|
|
173
|
final Element pathField = row.addElement("FIELD");
|
|
174
|
pathField.addAttribute("name", e.getKey());
|
|
175
|
pathField.setText(e.getValue());
|
|
176
|
}
|
|
177
|
}
|
|
178
|
};
|
|
179
|
}
|
|
180
|
|
|
181
|
}
|