1
|
package eu.dnetlib.msro.openaireplus.workflows.nodes.contexts;
|
2
|
|
3
|
import java.io.IOException;
|
4
|
import java.util.*;
|
5
|
|
6
|
import com.google.common.collect.Maps;
|
7
|
import com.ximpleware.AutoPilot;
|
8
|
import com.ximpleware.VTDException;
|
9
|
import com.ximpleware.VTDGen;
|
10
|
import com.ximpleware.VTDNav;
|
11
|
import org.apache.commons.logging.Log;
|
12
|
import org.apache.commons.logging.LogFactory;
|
13
|
import org.dom4j.Document;
|
14
|
import org.dom4j.DocumentHelper;
|
15
|
import org.dom4j.Element;
|
16
|
import org.joda.time.DateTime;
|
17
|
|
18
|
import static eu.dnetlib.data.transform.xml.AbstractDNetXsltFunctions.md5;
|
19
|
|
20
|
public class BuildH2020FETTaxonomy {
|
21
|
|
22
|
private static final Log log = LogFactory.getLog(BuildH2020FETTaxonomy.class);
|
23
|
|
24
|
private Iterator<String> projects;
|
25
|
private Map<String, Map<String, Map<String, List<Info>>>> dic;
|
26
|
|
27
|
public BuildH2020FETTaxonomy() {
|
28
|
dic = Maps.newHashMap();
|
29
|
}
|
30
|
|
31
|
public void setIterator(Iterator<String> projects) {
|
32
|
this.projects = projects;
|
33
|
}
|
34
|
|
35
|
public String parseProjects() throws Exception {
|
36
|
while (projects.hasNext())
|
37
|
parse(projects.next());
|
38
|
return getTaxonomy();
|
39
|
}
|
40
|
|
41
|
public String getTaxonomy() throws Exception {
|
42
|
Document taxonomy = DocumentHelper.createDocument();
|
43
|
Element root1 = taxonomy.addElement("RESOURCE_PROFILE");
|
44
|
createHeader(root1);
|
45
|
Element body = root1.addElement("BODY");
|
46
|
Element configurationElement = body.addElement("CONFIGURATION");
|
47
|
createContext(configurationElement);
|
48
|
body.addElement("STATUS");
|
49
|
body.addElement("SECURITY_PARAMETERS");
|
50
|
return taxonomy.getDocument().asXML();
|
51
|
|
52
|
}
|
53
|
|
54
|
private void createContext(final Element configurationElement) {
|
55
|
Element context = configurationElement.addElement("context");
|
56
|
context.addAttribute("type", "community");
|
57
|
context.addAttribute("id", "fet-h2020");
|
58
|
context.addAttribute("label", "FET H2020");
|
59
|
for (Map.Entry<String, Map<String, Map<String, List<Info>>>> entry : dic.entrySet()) {
|
60
|
String type = entry.getKey();
|
61
|
Element cat = context.addElement("category");
|
62
|
String catid = "fet-h2020::" + type.toLowerCase();
|
63
|
cat.addAttribute("id", catid);
|
64
|
cat.addAttribute("label", "FET " + type);
|
65
|
cat.addAttribute("claim", "false");
|
66
|
//int count = 0;
|
67
|
for (Map.Entry<String, Map<String, List<Info>>> entryst : dic.get(type).entrySet()) {
|
68
|
String subtype = entryst.getKey();
|
69
|
for (Map.Entry<String, List<Info>> entryp : dic.get(type).get(subtype).entrySet()) {
|
70
|
String opt = entryp.getKey();
|
71
|
// count++;
|
72
|
Element subcat = cat.addElement("concept");
|
73
|
String proj_md5 = md5(opt).substring(0,12);
|
74
|
//String subcatid = catid + "::" + count;
|
75
|
String subcatid = catid + "::" + proj_md5;
|
76
|
subcat.addAttribute("id", subcatid);
|
77
|
Element pa = subcat.addElement("param");
|
78
|
pa.addAttribute("name", "CALL_ID");
|
79
|
List<Info> projectRows = dic.get(type).get(subtype).get(opt);
|
80
|
if (!projectRows.isEmpty()) {
|
81
|
for (Info row : projectRows) {
|
82
|
createProjectConcept(row, subcat, subcatid);
|
83
|
}
|
84
|
subcat.addAttribute("label", projectRows.get(0).getCallname());
|
85
|
subcat.addAttribute("claim", "false");
|
86
|
pa.setText(projectRows.get(0).getCallID());
|
87
|
} else throw new RuntimeException("The lis of projects for " + subcatid + " is empty. This exception should never be thrown.");
|
88
|
}
|
89
|
}
|
90
|
}
|
91
|
}
|
92
|
|
93
|
private void createProjectConcept(Info row, Element father, String prefix) {
|
94
|
String code = row.getCode();
|
95
|
String acronym = row.getAcronym();
|
96
|
String title = row.getTitle();
|
97
|
String projcallid = row.getProjcallid();
|
98
|
Element concept = father.addElement("concept");
|
99
|
concept.addAttribute("id", prefix + "::" + code);
|
100
|
concept.addAttribute("label", title);
|
101
|
concept.addAttribute("claim", "true");
|
102
|
Element p = concept.addElement("param");
|
103
|
p.addAttribute("name", "CD_PROJ_ID");
|
104
|
p.setText(code);
|
105
|
p = concept.addElement("param");
|
106
|
p.addAttribute("name", "CD_CALL_ID");
|
107
|
p.setText(projcallid);
|
108
|
p = concept.addElement("param");
|
109
|
p.addAttribute("name", "CD_PROJECT_NUMBER");
|
110
|
p.setText(code);
|
111
|
p = concept.addElement("param");
|
112
|
p.addAttribute("name", "CD_ACRONYM");
|
113
|
p.setText(acronym);
|
114
|
p = concept.addElement("param");
|
115
|
p.addAttribute("name", "CD_FRAMEWORK");
|
116
|
p.setText("H2020");
|
117
|
p = concept.addElement("param");
|
118
|
p.addAttribute("name", "funder");
|
119
|
p.setText("EC");
|
120
|
}
|
121
|
|
122
|
private void createHeader(final Element profile) {
|
123
|
Element header = profile.addElement("HEADER");
|
124
|
Element rs = header.addElement("RESOURCE_IDENTIFIER");
|
125
|
rs.addAttribute("value", "");
|
126
|
Element rt = header.addElement("RESOURCE_TYPE");
|
127
|
rt.addAttribute("value", "ContextDSResourceType");
|
128
|
Element rk = header.addElement("RESOURCE_KIND");
|
129
|
rk.addAttribute("value", "ContextDSResources");
|
130
|
Element ru = header.addElement("RESOURCE_URI");
|
131
|
ru.addAttribute("value", "");
|
132
|
Element daoc = header.addElement("DATE_OF_CREATION");
|
133
|
daoc.addAttribute("value", DateTime.now().toString());
|
134
|
}
|
135
|
|
136
|
private boolean parse(String project) throws VTDException, IOException {
|
137
|
boolean fet = false;
|
138
|
String callid = "";
|
139
|
String projectid = "";
|
140
|
VTDGen vg = new VTDGen();
|
141
|
vg.setDoc(project.getBytes("UTF-8"));
|
142
|
vg.parse(false);
|
143
|
VTDNav vn = vg.getNav();
|
144
|
AutoPilot ap = new AutoPilot(vn);
|
145
|
ap.selectXPath("//metadata/ROWS/ROW[@table=\"projects\"]");
|
146
|
|
147
|
while (ap.evalXPath() != -1) {
|
148
|
AutoPilot ap1 = new AutoPilot(vn);
|
149
|
ap1.selectXPath("./FIELD[@name=\"call_identifier\"]");
|
150
|
if (ap1.evalXPath() != -1)
|
151
|
callid = vn.toNormalizedString(vn.getText());
|
152
|
ap1.clearVariableExprs();
|
153
|
vn.toElement(VTDNav.PARENT);
|
154
|
ap1.selectXPath("./FIELD[@name=\"optional1\"]");
|
155
|
if (ap1.evalXPath() != -1)
|
156
|
projectid = vn.toNormalizedString(vn.getText());
|
157
|
if (callid.contains("FET")) {
|
158
|
String type;
|
159
|
if (callid.contains("OPEN"))
|
160
|
type = "OPEN";
|
161
|
else if (callid.contains("PROACT"))
|
162
|
type = "PROACT";
|
163
|
else if (callid.contains("HPC"))
|
164
|
type = "HPC";
|
165
|
else
|
166
|
type = "FLAG";
|
167
|
vn.toElement(VTDNav.ROOT);
|
168
|
fet = true;
|
169
|
insert(type, callid, projectid, new Info(vn));
|
170
|
}
|
171
|
}
|
172
|
return fet;
|
173
|
}
|
174
|
|
175
|
private void insert(String type, String callid, String projcallid, Info row) {
|
176
|
Map<String, Map<String, List<Info>>> dopen;
|
177
|
Map<String, List<Info>> entry;
|
178
|
List<Info> projects;
|
179
|
if (dic.containsKey(type)) {
|
180
|
dopen = dic.get(type);
|
181
|
if (dopen.containsKey(callid)) {
|
182
|
entry = dopen.get(callid);
|
183
|
if (entry.containsKey(projcallid)) {
|
184
|
entry.get(projcallid).add(row);
|
185
|
} else {
|
186
|
projects = new ArrayList<>();
|
187
|
projects.add(row);
|
188
|
entry.put(projcallid, projects);
|
189
|
}
|
190
|
} else {
|
191
|
projects = new ArrayList<>();
|
192
|
entry = new HashMap<>();
|
193
|
projects.add(row);
|
194
|
entry.put(projcallid, projects);
|
195
|
dopen.put(callid, entry);
|
196
|
}
|
197
|
} else {
|
198
|
projects = new ArrayList<>();
|
199
|
entry = new HashMap<>();
|
200
|
dopen = new HashMap<>();
|
201
|
projects.add(row);
|
202
|
entry.put(projcallid, projects);
|
203
|
dopen.put(callid, entry);
|
204
|
dic.put(type, dopen);
|
205
|
}
|
206
|
}
|
207
|
|
208
|
public void parseFETProject(int projects_number) throws VTDException, IOException {
|
209
|
int parsed = 0;
|
210
|
while (projects.hasNext() && parsed < projects_number) {
|
211
|
if (parse(projects.next()))
|
212
|
parsed++;
|
213
|
}
|
214
|
log.debug(parsed);
|
215
|
}
|
216
|
|
217
|
private class Info {
|
218
|
|
219
|
VTDNav vn;
|
220
|
private String code, acronym, title, call_identifier, projcallid, callname;
|
221
|
|
222
|
public String getCode() {
|
223
|
return code;
|
224
|
}
|
225
|
|
226
|
public String getAcronym() {
|
227
|
return acronym;
|
228
|
}
|
229
|
|
230
|
public String getTitle() {
|
231
|
return title;
|
232
|
}
|
233
|
|
234
|
public String getCallID() {
|
235
|
return call_identifier;
|
236
|
}
|
237
|
|
238
|
public String getProjcallid() {
|
239
|
return projcallid;
|
240
|
}
|
241
|
|
242
|
public String getCallname() {
|
243
|
return callname;
|
244
|
}
|
245
|
|
246
|
public Info(VTDNav vn) {
|
247
|
this.vn = vn;
|
248
|
parse();
|
249
|
}
|
250
|
|
251
|
private void parse() {
|
252
|
try {
|
253
|
AutoPilot ap = new AutoPilot(vn);
|
254
|
ap.selectXPath("//FIELD");
|
255
|
while (ap.evalXPath() != -1) {
|
256
|
String aname = vn.toNormalizedString(vn.getAttrVal("name"));
|
257
|
String text = "";
|
258
|
if (vn.getText() != -1)
|
259
|
text = vn.toNormalizedString(vn.getText());
|
260
|
if (aname.equals("code"))
|
261
|
code = text;
|
262
|
else if (aname.equals("acronym"))
|
263
|
acronym = text;
|
264
|
else if (aname.equals("title"))
|
265
|
title = text;
|
266
|
else if (aname.equals("call_identifier"))
|
267
|
call_identifier = text;
|
268
|
else if (aname.equals("optional1"))
|
269
|
projcallid = text;
|
270
|
else if (aname.equals("optional2"))
|
271
|
callname = text;
|
272
|
}
|
273
|
} catch (Exception e) {
|
274
|
}
|
275
|
}
|
276
|
|
277
|
}
|
278
|
|
279
|
}
|