1 |
36921
|
eri.katsar
|
package eu.dnetlib.data.mapreduce.hbase.statsExport.utils;
|
2 |
37023
|
eri.katsar
|
|
3 |
36975
|
eri.katsar
|
import org.apache.log4j.Logger;
|
4 |
37023
|
eri.katsar
|
|
5 |
36921
|
eri.katsar
|
/**
|
6 |
|
|
* Created by envy17 j110ea on 4/5/2015.
|
7 |
|
|
*/
|
8 |
36975
|
eri.katsar
|
|
9 |
42734
|
eri.katsar
|
//Extract funder info for projects
|
10 |
|
|
|
11 |
36921
|
eri.katsar
|
public class FundingParser {
|
12 |
|
|
|
13 |
36982
|
eri.katsar
|
|
14 |
42734
|
eri.katsar
|
public static String getFundingLevel(String funding_level, int level, String DELIM, String ENCLOSING) {
|
15 |
36921
|
eri.katsar
|
|
16 |
37023
|
eri.katsar
|
if (funding_level.isEmpty()) {
|
17 |
42734
|
eri.katsar
|
return ENCLOSING + " " + ENCLOSING + DELIM;
|
18 |
37023
|
eri.katsar
|
}
|
19 |
36921
|
eri.katsar
|
|
20 |
37023
|
eri.katsar
|
if (!funding_level.contains("<funding_level_" + level + ">")) {
|
21 |
42734
|
eri.katsar
|
return ENCLOSING + " " + ENCLOSING + DELIM;
|
22 |
37023
|
eri.katsar
|
}
|
23 |
36921
|
eri.katsar
|
|
24 |
37023
|
eri.katsar
|
String[] split = funding_level.split("<funding_level_" + level + ">");
|
25 |
36921
|
eri.katsar
|
|
26 |
37023
|
eri.katsar
|
funding_level = split[1];
|
27 |
36921
|
eri.katsar
|
|
28 |
37023
|
eri.katsar
|
split = funding_level.split("<name>");
|
29 |
|
|
funding_level = split[1];
|
30 |
36921
|
eri.katsar
|
|
31 |
37023
|
eri.katsar
|
funding_level = funding_level.substring(0, funding_level.indexOf("</name>"));
|
32 |
|
|
funding_level = funding_level.replaceAll("\"", "");
|
33 |
42734
|
eri.katsar
|
funding_level = funding_level.replaceAll("/>", "");
|
34 |
|
|
funding_level = funding_level.replaceAll("<", "");
|
35 |
37023
|
eri.katsar
|
funding_level = funding_level.replaceAll("&", "");
|
36 |
37078
|
eri.katsar
|
|
37 |
|
|
if (level == 1) {
|
38 |
|
|
if (funding_level.equalsIgnoreCase("SP1")) {
|
39 |
|
|
funding_level = "SP1-Cooperation";
|
40 |
|
|
} else if (funding_level.equalsIgnoreCase("SP2")) {
|
41 |
|
|
funding_level = "SP2-Ideas";
|
42 |
|
|
}
|
43 |
|
|
if (funding_level.equalsIgnoreCase("SP3")) {
|
44 |
|
|
funding_level = "SP3-People";
|
45 |
|
|
} else if (funding_level.equalsIgnoreCase("SP4")) {
|
46 |
39008
|
eri.katsar
|
funding_level = "SP4-Capacities";
|
47 |
37078
|
eri.katsar
|
|
48 |
|
|
} else if (funding_level.equalsIgnoreCase("SP5")) {
|
49 |
|
|
funding_level = "SP5-Euratom";
|
50 |
|
|
}
|
51 |
|
|
}
|
52 |
|
|
|
53 |
38097
|
eri.katsar
|
|
54 |
|
|
funding_level = funding_level.replaceAll(">", "");
|
55 |
|
|
|
56 |
|
|
funding_level = funding_level.replaceAll("</", "");
|
57 |
42734
|
eri.katsar
|
funding_level = funding_level.replace(DELIM, " ");
|
58 |
|
|
funding_level = funding_level.replace(ENCLOSING, " ");
|
59 |
38097
|
eri.katsar
|
|
60 |
42734
|
eri.katsar
|
return ENCLOSING + funding_level + ENCLOSING + DELIM;
|
61 |
37023
|
eri.katsar
|
}
|
62 |
36921
|
eri.katsar
|
|
63 |
42734
|
eri.katsar
|
public static String getFundingInfo(String buff, String DELIM, String ENCLOSING) {
|
64 |
|
|
return getFunder(buff, DELIM, ENCLOSING) +
|
65 |
|
|
getFundingLevel(buff, 0, DELIM, ENCLOSING) + (getFundingLevel(buff, 1, DELIM, ENCLOSING) + getFundingLevel(buff, 2, DELIM, ENCLOSING)
|
66 |
|
|
+ getFundingLevel(buff, 3, DELIM, ENCLOSING));
|
67 |
37023
|
eri.katsar
|
}
|
68 |
36921
|
eri.katsar
|
|
69 |
42734
|
eri.katsar
|
public static String getFunder(String buff, String DELIM, String ENCLOSING) {
|
70 |
36921
|
eri.katsar
|
|
71 |
37023
|
eri.katsar
|
if (buff.isEmpty()) {
|
72 |
42734
|
eri.katsar
|
return ENCLOSING + " " + ENCLOSING + DELIM;
|
73 |
36921
|
eri.katsar
|
|
74 |
37023
|
eri.katsar
|
}
|
75 |
|
|
if (!buff.contains("<funder>")) {
|
76 |
42734
|
eri.katsar
|
return ENCLOSING + " " + ENCLOSING + DELIM;
|
77 |
37023
|
eri.katsar
|
}
|
78 |
42734
|
eri.katsar
|
|
79 |
37023
|
eri.katsar
|
String[] split = buff.split("<funder>");
|
80 |
|
|
String funder = split[1];
|
81 |
|
|
split = funder.split("<name>");
|
82 |
36982
|
eri.katsar
|
|
83 |
37023
|
eri.katsar
|
funder = split[1];
|
84 |
37008
|
eri.katsar
|
|
85 |
37023
|
eri.katsar
|
funder = funder.substring(0, funder.indexOf("</name>"));
|
86 |
38097
|
eri.katsar
|
|
87 |
|
|
funder = funder.replaceAll(">", "");
|
88 |
|
|
|
89 |
|
|
funder = funder.replaceAll("</", "");
|
90 |
|
|
|
91 |
37023
|
eri.katsar
|
funder = funder.replaceAll("\"", "");
|
92 |
|
|
funder = funder.replaceAll("&", "");
|
93 |
42734
|
eri.katsar
|
funder = funder.replace(ENCLOSING, " ");
|
94 |
37008
|
eri.katsar
|
|
95 |
42734
|
eri.katsar
|
return ENCLOSING + funder + ENCLOSING + DELIM;
|
96 |
37023
|
eri.katsar
|
}
|
97 |
36921
|
eri.katsar
|
|
98 |
|
|
}
|