Project

General

Profile

1
package eu.dnetlib.data.utils;
2

    
3
import java.net.MalformedURLException;
4
import java.net.URL;
5
import java.util.Arrays;
6
import java.util.regex.Matcher;
7
import java.util.regex.Pattern;
8
import java.util.stream.Collectors;
9

    
10
import org.apache.commons.codec.digest.DigestUtils;
11
import org.apache.commons.lang3.StringUtils;
12
import org.apache.commons.logging.Log;
13
import org.apache.commons.logging.LogFactory;
14

    
15
import eu.dnetlib.data.mdstore.plugins.objects.Project;
16

    
17
/**
18
 * Created by sandro on 12/6/16.
19
 */
20
public class XsltFunctions {
21

    
22
	private static final Log log = LogFactory.getLog(XsltFunctions.class);
23

    
24
	public static String cleanNames(final String s) {
25
		if (StringUtils.isBlank(s)) { return ""; }
26
		return cleanNames(s, StringUtils.countMatches(s, ",") > 1 ? "," : ";");
27
	}
28

    
29
	private static String cleanNames(final String s, final String sep) {
30
		return Arrays.stream(s.split(sep))
31
				.map(String::trim)
32
				.map(XsltFunctions::clean)
33
				.map(XsltFunctions::capitalize)
34
				.filter(StringUtils::isNotBlank)
35
				.collect(Collectors.joining("#"));
36
	}
37

    
38
	private static String clean(final String s) {
39
		return s.replaceAll("\\(.*\\)", "")
40
				.replaceAll("\\[.*\\]", "")
41
				.replaceAll("(?i)^et\\.? al(\\.|\\s)*$", "")
42
				.replaceAll("(?i)\\s*et\\.? al(\\.|\\s)*$", "")
43
				.replaceAll("\\d|\\*", "")
44
				.replaceAll("^(\\s|\\-|\\.)+", "")
45
				.replaceAll("\\,", "")
46
				.replaceAll("\\.", ". ")
47
				.trim();
48
	}
49

    
50
	public static String cleanDoi(final String doi) {
51
		if (StringUtils.isBlank(doi)) { return ""; }
52

    
53
		final String x = doi.replaceAll("\\?", "")
54
				.replaceAll("\\s*\\/\\s*", "/")
55
				.replaceAll("\\s*\\-\\s*", "-")
56
				.trim()
57
				.replaceAll("\\s", "_")
58
				.toLowerCase();
59

    
60
		if (!x.equals(doi)) {
61
			log.info("Cleaning doi: " + doi + " -> " + x);
62
		}
63

    
64
		return x;
65
	}
66

    
67
	public static String capitalize(final String s) {
68
		return Arrays.stream(s.split(" "))
69
				.map(String::toLowerCase)
70
				.map(StringUtils::capitalize)
71
				.map(XsltFunctions::fixApostrophes)
72
				.collect(Collectors.joining(" "));
73
	}
74

    
75
	public static String fixApostrophes(final String s) {
76
		return s.replaceAll("'a", "'A").replaceAll("'e", "'E").replaceAll("'i", "'I").replaceAll("'o", "'O").replaceAll("'u", "'U")
77
				.replaceAll("a'", "à").replaceAll("e'", "è").replaceAll("i'", "ì").replaceAll("o'", "ò").replaceAll("u'", "ù");
78
	}
79

    
80
	public static boolean isValidProject(final String id) {
81
		return Project.isValid(id);
82
	}
83

    
84
	public static String projectLongId(final String funder,
85
			final String program,
86
			final String code,
87
			final String jurisdiction,
88
			final String name,
89
			final String acronym) {
90

    
91
		return String.format("info:eu-repo/grantAgreement/%s/%s/%s/%s/%s/%s",
92
				funder.replaceAll("/", "%2F"),
93
				program.replaceAll("/", "%2F"),
94
				code.replaceAll("/", "%2F"),
95
				jurisdiction.replaceAll("/", "%2F"),
96
				name.replaceAll("/", "%2F"),
97
				acronym.replaceAll("/", "%2F"));
98

    
99
	}
100

    
101
	public static String calculatePersonName(final String s) {
102
		final Pattern pattern = Pattern.compile("info:cnr-pdr\\/author\\/(.+):(.+)\\/(.+)\\/(.+)");
103
		final Matcher matcher = pattern.matcher(s);
104
		return matcher.find() ? capitalize(String.format("%s, %s", matcher.group(3), matcher.group(4))) : "";
105
	}
106

    
107
	public static String serverName(final String s) {
108
		try {
109
			return new URL(s).getHost();
110
		} catch (final MalformedURLException e) {
111
			return "Unknown";
112
		}
113
	}
114

    
115
	public static String md5(final String s) {
116
		return DigestUtils.md5Hex(s);
117
	}
118

    
119
	// <xsl:for-each select="tokenize(istiFunction:cleanName(.), ';')">
120
	// <xsl:choose>
121
	// <xsl:when test="matches(normalize-space(.), ',(\s*[a-zA-Z]\.)+$')">
122
	// <creator>
123
	// <creatorName>
124
	// <xsl:value-of select="normalize-space(translate(.,',',' '))"/>
125
	// </creatorName>
126
	// </creator>
127
	// </xsl:when>
128
	//
129
	// <xsl:when test="matches(normalize-space(.), '(^[a-zA-Z\.]+,\s?[a-zA-Z\.\s]+$)|(^[a-zA-Z\.\s]+,\s?[a-zA-Z\.]+$)')">
130
	// <creator>
131
	// <creatorName>
132
	// <xsl:value-of select="normalize-space(translate(.,',',' '))"/>
133
	// </creatorName>
134
	// </creator>
135
	// </xsl:when>
136
	//
137
	// <xsl:otherwise>
138
	// <xsl:for-each select="tokenize(., ',')">
139
	// <creator>
140
	// <creatorName>
141
	// <xsl:value-of select="normalize-space(.)"/>
142
	// </creatorName>
143
	// </creator>
144
	// </xsl:for-each>
145
	// </xsl:otherwise>
146
	// </xsl:choose>
147
	// </xsl:for-each>
148
	//
149

    
150
}
(5-5/5)