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(fixApostrophes(s.toLowerCase()).split(" "))
69
				.map(p -> Arrays.stream(p.split("-"))
70
						.map(StringUtils::capitalize)
71
						.collect(Collectors.joining("-")))
72
				.map(StringUtils::capitalize)
73
				.collect(Collectors.joining(" "));
74
	}
75

    
76
	public static String fixApostrophes(final String s) {
77
		return s.replaceAll("'a", "'A").replaceAll("'e", "'E").replaceAll("'i", "'I").replaceAll("'o", "'O").replaceAll("'u", "'U")
78
				.replaceAll("a'", "à").replaceAll("e'", "è").replaceAll("i'", "ì").replaceAll("o'", "ò").replaceAll("u'", "ù");
79
	}
80

    
81
	public static boolean isValidProject(final String id) {
82
		return Project.isValid(id);
83
	}
84

    
85
	public static String projectLongId(final String funder,
86
			final String program,
87
			final String code,
88
			final String jurisdiction,
89
			final String name,
90
			final String acronym) {
91

    
92
		return String.format("info:eu-repo/grantAgreement/%s/%s/%s/%s/%s/%s",
93
				funder.replaceAll("/", "%2F"),
94
				program.replaceAll("/", "%2F"),
95
				code.replaceAll("/", "%2F"),
96
				jurisdiction.replaceAll("/", "%2F"),
97
				name.replaceAll("/", "%2F"),
98
				acronym.replaceAll("/", "%2F"));
99

    
100
	}
101

    
102
	public static String calculatePersonName(final String s) {
103
		final Pattern pattern = Pattern.compile("info:cnr-pdr\\/author\\/(.+):(.+)\\/(.+)\\/(.+)");
104
		final Matcher matcher = pattern.matcher(s);
105
		return matcher.find() ? capitalize(String.format("%s, %s", matcher.group(3), matcher.group(4))) : "";
106
	}
107

    
108
	public static String serverName(final String s) {
109
		try {
110
			return new URL(s).getHost();
111
		} catch (final MalformedURLException e) {
112
			return "Unknown";
113
		}
114
	}
115

    
116
	public static String md5(final String s) {
117
		return DigestUtils.md5Hex(s);
118
	}
119

    
120
	// <xsl:for-each select="tokenize(istiFunction:cleanName(.), ';')">
121
	// <xsl:choose>
122
	// <xsl:when test="matches(normalize-space(.), ',(\s*[a-zA-Z]\.)+$')">
123
	// <creator>
124
	// <creatorName>
125
	// <xsl:value-of select="normalize-space(translate(.,',',' '))"/>
126
	// </creatorName>
127
	// </creator>
128
	// </xsl:when>
129
	//
130
	// <xsl:when test="matches(normalize-space(.), '(^[a-zA-Z\.]+,\s?[a-zA-Z\.\s]+$)|(^[a-zA-Z\.\s]+,\s?[a-zA-Z\.]+$)')">
131
	// <creator>
132
	// <creatorName>
133
	// <xsl:value-of select="normalize-space(translate(.,',',' '))"/>
134
	// </creatorName>
135
	// </creator>
136
	// </xsl:when>
137
	//
138
	// <xsl:otherwise>
139
	// <xsl:for-each select="tokenize(., ',')">
140
	// <creator>
141
	// <creatorName>
142
	// <xsl:value-of select="normalize-space(.)"/>
143
	// </creatorName>
144
	// </creator>
145
	// </xsl:for-each>
146
	// </xsl:otherwise>
147
	// </xsl:choose>
148
	// </xsl:for-each>
149
	//
150

    
151
}
(5-5/5)