1
|
package eu.dnetlib.data.collector.plugins.schemaorg;
|
2
|
|
3
|
import org.json.JSONArray;
|
4
|
import org.json.JSONObject;
|
5
|
import org.w3c.dom.Document;
|
6
|
import org.w3c.dom.NodeList;
|
7
|
import org.xml.sax.InputSource;
|
8
|
|
9
|
import javax.xml.parsers.DocumentBuilder;
|
10
|
import javax.xml.parsers.DocumentBuilderFactory;
|
11
|
import javax.xml.xpath.XPath;
|
12
|
import javax.xml.xpath.XPathConstants;
|
13
|
import javax.xml.xpath.XPathExpression;
|
14
|
import javax.xml.xpath.XPathFactory;
|
15
|
import java.io.File;
|
16
|
import java.io.FileInputStream;
|
17
|
import java.io.FileOutputStream;
|
18
|
import java.io.StringReader;
|
19
|
import java.nio.charset.Charset;
|
20
|
import java.nio.charset.UnsupportedCharsetException;
|
21
|
import java.util.ArrayList;
|
22
|
import java.util.EnumSet;
|
23
|
import java.util.HashMap;
|
24
|
import java.util.List;
|
25
|
import java.util.zip.GZIPInputStream;
|
26
|
|
27
|
public class Utils {
|
28
|
|
29
|
public static List<String> collectAsStrings(String xml, String xpath) throws Exception{
|
30
|
DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
|
31
|
DocumentBuilder builder = factory.newDocumentBuilder();
|
32
|
Document doc = builder.parse(new InputSource(new StringReader(xml)));
|
33
|
return Utils.collectAsStrings(doc, xpath);
|
34
|
}
|
35
|
|
36
|
public static List<String> collectAsStrings(File file, String xpath) throws Exception{
|
37
|
DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
|
38
|
DocumentBuilder builder = factory.newDocumentBuilder();
|
39
|
Document doc = builder.parse(file);
|
40
|
return Utils.collectAsStrings(doc, xpath);
|
41
|
}
|
42
|
|
43
|
public static List<String> collectAsStrings(Document doc, String xpath) throws Exception{
|
44
|
XPathFactory xPathfactory = XPathFactory.newInstance();
|
45
|
XPath path = xPathfactory.newXPath();
|
46
|
XPathExpression expr = path.compile(xpath);
|
47
|
NodeList nodes = (NodeList) expr.evaluate(doc, XPathConstants.NODESET);
|
48
|
|
49
|
List<String> values = new ArrayList<>();
|
50
|
|
51
|
for (int i = 0; i < nodes.getLength(); i++)
|
52
|
values.add(nodes.item(i).getNodeValue());
|
53
|
|
54
|
return values;
|
55
|
}
|
56
|
|
57
|
public static void decompressGZipTo(File input, File output) throws Exception {
|
58
|
try (GZIPInputStream in = new GZIPInputStream(new FileInputStream(input))){
|
59
|
try (FileOutputStream out = new FileOutputStream(output)){
|
60
|
byte[] buffer = new byte[1024];
|
61
|
int len;
|
62
|
while((len = in.read(buffer)) != -1){
|
63
|
out.write(buffer, 0, len);
|
64
|
}
|
65
|
}
|
66
|
}
|
67
|
}
|
68
|
|
69
|
public static String getAsString(HashMap<String,String> map, String key, String defaultValue)
|
70
|
{
|
71
|
String value = map.get(key);
|
72
|
if(value == null) return defaultValue;
|
73
|
return value;
|
74
|
}
|
75
|
|
76
|
public static List<String> getAsStringCsv(HashMap<String,String> map, String key, List<String> defaultValue)
|
77
|
{
|
78
|
String value = map.get(key);
|
79
|
if(value == null) return defaultValue;
|
80
|
String[] splits = value.split(",");
|
81
|
List<String> curated = new ArrayList<>();
|
82
|
for(String item : splits){
|
83
|
if(item == null || item.trim().length() == 0) continue;
|
84
|
curated.add(item.trim());
|
85
|
}
|
86
|
return curated;
|
87
|
}
|
88
|
|
89
|
public static int getAsInt(HashMap<String,String> map, String key, int defaultValue)
|
90
|
{
|
91
|
String value = map.get(key);
|
92
|
if(value == null) return defaultValue;
|
93
|
try {
|
94
|
return Integer.parseInt(value);
|
95
|
} catch (NumberFormatException e) {
|
96
|
return defaultValue;
|
97
|
}
|
98
|
}
|
99
|
|
100
|
public static long getAsLong(HashMap<String,String> map, String key, long defaultValue)
|
101
|
{
|
102
|
String value = map.get(key);
|
103
|
if(value == null) return defaultValue;
|
104
|
try {
|
105
|
return Long.parseLong(value);
|
106
|
} catch (NumberFormatException e) {
|
107
|
return defaultValue;
|
108
|
}
|
109
|
}
|
110
|
|
111
|
public static <E extends Enum<E>> E getAsEnum(HashMap<String,String> map, String key, E defaultValue, Class<E> clazz) {
|
112
|
//EnumSet<E> values = EnumSet.allOf(defaultValue.getClass());
|
113
|
EnumSet<E> values = EnumSet.allOf(clazz);
|
114
|
String value = map.get(key);
|
115
|
if (value == null) return defaultValue;
|
116
|
for(E val : values){
|
117
|
if(!val.name().equalsIgnoreCase(value)) continue;
|
118
|
return val;
|
119
|
}
|
120
|
return defaultValue;
|
121
|
}
|
122
|
|
123
|
public static Boolean getAsBoolean(HashMap<String,String> map, String key, Boolean defaultValue) {
|
124
|
String value = map.get(key);
|
125
|
if (value == null) return defaultValue;
|
126
|
return Boolean.parseBoolean(value);
|
127
|
}
|
128
|
|
129
|
public static Charset getAsCharset(HashMap<String,String> map, String key, Charset defaultValue)
|
130
|
{
|
131
|
String value = map.get(key);
|
132
|
if(value == null) return defaultValue;
|
133
|
try {
|
134
|
return Charset.forName(value);
|
135
|
} catch (UnsupportedCharsetException e) {
|
136
|
return defaultValue;
|
137
|
}
|
138
|
}
|
139
|
|
140
|
}
|