Revision 53691
Added by Claudio Atzori over 5 years ago
SchemaOrgMainReactome.java | ||
---|---|---|
3 | 3 |
import eu.dnetlib.data.collector.plugins.schemaorg.sitemapindex.SitemapFileIterator; |
4 | 4 |
import eu.dnetlib.data.collector.rmi.InterfaceDescriptor; |
5 | 5 |
import org.apache.commons.io.FileUtils; |
6 |
import org.apache.commons.lang3.StringUtils; |
|
7 | 6 |
import org.apache.commons.logging.Log; |
8 | 7 |
import org.apache.commons.logging.LogFactory; |
9 | 8 |
import org.apache.log4j.ConsoleAppender; |
10 | 9 |
import org.apache.log4j.Level; |
11 | 10 |
import org.apache.log4j.Logger; |
12 | 11 |
import org.apache.log4j.PatternLayout; |
13 |
import org.dom4j.Document; |
|
14 |
import org.dom4j.io.SAXReader; |
|
15 | 12 |
|
16 | 13 |
import java.io.File; |
17 |
import java.io.FileWriter; |
|
18 |
import java.io.StringReader; |
|
19 | 14 |
import java.nio.charset.StandardCharsets; |
20 | 15 |
import java.util.HashMap; |
21 | 16 |
import java.util.concurrent.TimeUnit; |
... | ... | |
56 | 51 |
params.put("sitemap_FileCharset", StandardCharsets.UTF_8.name()); |
57 | 52 |
params.put("sitemap_FileSchema", SitemapFileIterator.Options.SitemapSchemaType.Text.toString()); |
58 | 53 |
params.put("sitemap_FileType", SitemapFileIterator.Options.SitemapFileType.GZ.toString()); |
54 |
params.put("sitemap_producerBlockPollingTimeout", "2"); |
|
55 |
params.put("sitemap_producerBlockPollingTimeoutUnit", TimeUnit.MINUTES.toString()); |
|
59 | 56 |
|
60 |
|
|
61 |
|
|
62 | 57 |
InterfaceDescriptor descriptor = new InterfaceDescriptor(); |
63 | 58 |
descriptor.setId("schema.org - reactome"); |
64 | 59 |
descriptor.setBaseUrl("https://reactome.org/sitemapindex.xml"); |
Also available in: Unified diff
Kaggle/Reactome: added configurable params