mirror of
https://github.com/internetarchive/heritrix3.git
synced 2026-11-09 07:24:41 +00:00
Initial checkin for [HER-1641] migration tool to convert Heritrix 1.14.x order.xml to H3 crawler-beans.cxml
This commit is contained in:
1 parent
ba5bc1a20c
commit
62690268b7
1 file changed
+46
@@ -0,0 +1,46 @@
|
||||
package org.archive.crawler.migrate;
|
||||
|
||||
import java.io.File;
|
||||
import javax.xml.parsers.DocumentBuilderFactory;
|
||||
import org.apache.commons.io.FileUtils;
|
||||
import org.w3c.dom.Document;
|
||||
|
||||
public class MigrateSettings {
|
||||
|
||||
public void instanceMain(String[] args) throws Exception {
|
||||
|
||||
File orderXmlFile = new File(args[0]);
|
||||
File newConfigDir = new File(args[1]);
|
||||
File defaultConfigDir = new File("dist/src/main/conf/jobs");
|
||||
|
||||
try {
|
||||
|
||||
// copy defaults.xml to configPath
|
||||
File srcDefaultsXml = new File(defaultConfigDir,"defaults.xml");
|
||||
File tgtDefaultsXml = new File(newConfigDir,"defaults.xml");
|
||||
FileUtils.copyFile(srcDefaultsXml, tgtDefaultsXml);
|
||||
|
||||
// copy default profile crawler-beans.xml to configPath
|
||||
File srcCrawlerBeansXml = new File(defaultConfigDir,"profile-crawler-beans.xml");
|
||||
File tgtCrawlerBeansXml = new File(newConfigDir,"crawler-beans.xml");
|
||||
FileUtils.copyFile(srcCrawlerBeansXml, tgtCrawlerBeansXml);
|
||||
|
||||
// parse order XML
|
||||
if (!orderXmlFile.canRead()) {
|
||||
throw new Exception("can not read file: "
|
||||
+ orderXmlFile.getAbsolutePath());
|
||||
} else {
|
||||
Document orderXmlDom = DocumentBuilderFactory.newInstance()
|
||||
.newDocumentBuilder().parse(orderXmlFile);
|
||||
}
|
||||
|
||||
} catch (Exception e) {
|
||||
System.out.println("Caught Exception: " + e.getMessage());
|
||||
}
|
||||
}
|
||||
|
||||
public static void main(String[] args) throws Exception {
|
||||
new MigrateSettings().instanceMain(args);
|
||||
}
|
||||
|
||||
}
|
||||
Reference in new issue
Block a user