160 lines
4.7 KiB
Java
160 lines
4.7 KiB
Java
package eu.dnetlib.common.oai.domain;
|
|
|
|
import java.io.StringReader;
|
|
import java.net.URLEncoder;
|
|
import java.nio.charset.StandardCharsets;
|
|
import java.time.LocalDateTime;
|
|
import java.time.format.DateTimeFormatter;
|
|
import java.util.Iterator;
|
|
import java.util.Queue;
|
|
import java.util.concurrent.PriorityBlockingQueue;
|
|
|
|
import org.apache.commons.lang3.StringUtils;
|
|
import org.apache.commons.logging.Log;
|
|
import org.apache.commons.logging.LogFactory;
|
|
import org.dom4j.Document;
|
|
import org.dom4j.DocumentException;
|
|
import org.dom4j.Node;
|
|
import org.dom4j.io.SAXReader;
|
|
import org.springframework.web.client.RestTemplate;
|
|
|
|
import eu.dnetlib.errors.CollectorException;
|
|
import eu.dnetlib.utils.XmlCleaner;
|
|
|
|
public class OaiIterator implements Iterator<String> {
|
|
|
|
private static final DateTimeFormatter oaiDateFormatter = DateTimeFormatter.ofPattern("yyyy-MM-dd");
|
|
|
|
private final Queue<String> queue = new PriorityBlockingQueue<>();
|
|
private final SAXReader reader = new SAXReader();
|
|
|
|
private final String baseUrl;
|
|
private final String set;
|
|
private final String mdFormat;
|
|
private final LocalDateTime fromDate;
|
|
private final LocalDateTime untilDate;
|
|
private String token;
|
|
private boolean started;
|
|
|
|
private static final Log log = LogFactory.getLog(OaiIterator.class);
|
|
|
|
public OaiIterator(final String baseUrl,
|
|
final String mdFormat,
|
|
final String set,
|
|
final LocalDateTime fromDate,
|
|
final LocalDateTime untilDate) {
|
|
this.baseUrl = baseUrl;
|
|
this.mdFormat = mdFormat;
|
|
this.set = set;
|
|
this.fromDate = fromDate;
|
|
this.untilDate = untilDate;
|
|
this.started = false;
|
|
}
|
|
|
|
private void verifyStarted() {
|
|
if (!this.started) {
|
|
this.started = true;
|
|
try {
|
|
this.token = firstPage();
|
|
} catch (final CollectorException e) {
|
|
throw new RuntimeException(e);
|
|
}
|
|
}
|
|
}
|
|
|
|
@Override
|
|
public boolean hasNext() {
|
|
synchronized (this.queue) {
|
|
verifyStarted();
|
|
return !this.queue.isEmpty();
|
|
}
|
|
}
|
|
|
|
@Override
|
|
public String next() {
|
|
synchronized (this.queue) {
|
|
verifyStarted();
|
|
final String res = this.queue.poll();
|
|
while (this.queue.isEmpty() && (this.token != null) && !this.token.isEmpty()) {
|
|
try {
|
|
this.token = otherPages(this.token);
|
|
} catch (final CollectorException e) {
|
|
throw new RuntimeException(e);
|
|
}
|
|
}
|
|
return res;
|
|
}
|
|
}
|
|
|
|
@Override
|
|
public void remove() {}
|
|
|
|
private String firstPage() throws CollectorException {
|
|
final StringBuilder url = new StringBuilder()
|
|
.append(this.baseUrl)
|
|
.append("?verb=ListRecords&metadataPrefix=")
|
|
.append(URLEncoder.encode(this.mdFormat, StandardCharsets.UTF_8));
|
|
|
|
if (StringUtils.isNotBlank(this.set)) {
|
|
url.append("&set=").append(URLEncoder.encode(this.set, StandardCharsets.UTF_8));
|
|
}
|
|
if (this.fromDate != null) {
|
|
url.append("&from=").append(URLEncoder.encode(this.fromDate.format(oaiDateFormatter), StandardCharsets.UTF_8));
|
|
}
|
|
if (this.untilDate != null) {
|
|
url.append("&until=").append(URLEncoder.encode(this.untilDate.format(oaiDateFormatter), StandardCharsets.UTF_8));
|
|
}
|
|
return downloadPage(url.toString());
|
|
}
|
|
|
|
private String otherPages(final String resumptionToken) throws CollectorException {
|
|
return downloadPage(this.baseUrl + "?verb=ListRecords&resumptionToken=" + URLEncoder.encode(resumptionToken, StandardCharsets.UTF_8));
|
|
}
|
|
|
|
private String extractResumptionToken(final String xml) {
|
|
|
|
final String s = StringUtils.substringAfter(xml, "<resumptionToken");
|
|
if (s == null) { return null; }
|
|
|
|
final String result = StringUtils.substringBetween(s, ">", "</");
|
|
if (result == null) { return null; }
|
|
return result.trim();
|
|
|
|
}
|
|
|
|
private String downloadPage(final String url) throws CollectorException {
|
|
|
|
log.debug("Invoking url: " + url);
|
|
|
|
final String xml = new RestTemplate().getForObject(url, String.class);
|
|
Document doc;
|
|
try {
|
|
doc = this.reader.read(new StringReader(xml));
|
|
} catch (final DocumentException e) {
|
|
final String cleaned = XmlCleaner.cleanAllEntities(xml);
|
|
try {
|
|
doc = this.reader.read(new StringReader(cleaned));
|
|
} catch (final DocumentException e1) {
|
|
final String resumptionToken = extractResumptionToken(xml);
|
|
if (resumptionToken == null) { throw new CollectorException("Error parsing cleaned document:" + cleaned, e1); }
|
|
return resumptionToken;
|
|
}
|
|
}
|
|
|
|
final Node errorNode = doc.selectSingleNode("/*[local-name()='OAI-PMH']/*[local-name()='error']");
|
|
if (errorNode != null) {
|
|
final String code = errorNode.valueOf("@code");
|
|
if ("noRecordsMatch".equalsIgnoreCase(code.trim())) { return null; }
|
|
throw new CollectorException(code + " - " + errorNode.getText());
|
|
}
|
|
|
|
for (final Object o : doc.selectNodes("//*[local-name()='ListRecords']/*[local-name()='record']")) {
|
|
this.queue.add(((Node) o).asXML());
|
|
}
|
|
|
|
return doc.valueOf("//*[local-name()='resumptionToken']");
|
|
|
|
}
|
|
|
|
}
|