forked from D-Net/dnet-hadoop
added a retry mechanism
This commit is contained in:
parent
c08a58bba8
commit
2615136efc
|
@ -18,7 +18,11 @@ import javax.xml.transform.TransformerConfigurationException;
|
||||||
import javax.xml.transform.TransformerFactory;
|
import javax.xml.transform.TransformerFactory;
|
||||||
import javax.xml.transform.dom.DOMSource;
|
import javax.xml.transform.dom.DOMSource;
|
||||||
import javax.xml.transform.stream.StreamResult;
|
import javax.xml.transform.stream.StreamResult;
|
||||||
import javax.xml.xpath.*;
|
import javax.xml.xpath.XPath;
|
||||||
|
import javax.xml.xpath.XPathConstants;
|
||||||
|
import javax.xml.xpath.XPathExpression;
|
||||||
|
import javax.xml.xpath.XPathExpressionException;
|
||||||
|
import javax.xml.xpath.XPathFactory;
|
||||||
|
|
||||||
import org.apache.commons.io.IOUtils;
|
import org.apache.commons.io.IOUtils;
|
||||||
import org.apache.commons.lang3.StringUtils;
|
import org.apache.commons.lang3.StringUtils;
|
||||||
|
@ -47,6 +51,7 @@ public class RestIterator implements Iterator<String> {
|
||||||
|
|
||||||
private static final Logger log = LoggerFactory.getLogger(RestIterator.class);
|
private static final Logger log = LoggerFactory.getLogger(RestIterator.class);
|
||||||
public static final String UTF_8 = "UTF-8";
|
public static final String UTF_8 = "UTF-8";
|
||||||
|
private static final int MAX_ATTEMPTS = 5;
|
||||||
|
|
||||||
private final HttpClientParams clientParams;
|
private final HttpClientParams clientParams;
|
||||||
|
|
||||||
|
@ -60,7 +65,7 @@ public class RestIterator implements Iterator<String> {
|
||||||
private final int resultSizeValue;
|
private final int resultSizeValue;
|
||||||
private int resumptionInt = 0; // integer resumption token (first record to harvest)
|
private int resumptionInt = 0; // integer resumption token (first record to harvest)
|
||||||
private int resultTotal = -1;
|
private int resultTotal = -1;
|
||||||
private String resumptionStr = Integer.toString(resumptionInt); // string resumption token (first record to harvest
|
private String resumptionStr = Integer.toString(this.resumptionInt); // string resumption token (first record to harvest
|
||||||
// or token scanned from results)
|
// or token scanned from results)
|
||||||
private InputStream resultStream;
|
private InputStream resultStream;
|
||||||
private Transformer transformer;
|
private Transformer transformer;
|
||||||
|
@ -73,18 +78,18 @@ public class RestIterator implements Iterator<String> {
|
||||||
private final String querySize;
|
private final String querySize;
|
||||||
private final String authMethod;
|
private final String authMethod;
|
||||||
private final String authToken;
|
private final String authToken;
|
||||||
private final Queue<String> recordQueue = new PriorityBlockingQueue<String>();
|
private final Queue<String> recordQueue = new PriorityBlockingQueue<>();
|
||||||
private int discoverResultSize = 0;
|
private int discoverResultSize = 0;
|
||||||
private int pagination = 1;
|
private int pagination = 1;
|
||||||
/*
|
/*
|
||||||
* While resultFormatValue is added to the request parameter, this is used to say that the results are retrieved in
|
* While resultFormatValue is added to the request parameter, this is used to say that the results are retrieved in json. useful for
|
||||||
* json. useful for cases when the target API expects a resultFormatValue != json, but the results are returned in
|
* cases when the target API expects a resultFormatValue != json, but the results are returned in json. An example is the EU Open Data
|
||||||
* json. An example is the EU Open Data Portal API: resultFormatValue=standard, results are in json format.
|
* Portal API: resultFormatValue=standard, results are in json format.
|
||||||
*/
|
*/
|
||||||
private final String resultOutputFormat;
|
private final String resultOutputFormat;
|
||||||
|
|
||||||
/** RestIterator class
|
/**
|
||||||
* compatible to version 1.3.33
|
* RestIterator class compatible to version 1.3.33
|
||||||
*/
|
*/
|
||||||
public RestIterator(
|
public RestIterator(
|
||||||
final HttpClientParams clientParams,
|
final HttpClientParams clientParams,
|
||||||
|
@ -108,40 +113,40 @@ public class RestIterator implements Iterator<String> {
|
||||||
this.resumptionType = resumptionType;
|
this.resumptionType = resumptionType;
|
||||||
this.resumptionParam = resumptionParam;
|
this.resumptionParam = resumptionParam;
|
||||||
this.resultFormatValue = resultFormatValue;
|
this.resultFormatValue = resultFormatValue;
|
||||||
this.resultSizeValue = Integer.valueOf(resultSizeValueStr);
|
this.resultSizeValue = Integer.parseInt(resultSizeValueStr);
|
||||||
this.queryParams = queryParams;
|
this.queryParams = queryParams;
|
||||||
this.authMethod = authMethod;
|
this.authMethod = authMethod;
|
||||||
this.authToken = authToken;
|
this.authToken = authToken;
|
||||||
this.resultOutputFormat = resultOutputFormat;
|
this.resultOutputFormat = resultOutputFormat;
|
||||||
|
|
||||||
queryFormat = StringUtils.isNotBlank(resultFormatParam) ? "&" + resultFormatParam + "=" + resultFormatValue
|
this.queryFormat = StringUtils.isNotBlank(resultFormatParam) ? "&" + resultFormatParam + "=" + resultFormatValue
|
||||||
: "";
|
: "";
|
||||||
querySize = StringUtils.isNotBlank(resultSizeParam) ? "&" + resultSizeParam + "=" + resultSizeValueStr : "";
|
this.querySize = StringUtils.isNotBlank(resultSizeParam) ? "&" + resultSizeParam + "=" + resultSizeValueStr : "";
|
||||||
|
|
||||||
try {
|
try {
|
||||||
initXmlTransformation(resultTotalXpath, resumptionXpath, entityXpath);
|
initXmlTransformation(resultTotalXpath, resumptionXpath, entityXpath);
|
||||||
} catch (Exception e) {
|
} catch (final Exception e) {
|
||||||
throw new IllegalStateException("xml transformation init failed: " + e.getMessage());
|
throw new IllegalStateException("xml transformation init failed: " + e.getMessage());
|
||||||
}
|
}
|
||||||
|
|
||||||
initQueue();
|
initQueue();
|
||||||
}
|
}
|
||||||
|
|
||||||
private void initXmlTransformation(String resultTotalXpath, String resumptionXpath, String entityXpath)
|
private void initXmlTransformation(final String resultTotalXpath, final String resumptionXpath, final String entityXpath)
|
||||||
throws TransformerConfigurationException, XPathExpressionException {
|
throws TransformerConfigurationException, XPathExpressionException {
|
||||||
final TransformerFactory factory = TransformerFactory.newInstance();
|
final TransformerFactory factory = TransformerFactory.newInstance();
|
||||||
transformer = factory.newTransformer();
|
this.transformer = factory.newTransformer();
|
||||||
transformer.setOutputProperty(OutputKeys.INDENT, "yes");
|
this.transformer.setOutputProperty(OutputKeys.INDENT, "yes");
|
||||||
transformer.setOutputProperty("{http://xml.apache.org/xslt}indent-amount", "3");
|
this.transformer.setOutputProperty("{http://xml.apache.org/xslt}indent-amount", "3");
|
||||||
xpath = XPathFactory.newInstance().newXPath();
|
this.xpath = XPathFactory.newInstance().newXPath();
|
||||||
xprResultTotalPath = xpath.compile(resultTotalXpath);
|
this.xprResultTotalPath = this.xpath.compile(resultTotalXpath);
|
||||||
xprResumptionPath = xpath.compile(StringUtils.isBlank(resumptionXpath) ? "/" : resumptionXpath);
|
this.xprResumptionPath = this.xpath.compile(StringUtils.isBlank(resumptionXpath) ? "/" : resumptionXpath);
|
||||||
xprEntity = xpath.compile(entityXpath);
|
this.xprEntity = this.xpath.compile(entityXpath);
|
||||||
}
|
}
|
||||||
|
|
||||||
private void initQueue() {
|
private void initQueue() {
|
||||||
query = baseUrl + "?" + queryParams + querySize + queryFormat;
|
this.query = this.baseUrl + "?" + this.queryParams + this.querySize + this.queryFormat;
|
||||||
log.info("REST calls starting with {}", query);
|
log.info("REST calls starting with {}", this.query);
|
||||||
}
|
}
|
||||||
|
|
||||||
private void disconnect() {
|
private void disconnect() {
|
||||||
|
@ -150,45 +155,60 @@ public class RestIterator implements Iterator<String> {
|
||||||
|
|
||||||
/*
|
/*
|
||||||
* (non-Javadoc)
|
* (non-Javadoc)
|
||||||
|
*
|
||||||
* @see java.util.Iterator#hasNext()
|
* @see java.util.Iterator#hasNext()
|
||||||
*/
|
*/
|
||||||
@Override
|
@Override
|
||||||
public boolean hasNext() {
|
public boolean hasNext() {
|
||||||
if (recordQueue.isEmpty() && query.isEmpty()) {
|
if (this.recordQueue.isEmpty() && this.query.isEmpty()) {
|
||||||
disconnect();
|
disconnect();
|
||||||
return false;
|
return false;
|
||||||
} else {
|
|
||||||
return true;
|
|
||||||
}
|
}
|
||||||
|
return true;
|
||||||
}
|
}
|
||||||
|
|
||||||
/*
|
/*
|
||||||
* (non-Javadoc)
|
* (non-Javadoc)
|
||||||
|
*
|
||||||
* @see java.util.Iterator#next()
|
* @see java.util.Iterator#next()
|
||||||
*/
|
*/
|
||||||
@Override
|
@Override
|
||||||
public String next() {
|
public String next() {
|
||||||
synchronized (recordQueue) {
|
synchronized (this.recordQueue) {
|
||||||
while (recordQueue.isEmpty() && !query.isEmpty()) {
|
while (this.recordQueue.isEmpty() && !this.query.isEmpty()) {
|
||||||
try {
|
try {
|
||||||
query = downloadPage(query);
|
this.query = downloadPage(this.query, 0);
|
||||||
} catch (CollectorException e) {
|
} catch (final CollectorException e) {
|
||||||
log.debug("CollectorPlugin.next()-Exception: {}", e);
|
log.debug("CollectorPlugin.next()-Exception: {}", e);
|
||||||
throw new RuntimeException(e);
|
throw new RuntimeException(e);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
return recordQueue.poll();
|
return this.recordQueue.poll();
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
/*
|
/*
|
||||||
* download page and return nextQuery
|
* download page and return nextQuery (with number of attempt)
|
||||||
*/
|
*/
|
||||||
private String downloadPage(String query) throws CollectorException {
|
private String downloadPage(String query, final int attempt) throws CollectorException {
|
||||||
|
|
||||||
|
if (attempt > MAX_ATTEMPTS) { throw new CollectorException("Max Number of attempts reached, query:" + query); }
|
||||||
|
|
||||||
|
if (attempt > 0) {
|
||||||
|
final int delay = (attempt * 5000);
|
||||||
|
log.debug("Attempt {} with delay {}", attempt, delay);
|
||||||
|
try {
|
||||||
|
Thread.sleep(delay);
|
||||||
|
} catch (final InterruptedException e) {
|
||||||
|
new CollectorException(e);
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
try {
|
||||||
String resultJson;
|
String resultJson;
|
||||||
String resultXml = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>";
|
String resultXml = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>";
|
||||||
String nextQuery = "";
|
String nextQuery = "";
|
||||||
String emptyXml = resultXml + "<" + JsonUtils.XML_WRAP_TAG + "></" + JsonUtils.XML_WRAP_TAG + ">";
|
final String emptyXml = resultXml + "<" + JsonUtils.XML_WRAP_TAG + "></" + JsonUtils.XML_WRAP_TAG + ">";
|
||||||
Node resultNode = null;
|
Node resultNode = null;
|
||||||
NodeList nodeList = null;
|
NodeList nodeList = null;
|
||||||
String qUrlArgument = "";
|
String qUrlArgument = "";
|
||||||
|
@ -196,7 +216,7 @@ public class RestIterator implements Iterator<String> {
|
||||||
InputStream theHttpInputStream;
|
InputStream theHttpInputStream;
|
||||||
|
|
||||||
// check if cursor=* is initial set otherwise add it to the queryParam URL
|
// check if cursor=* is initial set otherwise add it to the queryParam URL
|
||||||
if (resumptionType.equalsIgnoreCase("deep-cursor")) {
|
if ("deep-cursor".equalsIgnoreCase(this.resumptionType)) {
|
||||||
log.debug("check resumptionType deep-cursor and check cursor=*?{}", query);
|
log.debug("check resumptionType deep-cursor and check cursor=*?{}", query);
|
||||||
if (!query.contains("&cursor=")) {
|
if (!query.contains("&cursor=")) {
|
||||||
query += "&cursor=*";
|
query += "&cursor=*";
|
||||||
|
@ -204,21 +224,21 @@ public class RestIterator implements Iterator<String> {
|
||||||
}
|
}
|
||||||
|
|
||||||
try {
|
try {
|
||||||
log.info("requestig URL [{}]", query);
|
log.info("requesting URL [{}]", query);
|
||||||
|
|
||||||
URL qUrl = new URL(query);
|
final URL qUrl = new URL(query);
|
||||||
log.debug("authMethod: {}", authMethod);
|
log.debug("authMethod: {}", this.authMethod);
|
||||||
if ("bearer".equalsIgnoreCase(this.authMethod)) {
|
if ("bearer".equalsIgnoreCase(this.authMethod)) {
|
||||||
log.trace("authMethod before inputStream: {}", resultXml);
|
log.trace("authMethod before inputStream: {}", resultXml);
|
||||||
HttpURLConnection conn = (HttpURLConnection) qUrl.openConnection();
|
final HttpURLConnection conn = (HttpURLConnection) qUrl.openConnection();
|
||||||
conn.setRequestProperty(HttpHeaders.AUTHORIZATION, "Bearer " + authToken);
|
conn.setRequestProperty(HttpHeaders.AUTHORIZATION, "Bearer " + this.authToken);
|
||||||
conn.setRequestProperty(HttpHeaders.CONTENT_TYPE, ContentType.APPLICATION_JSON.getMimeType());
|
conn.setRequestProperty(HttpHeaders.CONTENT_TYPE, ContentType.APPLICATION_JSON.getMimeType());
|
||||||
conn.setRequestMethod("GET");
|
conn.setRequestMethod("GET");
|
||||||
theHttpInputStream = conn.getInputStream();
|
theHttpInputStream = conn.getInputStream();
|
||||||
} else if (BASIC.equalsIgnoreCase(this.authMethod)) {
|
} else if (this.BASIC.equalsIgnoreCase(this.authMethod)) {
|
||||||
log.trace("authMethod before inputStream: {}", resultXml);
|
log.trace("authMethod before inputStream: {}", resultXml);
|
||||||
HttpURLConnection conn = (HttpURLConnection) qUrl.openConnection();
|
final HttpURLConnection conn = (HttpURLConnection) qUrl.openConnection();
|
||||||
conn.setRequestProperty(HttpHeaders.AUTHORIZATION, "Basic " + authToken);
|
conn.setRequestProperty(HttpHeaders.AUTHORIZATION, "Basic " + this.authToken);
|
||||||
conn.setRequestProperty(HttpHeaders.ACCEPT, ContentType.APPLICATION_XML.getMimeType());
|
conn.setRequestProperty(HttpHeaders.ACCEPT, ContentType.APPLICATION_XML.getMimeType());
|
||||||
conn.setRequestMethod("GET");
|
conn.setRequestMethod("GET");
|
||||||
theHttpInputStream = conn.getInputStream();
|
theHttpInputStream = conn.getInputStream();
|
||||||
|
@ -226,51 +246,49 @@ public class RestIterator implements Iterator<String> {
|
||||||
theHttpInputStream = qUrl.openStream();
|
theHttpInputStream = qUrl.openStream();
|
||||||
}
|
}
|
||||||
|
|
||||||
resultStream = theHttpInputStream;
|
this.resultStream = theHttpInputStream;
|
||||||
if ("json".equals(resultOutputFormat)) {
|
if ("json".equals(this.resultOutputFormat)) {
|
||||||
resultJson = IOUtils.toString(resultStream, StandardCharsets.UTF_8);
|
resultJson = IOUtils.toString(this.resultStream, StandardCharsets.UTF_8);
|
||||||
resultXml = JsonUtils.convertToXML(resultJson);
|
resultXml = JsonUtils.convertToXML(resultJson);
|
||||||
resultStream = IOUtils.toInputStream(resultXml, UTF_8);
|
this.resultStream = IOUtils.toInputStream(resultXml, UTF_8);
|
||||||
}
|
}
|
||||||
|
|
||||||
if (!(emptyXml).equalsIgnoreCase(resultXml)) {
|
if (!(emptyXml).equalsIgnoreCase(resultXml)) {
|
||||||
resultNode = (Node) xpath.evaluate("/", new InputSource(resultStream), XPathConstants.NODE);
|
resultNode = (Node) this.xpath.evaluate("/", new InputSource(this.resultStream), XPathConstants.NODE);
|
||||||
nodeList = (NodeList) xprEntity.evaluate(resultNode, XPathConstants.NODESET);
|
nodeList = (NodeList) this.xprEntity.evaluate(resultNode, XPathConstants.NODESET);
|
||||||
log.debug("nodeList.length: {}", nodeList.getLength());
|
log.debug("nodeList.length: {}", nodeList.getLength());
|
||||||
for (int i = 0; i < nodeList.getLength(); i++) {
|
for (int i = 0; i < nodeList.getLength(); i++) {
|
||||||
StringWriter sw = new StringWriter();
|
final StringWriter sw = new StringWriter();
|
||||||
transformer.transform(new DOMSource(nodeList.item(i)), new StreamResult(sw));
|
this.transformer.transform(new DOMSource(nodeList.item(i)), new StreamResult(sw));
|
||||||
String toEnqueue = sw.toString();
|
final String toEnqueue = sw.toString();
|
||||||
if (toEnqueue == null || StringUtils.isBlank(toEnqueue) || emptyXml.equalsIgnoreCase(toEnqueue)) {
|
if ((toEnqueue == null) || StringUtils.isBlank(toEnqueue) || emptyXml.equalsIgnoreCase(toEnqueue)) {
|
||||||
log.warn("The following record resulted in empty item for the feeding queue: {}", resultXml);
|
log.warn("The following record resulted in empty item for the feeding queue: {}", resultXml);
|
||||||
} else {
|
} else {
|
||||||
recordQueue.add(sw.toString());
|
this.recordQueue.add(sw.toString());
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
} else {
|
} else {
|
||||||
log.warn("resultXml is equal with emptyXml");
|
log.warn("resultXml is equal with emptyXml");
|
||||||
}
|
}
|
||||||
|
|
||||||
resumptionInt += resultSizeValue;
|
this.resumptionInt += this.resultSizeValue;
|
||||||
|
|
||||||
switch (resumptionType.toLowerCase()) {
|
switch (this.resumptionType.toLowerCase()) {
|
||||||
case "scan": // read of resumptionToken , evaluate next results, e.g. OAI, iterate over items
|
case "scan": // read of resumptionToken , evaluate next results, e.g. OAI, iterate over items
|
||||||
resumptionStr = xprResumptionPath.evaluate(resultNode);
|
this.resumptionStr = this.xprResumptionPath.evaluate(resultNode);
|
||||||
break;
|
break;
|
||||||
|
|
||||||
case "count": // begin at one step for all records, iterate over items
|
case "count": // begin at one step for all records, iterate over items
|
||||||
resumptionStr = Integer.toString(resumptionInt);
|
this.resumptionStr = Integer.toString(this.resumptionInt);
|
||||||
break;
|
break;
|
||||||
|
|
||||||
case "discover": // size of result items unknown, iterate over items (for openDOAR - 201808)
|
case "discover": // size of result items unknown, iterate over items (for openDOAR - 201808)
|
||||||
if (resultSizeValue < 2) {
|
if (this.resultSizeValue < 2) { throw new CollectorException("Mode: discover, Param 'resultSizeValue' is less than 2"); }
|
||||||
throw new CollectorException("Mode: discover, Param 'resultSizeValue' is less than 2");
|
|
||||||
}
|
|
||||||
qUrlArgument = qUrl.getQuery();
|
qUrlArgument = qUrl.getQuery();
|
||||||
String[] arrayQUrlArgument = qUrlArgument.split("&");
|
final String[] arrayQUrlArgument = qUrlArgument.split("&");
|
||||||
for (String arrayUrlArgStr : arrayQUrlArgument) {
|
for (final String arrayUrlArgStr : arrayQUrlArgument) {
|
||||||
if (arrayUrlArgStr.startsWith(resumptionParam)) {
|
if (arrayUrlArgStr.startsWith(this.resumptionParam)) {
|
||||||
String[] resumptionKeyValue = arrayUrlArgStr.split("=");
|
final String[] resumptionKeyValue = arrayUrlArgStr.split("=");
|
||||||
if (isInteger(resumptionKeyValue[1])) {
|
if (isInteger(resumptionKeyValue[1])) {
|
||||||
urlOldResumptionSize = Integer.parseInt(resumptionKeyValue[1]);
|
urlOldResumptionSize = Integer.parseInt(resumptionKeyValue[1]);
|
||||||
log.debug("discover OldResumptionSize from Url (int): {}", urlOldResumptionSize);
|
log.debug("discover OldResumptionSize from Url (int): {}", urlOldResumptionSize);
|
||||||
|
@ -281,33 +299,33 @@ public class RestIterator implements Iterator<String> {
|
||||||
}
|
}
|
||||||
|
|
||||||
if (((emptyXml).equalsIgnoreCase(resultXml))
|
if (((emptyXml).equalsIgnoreCase(resultXml))
|
||||||
|| ((nodeList != null) && (nodeList.getLength() < resultSizeValue))) {
|
|| ((nodeList != null) && (nodeList.getLength() < this.resultSizeValue))) {
|
||||||
// resumptionStr = "";
|
// resumptionStr = "";
|
||||||
if (nodeList != null) {
|
if (nodeList != null) {
|
||||||
discoverResultSize += nodeList.getLength();
|
this.discoverResultSize += nodeList.getLength();
|
||||||
}
|
}
|
||||||
resultTotal = discoverResultSize;
|
this.resultTotal = this.discoverResultSize;
|
||||||
} else {
|
} else {
|
||||||
resumptionStr = Integer.toString(resumptionInt);
|
this.resumptionStr = Integer.toString(this.resumptionInt);
|
||||||
resultTotal = resumptionInt + 1;
|
this.resultTotal = this.resumptionInt + 1;
|
||||||
if (nodeList != null) {
|
if (nodeList != null) {
|
||||||
discoverResultSize += nodeList.getLength();
|
this.discoverResultSize += nodeList.getLength();
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
log.info("discoverResultSize: {}", discoverResultSize);
|
log.info("discoverResultSize: {}", this.discoverResultSize);
|
||||||
break;
|
break;
|
||||||
|
|
||||||
case "pagination":
|
case "pagination":
|
||||||
case "page": // pagination, iterate over page numbers
|
case "page": // pagination, iterate over page numbers
|
||||||
pagination += 1;
|
this.pagination += 1;
|
||||||
if (nodeList != null) {
|
if (nodeList != null) {
|
||||||
discoverResultSize += nodeList.getLength();
|
this.discoverResultSize += nodeList.getLength();
|
||||||
} else {
|
} else {
|
||||||
resultTotal = discoverResultSize;
|
this.resultTotal = this.discoverResultSize;
|
||||||
pagination = discoverResultSize;
|
this.pagination = this.discoverResultSize;
|
||||||
}
|
}
|
||||||
resumptionInt = pagination;
|
this.resumptionInt = this.pagination;
|
||||||
resumptionStr = Integer.toString(resumptionInt);
|
this.resumptionStr = Integer.toString(this.resumptionInt);
|
||||||
break;
|
break;
|
||||||
|
|
||||||
case "deep-cursor": // size of result items unknown, iterate over items (for supporting deep cursor in
|
case "deep-cursor": // size of result items unknown, iterate over items (for supporting deep cursor in
|
||||||
|
@ -315,26 +333,25 @@ public class RestIterator implements Iterator<String> {
|
||||||
// isn't relevant -- if (resultSizeValue < 2) {throw new CollectorServiceException("Mode:
|
// isn't relevant -- if (resultSizeValue < 2) {throw new CollectorServiceException("Mode:
|
||||||
// deep-cursor, Param 'resultSizeValue' is less than 2");}
|
// deep-cursor, Param 'resultSizeValue' is less than 2");}
|
||||||
|
|
||||||
resumptionStr = encodeValue(xprResumptionPath.evaluate(resultNode));
|
this.resumptionStr = encodeValue(this.xprResumptionPath.evaluate(resultNode));
|
||||||
queryParams = queryParams.replace("&cursor=*", "");
|
this.queryParams = this.queryParams.replace("&cursor=*", "");
|
||||||
|
|
||||||
// terminating if length of nodeList is 0
|
// terminating if length of nodeList is 0
|
||||||
if ((nodeList != null) && (nodeList.getLength() < discoverResultSize)) {
|
if ((nodeList != null) && (nodeList.getLength() < this.discoverResultSize)) {
|
||||||
resumptionInt += (nodeList.getLength() + 1 - resultSizeValue);
|
this.resumptionInt += ((nodeList.getLength() + 1) - this.resultSizeValue);
|
||||||
} else {
|
} else {
|
||||||
resumptionInt += (nodeList.getLength() - resultSizeValue); // subtract the resultSizeValue
|
this.resumptionInt += (nodeList.getLength() - this.resultSizeValue); // subtract the resultSizeValue
|
||||||
// because the iteration is over
|
// because the iteration is over
|
||||||
// real length and the
|
// real length and the
|
||||||
// resultSizeValue is added before
|
// resultSizeValue is added before
|
||||||
// the switch()
|
// the switch()
|
||||||
}
|
}
|
||||||
|
|
||||||
discoverResultSize = nodeList.getLength();
|
this.discoverResultSize = nodeList.getLength();
|
||||||
|
|
||||||
log
|
log
|
||||||
.debug(
|
.debug("downloadPage().deep-cursor: resumptionStr=" + this.resumptionStr + " ; queryParams="
|
||||||
"downloadPage().deep-cursor: resumptionStr=" + resumptionStr + " ; queryParams="
|
+ this.queryParams + " resumptionLengthIncreased: " + this.resumptionInt);
|
||||||
+ queryParams + " resumptionLengthIncreased: " + resumptionInt);
|
|
||||||
|
|
||||||
break;
|
break;
|
||||||
|
|
||||||
|
@ -343,28 +360,28 @@ public class RestIterator implements Iterator<String> {
|
||||||
break;
|
break;
|
||||||
}
|
}
|
||||||
|
|
||||||
} catch (Exception e) {
|
} catch (final Exception e) {
|
||||||
log.error(e.getMessage(), e);
|
log.error(e.getMessage(), e);
|
||||||
throw new IllegalStateException("collection failed: " + e.getMessage());
|
throw new IllegalStateException("collection failed: " + e.getMessage());
|
||||||
}
|
}
|
||||||
|
|
||||||
try {
|
try {
|
||||||
if (resultTotal == -1) {
|
if (this.resultTotal == -1) {
|
||||||
resultTotal = Integer.parseInt(xprResultTotalPath.evaluate(resultNode));
|
this.resultTotal = Integer.parseInt(this.xprResultTotalPath.evaluate(resultNode));
|
||||||
if (resumptionType.equalsIgnoreCase("page") && !BASIC.equalsIgnoreCase(authMethod)) {
|
if ("page".equalsIgnoreCase(this.resumptionType) && !this.BASIC.equalsIgnoreCase(this.authMethod)) {
|
||||||
resultTotal += 1;
|
this.resultTotal += 1;
|
||||||
} // to correct the upper bound
|
} // to correct the upper bound
|
||||||
log.info("resultTotal was -1 is now: " + resultTotal);
|
log.info("resultTotal was -1 is now: " + this.resultTotal);
|
||||||
}
|
}
|
||||||
} catch (Exception e) {
|
} catch (final Exception e) {
|
||||||
log.error(e.getMessage(), e);
|
log.error(e.getMessage(), e);
|
||||||
throw new IllegalStateException("downloadPage resultTotal couldn't parse: " + e.getMessage());
|
throw new IllegalStateException("downloadPage resultTotal couldn't parse: " + e.getMessage());
|
||||||
}
|
}
|
||||||
log.debug("resultTotal: " + resultTotal);
|
log.debug("resultTotal: " + this.resultTotal);
|
||||||
log.debug("resInt: " + resumptionInt);
|
log.debug("resInt: " + this.resumptionInt);
|
||||||
if (resumptionInt <= resultTotal) {
|
if (this.resumptionInt <= this.resultTotal) {
|
||||||
nextQuery = baseUrl + "?" + queryParams + querySize + "&" + resumptionParam + "=" + resumptionStr
|
nextQuery = this.baseUrl + "?" + this.queryParams + this.querySize + "&" + this.resumptionParam + "=" + this.resumptionStr
|
||||||
+ queryFormat;
|
+ this.queryFormat;
|
||||||
} else {
|
} else {
|
||||||
nextQuery = "";
|
nextQuery = "";
|
||||||
// if (resumptionType.toLowerCase().equals("deep-cursor")) { resumptionInt -= 1; } // correct the
|
// if (resumptionType.toLowerCase().equals("deep-cursor")) { resumptionInt -= 1; } // correct the
|
||||||
|
@ -372,10 +389,14 @@ public class RestIterator implements Iterator<String> {
|
||||||
}
|
}
|
||||||
log.debug("nextQueryUrl: " + nextQuery);
|
log.debug("nextQueryUrl: " + nextQuery);
|
||||||
return nextQuery;
|
return nextQuery;
|
||||||
|
} catch (final Throwable e) {
|
||||||
|
log.warn(e.getMessage(), e);
|
||||||
|
return downloadPage(query, attempt + 1);
|
||||||
|
}
|
||||||
|
|
||||||
}
|
}
|
||||||
|
|
||||||
private boolean isInteger(String s) {
|
private boolean isInteger(final String s) {
|
||||||
boolean isValidInteger = false;
|
boolean isValidInteger = false;
|
||||||
try {
|
try {
|
||||||
Integer.parseInt(s);
|
Integer.parseInt(s);
|
||||||
|
@ -383,7 +404,7 @@ public class RestIterator implements Iterator<String> {
|
||||||
// s is a valid integer
|
// s is a valid integer
|
||||||
|
|
||||||
isValidInteger = true;
|
isValidInteger = true;
|
||||||
} catch (NumberFormatException ex) {
|
} catch (final NumberFormatException ex) {
|
||||||
// s is not an integer
|
// s is not an integer
|
||||||
}
|
}
|
||||||
|
|
||||||
|
@ -391,20 +412,20 @@ public class RestIterator implements Iterator<String> {
|
||||||
}
|
}
|
||||||
|
|
||||||
// Method to encode a string value using `UTF-8` encoding scheme
|
// Method to encode a string value using `UTF-8` encoding scheme
|
||||||
private String encodeValue(String value) {
|
private String encodeValue(final String value) {
|
||||||
try {
|
try {
|
||||||
return URLEncoder.encode(value, StandardCharsets.UTF_8.toString());
|
return URLEncoder.encode(value, StandardCharsets.UTF_8.toString());
|
||||||
} catch (UnsupportedEncodingException ex) {
|
} catch (final UnsupportedEncodingException ex) {
|
||||||
throw new RuntimeException(ex.getCause());
|
throw new RuntimeException(ex.getCause());
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
public String getResultFormatValue() {
|
public String getResultFormatValue() {
|
||||||
return resultFormatValue;
|
return this.resultFormatValue;
|
||||||
}
|
}
|
||||||
|
|
||||||
public String getResultOutputFormat() {
|
public String getResultOutputFormat() {
|
||||||
return resultOutputFormat;
|
return this.resultOutputFormat;
|
||||||
}
|
}
|
||||||
|
|
||||||
}
|
}
|
||||||
|
|
Loading…
Reference in New Issue