sabeel
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Pull Requests Activity

formatting modified after commit

This commit is contained in:
Enrico Ottonello 2020-05-08 14:54:37 +02:00
parent 7e1c987370
commit b9d126dd1f
1 changed files with 138 additions and 142 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

280
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/orcid/OrcidDownloader.java
View File

@ -24,160 +24,156 @@ import org.mortbay.log.Log;
public class OrcidDownloader extends OrcidDSManager { public class OrcidDownloader extends OrcidDSManager {
static final int REQ_LIMIT = 24; static final int REQ_LIMIT = 24;
static final int REQ_MAX_TEST = 100; static final int REQ_MAX_TEST = 100;
static final int RECORD_PARSED_COUNTER_LOG_INTERVAL = 50000; static final int RECORD_PARSED_COUNTER_LOG_INTERVAL = 50000;
static final String DATE_FORMAT = "yyyy-MM-dd HH:mm:ss."; static final String DATE_FORMAT = "yyyy-MM-dd HH:mm:ss.";
static final String lastUpdate = "2019-09-30 00:00:00.000000"; static final String lastUpdate = "2019-09-30 00:00:00.000000";
private String lambdaFileName; private String lambdaFileName;
private String outputPath; private String outputPath;
private String token; private String token;
public static void main(String[] args) throws IOException, Exception { public static void main(String[] args) throws IOException, Exception {
OrcidDownloader orcidDownloader = new OrcidDownloader(); OrcidDownloader orcidDownloader = new OrcidDownloader();
orcidDownloader.loadArgs(args); orcidDownloader.loadArgs(args);
orcidDownloader.parseLambdaFile(); orcidDownloader.parseLambdaFile();
} }
private String downloadRecord(String orcidId) throws Exception { private String downloadRecord(String orcidId) throws Exception {
try (CloseableHttpClient client = HttpClients.createDefault()) { try (CloseableHttpClient client = HttpClients.createDefault()) {
HttpGet httpGet = new HttpGet("https://api.orcid.org/v3.0/" + orcidId + "/record"); HttpGet httpGet = new HttpGet("https://api.orcid.org/v3.0/" + orcidId + "/record");
httpGet.addHeader("Accept", "application/vnd.orcid+xml"); httpGet.addHeader("Accept", "application/vnd.orcid+xml");
httpGet.addHeader("Authorization", String.format("Bearer %s", token)); httpGet.addHeader("Authorization", String.format("Bearer %s", token));
CloseableHttpResponse response = client.execute(httpGet); CloseableHttpResponse response = client.execute(httpGet);
if (response.getStatusLine().getStatusCode() != 200) { if (response.getStatusLine().getStatusCode() != 200) {
Log.warn( Log.warn(
"Downloading " "Downloading " + orcidId + " status code: " + response.getStatusLine().getStatusCode());
+ orcidId
+ " status code: "
+ response.getStatusLine().getStatusCode());
return new String("");
}
return IOUtils.toString(response.getEntity().getContent());
} catch (Throwable e) {
Log.warn("Downloading " + orcidId, e);
}
return new String(""); return new String("");
}
return IOUtils.toString(response.getEntity().getContent());
} catch (Throwable e) {
Log.warn("Downloading " + orcidId, e);
} }
return new String("");
}
public void parseLambdaFile() throws Exception { public void parseLambdaFile() throws Exception {
int parsedRecordsCounter = 0; int parsedRecordsCounter = 0;
int downloadedRecordsCounter = 0; int downloadedRecordsCounter = 0;
int savedRecordsCounter = 0; int savedRecordsCounter = 0;
long startDownload = 0; long startDownload = 0;
Configuration conf = initConfigurationObject(); Configuration conf = initConfigurationObject();
FileSystem fs = initFileSystemObject(conf); FileSystem fs = initFileSystemObject(conf);
String lambdaFileUri = hdfsServerUri.concat(hdfsOrcidDefaultPath).concat(lambdaFileName); String lambdaFileUri = hdfsServerUri.concat(hdfsOrcidDefaultPath).concat(lambdaFileName);
Path hdfsreadpath = new Path(lambdaFileUri); Path hdfsreadpath = new Path(lambdaFileUri);
FSDataInputStream lambdaFileStream = fs.open(hdfsreadpath); FSDataInputStream lambdaFileStream = fs.open(hdfsreadpath);
Path hdfsoutputPath = Path hdfsoutputPath =
new Path( new Path(
hdfsServerUri hdfsServerUri
.concat(hdfsOrcidDefaultPath) .concat(hdfsOrcidDefaultPath)
.concat(outputPath) .concat(outputPath)
.concat("orcid_records.seq")); .concat("orcid_records.seq"));
try (SequenceFile.Writer writer = try (SequenceFile.Writer writer =
SequenceFile.createWriter( SequenceFile.createWriter(
conf, conf,
SequenceFile.Writer.file(hdfsoutputPath), SequenceFile.Writer.file(hdfsoutputPath),
SequenceFile.Writer.keyClass(Text.class), SequenceFile.Writer.keyClass(Text.class),
SequenceFile.Writer.valueClass(Text.class))) { SequenceFile.Writer.valueClass(Text.class))) {
try (BufferedReader br = new BufferedReader(new InputStreamReader(lambdaFileStream))) { try (BufferedReader br = new BufferedReader(new InputStreamReader(lambdaFileStream))) {
String line; String line;
int nReqTmp = 0; int nReqTmp = 0;
startDownload = System.currentTimeMillis(); startDownload = System.currentTimeMillis();
long startReqTmp = System.currentTimeMillis(); long startReqTmp = System.currentTimeMillis();
while ((line = br.readLine()) != null) { while ((line = br.readLine()) != null) {
parsedRecordsCounter++; parsedRecordsCounter++;
// skip headers line // skip headers line
if (parsedRecordsCounter == 1) { if (parsedRecordsCounter == 1) {
continue; continue;
} }
String[] values = line.split(","); String[] values = line.split(",");
List<String> recordInfo = Arrays.asList(values); List<String> recordInfo = Arrays.asList(values);
if (isModified(recordInfo.get(3))) { if (isModified(recordInfo.get(3))) {
String record = downloadRecord(recordInfo.get(0)); String record = downloadRecord(recordInfo.get(0));
downloadedRecordsCounter++; downloadedRecordsCounter++;
if (!record.isEmpty()) { if (!record.isEmpty()) {
String compressRecord = String compressRecord = ArgumentApplicationParser.compressArgument(record);
ArgumentApplicationParser.compressArgument(record); final Text key = new Text(recordInfo.get(0));
final Text key = new Text(recordInfo.get(0)); final Text value = new Text(compressRecord);
final Text value = new Text(compressRecord);
try { try {
writer.append(key, value); writer.append(key, value);
savedRecordsCounter++; savedRecordsCounter++;
} catch (IOException e) { } catch (IOException e) {
Log.debug("Writing to sequence file: " + e.getMessage()); Log.debug("Writing to sequence file: " + e.getMessage());
Log.debug(e); Log.debug(e);
throw new RuntimeException(e); throw new RuntimeException(e);
} }
}
}
long endReq = System.currentTimeMillis();
nReqTmp++;
if (nReqTmp == REQ_LIMIT) {
long reqSessionDuration = endReq - startReqTmp;
if (reqSessionDuration <= 1000) {
Log.warn(
"\nreqSessionDuration: "
+ reqSessionDuration
+ " nReqTmp: "
+ nReqTmp
+ " wait ....");
Thread.sleep(1000 - reqSessionDuration);
} else {
nReqTmp = 0;
startReqTmp = System.currentTimeMillis();
}
}
// if (parsedRecordsCounter>REQ_MAX_TEST) {
// break;
// }
if ((parsedRecordsCounter % RECORD_PARSED_COUNTER_LOG_INTERVAL) == 0) {
Log.info("Current record parsed: " + parsedRecordsCounter);
Log.info("Current record downloaded: " + downloadedRecordsCounter);
Log.info("Current record saved: " + savedRecordsCounter);
}
}
long endDownload = System.currentTimeMillis();
long downloadTime = endDownload - startDownload;
Log.info("Download time: " + ((downloadTime / 1000) / 60) + " minutes");
} }
}
long endReq = System.currentTimeMillis();
nReqTmp++;
if (nReqTmp == REQ_LIMIT) {
long reqSessionDuration = endReq - startReqTmp;
if (reqSessionDuration <= 1000) {
Log.warn(
"\nreqSessionDuration: "
+ reqSessionDuration
+ " nReqTmp: "
+ nReqTmp
+ " wait ....");
Thread.sleep(1000 - reqSessionDuration);
} else {
nReqTmp = 0;
startReqTmp = System.currentTimeMillis();
}
}
// if (parsedRecordsCounter>REQ_MAX_TEST) {
// break;
// }
if ((parsedRecordsCounter % RECORD_PARSED_COUNTER_LOG_INTERVAL) == 0) {
Log.info("Current record parsed: " + parsedRecordsCounter);
Log.info("Current record downloaded: " + downloadedRecordsCounter);
Log.info("Current record saved: " + savedRecordsCounter);
}
} }
lambdaFileStream.close(); long endDownload = System.currentTimeMillis();
Log.info("Download started at: " + new Date(startDownload).toString()); long downloadTime = endDownload - startDownload;
Log.info("Parsed Records Counter: " + parsedRecordsCounter); Log.info("Download time: " + ((downloadTime / 1000) / 60) + " minutes");
Log.info("Downloaded Records Counter: " + downloadedRecordsCounter); }
Log.info("Saved Records Counter: " + savedRecordsCounter);
} }
lambdaFileStream.close();
Log.info("Download started at: " + new Date(startDownload).toString());
Log.info("Parsed Records Counter: " + parsedRecordsCounter);
Log.info("Downloaded Records Counter: " + downloadedRecordsCounter);
Log.info("Saved Records Counter: " + savedRecordsCounter);
}
private void loadArgs(String[] args) throws IOException, Exception { private void loadArgs(String[] args) throws IOException, Exception {
final ArgumentApplicationParser parser = final ArgumentApplicationParser parser =
new ArgumentApplicationParser( new ArgumentApplicationParser(
IOUtils.toString( IOUtils.toString(
OrcidDownloader.class.getResourceAsStream( OrcidDownloader.class.getResourceAsStream(
"/eu/dnetlib/dhp/doiboost/download_orcid_data.json"))); "/eu/dnetlib/dhp/doiboost/download_orcid_data.json")));
parser.parseArgument(args); parser.parseArgument(args);
hdfsServerUri = parser.get("hdfsServerUri"); hdfsServerUri = parser.get("hdfsServerUri");
Log.info("HDFS URI: " + hdfsServerUri); Log.info("HDFS URI: " + hdfsServerUri);
hdfsOrcidDefaultPath = parser.get("hdfsOrcidDefaultPath"); hdfsOrcidDefaultPath = parser.get("hdfsOrcidDefaultPath");
Log.info("Default Path: " + hdfsOrcidDefaultPath); Log.info("Default Path: " + hdfsOrcidDefaultPath);
lambdaFileName = parser.get("lambdaFileName"); lambdaFileName = parser.get("lambdaFileName");
Log.info("Lambda File Name: " + lambdaFileName); Log.info("Lambda File Name: " + lambdaFileName);
outputPath = parser.get("outputPath"); outputPath = parser.get("outputPath");
Log.info("Output Data: " + outputPath); Log.info("Output Data: " + outputPath);
token = parser.get("token"); token = parser.get("token");
} }
private boolean isModified(String modifiedDate) throws ParseException { private boolean isModified(String modifiedDate) throws ParseException {
Date modifiedDateDt = new SimpleDateFormat(DATE_FORMAT).parse(modifiedDate); Date modifiedDateDt = new SimpleDateFormat(DATE_FORMAT).parse(modifiedDate);
Date lastUpdateDt = new SimpleDateFormat(DATE_FORMAT).parse(lastUpdate); Date lastUpdateDt = new SimpleDateFormat(DATE_FORMAT).parse(lastUpdate);
return modifiedDateDt.after(lastUpdateDt); return modifiedDateDt.after(lastUpdateDt);
} }
} }