Revision 59095
Added by Alessia Bardi almost 4 years ago
| modules/dnet-collector-plugins/trunk/src/test/java/eu/dnetlib/data/collector/plugins/doiresolver/DOIResolverIteratorTest.java | ||
|---|---|---|
| 1 | 1 |
package eu.dnetlib.data.collector.plugins.doiresolver; |
| 2 | 2 |
|
| 3 |
import org.apache.commons.logging.Log; |
|
| 4 |
import org.apache.commons.logging.LogFactory; |
|
| 3 | 5 |
import org.junit.Assert; |
| 4 | 6 |
import org.junit.Before; |
| 5 | 7 |
import org.junit.Test; |
| ... | ... | |
| 12 | 14 |
@RunWith(MockitoJUnitRunner.class) |
| 13 | 15 |
public class DOIResolverIteratorTest {
|
| 14 | 16 |
|
| 17 |
private static final Log log = LogFactory.getLog(DOIResolverIteratorTest.class); |
|
| 15 | 18 |
@Mock |
| 16 | 19 |
CrossrefResolver resolver; |
| 17 | 20 |
DOIResolverIterator it; |
| 18 | 21 |
|
| 22 |
String dirpath; |
|
| 23 |
|
|
| 19 | 24 |
@Before |
| 20 | 25 |
public void setup(){
|
| 21 | 26 |
when(resolver.resolve("1")).thenReturn("RECORD1");
|
| 22 | 27 |
when(resolver.resolve("2")).thenReturn(null);
|
| 23 | 28 |
when(resolver.resolve("3")).thenReturn("RECORD3");
|
| 24 |
String file = getClass().getResource("/eu/dnetlib/data/collector/plugins/doiresolver/doi_list.csv").getFile();
|
|
| 25 |
it = new DOIResolverIterator(file, resolver); |
|
| 29 |
dirpath = getClass().getResource("/eu/dnetlib/data/collector/plugins/doiresolver").getPath();
|
|
| 30 |
|
|
| 26 | 31 |
} |
| 27 | 32 |
|
| 28 | 33 |
@Test |
| 29 | 34 |
public void test(){
|
| 35 |
it = new DOIResolverIterator(dirpath, resolver, null); |
|
| 36 |
int count = 0; |
|
| 30 | 37 |
while(it.hasNext()){
|
| 31 |
System.out.println(it.next()); |
|
| 38 |
String res = it.next(); |
|
| 39 |
log.info(res); |
|
| 40 |
if(count == 0) Assert.assertEquals("RECORD1", res);
|
|
| 41 |
if(count == 1) Assert.assertEquals("RECORD3", res);
|
|
| 42 |
count++; |
|
| 32 | 43 |
} |
| 44 |
Assert.assertEquals(2, count); |
|
| 33 | 45 |
} |
| 34 | 46 |
|
| 35 | 47 |
@Test |
| 48 |
public void testIncremental(){
|
|
| 49 |
it = new DOIResolverIterator(dirpath, resolver, "2020-07-13"); |
|
| 50 |
int count = 0; |
|
| 51 |
while(it.hasNext()){
|
|
| 52 |
String res = it.next(); |
|
| 53 |
count++; |
|
| 54 |
} |
|
| 55 |
Assert.assertEquals(0, count); |
|
| 56 |
} |
|
| 57 |
|
|
| 58 |
@Test |
|
| 59 |
public void testIncremental2(){
|
|
| 60 |
it = new DOIResolverIterator(dirpath, resolver, "2020-01-13"); |
|
| 61 |
int count = 0; |
|
| 62 |
while(it.hasNext()){
|
|
| 63 |
String res = it.next(); |
|
| 64 |
count++; |
|
| 65 |
} |
|
| 66 |
Assert.assertEquals(2, count); |
|
| 67 |
} |
|
| 68 |
|
|
| 69 |
@Test |
|
| 36 | 70 |
public void testCleanOk(){
|
| 71 |
it = new DOIResolverIterator(dirpath, resolver, null); |
|
| 37 | 72 |
String doi = "10.1234/1234"; |
| 38 | 73 |
Assert.assertEquals(doi, it.cleanDOI(doi)); |
| 39 | 74 |
} |
| 40 | 75 |
|
| 41 | 76 |
@Test |
| 42 | 77 |
public void testCleanHttp(){
|
| 78 |
it = new DOIResolverIterator(dirpath, resolver, null); |
|
| 43 | 79 |
String doi = "10.1234/1234"; |
| 44 | 80 |
String doiURL = "http://dx.doi.org/"+doi; |
| 45 | 81 |
Assert.assertEquals(doi, it.cleanDOI(doiURL)); |
| ... | ... | |
| 47 | 83 |
|
| 48 | 84 |
@Test |
| 49 | 85 |
public void testCleanHttps(){
|
| 86 |
it = new DOIResolverIterator(dirpath, resolver, null); |
|
| 50 | 87 |
String doi = "10.1234/1234"; |
| 51 | 88 |
String doiURL = "https://dx.doi.org/"+doi; |
| 52 | 89 |
Assert.assertEquals(doi, it.cleanDOI(doiURL)); |
| modules/dnet-collector-plugins/trunk/src/test/resources/log4j.properties | ||
|---|---|---|
| 11 | 11 |
log4j.logger.eu.dnetlib.data.collector.plugins.projects.grist=DEBUG |
| 12 | 12 |
log4j.logger.eu.dnetlib.data.collector.plugins.projects.gtr2=DEBUG |
| 13 | 13 |
log4j.logger.eu.dnetlib.data.collector.plugins.doiresolver=DEBUG |
| 14 |
log4j.logger.eu.dnetlib.data.collector.plugins.filesystem=DEBUG |
|
| 14 | 15 |
|
| 15 | 16 |
|
| 16 | 17 |
|
| modules/dnet-collector-plugins/trunk/src/main/java/eu/dnetlib/data/collector/plugins/doiresolver/DOIResolverIterator.java | ||
|---|---|---|
| 1 | 1 |
package eu.dnetlib.data.collector.plugins.doiresolver; |
| 2 | 2 |
|
| 3 |
import eu.dnetlib.data.collector.plugins.filesystem.FileSystemIterator; |
|
| 3 | 4 |
import org.apache.commons.lang.StringUtils; |
| 4 | 5 |
import org.apache.commons.logging.Log; |
| 5 | 6 |
import org.apache.commons.logging.LogFactory; |
| ... | ... | |
| 9 | 10 |
import java.nio.file.Paths; |
| 10 | 11 |
import java.util.Iterator; |
| 11 | 12 |
import java.util.concurrent.ArrayBlockingQueue; |
| 13 |
import java.util.stream.Stream; |
|
| 12 | 14 |
|
| 13 | 15 |
public class DOIResolverIterator implements Iterator<String> {
|
| 14 | 16 |
|
| ... | ... | |
| 19 | 21 |
private static final String BAD_TERMINATOR = "BAD"; |
| 20 | 22 |
private static final String UNRESOLVED = "UNRESOLVED"; |
| 21 | 23 |
|
| 22 |
/** Path to the file that contains a list of DOIs, one per line. **/ |
|
| 23 |
private String filePath; |
|
| 24 |
/** Path to the dir that contains the files, each a csv with a list of DOIs, one per line. **/ |
|
| 25 |
private String baseDir; |
|
| 26 |
private String fromDate; |
|
| 24 | 27 |
|
| 25 | 28 |
private ArrayBlockingQueue<String> queue; |
| 26 | 29 |
|
| 27 | 30 |
private CrossrefResolver crossrefResolver; |
| 28 | 31 |
|
| 29 | 32 |
|
| 30 |
public DOIResolverIterator(final String filePath, final CrossrefResolver crossrefResolver) {
|
|
| 31 |
this.filePath = filePath; |
|
| 33 |
public DOIResolverIterator(final String baseDir, final CrossrefResolver crossrefResolver, final String fromDate) {
|
|
| 34 |
this.baseDir = baseDir; |
|
| 35 |
this.fromDate = fromDate; |
|
| 32 | 36 |
this.queue = new ArrayBlockingQueue<>(100); |
| 33 | 37 |
this.crossrefResolver = crossrefResolver; |
| 34 | 38 |
init(); |
| ... | ... | |
| 36 | 40 |
|
| 37 | 41 |
private void init(){
|
| 38 | 42 |
log.info("Init");
|
| 43 |
|
|
| 39 | 44 |
new Thread(() -> {
|
| 40 |
int count = 0; |
|
| 41 |
// put first item in the queue |
|
| 42 |
if(queue.offer(STARTER)) {
|
|
| 43 |
// read the file, ask the resolvers, put results in a shared queue |
|
| 44 |
//whatever exceptions, add terminator to the queue |
|
| 45 |
try{
|
|
| 46 |
Files.lines(Paths.get(filePath)).forEach(doi -> queue.offer(resolve(doi))); |
|
| 47 |
} catch (IOException e) {
|
|
| 48 |
log.error("DOI processing aborted");
|
|
| 49 |
log.error(e); |
|
| 50 |
queue.offer(BAD_TERMINATOR); |
|
| 45 |
try{
|
|
| 46 |
final FileSystemIterator fsi = new FileSystemIterator(baseDir, "csv", fromDate); |
|
| 47 |
// put first item in the queue |
|
| 48 |
if(queue.offer(STARTER)) {
|
|
| 49 |
// read the file, ask the resolvers, put results in a shared queue |
|
| 50 |
//whatever exceptions, add terminator to the queue |
|
| 51 |
while (fsi.hasNext()) {
|
|
| 52 |
String filePath = fsi.next(); |
|
| 53 |
try (Stream<String> stream = Files.lines(Paths.get(filePath))) {
|
|
| 54 |
|
|
| 55 |
stream.forEach(doi -> queue.offer(resolve(doi))); |
|
| 56 |
|
|
| 57 |
} catch (IOException e) {
|
|
| 58 |
log.error("DOI processing aborted");
|
|
| 59 |
log.error(e); |
|
| 60 |
queue.offer(BAD_TERMINATOR); |
|
| 61 |
} |
|
| 62 |
} |
|
| 51 | 63 |
} |
| 64 |
} catch (Exception e) {
|
|
| 65 |
log.error("DOI processing aborted");
|
|
| 66 |
log.error(e); |
|
| 67 |
queue.offer(BAD_TERMINATOR); |
|
| 52 | 68 |
} |
| 53 | 69 |
queue.offer(TERMINATOR); |
| 54 | 70 |
log.info("Finished processing DOI list");
|
| ... | ... | |
| 107 | 123 |
return queue.poll(); |
| 108 | 124 |
} |
| 109 | 125 |
|
| 110 |
public String getFilePath() {
|
|
| 111 |
return filePath;
|
|
| 126 |
public String getBaseDir() {
|
|
| 127 |
return baseDir;
|
|
| 112 | 128 |
} |
| 113 | 129 |
|
| 114 |
public void setFilePath(String filePath) {
|
|
| 115 |
this.filePath = filePath;
|
|
| 130 |
public void setBaseDir(String baseDir) {
|
|
| 131 |
this.baseDir = baseDir;
|
|
| 116 | 132 |
} |
| 117 | 133 |
|
| 118 | 134 |
public CrossrefResolver getCrossrefResolver() {
|
| modules/dnet-collector-plugins/trunk/src/main/java/eu/dnetlib/data/collector/plugins/doiresolver/DOIResolverPlugin.java | ||
|---|---|---|
| 11 | 11 |
|
| 12 | 12 |
@Override |
| 13 | 13 |
public Iterable<String> collect(final InterfaceDescriptor interfaceDescriptor, final String fromDate, final String untilDate) throws CollectorServiceException {
|
| 14 |
//check baseurl not blank |
|
| 15 |
return () -> new DOIResolverIterator(interfaceDescriptor.getBaseUrl(), crossrefResolver); |
|
| 14 |
final String baseUrl = interfaceDescriptor.getBaseUrl(); |
|
| 15 |
if ((baseUrl == null) || baseUrl.isEmpty()) {
|
|
| 16 |
throw new CollectorServiceException("Param 'baseurl' is null or empty");
|
|
| 17 |
} |
|
| 18 |
if (fromDate != null && !fromDate.matches("\\d{4}-\\d{2}-\\d{2}")) { throw new CollectorServiceException("Invalid date (YYYY-MM-DD): " + fromDate); }
|
|
| 19 |
return () -> new DOIResolverIterator(interfaceDescriptor.getBaseUrl(), crossrefResolver, fromDate); |
|
| 16 | 20 |
} |
| 17 | 21 |
|
| 18 | 22 |
public CrossrefResolver getCrossrefResolver() {
|
Also available in: Unified diff
DOIResolver plugin now supports multiple csv files in the input folder (baseURL) and incremental