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

improved crossref mapping

This commit is contained in:
Sandro La Bruzzo 2020-04-23 09:33:48 +02:00
parent bb6c9785b4
commit 4ba386d996
13 changed files with 1383 additions and 1048 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

86
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/crossref/Crossref2Oaf.scala
View File

@ -10,10 +10,11 @@ import org.json4s.jackson.JsonMethods._
import org.slf4j.Logger import org.slf4j.Logger
import scala.collection.JavaConverters._ import scala.collection.JavaConverters._
case class mappingAffiliation(name:String) case class mappingAffiliation(name: String)
case class mappingAuthor(given: Option[String], family: String, ORCID: Option[String], affiliation:Option[mappingAffiliation]) {} case class mappingAuthor(given: Option[String], family: String, ORCID: Option[String], affiliation: Option[mappingAffiliation]) {}
class Crossref2Oaf {
case object Crossref2Oaf {
//STATIC STRING //STATIC STRING
val MAG = "MAG" val MAG = "MAG"
@ -28,7 +29,6 @@ class Crossref2Oaf {
val DNET_LANGUAGES = "dnet:languages" val DNET_LANGUAGES = "dnet:languages"
val PID_TYPES = "dnet:pid_types" val PID_TYPES = "dnet:pid_types"
val mappingCrossrefType = Map( val mappingCrossrefType = Map(
"book-section" -> "publication", "book-section" -> "publication",
"book" -> "publication", "book" -> "publication",
@ -84,7 +84,7 @@ class Crossref2Oaf {
) )
def mappingResult(result: Result, json: JValue, cobjCategory:String): Result = { def mappingResult(result: Result, json: JValue, cobjCategory: String): Result = {
implicit lazy val formats: DefaultFormats.type = org.json4s.DefaultFormats implicit lazy val formats: DefaultFormats.type = org.json4s.DefaultFormats
//MAPPING Crossref DOI into PID //MAPPING Crossref DOI into PID
@ -111,7 +111,7 @@ class Crossref2Oaf {
result.setCollectedfrom(List(createCollectedFrom()).asJava) result.setCollectedfrom(List(createCollectedFrom()).asJava)
// Publisher ( Name of work's publisher mapped into Result/Publisher) // Publisher ( Name of work's publisher mapped into Result/Publisher)
val publisher = (json \ "publisher").extract[String] val publisher = (json \ "publisher").extractOrElse[String](null)
result.setPublisher(asField(publisher)) result.setPublisher(asField(publisher))
// TITLE // TITLE
@ -144,7 +144,7 @@ class Crossref2Oaf {
//Mapping AUthor //Mapping AUthor
val authorList:List[mappingAuthor] = (json \ "author").extract[List[mappingAuthor]] val authorList: List[mappingAuthor] = (json \ "author").extractOrElse[List[mappingAuthor]](List())
result.setAuthor(authorList.map(a => generateAuhtor(a.given.orNull, a.family, a.ORCID.orNull)).asJava) result.setAuthor(authorList.map(a => generateAuhtor(a.given.orNull, a.family, a.ORCID.orNull)).asJava)
@ -152,8 +152,8 @@ class Crossref2Oaf {
val instance = new Instance() val instance = new Instance()
val license = for { val license = for {
JString(lic) <- json \ "license" \ "URL" JString(lic) <- json \ "license" \ "URL"
} yield asField(lic) } yield asField(lic)
val l = license.filter(d => StringUtils.isNotBlank(d.getValue)) val l = license.filter(d => StringUtils.isNotBlank(d.getValue))
if (l.nonEmpty) if (l.nonEmpty)
instance.setLicense(l.head) instance.setLicense(l.head)
@ -161,24 +161,22 @@ class Crossref2Oaf {
instance.setAccessright(createQualifier("Restricted", "dnet:access_modes")) instance.setAccessright(createQualifier("Restricted", "dnet:access_modes"))
result.setInstance(List(instance).asJava) result.setInstance(List(instance).asJava)
instance.setInstancetype(createQualifier(cobjCategory.substring(0,4), cobjCategory.substring(5), "dnet:publication_resource", "dnet:publication_resource")) instance.setInstancetype(createQualifier(cobjCategory.substring(0, 4), cobjCategory.substring(5), "dnet:publication_resource", "dnet:publication_resource"))
instance.setCollectedfrom(createCollectedFrom()) instance.setCollectedfrom(createCollectedFrom())
if (StringUtils.isNotBlank(issuedDate)) { if (StringUtils.isNotBlank(issuedDate)) {
instance.setDateofacceptance(asField(issuedDate)) instance.setDateofacceptance(asField(issuedDate))
} }
val s: String =(json \ "URL").extract[String] val s: String = (json \ "URL").extract[String]
val links:List[String] = ((for {JString(url) <-json \ "link" \ "URL"} yield url) ::: List(s)).filter(p =>p != null).distinct val links: List[String] = ((for {JString(url) <- json \ "link" \ "URL"} yield url) ::: List(s)).filter(p => p != null).distinct
if (links.nonEmpty) if (links.nonEmpty)
instance.setUrl(links.asJava) instance.setUrl(links.asJava)
result result
} }
def generateAuhtor(given: String, family: String, orcid: String): Author = {
val a = new Author
def generateAuhtor(given:String, family:String, orcid:String):Author = {
val a =new Author
a.setName(given) a.setName(given)
a.setSurname(family) a.setSurname(family)
a.setFullname(s"${given} ${family}") a.setFullname(s"${given} ${family}")
@ -202,30 +200,28 @@ class Crossref2Oaf {
if (result == null) if (result == null)
return result return result
val cOBJCategory = mappingCrossrefSubType.getOrElse(objectType, mappingCrossrefSubType.getOrElse(objectSubType, "0038 Other literature type")); val cOBJCategory = mappingCrossrefSubType.getOrElse(objectType, mappingCrossrefSubType.getOrElse(objectSubType, "0038 Other literature type"));
logger.debug(mappingCrossrefType(objectType)) // logger.debug(mappingCrossrefType(objectType))
logger.debug(cOBJCategory) // logger.debug(cOBJCategory)
mappingResult(result, json, cOBJCategory) mappingResult(result, json, cOBJCategory)
result match { result match {
case publication: Publication => convertPublication(publication) case publication: Publication => convertPublication(publication, json, cOBJCategory)
case dataset: Dataset => convertDataset(dataset) case dataset: Dataset => convertDataset(dataset)
} }
result result
} }
def convertDataset(dataset: Dataset): Unit = { def convertDataset(dataset: Dataset): Unit = {
//TODO probably we need to add relation and other stuff here
} }
def convertPublication(publication: Publication, json: JValue, cobjCategory:String): Unit = { def convertPublication(publication: Publication, json: JValue, cobjCategory: String): Unit = {
implicit lazy val formats: DefaultFormats.type = org.json4s.DefaultFormats
val containerTitles = for {JString(ct) <- json \ "container-title"} yield ct val containerTitles = for {JString(ct) <- json \ "container-title"} yield ct
@ -243,12 +239,44 @@ class Crossref2Oaf {
publication.setSource(List(asField(source)).asJava) publication.setSource(List(asField(source)).asJava)
} }
} else { } else {
val issn = // Mapping Journal
val issnInfos = for {JArray(issn_types) <- json \ "issn-type"
JObject(issn_type) <- issn_types
JField("type", JString(tp)) <- issn_type
JField("value", JString(vl)) <- issn_type
} yield Tuple2(tp, vl)
val volume = (json \ "volume").extractOrElse[String] (null)
if (containerTitles.nonEmpty) {
val journal = new Journal
journal.setName(containerTitles.head)
if (issnInfos.nonEmpty) {
issnInfos.foreach(tp => {
tp._1 match {
case "electronic" => journal.setIssnOnline(tp._2)
case "print" => journal.setIssnPrinted(tp._2)
}
})
}
journal.setVol(volume)
val page = (json \ "page").extractOrElse[String] (null)
if(page!= null ) {
val pp = page.split("-")
journal.setSp(pp.head)
if (pp.size > 1)
journal.setEp(pp(1))
}
publication.setJournal(journal)
}
} }
// Mapping other types of publications
} }
@ -322,7 +350,7 @@ class Crossref2Oaf {
} }
def createQualifier(clsName: String,clsValue: String, schName: String, schValue: String): Qualifier = { def createQualifier(clsName: String, clsValue: String, schName: String, schValue: String): Qualifier = {
val q = new Qualifier val q = new Qualifier
q.setClassid(clsName) q.setClassid(clsName)
q.setClassname(clsValue) q.setClassname(clsValue)

53
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/crossref/CrossrefImporter.java
View File

@ -1,6 +1,8 @@
package eu.dnetlib.doiboost.crossref; package eu.dnetlib.doiboost.crossref;
import eu.dnetlib.dhp.application.ArgumentApplicationParser; import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import java.io.ByteArrayOutputStream;
import java.util.zip.Inflater;
import org.apache.commons.codec.binary.Base64; import org.apache.commons.codec.binary.Base64;
import org.apache.commons.io.IOUtils; import org.apache.commons.io.IOUtils;
import org.apache.commons.lang3.StringUtils; import org.apache.commons.lang3.StringUtils;
@ -12,30 +14,29 @@ import org.apache.hadoop.io.Text;
import org.slf4j.Logger; import org.slf4j.Logger;
import org.slf4j.LoggerFactory; import org.slf4j.LoggerFactory;
import java.io.ByteArrayOutputStream;
import java.util.zip.Inflater;
public class CrossrefImporter { public class CrossrefImporter {
public static void main(String[] args) throws Exception { public static void main(String[] args) throws Exception {
final ArgumentApplicationParser parser = new ArgumentApplicationParser(IOUtils.toString(CrossrefImporter.class.getResourceAsStream("/eu/dnetlib/dhp/doiboost/import_from_es.json"))); final ArgumentApplicationParser parser =
new ArgumentApplicationParser(
IOUtils.toString(
CrossrefImporter.class.getResourceAsStream(
"/eu/dnetlib/dhp/doiboost/import_from_es.json")));
Logger logger = LoggerFactory.getLogger(CrossrefImporter.class); Logger logger = LoggerFactory.getLogger(CrossrefImporter.class);
parser.parseArgument(args); parser.parseArgument(args);
final String hdfsuri = parser.get("namenode"); final String hdfsuri = parser.get("namenode");
logger.info("HDFS URI"+hdfsuri); logger.info("HDFS URI" + hdfsuri);
Path hdfswritepath = new Path(parser.get("targetPath")); Path hdfswritepath = new Path(parser.get("targetPath"));
logger.info("TargetPath: "+hdfsuri); logger.info("TargetPath: " + hdfsuri);
final Long timestamp = StringUtils.isNotBlank(parser.get("timestamp"))?Long.parseLong(parser.get("timestamp")):-1; final Long timestamp =
StringUtils.isNotBlank(parser.get("timestamp"))
if(timestamp>0) ? Long.parseLong(parser.get("timestamp"))
logger.info("Timestamp added "+timestamp); : -1;
if (timestamp > 0) logger.info("Timestamp added " + timestamp);
// ====== Init HDFS File System Object // ====== Init HDFS File System Object
Configuration conf = new Configuration(); Configuration conf = new Configuration();
@ -45,16 +46,21 @@ public class CrossrefImporter {
conf.set("fs.hdfs.impl", org.apache.hadoop.hdfs.DistributedFileSystem.class.getName()); conf.set("fs.hdfs.impl", org.apache.hadoop.hdfs.DistributedFileSystem.class.getName());
conf.set("fs.file.impl", org.apache.hadoop.fs.LocalFileSystem.class.getName()); conf.set("fs.file.impl", org.apache.hadoop.fs.LocalFileSystem.class.getName());
ESClient client =
timestamp > 0
? new ESClient(
"ip-90-147-167-25.ct1.garrservices.it", "crossref", timestamp)
: new ESClient("ip-90-147-167-25.ct1.garrservices.it", "crossref");
try (SequenceFile.Writer writer =
ESClient client = timestamp>0?new ESClient("ip-90-147-167-25.ct1.garrservices.it", "crossref", timestamp):new ESClient("ip-90-147-167-25.ct1.garrservices.it", "crossref"); SequenceFile.createWriter(
conf,
try (SequenceFile.Writer writer = SequenceFile.createWriter(conf, SequenceFile.Writer.file(hdfswritepath),
SequenceFile.Writer.file(hdfswritepath), SequenceFile.Writer.keyClass(IntWritable.class), SequenceFile.Writer.keyClass(IntWritable.class),
SequenceFile.Writer.valueClass(Text.class))) { SequenceFile.Writer.valueClass(Text.class))) {
int i = 0; int i = 0;
long start= System.currentTimeMillis(); long start = System.currentTimeMillis();
long end = 0; long end = 0;
final IntWritable key = new IntWritable(i); final IntWritable key = new IntWritable(i);
final Text value = new Text(); final Text value = new Text();
@ -65,7 +71,10 @@ public class CrossrefImporter {
if (i % 1000000 == 0) { if (i % 1000000 == 0) {
end = System.currentTimeMillis(); end = System.currentTimeMillis();
final float time = (end - start) / 1000.0F; final float time = (end - start) / 1000.0F;
logger.info(String.format("Imported %d records last 100000 imported in %f seconds", i, time)); logger.info(
String.format(
"Imported %d records last 100000 imported in %f seconds",
i, time));
start = System.currentTimeMillis(); start = System.currentTimeMillis();
} }
} }
@ -87,7 +96,7 @@ public class CrossrefImporter {
decompresser.end(); decompresser.end();
return new String(unzippeddata); return new String(unzippeddata);
} catch (Throwable e) { } catch (Throwable e) {
throw new RuntimeException("Wrong record:" + blob,e); throw new RuntimeException("Wrong record:" + blob, e);
} }
} }
} }

63
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/crossref/ESClient.java
View File

@ -1,6 +1,9 @@
package eu.dnetlib.doiboost.crossref; package eu.dnetlib.doiboost.crossref;
import com.jayway.jsonpath.JsonPath; import com.jayway.jsonpath.JsonPath;
import java.io.IOException;
import java.util.Iterator;
import java.util.List;
import org.apache.commons.io.IOUtils; import org.apache.commons.io.IOUtils;
import org.apache.http.client.methods.CloseableHttpResponse; import org.apache.http.client.methods.CloseableHttpResponse;
import org.apache.http.client.methods.HttpPost; import org.apache.http.client.methods.HttpPost;
@ -10,18 +13,15 @@ import org.apache.http.impl.client.HttpClients;
import org.slf4j.Logger; import org.slf4j.Logger;
import org.slf4j.LoggerFactory; import org.slf4j.LoggerFactory;
import java.io.IOException;
import java.util.Iterator;
import java.util.List;
public class ESClient implements Iterator<String> { public class ESClient implements Iterator<String> {
private final static Logger logger = LoggerFactory.getLogger(ESClient.class); private static final Logger logger = LoggerFactory.getLogger(ESClient.class);
final static String blobPath = "$.hits[*].hits[*]._source.blob"; static final String blobPath = "$.hits[*].hits[*]._source.blob";
final static String scrollIdPath = "$._scroll_id"; static final String scrollIdPath = "$._scroll_id";
final static String JSON_NO_TS ="{\"size\":1000}"; static final String JSON_NO_TS = "{\"size\":1000}";
final static String JSON_WITH_TS ="{\"size\":1000, \"query\":{\"range\":{\"timestamp\":{\"gte\":%d}}}}"; static final String JSON_WITH_TS =
final static String JSON_SCROLL = "{\"scroll_id\":\"%s\",\"scroll\" : \"1m\"}"; "{\"size\":1000, \"query\":{\"range\":{\"timestamp\":{\"gte\":%d}}}}";
static final String JSON_SCROLL = "{\"scroll_id\":\"%s\",\"scroll\" : \"1m\"}";
private final String scrollId; private final String scrollId;
@ -29,29 +29,34 @@ public class ESClient implements Iterator<String> {
private final String esHost; private final String esHost;
public ESClient(final String esHost, final String esIndex) throws IOException { public ESClient(final String esHost, final String esIndex) throws IOException {
this.esHost = esHost; this.esHost = esHost;
final String body =getResponse(String.format("http://%s:9200/%s/_search?scroll=1m", esHost, esIndex), JSON_NO_TS); final String body =
scrollId= getJPathString(scrollIdPath, body); getResponse(
String.format("http://%s:9200/%s/_search?scroll=1m", esHost, esIndex),
JSON_NO_TS);
scrollId = getJPathString(scrollIdPath, body);
buffer = getBlobs(body); buffer = getBlobs(body);
} }
public ESClient(final String esHost, final String esIndex, final long timestamp)
public ESClient(final String esHost, final String esIndex, final long timestamp) throws IOException { throws IOException {
this.esHost = esHost; this.esHost = esHost;
final String body =getResponse(String.format("http://%s:9200/%s/_search?scroll=1m", esHost, esIndex), String.format(JSON_WITH_TS, timestamp)); final String body =
scrollId= getJPathString(scrollIdPath, body); getResponse(
String.format("http://%s:9200/%s/_search?scroll=1m", esHost, esIndex),
String.format(JSON_WITH_TS, timestamp));
scrollId = getJPathString(scrollIdPath, body);
buffer = getBlobs(body); buffer = getBlobs(body);
} }
private String getResponse(final String url,final String json ) { private String getResponse(final String url, final String json) {
CloseableHttpClient client = HttpClients.createDefault(); CloseableHttpClient client = HttpClients.createDefault();
try { try {
HttpPost httpPost = new HttpPost(url); HttpPost httpPost = new HttpPost(url);
if (json!= null) { if (json != null) {
StringEntity entity = new StringEntity(json); StringEntity entity = new StringEntity(json);
httpPost.setEntity(entity); httpPost.setEntity(entity);
httpPost.setHeader("Accept", "application/json"); httpPost.setHeader("Accept", "application/json");
@ -61,22 +66,20 @@ public class ESClient implements Iterator<String> {
return IOUtils.toString(response.getEntity().getContent()); return IOUtils.toString(response.getEntity().getContent());
} catch (Throwable e) { } catch (Throwable e) {
throw new RuntimeException("Error on executing request ",e); throw new RuntimeException("Error on executing request ", e);
} finally { } finally {
try { try {
client.close(); client.close();
} catch (IOException e) { } catch (IOException e) {
throw new RuntimeException("Unable to close client ",e); throw new RuntimeException("Unable to close client ", e);
} }
} }
} }
private String getJPathString(final String jsonPath, final String json) { private String getJPathString(final String jsonPath, final String json) {
try { try {
Object o = JsonPath.read(json, jsonPath); Object o = JsonPath.read(json, jsonPath);
if (o instanceof String) if (o instanceof String) return (String) o;
return (String) o;
return null; return null;
} catch (Exception e) { } catch (Exception e) {
return ""; return "";
@ -84,14 +87,13 @@ public class ESClient implements Iterator<String> {
} }
private List<String> getBlobs(final String body) { private List<String> getBlobs(final String body) {
final List<String > res = JsonPath.read(body, "$.hits.hits[*]._source.blob"); final List<String> res = JsonPath.read(body, "$.hits.hits[*]._source.blob");
return res; return res;
} }
@Override @Override
public boolean hasNext() { public boolean hasNext() {
return (buffer!= null && !buffer.isEmpty()); return (buffer != null && !buffer.isEmpty());
} }
@Override @Override
@ -100,11 +102,12 @@ public class ESClient implements Iterator<String> {
if (buffer.isEmpty()) { if (buffer.isEmpty()) {
final String json_param = String.format(JSON_SCROLL, scrollId); final String json_param = String.format(JSON_SCROLL, scrollId);
final String body =getResponse(String.format("http://%s:9200/_search/scroll", esHost), json_param); final String body =
getResponse(String.format("http://%s:9200/_search/scroll", esHost), json_param);
try { try {
buffer = getBlobs(body); buffer = getBlobs(body);
} catch (Throwable e) { } catch (Throwable e) {
logger.error("Error on get next page: body:"+body); logger.error("Error on get next page: body:" + body);
} }
} }
return nextItem; return nextItem;

59
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/crossref/SparkMapDumpIntoOAF.scala
View File

@ -1,22 +1,22 @@
package eu.dnetlib.doiboost.crossref package eu.dnetlib.doiboost.crossref
import eu.dnetlib.dhp.application.ArgumentApplicationParser import eu.dnetlib.dhp.application.ArgumentApplicationParser
import eu.dnetlib.dhp.schema.oaf.Publication
import org.apache.commons.io.IOUtils import org.apache.commons.io.IOUtils
import org.apache.hadoop.io.{IntWritable, Text} import org.apache.hadoop.io.{IntWritable, Text}
import org.apache.spark.SparkConf import org.apache.spark.SparkConf
import org.apache.spark.sql.SparkSession import org.apache.spark.sql.{Dataset, Encoders, SaveMode, SparkSession}
import org.slf4j.{Logger, LoggerFactory} import org.slf4j.{Logger, LoggerFactory}
case class Reference(author: String, firstPage: String) {}
case class Reference(author:String, firstPage:String) {}
object SparkMapDumpIntoOAF { object SparkMapDumpIntoOAF {
def main(args: Array[String]): Unit = { def main(args: Array[String]): Unit = {
val logger:Logger = LoggerFactory.getLogger(SparkMapDumpIntoOAF.getClass) val logger: Logger = LoggerFactory.getLogger(SparkMapDumpIntoOAF.getClass)
val conf: SparkConf = new SparkConf() val conf: SparkConf = new SparkConf()
val parser = new ArgumentApplicationParser(IOUtils.toString(SparkMapDumpIntoOAF.getClass.getResourceAsStream("/eu/dnetlib/dhp/doiboost/convert_map_to_oaf_params.json"))) val parser = new ArgumentApplicationParser(IOUtils.toString(SparkMapDumpIntoOAF.getClass.getResourceAsStream("/eu/dnetlib/dhp/doiboost/convert_map_to_oaf_params.json")))
parser.parseArgument(args) parser.parseArgument(args)
@ -26,33 +26,42 @@ object SparkMapDumpIntoOAF {
.config(conf) .config(conf)
.appName(SparkMapDumpIntoOAF.getClass.getSimpleName) .appName(SparkMapDumpIntoOAF.getClass.getSimpleName)
.master(parser.get("master")).getOrCreate() .master(parser.get("master")).getOrCreate()
import spark.implicits._
implicit val mapEncoder = Encoders.bean(classOf[Publication])
val sc = spark.sparkContext val sc = spark.sparkContext
val x: String = sc.sequenceFile(parser.get("sourcePath"), classOf[IntWritable], classOf[Text])
.map(k => k._2.toString).first()
val item =CrossrefImporter.decompressBlob(x) val total = sc.sequenceFile(parser.get("sourcePath"), classOf[IntWritable], classOf[Text])
.map(k => k._2.toString).map(CrossrefImporter.decompressBlob)
.map(k => Crossref2Oaf.convert(k, logger))
.filter(k => k != null && k.isInstanceOf[Publication])
.map(k => k.asInstanceOf[Publication])
logger.info(item) val ds: Dataset[Publication] = spark.createDataset(total)
val targetPath = parser.get("targetPath")
ds.write.mode(SaveMode.Overwrite).save(s"${targetPath}/publication")
// lazy val json: json4s.JValue = parse(item)
// logger.info(s"total Item :${total}")
//
// val references = for { // lazy val json: json4s.JValue = parse(item)
// JArray(references) <- json \\ "reference" //
// JObject(reference) <- references //
// JField("first-page", JString(firstPage)) <- reference // val references = for {
// JField("author", JString(author)) <- reference // JArray(references) <- json \\ "reference"
// } yield Reference(author, firstPage) // JObject(reference) <- references
// // JField("first-page", JString(firstPage)) <- reference
// // JField("author", JString(author)) <- reference
// // } yield Reference(author, firstPage)
// //
// logger.info((json \ "created" \ "timestamp").extractOrElse("missing")) //
// logger.info(references.toString()) //
// //
// logger.info((json \ "type").extractOrElse("missing")) // logger.info((json \ "created" \ "timestamp").extractOrElse("missing"))
// logger.info(references.toString())
//
// logger.info((json \ "type").extractOrElse("missing"))
} }

22
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/orcid/json/JsonWriter.java
View File

@ -1,20 +1,18 @@
package eu.dnetlib.doiboost.orcid.json; package eu.dnetlib.doiboost.orcid.json;
import com.google.gson.JsonObject; import com.google.gson.JsonObject;
import eu.dnetlib.doiboost.orcid.model.AuthorData; import eu.dnetlib.doiboost.orcid.model.AuthorData;
public class JsonWriter { public class JsonWriter {
public static String create(AuthorData authorData) { public static String create(AuthorData authorData) {
JsonObject author = new JsonObject(); JsonObject author = new JsonObject();
author.addProperty("oid", authorData.getOid()); author.addProperty("oid", authorData.getOid());
author.addProperty("name", authorData.getName()); author.addProperty("name", authorData.getName());
author.addProperty("surname", authorData.getSurname()); author.addProperty("surname", authorData.getSurname());
if (authorData.getCreditName()!=null) { if (authorData.getCreditName() != null) {
author.addProperty("creditname", authorData.getCreditName()); author.addProperty("creditname", authorData.getCreditName());
} }
return author.toString(); return author.toString();
} }
} }

79
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/orcid/model/AuthorData.java
View File

@ -2,40 +2,49 @@ package eu.dnetlib.doiboost.orcid.model;
public class AuthorData { public class AuthorData {
private String oid; private String oid;
private String name; private String name;
private String surname; private String surname;
private String creditName; private String creditName;
private String errorCode; private String errorCode;
public String getErrorCode() { public String getErrorCode() {
return errorCode; return errorCode;
} }
public void setErrorCode(String errorCode) {
this.errorCode = errorCode; public void setErrorCode(String errorCode) {
} this.errorCode = errorCode;
public String getName() { }
return name;
} public String getName() {
public void setName(String name) { return name;
this.name = name; }
}
public String getSurname() { public void setName(String name) {
return surname; this.name = name;
} }
public void setSurname(String surname) {
this.surname = surname; public String getSurname() {
} return surname;
public String getCreditName() { }
return creditName;
} public void setSurname(String surname) {
public void setCreditName(String creditName) { this.surname = surname;
this.creditName = creditName; }
}
public String getOid() { public String getCreditName() {
return oid; return creditName;
} }
public void setOid(String oid) {
this.oid = oid; public void setCreditName(String creditName) {
} this.creditName = creditName;
}
public String getOid() {
return oid;
}
public void setOid(String oid) {
this.oid = oid;
}
} }

1
dhp-workflows/dhp-doiboost/src/main/resources/eu/dnetlib/dhp/doiboost/convert_map_to_oaf_params.json
View File

@ -1,5 +1,6 @@
[ [
{"paramName":"s", "paramLongName":"sourcePath", "paramDescription": "the path of the sequencial file to read", "paramRequired": true}, {"paramName":"s", "paramLongName":"sourcePath", "paramDescription": "the path of the sequencial file to read", "paramRequired": true},
{"paramName":"t", "paramLongName":"targetPath", "paramDescription": "the working dir path", "paramRequired": true},
{"paramName":"m", "paramLongName":"master", "paramDescription": "the master name", "paramRequired": true} {"paramName":"m", "paramLongName":"master", "paramDescription": "the master name", "paramRequired": true}
] ]

100
dhp-workflows/dhp-doiboost/src/test/java/eu/dnetlib/doiboost/DoiBoostTest.java
View File

@ -26,18 +26,98 @@ public class DoiBoostTest {
// CrossrefImporter.main("-n file:///tmp -t file:///tmp/p.seq -ts 1586110000749".split(" // CrossrefImporter.main("-n file:///tmp -t file:///tmp/p.seq -ts 1586110000749".split("
// ")); // "));
SparkMapDumpIntoOAF.main( SparkMapDumpIntoOAF.main(
"-m local[*] -s file:///data/doiboost/crossref_dump.seq".split(" ")); "-m local[*] -s file:///data/doiboost/crossref_dump.seq -t /data/doiboost"
.split(" "));
}
@Test
public void testConvertDatasetCrossRef2Oaf() throws IOException {
final String json = IOUtils.toString(getClass().getResourceAsStream("dataset.json"));
ObjectMapper mapper = new ObjectMapper().enable(SerializationFeature.INDENT_OUTPUT);
assertNotNull(json);
assertFalse(StringUtils.isBlank(json));
final Result result = Crossref2Oaf.convert(json, logger);
logger.info(mapper.writeValueAsString(result));
} }
@Test @Test
public void testConvertPreprintCrossRef2Oaf() throws IOException { public void testConvertPreprintCrossRef2Oaf() throws IOException {
final String json = IOUtils.toString(getClass().getResourceAsStream("article.json")); final String json = IOUtils.toString(getClass().getResourceAsStream("preprint.json"));
ObjectMapper mapper = new ObjectMapper().enable(SerializationFeature.INDENT_OUTPUT); ObjectMapper mapper = new ObjectMapper().enable(SerializationFeature.INDENT_OUTPUT);
assertNotNull(json); assertNotNull(json);
assertFalse(StringUtils.isBlank(json)); assertFalse(StringUtils.isBlank(json));
Crossref2Oaf cf = new Crossref2Oaf();
final Result result = cf.convert(json, logger); final Result result = Crossref2Oaf.convert(json, logger);
assertNotNull(result);
assertNotNull(result.getDataInfo(), "Datainfo test not null Failed");
assertNotNull(
result.getDataInfo().getProvenanceaction(),
"DataInfo/Provenance test not null Failed");
assertFalse(
StringUtils.isBlank(result.getDataInfo().getProvenanceaction().getClassid()),
"DataInfo/Provenance/classId test not null Failed");
assertFalse(
StringUtils.isBlank(result.getDataInfo().getProvenanceaction().getClassname()),
"DataInfo/Provenance/className test not null Failed");
assertFalse(
StringUtils.isBlank(result.getDataInfo().getProvenanceaction().getSchemeid()),
"DataInfo/Provenance/SchemeId test not null Failed");
assertFalse(
StringUtils.isBlank(result.getDataInfo().getProvenanceaction().getSchemename()),
"DataInfo/Provenance/SchemeName test not null Failed");
assertNotNull(result.getCollectedfrom(), "CollectedFrom test not null Failed");
assertTrue(result.getCollectedfrom().size() > 0);
assertTrue(
result.getCollectedfrom().stream()
.anyMatch(
c ->
c.getKey()
.equalsIgnoreCase(
"10|openaire____::081b82f96300b6a6e3d282bad31cb6e2")));
assertTrue(
result.getCollectedfrom().stream()
.anyMatch(c -> c.getValue().equalsIgnoreCase("crossref")));
assertTrue(
result.getRelevantdate().stream()
.anyMatch(d -> d.getQualifier().getClassid().equalsIgnoreCase("created")));
assertTrue(
result.getRelevantdate().stream()
.anyMatch(
d -> d.getQualifier().getClassid().equalsIgnoreCase("available")));
assertTrue(
result.getRelevantdate().stream()
.anyMatch(d -> d.getQualifier().getClassid().equalsIgnoreCase("accepted")));
assertTrue(
result.getRelevantdate().stream()
.anyMatch(
d ->
d.getQualifier()
.getClassid()
.equalsIgnoreCase("published-online")));
assertTrue(
result.getRelevantdate().stream()
.anyMatch(
d ->
d.getQualifier()
.getClassid()
.equalsIgnoreCase("published-print")));
logger.info(mapper.writeValueAsString(result));
}
@Test
public void testConvertArticleCrossRef2Oaf() throws IOException {
final String json = IOUtils.toString(getClass().getResourceAsStream("article.json"));
ObjectMapper mapper = new ObjectMapper().enable(SerializationFeature.INDENT_OUTPUT);
assertNotNull(json);
assertFalse(StringUtils.isBlank(json));
final Result result = Crossref2Oaf.convert(json, logger);
assertNotNull(result); assertNotNull(result);
assertNotNull(result.getDataInfo(), "Datainfo test not null Failed"); assertNotNull(result.getDataInfo(), "Datainfo test not null Failed");
@ -73,15 +153,6 @@ public class DoiBoostTest {
assertTrue( assertTrue(
result.getRelevantdate().stream() result.getRelevantdate().stream()
.anyMatch(d -> d.getQualifier().getClassid().equalsIgnoreCase("created"))); .anyMatch(d -> d.getQualifier().getClassid().equalsIgnoreCase("created")));
// assertTrue(
// result.getRelevantdate().stream()
// .anyMatch(
// d ->
// d.getQualifier().getClassid().equalsIgnoreCase("available")));
// assertTrue(
// result.getRelevantdate().stream()
// .anyMatch(d ->
// d.getQualifier().getClassid().equalsIgnoreCase("accepted")));
assertTrue( assertTrue(
result.getRelevantdate().stream() result.getRelevantdate().stream()
.anyMatch( .anyMatch(
@ -107,8 +178,7 @@ public class DoiBoostTest {
ObjectMapper mapper = new ObjectMapper().enable(SerializationFeature.INDENT_OUTPUT); ObjectMapper mapper = new ObjectMapper().enable(SerializationFeature.INDENT_OUTPUT);
assertNotNull(json); assertNotNull(json);
assertFalse(StringUtils.isBlank(json)); assertFalse(StringUtils.isBlank(json));
Crossref2Oaf cf = new Crossref2Oaf(); final Result result = Crossref2Oaf.convert(json, logger);
final Result result = cf.convert(json, logger);
assertNotNull(result); assertNotNull(result);
logger.info(mapper.writeValueAsString(result)); logger.info(mapper.writeValueAsString(result));

2
dhp-workflows/dhp-doiboost/src/test/resources/eu/dnetlib/doiboost/article.json
View File

@ -170,5 +170,5 @@
"container-title": [ "container-title": [
"Ecl\u00e9tica Qu\u00edmica Journal" "Ecl\u00e9tica Qu\u00edmica Journal"
], ],
"page": "41" "page": "41-50"
} }

105
dhp-workflows/dhp-doiboost/src/test/resources/eu/dnetlib/doiboost/dataset.json Normal file
View File

@ -0,0 +1,105 @@
{
"DOI": "10.1037/e522512014-096",
"subtitle": [
"(522512014-096)"
],
"issued": {
"date-parts": [
[
2012
]
]
},
"prefix": "10.1037",
"author": [
{
"affiliation": [],
"given": "Jessica",
"family": "Trudeau",
"sequence": "first"
},
{
"affiliation": [],
"given": "Amy",
"family": "McShane",
"sequence": "additional"
},
{
"affiliation": [],
"given": "Renee",
"family": "McDonald",
"sequence": "additional"
}
],
"reference-count": 0,
"member": "15",
"source": "Crossref",
"score": 1.0,
"deposited": {
"timestamp": 1413827035000,
"date-parts": [
[
2014,
10,
20
]
],
"date-time": "2014-10-20T17:43:55Z"
},
"indexed": {
"timestamp": 1550142454710,
"date-parts": [
[
2019,
2,
14
]
],
"date-time": "2019-02-14T11:07:34Z"
},
"type": "dataset",
"URL": "http://dx.doi.org/10.1037/e522512014-096",
"is-referenced-by-count": 0,
"published-print": {
"date-parts": [
[
2012
]
]
},
"references-count": 0,
"institution": {
"acronym": [
"APA"
],
"place": [
"-"
],
"name": "American Psychological Association"
},
"publisher": "American Psychological Association (APA)",
"content-domain": {
"domain": [],
"crossmark-restriction": false
},
"created": {
"timestamp": 1413826121000,
"date-parts": [
[
2014,
10,
20
]
],
"date-time": "2014-10-20T17:28:41Z"
},
"title": [
"Project Support: A Randomized Control Study to Evaluate the Translation of an Evidence- Based Program"
],
"alternative-id": [
"522512014-096"
],
"container-title": [
"PsycEXTRA Dataset"
]
}

826
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/AbstractMdRecordToOafMapper.java
View File

@ -10,19 +10,6 @@ import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.oaiIProvenance;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.qualifier; import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.qualifier;
import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.structuredProperty; import static eu.dnetlib.dhp.oa.graph.raw.common.OafMapperUtils.structuredProperty;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Date;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import org.apache.commons.lang3.StringUtils;
import org.dom4j.Document;
import org.dom4j.DocumentFactory;
import org.dom4j.DocumentHelper;
import org.dom4j.Node;
import eu.dnetlib.dhp.oa.graph.raw.common.MigrationConstants; import eu.dnetlib.dhp.oa.graph.raw.common.MigrationConstants;
import eu.dnetlib.dhp.schema.oaf.Author; import eu.dnetlib.dhp.schema.oaf.Author;
import eu.dnetlib.dhp.schema.oaf.DataInfo; import eu.dnetlib.dhp.schema.oaf.DataInfo;
@ -41,388 +28,439 @@ import eu.dnetlib.dhp.schema.oaf.Relation;
import eu.dnetlib.dhp.schema.oaf.Result; import eu.dnetlib.dhp.schema.oaf.Result;
import eu.dnetlib.dhp.schema.oaf.Software; import eu.dnetlib.dhp.schema.oaf.Software;
import eu.dnetlib.dhp.schema.oaf.StructuredProperty; import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Date;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import org.apache.commons.lang3.StringUtils;
import org.dom4j.Document;
import org.dom4j.DocumentFactory;
import org.dom4j.DocumentHelper;
import org.dom4j.Node;
public abstract class AbstractMdRecordToOafMapper { public abstract class AbstractMdRecordToOafMapper {
protected final Map<String, String> code2name; protected final Map<String, String> code2name;
protected static final Qualifier MAIN_TITLE_QUALIFIER = protected static final Qualifier MAIN_TITLE_QUALIFIER =
qualifier("main title", "main title", "dnet:dataCite_title", "dnet:dataCite_title"); qualifier("main title", "main title", "dnet:dataCite_title", "dnet:dataCite_title");
protected AbstractMdRecordToOafMapper(final Map<String, String> code2name) { protected AbstractMdRecordToOafMapper(final Map<String, String> code2name) {
this.code2name = code2name; this.code2name = code2name;
} }
public List<Oaf> processMdRecord(final String xml) { public List<Oaf> processMdRecord(final String xml) {
try { try {
final Map<String, String> nsContext = new HashMap<>(); final Map<String, String> nsContext = new HashMap<>();
nsContext.put("dr", "http://www.driver-repository.eu/namespace/dr"); nsContext.put("dr", "http://www.driver-repository.eu/namespace/dr");
nsContext.put("dri", "http://www.driver-repository.eu/namespace/dri"); nsContext.put("dri", "http://www.driver-repository.eu/namespace/dri");
nsContext.put("oaf", "http://namespace.openaire.eu/oaf"); nsContext.put("oaf", "http://namespace.openaire.eu/oaf");
nsContext.put("oai", "http://www.openarchives.org/OAI/2.0/"); nsContext.put("oai", "http://www.openarchives.org/OAI/2.0/");
nsContext.put("prov", "http://www.openarchives.org/OAI/2.0/provenance"); nsContext.put("prov", "http://www.openarchives.org/OAI/2.0/provenance");
nsContext.put("dc", "http://purl.org/dc/elements/1.1/"); nsContext.put("dc", "http://purl.org/dc/elements/1.1/");
nsContext.put("datacite", "http://datacite.org/schema/kernel-3"); nsContext.put("datacite", "http://datacite.org/schema/kernel-3");
DocumentFactory.getInstance().setXPathNamespaceURIs(nsContext); DocumentFactory.getInstance().setXPathNamespaceURIs(nsContext);
final Document doc = final Document doc =
DocumentHelper.parseText(xml.replaceAll("http://datacite.org/schema/kernel-4", "http://datacite.org/schema/kernel-3")); DocumentHelper.parseText(
xml.replaceAll(
final String type = doc.valueOf("//dr:CobjCategory/@type"); "http://datacite.org/schema/kernel-4",
final KeyValue collectedFrom = "http://datacite.org/schema/kernel-3"));
keyValue(createOpenaireId(10, doc.valueOf("//oaf:collectedFrom/@id"), true), doc.valueOf("//oaf:collectedFrom/@name"));
final KeyValue hostedBy = final String type = doc.valueOf("//dr:CobjCategory/@type");
StringUtils.isBlank(doc.valueOf("//oaf:hostedBy/@id")) final KeyValue collectedFrom =
? collectedFrom keyValue(
: keyValue(createOpenaireId(10, doc.valueOf("//oaf:hostedBy/@id"), true), doc.valueOf("//oaf:hostedBy/@name")); createOpenaireId(10, doc.valueOf("//oaf:collectedFrom/@id"), true),
doc.valueOf("//oaf:collectedFrom/@name"));
final DataInfo info = prepareDataInfo(doc); final KeyValue hostedBy =
final long lastUpdateTimestamp = new Date().getTime(); StringUtils.isBlank(doc.valueOf("//oaf:hostedBy/@id"))
? collectedFrom
return createOafs(doc, type, collectedFrom, hostedBy, info, lastUpdateTimestamp); : keyValue(
} catch (final Exception e) { createOpenaireId(10, doc.valueOf("//oaf:hostedBy/@id"), true),
throw new RuntimeException(e); doc.valueOf("//oaf:hostedBy/@name"));
}
} final DataInfo info = prepareDataInfo(doc);
final long lastUpdateTimestamp = new Date().getTime();
protected List<Oaf> createOafs(
final Document doc, return createOafs(doc, type, collectedFrom, hostedBy, info, lastUpdateTimestamp);
final String type, } catch (final Exception e) {
final KeyValue collectedFrom, throw new RuntimeException(e);
final KeyValue hostedBy, }
final DataInfo info, }
final long lastUpdateTimestamp) {
protected List<Oaf> createOafs(
final List<Oaf> oafs = new ArrayList<>(); final Document doc,
final String type,
switch (type.toLowerCase()) { final KeyValue collectedFrom,
case "": final KeyValue hostedBy,
case "publication": final DataInfo info,
final Publication p = new Publication(); final long lastUpdateTimestamp) {
populateResultFields(p, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
p.setResulttype(MigrationConstants.PUBLICATION_RESULTTYPE_QUALIFIER); final List<Oaf> oafs = new ArrayList<>();
p.setJournal(prepareJournal(doc, info));
oafs.add(p); switch (type.toLowerCase()) {
break; case "":
case "dataset": case "publication":
final Dataset d = new Dataset(); final Publication p = new Publication();
populateResultFields(d, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp); populateResultFields(p, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
d.setResulttype(MigrationConstants.DATASET_RESULTTYPE_QUALIFIER); p.setResulttype(MigrationConstants.PUBLICATION_RESULTTYPE_QUALIFIER);
d.setStoragedate(prepareDatasetStorageDate(doc, info)); p.setJournal(prepareJournal(doc, info));
d.setDevice(prepareDatasetDevice(doc, info)); oafs.add(p);
d.setSize(prepareDatasetSize(doc, info)); break;
d.setVersion(prepareDatasetVersion(doc, info)); case "dataset":
d.setLastmetadataupdate(prepareDatasetLastMetadataUpdate(doc, info)); final Dataset d = new Dataset();
d.setMetadataversionnumber(prepareDatasetMetadataVersionNumber(doc, info)); populateResultFields(d, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
d.setGeolocation(prepareDatasetGeoLocations(doc, info)); d.setResulttype(MigrationConstants.DATASET_RESULTTYPE_QUALIFIER);
oafs.add(d); d.setStoragedate(prepareDatasetStorageDate(doc, info));
break; d.setDevice(prepareDatasetDevice(doc, info));
case "software": d.setSize(prepareDatasetSize(doc, info));
final Software s = new Software(); d.setVersion(prepareDatasetVersion(doc, info));
populateResultFields(s, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp); d.setLastmetadataupdate(prepareDatasetLastMetadataUpdate(doc, info));
s.setResulttype(MigrationConstants.SOFTWARE_RESULTTYPE_QUALIFIER); d.setMetadataversionnumber(prepareDatasetMetadataVersionNumber(doc, info));
s.setDocumentationUrl(prepareSoftwareDocumentationUrls(doc, info)); d.setGeolocation(prepareDatasetGeoLocations(doc, info));
s.setLicense(prepareSoftwareLicenses(doc, info)); oafs.add(d);
s.setCodeRepositoryUrl(prepareSoftwareCodeRepositoryUrl(doc, info)); break;
s.setProgrammingLanguage(prepareSoftwareProgrammingLanguage(doc, info)); case "software":
oafs.add(s); final Software s = new Software();
break; populateResultFields(s, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
case "otherresearchproducts": s.setResulttype(MigrationConstants.SOFTWARE_RESULTTYPE_QUALIFIER);
default: s.setDocumentationUrl(prepareSoftwareDocumentationUrls(doc, info));
final OtherResearchProduct o = new OtherResearchProduct(); s.setLicense(prepareSoftwareLicenses(doc, info));
populateResultFields(o, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp); s.setCodeRepositoryUrl(prepareSoftwareCodeRepositoryUrl(doc, info));
o.setResulttype(MigrationConstants.OTHER_RESULTTYPE_QUALIFIER); s.setProgrammingLanguage(prepareSoftwareProgrammingLanguage(doc, info));
o.setContactperson(prepareOtherResearchProductContactPersons(doc, info)); oafs.add(s);
o.setContactgroup(prepareOtherResearchProductContactGroups(doc, info)); break;
o.setTool(prepareOtherResearchProductTools(doc, info)); case "otherresearchproducts":
oafs.add(o); default:
break; final OtherResearchProduct o = new OtherResearchProduct();
} populateResultFields(o, doc, collectedFrom, hostedBy, info, lastUpdateTimestamp);
o.setResulttype(MigrationConstants.OTHER_RESULTTYPE_QUALIFIER);
if (!oafs.isEmpty()) { o.setContactperson(prepareOtherResearchProductContactPersons(doc, info));
oafs.addAll(addProjectRels(doc, collectedFrom, info, lastUpdateTimestamp)); o.setContactgroup(prepareOtherResearchProductContactGroups(doc, info));
oafs.addAll(addOtherResultRels(doc, collectedFrom, info, lastUpdateTimestamp)); o.setTool(prepareOtherResearchProductTools(doc, info));
} oafs.add(o);
break;
return oafs; }
}
if (!oafs.isEmpty()) {
private List<Oaf> addProjectRels( oafs.addAll(addProjectRels(doc, collectedFrom, info, lastUpdateTimestamp));
final Document doc, oafs.addAll(addOtherResultRels(doc, collectedFrom, info, lastUpdateTimestamp));
final KeyValue collectedFrom, }
final DataInfo info,
final long lastUpdateTimestamp) { return oafs;
}
final List<Oaf> res = new ArrayList<>();
private List<Oaf> addProjectRels(
final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false); final Document doc,
final KeyValue collectedFrom,
for (final Object o : doc.selectNodes("//oaf:projectid")) { final DataInfo info,
final String projectId = createOpenaireId(40, ((Node) o).getText(), true); final long lastUpdateTimestamp) {
final Relation r1 = new Relation(); final List<Oaf> res = new ArrayList<>();
r1.setRelType("resultProject");
r1.setSubRelType("outcome"); final String docId = createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false);
r1.setRelClass("isProducedBy");
r1.setSource(docId); for (final Object o : doc.selectNodes("//oaf:projectid")) {
r1.setTarget(projectId); final String projectId = createOpenaireId(40, ((Node) o).getText(), true);
r1.setCollectedfrom(Arrays.asList(collectedFrom));
r1.setDataInfo(info); final Relation r1 = new Relation();
r1.setLastupdatetimestamp(lastUpdateTimestamp); r1.setRelType("resultProject");
res.add(r1); r1.setSubRelType("outcome");
r1.setRelClass("isProducedBy");
final Relation r2 = new Relation(); r1.setSource(docId);
r2.setRelType("resultProject"); r1.setTarget(projectId);
r2.setSubRelType("outcome"); r1.setCollectedfrom(Arrays.asList(collectedFrom));
r2.setRelClass("produces"); r1.setDataInfo(info);
r2.setSource(projectId); r1.setLastupdatetimestamp(lastUpdateTimestamp);
r2.setTarget(docId); res.add(r1);
r2.setCollectedfrom(Arrays.asList(collectedFrom));
r2.setDataInfo(info); final Relation r2 = new Relation();
r2.setLastupdatetimestamp(lastUpdateTimestamp); r2.setRelType("resultProject");
res.add(r2); r2.setSubRelType("outcome");
} r2.setRelClass("produces");
r2.setSource(projectId);
return res; r2.setTarget(docId);
} r2.setCollectedfrom(Arrays.asList(collectedFrom));
r2.setDataInfo(info);
protected abstract List<Oaf> addOtherResultRels( r2.setLastupdatetimestamp(lastUpdateTimestamp);
final Document doc, res.add(r2);
final KeyValue collectedFrom, }
final DataInfo info,
final long lastUpdateTimestamp); return res;
}
private void populateResultFields(
final Result r, protected abstract List<Oaf> addOtherResultRels(
final Document doc, final Document doc,
final KeyValue collectedFrom, final KeyValue collectedFrom,
final KeyValue hostedBy, final DataInfo info,
final DataInfo info, final long lastUpdateTimestamp);
final long lastUpdateTimestamp) {
r.setDataInfo(info); private void populateResultFields(
r.setLastupdatetimestamp(lastUpdateTimestamp); final Result r,
r.setId(createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false)); final Document doc,
r.setOriginalId(Arrays.asList(doc.valueOf("//dri:objIdentifier"))); final KeyValue collectedFrom,
r.setCollectedfrom(Arrays.asList(collectedFrom)); final KeyValue hostedBy,
r.setPid(prepareListStructProps(doc, "//oaf:identifier", "@identifierType", "dnet:pid_types", "dnet:pid_types", info)); final DataInfo info,
r.setDateofcollection(doc.valueOf("//dr:dateOfCollection")); final long lastUpdateTimestamp) {
r.setDateoftransformation(doc.valueOf("//dr:dateOfTransformation")); r.setDataInfo(info);
r.setExtraInfo(new ArrayList<>()); // NOT PRESENT IN MDSTORES r.setLastupdatetimestamp(lastUpdateTimestamp);
r.setOaiprovenance(prepareOAIprovenance(doc)); r.setId(createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false));
r.setAuthor(prepareAuthors(doc, info)); r.setOriginalId(Arrays.asList(doc.valueOf("//dri:objIdentifier")));
r.setLanguage(prepareLanguages(doc)); r.setCollectedfrom(Arrays.asList(collectedFrom));
r.setCountry(new ArrayList<>()); // NOT PRESENT IN MDSTORES r.setPid(
r.setSubject(prepareSubjects(doc, info)); prepareListStructProps(
r.setTitle(prepareTitles(doc, info)); doc,
r.setRelevantdate(prepareRelevantDates(doc, info)); "//oaf:identifier",
r.setDescription(prepareDescriptions(doc, info)); "@identifierType",
r.setDateofacceptance(prepareField(doc, "//oaf:dateAccepted", info)); "dnet:pid_types",
r.setPublisher(preparePublisher(doc, info)); "dnet:pid_types",
r.setEmbargoenddate(prepareField(doc, "//oaf:embargoenddate", info)); info));
r.setSource(prepareSources(doc, info)); r.setDateofcollection(doc.valueOf("//dr:dateOfCollection"));
r.setFulltext(new ArrayList<>()); // NOT PRESENT IN MDSTORES r.setDateoftransformation(doc.valueOf("//dr:dateOfTransformation"));
r.setFormat(prepareFormats(doc, info)); r.setExtraInfo(new ArrayList<>()); // NOT PRESENT IN MDSTORES
r.setContributor(prepareContributors(doc, info)); r.setOaiprovenance(prepareOAIprovenance(doc));
r.setResourcetype(prepareResourceType(doc, info)); r.setAuthor(prepareAuthors(doc, info));
r.setCoverage(prepareCoverages(doc, info)); r.setLanguage(prepareLanguages(doc));
r.setContext(new ArrayList<>()); // NOT PRESENT IN MDSTORES r.setCountry(new ArrayList<>()); // NOT PRESENT IN MDSTORES
r.setExternalReference(new ArrayList<>()); // NOT PRESENT IN MDSTORES r.setSubject(prepareSubjects(doc, info));
r.setInstance(prepareInstances(doc, info, collectedFrom, hostedBy)); r.setTitle(prepareTitles(doc, info));
} r.setRelevantdate(prepareRelevantDates(doc, info));
r.setDescription(prepareDescriptions(doc, info));
protected abstract Qualifier prepareResourceType(Document doc, DataInfo info); r.setDateofacceptance(prepareField(doc, "//oaf:dateAccepted", info));
r.setPublisher(preparePublisher(doc, info));
protected abstract List<Instance> prepareInstances( r.setEmbargoenddate(prepareField(doc, "//oaf:embargoenddate", info));
Document doc, r.setSource(prepareSources(doc, info));
DataInfo info, r.setFulltext(new ArrayList<>()); // NOT PRESENT IN MDSTORES
KeyValue collectedfrom, r.setFormat(prepareFormats(doc, info));
KeyValue hostedby); r.setContributor(prepareContributors(doc, info));
r.setResourcetype(prepareResourceType(doc, info));
protected abstract List<Field<String>> prepareSources(Document doc, DataInfo info); r.setCoverage(prepareCoverages(doc, info));
r.setContext(new ArrayList<>()); // NOT PRESENT IN MDSTORES
protected abstract List<StructuredProperty> prepareRelevantDates(Document doc, DataInfo info); r.setExternalReference(new ArrayList<>()); // NOT PRESENT IN MDSTORES
r.setInstance(prepareInstances(doc, info, collectedFrom, hostedBy));
protected abstract List<Field<String>> prepareCoverages(Document doc, DataInfo info); }
protected abstract List<Field<String>> prepareContributors(Document doc, DataInfo info); protected abstract Qualifier prepareResourceType(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareFormats(Document doc, DataInfo info); protected abstract List<Instance> prepareInstances(
Document doc, DataInfo info, KeyValue collectedfrom, KeyValue hostedby);
protected abstract Field<String> preparePublisher(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareSources(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareDescriptions(Document doc, DataInfo info);
protected abstract List<StructuredProperty> prepareRelevantDates(Document doc, DataInfo info);
protected abstract List<StructuredProperty> prepareTitles(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareCoverages(Document doc, DataInfo info);
protected abstract List<StructuredProperty> prepareSubjects(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareContributors(Document doc, DataInfo info);
protected abstract Qualifier prepareLanguages(Document doc);
protected abstract List<Field<String>> prepareFormats(Document doc, DataInfo info);
protected abstract List<Author> prepareAuthors(Document doc, DataInfo info);
protected abstract Field<String> preparePublisher(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareOtherResearchProductTools(
Document doc, protected abstract List<Field<String>> prepareDescriptions(Document doc, DataInfo info);
DataInfo info);
protected abstract List<StructuredProperty> prepareTitles(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareOtherResearchProductContactGroups(
Document doc, protected abstract List<StructuredProperty> prepareSubjects(Document doc, DataInfo info);
DataInfo info);
protected abstract Qualifier prepareLanguages(Document doc);
protected abstract List<Field<String>> prepareOtherResearchProductContactPersons(
Document doc, protected abstract List<Author> prepareAuthors(Document doc, DataInfo info);
DataInfo info);
protected abstract List<Field<String>> prepareOtherResearchProductTools(
protected abstract Qualifier prepareSoftwareProgrammingLanguage(Document doc, DataInfo info); Document doc, DataInfo info);
protected abstract Field<String> prepareSoftwareCodeRepositoryUrl(Document doc, DataInfo info); protected abstract List<Field<String>> prepareOtherResearchProductContactGroups(
Document doc, DataInfo info);
protected abstract List<StructuredProperty> prepareSoftwareLicenses(
Document doc, protected abstract List<Field<String>> prepareOtherResearchProductContactPersons(
DataInfo info); Document doc, DataInfo info);
protected abstract List<Field<String>> prepareSoftwareDocumentationUrls( protected abstract Qualifier prepareSoftwareProgrammingLanguage(Document doc, DataInfo info);
Document doc,
DataInfo info); protected abstract Field<String> prepareSoftwareCodeRepositoryUrl(Document doc, DataInfo info);
protected abstract List<GeoLocation> prepareDatasetGeoLocations(Document doc, DataInfo info); protected abstract List<StructuredProperty> prepareSoftwareLicenses(
Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetMetadataVersionNumber(
Document doc, protected abstract List<Field<String>> prepareSoftwareDocumentationUrls(
DataInfo info); Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetLastMetadataUpdate(Document doc, DataInfo info); protected abstract List<GeoLocation> prepareDatasetGeoLocations(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetVersion(Document doc, DataInfo info); protected abstract Field<String> prepareDatasetMetadataVersionNumber(
Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetSize(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetLastMetadataUpdate(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetDevice(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetVersion(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetStorageDate(Document doc, DataInfo info);
protected abstract Field<String> prepareDatasetSize(Document doc, DataInfo info);
private Journal prepareJournal(final Document doc, final DataInfo info) {
final Node n = doc.selectSingleNode("//oaf:journal"); protected abstract Field<String> prepareDatasetDevice(Document doc, DataInfo info);
if (n != null) {
final String name = n.getText(); protected abstract Field<String> prepareDatasetStorageDate(Document doc, DataInfo info);
final String issnPrinted = n.valueOf("@issn");
final String issnOnline = n.valueOf("@eissn"); private Journal prepareJournal(final Document doc, final DataInfo info) {
final String issnLinking = n.valueOf("@lissn"); final Node n = doc.selectSingleNode("//oaf:journal");
final String ep = n.valueOf("@ep"); if (n != null) {
final String iss = n.valueOf("@iss"); final String name = n.getText();
final String sp = n.valueOf("@sp"); final String issnPrinted = n.valueOf("@issn");
final String vol = n.valueOf("@vol"); final String issnOnline = n.valueOf("@eissn");
final String edition = n.valueOf("@edition"); final String issnLinking = n.valueOf("@lissn");
if (StringUtils.isNotBlank(name)) { return journal(name, issnPrinted, issnOnline, issnLinking, ep, iss, sp, vol, edition, null, null, info); } final String ep = n.valueOf("@ep");
} final String iss = n.valueOf("@iss");
return null; final String sp = n.valueOf("@sp");
} final String vol = n.valueOf("@vol");
final String edition = n.valueOf("@edition");
protected Qualifier prepareQualifier( if (StringUtils.isNotBlank(name)) {
final Node node, return journal(
final String xpath, name,
final String schemeId, issnPrinted,
final String schemeName) { issnOnline,
final String classId = node.valueOf(xpath); issnLinking,
final String className = code2name.get(classId); ep,
return qualifier(classId, className, schemeId, schemeName); iss,
} sp,
vol,
protected List<StructuredProperty> prepareListStructProps( edition,
final Node node, null,
final String xpath, null,
final String xpathClassId, info);
final String schemeId, }
final String schemeName, }
final DataInfo info) { return null;
final List<StructuredProperty> res = new ArrayList<>(); }
for (final Object o : node.selectNodes(xpath)) {
final Node n = (Node) o; protected Qualifier prepareQualifier(
final String classId = n.valueOf(xpathClassId); final Node node, final String xpath, final String schemeId, final String schemeName) {
final String className = code2name.get(classId); final String classId = node.valueOf(xpath);
res.add(structuredProperty(n.getText(), classId, className, schemeId, schemeName, info)); final String className = code2name.get(classId);
} return qualifier(classId, className, schemeId, schemeName);
return res; }
}
protected List<StructuredProperty> prepareListStructProps(
protected List<StructuredProperty> prepareListStructProps( final Node node,
final Node node, final String xpath,
final String xpath, final String xpathClassId,
final Qualifier qualifier, final String schemeId,
final DataInfo info) { final String schemeName,
final List<StructuredProperty> res = new ArrayList<>(); final DataInfo info) {
for (final Object o : node.selectNodes(xpath)) { final List<StructuredProperty> res = new ArrayList<>();
final Node n = (Node) o; for (final Object o : node.selectNodes(xpath)) {
res.add(structuredProperty(n.getText(), qualifier, info)); final Node n = (Node) o;
} final String classId = n.valueOf(xpathClassId);
return res; final String className = code2name.get(classId);
} res.add(
structuredProperty(
protected List<StructuredProperty> prepareListStructProps( n.getText(), classId, className, schemeId, schemeName, info));
final Node node, }
final String xpath, return res;
final DataInfo info) { }
final List<StructuredProperty> res = new ArrayList<>();
for (final Object o : node.selectNodes(xpath)) { protected List<StructuredProperty> prepareListStructProps(
final Node n = (Node) o; final Node node, final String xpath, final Qualifier qualifier, final DataInfo info) {
res.add(structuredProperty(n.getText(), n.valueOf("@classid"), n.valueOf("@classname"), n.valueOf("@schemeid"), n.valueOf("@schemename"), info)); final List<StructuredProperty> res = new ArrayList<>();
} for (final Object o : node.selectNodes(xpath)) {
return res; final Node n = (Node) o;
} res.add(structuredProperty(n.getText(), qualifier, info));
}
protected OAIProvenance prepareOAIprovenance(final Document doc) { return res;
final Node n = }
doc.selectSingleNode("//*[local-name()='provenance']/*[local-name()='originDescription']");
protected List<StructuredProperty> prepareListStructProps(
if (n == null) { return null; } final Node node, final String xpath, final DataInfo info) {
final List<StructuredProperty> res = new ArrayList<>();
final String identifier = n.valueOf("./*[local-name()='identifier']"); for (final Object o : node.selectNodes(xpath)) {
final String baseURL = n.valueOf("./*[local-name()='baseURL']");; final Node n = (Node) o;
final String metadataNamespace = n.valueOf("./*[local-name()='metadataNamespace']");; res.add(
final boolean altered = n.valueOf("@altered").equalsIgnoreCase("true"); structuredProperty(
final String datestamp = n.valueOf("./*[local-name()='datestamp']");; n.getText(),
final String harvestDate = n.valueOf("@harvestDate");; n.valueOf("@classid"),
n.valueOf("@classname"),
return oaiIProvenance(identifier, baseURL, metadataNamespace, altered, datestamp, harvestDate); n.valueOf("@schemeid"),
} n.valueOf("@schemename"),
info));
protected DataInfo prepareDataInfo(final Document doc) { }
final Node n = doc.selectSingleNode("//oaf:datainfo"); return res;
}
if (n == null) { return dataInfo(false, null, false, false, MigrationConstants.REPOSITORY_PROVENANCE_ACTIONS, "0.9"); }
protected OAIProvenance prepareOAIprovenance(final Document doc) {
final String paClassId = n.valueOf("./oaf:provenanceaction/@classid"); final Node n =
final String paClassName = n.valueOf("./oaf:provenanceaction/@classname"); doc.selectSingleNode(
final String paSchemeId = n.valueOf("./oaf:provenanceaction/@schemeid"); "//*[local-name()='provenance']/*[local-name()='originDescription']");
final String paSchemeName = n.valueOf("./oaf:provenanceaction/@schemename");
if (n == null) {
final boolean deletedbyinference = return null;
Boolean.parseBoolean(n.valueOf("./oaf:deletedbyinference")); }
final String inferenceprovenance = n.valueOf("./oaf:inferenceprovenance");
final Boolean inferred = Boolean.parseBoolean(n.valueOf("./oaf:inferred")); final String identifier = n.valueOf("./*[local-name()='identifier']");
final String trust = n.valueOf("./oaf:trust"); final String baseURL = n.valueOf("./*[local-name()='baseURL']");
;
return dataInfo(deletedbyinference, inferenceprovenance, inferred, false, qualifier(paClassId, paClassName, paSchemeId, paSchemeName), trust); final String metadataNamespace = n.valueOf("./*[local-name()='metadataNamespace']");
} ;
final boolean altered = n.valueOf("@altered").equalsIgnoreCase("true");
protected Field<String> prepareField(final Node node, final String xpath, final DataInfo info) { final String datestamp = n.valueOf("./*[local-name()='datestamp']");
return field(node.valueOf(xpath), info); ;
} final String harvestDate = n.valueOf("@harvestDate");
;
protected List<Field<String>> prepareListFields(
final Node node, return oaiIProvenance(
final String xpath, identifier, baseURL, metadataNamespace, altered, datestamp, harvestDate);
final DataInfo info) { }
return listFields(info, prepareListString(node, xpath));
} protected DataInfo prepareDataInfo(final Document doc) {
final Node n = doc.selectSingleNode("//oaf:datainfo");
protected List<String> prepareListString(final Node node, final String xpath) {
final List<String> res = new ArrayList<>(); if (n == null) {
for (final Object o : node.selectNodes(xpath)) { return dataInfo(
final String s = ((Node) o).getText().trim(); false,
if (StringUtils.isNotBlank(s)) { null,
res.add(s); false,
} false,
} MigrationConstants.REPOSITORY_PROVENANCE_ACTIONS,
return res; "0.9");
} }
final String paClassId = n.valueOf("./oaf:provenanceaction/@classid");
final String paClassName = n.valueOf("./oaf:provenanceaction/@classname");
final String paSchemeId = n.valueOf("./oaf:provenanceaction/@schemeid");
final String paSchemeName = n.valueOf("./oaf:provenanceaction/@schemename");
final boolean deletedbyinference =
Boolean.parseBoolean(n.valueOf("./oaf:deletedbyinference"));
final String inferenceprovenance = n.valueOf("./oaf:inferenceprovenance");
final Boolean inferred = Boolean.parseBoolean(n.valueOf("./oaf:inferred"));
final String trust = n.valueOf("./oaf:trust");
return dataInfo(
deletedbyinference,
inferenceprovenance,
inferred,
false,
qualifier(paClassId, paClassName, paSchemeId, paSchemeName),
trust);
}
protected Field<String> prepareField(final Node node, final String xpath, final DataInfo info) {
return field(node.valueOf(xpath), info);
}
protected List<Field<String>> prepareListFields(
final Node node, final String xpath, final DataInfo info) {
return listFields(info, prepareListString(node, xpath));
}
protected List<String> prepareListString(final Node node, final String xpath) {
final List<String> res = new ArrayList<>();
for (final Object o : node.selectNodes(xpath)) {
final String s = ((Node) o).getText().trim();
if (StringUtils.isNotBlank(s)) {
res.add(s);
}
}
return res;
}
} }

996
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/MigrateDbEntitiesApplication.java
View File

File diff suppressed because it is too large Load Diff

37
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/common/MigrationConstants.java
View File

@ -6,17 +6,28 @@ import eu.dnetlib.dhp.schema.oaf.Qualifier;
public class MigrationConstants { public class MigrationConstants {
public static final Qualifier PUBLICATION_RESULTTYPE_QUALIFIER = public static final Qualifier PUBLICATION_RESULTTYPE_QUALIFIER =
qualifier("publication", "publication", "dnet:result_typologies", "dnet:result_typologies"); qualifier(
public static final Qualifier DATASET_RESULTTYPE_QUALIFIER = "publication",
qualifier("dataset", "dataset", "dnet:result_typologies", "dnet:result_typologies"); "publication",
public static final Qualifier SOFTWARE_RESULTTYPE_QUALIFIER = "dnet:result_typologies",
qualifier("software", "software", "dnet:result_typologies", "dnet:result_typologies"); "dnet:result_typologies");
public static final Qualifier OTHER_RESULTTYPE_QUALIFIER = public static final Qualifier DATASET_RESULTTYPE_QUALIFIER =
qualifier("other", "other", "dnet:result_typologies", "dnet:result_typologies"); qualifier("dataset", "dataset", "dnet:result_typologies", "dnet:result_typologies");
public static final Qualifier REPOSITORY_PROVENANCE_ACTIONS = public static final Qualifier SOFTWARE_RESULTTYPE_QUALIFIER =
qualifier("sysimport:crosswalk:repository", "sysimport:crosswalk:repository", "dnet:provenanceActions", "dnet:provenanceActions"); qualifier("software", "software", "dnet:result_typologies", "dnet:result_typologies");
public static final Qualifier ENTITYREGISTRY_PROVENANCE_ACTION = public static final Qualifier OTHER_RESULTTYPE_QUALIFIER =
qualifier("sysimport:crosswalk:entityregistry", "sysimport:crosswalk:entityregistry", "dnet:provenanceActions", "dnet:provenanceActions"); qualifier("other", "other", "dnet:result_typologies", "dnet:result_typologies");
public static final Qualifier REPOSITORY_PROVENANCE_ACTIONS =
qualifier(
"sysimport:crosswalk:repository",
"sysimport:crosswalk:repository",
"dnet:provenanceActions",
"dnet:provenanceActions");
public static final Qualifier ENTITYREGISTRY_PROVENANCE_ACTION =
qualifier(
"sysimport:crosswalk:entityregistry",
"sysimport:crosswalk:entityregistry",
"dnet:provenanceActions",
"dnet:provenanceActions");
} }