From 009dcf6aea063ff3ebd16967b08791a6b5e1c812 Mon Sep 17 00:00:00 2001 From: Claudio Atzori Date: Mon, 5 Feb 2024 16:43:40 +0200 Subject: [PATCH 1/6] [actiosets] introduced support for the PromoteAction strategy --- .../actionmanager/promote/PromoteAction.java | 39 +++++++++++++++++++ .../PromoteActionPayloadForGraphTableJob.java | 29 ++++++++++---- .../PromoteActionPayloadFunctions.java | 4 +- ...load_for_graph_table_input_parameters.json | 6 +++ .../wf/dataset/oozie_app/workflow.xml | 2 + .../wf/datasource/oozie_app/workflow.xml | 1 + .../wf/organization/oozie_app/workflow.xml | 1 + .../oozie_app/workflow.xml | 2 + .../wf/project/oozie_app/workflow.xml | 1 + .../wf/publication/oozie_app/workflow.xml | 2 + .../wf/relation/oozie_app/workflow.xml | 1 + .../wf/software/oozie_app/workflow.xml | 2 + .../PromoteActionPayloadFunctionsTest.java | 4 +- 13 files changed, 84 insertions(+), 10 deletions(-) create mode 100644 dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteAction.java diff --git a/dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteAction.java b/dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteAction.java new file mode 100644 index 0000000000..163a8708e2 --- /dev/null +++ b/dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteAction.java @@ -0,0 +1,39 @@ +/* + * Copyright (c) 2024. + * SPDX-FileCopyrightText: © 2023 Consiglio Nazionale delle Ricerche + * SPDX-License-Identifier: AGPL-3.0-or-later + */ + +package eu.dnetlib.dhp.actionmanager.promote; + +/** Encodes the Actionset promotion strategies */ +public class PromoteAction { + + /** The supported actionset promotion strategies + * + * ENRICH: promotes only records in the actionset matching another record in the + * graph and enriches them applying the given MergeAndGet strategy + * UPSERT: promotes all the records in an actionset, matching records are updated + * using the given MergeAndGet strategy, the non-matching record as inserted as they are. + */ + public enum Strategy { + ENRICH, UPSERT + } + + /** + * Returns the string representation of the join type implementing the given PromoteAction. + * + * @param strategy the strategy to be used to promote the Actionset contents + * @return the join type used to implement the promotion strategy + */ + public static String joinTypeForStrategy(PromoteAction.Strategy strategy) { + switch (strategy) { + case ENRICH: + return "join"; + case UPSERT: + return "full_outer"; + default: + throw new IllegalStateException("unsupported PromoteAction: " + strategy.toString()); + } + } +} diff --git a/dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteActionPayloadForGraphTableJob.java b/dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteActionPayloadForGraphTableJob.java index 7b024bea8b..56cbda4d67 100644 --- a/dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteActionPayloadForGraphTableJob.java +++ b/dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteActionPayloadForGraphTableJob.java @@ -67,8 +67,9 @@ public class PromoteActionPayloadForGraphTableJob { String outputGraphTablePath = parser.get("outputGraphTablePath"); logger.info("outputGraphTablePath: {}", outputGraphTablePath); - MergeAndGet.Strategy strategy = MergeAndGet.Strategy.valueOf(parser.get("mergeAndGetStrategy").toUpperCase()); - logger.info("strategy: {}", strategy); + MergeAndGet.Strategy mergeAndGetStrategy = MergeAndGet.Strategy + .valueOf(parser.get("mergeAndGetStrategy").toUpperCase()); + logger.info("mergeAndGetStrategy: {}", mergeAndGetStrategy); Boolean shouldGroupById = Optional .ofNullable(parser.get("shouldGroupById")) @@ -76,6 +77,12 @@ public class PromoteActionPayloadForGraphTableJob { .orElse(true); logger.info("shouldGroupById: {}", shouldGroupById); + PromoteAction.Strategy promoteActionStrategy = Optional + .ofNullable(parser.get("promoteActionStrategy")) + .map(PromoteAction.Strategy::valueOf) + .orElse(PromoteAction.Strategy.UPSERT); + logger.info("promoteActionStrategy: {}", promoteActionStrategy); + @SuppressWarnings("unchecked") Class rowClazz = (Class) Class.forName(graphTableClassName); @SuppressWarnings("unchecked") @@ -97,7 +104,8 @@ public class PromoteActionPayloadForGraphTableJob { inputGraphTablePath, inputActionPayloadPath, outputGraphTablePath, - strategy, + mergeAndGetStrategy, + promoteActionStrategy, rowClazz, actionPayloadClazz, shouldGroupById); @@ -124,14 +132,16 @@ public class PromoteActionPayloadForGraphTableJob { String inputGraphTablePath, String inputActionPayloadPath, String outputGraphTablePath, - MergeAndGet.Strategy strategy, + MergeAndGet.Strategy mergeAndGetStrategy, + PromoteAction.Strategy promoteActionStrategy, Class rowClazz, Class actionPayloadClazz, Boolean shouldGroupById) { Dataset rowDS = readGraphTable(spark, inputGraphTablePath, rowClazz); Dataset actionPayloadDS = readActionPayload(spark, inputActionPayloadPath, actionPayloadClazz); Dataset result = promoteActionPayloadForGraphTable( - rowDS, actionPayloadDS, strategy, rowClazz, actionPayloadClazz, shouldGroupById) + rowDS, actionPayloadDS, mergeAndGetStrategy, promoteActionStrategy, rowClazz, actionPayloadClazz, + shouldGroupById) .map((MapFunction) value -> value, Encoders.bean(rowClazz)); saveGraphTable(result, outputGraphTablePath); @@ -183,7 +193,8 @@ public class PromoteActionPayloadForGraphTableJob { private static Dataset promoteActionPayloadForGraphTable( Dataset rowDS, Dataset actionPayloadDS, - MergeAndGet.Strategy strategy, + MergeAndGet.Strategy mergeAndGetStrategy, + PromoteAction.Strategy promoteActionStrategy, Class rowClazz, Class actionPayloadClazz, Boolean shouldGroupById) { @@ -195,8 +206,9 @@ public class PromoteActionPayloadForGraphTableJob { SerializableSupplier> rowIdFn = ModelSupport::idFn; SerializableSupplier> actionPayloadIdFn = ModelSupport::idFn; - SerializableSupplier> mergeRowWithActionPayloadAndGetFn = MergeAndGet.functionFor(strategy); - SerializableSupplier> mergeRowsAndGetFn = MergeAndGet.functionFor(strategy); + SerializableSupplier> mergeRowWithActionPayloadAndGetFn = MergeAndGet + .functionFor(mergeAndGetStrategy); + SerializableSupplier> mergeRowsAndGetFn = MergeAndGet.functionFor(mergeAndGetStrategy); SerializableSupplier zeroFn = zeroFn(rowClazz); SerializableSupplier> isNotZeroFn = PromoteActionPayloadForGraphTableJob::isNotZeroFnUsingIdOrSourceAndTarget; @@ -207,6 +219,7 @@ public class PromoteActionPayloadForGraphTableJob { rowIdFn, actionPayloadIdFn, mergeRowWithActionPayloadAndGetFn, + promoteActionStrategy, rowClazz, actionPayloadClazz); diff --git a/dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteActionPayloadFunctions.java b/dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteActionPayloadFunctions.java index d799c646bc..f0b094240e 100644 --- a/dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteActionPayloadFunctions.java +++ b/dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteActionPayloadFunctions.java @@ -34,6 +34,7 @@ public class PromoteActionPayloadFunctions { * @param rowIdFn Function used to get the id of graph table row * @param actionPayloadIdFn Function used to get id of action payload instance * @param mergeAndGetFn Function used to merge graph table row and action payload instance + * @param promoteActionStrategy the Actionset promotion strategy * @param rowClazz Class of graph table * @param actionPayloadClazz Class of action payload * @param Type of graph table row @@ -46,6 +47,7 @@ public class PromoteActionPayloadFunctions { SerializableSupplier> rowIdFn, SerializableSupplier> actionPayloadIdFn, SerializableSupplier> mergeAndGetFn, + PromoteAction.Strategy promoteActionStrategy, Class rowClazz, Class actionPayloadClazz) { if (!isSubClass(rowClazz, actionPayloadClazz)) { @@ -61,7 +63,7 @@ public class PromoteActionPayloadFunctions { .joinWith( actionPayloadWithIdDS, rowWithIdDS.col("_1").equalTo(actionPayloadWithIdDS.col("_1")), - "full_outer") + PromoteAction.joinTypeForStrategy(promoteActionStrategy)) .map( (MapFunction, Tuple2>, G>) value -> { Optional rowOpt = Optional.ofNullable(value._1()).map(Tuple2::_2); diff --git a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/promote/promote_action_payload_for_graph_table_input_parameters.json b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/promote/promote_action_payload_for_graph_table_input_parameters.json index 00c9404ef0..81a7c77d7b 100644 --- a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/promote/promote_action_payload_for_graph_table_input_parameters.json +++ b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/promote/promote_action_payload_for_graph_table_input_parameters.json @@ -41,6 +41,12 @@ "paramDescription": "strategy for merging graph table objects with action payload instances, MERGE_FROM_AND_GET or SELECT_NEWER_AND_GET", "paramRequired": true }, + { + "paramName": "pas", + "paramLongName": "promoteActionStrategy", + "paramDescription": "strategy for promoting the actionset contents into the graph tables, ENRICH or UPSERT (default)", + "paramRequired": false + }, { "paramName": "sgid", "paramLongName": "shouldGroupById", diff --git a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/dataset/oozie_app/workflow.xml b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/dataset/oozie_app/workflow.xml index 4f374a75a0..5401b45cac 100644 --- a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/dataset/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/dataset/oozie_app/workflow.xml @@ -115,6 +115,7 @@ --actionPayloadClassNameeu.dnetlib.dhp.schema.oaf.Dataset --outputGraphTablePath${workingDir}/dataset --mergeAndGetStrategy${mergeAndGetStrategy} + --promoteActionStrategy${promoteActionStrategy} --shouldGroupById${shouldGroupById} @@ -167,6 +168,7 @@ --actionPayloadClassNameeu.dnetlib.dhp.schema.oaf.Result --outputGraphTablePath${outputGraphRootPath}/dataset --mergeAndGetStrategy${mergeAndGetStrategy} + --promoteActionStrategy${promoteActionStrategy} --shouldGroupById${shouldGroupById} diff --git a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/datasource/oozie_app/workflow.xml b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/datasource/oozie_app/workflow.xml index c85ba4ac1c..f9bd66ae3a 100644 --- a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/datasource/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/datasource/oozie_app/workflow.xml @@ -106,6 +106,7 @@ --actionPayloadClassNameeu.dnetlib.dhp.schema.oaf.Datasource --outputGraphTablePath${outputGraphRootPath}/datasource --mergeAndGetStrategy${mergeAndGetStrategy} + --promoteActionStrategy${promoteActionStrategy} diff --git a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/organization/oozie_app/workflow.xml b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/organization/oozie_app/workflow.xml index 412cad70bc..ebfdeee312 100644 --- a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/organization/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/organization/oozie_app/workflow.xml @@ -106,6 +106,7 @@ --actionPayloadClassNameeu.dnetlib.dhp.schema.oaf.Organization --outputGraphTablePath${outputGraphRootPath}/organization --mergeAndGetStrategy${mergeAndGetStrategy} + --promoteActionStrategy${promoteActionStrategy} diff --git a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/otherresearchproduct/oozie_app/workflow.xml b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/otherresearchproduct/oozie_app/workflow.xml index 7bac760e2c..02399ed9b9 100644 --- a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/otherresearchproduct/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/otherresearchproduct/oozie_app/workflow.xml @@ -114,6 +114,7 @@ --actionPayloadClassNameeu.dnetlib.dhp.schema.oaf.OtherResearchProduct --outputGraphTablePath${workingDir}/otherresearchproduct --mergeAndGetStrategy${mergeAndGetStrategy} + --promoteActionStrategy${promoteActionStrategy} --shouldGroupById${shouldGroupById} @@ -166,6 +167,7 @@ --actionPayloadClassNameeu.dnetlib.dhp.schema.oaf.Result --outputGraphTablePath${outputGraphRootPath}/otherresearchproduct --mergeAndGetStrategy${mergeAndGetStrategy} + --promoteActionStrategy${promoteActionStrategy} --shouldGroupById${shouldGroupById} diff --git a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/project/oozie_app/workflow.xml b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/project/oozie_app/workflow.xml index daf48e9d78..57c2357b4a 100644 --- a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/project/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/project/oozie_app/workflow.xml @@ -106,6 +106,7 @@ --actionPayloadClassNameeu.dnetlib.dhp.schema.oaf.Project --outputGraphTablePath${outputGraphRootPath}/project --mergeAndGetStrategy${mergeAndGetStrategy} + --promoteActionStrategy${promoteActionStrategy} diff --git a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/publication/oozie_app/workflow.xml b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/publication/oozie_app/workflow.xml index b76dc82f14..92b114776a 100644 --- a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/publication/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/publication/oozie_app/workflow.xml @@ -115,6 +115,7 @@ --actionPayloadClassNameeu.dnetlib.dhp.schema.oaf.Publication --outputGraphTablePath${workingDir}/publication --mergeAndGetStrategy${mergeAndGetStrategy} + --promoteActionStrategy${promoteActionStrategy} --shouldGroupById${shouldGroupById} @@ -167,6 +168,7 @@ --actionPayloadClassNameeu.dnetlib.dhp.schema.oaf.Result --outputGraphTablePath${outputGraphRootPath}/publication --mergeAndGetStrategy${mergeAndGetStrategy} + --promoteActionStrategy${promoteActionStrategy} --shouldGroupById${shouldGroupById} diff --git a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/relation/oozie_app/workflow.xml b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/relation/oozie_app/workflow.xml index d3086dbdc0..e9e5f0b454 100644 --- a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/relation/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/relation/oozie_app/workflow.xml @@ -107,6 +107,7 @@ --actionPayloadClassNameeu.dnetlib.dhp.schema.oaf.Relation --outputGraphTablePath${outputGraphRootPath}/relation --mergeAndGetStrategy${mergeAndGetStrategy} + --promoteActionStrategy${promoteActionStrategy} diff --git a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/software/oozie_app/workflow.xml b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/software/oozie_app/workflow.xml index b5673b18f4..1d36ddf94b 100644 --- a/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/software/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/software/oozie_app/workflow.xml @@ -114,6 +114,7 @@ --actionPayloadClassNameeu.dnetlib.dhp.schema.oaf.Software --outputGraphTablePath${workingDir}/software --mergeAndGetStrategy${mergeAndGetStrategy} + --promoteActionStrategy${promoteActionStrategy} --shouldGroupById${shouldGroupById} @@ -166,6 +167,7 @@ --actionPayloadClassNameeu.dnetlib.dhp.schema.oaf.Result --outputGraphTablePath${outputGraphRootPath}/software --mergeAndGetStrategy${mergeAndGetStrategy} + --promoteActionStrategy${promoteActionStrategy} --shouldGroupById${shouldGroupById} diff --git a/dhp-workflows/dhp-actionmanager/src/test/java/eu/dnetlib/dhp/actionmanager/promote/PromoteActionPayloadFunctionsTest.java b/dhp-workflows/dhp-actionmanager/src/test/java/eu/dnetlib/dhp/actionmanager/promote/PromoteActionPayloadFunctionsTest.java index cbc1bfaba7..777e2fa1ce 100644 --- a/dhp-workflows/dhp-actionmanager/src/test/java/eu/dnetlib/dhp/actionmanager/promote/PromoteActionPayloadFunctionsTest.java +++ b/dhp-workflows/dhp-actionmanager/src/test/java/eu/dnetlib/dhp/actionmanager/promote/PromoteActionPayloadFunctionsTest.java @@ -54,7 +54,7 @@ public class PromoteActionPayloadFunctionsTest { RuntimeException.class, () -> PromoteActionPayloadFunctions .joinGraphTableWithActionPayloadAndMerge( - null, null, null, null, null, OafImplSubSub.class, OafImpl.class)); + null, null, null, null, null, null, OafImplSubSub.class, OafImpl.class)); } @Test @@ -104,6 +104,7 @@ public class PromoteActionPayloadFunctionsTest { rowIdFn, actionPayloadIdFn, mergeAndGetFn, + PromoteAction.Strategy.UPSERT, OafImplSubSub.class, OafImplSubSub.class) .collectAsList(); @@ -183,6 +184,7 @@ public class PromoteActionPayloadFunctionsTest { rowIdFn, actionPayloadIdFn, mergeAndGetFn, + PromoteAction.Strategy.UPSERT, OafImplSubSub.class, OafImplSub.class) .collectAsList(); From fd17c1f17c4470be0e45dfdf3c13255087ce8e59 Mon Sep 17 00:00:00 2001 From: Claudio Atzori Date: Mon, 5 Feb 2024 16:55:36 +0200 Subject: [PATCH 2/6] [actiosets] fixed join type --- .../eu/dnetlib/dhp/actionmanager/promote/PromoteAction.java | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteAction.java b/dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteAction.java index 163a8708e2..8fb9c8c953 100644 --- a/dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteAction.java +++ b/dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteAction.java @@ -29,7 +29,7 @@ public class PromoteAction { public static String joinTypeForStrategy(PromoteAction.Strategy strategy) { switch (strategy) { case ENRICH: - return "join"; + return "left_outer"; case UPSERT: return "full_outer"; default: From b19643f6eb0d0f7bf798ffbff0427a6c2c9671e0 Mon Sep 17 00:00:00 2001 From: Giambattista Bloisi Date: Thu, 8 Feb 2024 15:12:16 +0100 Subject: [PATCH 3/6] Dedup aliases, created when a dedup in a previous build has been merged in a new dedup, need to be marked as "deletedbyinference", since they are "merged" in the new dedup --- .../dhp/oa/dedup/DedupRecordFactory.java | 43 +++++++++++++------ 1 file changed, 31 insertions(+), 12 deletions(-) diff --git a/dhp-workflows/dhp-dedup-openaire/src/main/java/eu/dnetlib/dhp/oa/dedup/DedupRecordFactory.java b/dhp-workflows/dhp-dedup-openaire/src/main/java/eu/dnetlib/dhp/oa/dedup/DedupRecordFactory.java index eddfba309d..d5b106c81b 100644 --- a/dhp-workflows/dhp-dedup-openaire/src/main/java/eu/dnetlib/dhp/oa/dedup/DedupRecordFactory.java +++ b/dhp-workflows/dhp-dedup-openaire/src/main/java/eu/dnetlib/dhp/oa/dedup/DedupRecordFactory.java @@ -122,22 +122,41 @@ public class DedupRecordFactory { } return Stream - .concat(Stream.of(agg.getDedupId()), agg.aliases.stream()) - .map(id -> { - try { - OafEntity res = (OafEntity) BeanUtils.cloneBean(agg.entity); - res.setId(id); - res.setDataInfo(dataInfo); - res.setLastupdatetimestamp(ts); - return res; - } catch (Exception e) { - throw new RuntimeException(e); - } - }) + .concat( + Stream + .of(agg.getDedupId()) + .map(id -> createDedupOafEntity(id, agg.entity, dataInfo, ts)), + agg.aliases + .stream() + .map(id -> createMergedDedupAliasOafEntity(id, agg.entity, dataInfo, ts))) .iterator(); }, beanEncoder); } + private static OafEntity createDedupOafEntity(String id, OafEntity base, DataInfo dataInfo, long ts) { + try { + OafEntity res = (OafEntity) BeanUtils.cloneBean(base); + res.setId(id); + res.setDataInfo(dataInfo); + res.setLastupdatetimestamp(ts); + return res; + } catch (Exception e) { + throw new RuntimeException(e); + } + } + + private static OafEntity createMergedDedupAliasOafEntity(String id, OafEntity base, DataInfo dataInfo, long ts) { + try { + OafEntity res = createDedupOafEntity(id, base, dataInfo, ts); + DataInfo ds = (DataInfo) BeanUtils.cloneBean(dataInfo); + ds.setDeletedbyinference(true); + res.setDataInfo(ds); + return res; + } catch (Exception e) { + throw new RuntimeException(e); + } + } + private static OafEntity reduceEntity(OafEntity entity, OafEntity duplicate) { if (duplicate == null) { From d85d2df6ad989abe32a3ca3b509c2c264e8d3553 Mon Sep 17 00:00:00 2001 From: Claudio Atzori Date: Fri, 9 Feb 2024 10:19:53 +0100 Subject: [PATCH 4/6] [graph raw] fixed mapping of the original resource type from the Datacite format --- .../dhp/oa/graph/raw/OdfToOafMapper.java | 22 +++++-- .../dnetlib/dhp/oa/graph/raw/MappersTest.java | 32 +++++++++- .../dnetlib/dhp/oa/graph/raw/odf_zenodo2.xml | 59 +++++++++++++++++++ 3 files changed, 106 insertions(+), 7 deletions(-) create mode 100644 dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/odf_zenodo2.xml diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java index 08529125c8..57e0d29555 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java @@ -238,11 +238,23 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper { (Element) doc .selectSingleNode( "//*[local-name()='metadata']/*[local-name() = 'resource']/*[local-name() = 'resourceType']")) - .map(element -> { - final String resourceTypeURI = element.attributeValue("uri"); - final String resourceTypeAnyURI = element.attributeValue("anyURI"); - final String resourceTypeTxt = element.getText(); - final String resourceTypeGeneral = element.attributeValue("resourceTypeGeneral"); + .map(e -> { + final String resourceTypeURI = Optional + .ofNullable(e.attributeValue("uri")) + .filter(StringUtils::isNotBlank) + .orElse(null); + final String resourceTypeAnyURI = Optional + .ofNullable(e.attributeValue("anyURI")) + .filter(StringUtils::isNotBlank) + .orElse(null); + final String resourceTypeTxt = Optional + .ofNullable(e.getText()) + .filter(StringUtils::isNotBlank) + .orElse(null); + final String resourceTypeGeneral = Optional + .ofNullable(e.attributeValue("resourceTypeGeneral")) + .filter(StringUtils::isNotBlank) + .orElse(null); return ObjectUtils .firstNonNull(resourceTypeURI, resourceTypeAnyURI, resourceTypeTxt, resourceTypeGeneral); diff --git a/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/oa/graph/raw/MappersTest.java b/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/oa/graph/raw/MappersTest.java index bbb8e7d573..a6bbd30cff 100644 --- a/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/oa/graph/raw/MappersTest.java +++ b/dhp-workflows/dhp-graph-mapper/src/test/java/eu/dnetlib/dhp/oa/graph/raw/MappersTest.java @@ -1171,6 +1171,34 @@ class MappersTest { } + @Test + void test_Zenodo2() throws IOException { + final String xml = IOUtils.toString(Objects.requireNonNull(getClass().getResourceAsStream("odf_zenodo2.xml"))); + final List list = new OdfToOafMapper(vocs, false, true).processMdRecord(xml); + + assertEquals(3, list.size()); + Publication p = cleanup((Publication) list.get(0), vocs); + + assertNotNull(p.getInstance()); + assertEquals(1, p.getInstance().size()); + + final Instance instance = p.getInstance().get(0); + + assertNotNull(instance.getInstanceTypeMapping()); + assertEquals(1, instance.getInstanceTypeMapping().size()); + + Optional coarType = instance + .getInstanceTypeMapping() + .stream() + .filter(itm -> ModelConstants.OPENAIRE_COAR_RESOURCE_TYPES_3_1.equals(itm.getVocabularyName())) + .findFirst(); + + assertTrue(coarType.isPresent()); + assertNotNull(coarType.get().getOriginalType()); + assertNull(coarType.get().getTypeCode()); + assertNull(coarType.get().getTypeLabel()); + } + @Test void testROHub2() throws IOException { final String xml = IOUtils @@ -1229,7 +1257,7 @@ class MappersTest { } @Test - public void testD4ScienceTraining() throws IOException { + void testD4ScienceTraining() throws IOException { final String xml = IOUtils .toString(Objects.requireNonNull(getClass().getResourceAsStream("d4science-1-training.xml"))); final List list = new OdfToOafMapper(vocs, false, true).processMdRecord(xml); @@ -1240,7 +1268,7 @@ class MappersTest { } @Test - public void testD4ScienceDataset() throws IOException { + void testD4ScienceDataset() throws IOException { final String xml = IOUtils .toString(Objects.requireNonNull(getClass().getResourceAsStream("d4science-2-dataset.xml"))); final List list = new OdfToOafMapper(vocs, false, true).processMdRecord(xml); diff --git a/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/odf_zenodo2.xml b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/odf_zenodo2.xml new file mode 100644 index 0000000000..ebe105de86 --- /dev/null +++ b/dhp-workflows/dhp-graph-mapper/src/test/resources/eu/dnetlib/dhp/oa/graph/raw/odf_zenodo2.xml @@ -0,0 +1,59 @@ + + +
+ oai:zenodo.org:1596086 + 2020-01-20T13:50:28Z + openaire + 2024-02-08T11:03:10.994Z + od______2659::036d5555a6688ed00c8d0da97bdece3b + 2024-02-08T11:03:10.994Z + 2024-02-08T11:03:10.994Z +
+ + + https://zenodo.org/record/1596086 + + + + Bonney, T. G. + T. G. + Bonney + + + + Ice Blocks on a Moraine + + Zenodo + 1889 + + 1889-08-22 + + + + 10.1038/040391a0 + + + Creative Commons Zero v1.0 Universal + Open Access + + + n/a + + + 0001 + 1889-08-22 + OPEN + http://creativecommons.org/publicdomain/zero/1.0/legalcode + + + + + From 29194472a71e456f74d5b88aa681c56ebf869acf Mon Sep 17 00:00:00 2001 From: Giambattista Bloisi Date: Sun, 4 Feb 2024 10:19:33 +0100 Subject: [PATCH 5/6] Promote "Research" to a jolly instanceType in dedup comparisons Compare Part of book or chapter of book with Article --- .../java/eu/dnetlib/pace/tree/InstanceTypeMatch.java | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/dhp-pace-core/src/main/java/eu/dnetlib/pace/tree/InstanceTypeMatch.java b/dhp-pace-core/src/main/java/eu/dnetlib/pace/tree/InstanceTypeMatch.java index 34ebcf7a7d..e9fb13e7c1 100644 --- a/dhp-pace-core/src/main/java/eu/dnetlib/pace/tree/InstanceTypeMatch.java +++ b/dhp-pace-core/src/main/java/eu/dnetlib/pace/tree/InstanceTypeMatch.java @@ -21,17 +21,26 @@ public class InstanceTypeMatch extends AbstractListComparator { public InstanceTypeMatch(Map params) { super(params); + + // chaper of book in article + // in article + // jolly types translationMap.put("Conference object", "*"); + translationMap.put("Research", "*"); translationMap.put("Other literature type", "*"); translationMap.put("Unknown", "*"); translationMap.put("UNKNOWN", "*"); // article types translationMap.put("Article", "Article"); + // Book translationMap.put("Data Paper", "Article"); translationMap.put("Software Paper", "Article"); translationMap.put("Preprint", "Article"); + translationMap.put("Part of book or chapter of book", "Article"); + //Journal + // thesis types translationMap.put("Thesis", "Thesis"); From d65285da7f1340121814413c10f3e896a0debe0a Mon Sep 17 00:00:00 2001 From: Giambattista Bloisi Date: Thu, 15 Feb 2024 12:10:42 +0100 Subject: [PATCH 6/6] Promote "Research" to a jolly instanceType in dedup comparisons Compare "Journal" and "Part of book or chapter of book" with "Article" --- .../main/java/eu/dnetlib/pace/tree/InstanceTypeMatch.java | 8 +------- 1 file changed, 1 insertion(+), 7 deletions(-) diff --git a/dhp-pace-core/src/main/java/eu/dnetlib/pace/tree/InstanceTypeMatch.java b/dhp-pace-core/src/main/java/eu/dnetlib/pace/tree/InstanceTypeMatch.java index e9fb13e7c1..42459ef00e 100644 --- a/dhp-pace-core/src/main/java/eu/dnetlib/pace/tree/InstanceTypeMatch.java +++ b/dhp-pace-core/src/main/java/eu/dnetlib/pace/tree/InstanceTypeMatch.java @@ -21,10 +21,6 @@ public class InstanceTypeMatch extends AbstractListComparator { public InstanceTypeMatch(Map params) { super(params); - - // chaper of book in article - // in article - // jolly types translationMap.put("Conference object", "*"); translationMap.put("Research", "*"); @@ -34,13 +30,11 @@ public class InstanceTypeMatch extends AbstractListComparator { // article types translationMap.put("Article", "Article"); - // Book + translationMap.put("Journal", "Article"); translationMap.put("Data Paper", "Article"); translationMap.put("Software Paper", "Article"); translationMap.put("Preprint", "Article"); translationMap.put("Part of book or chapter of book", "Article"); - //Journal - // thesis types translationMap.put("Thesis", "Thesis");