From 5489ed9834b4ad9d20b6c42cbaf2ca49c07b1846 Mon Sep 17 00:00:00 2001 From: "miriam.baglioni" Date: Tue, 4 Oct 2022 10:10:59 +0200 Subject: [PATCH] [EOSC Dump] refactoring --- .../jsonschemas/eosc_result_schema.json | 946 +++++++++--------- .../dump/eoscdump/oozie_app/workflow.xml | 2 +- .../dump/projectsubset/oozie_app/workflow.xml | 4 +- 3 files changed, 465 insertions(+), 487 deletions(-) diff --git a/dump-schema/src/main/resources/jsonschemas/eosc_result_schema.json b/dump-schema/src/main/resources/jsonschemas/eosc_result_schema.json index 4c8d98e..886a173 100644 --- a/dump-schema/src/main/resources/jsonschemas/eosc_result_schema.json +++ b/dump-schema/src/main/resources/jsonschemas/eosc_result_schema.json @@ -1,596 +1,574 @@ { - "$schema" : "http://json-schema.org/draft-07/schema#", - "definitions" : { - "CfHbKeyValue" : { - "type" : "object", - "properties" : { - "key" : { - "type" : "string", - "description" : "the OpenAIRE identifier of the data source" + "$schema": "http://json-schema.org/draft-07/schema#", + "definitions": { + "CfHbKeyValue": { + "type": "object", + "properties": { + "key": { + "type": "string", + "description": "the OpenAIRE identifier of the data source" }, - "value" : { - "type" : "string", - "description" : "the name of the data source" + "value": { + "type": "string", + "description": "the name of the data source" } } }, - "Provenance" : { - "type" : "object", - "properties" : { - "provenance" : { - "type" : "string" - }, - "trust" : { - "type" : "string" - } + "Provenance": { + "type": "object", + "properties": { + "provenance": {"type": "string"}, + "trust": {"type": "string"} } }, - "ResultPid" : { - "type" : "object", - "properties" : { - "scheme" : { - "type" : "string", - "description" : "The scheme of the persistent identifier for the result (i.e. doi). If the pid is here it means the information for the pid has been collected from an authority for that pid type (i.e. Crossref/Datacite for doi). The set of authoritative pid is: doi when collected from Crossref or Datacite pmid when collected from EuroPubmed, arxiv when collected from arXiv, handle from the repositories" + "ResultPid": { + "type": "object", + "properties": { + "scheme": { + "type": "string", + "description": "The scheme of the persistent identifier for the result (i.e. doi). If the pid is here it means the information for the pid has been collected from an authority for that pid type (i.e. Crossref/Datacite for doi). The set of authoritative pid is: doi when collected from Crossref or Datacite pmid when collected from EuroPubmed, arxiv when collected from arXiv, handle from the repositories" }, - "value" : { - "type" : "string", - "description" : "The value expressed in the scheme (i.e. 10.1000/182)" + "value": { + "type": "string", + "description": "The value expressed in the scheme (i.e. 10.1000/182)" } } } }, - "type" : "object", - "properties" : { - "author" : { - "type" : "array", - "items" : { - "type" : "object", - "properties" : { - "fullname" : { - "type" : "string" + "type": "object", + "properties": { + "affiliation": { + "description": "The list of organizations the result is affiliated to", + "type": "array", + "items": { + "type": "object", + "properties": { + "id": { + "type": "string", + "description": "the OpenAIRE id of the organizaiton" }, - "name" : { - "type" : "string" + "name": { + "type": "string", + "description": "the name of the organization" }, - "pid" : { - "type" : "object", - "properties" : { - "id" : { - "type" : "object", - "properties" : { - "scheme" : { - "type" : "string", - "description" : "The author's pid scheme. OpenAIRE currently supports 'ORCID'" + "pid": { + "description": "the list of pids we have in OpenAIRE for the organization", + "type": "array", + "items": { + "type": "object", + "properties": { + "type": { + "type": "string", + "description": "the type of the organization pid" + }, + "value": { + "type": "string", + "description": "the value of the organization pid" + } + }, + "description": "the list of pids we have in OpenAIRE for the organization" + } + } + }, + "description": "The list of organizations the result is affiliated to" + } + }, + "author": { + "type": "array", + "items": { + "type": "object", + "properties": { + "fullname": {"type": "string"}, + "name": {"type": "string"}, + "pid": { + "type": "object", + "properties": { + "id": { + "type": "object", + "properties": { + "scheme": { + "type": "string", + "description": "The author's pid scheme. OpenAIRE currently supports 'ORCID'" }, - "value" : { - "type" : "string", - "description" : "The author's pid value in that scheme (i.e. 0000-1111-2222-3333)" + "value": { + "type": "string", + "description": "The author's pid value in that scheme (i.e. 0000-1111-2222-3333)" } } }, - "provenance" : { - "allOf" : [ { - "$ref" : "#/definitions/Provenance" - }, { - "description" : "The reason why the pid was associated to the author" - } ] + "provenance": { + "allOf": [ + {"$ref": "#/definitions/Provenance"}, + {"description": "The reason why the pid was associated to the author"} + ] } }, - "description" : "The author's persistent identifiers" + "description": "The author's persistent identifiers" }, - "rank" : { - "type" : "integer" - }, - "surname" : { - "type" : "string" - } + "rank": {"type": "integer"}, + "surname": {"type": "string"} } } }, - "bestaccessright" : { - "type" : "object", - "properties" : { - "code" : { - "type" : "string", - "description" : "COAR access mode code: http://vocabularies.coar-repositories.org/documentation/access_rights/" + "bestaccessright": { + "type": "object", + "properties": { + "code": { + "type": "string", + "description": "COAR access mode code: http://vocabularies.coar-repositories.org/documentation/access_rights/" }, - "label" : { - "type" : "string", - "description" : "Label for the access mode" + "label": { + "type": "string", + "description": "Label for the access mode" }, - "scheme" : { - "type" : "string", - "description" : "Scheme of reference for access right code. Always set to COAR access rights vocabulary: http://vocabularies.coar-repositories.org/documentation/access_rights/" + "scheme": { + "type": "string", + "description": "Scheme of reference for access right code. Always set to COAR access rights vocabulary: http://vocabularies.coar-repositories.org/documentation/access_rights/" } }, - "description" : "The openest of the access rights of this result." + "description": "The openest of the access rights of this result." }, - "codeRepositoryUrl" : { - "type" : "string", - "description" : "Only for results with type 'software': the URL to the repository with the source code" + "codeRepositoryUrl": { + "type": "string", + "description": "Only for results with type 'software': the URL to the repository with the source code" }, - "contactgroup" : { - "description" : "Only for results with type 'software': Information on the group responsible for providing further information regarding the resource", - "type" : "array", - "items" : { - "type" : "string", - "description" : "Only for results with type 'software': Information on the group responsible for providing further information regarding the resource" + "contactgroup": { + "description": "Only for results with type 'software': Information on the group responsible for providing further information regarding the resource", + "type": "array", + "items": { + "type": "string", + "description": "Only for results with type 'software': Information on the group responsible for providing further information regarding the resource" } }, - "contactperson" : { - "description" : "Only for results with type 'software': Information on the person responsible for providing further information regarding the resource", - "type" : "array", - "items" : { - "type" : "string", - "description" : "Only for results with type 'software': Information on the person responsible for providing further information regarding the resource" + "contactperson": { + "description": "Only for results with type 'software': Information on the person responsible for providing further information regarding the resource", + "type": "array", + "items": { + "type": "string", + "description": "Only for results with type 'software': Information on the person responsible for providing further information regarding the resource" } }, - "container" : { - "type" : "object", - "properties" : { - "conferencedate" : { - "type" : "string" + "container": { + "type": "object", + "properties": { + "conferencedate": {"type": "string"}, + "conferenceplace": {"type": "string"}, + "edition": { + "type": "string", + "description": "Edition of the journal or conference proceeding" }, - "conferenceplace" : { - "type" : "string" + "ep": { + "type": "string", + "description": "End page" }, - "edition" : { - "type" : "string", - "description" : "Edition of the journal or conference proceeding" + "iss": { + "type": "string", + "description": "Journal issue number" }, - "ep" : { - "type" : "string", - "description" : "End page" + "issnLinking": {"type": "string"}, + "issnOnline": {"type": "string"}, + "issnPrinted": {"type": "string"}, + "name": { + "type": "string", + "description": "Name of the journal or conference" }, - "iss" : { - "type" : "string", - "description" : "Journal issue number" + "sp": { + "type": "string", + "description": "Start page" }, - "issnLinking" : { - "type" : "string" - }, - "issnOnline" : { - "type" : "string" - }, - "issnPrinted" : { - "type" : "string" - }, - "name" : { - "type" : "string", - "description" : "Name of the journal or conference" - }, - "sp" : { - "type" : "string", - "description" : "Start page" - }, - "vol" : { - "type" : "string", - "description" : "Volume" + "vol": { + "type": "string", + "description": "Volume" } }, - "description" : "Container has information about the conference or journal where the result has been presented or published" + "description": "Container has information about the conference or journal where the result has been presented or published" }, - "context" : { - "description" : "Reference to a relevant research infrastructure, initiative or community (RI/RC) among those collaborating with OpenAIRE. Please see https://connect.openaire.eu", - "type" : "array", - "items" : { - "type" : "object", - "properties" : { - "code" : { - "type" : "string", - "description" : "Code identifying the RI/RC" + "context": { + "description": "Reference to a relevant research infrastructure, initiative or community (RI/RC) among those collaborating with OpenAIRE. Please see https://connect.openaire.eu", + "type": "array", + "items": { + "type": "object", + "properties": { + "code": { + "type": "string", + "description": "Code identifying the RI/RC" }, - "label" : { - "type" : "string", - "description" : "Label of the RI/RC" + "label": { + "type": "string", + "description": "Label of the RI/RC" }, - "provenance" : { - "description" : "Why this result is associated to the RI/RC.", - "type" : "array", - "items" : { - "allOf" : [ { - "$ref" : "#/definitions/Provenance" - }, { - "description" : "Why this result is associated to the RI/RC." - } ] + "provenance": { + "description": "Why this result is associated to the RI/RC.", + "type": "array", + "items": { + "allOf": [ + {"$ref": "#/definitions/Provenance"}, + {"description": "Why this result is associated to the RI/RC."} + ] } } }, - "description" : "Reference to a relevant research infrastructure, initiative or community (RI/RC) among those collaborating with OpenAIRE. Please see https://connect.openaire.eu" + "description": "Reference to a relevant research infrastructure, initiative or community (RI/RC) among those collaborating with OpenAIRE. Please see https://connect.openaire.eu" } }, - "contributor" : { - "description" : "Contributors for the result", - "type" : "array", - "items" : { - "type" : "string", - "description" : "Contributors for the result" + "contributor": { + "description": "Contributors for the result", + "type": "array", + "items": { + "type": "string", + "description": "Contributors for the result" } }, - "country" : { - "description" : "The list of countries associated to this result", - "type" : "array", - "items" : { - "type" : "object", - "properties" : { - "code" : { - "type" : "string", - "description" : "ISO 3166-1 alpha-2 country code (i.e. IT)" + "country": { + "description": "The list of countries associated to this result", + "type": "array", + "items": { + "type": "object", + "properties": { + "code": { + "type": "string", + "description": "ISO 3166-1 alpha-2 country code (i.e. IT)" }, - "label" : { - "type" : "string", - "description" : "The label for that code (i.e. Italy)" + "label": { + "type": "string", + "description": "The label for that code (i.e. Italy)" }, - "provenance" : { - "allOf" : [ { - "$ref" : "#/definitions/Provenance" - }, { - "description" : "Why this result is associated to the country." - } ] + "provenance": { + "allOf": [ + {"$ref": "#/definitions/Provenance"}, + {"description": "Why this result is associated to the country."} + ] } }, - "description" : "The list of countries associated to this result" + "description": "The list of countries associated to this result" } }, - "coverage" : { - "type" : "array", - "items" : { - "type" : "string" + "coverage": { + "type": "array", + "items": {"type": "string"} + }, + "dateofcollection": { + "type": "string", + "description": "When OpenAIRE collected the record the last time" + }, + "description": { + "type": "array", + "items": {"type": "string"} + }, + "documentationUrl": { + "description": "Only for results with type 'software': URL to the software documentation", + "type": "array", + "items": { + "type": "string", + "description": "Only for results with type 'software': URL to the software documentation" } }, - "dateofcollection" : { - "type" : "string", - "description" : "When OpenAIRE collected the record the last time" + "embargoenddate": { + "type": "string", + "description": "Date when the embargo ends and this result turns Open Access" }, - "description" : { - "type" : "array", - "items" : { - "type" : "string" - } - }, - "documentationUrl" : { - "description" : "Only for results with type 'software': URL to the software documentation", - "type" : "array", - "items" : { - "type" : "string", - "description" : "Only for results with type 'software': URL to the software documentation" - } - }, - "embargoenddate" : { - "type" : "string", - "description" : "Date when the embargo ends and this result turns Open Access" - }, - "eoscIF" : { - "type" : "object", - "properties" : { - "code" : { - "type" : "string", - "description" : "EOSC-IF local code. Later on it could be populated with a PID (e.g. DOI), but for the time being we stick to a more loose definition." + "eoscIF": { + "type": "object", + "properties": { + "code": { + "type": "string", + "description": "EOSC-IF local code. Later on it could be populated with a PID (e.g. DOI), but for the time being we stick to a more loose definition." }, - "label" : { - "type" : "string", - "description" : "EOSC-IF label" + "label": { + "type": "string", + "description": "EOSC-IF label" }, - "semanticRelation" : { - "type" : "string", - "description" : "EOSC-IF semantic relation (e.g. compliesWith)" + "semanticRelation": { + "type": "string", + "description": "EOSC-IF semantic relation (e.g. compliesWith)" }, - "url" : { - "type" : "string", - "description" : "EOSC-IF url to the guidelines" + "url": { + "type": "string", + "description": "EOSC-IF url to the guidelines" } }, - "description" : "Describes a reference to the EOSC Interoperability Framework (IF) Guidelines" + "description": "Describes a reference to the EOSC Interoperability Framework (IF) Guidelines" }, - "format" : { - "type" : "array", - "items" : { - "type" : "string" + "format": { + "type": "array", + "items": {"type": "string"} + }, + "geolocation": { + "description": "Geolocation information", + "type": "array", + "items": { + "type": "object", + "properties": { + "box": {"type": "string"}, + "place": {"type": "string"}, + "point": {"type": "string"} + }, + "description": "Geolocation information" } }, - "geolocation" : { - "description" : "Geolocation information", - "type" : "array", - "items" : { - "type" : "object", - "properties" : { - "box" : { - "type" : "string" + "id": { + "type": "string", + "description": "The OpenAIRE identifiers for this result" + }, + "instance": { + "description": "Each instance is one specific materialisation or version of the result. For example, you can have one result with three instance: one is the pre-print, one is the post-print, one is te published version", + "type": "array", + "items": { + "type": "object", + "properties": { + "accessright": { + "type": "object", + "properties": { + "code": { + "type": "string", + "description": "COAR access mode code: http://vocabularies.coar-repositories.org/documentation/access_rights/" + }, + "label": { + "type": "string", + "description": "Label for the access mode" + }, + "openAccessRoute": { + "type": "string", + "enum": [ + "gold", + "green", + "hybrid", + "bronze" + ] + }, + "scheme": { + "type": "string", + "description": "Scheme of reference for access right code. Always set to COAR access rights vocabulary: http://vocabularies.coar-repositories.org/documentation/access_rights/" + } + }, + "description": "The accessRights for this materialization of the result" }, - "place" : { - "type" : "string" + "alternateIdentifier": { + "description": "All the identifiers other than pids forged by an authorithy for the pid type (i.e. Crossref for DOIs", + "type": "array", + "items": { + "type": "object", + "properties": { + "scheme": { + "type": "string", + "description": "The scheme of the identifier. It can be a persistent identifier (i.e. doi). If it is present in the alternate identifiers it means it has not been forged by an authority for that pid. For example we collect metadata from an institutional repository that provides as identifier for the result also the doi" + }, + "value": { + "type": "string", + "description": "The value expressed in the scheme" + } + }, + "description": "All the identifiers other than pids forged by an authorithy for the pid type (i.e. Crossref for DOIs" + } }, - "point" : { - "type" : "string" + "articleprocessingcharge": { + "type": "object", + "properties": { + "amount": {"type": "string"}, + "currency": {"type": "string"} + }, + "description": "The money spent to make this book or article available in Open Access. Source for this information is the OpenAPC initiative." + }, + "hostedby": { + "allOf": [ + {"$ref": "#/definitions/CfHbKeyValue"}, + {"description": "Information about the source from which the instance can be viewed or downloaded."} + ] + }, + "license": {"type": "string"}, + "measures": { + "description": "Measures computed for this instance, for example Bip!Finder ones", + "type": "array", + "items": { + "type": "object", + "properties": { + "key": { + "type": "string", + "description": "The measure (i.e. popularity)" + }, + "value": { + "type": "string", + "description": "The value for that measure" + } + }, + "description": "Measures computed for this instance, for example Bip!Finder ones" + } + }, + "pid": { + "type": "array", + "items": {"$ref": "#/definitions/ResultPid"} + }, + "publicationdate": { + "type": "string", + "description": "Date of the research product" + }, + "refereed": { + "type": "string", + "description": "If this instance has been peer-reviewed or not. Allowed values are peerReviewed, nonPeerReviewed, UNKNOWN (as defined in https://api.openaire.eu/vocabularies/dnet:review_levels)" + }, + "type": { + "type": "string", + "description": "The specific sub-type of this instance (see https://api.openaire.eu/vocabularies/dnet:result_typologies following the links)" + }, + "url": { + "description": "URLs to the instance. They may link to the actual full-text or to the landing page at the hosting source. ", + "type": "array", + "items": { + "type": "string", + "description": "URLs to the instance. They may link to the actual full-text or to the landing page at the hosting source. " + } } }, - "description" : "Geolocation information" + "description": "Each instance is one specific materialisation or version of the result. For example, you can have one result with three instance: one is the pre-print, one is the post-print, one is te published version" } }, "keywords": { - }, - "id" : { - "type" : "string", - "description" : "The OpenAIRE identifiers for this result" - }, - "instance" : { - "description" : "Each instance is one specific materialisation or version of the result. For example, you can have one result with three instance: one is the pre-print, one is the post-print, one is te published version", - "type" : "array", - "items" : { - "type" : "object", - "properties" : { - "accessright" : { - "type" : "object", - "properties" : { - "code" : { - "type" : "string", - "description" : "COAR access mode code: http://vocabularies.coar-repositories.org/documentation/access_rights/" - }, - "label" : { - "type" : "string", - "description" : "Label for the access mode" - }, - "openAccessRoute" : { - "type" : "string", - "enum" : [ "gold", "green", "hybrid", "bronze" ] - }, - "scheme" : { - "type" : "string", - "description" : "Scheme of reference for access right code. Always set to COAR access rights vocabulary: http://vocabularies.coar-repositories.org/documentation/access_rights/" - } - }, - "description" : "The accessRights for this materialization of the result" - }, - "alternateIdentifier" : { - "description" : "All the identifiers other than pids forged by an authorithy for the pid type (i.e. Crossref for DOIs", - "type" : "array", - "items" : { - "type" : "object", - "properties" : { - "scheme" : { - "type" : "string", - "description" : "The scheme of the identifier. It can be a persistent identifier (i.e. doi). If it is present in the alternate identifiers it means it has not been forged by an authority for that pid. For example we collect metadata from an institutional repository that provides as identifier for the result also the doi" - }, - "value" : { - "type" : "string", - "description" : "The value expressed in the scheme" - } - }, - "description" : "All the identifiers other than pids forged by an authorithy for the pid type (i.e. Crossref for DOIs" - } - }, - "articleprocessingcharge" : { - "type" : "object", - "properties" : { - "amount" : { - "type" : "string" - }, - "currency" : { - "type" : "string" - } - }, - "description" : "The money spent to make this book or article available in Open Access. Source for this information is the OpenAPC initiative." - }, - - "hostedby" : { - "allOf" : [ { - "$ref" : "#/definitions/CfHbKeyValue" - }, { - "description" : "Information about the source from which the instance can be viewed or downloaded." - } ] - }, - "license" : { - "type" : "string" - }, - "measures" : { - "description" : "Measures computed for this instance, for example Bip!Finder ones", - "type" : "array", - "items" : { - "type" : "object", - "properties" : { - "key" : { - "type" : "string", - "description" : "The measure (i.e. popularity)" - }, - "value" : { - "type" : "string", - "description" : "The value for that measure" - } - }, - "description" : "Measures computed for this instance, for example Bip!Finder ones" - } - }, - "pid" : { - "type" : "array", - "items" : { - "$ref" : "#/definitions/ResultPid" - } - }, - "publicationdate" : { - "type" : "string", - "description" : "Date of the research product" - }, - "refereed" : { - "type" : "string", - "description" : "If this instance has been peer-reviewed or not. Allowed values are peerReviewed, nonPeerReviewed, UNKNOWN (as defined in https://api.openaire.eu/vocabularies/dnet:review_levels)" - }, - "type" : { - "type" : "string", - "description" : "The specific sub-type of this instance (see https://api.openaire.eu/vocabularies/dnet:result_typologies following the links)" - }, - "url" : { - "description" : "URLs to the instance. They may link to the actual full-text or to the landing page at the hosting source. ", - "type" : "array", - "items" : { - "type" : "string", - "description" : "URLs to the instance. They may link to the actual full-text or to the landing page at the hosting source. " - } - } - }, - "description" : "Each instance is one specific materialisation or version of the result. For example, you can have one result with three instance: one is the pre-print, one is the post-print, one is te published version" + "description": "Te list of keywords associated to the result", + "type": "array", + "items": { + "type": "string", + "description": "Te list of keywords associated to the result" } }, - "language" : { - "type" : "object", - "properties" : { - "code" : { - "type" : "string", - "description" : "alpha-3/ISO 639-2 code of the language" + "language": { + "type": "object", + "properties": { + "code": { + "type": "string", + "description": "alpha-3/ISO 639-2 code of the language" }, - "label" : { - "type" : "string", - "description" : "Language label in English" + "label": { + "type": "string", + "description": "Language label in English" } } }, - "lastupdatetimestamp" : { - "type" : "integer", - "description" : "Timestamp of last update of the record in OpenAIRE" + "lastupdatetimestamp": { + "type": "integer", + "description": "Timestamp of last update of the record in OpenAIRE" }, - "maintitle" : { - "type" : "string", - "description" : "A name or title by which a scientific result is known. May be the title of a publication, of a dataset or the name of a piece of software." + "maintitle": { + "type": "string", + "description": "A name or title by which a scientific result is known. May be the title of a publication, of a dataset or the name of a piece of software." }, - "originalId" : { - "description" : "Identifiers of the record at the original sources", - "type" : "array", - "items" : { - "type" : "string", - "description" : "Identifiers of the record at the original sources" + "originalId": { + "description": "Identifiers of the record at the original sources", + "type": "array", + "items": { + "type": "string", + "description": "Identifiers of the record at the original sources" } }, - "pid" : { - "description" : "Persistent identifiers of the result", - "type" : "array", - "items" : { - "allOf" : [ { - "$ref" : "#/definitions/ResultPid" - }, { - "description" : "Persistent identifiers of the result" - } ] + "pid": { + "description": "Persistent identifiers of the result", + "type": "array", + "items": { + "allOf": [ + {"$ref": "#/definitions/ResultPid"}, + {"description": "Persistent identifiers of the result"} + ] } }, - "programmingLanguage" : { - "type" : "string", - "description" : "Only for results with type 'software': the programming language" + "programmingLanguage": { + "type": "string", + "description": "Only for results with type 'software': the programming language" }, - "projects" : { - "description" : "List of projects (i.e. grants) that (co-)funded the production ofn the research results", - "type" : "array", - "items" : { - "type" : "object", - "properties" : { - "acronym" : { - "type" : "string", - "description" : "The acronym of the project" + "projects": { + "description": "List of projects (i.e. grants) that (co-)funded the production ofn the research results", + "type": "array", + "items": { + "type": "object", + "properties": { + "acronym": { + "type": "string", + "description": "The acronym of the project" }, - "code" : { - "type" : "string", - "description" : "The grant agreement number" + "code": { + "type": "string", + "description": "The grant agreement number" }, - "funder" : { - "type" : "object", - "properties" : { - "fundingStream" : { - "type" : "string", - "description" : "Stream of funding (e.g. for European Commission can be H2020 or FP7)" + "funder": { + "type": "object", + "properties": { + "fundingStream": { + "type": "string", + "description": "Stream of funding (e.g. for European Commission can be H2020 or FP7)" }, - "jurisdiction" : { - "type" : "string", - "description" : "Geographical jurisdiction (e.g. for European Commission is EU, for Croatian Science Foundation is HR)" + "jurisdiction": { + "type": "string", + "description": "Geographical jurisdiction (e.g. for European Commission is EU, for Croatian Science Foundation is HR)" }, - "name" : { - "type" : "string", - "description" : "The name of the funder (European Commission)" + "name": { + "type": "string", + "description": "The name of the funder (European Commission)" }, - "shortName" : { - "type" : "string", - "description" : "The short name of the funder (EC)" + "shortName": { + "type": "string", + "description": "The short name of the funder (EC)" } }, - "description" : "Information about the funder funding the project" + "description": "Information about the funder funding the project" }, - "id" : { - "type" : "string", - "description" : "The OpenAIRE id for the project" + "id": { + "type": "string", + "description": "The OpenAIRE id for the project" }, - "provenance" : { - "$ref" : "#/definitions/Provenance" - }, - "title" : { - "type" : "string" - }, - "validated" : { - "type" : "object", - "properties" : { - "validatedByFunder" : { - "type" : "boolean" - }, - "validationDate" : { - "type" : "string" - } + "provenance": {"$ref": "#/definitions/Provenance"}, + "title": {"type": "string"}, + "validated": { + "type": "object", + "properties": { + "validatedByFunder": {"type": "boolean"}, + "validationDate": {"type": "string"} } } }, - "description" : "List of projects (i.e. grants) that (co-)funded the production ofn the research results" + "description": "List of projects (i.e. grants) that (co-)funded the production ofn the research results" } }, - "publicationdate" : { - "type" : "string", - "description" : "Main date of the research product: typically the publication or issued date. In case of a research result with different versions with different dates, the date of the result is selected as the most frequent well-formatted date. If not available, then the most recent and complete date among those that are well-formatted. For statistics, the year is extracted and the result is counted only among the result of that year. Example: Pre-print date: 2019-02-03, Article date provided by repository: 2020-02, Article date provided by Crossref: 2020, OpenAIRE will set as date 2019-02-03, because it’s the most recent among the complete and well-formed dates. If then the repository updates the metadata and set a complete date (e.g. 2020-02-12), then this will be the new date for the result because it becomes the most recent most complete date. However, if OpenAIRE then collects the pre-print from another repository with date 2019-02-03, then this will be the “winning date” because it becomes the most frequent well-formatted date." + "publicationdate": { + "type": "string", + "description": "Main date of the research product: typically the publication or issued date. In case of a research result with different versions with different dates, the date of the result is selected as the most frequent well-formatted date. If not available, then the most recent and complete date among those that are well-formatted. For statistics, the year is extracted and the result is counted only among the result of that year. Example: Pre-print date: 2019-02-03, Article date provided by repository: 2020-02, Article date provided by Crossref: 2020, OpenAIRE will set as date 2019-02-03, because it\u2019s the most recent among the complete and well-formed dates. If then the repository updates the metadata and set a complete date (e.g. 2020-02-12), then this will be the new date for the result because it becomes the most recent most complete date. However, if OpenAIRE then collects the pre-print from another repository with date 2019-02-03, then this will be the \u201cwinning date\u201d because it becomes the most frequent well-formatted date." }, - "publisher" : { - "type" : "string", - "description" : "The name of the entity that holds, archives, publishes prints, distributes, releases, issues, or produces the resource." + "publisher": { + "type": "string", + "description": "The name of the entity that holds, archives, publishes prints, distributes, releases, issues, or produces the resource." }, - "size" : { - "type" : "string", - "description" : "Only for results with type 'dataset': the declared size of the dataset" + "size": { + "type": "string", + "description": "Only for results with type 'dataset': the declared size of the dataset" }, - "source" : { - "description" : "See definition of Dublin Core field dc:source", - "type" : "array", - "items" : { - "type" : "string", - "description" : "See definition of Dublin Core field dc:source" + "source": { + "description": "See definition of Dublin Core field dc:source", + "type": "array", + "items": { + "type": "string", + "description": "See definition of Dublin Core field dc:source" } }, - "subject" : { - "type" : "object", - "properties" : { - "scheme" : { - "type" : "string", - "description" : "OpenAIRE subject classification scheme (https://api.openaire.eu/vocabularies/dnet:subject_classification_typologies)." - }, - "value" : { - "type" : "string", - "description" : "The value for the subject in the selected scheme. When the scheme is 'keyword', it means that the subject is free-text (i.e. not a term from a controlled vocabulary)." - } - } - } - }, - "description" : "Keywords associated to the result" + "subject": { + "type": "object", + "description": "The subject dumped by type associated to the result" + }, + "subtitle": { + "type": "string", + "description": "Explanatory or alternative name by which a scientific result is known." + }, + "tool": { + "description": "Only for results with type 'other': tool useful for the interpretation and/or re-used of the research product", + "type": "array", + "items": { + "type": "string", + "description": "Only for results with type 'other': tool useful for the interpretation and/or re-used of the research product" } }, - "subtitle" : { - "type" : "string", - "description" : "Explanatory or alternative name by which a scientific result is known." + "type": { + "type": "string", + "description": "Type of the result: one of 'publication', 'dataset', 'software', 'other' (see also https://api.openaire.eu/vocabularies/dnet:result_typologies)" }, - "tool" : { - "description" : "Only for results with type 'other': tool useful for the interpretation and/or re-used of the research product", - "type" : "array", - "items" : { - "type" : "string", - "description" : "Only for results with type 'other': tool useful for the interpretation and/or re-used of the research product" - } - }, - "type" : { - "type" : "string", - "description" : "Type of the result: one of 'publication', 'dataset', 'software', 'other' (see also https://api.openaire.eu/vocabularies/dnet:result_typologies)" - }, - "version" : { - "type" : "string", - "description" : "Version of the result" + "version": { + "type": "string", + "description": "Version of the result" } } -} +} \ No newline at end of file diff --git a/dump/src/main/resources/eu/dnetlib/dhp/oa/graph/dump/eoscdump/oozie_app/workflow.xml b/dump/src/main/resources/eu/dnetlib/dhp/oa/graph/dump/eoscdump/oozie_app/workflow.xml index 3e4cfd8..f0e4970 100644 --- a/dump/src/main/resources/eu/dnetlib/dhp/oa/graph/dump/eoscdump/oozie_app/workflow.xml +++ b/dump/src/main/resources/eu/dnetlib/dhp/oa/graph/dump/eoscdump/oozie_app/workflow.xml @@ -96,7 +96,7 @@ --outputPath${workingDir}/communityMap --nameNode${nameNode} --isLookUpUrl${isLookUpUrl} - --singleDeposition${singleDeposition} + --singleDepositionfalse --communityId${communityId} diff --git a/dump/src/main/resources/eu/dnetlib/dhp/oa/graph/dump/projectsubset/oozie_app/workflow.xml b/dump/src/main/resources/eu/dnetlib/dhp/oa/graph/dump/projectsubset/oozie_app/workflow.xml index 4013db9..620f761 100644 --- a/dump/src/main/resources/eu/dnetlib/dhp/oa/graph/dump/projectsubset/oozie_app/workflow.xml +++ b/dump/src/main/resources/eu/dnetlib/dhp/oa/graph/dump/projectsubset/oozie_app/workflow.xml @@ -99,7 +99,7 @@ cluster Dump table project eu.dnetlib.dhp.oa.graph.dump.complete.SparkDumpEntitiesJob - dhp-graph-dump-${projectVersion}.jar + dump-${projectVersion}.jar --executor-memory=${sparkExecutorMemory} --executor-cores=${sparkExecutorCores} @@ -124,7 +124,7 @@ cluster Dump table project eu.dnetlib.dhp.oa.graph.dump.projectssubset.ProjectsSubsetSparkJob - dhp-graph-dump-${projectVersion}.jar + dump-${projectVersion}.jar --executor-memory=${sparkExecutorMemory} --executor-cores=${sparkExecutorCores}