From 9709eca5c76aabe823ba34d976c5d11a9d150b76 Mon Sep 17 00:00:00 2001
From: Pjotr Prins
Date: Sat, 7 Nov 2020 10:06:30 +0000
Subject: RDF: use URIs

---
 bh20sequploader/bh20seq-schema.yml |  8 ++++-
 bh20sequploader/bh20seq-shex.rdf   |  5 +--
 test/data/input/TN_UT2.yaml        |  2 +-
 test/data/regression/TN_UT2.rdf    | 64 +++++++++++++++++++-------------------
 4 files changed, 43 insertions(+), 36 deletions(-)

diff --git a/bh20sequploader/bh20seq-schema.yml b/bh20sequploader/bh20seq-schema.yml
index 861a5ca..645be5e 100644
--- a/bh20sequploader/bh20seq-schema.yml
+++ b/bh20sequploader/bh20seq-schema.yml
@@ -1,6 +1,6 @@
 $base: http://biohackathon.org/bh20-seq-schema
 $namespaces:
-  cc:  http://creativecommons.org/ns#
+  cc:  https://creativecommons.org/ns#
   dc:  http://purl.org/metadata/dublin_core_elements#
   sch: https://schema.org/
   efo: http://www.ebi.ac.uk/efo/
@@ -19,6 +19,8 @@ $graph:
       type: string
       jsonldPredicate:
           _id: https://creativecommons.org/ns#License
+          _type: "@id"
+          noLinkCheck: true
     title:
       doc: Attribution title related to data license
       type: string?
@@ -34,11 +36,15 @@ $graph:
       type: string?
       jsonldPredicate:
           _id: https://creativecommons.org/ns#attributionURL
+          _type: "@id"
+          noLinkCheck: true
     attribution_source:
       doc: Attribution source URL related to data license
       type: string?
       jsonldPredicate:
           _id: https://creativecommons.org/ns#attributionSource
+          _type: "@id"
+          noLinkCheck: true
 
 - name: hostSchema
   type: record
diff --git a/bh20sequploader/bh20seq-shex.rdf b/bh20sequploader/bh20seq-shex.rdf
index fa765dd..1b15c58 100644
--- a/bh20sequploader/bh20seq-shex.rdf
+++ b/bh20sequploader/bh20seq-shex.rdf
@@ -2,6 +2,7 @@ PREFIX : <https://raw.githubusercontent.com/arvados/bh20-seq-resource/master/bh2
 PREFIX MainSchema: <http://biohackathon.org/bh20-seq-schema#MainSchema/>
 PREFIX hostSchema: <http://biohackathon.org/bh20-seq-schema#hostSchema/>
 PREFIX cc:  <https://creativecommons.org/ns#>
+PREFIX cclicenses:  <https://creativecommons.org/licenses/>
 PREFIX dc:  <http://purl.org/metadata/dublin_core_elements#>
 PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>
 PREFIX obo: <http://purl.obolibrary.org/obo/>
@@ -71,9 +72,9 @@ PREFIX wikidata: <http://www.wikidata.org/entity/>
 }
 
 :licenseShape{
-    cc:License xsd:string ;
+    cc:License [ cclicenses:~ ] ;
     dc:Title xsd:string ?;
     cc:attributionName xsd:string ?;
-    cc:attributionURL xsd:string ?;
+    cc:attributionURL /^http/ ;
     cc:attributionSource xsd:string ?;
 }
diff --git a/test/data/input/TN_UT2.yaml b/test/data/input/TN_UT2.yaml
index 2c5f987..9c4370d 100644
--- a/test/data/input/TN_UT2.yaml
+++ b/test/data/input/TN_UT2.yaml
@@ -1,7 +1,7 @@
 id: placeholder
 
 license:
-    license_type: http://creativecommons.org/licenses/by/4.0/
+    license_type: https://creativecommons.org/licenses/by/4.0/
     title: "TN_UT2 - Pegram, Tennessee, USA"
     attribution_name: "Mariah Taylor, Colleen Jonsson"
     attribution_url: https://www.uthsc.edu/medicine/molecular-sciences/faculty-directory/jonsson.php
diff --git a/test/data/regression/TN_UT2.rdf b/test/data/regression/TN_UT2.rdf
index aa1356a..84353a5 100644
--- a/test/data/regression/TN_UT2.rdf
+++ b/test/data/regression/TN_UT2.rdf
@@ -1,33 +1,33 @@
-_:Nc519408f538a4bb39ea09f56ce9bb88e <http://semanticscience.org/resource/SIO_000115> "TN_UT2" .
-_:N299826fb66794f11991cebc5f8d8b24a <http://www.ebi.ac.uk/efo/EFO_0002699> "https://bio.tools/BWA#!" .
-_:N982774a13a39474aae39272447a09574 <https://creativecommons.org/ns#attributionURL> "https://www.uthsc.edu/medicine/molecular-sciences/faculty-directory/jonsson.php" .
-_:N7857c4718ab741169bae9fafa9b47c75 <http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#C25164> "2020-04-26" .
-_:N708d19e87323455c9f9bb317538f742d <http://semanticscience.org/resource/SIO_000116> "Pjotr Prins" .
-_:N95d4a000e14d4bd99b46d6ec84989aaf <http://semanticscience.org/resource/SIO_010055> "SARS-CoV-2/human/USA/TN_UT2/2020" .
-<file:///export/local/home/wrk/iwrk/opensource/code/vg/bh20-seq-resource/test/data/input/placeholder> <http://biohackathon.org/bh20-seq-schema#MainSchema/license> _:N982774a13a39474aae39272447a09574 .
-_:N95d4a000e14d4bd99b46d6ec84989aaf <http://edamontology.org/data_1875> <http://purl.obolibrary.org/obo/NCBITaxon_2697049> .
-_:Nc519408f538a4bb39ea09f56ce9bb88e <http://www.ebi.ac.uk/efo/EFO_0000532> <http://purl.obolibrary.org/obo/NCBITaxon_9606> .
-_:N708d19e87323455c9f9bb317538f742d <http://purl.obolibrary.org/obo/NCIT_C42781> "Colleen Jonsson" .
-_:N982774a13a39474aae39272447a09574 <https://creativecommons.org/ns#attributionName> "Mariah Taylor, Colleen Jonsson" .
-<file:///export/local/home/wrk/iwrk/opensource/code/vg/bh20-seq-resource/test/data/input/placeholder> <http://biohackathon.org/bh20-seq-schema#MainSchema/sample> _:N7857c4718ab741169bae9fafa9b47c75 .
-_:N708d19e87323455c9f9bb317538f742d <http://purl.obolibrary.org/obo/NCIT_C37984> "Regional Biocontainment Laboratory, Memphis, TN" .
-_:N299826fb66794f11991cebc5f8d8b24a <http://purl.obolibrary.org/obo/OBI_0600047> <http://www.ebi.ac.uk/efo/EFO_0008632> .
-_:N708d19e87323455c9f9bb317538f742d <http://semanticscience.org/resource/SIO_000172> "UTHSC, Memphis, Tennessee 38163, USA" .
-_:N7857c4718ab741169bae9fafa9b47c75 <http://purl.obolibrary.org/obo/OBI_0001479> <http://purl.obolibrary.org/obo/NCIT_C155831> .
-_:N708d19e87323455c9f9bb317538f742d <http://purl.obolibrary.org/obo/NCIT_C42781> "TN_UT2" .
-_:N7857c4718ab741169bae9fafa9b47c75 <http://semanticscience.org/resource/SIO_000115> "TN_UT2" .
-_:N708d19e87323455c9f9bb317538f742d <http://semanticscience.org/resource/SIO_000115> <https://orcid.org/0000-0002-2640-7672> .
-<file:///export/local/home/wrk/iwrk/opensource/code/vg/bh20-seq-resource/test/data/input/placeholder> <http://biohackathon.org/bh20-seq-schema#MainSchema/submitter> _:N708d19e87323455c9f9bb317538f742d .
-<file:///export/local/home/wrk/iwrk/opensource/code/vg/bh20-seq-resource/test/data/input/placeholder> <http://biohackathon.org/bh20-seq-schema#MainSchema/technology> _:N299826fb66794f11991cebc5f8d8b24a .
-_:N982774a13a39474aae39272447a09574 <http://purl.org/metadata/dublin_core_elements#Title> "TN_UT2 - Pegram, Tennessee, USA" .
-_:N299826fb66794f11991cebc5f8d8b24a <http://semanticscience.org/resource/SIO_001167> "Oxford Nanopore MiniIon RNA long reads" .
-_:N7857c4718ab741169bae9fafa9b47c75 <http://purl.obolibrary.org/obo/GAZ_00000448> <http://www.wikidata.org/entity/Q3289517> .
-<file:///export/local/home/wrk/iwrk/opensource/code/vg/bh20-seq-resource/test/data/input/placeholder> <http://biohackathon.org/bh20-seq-schema#MainSchema/host> _:Nc519408f538a4bb39ea09f56ce9bb88e .
-_:N708d19e87323455c9f9bb317538f742d <http://purl.obolibrary.org/obo/NCIT_C42781> "Mariah Taylor" .
-_:N708d19e87323455c9f9bb317538f742d <http://semanticscience.org/resource/SIO_000115> <https://orcid.org/0000-0002-8021-9162> .
-<file:///export/local/home/wrk/iwrk/opensource/code/vg/bh20-seq-resource/test/data/input/placeholder> <http://biohackathon.org/bh20-seq-schema#MainSchema/virus> _:N95d4a000e14d4bd99b46d6ec84989aaf .
-_:N708d19e87323455c9f9bb317538f742d <http://semanticscience.org/resource/SIO_000116> "Colleen B. Jonsson" .
-_:N708d19e87323455c9f9bb317538f742d <http://www.ebi.ac.uk/efo/EFO_0001741> "TN_UT2" .
-_:N982774a13a39474aae39272447a09574 <https://creativecommons.org/ns#License> "http://creativecommons.org/licenses/by/4.0/" .
-_:N708d19e87323455c9f9bb317538f742d <http://semanticscience.org/resource/SIO_000116> "Mariah Taylor" .
+_:N850430bbc537473e8fbab09c300daf2b <https://creativecommons.org/ns#License> <https://creativecommons.org/licenses/by/4.0/> .
+_:Ncab40e8bd4b84431aa6782f7e7480ef3 <http://www.ebi.ac.uk/efo/EFO_0000532> <http://purl.obolibrary.org/obo/NCBITaxon_9606> .
+_:N3c28bc8014134d77823451de114c6085 <http://purl.obolibrary.org/obo/NCIT_C42781> "Mariah Taylor" .
+_:Ncab40e8bd4b84431aa6782f7e7480ef3 <http://semanticscience.org/resource/SIO_000115> "TN_UT2" .
+<file:///export/local/home/wrk/iwrk/opensource/code/vg/bh20-seq-resource/test/data/input/placeholder> <http://biohackathon.org/bh20-seq-schema#MainSchema/license> _:N850430bbc537473e8fbab09c300daf2b .
+_:N3c28bc8014134d77823451de114c6085 <http://purl.obolibrary.org/obo/NCIT_C42781> "TN_UT2" .
+_:N850430bbc537473e8fbab09c300daf2b <https://creativecommons.org/ns#attributionURL> <https://www.uthsc.edu/medicine/molecular-sciences/faculty-directory/jonsson.php> .
+_:N3c28bc8014134d77823451de114c6085 <http://semanticscience.org/resource/SIO_000172> "UTHSC, Memphis, Tennessee 38163, USA" .
+_:Nebe62f68542a47d0936e953195cf25d7 <http://purl.obolibrary.org/obo/OBI_0001479> <http://purl.obolibrary.org/obo/NCIT_C155831> .
+_:N3c28bc8014134d77823451de114c6085 <http://semanticscience.org/resource/SIO_000116> "Colleen B. Jonsson" .
+_:N78bfdbe9d12745e99164d619cb125277 <http://edamontology.org/data_1875> <http://purl.obolibrary.org/obo/NCBITaxon_2697049> .
+_:Nbf9e7b47381b4965926a47bf5b34e489 <http://semanticscience.org/resource/SIO_001167> "Oxford Nanopore MiniIon RNA long reads" .
+<file:///export/local/home/wrk/iwrk/opensource/code/vg/bh20-seq-resource/test/data/input/placeholder> <http://biohackathon.org/bh20-seq-schema#MainSchema/host> _:Ncab40e8bd4b84431aa6782f7e7480ef3 .
+_:Nbf9e7b47381b4965926a47bf5b34e489 <http://www.ebi.ac.uk/efo/EFO_0002699> "https://bio.tools/BWA#!" .
+_:Nebe62f68542a47d0936e953195cf25d7 <http://semanticscience.org/resource/SIO_000115> "TN_UT2" .
+<file:///export/local/home/wrk/iwrk/opensource/code/vg/bh20-seq-resource/test/data/input/placeholder> <http://biohackathon.org/bh20-seq-schema#MainSchema/sample> _:Nebe62f68542a47d0936e953195cf25d7 .
+_:N850430bbc537473e8fbab09c300daf2b <https://creativecommons.org/ns#attributionName> "Mariah Taylor, Colleen Jonsson" .
+_:N78bfdbe9d12745e99164d619cb125277 <http://semanticscience.org/resource/SIO_010055> "SARS-CoV-2/human/USA/TN_UT2/2020" .
+_:N3c28bc8014134d77823451de114c6085 <http://purl.obolibrary.org/obo/NCIT_C42781> "Colleen Jonsson" .
+_:N3c28bc8014134d77823451de114c6085 <http://semanticscience.org/resource/SIO_000115> <https://orcid.org/0000-0002-2640-7672> .
+_:Nebe62f68542a47d0936e953195cf25d7 <http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#C25164> "2020-04-26" .
+_:Nebe62f68542a47d0936e953195cf25d7 <http://purl.obolibrary.org/obo/GAZ_00000448> <http://www.wikidata.org/entity/Q3289517> .
+_:N850430bbc537473e8fbab09c300daf2b <http://purl.org/metadata/dublin_core_elements#Title> "TN_UT2 - Pegram, Tennessee, USA" .
+_:Nbf9e7b47381b4965926a47bf5b34e489 <http://purl.obolibrary.org/obo/OBI_0600047> <http://www.ebi.ac.uk/efo/EFO_0008632> .
+_:N3c28bc8014134d77823451de114c6085 <http://semanticscience.org/resource/SIO_000116> "Mariah Taylor" .
+_:N3c28bc8014134d77823451de114c6085 <http://www.ebi.ac.uk/efo/EFO_0001741> "TN_UT2" .
+_:N3c28bc8014134d77823451de114c6085 <http://semanticscience.org/resource/SIO_000116> "Pjotr Prins" .
+<file:///export/local/home/wrk/iwrk/opensource/code/vg/bh20-seq-resource/test/data/input/placeholder> <http://biohackathon.org/bh20-seq-schema#MainSchema/submitter> _:N3c28bc8014134d77823451de114c6085 .
+_:N3c28bc8014134d77823451de114c6085 <http://semanticscience.org/resource/SIO_000115> <https://orcid.org/0000-0002-8021-9162> .
+<file:///export/local/home/wrk/iwrk/opensource/code/vg/bh20-seq-resource/test/data/input/placeholder> <http://biohackathon.org/bh20-seq-schema#MainSchema/virus> _:N78bfdbe9d12745e99164d619cb125277 .
+_:N3c28bc8014134d77823451de114c6085 <http://purl.obolibrary.org/obo/NCIT_C37984> "Regional Biocontainment Laboratory, Memphis, TN" .
+<file:///export/local/home/wrk/iwrk/opensource/code/vg/bh20-seq-resource/test/data/input/placeholder> <http://biohackathon.org/bh20-seq-schema#MainSchema/technology> _:Nbf9e7b47381b4965926a47bf5b34e489 .
 
-- 
cgit v1.2.3