Merge pull request #52 from cthoyt/add-source

Add source column to merged context CSVs

src/prefixmaps/data/bioregistry.csv

@@ -776,6 +776,7 @@ bioregistry,INO,http://purl.obolibrary.org/obo/INO_,canonical
 bioregistry,insdc,http://identifiers.org/insdc/,canonical
 bioregistry,insdc.cds,http://identifiers.org/ncbiprotein/,namespace_alias
 bioregistry,insdc.gca,http://identifiers.org/assembly/,namespace_alias
+bioregistry,insdc.gcf,http://identifiers.org/assembly/,namespace_alias
 bioregistry,insdc.run,https://www.ebi.ac.uk/ena/browser/view/,canonical
 bioregistry,insdc.sra,http://identifiers.org/insdc.sra/,canonical
 bioregistry,intact,http://identifiers.org/intact/,canonical
@@ -1034,6 +1035,7 @@ bioregistry,NCBI_gi,https://www.ncbi.nlm.nih.gov/nucleotide/,namespace_alias
 bioregistry,NCBI_taxid,http://purl.obolibrary.org/obo/NCBITaxon_,namespace_alias
 bioregistry,NCBI_Taxon_ID,http://purl.obolibrary.org/obo/NCBITaxon_,namespace_alias
 bioregistry,ncbibook,https://www.ncbi.nlm.nih.gov/books/,canonical
+bioregistry,ncbidrs,http://identifiers.org/ncbidrs/,canonical
 bioregistry,NCBIGene,http://identifiers.org/ncbigene/,canonical
 bioregistry,ncbigi,https://www.ncbi.nlm.nih.gov/nucleotide/,namespace_alias
 bioregistry,NCBIProtein,http://identifiers.org/ncbiprotein/,canonical

src/prefixmaps/data/bioregistry.upper.csv

@@ -776,6 +776,7 @@ bioregistry,INO,http://purl.obolibrary.org/obo/INO_,canonical
 bioregistry,INSDC,http://identifiers.org/insdc/,canonical
 bioregistry,insdc.cds,http://identifiers.org/ncbiprotein/,namespace_alias
 bioregistry,INSDC.GCA,http://identifiers.org/assembly/,namespace_alias
+bioregistry,INSDC.GCF,http://identifiers.org/assembly/,namespace_alias
 bioregistry,INSDC.RUN,https://www.ebi.ac.uk/ena/browser/view/,canonical
 bioregistry,INSDC.SRA,http://identifiers.org/insdc.sra/,canonical
 bioregistry,INTACT,http://identifiers.org/intact/,canonical
@@ -1034,6 +1035,7 @@ bioregistry,NCBI_GI,https://www.ncbi.nlm.nih.gov/nucleotide/,namespace_alias
 bioregistry,NCBI_taxid,http://purl.obolibrary.org/obo/NCBITaxon_,namespace_alias
 bioregistry,NCBI_Taxon_ID,http://purl.obolibrary.org/obo/NCBITaxon_,namespace_alias
 bioregistry,NCBIBOOK,https://www.ncbi.nlm.nih.gov/books/,canonical
+bioregistry,NCBIDRS,http://identifiers.org/ncbidrs/,canonical
 bioregistry,NCBIGene,http://identifiers.org/ncbigene/,canonical
 bioregistry,NCBIGI,https://www.ncbi.nlm.nih.gov/nucleotide/,namespace_alias
 bioregistry,NCBIProtein,http://identifiers.org/ncbiprotein/,canonical

src/prefixmaps/data/prefixcc.csv

@@ -196,6 +196,7 @@ prefixcc,biol,http://purl.org/NET/biol/ns#,canonical
 prefixcc,biolink,https://w3id.org/biolink/vocab/,namespace_alias
 prefixcc,biopax,http://www.biopax.org/release/biopax-level3.owl#,canonical
 prefixcc,biordf,http://purl.org/net/biordfmicroarray/ns#,canonical
+prefixcc,bioschemas,https://bioschemas.org/,canonical
 prefixcc,bioskos,http://eulersharp.sourceforge.net/2003/03swap/bioSKOSSchemes#,canonical
 prefixcc,biotop,http://purl.org/biotop/biotop.owl#,canonical
 prefixcc,biro,http://purl.org/spar/biro/,canonical
@@ -287,6 +288,7 @@ prefixcc,cdtype,http://purl.org/cld/cdtype/,canonical
 prefixcc,centrifuge,http://purl.org/twc/vocab/centrifuge#,canonical
 prefixcc,ceo,https://linkeddata.cultureelerfgoed.nl/def/ceo#,canonical
 prefixcc,ceox,https://linkeddata.cultureelerfgoed.nl/def/ceox#,canonical
+prefixcc,cercabib,https://cercabib.ub.edu/,canonical
 prefixcc,cerealstoo,http://rdf.ag/o/cerealstoo#,canonical
 prefixcc,cerif,http://spi-fm.uca.es/neologism/cerif#,canonical
 prefixcc,cert,http://www.w3.org/ns/auth/cert#,canonical
@@ -773,6 +775,7 @@ prefixcc,evset,http://dsnotify.org/vocab/eventset/0.1/,canonical
 prefixcc,ewg,http://ethoinformatics.org/,canonical
 prefixcc,ex,http://example.org/,canonical
 prefixcc,example,http://www.example.org/rdf#,canonical
+prefixcc,exekg,https://raw.githubusercontent.com/nsai-uio/ExeKGOntology/main/ds_exeKGOntology.ttl#,canonical
 prefixcc,exif,http://www.w3.org/2003/12/exif/ns#,canonical
 prefixcc,exo,https://w3id.org/example#,canonical
 prefixcc,experts,http://emmo.info/emmo/application/maeo/experts#,canonical
@@ -1407,6 +1410,7 @@ prefixcc,mi,http://www.marineinfo.org/ns/ontology#,canonical
 prefixcc,mibc,http://marineinfo.org/ns/library/bibcodes#,canonical
 prefixcc,mibt,https://marineinfo.org/ns/library/bibtypes#,canonical
 prefixcc,mico,http://www.mico-project.eu/ns/platform/1.0/schema#,canonical
+prefixcc,mifesto,https://w3id.org/mifesto#,canonical
 prefixcc,mil,http://rdf.muninn-project.org/ontologies/military#,canonical
 prefixcc,mime,https://www.iana.org/assignments/media-types/,canonical
 prefixcc,mindat,https://www.mindat.org/,canonical
@@ -1489,8 +1493,8 @@ prefixcc,ncal,http://www.semanticdesktop.org/ontologies/2007/04/02/ncal#,canonic
 prefixcc,ncbi,https://www.ncbi.nlm.nih.gov/,canonical
 prefixcc,ncbigene,http://identifiers.org/ncbigene/,canonical
 prefixcc,ncbitaxon,http://purl.org/obo/owl/NCBITaxon#,canonical
-prefixcc,ncicp,http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#,namespace_alias
-prefixcc,ncit,http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#,canonical
+prefixcc,ncicp,http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#,canonical
+prefixcc,ncit,https://ncit.nci.nih.gov/ncitbrowser/ConceptReport.jsp?dictionary=NCI_Thesaurus&ns=ncit&code=,canonical
 prefixcc,nco,http://www.semanticdesktop.org/ontologies/2007/03/22/nco#,canonical
 prefixcc,ndl,http://schemas.ogf.org/nml/2013/05/base#,canonical
 prefixcc,ndnp,http://chroniclingamerica.loc.gov/terms#,canonical
@@ -1779,6 +1783,7 @@ prefixcc,pattern,http://www.essepuntato.it/2008/12/pattern#,canonical
 prefixcc,pav,http://purl.org/pav/,canonical
 prefixcc,pay,http://reference.data.gov.uk/def/payment#,namespace_alias
 prefixcc,payment,http://reference.data.gov.uk/def/payment#,canonical
+prefixcc,pbac,https://w3id.org/pbac#,canonical
 prefixcc,pbo,http://purl.org/ontology/pbo/core#,canonical
 prefixcc,pbody,http://reference.data.gov.uk/def/public-body/,canonical
 prefixcc,pc,http://purl.org/procurement/public-contracts#,canonical
@@ -1955,6 +1960,7 @@ prefixcc,qud,http://qudt.org/1.1/schema/qudt#,canonical
 prefixcc,qudt,http://qudt.org/schema/qudt/,canonical
 prefixcc,quest,https://rb.gy/ntg7l/,canonical
 prefixcc,quid,https://w3id.org/quid/,canonical
+prefixcc,quit,http://quit.aksw.org/vocab/,canonical
 prefixcc,quran,http://khalidaloufi.sa/quran#,canonical
 prefixcc,quty,http://www.telegraphis.net/ontology/measurement/quantity#,canonical
 prefixcc,qvoc,http://mlode.nlp2rdf.org/quranvocab#,canonical
@@ -2259,6 +2265,7 @@ prefixcc,security,http://securitytoolbox.appspot.com/securityMain#,canonical
 prefixcc,sede,http://eventography.org/sede/0.1/,canonical
 prefixcc,seeds,http://deductions.github.io/seeds.owl.ttl#,canonical
 prefixcc,sem,http://semanticweb.cs.vu.nl/2009/11/sem/,canonical
+prefixcc,semapv,https://w3id.org/semapv/vocab/,canonical
 prefixcc,semio,http://www.lingvoj.org/semio#,canonical
 prefixcc,semiot,http://w3id.org/semiot/ontologies/semiot#,canonical
 prefixcc,semsur,http://purl.org/SemSur/,canonical
@@ -2601,6 +2608,7 @@ prefixcc,states,http://www.w3.org/2005/07/aaa#,canonical
 prefixcc,static,http://vocab-ld.org/vocab/static-ld#,canonical
 prefixcc,stats,http://purl.org/rdfstats/stats#,canonical
 prefixcc,status,http://www.w3.org/2003/06/sw-vocab-status/ns#,namespace_alias
+prefixcc,stax,https://w3id.org/stax/ontology#,canonical
 prefixcc,steel,http://ontorule-project.eu/resources/steel-30#,canonical
 prefixcc,stencila,http://schema.stenci.la/,canonical
 prefixcc,step,http://purl.org/net/step#,canonical
@@ -2937,6 +2945,7 @@ prefixcc,wotc,http://purl.org/wot-catalogue#,canonical
 prefixcc,wotsec,https://www.w3.org/2019/wot/security#,canonical
 prefixcc,wp,http://vocabularies.wikipathways.org/wp#,canonical
 prefixcc,wro,http://purl.org/net/wf4ever/ro#,canonical
+prefixcc,wrroc,https://w3id.org/ro/terms/workflow-run#,canonical
 prefixcc,ws,http://www.w3.org/ns/pim/space#,namespace_alias
 prefixcc,wsc,http://www.openk.org/wscaim.owl#,canonical
 prefixcc,wscaim,http://www.openk.org/wscaim.owl#,namespace_alias

src/prefixmaps/data/w3id.csv

@@ -139,6 +139,7 @@ w3id,cerrotti,https://w3id.org/cerrotti/,canonical
 w3id,certainty_nanopubs,https://w3id.org/certainty_nanopubs/,canonical
 w3id,cevo,https://w3id.org/cevo/,canonical
 w3id,chainpoint,https://w3id.org/chainpoint/,canonical
+w3id,chalkgrp,https://w3id.org/chalkgrp/,canonical
 w3id,character-computing,https://w3id.org/character-computing/,canonical
 w3id,charity-organization,https://w3id.org/charity-organization/,canonical
 w3id,charta77,https://w3id.org/charta77/,canonical
@@ -161,6 +162,7 @@ w3id,cld,https://w3id.org/cld/,canonical
 w3id,clinga,https://w3id.org/clinga/,canonical
 w3id,clipc,https://w3id.org/clipc/,canonical
 w3id,clodg,https://w3id.org/clodg/,canonical
+w3id,CMECS,https://w3id.org/CMECS/,canonical
 w3id,cmip6dr,https://w3id.org/cmip6dr/,canonical
 w3id,cntf,https://w3id.org/cntf/,canonical
 w3id,cocoon,https://w3id.org/cocoon/,canonical
@@ -312,6 +314,7 @@ w3id,ecodigit,https://w3id.org/ecodigit/,canonical
 w3id,ecsel-dr,https://w3id.org/ecsel-dr/,canonical
 w3id,ecsel-dr-prc-PMV,https://w3id.org/ecsel-dr-prc-PMV/,canonical
 w3id,ecsel-dr-sn-SSP,https://w3id.org/ecsel-dr-sn-SSP/,canonical
+w3id,EDH_Ctagged,https://w3id.org/EDH_Ctagged/,canonical
 w3id,edu-sharing,https://w3id.org/edu-sharing/,canonical
 w3id,education,https://w3id.org/education/,canonical
 w3id,edukg,https://w3id.org/edukg/,canonical
@@ -631,6 +634,7 @@ w3id,mgkb,https://w3id.org/mgkb/,canonical
 w3id,mica,https://w3id.org/mica/,canonical
 w3id,midas-catalog,https://w3id.org/midas-catalog/,canonical
 w3id,midas-metadata,https://w3id.org/midas-metadata/,canonical
+w3id,mifesto,https://w3id.org/mifesto/,canonical
 w3id,minerva,https://w3id.org/minerva/,canonical
 w3id,mint,https://w3id.org/mint/,canonical
 w3id,minte,https://w3id.org/minte/,canonical
@@ -787,10 +791,12 @@ w3id,paradise,https://w3id.org/paradise/,canonical
 w3id,patent_ontologies,https://w3id.org/patent_ontologies/,canonical
 w3id,payments,https://w3id.org/payments/,canonical
 w3id,payswarm,https://w3id.org/payswarm/,canonical
+w3id,pbac,https://w3id.org/pbac/,canonical
 w3id,pbs,https://w3id.org/pbs/,canonical
 w3id,pc,https://w3id.org/pc/,canonical
 w3id,peco,https://w3id.org/peco/,canonical
 w3id,pedigree,https://w3id.org/pedigree/,canonical
+w3id,peh,https://w3id.org/peh/,canonical
 w3id,people,https://w3id.org/people/,canonical
 w3id,pep,https://w3id.org/pep/,canonical
 w3id,per,https://w3id.org/per/,canonical
@@ -834,6 +840,7 @@ w3id,qb4solap,https://w3id.org/qb4solap/,canonical
 w3id,quality,https://w3id.org/quality/,canonical
 w3id,qudt,https://w3id.org/qudt/,canonical
 w3id,quid,https://w3id.org/quid/,canonical
+w3id,r74n,https://w3id.org/r74n/,canonical
 w3id,rai,https://w3id.org/rai/,canonical
 w3id,rail,https://w3id.org/rail/,canonical
 w3id,rains,https://w3id.org/rains/,canonical
@@ -929,6 +936,7 @@ w3id,seneca,https://w3id.org/seneca/,canonical
 w3id,sense,https://w3id.org/sense/,canonical
 w3id,sentitrack,https://w3id.org/sentitrack/,canonical
 w3id,seo,https://w3id.org/seo/,canonical
+w3id,seovoc,https://w3id.org/seovoc/,canonical
 w3id,sepses,https://w3id.org/sepses/,canonical
 w3id,serdif,https://w3id.org/serdif/,canonical
 w3id,sfs-ontology,https://w3id.org/sfs-ontology/,canonical
@@ -975,6 +983,7 @@ w3id,SpOTy,https://w3id.org/SpOTy/,canonical
 w3id,sqo,https://w3id.org/sqo/,canonical
 w3id,squap,https://w3id.org/squap/,canonical
 w3id,squirrel,https://w3id.org/squirrel/,canonical
+w3id,sri,https://w3id.org/sri/,canonical
 w3id,sri-lanka,https://w3id.org/sri-lanka/,canonical
 w3id,srmo,https://w3id.org/srmo/,canonical
 w3id,srr,https://w3id.org/srr/,canonical
@@ -984,6 +993,7 @@ w3id,sssom,https://w3id.org/sssom/,canonical
 w3id,stahl,https://w3id.org/stahl/,canonical
 w3id,stargate-h2020,https://w3id.org/stargate-h2020/,canonical
 w3id,stav,https://w3id.org/stav/,canonical
+w3id,stax,https://w3id.org/stax/,canonical
 w3id,steel,https://w3id.org/steel/,canonical
 w3id,stirdata,https://w3id.org/stirdata/,canonical
 w3id,stlab,https://w3id.org/stlab/,canonical
@@ -1033,6 +1043,7 @@ w3id,tso,https://w3id.org/tso/,canonical
 w3id,tsso,https://w3id.org/tsso/,canonical
 w3id,ttla,https://w3id.org/ttla/,canonical
 w3id,TTRpg,https://w3id.org/TTRpg/,canonical
+w3id,tvstationjp,https://w3id.org/tvstationjp/,canonical
 w3id,twins,https://w3id.org/twins/,canonical
 w3id,uco,https://w3id.org/uco/,canonical
 w3id,ufo,https://w3id.org/ufo/,canonical

src/prefixmaps/datamodel/context.py

@@ -71,6 +71,9 @@ class PrefixExpansion:
     status: StatusType
     """Indicates whether the expansion is canonical, a prefix alias, a namespace alias, or both."""
 
+    expansion_source: Optional[str] = None
+    """Indicates the source of the prefix expansion."""
+
     def canonical(self) -> bool:
         """
         True if this is the canonical mapping in both directions.
@@ -153,14 +156,15 @@ def combine(self, context: "Context"):
         :return:
         """
         for pe in context.prefix_expansions:
-            self.add_prefix(pe.prefix, pe.namespace, pe.status)
+            self.add_prefix(pe.prefix, pe.namespace, pe.status, expansion_source=context.name)
 
     def add_prefix(
         self,
         prefix: PREFIX,
         namespace: NAMESPACE,
         status: StatusType = StatusType.canonical,
         preferred: bool = False,
+        expansion_source: Optional[str] = None,
     ):
         """
         Adds a prefix expansion to this context.
@@ -176,6 +180,9 @@ def add_prefix(
         :param namespace: namespace to be added
         :param status: the status of the prefix being added
         :param preferred:
+        :param expansion_source: An optional annotation to be used when merging contexts together.
+            The source will keep track of the original context that a given prefix
+            expansion came from. This is used in :meth:`Context.combine`.
         :return:
         """
         # TODO: check status
@@ -203,6 +210,7 @@ def add_prefix(
                 prefix=prefix,
                 namespace=namespace,
                 status=status,
+                expansion_source=expansion_source,
             )
         )
 

src/prefixmaps/ingest/etl_runner.py

@@ -90,7 +90,7 @@ def run_etl(output_directory: Union[str, Path]) -> None:
     # Write all contexts
     for name, context in contexts.items():
         with output_directory.joinpath(f"{name}.csv").open("w", encoding="UTF-8") as file:
-            context_to_file(context, file)
+            context_to_file(context, file, include_expansion_source=context.name in COMBINED)
 
 
 @click.command

src/prefixmaps/io/writer.py

@@ -15,17 +15,27 @@ def _key(pe: PrefixExpansion):
     return pe.prefix.casefold(), STATUS_TYPE_ORDER[pe.status]
 
 
-def context_to_file(context: Context, file: TextIO) -> None:
+def context_to_file(
+    context: Context, file: TextIO, *, include_expansion_source: bool = False
+) -> None:
     """
     Writes a context to a file
 
     :param context:
     :param file:
+    :param include_expansion_source: If true, include a "source" column. This is useful for
+        writing merged contexts since it says the highest priority simple context
+        from which the row corresponding to a :class:`PrefixExpansion` came.
     :return:
     """
-    writer = DictWriter(file, fieldnames=["context", "prefix", "namespace", "status"])
+    field_names = ["context", "prefix", "namespace", "status"]
+    if include_expansion_source:
+        field_names.append("expansion_source")
+    writer = DictWriter(file, fieldnames=field_names)
     writer.writeheader()
     for pe in sorted(context.prefix_expansions, key=_key):
         row = vars(pe)
         row["status"] = pe.status.value
+        if not include_expansion_source:
+            row.pop("expansion_source", None)
         writer.writerow(row)