def graph(self):
if not hasattr(self, '_graph'):
self._graph = populateFromJsonLd(OntGraph(), self.asJsonLd())
OntCuries.populate(self._graph)
self.populateHeader(self._graph)
return self._graph
def export_protcur(self, dump_path, *hypothesis_groups, no_network=False):
# FIXME no_network passed in here is dumb
#if (self.latest and # FIXME NOTE this only points to the latest integrated release
#self.latest_protcur_path.exists()):
#blob_protcur = self.latest_protocols
#else:
pipeline = pipes.ProtcurPipeline(*hypothesis_groups,
no_network=no_network)
# FIXME NOTE this does not do the identifier expansion pass
protcur = pipeline.data
context = {
**sc.base_context,
**sc.protcur_context,
}
for f in ('meta', 'subjects', 'samples', 'contributors'):
context.pop(f) # FIXME HACK meta @graph for datasets
ontology_header = { # FIXME should probably not be added here since it is obscure ...
'@id': 'https://cassava.ucsd.edu/sparc/ontologies/protcur.ttl',
'@type': 'owl:Ontology',
}
protcur.append(ontology_header)
blob_protcur = { # FIXME this should not be defined here so confusing that it is not with the pipeline ...
'@context': context,
'meta': {
'count': len(protcur)
}, # FIXME adjust to structure
'prov': {
'timestamp_export_start': self.timestamp,
'export_system_identifier': Path.sysid,
'export_hostname': gethostname(),
},
'@graph': protcur, # FIXME regularize elements ?
}
dump_path.mkdir(parents=True, exist_ok=True)
# FIXME TODO make these latest paths accessible
# probably by splitting protcur export out into
# its own class
latest_path = dump_path.parent / 'LATEST'
latest_partial_path = dump_path.parent / 'LATEST_PARTIAL'
fn = dump_path / 'protcur.json'
with open(fn, 'wt') as f:
json.dump(blob_protcur, f, sort_keys=True, indent=2, cls=JEncode)
symlink_latest(dump_path, latest_partial_path)
g = populateFromJsonLd(OntGraph(), fn).write(fn.with_suffix('.ttl'))
symlink_latest(dump_path, latest_path)
return blob_protcur
def export_other_formats(self, dump_path, filepath_json, blob_ir,
blob_export_json, *rest):
summary, previous_latest, previous_latest_datasets = rest
dataset_blobs = blob_ir['datasets']
# jsonld
blob_export_jsonld = self.export_jsonld(filepath_json,
blob_export_json)
# identifier metadata
blob_id_met = self.export_identifier_metadata(dump_path,
previous_latest,
dataset_blobs)
teim = self.export_identifier_rdf(dump_path, blob_id_met)
# rdf
teds = self.export_rdf(dump_path, previous_latest_datasets,
dataset_blobs)
tes = ex.TriplesExportSummary(blob_ir, teds=teds + [teim])
# protcur # FIXME running after because rdf export side effects anno sync
blob_protcur = self.export_protcur(
dump_path, 'sparc-curation') # FIXME # handle orthogonally
blob_protcur_path = dump_path / 'protcur.json' # FIXME SIGH
populateFromJsonLd(tes.graph,
blob_protcur_path) # this makes me so happy
with open(filepath_json.with_suffix('.ttl'), 'wb') as f:
f.write(tes.ttl)
# protocol # handled orthogonally ??
#blob_protocol = self.export_protocols(dump_path, dataset_blobs, blob_protcur)
# xml
self.export_xml(filepath_json, dataset_blobs)
# disco
self.export_disco(filepath_json, dataset_blobs, teds)
def main():
dandi_terms_path = aug.LocalPath.cwd()
g = OntGraph()
_ = [
populateFromJsonLd(g, path_yaml(p))
for p in dandi_terms_path.rglob('*.yaml')
]
g.write('dandi-raw.ttl')
remove = [(s, p, o) for p in (schema.domainIncludes, schema.rangeIncludes,
rdfs.subClassOf, rdf.type)
for s, o in g[:p:]]
add = [(s, p, (g.namespace_manager.expand(o.toPython()) if isinstance(
o, rdflib.Literal) else o)) for s, p, o in remove]
_ = [g.remove(t) for t in remove]
_ = [g.add(t) for t in add]
# TODO ontology metadata header section
g.write('dandi.ttl')
def export_protcur(
self,
dump_path,
*hypothesis_groups,
rerun_protcur_export=False,
# FIXME no_network passed in here is dumb
no_network=False,
# FIXME direct= is a hack
direct=False):
if not direct and self.export_base != self.export_protcur_base:
# workaround to set the correct export base path
nargs = {**self._args}
nargs['export_base'] = self.export_protcur_base
export = ExportProtcur(**nargs)
return export.export_protcur(export.dump_path,
*hypothesis_groups,
no_network=no_network), export
pipeline = pipes.ProtcurPipeline(*hypothesis_groups,
no_network=no_network)
annos = pipeline.load()
if not annos:
msg = ('No annos. Did you remember to run\n'
'python -m sparcur.simple.fetch_annotations')
raise ValueError(msg)
if self.latest_export_path.exists():
# FIXME this only points to the latest integrated release
# which is not what we want, we need the latest protcur to be independent
#self.latest and
blob_protcur = self.latest_export
t_lex = blob_protcur['prov']['timestamp_export_start']
t_lup = max(a.updated for a in annos).replace('+00:00', 'Z')
new_annos_here = t_lex < t_lup # <= is pretty much impossible
if not (new_annos_here or rerun_protcur_export):
return blob_protcur
# FIXME NOTE this does not do the identifier expansion pass
protcur = pipeline._make_blob(annos=annos)
context = {
**sc.base_context,
**sc.protcur_context,
}
for f in ('meta', 'subjects', 'samples', 'contributors'):
# subjects samples and contributors no longer included in context directly
if f in context:
context.pop(f) # FIXME HACK meta @graph for datasets
ontology_header = { # FIXME should probably not be added here since it is obscure ...
'@id': 'https://cassava.ucsd.edu/sparc/ontologies/protcur.ttl',
'@type': 'owl:Ontology',
}
protcur.append(ontology_header)
blob_protcur = { # FIXME this should not be defined here so confusing that it is not with the pipeline ...
'@context': context,
'meta': {
'count': len(protcur)
}, # FIXME adjust to structure
'prov': {
'timestamp_export_start': self.timestamp,
'export_system_identifier': Path.sysid,
'export_hostname': gethostname(),
},
'@graph': protcur, # FIXME regularize elements ?
}
dump_path.mkdir(parents=True, exist_ok=True)
# FIXME TODO make these latest paths accessible
# probably by splitting protcur export out into
# its own class
latest_path = dump_path.parent / 'LATEST'
latest_partial_path = dump_path.parent / 'LATEST_PARTIAL'
fn = dump_path / 'protcur.json'
with open(fn, 'wt') as f:
json.dump(blob_protcur, f, sort_keys=True, indent=2, cls=JEncode)
symlink_latest(dump_path, latest_partial_path)
g = populateFromJsonLd(OntGraph(), fn).write(fn.with_suffix('.ttl'))
symlink_latest(dump_path, latest_path)
return blob_protcur