def org_top_categories(orgs):
g = Graph()
for org_name in orgs:
org_uri = waan_uri(org_name)
ln = local_name(org_uri)
top_cat = load_incites_json_file(org_name, 'categories-by-year')
if len(top_cat) == 0:
logger.warning("{} file is empty.".format(org_name))
continue
for item in top_cat:
cat = item['category']
for tc_yr in item['counts']:
count = tc_yr['count']
year = tc_yr['year']
category_uri = get_category_uri(cat)
curi = D['topcategory-'] + ln + slugify(cat) + '-{}'.format(
year)
g.add((curi, RDF.type, WOS.InCitesTopCategory))
g.add(
(curi, RDFS.label, Literal("{} - {}".format(org_name,
cat))))
g.add((curi, WOS.number, Literal(count)))
g.add((curi, WOS.year, Literal(year)))
g.add((curi, VIVO.relates, category_uri))
g.add((curi, VIVO.relates, org_uri))
ng = settings.INCITES_TOP_CATEGORIES
backend.sync_updates(ng, g)
return True
def map_categories():
jrn_key = get_journals()
cat_key = read_categories(CATEGORY_FILE)
g = Graph()
# jrnls to categories
g += map_journals_to_categories(jrn_key, cat_key)
added, removed = backend.sync_updates(CATEGORY_NG, g)
return added, removed
def org_total_counts(orgs):
g = Graph()
for org_name in orgs:
org_uri = waan_uri(org_name)
ln = local_name(org_uri)
pcounts = load_incites_json_file(org_name, 'total')
if len(pcounts) == 0:
logger.warning("{} file is empty.".format(org_name))
continue
for item in pcounts:
curi = D['pubcount-' + ln + '-' + str(item['year'])]
g.add((curi, RDF.type, WOS.InCitesPubPerYear))
g.add((curi, RDFS.label,
Literal("{} - {}".format(item['year'], item['count']))))
g.add((curi, WOS.number, Literal(item['count'])))
g.add((curi, WOS.year, Literal(item['year'])))
g.add((org_uri, VIVO.relates, curi))
ng = settings.INCITES_PUB_YEAR_COUNTS
backend.sync_updates(ng, g)
return True
def process(triple_files, format="nt", dry=False, sync=False, sleep=10, size=DEFAULT_BATCH_SIZE):
vstore = backend.get_store()
for fpath in triple_files:
g = Graph()
g.parse(source=fpath, format=format)
named_graph = NG_BASE + fpath.split("/")[-1].split(".")[0]
logger.info("Processing updates with {} triples to {} and batch size {}.".format(len(g), named_graph, size))
if dry is True:
logger.info("Dry run. No changes made.")
else:
if sync is True:
logger.info("Syncing graph to {}.".format(named_graph))
added, removed = backend.sync_updates(named_graph, g, size=size)
else:
logger.info("Posting graph as updates to {}.".format(named_graph))
added = vstore.bulk_add(named_graph, g, size=size)
removed = 0
if (added > 0) or (removed > 0):
if sleep > 0:
logger.info("Sleeping for {} seconds between files.".format(sleep))
time.sleep(sleep)
else:
logger.info("No changes made to {}.".format(named_graph))
return True