def time_store_select(self,
store,
query_template,
basefile,
context=None,
label="things"):
values = {'basefile': basefile, 'label': label, 'count': None}
uri = self.canonical_uri(basefile)
msg = ("%(basefile)s: selected %(count)s %(label)s "
"(%(elapsed).3f sec)")
with util.logtime(self.log.debug, msg, values):
result = self.store_select(store, query_template, uri, context)
values['count'] = len(result)
return result
def select(self, template, uri, format="json"):
sq = util.readfile(template) % {'uri': uri}
ts = TripleStore.connect(self.config.storetype,
self.config.storelocation,
self.config.storerepository)
print("# Constructing the following from %s, repository %s, type %s" %
(self.config.storelocation,
self.config.storerepository,
self.config.storetype))
print("".join(["# %s\n" % x for x in sq.split("\n")]))
p = {}
with util.logtime(print,
"# Selected in %(elapsed).3fs",
p):
res = ts.select(sq, format=format)
print(res.decode('utf-8'))
def parse(self, basefile):
with util.logtime(self.log.info, "%(basefile)s OK (%(elapsed).3f sec)",
{'basefile': basefile}):
ret = False
for c in self.subrepos:
inst = self.get_instance(c, self.myoptions)
try:
# each parse method should be smart about whether to re-parse
# or not (i.e. use the @managedparsing decorator)
ret = inst.parse(basefile)
except errors.ParseError: # or others
ret = False
if ret:
break
if ret:
self.copy_parsed(basefile, inst)
return ret
def time_store_select(
self, store, query_template, basefile, context=None, label="things"):
values = {'basefile': basefile,
'label': label,
'count': None}
uri = self.canonical_uri(basefile)
msg = ("%(basefile)s: selected %(count)s %(label)s "
"(%(elapsed).3f sec)")
with util.logtime(self.log.debug,
msg,
values):
result = self.store_select(store,
query_template,
uri,
context)
values['count'] = len(result)
return result
def get_treenotice_graph(self, cellarurl, celexid):
# avoid HTTP call if we already have the data
if os.path.exists(self.store.intermediate_path(celexid, suffix=".ttl")):
self.log.info("%s: Opening existing TTL file" % celexid)
with self.store.open_intermediate(celexid, suffix=".ttl") as fp:
return Graph().parse(data=fp.read(), format="ttl")
# FIXME: read the rdf-xml data line by line and construct a
# graph by regex-parsing interesting lines with a very simple
# state machine, rather than doing a full parse, to speed
# things up
resp = util.robust_fetch(self.session.get, cellarurl, self.log, headers={"Accept": "application/rdf+xml;notice=tree"}, timeout=10)
if not resp:
return None
with util.logtime(self.log.info,
"%(basefile)s: parsing the tree notice took %(elapsed).3f s",
{'basefile': celexid}):
graph = Graph().parse(data=resp.content)
return graph
