def __init__(self, db, mapping_db):
self.mime_types = dict(
[('html', 'text/html'), ('pretty-xml', 'application/rdf+xml'), ('turtle', 'text/turtle'),
('nt', 'text/plain'), ('json-ld', 'application/ld+json'), ('sparql', 'application/sparql-results+xml'
), ('json-sparql', 'application/sparql-results+json')])
self.mime_ext = dict(
[('html','.html'), ('pretty-xml', '.rdf'), ('turtle', '.ttl'), ('nt', '.nt')])
self.wordnet_context = WNRDF.WNRDFContext(db, mapping_db)
self.header = open(resolve("header")).read()
self.footer = open(resolve("footer")).read()
def test_synset(self):
context = WNRDF.WNRDFContext("wordnet_3.1+.db")
graph = WNRDF.synset(context, 100001740)
print graph.serialize()
assert(graph)
def test_sample(self):
context = WNRDF.WNRDFContext("wordnet_3.1+.db")
graph = WNRDF.synset(context, 100003993)
serialization = graph.serialize()
assert("American shopkeeper" in serialization)
def test_fail(self):
context = WNRDF.WNRDFContext("wordnet_3.1+.db")
graph = WNRDF.synset(context, 0)
assert(graph is None)
def test_morphs(self):
context = WNRDF.WNRDFContext("wordnet_3.1+.db")
graph = WNRDF.entry(context, "abhor","v")
serialization = graph.serialize()
assert("abhorred" in serialization)
assert("abhorring" in serialization)
def test_adjposition(self):
context = WNRDF.WNRDFContext("wordnet_3.1+.db")
graph = WNRDF.entry(context, "inclined","a")
serialization = graph.serialize()
assert("adjposition" in serialization)
def test_entry(self):
context = WNRDF.WNRDFContext("wordnet_3.1+.db")
graph = WNRDF.entry(context, "cat","n")
print graph.serialize()
assert(graph)
def test_synset(self):
context = WNRDF.WNRDFContext("wordnet_3.1+.db")
graph = WNRDF.synset(context, 100001740)
print graph.serialize()
assert (graph)
def test_sample(self):
context = WNRDF.WNRDFContext("wordnet_3.1+.db")
graph = WNRDF.synset(context, 100003993)
serialization = graph.serialize()
assert ("American shopkeeper" in serialization)
def test_fail(self):
context = WNRDF.WNRDFContext("wordnet_3.1+.db")
graph = WNRDF.synset(context, 0)
assert (graph is None)
def test_morphs(self):
context = WNRDF.WNRDFContext("wordnet_3.1+.db")
graph = WNRDF.entry(context, "abhor", "v")
serialization = graph.serialize()
assert ("abhorred" in serialization)
assert ("abhorring" in serialization)
def test_adjposition(self):
context = WNRDF.WNRDFContext("wordnet_3.1+.db")
graph = WNRDF.entry(context, "inclined", "a")
serialization = graph.serialize()
assert ("adjposition" in serialization)
def test_entry(self):
context = WNRDF.WNRDFContext("wordnet_3.1+.db")
graph = WNRDF.entry(context, "cat", "n")
print graph.serialize()
assert (graph)
def application(self, environ, start_response):
uri = environ['PATH_INFO']
if re.match(".*\.html", uri):
mime = "html"
elif re.match(".*\.rdf", uri):
mime = "pretty-xml"
elif re.match(".*\.ttl", uri):
mime = "turtle"
elif re.match(".*\.nt", uri):
mime = "nt"
elif re.match(".*\.json", uri):
mime = "json-ld"
elif 'HTTP_ACCEPT' in environ:
if uri.startswith("/sparql"):
mime = self.best_mime_type(environ['HTTP_ACCEPT'], "sparql")
else:
mime = self.best_mime_type(environ['HTTP_ACCEPT'], "html")
else:
mime = "html"
if uri == "/" or uri == "/index.html":
start_response('200 OK', [('Content-type', 'text/html')])
return [
self.render_html("WordNet RDF",
open(resolve("index.html")).read())
]
if uri == "/license.html":
start_response('200 OK', [('Content-type', 'text/html')])
return [
self.render_html("WordNet RDF",
open(resolve("license.html")).read())
]
elif uri == "/wnrdf.css" or uri == "/sparql/wnrdf.css":
start_response('200 OK', [('Content-type', 'text/css')])
return [open(resolve("wnrdf.css")).read()]
elif re.match(
"/%s/(\d+)\-[nvarspNVARSP](|\.nt|\.html|\.rdf|\.ttl|\.json)$" %
WNRDF.wn_version, uri):
synset_id = re.findall(
"/%s/(\d+)\-[nvarspNVARSP]" % WNRDF.wn_version, uri)[0]
if len(synset_id) == 8:
synset_id = str(WNRDF.pos2number(uri[-1])) + synset_id
return self.send302(
start_response,
"/%s/%s-%s" % (WNRDF.wn_version, synset_id, uri[-1]))
translate = True
if uri[-1].isupper():
translate = False
graph = WNRDF.synset(self.wordnet_context,
int(synset_id),
extras=mime == "html",
translate=translate)
if graph is None:
return self.send404(start_response)
title = ', '.join(
sorted([
str(o)
for _, _, o in graph.triples((None, RDFS.label, None))
]))
if mime == "html":
content = self.rdfxml_to_html(graph, title)
else:
try:
content = graph.serialize(
format=mime,
context=self.wordnet_context.jsonld_context)
except Exception:
return self.send501(start_response)
start_response('200 OK', [('Content-type', self.mime_types[mime]),
('Vary', 'Accept'),
('Content-length', str(len(content)))])
return [content]
# elif re.match("/title/%s/(\d+)\-[nvarsp](|\.nt|\.html|\.rdf|\.ttl|\.json)$" % WNRDF.wn_version, uri):
# synset_id = re.findall("/%s\-(\d+)\-[nvarsp]" % WNRDF.wn_version, uri)[0]
# graph = WNRDF.synset(self.wordnet_context, int(synset_id))
# if graph is None:
# return self.send404(start_response)
# title = ', '.join(sorted([str(o) for _, _, o in graph.triples((None, RDFS.label, None))]))
# start_response('200 OK', [('Content-type', 'text/plain')])
# return [title]
elif re.match(
"/%s/(.*)\-[nvarsp](|\.nt|\.html|\.rdf|\.ttl|\.json)$" %
WNRDF.wn_version, uri):
lemma_pos, = re.findall("^/%s/(.*)\-([nvarsp])" % WNRDF.wn_version,
uri)
lemma, pos = lemma_pos
graph = WNRDF.entry(self.wordnet_context, unquote_plus(lemma), pos)
if graph is None:
return self.send404(start_response)
can_form = graph.value(WNRDF.entry_name(unquote_plus(lemma), pos),
WNRDF.lemon.canonicalForm)
title = graph.value(can_form, WNRDF.lemon.writtenRep)
if mime == "html":
content = self.rdfxml_to_html(graph, str(title))
else:
try:
content = graph.serialize(
format=mime,
context=self.wordnet_context.jsonld_context,
base=WNRDF.entry_name(unquote_plus(lemma), pos))
except Exception as x:
return self.send501(start_response)
start_response('200 OK', [('Content-type', self.mime_types[mime]),
('Vary', 'Accept'),
('Content-length', str(len(content)))])
return [content]
elif re.match(
"/wn30/(\d{8}\-[nvarsp])(|\.nt|\.html|\.rdf|\.ttl|\.json)$",
uri):
(synset_id, ext), = re.findall(
"/wn30/(\d{8}\-[nvarsp])(|\.nt|\.html|\.rdf|\.ttl|\.json)$",
uri)
with sqlite3.connect(resolve('mapping/mapping.db')) as conn:
c = conn.cursor()
c.execute("select wn31 from wn30 where wn30=?", (synset_id, ))
row = c.fetchone()
if row:
wn31, = row
return self.send302(start_response,
"/wn31/%s-%s" % (wn31, synset_id[-1]))
else:
return self.send404(start_response)
elif re.match(
"/wn20/(\d{8}\-[nvarsp])(|\.nt|\.html|\.rdf|\.ttl|\.json)$",
uri):
(synset_id, ext), = re.findall(
"/wn20/(\d{8}\-[nvarsp])(|\.nt|\.html|\.rdf|\.ttl|\.json)$",
uri)
print(synset_id)
with sqlite3.connect(resolve('mapping/mapping.db')) as conn:
c = conn.cursor()
c.execute(
"select wn31 from wn30 join wn20 on wn30.wn30 = wn20.wn30 where wn20=?",
(synset_id, ))
row = c.fetchone()
if row:
wn31, = row
return self.send302(start_response,
"/wn31/%s-%s" % (wn31, synset_id[-1]))
else:
return self.send404(start_response)
elif uri == "/search" or uri == "/search/":
start_response('200 OK', [('Content-type', 'text/html')])
if 'QUERY_STRING' in environ:
qs_parsed = parse_qs(environ['QUERY_STRING'])
if 'query' in qs_parsed:
lemma = qs_parsed['query'][0]
result = self.search(self.wordnet_context, lemma)
return [result]
else:
return ["No query"]
else:
return ["No query string"]
elif re.match("/ontology(|.nt|.html|.ttl|.rdf)", uri):
start_response('200 OK', [
('Content-type', self.mime_types[mime]),
('Content-length',
str(
os.stat(
resolve("ontology" + self.mime_ext[mime])).st_size))
])
return [open(resolve("ontology" + self.mime_ext[mime])).read()]
elif uri == ("/%s.nt.gz" % WNRDF.wn_version):
start_response(
'200 OK',
[('Content-type', 'appliction/x-gzip'),
('Content-length', str(os.stat("wordnet.nt.gz").st_size))])
return open(resolve("wordnet.nt.gz"), "rb").read()
elif uri.startswith("/flag/") and exists(resolve(uri[1:])):
start_response(
'200 OK',
[('Content-type', 'image/gif'),
('Content-length', str(os.stat(resolve(uri[1:])).st_size))])
return open(resolve(uri[1:]), "rb").read()
elif uri == "/sparql/":
if 'QUERY_STRING' in environ:
qs = parse_qs(environ['QUERY_STRING'])
if 'query' in qs:
return self.sparql_query(
qs['query'][0], mime,
qs.get('default-graph-uri', [None])[0], start_response)
else:
start_response('200 OK', [('Content-type', 'text/html')])
return [
self.render_html("WordNet RDF",
open(resolve("sparql.html")).read())
]
else:
start_response('200 OK', [('Content-type', 'text/html')])
return [
self.render_html("WordNet RDF",
open(resolve("sparql.html")).read())
]
else:
return self.send404(start_response)
def application(self, environ, start_response):
uri = environ['PATH_INFO']
if re.match(".*\.html", uri):
mime = "html"
elif re.match(".*\.rdf", uri):
mime = "pretty-xml"
elif re.match(".*\.ttl", uri):
mime = "turtle"
elif re.match(".*\.nt", uri):
mime = "nt"
elif re.match(".*\.json", uri):
mime = "json-ld"
elif 'HTTP_ACCEPT' in environ:
if uri.startswith("/sparql"):
mime = self.best_mime_type(environ['HTTP_ACCEPT'], "sparql")
else:
mime = self.best_mime_type(environ['HTTP_ACCEPT'], "html")
else:
mime = "html"
if uri == "/" or uri == "/index.html":
start_response('200 OK', [('Content-type', 'text/html')])
return [self.render_html("WordNet RDF",open(resolve("index.html")).read())]
if uri == "/license.html":
start_response('200 OK', [('Content-type', 'text/html')])
return [self.render_html("WordNet RDF",open(resolve("license.html")).read())]
elif uri == "/wnrdf.css" or uri == "/sparql/wnrdf.css":
start_response('200 OK', [('Content-type', 'text/css')])
return [open(resolve("wnrdf.css")).read()]
elif re.match("/%s/(\d+)\-[nvarspNVARSP](|\.nt|\.html|\.rdf|\.ttl|\.json)$" % WNRDF.wn_version, uri):
synset_id = re.findall("/%s/(\d+)\-[nvarspNVARSP]" % WNRDF.wn_version, uri)[0]
if len(synset_id) == 8:
synset_id = str(WNRDF.pos2number(uri[-1])) + synset_id
return self.send302(start_response,"/%s/%s-%s" % (WNRDF.wn_version, synset_id, uri[-1]))
translate = True
if uri[-1].isupper():
translate = False
graph = WNRDF.synset(self.wordnet_context, int(synset_id), extras=mime == "html", translate=translate)
if graph is None:
return self.send404(start_response)
title = ', '.join(sorted([str(o) for _, _, o in graph.triples((None, RDFS.label, None))]))
if mime == "html":
content = self.rdfxml_to_html(graph, title)
else:
try:
content = graph.serialize(format=mime, context=self.wordnet_context.jsonld_context)
except Exception:
return self.send501(start_response)
start_response('200 OK', [('Content-type', self.mime_types[mime]),('Vary','Accept'), ('Content-length', str(len(content)))])
return [content]
# elif re.match("/title/%s/(\d+)\-[nvarsp](|\.nt|\.html|\.rdf|\.ttl|\.json)$" % WNRDF.wn_version, uri):
# synset_id = re.findall("/%s\-(\d+)\-[nvarsp]" % WNRDF.wn_version, uri)[0]
# graph = WNRDF.synset(self.wordnet_context, int(synset_id))
# if graph is None:
# return self.send404(start_response)
# title = ', '.join(sorted([str(o) for _, _, o in graph.triples((None, RDFS.label, None))]))
# start_response('200 OK', [('Content-type', 'text/plain')])
# return [title]
elif re.match("/%s/(.*)\-[nvarsp](|\.nt|\.html|\.rdf|\.ttl|\.json)$" % WNRDF.wn_version, uri):
lemma_pos, = re.findall("^/%s/(.*)\-([nvarsp])" % WNRDF.wn_version, uri)
lemma, pos = lemma_pos
graph = WNRDF.entry(self.wordnet_context, unquote_plus(lemma), pos)
if graph is None:
return self.send404(start_response)
can_form = graph.value(WNRDF.entry_name(unquote_plus(lemma), pos), WNRDF.lemon.canonicalForm)
title = graph.value(can_form, WNRDF.lemon.writtenRep)
if mime == "html":
content = self.rdfxml_to_html(graph, str(title))
else:
try:
content = graph.serialize(format=mime, context=self.wordnet_context.jsonld_context,
base=WNRDF.entry_name(unquote_plus(lemma), pos))
except Exception as x:
return self.send501(start_response)
start_response('200 OK', [('Content-type', self.mime_types[mime]),('Vary','Accept'), ('Content-length', str(len(content)))])
return [content]
elif re.match("/wn30/(\d{8}\-[nvarsp])(|\.nt|\.html|\.rdf|\.ttl|\.json)$", uri):
(synset_id, ext), = re.findall("/wn30/(\d{8}\-[nvarsp])(|\.nt|\.html|\.rdf|\.ttl|\.json)$", uri)
with sqlite3.connect(resolve('mapping/mapping.db')) as conn:
c = conn.cursor()
c.execute("select wn31 from wn30 where wn30=?", (synset_id,))
row = c.fetchone()
if row:
wn31, = row
return self.send302(start_response, "/wn31/%s-%s" % (wn31, synset_id[-1]))
elif synset_id.endswith("s"):
synset_id = synset_id[:-1] + "a"
c.execute("select wn31 from wn30 where wn30=?", (synset_id,))
row = c.fetchone()
if row:
wn31, = row
return self.send302(start_response, "/wn31/%s-%s" % (wn31, synset_id[-1]))
else:
return self.send404(start_response)
else:
return self.send404(start_response)
elif re.match("/wn20/(\d{8}\-[nvarsp])(|\.nt|\.html|\.rdf|\.ttl|\.json)$", uri):
(synset_id, ext), = re.findall("/wn20/(\d{8}\-[nvarsp])(|\.nt|\.html|\.rdf|\.ttl|\.json)$", uri)
print(synset_id)
with sqlite3.connect(resolve('mapping/mapping.db')) as conn:
c = conn.cursor()
c.execute("select wn31 from wn30 join wn20 on wn30.wn30 = wn20.wn30 where wn20=?", (synset_id,))
row = c.fetchone()
if row:
wn31, = row
return self.send302(start_response, "/wn31/%s-%s" % (wn31, synset_id[-1]))
else:
return self.send404(start_response)
elif uri == "/search" or uri == "/search/":
start_response('200 OK', [('Content-type', 'text/html')])
if 'QUERY_STRING' in environ:
qs_parsed = parse_qs(environ['QUERY_STRING'])
if 'query' in qs_parsed:
lemma = qs_parsed['query'][0]
result = self.search(self.wordnet_context, lemma)
return [result]
else:
return ["No query"]
else:
return ["No query string"]
elif re.match("/ontology(|.nt|.html|.ttl|.rdf)",uri):
start_response('200 OK', [('Content-type', self.mime_types[mime]),
('Content-length', str(os.stat(resolve("ontology" + self.mime_ext[mime])).st_size))])
return [open(resolve("ontology" + self.mime_ext[mime])).read()]
elif uri == ("/%s.nt.gz" % WNRDF.wn_version):
start_response('200 OK', [('Content-type', 'appliction/x-gzip'),
('Content-length', str(os.stat("wordnet.nt.gz").st_size))])
return open(resolve("wordnet.nt.gz"), "rb").read()
elif uri.startswith("/flag/") and exists(resolve(uri[1:])):
start_response('200 OK', [('Content-type', 'image/gif'),
('Content-length', str(os.stat(resolve(uri[1:])).st_size))])
return open(resolve(uri[1:]), "rb").read()
elif uri == "/sparql/":
if 'QUERY_STRING' in environ:
qs = parse_qs(environ['QUERY_STRING'])
if 'query' in qs:
return self.sparql_query(qs['query'][0], mime, qs.get('default-graph-uri',[None])[0], start_response)
else:
start_response('200 OK', [('Content-type', 'text/html')])
return [self.render_html("WordNet RDF",open(resolve("sparql.html")).read())]
else:
start_response('200 OK', [('Content-type', 'text/html')])
return [self.render_html("WordNet RDF",open(resolve("sparql.html")).read())]
else:
return self.send404(start_response)
