class TraphServerFactory(Factory):
default_WECR = '(s:[a-zA-Z]+\\|(t:[0-9]+\\|)?(h:[^\\|]+\\|(h:[^\\|]+\\|)|h:(localhost|(\\d{1,3}\\.){3}\\d{1,3}|\\[[\\da-f]*:[\\da-f:]*\\])\\|))'
WECRs = {
's:http|h:com|h:world|': '(s:[a-zA-Z]+\\|(t:[0-9]+\\|)?(h:[^\\|]+\\|(h:[^\\|]+\\|)+|h:(localhost|(\\d{1,3}\\.){3}\\d{1,3}|\\[[\\da-f]*:[\\da-f:]*\\])\\|)(p:[^\\|]+\\|){1})'
}
def __init__(self, corpus, traph_dir="traph-data", default_WECR=None, WECRs=None):
self.traph_dir = traph_dir
self.corpus = corpus
if not os.path.isdir(self.traph_dir):
os.makedirs(self.traph_dir)
self.traph = Traph(
folder=os.path.join(self.traph_dir, corpus),
default_webentity_creation_rule=default_WECR or self.default_WECR,
webentity_creation_rules=WECRs or self.WECRs
)
def ready(self):
# stdin message received by childprocess to know when traph is ready
print "READY"
def buildProtocol(self, addr):
return TraphProtocol(self.traph)
def close(self):
self.traph.close()
class TraphServerFactory(Factory):
default_WECR = '(s:[a-zA-Z]+\\|(t:[0-9]+\\|)?(h:[^\\|]+\\|(h:[^\\|]+\\|)|h:(localhost|(\\d{1,3}\\.){3}\\d{1,3}|\\[[\\da-f]*:[\\da-f:]*\\])\\|))'
WECRs = {
's:http|h:com|h:world|':
'(s:[a-zA-Z]+\\|(t:[0-9]+\\|)?(h:[^\\|]+\\|(h:[^\\|]+\\|)+|h:(localhost|(\\d{1,3}\\.){3}\\d{1,3}|\\[[\\da-f]*:[\\da-f:]*\\])\\|)(p:[^\\|]+\\|){1})'
}
def __init__(self,
corpus,
traph_dir="traph-data",
default_WECR=None,
WECRs=None):
self.traph_dir = traph_dir
self.corpus = corpus
if not os.path.isdir(self.traph_dir):
os.makedirs(self.traph_dir)
self.traph = Traph(folder=os.path.join(self.traph_dir, corpus),
default_webentity_creation_rule=default_WECR
or self.default_WECR,
webentity_creation_rules=WECRs or self.WECRs)
def ready(self):
# stdin message received by childprocess to know when traph is ready
print "READY"
def buildProtocol(self, addr):
return TraphProtocol(self.traph)
def close(self):
self.traph.close()
include_internal=False,
include_outbound=False)
for source_lru, lru, weight in valjean_inlinks:
print '\t<- (weight %s) \t%s' % (weight, source_lru)
print ''
valjean_outlinks = traph.get_page_links('s:http|h:com|h:valjean|',
include_inbound=False,
include_internal=False,
include_outbound=True)
for lru, target_lru, weight in valjean_outlinks:
print '\t-> (weight %s) \t%s' % (weight, target_lru)
# import networkx as nx
# g = nx.Graph()
# w = traph.get_webentities_links()
# for source, targets in w.items():
# source_label = webentity_store.data['webentities'][source][1]
# g.add_node(source, label=source_label)
# for target in targets:
# target_label = webentity_store.data['webentities'][target][1]
# g.add_node(target, label=target_label)
# g.add_edge(source, target)
# nx.write_gexf(g, './scripts/data/dump.gexf')
traph.close()