def initTries(trie_dict): for k,v in trie_dict.iteritems(): v_arr = [] for value in v: v_arr.append(value) v_arr = pyannotate.toLowerCase(v_arr) tries_all[k] = marisa_trie.Trie(v_arr)
def initTries(trie_dict):
for k,v in trie_dict.iteritems():
v_arr = []
for value in v:
v_arr.append(unicode(value,errors="ignore"))
v_arr = pyannotate.toLowerCase(v_arr)
tries_all[unicode(k,errors="ignore")] = marisa_trie.Trie(v_arr)
tries_all["PERSON"] = marisa_trie.Trie().load("person.marisa")
def initTries(trie_dict): for k,v in trie_dict.iteritems(): v_arr = [] for value in v: v_arr.append(unicode(value,errors="ignore")) v_arr = pyannotate.toLowerCase(v_arr) tries_all[unicode(k,errors="ignore")] = marisa_trie.Trie(v_arr) for name in tries_all: try: tries_all[name] = marisa_trie.Trie().load(name+".marisa") except: pass
def resolveRedirection(sent,title):
redirect_labels = ['he',"she","it","we","they"]
redirect = pyannotate.toLowerCase(redirect_labels)
arr_arr = sent.split(" ")
arr = arr_arr[0].lower()
if arr in redirect:
return " ".join(title.split("_")) + " " + " ".join(arr_arr[1:])
if title.find("_") > -1:
if arr in title.split("_"):
return title + " " + " ".join(arr_arr[1:])
else:
if arr in title.split(" "):
return title + " " + " ".join(arr_arr[1:])
return " ".join(arr_arr)
#json.dumps(extractions,indent = 4,separators=(',', ': '))
if __name__ == '__main__':
sent = "Charles Dickens wrote books like A Christmas Carol, Anthony and Mayan, A Chritmas Carol"
sent1 = "hello from book A Christmas Carol, Anthony to kill me"
persons = ["Bill Gates"]
authors = ["Charles Dickens"]
books = ["A Christmas Carol",""]
location = ["Seattle"]
company = ["Microsoft","Apple Inc","Apple","Mcafee","Amazon.com","Intel","Google","Nvidia","AMD","Oracle","Sun Microsystems"]
trie = {"PERSON":persons, "BOOK":books, "AUTHOR":authors, "COMPANY":company, "LOCATION":location}
#test annotation
ptrie = marisa_trie.Trie(pyannotate.toLowerCase(persons))
btrie = marisa_trie.Trie(pyannotate.toLowerCase(books))
#pyannotate.annotate(sent,ptrie,1)
#test pattern indices
start,end = getIndices(sent,"[C|c]arol")
#print start, end
#read JSON
exp = readJson()
#main
initTries(trie)
"""
sent = "Bill Gates was born in August 14, 1897 - December 14, 1945. He founded Microsoft"