def get_targets(sent,patt_lines): pot_target_list = [] matched_patt_word = [] for patt in patt_lines: if (not patt): continue patt = re.sub(COLL_SPACES,SPACES_REPL,patt) m = re.findall(patt,sent) if m: # check forward or backward if(DEBUG): print "pattern matched ",m,"for patt ",patt,"and sent",sent # Now parse this line parsed_sent = parse_file(sent) if (not parsed_sent): print "could not parse line"+parsed_sent continue # NOW NP CHUNK THE SENTENCE # First make sense of parsed input pos_dict,parse_dict,_ = process_parse.pprocess_pline(parsed_sent) # the above might return multiple lines for i in xrange(len(pos_dict.keys())): pos_sent = pos_dict[i] parsed_sent = parse_dict[i] # NP chunking algo np_sent= process_parse.extract_np(pos_sent,parsed_sent) np_chunk_sent = process_parse.assemble_extracts(np_sent) # MATCHES BACK PATTERN # use the first word in the patt to split the parsed sentence patt = patt.strip() if(not patt): print "patt was empty line move to next" continue split_patt = patt.split() split_word = split_patt[0] split_word = split_word.strip() if split_word in matched_patt_word: if(DEBUG): print "###not matching back pattern since back pattern with same key word was matched ,back key word =",split_word continue m_temp = re.search(split_word,np_chunk_sent) if(not m_temp): print "split word=",split_word,"not in sent" continue pot_target_list = get_np(split_word,np_chunk_sent,BACK,'target') # search for AND IN THE np if it exists divide the np into two parts new_list = and_detector(pot_target_list) return new_list
def get_perpi(sent,patt_lines):
pot_perpi_list = []
matched_patt_word = []
for patt in patt_lines:
if (not patt):
continue
#m2 = re.search('MURDERED',patt)
#if m2:
# print "patt",patt
# collapse multiple white spaces
patt = re.sub(COLL_SPACES,SPACES_REPL,patt)
# check if any of the victim patterns exist for this line
m = re.findall(patt,sent)
if m:
# check forward or backward
if(DEBUG):
print "pattern matched ",m,"for patt ",patt,"and sent",sent
# Now parse this line
parsed_sent = parse_file(sent)
if (not parsed_sent):
print "could not parse line"+parsed_sent
continue
# NOW NP CHUNK THE SENTENCE
# First make sense of parsed input
pos_dict,parse_dict,_ = process_parse.pprocess_pline(parsed_sent)
# the above might return multiple lines
for i in xrange(len(pos_dict.keys())):
pos_sent = pos_dict[i]
parsed_sent = parse_dict[i]
# NP chunking algo
np_sent= process_parse.extract_np(pos_sent,parsed_sent)
np_chunk_sent = process_parse.assemble_extracts(np_sent)
if(is_front(patt)):
#perpi just have one word ( as of now) so just split by word
patt = patt.strip()
if(not patt):
print "patt was empty line move to next"
continue
split_patt = patt.split()
split_word = split_patt[0]
split_word = split_word.strip()
# THIS MAKES SURE THAT a FONT PATT IS NOT MATCHED AGAIN BY BACK PATT
#matched_patt_word.append(split_word)
m_temp = re.search(split_word,np_chunk_sent)
if(not m_temp):
print "split word=",split_word,"not in sent"
continue
pot_perpi_list = get_np(split_word,np_chunk_sent,FRONT,'perpi')
if(len(pot_perpi_list) > 0):
# THIS MAKES SURE THAT a FONT PATT IS NOT MATCHED AGAIN BY BACK PATT
matched_patt_word.append(split_word)
else:
# MATCHES BACK PATTERN
# Back patterns have three words ..second last word is the main word / split word
patt = patt.strip()
if(not patt):
print "patt was empty line move to next"
continue
split_patt = patt.split()
# second last word or second word is the main word
split_word = split_patt[1]
split_word = split_word.strip()
if split_word in matched_patt_word:
print "###not matching back pattern since back pattern with same key word was matched ,back key word =",split_word
continue
m_temp = re.search(split_word,np_chunk_sent)
if(not m_temp):
print "split word=",split_word,"not in sent"
continue
pot_perpi_list = get_np(split_word,np_chunk_sent,BACK,'perpi')
# search for AND IN THE np if it exists divide the np into two parts
new_list = and_detector(pot_perpi_list)
return new_list
