get_resource.py

import re
# import CMUTweetTagger
#import cPickle
from collections import defaultdict
import pickle
from nltk.corpus import wordnet as wn
from itertools import product
import spacy
from spacy.symbols import *
from nltk import Tree
from nltk import *
import nltk
import location
import time
import sys
ps_stemmer=stem.porter.PorterStemmer()



nlp=spacy.load('en')
np_labels=set(['nsubj','dobj','pobj','iobj','conj','nsubjpass','appos','nmod','poss','parataxis','advmod','advcl'])
subj_labels=set(['nsubj','nsubjpass','csubj','csubjpass'])
modifiers=['nummod','compound','amod','punct']
after_clause_modifier=['relcl','acl','ccomp','xcomp','acomp','punct','advcl','rcmod']
tel_no="([+]?[0]?[1-9][0-9\s]*[-]?[0-9\s]+)"
email="([a-zA-Z0-9]?[a-zA-Z0-9_.]+[@][a-zA-Z]+[.](com|net|edu|in|org|en))"
web_url="http:[a-zA-Z._0-9/]+[a-zA-Z0-9]"
http_url='http[s]?://(?:[a-zA-Z]|[0-9]|[$-_@.&+]|[!*\(\)]|(?:%[0-9a-fA-F][0-9a-fA-F]))+'
entity_type_list=['NORP','ORG','GPE','PERSON']
quant_no="([0-9]*[,.]?[0-9]+[km]?)"
alphanum="[^0-9a-zA-Z ]"

stop_list=list(location.false_names)
#,'nn','quantmod','nmod','hmod','infmod']

need_file=open('DATA/Process_resources/need.txt')
offer_file=open('DATA/Process_resources/offer.txt')
shelter_file=open('DATA/Process_resources/shelter.txt')
food_file=open('DATA/Process_resources/food.txt')
medical_file=open('DATA/Process_resources/medical.txt')
cash_file=open('DATA/Process_resources/cash.txt')
logistics_file=open('DATA/Process_resources/logistics.txt')
disaster_events_file=open('DATA/Process_resources/disaster_events.txt')


basic_resource=['medical','water','sanitation','shelter','cloth','food','transport','infrastructure','volunteers','logistic']

need_verb_list=set()
for line in need_file:
	line=line.rstrip().lower()
	need_verb_list.add(line)
send_verb_list=set()	
for line in offer_file:
	line=line.rstrip().lower()
	send_verb_list.add(line)

need_send_verb_list=list(need_verb_list)
need_send_verb_list.extend(list(send_verb_list))	
common_resource=set()
dis_events=set()

for line in shelter_file:
	line=line.rstrip().lower()
	common_resource.add(line)
for line in cash_file:
	line=line.rstrip().lower()
	common_resource.add(line)
for line in food_file:
	line=line.rstrip().lower()
	common_resource.add(line)
for line in medical_file:
	line=line.rstrip().lower()
	common_resource.add(line)
for line in logistics_file:
	line=line.rstrip().lower()
	common_resource.add(line)		
for line in disaster_events_file:
	line=line.rstrip().lower()
	dis_events.add(line)

dis_events_stem=[ps_stemmer.stem(i) for i in list(dis_events)]

common_resource=list(common_resource)
common_resource.extend([ps_stemmer.stem(i) for i in common_resource])

# try:
# 	input_name=sys.argv[1]
# except:
# 	input_name='nepal_needs'	

# print(input_name)
# input_file='DATA/INPUT/'+input_name+'.txt'

# print(tweet_preprocess2(text,[]))

stop_list.extend(need_send_verb_list)
print("Loading done")

def get_contact(text):
	contacts=[]
	flag=0
	numbers=re.findall(tel_no,text)
	temp=set()
	for i in numbers:
		if len(i.replace(' ',''))>=7:
			temp.add(i)
			# print("Contact information:" +i)
	contacts.append(temp)		
	temp=set()
	mails= re.findall(email,text)
	for i in mails:
		temp.add(i)
		# print("Mail: "+i[0])	
	
	contacts.append(temp)	
	temp=set()	
	urls= re.findall(http_url,text)	
	for i in urls:
		temp.add(i)
		# print("URL: "+i)	
		
	contacts.append(temp)
	return contacts	

# f=open(input_file,'r')

# print(contacts)

def modifier_word(word):
	modified_word=word.orth_
	while word.n_lefts+word.n_rights==1 and word.dep_.lower() in modifiers:
		word=[child for child in word.children][0]
		modified_word=word.orth_+" "+modified_word
	return modified_word	

def tok_format(tok):
	return "_".join([tok.orth_, tok.dep_,tok.ent_type_])

def to_nltk_tree(node):
	if node.n_lefts + node.n_rights > 0:
		return Tree(tok_format(node), [to_nltk_tree(child) for child in node.children])
	else:
		return tok_format(node)		

def get_verb_similarity_score(word,given_list,given_list_2):
	max_verb_similarity=0
	if word.lower() in given_list:
		max_verb_similarity=1
	else:	
		current_verb_list=wn.synsets(word.lower())
		for verb in given_list_2:
			related_verbs=wn.synsets(verb)
			for a,b in product(related_verbs,current_verb_list):
				d=wn.wup_similarity(a,b)
				try:
					if d> max_verb_similarity:
						max_verb_similarity=d
				except:
					continue
	return max_verb_similarity			


def resource_in_list(resource):
	
	related_resources=wn.synsets(resource)
	max_similarity=0
	chosen_word=""
	if ps_stemmer.stem(resource.lower()) in common_resource:
		return 1,resource

	for word in basic_resource:
		related_words=wn.synsets(word)
		for a,b in product(related_words,related_resources):
			d=wn.wup_similarity(a,b)
			try:
				if d> max_similarity:
					max_similarity=d
					chosen_word=word
			except:
				continue
	return max_similarity, chosen_word

def get_children(word,resource_array,modified_array):
	#print(word,word.dep_)

	for child in word.children:
		if child.dep_.lower() in modifiers:
			get_word=modifier_word(child)+" "+word.orth_+"<_>"+word.dep_
			modified_array.append(get_word)
		if child.dep_.lower()=='prep' or child.dep_.lower()=='punct':
			get_children(child,resource_array,modified_array)	
		if child.dep_.lower() in after_clause_modifier:	
			#print(child, child.dep_)
			get_children(child,resource_array,modified_array)	
		if child.dep_.lower() in np_labels:			
			get_children(child,resource_array,modified_array)
			resource_array.append(child.orth_+"<_>"+child.dep_)		
		else:
			if get_verb_similarity_score(child.orth_,common_resource,basic_resource)>0.9:
				get_children(child,resource_array,modified_array)


def get_resource(text):
	doc=nlp(text)
	# try:
	# 	[to_nltk_tree(sent.root).pretty_print() for sent in doc.sents]
	# except:
	# 	print("Exception here")

	# print(time.time()-start_time,1)
	org_list=[]
	prev_word=""
	prev_word_type=""
	for word in doc:
		if word.ent_type_ in entity_type_list:
			org_list.append(word.orth_+"<_>"+word.ent_type_)
		else:
			org_list.append("<_>")

	resource_array=[]
	modified_array=[]
	for word in doc:
		if  get_verb_similarity_score(word.orth_,need_send_verb_list,need_send_verb_list)>0.9 or word.dep_=='ROOT':
			get_children(word,resource_array,modified_array)

		if word.dep_=='cc' and word.n_lefts+word.n_rights==0:
			ancestor=word.head.orth_
			#print(ancestor)
			if get_verb_similarity_score(ancestor,common_resource,basic_resource)>0.9:
				get_children(word.head,resource_array,modified_array)		
				
	last_word=[]
	final_resource={}
	modified_array_2=[]
	resource_array_2=[]
	n_subj_list=[]

	# print(time.time()-start_time,2)

	# print("Modified array", modified_array)
	# print("Resource array", resource_array)
	
	for i in modified_array:
		modified_array_2.append(i[:(i.index("<_>"))])

	for i in resource_array:
		resource_array_2.append(i[:(i.index("<_>"))])

	modified_array_2=[re.sub(alphanum,"",i.strip()) for i in modified_array_2]
	modified_array_2=list(set([i.strip() for i in modified_array_2]))
	resource_array_2=[re.sub(alphanum,"",i.strip()) for i in resource_array_2]
	resource_array_2=list(set([i.strip() for i in resource_array_2]))

	# print("Resource array: ",resource_array_2)	
	# print("Modified array: ", modified_array_2)

	for resources in modified_array_2:
		max_val_resource=-1
		val_type=""
		resource_list=resources.strip().split(" ")
		for resource in resource_list:
			pres_res_val,pres_res_type=resource_in_list(resource)

			# print(resource,pres_res_val,pres_res_type)

			if max_val_resource==-1:
				max_val_resource=pres_res_val

			if pres_res_val> max_val_resource:
				val_type=pres_res_type
				max_val_resource=pres_res_val			

			if pres_res_val> 0.8:
				final_resource[resource]=pres_res_type	
				
		if max_val_resource > 0.9:
			final_resource[resources]=val_type

	# print(time.time()-start_time,3)
			
	for resource in resource_array_2:
		#print(resource)
		pres_res_val,pres_res_type=resource_in_list(resource)
		
		if pres_res_val> 0.8:
			if resource not in final_resource:
				final_resource[resource]=pres_res_type

	final_resource_keys=list(final_resource.keys())
	
		
	prev_word_type=""
	prev_word=""
	org_list_2=[]
	
	for i in org_list:
		index=i.index("<_>")
		if i[index+3:]=="ORG" and prev_word_type=="ORG":
			prev_word=prev_word+" "+i[:index]
		elif i[index+3:]=="PERSON" and prev_word_type=="PERSON":	
			prev_word=prev_word+" "+i[:index]
		else:
			if prev_word !='':
				org_list_2.append(prev_word+"<_>"+prev_word_type)
			prev_word_type=i[index+3:]
			prev_word=i[:index]


	source_list=[]
	org_person_list=[]
	
	for i in org_list_2:
		tag=i[i.index("<_>")+3:]
		j=i[:i.index("<_>")]

		if tag=="ORG" or tag=="PERSON" or tag=='GPE' or tag=='LOC':
			if j.lower() not in stop_list:
				org_person_list.append(j)
		elif j.lower() not in stop_list :
			source_list.append(j)
		else:
			continue	
	
	for i in modified_array:
		pos_res=i[:i.index("<_>")]
		pos_tag=i[i.index("<_>")+3:]
		if pos_tag in subj_labels:
			if pos_res not in source_list and pos_res not in final_resource_keys and pos_res.lower() not in  stop_list:
				#print(pos_tag,pos_res)
				source_list.append(pos_res)

	for i in resource_array:
		pos_res=i[:i.index("<_>")]
		pos_tag=i[i.index("<_>")+3:]
		if pos_tag in subj_labels:
			if pos_res not in source_list and pos_res not in final_resource_keys and pos_res.lower() not in  stop_list:
				#print(pos_tag,pos_res)
				source_list.append(pos_res)				

	pos_tags_dict={}	
	doc2=nlp(text.lower())		

	for word in doc2:
		try:
			pos_tags_dict[word.orth_]=word.pos_
		except:
			continue	

	final_resource_keys_2=[]

	for elem in final_resource_keys:
		elem2=elem.split()
		poss=[]

		for i in elem2:
			try:
				poss.append(pos_tags_dict[i.lower()])
			except Exception as e:
				continue	
		# poss=[pos_tags_dict[i.lower()] for i in elem2]

		if poss==[]:
			continue
			
		if 'VERB' not in poss and( poss[-1]=='NOUN'):
			final_resource_keys_2.append(elem)	


	return final_resource_keys_2,source_list,org_person_list,modified_array	

def jumble(text,items):

	final_items=[]

	for item in items:
		if item in text:
			final_items.append(item)

	temp_list=[]
	for item1 in final_items:
		for item2 in final_items:
			if item1+' '+item2 in text:
				temp_list.append(item1+' '+item2)

	final_items.extend(temp_list)			

	items=list(set(items)-set(final_items))
	
	while True:
		add_list=[]
		rem_list=[]
		item_list=[]

		for item in items:		
			item_split=item.split()
			for elem in final_items:
				for k in item_split:
					if k+' '+elem in text:
						add_list.append(k+' '+elem)
						rem_list.append(elem)
						item_list.append(item)
					if elem+' '+k in text:	
						add_list.append(elem+' '+k)
						rem_list.append(elem)
						item_list.append(item)						

		if add_list==[]:
			break
		else:
			final_items.extend(add_list)
			items= list(set(items)-set(item_list))

	return final_items		

def post_process(text,final_resource_keys,source_list,loc_list):

	source_dis=set()
	resource_dis=set()

	for loc in loc_list:
		for elem in source_list:
			elem2=elem
			elem=elem.lower()
			if loc in elem or elem in loc or elem in stop_list:
				source_dis.add(elem2)
				continue

		for elem in final_resource_keys:
			elem2=elem
			elem=elem.lower()
			if loc in elem or elem in loc or elem in stop_list:
				resource_dis.add(elem2)
				continue


	source_list=list(set(source_list)- source_dis)
	final_resource_keys=list(set(final_resource_keys)- resource_dis)

	source_list_2=[]
	source_dis=set()
	
	for elem in source_list:	
		elem_split=[ps_stemmer.stem(i) for i in elem.lower().split()]
		flag=False
		for i in elem_split:
			if i in dis_events_stem:
				flag=True
				break
		if flag==True:
			source_dis.add(elem)
			continue

		for elem2 in source_list:
			if elem2 ==elem:
				continue

			if elem2 in source_dis:
				continue

			if elem2 in elem :
				source_dis.add(elem2)
			if elem in elem2:
				source_dis.add(elem)	

	source_list=list(set(source_list)- source_dis)	
	source_list=jumble(text,source_list)

	dup_final_resource_keys=list(final_resource_keys)

	final_resource_keys=jumble(text,final_resource_keys)


	source_dis=set()
	resource_dis=set()

	for elem in source_list:	
		for elem2 in source_list:
			if elem2 ==elem:
				continue
			if elem2 in source_dis:
				continue
			if elem2 in elem :
				source_dis.add(elem2)
			if elem in elem2:
				source_dis.add(elem)

		for elem3 in final_resource_keys:
			if elem in elem3 or elem3 in elem:
				source_dis.add(elem)		



	for elem in final_resource_keys:
		for elem2 in final_resource_keys:
			if elem2 ==elem:
				continue
			if elem2 in resource_dis:
				continue
			if elem2 in elem :
				resource_dis.add(elem2)
			if elem in elem2:
				resource_dis.add(elem)			

	source_list=list(set(source_list)- source_dis)			
	final_resource_keys=list(set(final_resource_keys)- resource_dis)

	return source_list,final_resource_keys,loc_list	,dup_final_resource_keys
	# print(source_list)
	# print(final_resource_keys)
	# print(loc_list)
				

def create_resource_list(text):
	count=0
	
	quantity_dict={}
	final_resource_keys=[]
	source_list=[]
	loc_list=[]
	org_person_list=[]

	loc_list_2=location.return_location_list(text)
	
	final_resource_keys,source_list,org_person_list,modified_array= get_resource(text)	

	doc=nlp(text)		
	

	for elem in source_list:
		if elem.lower() in location.curr_loc_dict and elem.lower() not in stop_list:
			loc_list_2.append((elem.lower(),location.curr_loc_dict[elem.lower()]))

	for elem in org_person_list:
		if elem.lower() in location.curr_loc_dict and elem.lower() not in stop_list:
			loc_list_2.append((elem.lower(),location.curr_loc_dict[elem.lower()]))

	loc_list=list(set([i[0] for i in loc_list_2]))

	source_list= [i for i in source_list if i.lower() not in loc_list] 
	org_person_list=[i for i in org_person_list if i.lower() not in loc_list] 

	source_list=list(set(source_list) | set(org_person_list))

	final_resource_keys=[i for i in final_resource_keys if i.lower() not in loc_list]

		
	# print(text)
	a,b,c,d=post_process(text, final_resource_keys,source_list,loc_list)
	# print(a)
	# print(b)
	# print(c)
	
	return a,b,loc_list_2,modified_array,d


global_resource_list={}

def show_resource(text):
	contacts=get_contact(text)
	text=location.tweet_preprocess2(text,[])
	a,b,c,modified_array,d=create_resource_list(text)

	'''
	a= source list
	b= final resource keys
	c= loc_list
	d= dup_final_resource keys
	'''

	print("Phone number ", contacts[0])
	print("Email ", contacts[1])
	print("List of sources ", a)
	print("Resource list ", b)
	print("List of possible locations ", set(c))
	# print(d)


sample_text='We need food in Nepal'
show_resource(sample_text)

# for line in f:
# 	start_time=time.time()
# 	line=line.rstrip().split('<||>')		
# 	tid=line[0]
# 	text=line[1]
# 	contacts=get_contact(text)
# 	text=location.tweet_preprocess2(text,[])
# 	a,b,c,modified_array,d=create_resource_list(text)
# 	global_resource_list[tid]=((text,a,b,c,contacts,modified_array,d))


# with open('DATA/OUTPUT/'+input_name+'.p','wb')as handle:
# 	pickle.dump(global_resource_list,handle)