def do_stem(self, word):
fsm = Fysom(initial='start', events=self.events)
i = len(word) - 1
j = len(word)
while (True):
if i <= 0:
break
v = word[i:j]
# print v
res = fsm.can(v)
if (res):
if v == 'i' and fsm.can(word[i - 1:j]):
i = i - 1
continue
fsm.trigger(v)
if fsm.current == 'h':
if word[i - 1:i] == 'i':
i = i - 1 # skip i
if word[i - 1:i] == 'n':
# ning qushimchasi
fsm.current = 'start'
continue
elif fsm.current == 'b':
fsm.current = 'start'
j = i
i = i - 1
return word[:j]
def do_stem(self, word):
fsm = Fysom(initial='start', events=self.events)
# FIXME: uncomment below and make sanitize functions support
# both Python 2 and 3 versions
# word = WordProcessor.sanitize(word)
i = len(word) - 1
j = len(word)
while(True):
if i <= 0:
break
v = word[i:j]
# print v
res = fsm.can(v)
if (res):
if v == 'i' and fsm.can(word[i-1:j]):
i = i - 1
continue
fsm.trigger(v)
if fsm.current == 'h':
if word[i-1:i] == 'i':
i = i - 1 # skip i
if word[i-1:i] == 'n':
# ning qushimchasi
fsm.current = 'start'
continue
elif fsm.current == 'b':
fsm.current = 'start'
j = i
i = i - 1
return word[:j]
def do_stem(self, word):
fsm = Fysom(initial='start', events=self.events)
# FIXME: uncomment below and make sanitize functions support
# both Python 2 and 3 versions
# word = WordProcessor.sanitize(word)
i = len(word) - 1
j = len(word)
while (True):
if i <= 0:
break
v = word[i:j]
# print v
res = fsm.can(v)
if (res):
if v == 'i' and fsm.can(word[i - 1:j]):
i = i - 1
continue
fsm.trigger(v)
if fsm.current == 'h':
if word[i - 1:i] == 'i':
i = i - 1 # skip i
if word[i - 1:i] == 'n':
# ning qushimchasi
fsm.current = 'start'
continue
elif fsm.current == 'b':
fsm.current = 'start'
j = i
i = i - 1
return word[:j]
def SenToPhrase (tagged_sentence):
fsm = Fysom({'initial': '0',
'events': [
{'name': 'IN', 'src': '0', 'dst': '1'},{'name': 'NN', 'src': '1', 'dst': '3'},{'name': 'NNS', 'src': '1', 'dst': '3'},
{'name': 'NNP', 'src': '1', 'dst': '3'},{'name': 'NNPS', 'src': '1', 'dst': '3'},{'name': 'DT', 'src': '1', 'dst': '2'},
{'name': 'NN', 'src': '2', 'dst': '3'},{'name': 'NNS', 'src': '2', 'dst': '3'},{'name': 'NNP', 'src': '2', 'dst': '3'},
{'name': 'NNPS', 'src': '2', 'dst': '3'},{'name': 'PRP$', 'src': '1', 'dst': '4'},{'name': 'PRP$', 'src': '2', 'dst': '4'},
{'name': 'JJ', 'src': '1', 'dst': '5'},{'name': 'JJ', 'src': '2', 'dst': '5'},{'name': 'JJR', 'src': '1', 'dst': '6'},
{'name': 'JJR', 'src': '2', 'dst': '6'},{'name': 'JJS', 'src': '1', 'dst': '7'},{'name': 'JJS', 'src': '2', 'dst': '7'},
{'name': 'NN', 'src': '5', 'dst': '3'},{'name': 'NN', 'src': '6', 'dst': '3'},{'name': 'NN', 'src': '7', 'dst': '3'},
{'name': 'NNS', 'src': '5', 'dst': '3'},{'name': 'NNS', 'src': '6', 'dst': '3'},{'name': 'NNS', 'src': '7', 'dst': '3'},
{'name': 'NNP', 'src': '5', 'dst': '3'},{'name': 'NNP', 'src': '6', 'dst': '3'},{'name': 'NNP', 'src': '7', 'dst': '3'},
{'name': 'NNPS', 'src': '5', 'dst': '3'},{'name': 'NNPS', 'src': '6', 'dst': '3'},{'name': 'NNPS', 'src': '7', 'dst': '3'},
{'name': 'PRP', 'src': '1', 'dst': '4'},{'name': 'PRP', 'src': '2', 'dst': '4'},{'name': 'NN', 'src': '4', 'dst': '3'},
{'name': 'NNS', 'src': '4', 'dst': '3'},{'name': 'NNP', 'src': '4', 'dst': '3'},{'name': 'NNPS', 'src': '4', 'dst': '3'},
{'name': 'TO', 'src': '0', 'dst': '1'},{'name': 'NN', 'src': '3', 'dst': '4'},{'name': 'NNS', 'src': '3', 'dst': '4'},
{'name': 'NNP', 'src': '3', 'dst': '4'},{'name': 'NNPS', 'src': '3', 'dst': '4'},
#######VERB################
{'name': 'MD', 'src': '0', 'dst': '8'},{'name': 'VB', 'src': '8', 'dst': '9'},{'name': 'VBN', 'src': '9', 'dst': '21'},
{'name': 'VBG', 'src': '9', 'dst': '10'},{'name': 'JJ', 'src': '9', 'dst': '10'},{'name': 'RB', 'src': '9', 'dst': '11'},
{'name': 'VBD', 'src': '0', 'dst': '12'},{'name': 'VBG', 'src': '12', 'dst': '13'},{'name': 'RB', 'src': '13', 'dst': '11'},
{'name': 'RB', 'src': '12', 'dst': '11'},{'name': 'VBN', 'src': '12', 'dst': '14'},{'name': 'VBG', 'src': '14', 'dst': '15'},
{'name': 'JJ', 'src': '14', 'dst': '16'},{'name': 'VBZ', 'src': '0', 'dst': '17'},{'name': 'VBP', 'src': '0', 'dst': '17'},
{'name': 'RB', 'src': '17', 'dst': '11'},{'name': 'VBG', 'src': '17', 'dst': '18'},{'name': 'VBN', 'src': '17', 'dst': '19'},
{'name': 'RB', 'src': '18', 'dst': '11'},{'name': 'VBG', 'src': '19', 'dst': '20'},{'name': 'RB', 'src': '20', 'dst': '11'},
{'name': 'VBG', 'src': '21', 'dst': '22'},{'name': 'RB', 'src': '14', 'dst': '11'}
]})
high_final_states = ['3','4','9','10','13','14','15','16','18','19','20','21','22']
phras_rules={'R1':[],'R2':[],'R3':[],'R4':[],'R5':[],'R6':[],'R7':[],'R8':[],'R9':[],'R10':[],'R11':[],'R12':[]}
to_rb = ['9','12','13','14','17','18','20']
fsm.current = '0'
# new_temp = ""
t=[]
k = 0
phrase_count=0
while(k<len(tagged_sentence)):
flag = 0
rbflag = 0
fsm.current = '0'
temp_current='0'
count = 0
# new_temp = ""
t=[]
j = k
s=[]
for j in range(k,len(tagged_sentence)):
# print("-----For loop j----")
# print(tagged_sentence[j])
try:
fsm.trigger(tagged_sentence[j][1])
# print("\n",fsm.current)
temp_current=fsm.current
# new_temp += tagged_sentence[j][0] + " "
t.append(tagged_sentence[j])
count += 1
except:
break
finally:
if(fsm.current=='3' and j!=len(tagged_sentence)-1 and tagged_sentence[j][0][-1]!=","):
try:
fsm.trigger(tagged_sentence[j+1][1])
# print(tagged_sentence[j+1])
# print("\n",fsm.current)
# new_temp += tagged_sentence[j+1][0] + " "
t.append(tagged_sentence[j+1])
count += 1
except:
oops = 2
if(fsm.current=='9' and j!=len(tagged_sentence)-1):
try:
fsm.trigger(tagged_sentence[j+1][1])
# print("\n",fsm.current)
# print(tagged_sentence[j+1])
if(fsm.current=='11'):
fsm.current = '9'
else:
# new_temp += i[j+1][0] + " "
t.append(tagged_sentence[j+1])
count += 1
except:
oops = 3
if(fsm.current=='14' and j!=len(tagged_sentence)-1):
try:
fsm.trigger(tagged_sentence[j+1][1])
# print("\n",fsm.current)
# print(tagged_sentence[j+1])
if(fsm.current=='11'):
fsm.current = '14'
else:
# new_temp += i[j+1][0] + " "
t.append(tagged_sentence[j+1])
count += 1
except:
oops = 3
if(fsm.current=='19' and j!=len(tagged_sentence)-1):
try:
fsm.trigger(tagged_sentence[j+1][1])
# print("\n",fsm.current)
# print(tagged_sentence[j+1])
# new_temp += i[j+1][0] + " "
temp_current=fsm.current
t.append(tagged_sentence[j+1])
count += 1
except:
oops = 1
if(fsm.current=='21' and j!=len(tagged_sentence)-1):
try:
fsm.trigger(tagged_sentence[j+1][1])
# print("\n",fsm.current)
# print(tagged_sentence[j+1])
# new_temp += i[j+1][0] + " "
t.append(tagged_sentence[j+1])
count += 1
except:
oops = 1
c=int(fsm.current)
if(fsm.current in high_final_states):
phrase_count+=1
# new_temp = new_temp[:-1]
# if(new_temp[-1]==','):
# new_temp = new_temp[:-1]
if(c==3 or c==4):
phras_rules['R1'].append(t)
#PREPOSITION
elif(c==18):
phras_rules['R2'].append(t)
#PRESENT CONTINUOUS
elif(c==19):
phras_rules['R3'].append(t)
#PRESENT PERFECT
elif(c==20):
phras_rules['R4'].append(t)
#PRESENT PERFECT CONTINUOUS
elif(c==13):
phras_rules['R5'].append(t)
#PAST CONTINUOUS
elif(c==14):
phras_rules['R6'].append(t)
#PAST PERFECT
elif(c==15):
phras_rules['R7'].append(t)
#PAST PERFECT CONTINUOUS
elif(c==9):
phras_rules['R8'].append(t)
#SIMPLE FUTURE
elif(c==10):
phras_rules['R9'].append(t)
#FUTURE CONTINUOUS
elif(c==21):
phras_rules['R10'].append(t)
#FUTURE PERFECT
elif(c==22):
phras_rules['R11'].append(t)
#FUTURE PERFECT CONTINUOUS
t = []
fsm.current = '0'
# backup_k = k
k = count + k
flag = 1
# break
if(temp_current in to_rb and j!=len(tagged_sentence)-1):
# print("----temp-----")
# print("\n",temp_current)
fsm.current = temp_current
try:
if(fsm.current=='20'):
check = fsm.current
s.append(tagged_sentence[j+1])
fsm.trigger(tagged_sentence[j+2][1])
else:
check = fsm.current
s.append(tagged_sentence[j])
fsm.trigger(tagged_sentence[j+1][1])
# print("-------ENTERED TRYYYYY-------"
# print(tagged_sentence[j+1])
# print("\n",fsm.current)
if(fsm.current=='11'):
# print("----ENTERED IF----")
if(check=='20'):
s.append(tagged_sentence[j+2])
else:
s.append(tagged_sentence[j+1])
# count += 1
else:
# print("-----OOPS ELSEEE---")
s=[]
fsm.current =temp_current
rbflag = 1
except:
# print("----Uhohhhh------")
oops = 1
t = []
fsm.current = '0'
if(fsm.current=='11'):
phrase_count+=1
phras_rules['R12'].append(s)
fsm.current = '0'
s = []
if(flag==0):
k += 1
if(rbflag==1 and flag==1):
break
english_sentence_structure[sentence].append(phrase_count)
english_sentence_structure[sentence].append(phras_rules)
def SenToPhrase(tagged_sentence):
fsm = Fysom({
'initial':
'0',
'events': [{
'name': 'IN',
'src': '0',
'dst': '1'
}, {
'name': 'NN',
'src': '1',
'dst': '3'
}, {
'name': 'NNS',
'src': '1',
'dst': '3'
}, {
'name': 'NNP',
'src': '1',
'dst': '3'
}, {
'name': 'NNPS',
'src': '1',
'dst': '3'
}, {
'name': 'DT',
'src': '1',
'dst': '2'
}, {
'name': 'NN',
'src': '2',
'dst': '3'
}, {
'name': 'NNS',
'src': '2',
'dst': '3'
}, {
'name': 'NNP',
'src': '2',
'dst': '3'
}, {
'name': 'NNPS',
'src': '2',
'dst': '3'
}, {
'name': 'PRP$',
'src': '1',
'dst': '4'
}, {
'name': 'PRP$',
'src': '2',
'dst': '4'
}, {
'name': 'JJ',
'src': '1',
'dst': '5'
}, {
'name': 'JJ',
'src': '2',
'dst': '5'
}, {
'name': 'JJR',
'src': '1',
'dst': '6'
}, {
'name': 'JJR',
'src': '2',
'dst': '6'
}, {
'name': 'JJS',
'src': '1',
'dst': '7'
}, {
'name': 'JJS',
'src': '2',
'dst': '7'
}, {
'name': 'NN',
'src': '5',
'dst': '3'
}, {
'name': 'NN',
'src': '6',
'dst': '3'
}, {
'name': 'NN',
'src': '7',
'dst': '3'
}, {
'name': 'NNS',
'src': '5',
'dst': '3'
}, {
'name': 'NNS',
'src': '6',
'dst': '3'
}, {
'name': 'NNS',
'src': '7',
'dst': '3'
}, {
'name': 'NNP',
'src': '5',
'dst': '3'
}, {
'name': 'NNP',
'src': '6',
'dst': '3'
}, {
'name': 'NNP',
'src': '7',
'dst': '3'
}, {
'name': 'NNPS',
'src': '5',
'dst': '3'
}, {
'name': 'NNPS',
'src': '6',
'dst': '3'
}, {
'name': 'NNPS',
'src': '7',
'dst': '3'
}, {
'name': 'PRP',
'src': '1',
'dst': '4'
}, {
'name': 'PRP',
'src': '2',
'dst': '4'
}, {
'name': 'NN',
'src': '4',
'dst': '3'
}, {
'name': 'NNS',
'src': '4',
'dst': '3'
}, {
'name': 'NNP',
'src': '4',
'dst': '3'
}, {
'name': 'NNPS',
'src': '4',
'dst': '3'
}, {
'name': 'TO',
'src': '0',
'dst': '1'
}, {
'name': 'MD',
'src': '0',
'dst': '8'
}, {
'name': 'VB',
'src': '8',
'dst': '9'
}, {
'name': 'VBD',
'src': '8',
'dst': '10'
}, {
'name': 'VBG',
'src': '8',
'dst': '11'
}, {
'name': 'VBN',
'src': '8',
'dst': '12'
}, {
'name': 'VBP',
'src': '8',
'dst': '13'
}, {
'name': 'VBZ',
'src': '8',
'dst': '14'
}, {
'name': 'VB',
'src': '0',
'dst': '9'
}, {
'name': 'VBD',
'src': '0',
'dst': '10'
}, {
'name': 'VBG',
'src': '0',
'dst': '11'
}, {
'name': 'VBN',
'src': '0',
'dst': '12'
}, {
'name': 'VBP',
'src': '0',
'dst': '13'
}, {
'name': 'VBZ',
'src': '0',
'dst': '14'
}, {
'name': 'RB',
'src': '9',
'dst': '15'
}, {
'name': 'RB',
'src': '10',
'dst': '15'
}, {
'name': 'RB',
'src': '11',
'dst': '15'
}, {
'name': 'RB',
'src': '12',
'dst': '15'
}, {
'name': 'RB',
'src': '13',
'dst': '15'
}, {
'name': 'RB',
'src': '14',
'dst': '15'
}, {
'name': 'JJ',
'src': '9',
'dst': '16'
}, {
'name': 'JJ',
'src': '10',
'dst': '16'
}, {
'name': 'JJ',
'src': '11',
'dst': '16'
}, {
'name': 'JJ',
'src': '12',
'dst': '16'
}, {
'name': 'JJ',
'src': '13',
'dst': '16'
}, {
'name': 'JJ',
'src': '14',
'dst': '16'
}, {
'name': 'VB',
'src': '15',
'dst': '17'
}, {
'name': 'VBD',
'src': '15',
'dst': '18'
}, {
'name': 'VBG',
'src': '15',
'dst': '19'
}, {
'name': 'VBN',
'src': '15',
'dst': '20'
}, {
'name': 'VBP',
'src': '15',
'dst': '21'
}, {
'name': 'VBZ',
'src': '15',
'dst': '22'
}, {
'name': 'VB',
'src': '9',
'dst': '17'
}, {
'name': 'VBD',
'src': '9',
'dst': '18'
}, {
'name': 'VBG',
'src': '9',
'dst': '19'
}, {
'name': 'VBN',
'src': '9',
'dst': '20'
}, {
'name': 'VBP',
'src': '9',
'dst': '21'
}, {
'name': 'VBZ',
'src': '9',
'dst': '22'
}, {
'name': 'VB',
'src': '10',
'dst': '17'
}, {
'name': 'VBD',
'src': '10',
'dst': '18'
}, {
'name': 'VBG',
'src': '10',
'dst': '19'
}, {
'name': 'VBN',
'src': '10',
'dst': '20'
}, {
'name': 'VBP',
'src': '10',
'dst': '21'
}, {
'name': 'VBZ',
'src': '10',
'dst': '22'
}, {
'name': 'VB',
'src': '11',
'dst': '17'
}, {
'name': 'VBD',
'src': '11',
'dst': '18'
}, {
'name': 'VBG',
'src': '11',
'dst': '19'
}, {
'name': 'VBN',
'src': '11',
'dst': '20'
}, {
'name': 'VBP',
'src': '11',
'dst': '21'
}, {
'name': 'VBZ',
'src': '11',
'dst': '22'
}, {
'name': 'VB',
'src': '12',
'dst': '17'
}, {
'name': 'VBD',
'src': '12',
'dst': '18'
}, {
'name': 'VBG',
'src': '12',
'dst': '19'
}, {
'name': 'VBN',
'src': '12',
'dst': '20'
}, {
'name': 'VBP',
'src': '12',
'dst': '21'
}, {
'name': 'VBZ',
'src': '12',
'dst': '22'
}, {
'name': 'VB',
'src': '13',
'dst': '17'
}, {
'name': 'VBD',
'src': '13',
'dst': '18'
}, {
'name': 'VBG',
'src': '13',
'dst': '19'
}, {
'name': 'VBN',
'src': '13',
'dst': '20'
}, {
'name': 'VBP',
'src': '13',
'dst': '21'
}, {
'name': 'VBZ',
'src': '13',
'dst': '22'
}, {
'name': 'VB',
'src': '14',
'dst': '17'
}, {
'name': 'VBD',
'src': '14',
'dst': '18'
}, {
'name': 'VBG',
'src': '14',
'dst': '19'
}, {
'name': 'VBN',
'src': '14',
'dst': '20'
}, {
'name': 'VBP',
'src': '14',
'dst': '21'
}, {
'name': 'VBZ',
'src': '14',
'dst': '22'
}]
})
high_final_states = ['3', '15', '16', '17', '18', '19', '20', '21', '22']
low_final_states = ['4']
final = []
fsm.current = '0'
new_temp = ""
for j in tagged_sentence:
try:
fsm.trigger(j[1])
new_temp += j[0] + " "
except:
fsm.current = '0'
new_temp = ""
finally:
if (fsm.current in high_final_states):
fsm.current = "0"
new_temp = new_temp[:-1]
#final.append(new_temp)
final.append(english_postagger.tag(new_temp.split()))
new_temp = ""
elif (fsm.current in low_final_states):
fsm.current = "0"
new_temp = new_temp[:-1]
#final.append(new_temp)
final.append(english_postagger.tag(new_temp.split()))
new_temp = ""
s = " "
return final
