def extractPhrases(normalToSimple, simpleToNormal):
data = []
fns = open(normalToSimple, 'r')
fsn = open(simpleToNormal, 'r')
count = 0
while True:
count += 1
print(count)
line = fns.readline()
if line == "":
break
normalToSimple1 = fns.readline()
normalToSimple2 = fns.readline()
line = fsn.readline()
simpleToNormal1 = fsn.readline()
simpleToNormal2 = fsn.readline()
wordAlignment, normal, simple = findAlignment(normalToSimple1,
normalToSimple2,
simpleToNormal1,
simpleToNormal2)
lNormal = len(normal)
lSimple = len(simple)
# print(wordAlignment)
phrases = []
for estart in range(lNormal):
for eend in range(estart, (lNormal)):
fstart = lSimple
fend = -1
for i in wordAlignment:
if i <= eend and i >= estart:
for j in wordAlignment[i]:
fstart = min(j, fstart)
fend = max(j, fend)
if ((eend - estart) <= 20) or ((fend - fstart) <= 20):
phrases.append([estart, eend, fstart, fend])
# print phrases
for key in phrases:
estart = key[0]
eend = key[1]
fstart = key[2]
fend = key[3]
phrase = extract(fstart, fend, estart, eend, wordAlignment, normal,
simple)
# print (phrase)
if phrase != 'NULL':
# print (phrase)
data.append(phrase[0] + '\t' + phrase[1])
fns.close()
fsn.close()
f = open('phrases.txt', 'w')
f.write('\n'.join(data))
f.close()
def extractPhrases(englishToGerman, germanToEnglish):
'''this method reads the file for both the source and target language and returns the phrases extracted from the
sentences. The phrases are consistent in nature'''
data=[]
feg = open(englishToGerman, 'r')
fge = open(germanToEnglish,'r')
count = 0
while True:
count+=1
print count
line = feg.readline()
if line == "":
break
englishToGerman1 = feg.readline()
englishToGerman2 = feg.readline()
#print englishToGerman1
line = fge.readline()
germanToEnglish1 = fge.readline()
germanToEnglish2 = fge.readline()
#print germanToEnglish1
wordAlignment, english, german = findAlignment(englishToGerman1, englishToGerman2, germanToEnglish1, germanToEnglish2)
lEnglish = len(english)
lGerman = len(german)
phrases = []
for estart in range(lEnglish):
for eend in range(estart,(lEnglish)):
fstart = lGerman
fend = -1
for i in wordAlignment:
if i <= eend and i >= estart:
for j in wordAlignment[i]:
fstart = min(j, fstart)
fend = max(j, fend)
if ((eend - estart) <= 20) or ((fend -fstart) <= 20) :
phrases.append([estart, eend, fstart, fend])
#print phrases
for key in phrases:
estart = key[0]
eend = key[1]
fstart = key [2]
fend = key[3]
phrase = extract (fstart, fend,estart, eend,wordAlignment, english, german)
if phrase!= 'NULL':
#print phrase
data.append(phrase[0]+'\t'+phrase[1])
feg.close()
fge.close()
f=open('phrases.txt','w')
f.write('\n'.join(data))
f.close()
def extractPhrases(sourceToTarget, targetToSource):
'''this method reads the file for both the source and target language and returns the phrases extracted from the
sentences. The phrases are consistent in nature'''
data = []
feg = open(sourceToTarget, 'r')
fge = open(targetToSource, 'r')
count = 0
while True:
count += 1
print count
line = feg.readline()
if line == "":
break
sourceToTarget1 = feg.readline()
sourceToTarget2 = feg.readline()
#print sourceToTarget1
line = fge.readline()
targetToSource1 = fge.readline()
targetToSource2 = fge.readline()
#print targetToSource1
wordAlignment, source, target = findAlignment(sourceToTarget1,
sourceToTarget2,
targetToSource1,
targetToSource2)
lSource = len(source)
lTarget = len(target)
phrases = []
for estart in range(lSource):
for eend in range(estart, (lSource)):
fstart = lTarget
fend = -1
for i in wordAlignment:
if i <= eend and i >= estart:
for j in wordAlignment[i]:
fstart = min(j, fstart)
fend = max(j, fend)
if ((eend - estart) <= 20) or ((fend - fstart) <= 20):
phrases.append([estart, eend, fstart, fend])
# print phrases
for key in phrases:
estart = key[0]
eend = key[1]
fstart = key[2]
fend = key[3]
phrase = extract(fstart, fend, estart, eend, wordAlignment, source,
target)
if phrase != 'NULL':
#print phrase
tmp = (phrase[0].strip() + '\t' + phrase[1].strip())
if tmp.strip() != "" and tmp not in data:
data.append(tmp)
feg.close()
fge.close()
f = open('phrases.txt', 'w')
f.write('\n'.join(data))
f.close()
_ = cv2.circle(disp, (p[0] - rect[0], p[1] - rect[1]), r, (0, 0, v),
cv2.FILLED)
for i, p in enumerate(np.asarray(_2[:, :2] * 20, np.int32)):
v = 128 + int(127 * (_2[i, 2] - mini) / (maxi - mini + 1))
r = 5 if i == 0 else 1
_ = cv2.circle(disp, (p[0] - rect[0], p[1] - rect[1]), r, (0, v, 0),
cv2.FILLED)
return disp
disp = display(dst, src)
cv2.imshow("disp", disp)
cv2.waitKey(0)
cv2.destroyAllWindows()
R, t = findRigidTransform(dst, src)
src1 = dst @ R.T + t
disp = display(src1, src)
cv2.imshow("disp", disp)
cv2.waitKey(0)
cv2.destroyAllWindows()
R, t = findAlignment(dst, src)
src2 = dst @ R.T + t
disp = display(src2, src)
cv2.imshow("disp", disp)
cv2.waitKey(0)
cv2.destroyAllWindows()
def extractPhrases(source_target, target_source):
'''
Extract consistent phrases from the bi-lingual sentences
'''
data = []
src_align_file = open(source_target, 'r')
trg_align_file = open(target_source, 'r')
count = 0
while True:
count += 1
#print(count)
line = src_align_file.readline()
if line == "":
break
target_txt = src_align_file.readline()
source_align_idx = src_align_file.readline()
#print("Target Text:", target_txt.rstrip('\n'))
line = trg_align_file.readline()
source_txt = trg_align_file.readline()
target_align_idx = trg_align_file.readline()
#print("Source Text:", source_txt.rstrip('\n'))
if float(line.strip().split(':')[1].strip()) < 1e-18:
continue
word_alignment, source, target = findAlignment(target_txt, source_align_idx, source_txt, target_align_idx)
print("Source Text:", source)
print("Target Text:", target)
print("Word Alignment:", word_alignment, end = "\n\n")
src_len = len(source)
trg_len = len(target)
phrases = []
for src_start in range(src_len):
for src_end in range(src_start,(src_len)):
trg_start = trg_len
trg_end = -1
for i in word_alignment:
if i <= src_end and i >= src_start:
for j in word_alignment[i]:
trg_start = min(j, trg_start)
trg_end = max(j, trg_end)
if ((src_end - src_start) <= 5) or ((trg_end - trg_start) <= 5) :
phrases.append([src_start, src_end, trg_start, trg_end])
#print(phrases)
for key in phrases:
src_start = key[0]
src_end = key[1]
trg_start = key [2]
trg_end = key[3]
phrase = extract (src_start, src_end, trg_start, trg_end, word_alignment, source, target)
if phrase is not None:
#print(phrase)
data.append(phrase[0] + '\t' + phrase[1])
src_align_file.close()
trg_align_file.close()
phrase_file = open('phrases.txt','w')
phrase_file.write('\n'.join(data))
phrase_file.close()