deshanksuman
commited on
Commit
•
64766e7
1
Parent(s):
4010517
Upload 3 files
Browse filesThe files been uploaded.
- TranslaterLogic.py +276 -0
- Transliterator.py +23 -0
- trigramTrans.pickle +3 -0
TranslaterLogic.py
ADDED
@@ -0,0 +1,276 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
nVowels =26
|
3 |
+
# 46
|
4 |
+
consonants = []
|
5 |
+
consonantsUni = []
|
6 |
+
|
7 |
+
# 26
|
8 |
+
vowels = []
|
9 |
+
vowelsUni = []
|
10 |
+
vowelModifiersUni = []
|
11 |
+
|
12 |
+
# 6
|
13 |
+
specialConsonants = []
|
14 |
+
specialConsonantsUni = []
|
15 |
+
# 2
|
16 |
+
specialCharUni = []
|
17 |
+
specialChar = []
|
18 |
+
|
19 |
+
|
20 |
+
def initializeVar():
|
21 |
+
|
22 |
+
vowelsUni.append("ඌ")
|
23 |
+
vowels.append("oo")
|
24 |
+
vowelModifiersUni.append("ූ")
|
25 |
+
|
26 |
+
vowelsUni.append("ඕ")
|
27 |
+
vowels.append("o\\)")
|
28 |
+
vowelModifiersUni.append("ෝ")
|
29 |
+
vowelsUni.append("ඕ")
|
30 |
+
vowels.append("oe")
|
31 |
+
vowelModifiersUni.append("ෝ")
|
32 |
+
vowelsUni.append("ආ")
|
33 |
+
vowels.append("aa")
|
34 |
+
vowelModifiersUni.append("ා")
|
35 |
+
vowelsUni.append("ආ")
|
36 |
+
vowels.append("a\\)")
|
37 |
+
vowelModifiersUni.append("ා")
|
38 |
+
vowelsUni.append("ඈ")
|
39 |
+
vowels.append("Aa")
|
40 |
+
vowelModifiersUni.append("ෑ")
|
41 |
+
vowelsUni.append("ඈ")
|
42 |
+
vowels.append("A\\)")
|
43 |
+
vowelModifiersUni.append("ෑ")
|
44 |
+
vowelsUni.append("ඈ")
|
45 |
+
vowels.append("ae")
|
46 |
+
vowelModifiersUni.append("ෑ")
|
47 |
+
vowelsUni.append("ඊ")
|
48 |
+
vowels.append("ii")
|
49 |
+
vowelModifiersUni.append("ී")
|
50 |
+
vowelsUni.append("ඊ")
|
51 |
+
vowels.append("i\\)")
|
52 |
+
vowelModifiersUni.append("ී")
|
53 |
+
vowelsUni.append("ඊ")
|
54 |
+
vowels.append("ie")
|
55 |
+
vowelModifiersUni.append("ී")
|
56 |
+
vowelsUni.append("ඊ")
|
57 |
+
vowels.append("ee")
|
58 |
+
vowelModifiersUni.append("ී")
|
59 |
+
vowelsUni.append("ඒ")
|
60 |
+
vowels.append("ea")
|
61 |
+
vowelModifiersUni.append("ේ")
|
62 |
+
vowelsUni.append("ඒ")
|
63 |
+
vowels.append("e\\)")
|
64 |
+
vowelModifiersUni.append("ේ")
|
65 |
+
vowelsUni.append("ඒ")
|
66 |
+
vowels.append("ei")
|
67 |
+
vowelModifiersUni.append("ේ")
|
68 |
+
vowelsUni.append("ඌ")
|
69 |
+
vowels.append("uu")
|
70 |
+
vowelModifiersUni.append("ූ")
|
71 |
+
vowelsUni.append("ඌ")
|
72 |
+
vowels.append("u\\)")
|
73 |
+
vowelModifiersUni.append("ූ")
|
74 |
+
|
75 |
+
vowelsUni.append("ඖ")
|
76 |
+
vowels.append("au")
|
77 |
+
vowelModifiersUni.append("ෞ")
|
78 |
+
|
79 |
+
vowelsUni.append("ඇ")
|
80 |
+
vowels.append("\\a")
|
81 |
+
vowelModifiersUni.append("ැ")
|
82 |
+
|
83 |
+
vowelsUni.append("අ")
|
84 |
+
vowels.append("a")
|
85 |
+
vowelModifiersUni.append("")
|
86 |
+
|
87 |
+
vowelsUni.append("ඇ")
|
88 |
+
vowels.append("A")
|
89 |
+
vowelModifiersUni.append("ැ")
|
90 |
+
vowelsUni.append("ඉ")
|
91 |
+
vowels.append("i")
|
92 |
+
vowelModifiersUni.append("ි")
|
93 |
+
vowelsUni.append("එ")
|
94 |
+
vowels.append("e")
|
95 |
+
vowelModifiersUni.append("ෙ")
|
96 |
+
vowelsUni.append("උ")
|
97 |
+
vowels.append("u")
|
98 |
+
vowelModifiersUni.append("ු")
|
99 |
+
vowelsUni.append("ඔ")
|
100 |
+
vowels.append("o")
|
101 |
+
vowelModifiersUni.append("ො")
|
102 |
+
vowelsUni.append("ඓ")
|
103 |
+
vowels.append("I")
|
104 |
+
vowelModifiersUni.append("ෛ")
|
105 |
+
|
106 |
+
specialConsonantsUni.append("ං")
|
107 |
+
specialConsonants.append("\\n")
|
108 |
+
|
109 |
+
specialConsonantsUni.append("ඃ")
|
110 |
+
specialConsonants.append("\\h")
|
111 |
+
specialConsonantsUni.append("ඞ")
|
112 |
+
specialConsonants.append("\\N")
|
113 |
+
specialConsonantsUni.append("ඍ")
|
114 |
+
specialConsonants.append("\\R")
|
115 |
+
# special characher Repaya
|
116 |
+
specialConsonantsUni.append("ර්"+"\u200D")
|
117 |
+
specialConsonants.append("R")
|
118 |
+
specialConsonantsUni.append("ර්"+"\u200D")
|
119 |
+
specialConsonants.append("\\r")
|
120 |
+
|
121 |
+
consonantsUni.append("ඬ")
|
122 |
+
consonants.append("nnd")
|
123 |
+
|
124 |
+
consonantsUni.append("ඳ")
|
125 |
+
consonants.append("nndh")
|
126 |
+
|
127 |
+
consonantsUni.append("ඟ")
|
128 |
+
consonants.append("nng")
|
129 |
+
|
130 |
+
consonantsUni.append("ත")
|
131 |
+
consonants.append("th")
|
132 |
+
|
133 |
+
consonantsUni.append("ධ")
|
134 |
+
consonants.append("dh")
|
135 |
+
consonantsUni.append("ඝ")
|
136 |
+
consonants.append("gh")
|
137 |
+
consonantsUni.append("ච")
|
138 |
+
consonants.append("ch")
|
139 |
+
consonantsUni.append("ඵ")
|
140 |
+
consonants.append("ph")
|
141 |
+
consonantsUni.append("භ")
|
142 |
+
consonants.append("bh")
|
143 |
+
consonantsUni.append("ඣ")
|
144 |
+
consonants.append("jh")
|
145 |
+
consonantsUni.append("ෂ")
|
146 |
+
consonants.append("sh")
|
147 |
+
consonantsUni.append("ඥ")
|
148 |
+
consonants.append("GN")
|
149 |
+
consonantsUni.append("ඤ")
|
150 |
+
consonants.append("KN")
|
151 |
+
consonantsUni.append("ළු")
|
152 |
+
consonants.append("Lu")
|
153 |
+
consonantsUni.append("ඛ")
|
154 |
+
consonants.append("kh")
|
155 |
+
consonantsUni.append("ඨ")
|
156 |
+
consonants.append("Th")
|
157 |
+
consonantsUni.append("ඪ")
|
158 |
+
consonants.append("Dh")
|
159 |
+
consonantsUni.append("ශ")
|
160 |
+
consonants.append("S")
|
161 |
+
consonantsUni.append("ද")
|
162 |
+
consonants.append("d")
|
163 |
+
consonantsUni.append("ච")
|
164 |
+
consonants.append("c")
|
165 |
+
consonantsUni.append("ත")
|
166 |
+
consonants.append("th")
|
167 |
+
consonantsUni.append("ට")
|
168 |
+
consonants.append("t")
|
169 |
+
consonantsUni.append("ක")
|
170 |
+
consonants.append("k")
|
171 |
+
consonantsUni.append("ඩ")
|
172 |
+
consonants.append("D")
|
173 |
+
consonantsUni.append("න")
|
174 |
+
consonants.append("n")
|
175 |
+
consonantsUni.append("ප")
|
176 |
+
consonants.append("p")
|
177 |
+
consonantsUni.append("බ")
|
178 |
+
consonants.append("b")
|
179 |
+
consonantsUni.append("ම")
|
180 |
+
consonants.append("m")
|
181 |
+
consonantsUni.append("ය")
|
182 |
+
consonants.append("\\u005C" + "y")
|
183 |
+
consonantsUni.append("ය")
|
184 |
+
consonants.append("Y")
|
185 |
+
consonantsUni.append("ය")
|
186 |
+
consonants.append("y")
|
187 |
+
consonantsUni.append("ජ")
|
188 |
+
consonants.append("j")
|
189 |
+
consonantsUni.append("ල")
|
190 |
+
consonants.append("l")
|
191 |
+
consonantsUni.append("ව")
|
192 |
+
consonants.append("v")
|
193 |
+
consonantsUni.append("ව")
|
194 |
+
consonants.append("w")
|
195 |
+
consonantsUni.append("ස")
|
196 |
+
consonants.append("s")
|
197 |
+
consonantsUni.append("හ")
|
198 |
+
consonants.append("h")
|
199 |
+
consonantsUni.append("ණ")
|
200 |
+
consonants.append("N")
|
201 |
+
consonantsUni.append("ළ")
|
202 |
+
consonants.append("L")
|
203 |
+
consonantsUni.append("ඛ")
|
204 |
+
consonants.append("K")
|
205 |
+
consonantsUni.append("ඝ")
|
206 |
+
consonants.append("G")
|
207 |
+
consonantsUni.append("ඵ")
|
208 |
+
consonants.append("P")
|
209 |
+
consonantsUni.append("ඹ")
|
210 |
+
consonants.append("B")
|
211 |
+
consonantsUni.append("ෆ")
|
212 |
+
consonants.append("f")
|
213 |
+
consonantsUni.append("ග")
|
214 |
+
consonants.append("g")
|
215 |
+
# last because we need to ommit this in dealing with Rakaransha
|
216 |
+
consonantsUni.append("ර")
|
217 |
+
consonants.append("r")
|
218 |
+
specialCharUni.append("ෲ")
|
219 |
+
specialChar.append("ruu")
|
220 |
+
specialCharUni.append("ෘ")
|
221 |
+
specialChar.append("ru")
|
222 |
+
# specialCharUni[2]="්ර" specialChar[2]="ra"
|
223 |
+
|
224 |
+
initializeVar()
|
225 |
+
|
226 |
+
def convertText(text):
|
227 |
+
s=""
|
228 |
+
r=""
|
229 |
+
v=""
|
230 |
+
# text = document.txtBox.box1.value;
|
231 |
+
# special consonents
|
232 |
+
for i in range(len(specialConsonants)):
|
233 |
+
text = text.replace(specialConsonants[i], specialConsonantsUni[i])
|
234 |
+
# consonents + special Chars
|
235 |
+
for i in range(len(specialCharUni)):
|
236 |
+
for j in range(len(consonants)):
|
237 |
+
s = consonants[j] + specialChar[i]
|
238 |
+
v = consonantsUni[j] + specialCharUni[i]
|
239 |
+
# r = new RegExp(s, "g")
|
240 |
+
r = s.replace(s+"/G", "")
|
241 |
+
text = text.replace(r, v)
|
242 |
+
|
243 |
+
# consonants + Rakaransha + vowel modifiers
|
244 |
+
for j in range(len(consonants)):
|
245 |
+
for i in range(len(vowels)):
|
246 |
+
s = consonants[j] + "r" + vowels[i]
|
247 |
+
v = consonantsUni[j] + "්ර" + vowelModifiersUni[i]
|
248 |
+
# r = new RegExp(s, "g")
|
249 |
+
r = s.replace(s+"/G", "")
|
250 |
+
text = text.replace(r, v)
|
251 |
+
s = consonants[j] + "r"
|
252 |
+
v = consonantsUni[j] + "්ර"
|
253 |
+
# r = new RegExp(s, "g")
|
254 |
+
r = s.replace(s+"/G", "")
|
255 |
+
text = text.replace(r, v)
|
256 |
+
# consonents + vowel modifiers
|
257 |
+
for i in range(len(consonants)):
|
258 |
+
for j in range(nVowels):
|
259 |
+
s = consonants[i] + vowels[j]
|
260 |
+
v = consonantsUni[i] + vowelModifiersUni[j]
|
261 |
+
# r = new RegExp(s, "g")
|
262 |
+
r = s.replace(s+"/G", "")
|
263 |
+
text = text.replace(r, v)
|
264 |
+
|
265 |
+
# consonents + HAL
|
266 |
+
for i in range(len(consonants)):
|
267 |
+
r = consonants[i].replace(consonants[i]+"/G", "")
|
268 |
+
text = text.replace(r, consonantsUni[i] + "්")
|
269 |
+
# vowels
|
270 |
+
for i in range(len(vowels)):
|
271 |
+
# r = new RegExp(vowels[i], "g")
|
272 |
+
r = vowels[i].replace(vowels[i]+"/G", "")
|
273 |
+
text = text.replace(r, vowelsUni[i])
|
274 |
+
|
275 |
+
return text
|
276 |
+
|
Transliterator.py
ADDED
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import pickle
|
2 |
+
import nltk
|
3 |
+
import TranslaterLogic
|
4 |
+
|
5 |
+
translatorPic = open("trigramTrans.pickle", "rb")
|
6 |
+
translator = pickle.load(translatorPic)
|
7 |
+
|
8 |
+
def triGramTranslate(sentence):
|
9 |
+
sentence_romanized=sentence.split(" ")
|
10 |
+
translation = ""
|
11 |
+
translated = translator.tag(nltk.word_tokenize(sentence.lower()))
|
12 |
+
print(translated)
|
13 |
+
i=-1
|
14 |
+
for word, trans in translated:
|
15 |
+
i+=1
|
16 |
+
if trans in ('NNN'):
|
17 |
+
translation = translation + str(TranslaterLogic.convertText(str(sentence_romanized[i])) + " ")
|
18 |
+
else:
|
19 |
+
translation = translation + str(trans + " ")
|
20 |
+
return translation
|
21 |
+
|
22 |
+
inputStr=input("Enter the input")
|
23 |
+
print(triGramTranslate(inputStr))
|
trigramTrans.pickle
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9ad177fa31545923ab983f2d7eae186ccb7f0eee1ac36f8fadd8f77485d01a4b
|
3 |
+
size 397988884
|