Commit
·
75aab4f
1
Parent(s):
03e97cd
Refactor get_better_translation function in translators.py
Browse files- helper/translators.py +2 -128
helper/translators.py
CHANGED
|
@@ -48,139 +48,13 @@ def google_translation(sentence: str, source="auto", target="en") -> str:
|
|
| 48 |
return translated_sentence
|
| 49 |
|
| 50 |
|
| 51 |
-
def get_better_translation(
|
| 52 |
src_mod = get_translated_digit(src)
|
| 53 |
-
tgt =
|
| 54 |
tgt = decontracting_words(tgt)
|
| 55 |
tgt = tgt.replace('rupees', 'takas').replace('Rs', 'takas')
|
| 56 |
return tgt
|
| 57 |
|
| 58 |
-
# def select_translator(src, translator):
|
| 59 |
-
# """
|
| 60 |
-
# Select the translator
|
| 61 |
-
# """
|
| 62 |
-
# tgt = None
|
| 63 |
-
# tgt_base = None
|
| 64 |
-
|
| 65 |
-
# if translator == "Google":
|
| 66 |
-
# tgt = get_better_translation(google_translation, src)
|
| 67 |
-
# tgt = space_punc(tgt)
|
| 68 |
-
# tgt_base = google_translation(src)
|
| 69 |
-
# elif translator == "BanglaNMT":
|
| 70 |
-
# tgt = get_better_translation(banglanmt_translation, src)
|
| 71 |
-
# tgt = space_punc(tgt)
|
| 72 |
-
# tgt_base = banglanmt_translation(src)
|
| 73 |
-
# elif translator == "MyMemory":
|
| 74 |
-
# tgt = get_better_translation(mymemory_translation, src)
|
| 75 |
-
# tgt = space_punc(tgt)
|
| 76 |
-
# tgt_base = mymemory_translation(src)
|
| 77 |
-
|
| 78 |
-
# return tgt_base, tgt
|
| 79 |
-
|
| 80 |
-
# Afrikaans
|
| 81 |
-
# Albanian
|
| 82 |
-
# Arabic
|
| 83 |
-
# Aragonese
|
| 84 |
-
# Armenian
|
| 85 |
-
# Asturian
|
| 86 |
-
# Azerbaijani
|
| 87 |
-
# Bashkir
|
| 88 |
-
# Basque
|
| 89 |
-
# Bavarian
|
| 90 |
-
# Belarusian
|
| 91 |
-
# Bengali
|
| 92 |
-
# Bishnupriya Manipuri
|
| 93 |
-
# Bosnian
|
| 94 |
-
# Breton
|
| 95 |
-
# Bulgarian
|
| 96 |
-
# Burmese
|
| 97 |
-
# Catalan
|
| 98 |
-
# Cebuano
|
| 99 |
-
# Chechen
|
| 100 |
-
# Chinese (Simplified)
|
| 101 |
-
# Chinese (Traditional)
|
| 102 |
-
# Chuvash
|
| 103 |
-
# Croatian
|
| 104 |
-
# Czech
|
| 105 |
-
# Danish
|
| 106 |
-
# Dutch
|
| 107 |
-
# English
|
| 108 |
-
# Estonian
|
| 109 |
-
# Finnish
|
| 110 |
-
# French
|
| 111 |
-
# Galician
|
| 112 |
-
# Georgian
|
| 113 |
-
# German
|
| 114 |
-
# Greek
|
| 115 |
-
# Gujarati
|
| 116 |
-
# Haitian
|
| 117 |
-
# Hebrew
|
| 118 |
-
# Hindi
|
| 119 |
-
# Hungarian
|
| 120 |
-
# Icelandic
|
| 121 |
-
# Ido
|
| 122 |
-
# Indonesian
|
| 123 |
-
# Irish
|
| 124 |
-
# Italian
|
| 125 |
-
# Japanese
|
| 126 |
-
# Javanese
|
| 127 |
-
# Kannada
|
| 128 |
-
# Kazakh
|
| 129 |
-
# Kirghiz
|
| 130 |
-
# Korean
|
| 131 |
-
# Latin
|
| 132 |
-
# Latvian
|
| 133 |
-
# Lithuanian
|
| 134 |
-
# Lombard
|
| 135 |
-
# Low Saxon
|
| 136 |
-
# Luxembourgish
|
| 137 |
-
# Macedonian
|
| 138 |
-
# Malagasy
|
| 139 |
-
# Malay
|
| 140 |
-
# Malayalam
|
| 141 |
-
# Marathi
|
| 142 |
-
# Minangkabau
|
| 143 |
-
# Nepali
|
| 144 |
-
# Newar
|
| 145 |
-
# Norwegian (Bokmal)
|
| 146 |
-
# Norwegian (Nynorsk)
|
| 147 |
-
# Occitan
|
| 148 |
-
# Persian (Farsi)
|
| 149 |
-
# Piedmontese
|
| 150 |
-
# Polish
|
| 151 |
-
# Portuguese
|
| 152 |
-
# Punjabi
|
| 153 |
-
# Romanian
|
| 154 |
-
# Russian
|
| 155 |
-
# Scots
|
| 156 |
-
# Serbian
|
| 157 |
-
# Serbo-Croatian
|
| 158 |
-
# Sicilian
|
| 159 |
-
# Slovak
|
| 160 |
-
# Slovenian
|
| 161 |
-
# South Azerbaijani
|
| 162 |
-
# Spanish
|
| 163 |
-
# Sundanese
|
| 164 |
-
# Swahili
|
| 165 |
-
# Swedish
|
| 166 |
-
# Tagalog
|
| 167 |
-
# Tajik
|
| 168 |
-
# Tamil
|
| 169 |
-
# Tatar
|
| 170 |
-
# Telugu
|
| 171 |
-
# Turkish
|
| 172 |
-
# Ukrainian
|
| 173 |
-
# Urdu
|
| 174 |
-
# Uzbek
|
| 175 |
-
# Vietnamese
|
| 176 |
-
# Volapük
|
| 177 |
-
# Waray-Waray
|
| 178 |
-
# Welsh
|
| 179 |
-
# West Frisian
|
| 180 |
-
# Western Punjabi
|
| 181 |
-
# Yoruba
|
| 182 |
-
# Thai
|
| 183 |
-
# Mongolian
|
| 184 |
|
| 185 |
target_lang_dict = {
|
| 186 |
"Afrikaans": "af",
|
|
|
|
| 48 |
return translated_sentence
|
| 49 |
|
| 50 |
|
| 51 |
+
def get_better_translation(src=""):
|
| 52 |
src_mod = get_translated_digit(src)
|
| 53 |
+
tgt = google_translation(src_mod)
|
| 54 |
tgt = decontracting_words(tgt)
|
| 55 |
tgt = tgt.replace('rupees', 'takas').replace('Rs', 'takas')
|
| 56 |
return tgt
|
| 57 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 58 |
|
| 59 |
target_lang_dict = {
|
| 60 |
"Afrikaans": "af",
|