Spaces:
Running
Running
Yurii Paniv
commited on
Commit
•
c010ef4
1
Parent(s):
8472231
Add simple English letter handling
Browse files- formatter.py +39 -0
formatter.py
CHANGED
@@ -13,4 +13,43 @@ def preprocess_text(text):
|
|
13 |
text = text.replace("8", "вісім ")
|
14 |
text = text.replace("9", "дев'ять ")
|
15 |
text = text.replace("0", "нуль ")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
16 |
return text
|
|
|
|
|
|
|
|
|
|
13 |
text = text.replace("8", "вісім ")
|
14 |
text = text.replace("9", "дев'ять ")
|
15 |
text = text.replace("0", "нуль ")
|
16 |
+
# speak english alphabet using brute force transliteration
|
17 |
+
english = {
|
18 |
+
"a": "а",
|
19 |
+
"b": "б",
|
20 |
+
"c": "ц",
|
21 |
+
"d": "д",
|
22 |
+
"e": "е",
|
23 |
+
"f": "ф",
|
24 |
+
"g": "ґ",
|
25 |
+
"h": "г",
|
26 |
+
"i": "і",
|
27 |
+
"j": "дж",
|
28 |
+
"k": "к",
|
29 |
+
"l": "л",
|
30 |
+
"m": "м",
|
31 |
+
"n": "н",
|
32 |
+
"o": "о",
|
33 |
+
"p": "п",
|
34 |
+
"q": "кв",
|
35 |
+
"r": "р",
|
36 |
+
"s": "с",
|
37 |
+
"t": "т",
|
38 |
+
"u": "ю",
|
39 |
+
"v": "в",
|
40 |
+
"w": "в",
|
41 |
+
"x": "кс",
|
42 |
+
"y": "й",
|
43 |
+
"z": "з",
|
44 |
+
}
|
45 |
+
for english_char in english.keys():
|
46 |
+
# uppercase
|
47 |
+
text = text.replace(english_char.upper(), english[english_char].upper())
|
48 |
+
text = text.replace(english_char, english[english_char])
|
49 |
+
|
50 |
+
# TODO: autostress support here
|
51 |
return text
|
52 |
+
|
53 |
+
|
54 |
+
if __name__ == "__main__":
|
55 |
+
print(preprocess_text("Quality of life update"))
|