Spaces:
Sleeping
Sleeping
Update asr-tts_service.py
Browse files- asr-tts_service.py +29 -8
asr-tts_service.py
CHANGED
|
@@ -73,18 +73,39 @@ def number_to_french(n: int) -> str:
|
|
| 73 |
base = "quatre-vingt"
|
| 74 |
if n == 80: return base
|
| 75 |
return base + "-" + number_to_french(n - 80)
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 79 |
|
| 80 |
def convert_digits_in_text(text: str) -> str:
|
|
|
|
|
|
|
| 81 |
def repl(match):
|
| 82 |
s = match.group(0)
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 88 |
return re.sub(r"\d+", repl, text)
|
| 89 |
|
| 90 |
def split_by_sentences(text: str, max_chars: int = 150) -> list:
|
|
|
|
| 73 |
base = "quatre-vingt"
|
| 74 |
if n == 80: return base
|
| 75 |
return base + "-" + number_to_french(n - 80)
|
| 76 |
+
|
| 77 |
+
# GESTION DES CENTAINES
|
| 78 |
+
if n < 1000:
|
| 79 |
+
hundreds, rest = divmod(n, 100)
|
| 80 |
+
base = "cent" if hundreds == 1 else UNITS.get(hundreds, str(hundreds)) + " cent"
|
| 81 |
+
return base if rest == 0 else base + " " + number_to_french(rest)
|
| 82 |
+
|
| 83 |
+
# GESTION DES MILLIERS ET PLUS (Sécurité pour éviter le KeyError)
|
| 84 |
+
if n < 1000000:
|
| 85 |
+
thousands, rest = divmod(n, 1000)
|
| 86 |
+
base = "mille" if thousands == 1 else number_to_french(thousands) + " mille"
|
| 87 |
+
return base if rest == 0 else base + " " + number_to_french(rest)
|
| 88 |
+
|
| 89 |
+
# Par sécurité, si le nombre est trop grand (millions), on le renvoie tel quel
|
| 90 |
+
return str(n)
|
| 91 |
|
| 92 |
def convert_digits_in_text(text: str) -> str:
|
| 93 |
+
if not isinstance(text, str): return ""
|
| 94 |
+
|
| 95 |
def repl(match):
|
| 96 |
s = match.group(0)
|
| 97 |
+
# SI C'EST UN LONG NOMBRE (Téléphone ou Réf > 4 chiffres)
|
| 98 |
+
# On épèle : "7 7 1 2 3..." au lieu de "Sept milliards..."
|
| 99 |
+
if len(s) > 4:
|
| 100 |
+
return " ".join([UNITS.get(int(digit), digit) for digit in s])
|
| 101 |
+
|
| 102 |
+
# SINON (Petit nombre comme 15, 100, 2024)
|
| 103 |
+
try:
|
| 104 |
+
val = int(s)
|
| 105 |
+
return number_to_french(val)
|
| 106 |
+
except:
|
| 107 |
+
return s
|
| 108 |
+
|
| 109 |
return re.sub(r"\d+", repl, text)
|
| 110 |
|
| 111 |
def split_by_sentences(text: str, max_chars: int = 150) -> list:
|