Spaces:
Running
Running
Upload intents.py
Browse files- intents.py +141 -0
intents.py
ADDED
|
@@ -0,0 +1,141 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import re
|
| 2 |
+
import unicodedata
|
| 3 |
+
|
| 4 |
+
SI_GREET = ["ආයුබෝවන්", "හෙලෝ", "හලෝ", "හායි", "සුභ උදෑසනක්", "සුභ සන්ධ්යාවක්"]
|
| 5 |
+
TA_GREET = ["வணக்கம்", "ஹலோ", "ஹாய்", "காலை வணக்கம்", "மாலை வணக்கம்"]
|
| 6 |
+
|
| 7 |
+
SI_THANKS = ["ස්තුතියි", "බොහොම ස්තුතියි", "තෑන්ක්ස්"]
|
| 8 |
+
TA_THANKS = ["நன்றி", "மிக்க நன்றி", "தாங்க்ஸ்"]
|
| 9 |
+
|
| 10 |
+
SI_HOW_ARE_YOU = [
|
| 11 |
+
"කොහොමද",
|
| 12 |
+
"ඔබ කොහොමද",
|
| 13 |
+
"ඔයා හොඳින්ද",
|
| 14 |
+
"ඔබ හොඳින්ද",
|
| 15 |
+
]
|
| 16 |
+
|
| 17 |
+
TA_HOW_ARE_YOU = [
|
| 18 |
+
"நீங்கள் எப்படி இருக்கிறீர்கள்",
|
| 19 |
+
"நீ எப்படி இருக்கிறாய்",
|
| 20 |
+
"நலமா",
|
| 21 |
+
"சௌக்கியமா",
|
| 22 |
+
]
|
| 23 |
+
|
| 24 |
+
SI_WHO_ARE_YOU = [
|
| 25 |
+
"ඔයා කවුද",
|
| 26 |
+
"ඔබ කවුද",
|
| 27 |
+
]
|
| 28 |
+
|
| 29 |
+
TA_WHO_ARE_YOU = [
|
| 30 |
+
"நீங்கள் யார்",
|
| 31 |
+
"நீ யார்",
|
| 32 |
+
]
|
| 33 |
+
|
| 34 |
+
SI_WHAT_CAN_YOU_DO = [
|
| 35 |
+
"ඔයාට මොනවා කරන්න පුළුවන්",
|
| 36 |
+
"ඔබට මොනවා කරන්න පුළුවන්",
|
| 37 |
+
"ඔයා මොකක්ද කරන්නෙ",
|
| 38 |
+
"ඔබට මොකක්ද කරන්න පුළුවන්",
|
| 39 |
+
]
|
| 40 |
+
|
| 41 |
+
TA_WHAT_CAN_YOU_DO = [
|
| 42 |
+
"நீங்கள் என்ன செய்ய முடியும்",
|
| 43 |
+
"நீ என்ன செய்ய முடியும்",
|
| 44 |
+
"உங்களுக்கு என்ன செய்ய முடியும்",
|
| 45 |
+
]
|
| 46 |
+
|
| 47 |
+
EN_GREET = {"hi", "hello"}
|
| 48 |
+
EN_THANKS = {"thanks", "thank you"}
|
| 49 |
+
EN_HOW_ARE_YOU = {"how are you"}
|
| 50 |
+
|
| 51 |
+
|
| 52 |
+
def _norm(text: str) -> str:
|
| 53 |
+
text = unicodedata.normalize("NFC", str(text))
|
| 54 |
+
text = re.sub(r"\s+", " ", text).strip()
|
| 55 |
+
text = text.strip(" ,.!?;:-")
|
| 56 |
+
return text
|
| 57 |
+
|
| 58 |
+
|
| 59 |
+
def _is_only_smalltalk(t: str, phrases: list[str]) -> bool:
|
| 60 |
+
for p in phrases:
|
| 61 |
+
p_norm = _norm(p)
|
| 62 |
+
if t == p_norm:
|
| 63 |
+
return True
|
| 64 |
+
if t.startswith(p_norm):
|
| 65 |
+
rest = t[len(p_norm):].strip(" ,.!?;:-")
|
| 66 |
+
if not rest:
|
| 67 |
+
return True
|
| 68 |
+
return False
|
| 69 |
+
|
| 70 |
+
|
| 71 |
+
def _matches_any_normalized(t: str, phrases: list[str]) -> bool:
|
| 72 |
+
normalized_phrases = {_norm(p) for p in phrases}
|
| 73 |
+
return t in normalized_phrases
|
| 74 |
+
|
| 75 |
+
|
| 76 |
+
def detect_smalltalk(text: str, lang: str):
|
| 77 |
+
t = _norm(text)
|
| 78 |
+
tl = t.lower()
|
| 79 |
+
|
| 80 |
+
if lang == "si":
|
| 81 |
+
if _is_only_smalltalk(t, SI_GREET):
|
| 82 |
+
return "greet"
|
| 83 |
+
if _is_only_smalltalk(t, SI_THANKS):
|
| 84 |
+
return "thanks"
|
| 85 |
+
if _matches_any_normalized(t, SI_HOW_ARE_YOU):
|
| 86 |
+
return "how_are_you"
|
| 87 |
+
if _matches_any_normalized(t, SI_WHO_ARE_YOU):
|
| 88 |
+
return "who_are_you"
|
| 89 |
+
if _matches_any_normalized(t, SI_WHAT_CAN_YOU_DO):
|
| 90 |
+
return "what_can_you_do"
|
| 91 |
+
|
| 92 |
+
if lang == "ta":
|
| 93 |
+
if _is_only_smalltalk(t, TA_GREET):
|
| 94 |
+
return "greet"
|
| 95 |
+
if _is_only_smalltalk(t, TA_THANKS):
|
| 96 |
+
return "thanks"
|
| 97 |
+
if _matches_any_normalized(t, TA_HOW_ARE_YOU):
|
| 98 |
+
return "how_are_you"
|
| 99 |
+
if _matches_any_normalized(t, TA_WHO_ARE_YOU):
|
| 100 |
+
return "who_are_you"
|
| 101 |
+
if _matches_any_normalized(t, TA_WHAT_CAN_YOU_DO):
|
| 102 |
+
return "what_can_you_do"
|
| 103 |
+
|
| 104 |
+
if tl in EN_GREET:
|
| 105 |
+
return "greet"
|
| 106 |
+
if tl in EN_THANKS:
|
| 107 |
+
return "thanks"
|
| 108 |
+
if tl in EN_HOW_ARE_YOU:
|
| 109 |
+
return "how_are_you"
|
| 110 |
+
|
| 111 |
+
return None
|
| 112 |
+
|
| 113 |
+
|
| 114 |
+
def smalltalk_reply(kind: str, lang: str):
|
| 115 |
+
if lang == "si":
|
| 116 |
+
if kind == "greet":
|
| 117 |
+
return "ආයුබෝවන්! ඔබට පොල් වගාව ගැන මොනවාද අහන්න තියෙන්නේ?"
|
| 118 |
+
if kind == "thanks":
|
| 119 |
+
return "ස්තුතියි! තවත් ප්රශ්නයක් තිබ්බොත් අහන්න."
|
| 120 |
+
if kind == "how_are_you":
|
| 121 |
+
return "මම හොඳින්. ඔබට පොල් වගාව සම්බන්ධ ප්රශ්නයක් ඇත්නම් අහන්න."
|
| 122 |
+
if kind == "who_are_you":
|
| 123 |
+
return "මම පොල් වගාව සම්බන්ධ උපදෙස් ලබාදෙන සහායක පද්ධතියක්."
|
| 124 |
+
if kind == "what_can_you_do":
|
| 125 |
+
return "මට පොල් වගාව, රෝග, පළිබෝධ, පොහොර සහ වගා කළමනාකරණය සම්බන්ධ උපදෙස් ලබාදිය හැක."
|
| 126 |
+
return "කරුණාකර පොල් වගාවට අදාළ ප්රශ්නයක් අහන්න."
|
| 127 |
+
|
| 128 |
+
if lang == "ta":
|
| 129 |
+
if kind == "greet":
|
| 130 |
+
return "வணக்கம்! தேங்காய் சாகுபடி பற்றி என்ன கேட்க விரும்புகிறீர்கள்?"
|
| 131 |
+
if kind == "thanks":
|
| 132 |
+
return "நன்றி! இன்னும் கேள்விகள் இருந்தால் கேளுங்கள்."
|
| 133 |
+
if kind == "how_are_you":
|
| 134 |
+
return "நான் நலமாக இருக்கிறேன். தென்னைச் சாகுபடி தொடர்பான கேள்வி இருந்தால் கேளுங்கள்."
|
| 135 |
+
if kind == "who_are_you":
|
| 136 |
+
return "நான் தென்னைச் சாகுபடி தொடர்பான ஆலோசனைகளை வழங்கும் உதவி அமைப்பு."
|
| 137 |
+
if kind == "what_can_you_do":
|
| 138 |
+
return "நான் தென்னைச் சாகுபடி, நோய்கள், பூச்சிகள், உரம் மற்றும் மேலாண்மை தொடர்பான ஆலோசனைகளை வழங்க முடியும்."
|
| 139 |
+
return "தயவுசெய்து தென்னைச் சாகுபடி தொடர்பான கேள்வியை கேளுங்கள்."
|
| 140 |
+
|
| 141 |
+
return ""
|