Spaces:
Sleeping
Sleeping
Update entity_dictionary.py
Browse files- entity_dictionary.py +8 -30
entity_dictionary.py
CHANGED
|
@@ -4,18 +4,12 @@
|
|
| 4 |
|
| 5 |
ENTITY_DICTIONARY = {
|
| 6 |
|
| 7 |
-
# ========================================================
|
| 8 |
-
# 1️⃣ DISTRICTS OF MAURITIUS
|
| 9 |
-
# ========================================================
|
| 10 |
"DISTRICT": [
|
| 11 |
"Port Louis", "Pamplemousses", "Riviere du Rempart",
|
| 12 |
"Flacq", "Grand Port", "Savanne", "Moka",
|
| 13 |
"Plaines Wilhems", "Riviere Noire", "Black River"
|
| 14 |
],
|
| 15 |
|
| 16 |
-
# ========================================================
|
| 17 |
-
# 2️⃣ CITIES & TOWNS
|
| 18 |
-
# ========================================================
|
| 19 |
"CITY": [
|
| 20 |
"Curepipe", "Vacoas", "Quatre Bornes",
|
| 21 |
"Beau Bassin", "Rose Hill", "Ebene", "Phoenix",
|
|
@@ -24,9 +18,7 @@ ENTITY_DICTIONARY = {
|
|
| 24 |
"Roche Bois", "Piton", "Triolet"
|
| 25 |
],
|
| 26 |
|
| 27 |
-
|
| 28 |
-
# 3️⃣ VILLAGES
|
| 29 |
-
# ========================================================
|
| 30 |
"VILLAGE": [
|
| 31 |
"Albion", "Amaury", "Amitie-Gokhoola", "Arsenal",
|
| 32 |
"Baie du Tombeau", "Baie du Cap", "Bambous",
|
|
@@ -75,9 +67,7 @@ ENTITY_DICTIONARY = {
|
|
| 75 |
"Tyack", "Union Park", "Verdun", "Ville Bague"
|
| 76 |
],
|
| 77 |
|
| 78 |
-
|
| 79 |
-
# 4️⃣ LANDMARKS & REGIONS
|
| 80 |
-
# ========================================================
|
| 81 |
"LANDMARK": [
|
| 82 |
"Bagatelle", "Caudan Waterfront", "Le Caudan",
|
| 83 |
"Ebene Cybercity", "Mont Choisy", "Chamarel",
|
|
@@ -86,9 +76,7 @@ ENTITY_DICTIONARY = {
|
|
| 86 |
"Plaisance", "Cascavelle", "Medine"
|
| 87 |
],
|
| 88 |
|
| 89 |
-
|
| 90 |
-
# 5️⃣ COMMON MAURITIAN NAMES
|
| 91 |
-
# ========================================================
|
| 92 |
"PERSON": [
|
| 93 |
# Hindu / Indian
|
| 94 |
"Amit", "Vikash", "Rishi", "Kiran", "Anil",
|
|
@@ -112,9 +100,7 @@ ENTITY_DICTIONARY = {
|
|
| 112 |
"Sidharth", "Sameer", "Sameerchand", "Karuna"
|
| 113 |
],
|
| 114 |
|
| 115 |
-
|
| 116 |
-
# 6️⃣ PRODUCTS & VEHICLES
|
| 117 |
-
# ========================================================
|
| 118 |
"PRODUCT": [
|
| 119 |
"Honda Fit", "Honda Vezel", "Suzuki Swift",
|
| 120 |
"Toyota Aqua", "Toyota Vitz", "Nissan Note",
|
|
@@ -122,26 +108,18 @@ ENTITY_DICTIONARY = {
|
|
| 122 |
"OnePlus", "Pixel", "Redmi"
|
| 123 |
],
|
| 124 |
|
| 125 |
-
|
| 126 |
-
# 7️⃣ REGEX PATTERNS
|
| 127 |
-
# ========================================================
|
| 128 |
"REGEX_PATTERNS": {
|
| 129 |
-
# ⏰ TIME must come BEFORE NUMBER
|
| 130 |
"TIME": r"\b\d{1,2}(?::\d{2})?\s?(?:am|pm)\b",
|
| 131 |
-
|
| 132 |
-
# 📅 DATE
|
| 133 |
"DATE": r"\b\d{1,2}[\/\-.]\d{1,2}[\/\-.]\d{2,4}\b",
|
| 134 |
|
| 135 |
-
# 💰 MONEY
|
| 136 |
"MONEY": r"\b(?:Rs|rs|RS)\s?\d+(?:,\d{3})*\b",
|
| 137 |
-
|
| 138 |
-
# 📞 PHONE
|
| 139 |
"PHONE": r"\b\d{8}\b",
|
| 140 |
|
| 141 |
-
# 🔢 NUMBER (LOWEST PRIORITY)
|
| 142 |
"NUMBER": r"\b\d+(?:,\d{3})*(?:\.\d+)?\b",
|
| 143 |
-
|
| 144 |
-
# 📧 EMAIL
|
| 145 |
"EMAIL": r"[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}"
|
| 146 |
}
|
| 147 |
|
|
|
|
| 4 |
|
| 5 |
ENTITY_DICTIONARY = {
|
| 6 |
|
|
|
|
|
|
|
|
|
|
| 7 |
"DISTRICT": [
|
| 8 |
"Port Louis", "Pamplemousses", "Riviere du Rempart",
|
| 9 |
"Flacq", "Grand Port", "Savanne", "Moka",
|
| 10 |
"Plaines Wilhems", "Riviere Noire", "Black River"
|
| 11 |
],
|
| 12 |
|
|
|
|
|
|
|
|
|
|
| 13 |
"CITY": [
|
| 14 |
"Curepipe", "Vacoas", "Quatre Bornes",
|
| 15 |
"Beau Bassin", "Rose Hill", "Ebene", "Phoenix",
|
|
|
|
| 18 |
"Roche Bois", "Piton", "Triolet"
|
| 19 |
],
|
| 20 |
|
| 21 |
+
|
|
|
|
|
|
|
| 22 |
"VILLAGE": [
|
| 23 |
"Albion", "Amaury", "Amitie-Gokhoola", "Arsenal",
|
| 24 |
"Baie du Tombeau", "Baie du Cap", "Bambous",
|
|
|
|
| 67 |
"Tyack", "Union Park", "Verdun", "Ville Bague"
|
| 68 |
],
|
| 69 |
|
| 70 |
+
|
|
|
|
|
|
|
| 71 |
"LANDMARK": [
|
| 72 |
"Bagatelle", "Caudan Waterfront", "Le Caudan",
|
| 73 |
"Ebene Cybercity", "Mont Choisy", "Chamarel",
|
|
|
|
| 76 |
"Plaisance", "Cascavelle", "Medine"
|
| 77 |
],
|
| 78 |
|
| 79 |
+
|
|
|
|
|
|
|
| 80 |
"PERSON": [
|
| 81 |
# Hindu / Indian
|
| 82 |
"Amit", "Vikash", "Rishi", "Kiran", "Anil",
|
|
|
|
| 100 |
"Sidharth", "Sameer", "Sameerchand", "Karuna"
|
| 101 |
],
|
| 102 |
|
| 103 |
+
|
|
|
|
|
|
|
| 104 |
"PRODUCT": [
|
| 105 |
"Honda Fit", "Honda Vezel", "Suzuki Swift",
|
| 106 |
"Toyota Aqua", "Toyota Vitz", "Nissan Note",
|
|
|
|
| 108 |
"OnePlus", "Pixel", "Redmi"
|
| 109 |
],
|
| 110 |
|
| 111 |
+
|
|
|
|
|
|
|
| 112 |
"REGEX_PATTERNS": {
|
|
|
|
| 113 |
"TIME": r"\b\d{1,2}(?::\d{2})?\s?(?:am|pm)\b",
|
| 114 |
+
|
|
|
|
| 115 |
"DATE": r"\b\d{1,2}[\/\-.]\d{1,2}[\/\-.]\d{2,4}\b",
|
| 116 |
|
|
|
|
| 117 |
"MONEY": r"\b(?:Rs|rs|RS)\s?\d+(?:,\d{3})*\b",
|
| 118 |
+
|
|
|
|
| 119 |
"PHONE": r"\b\d{8}\b",
|
| 120 |
|
|
|
|
| 121 |
"NUMBER": r"\b\d+(?:,\d{3})*(?:\.\d+)?\b",
|
| 122 |
+
|
|
|
|
| 123 |
"EMAIL": r"[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}"
|
| 124 |
}
|
| 125 |
|