Create qa_store.py
Browse files- qa_store.py +32 -0
qa_store.py
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# qa_store.py
|
| 2 |
+
from typing import List, Dict, Any
|
| 3 |
+
import re
|
| 4 |
+
|
| 5 |
+
# Textbook chunks
|
| 6 |
+
ENTRIES: List[Dict[str, Any]] = []
|
| 7 |
+
RAW_KNOWLEDGE: str = ""
|
| 8 |
+
|
| 9 |
+
# QA from textbook JSONL
|
| 10 |
+
AUTO_QA_KNOWLEDGE: List[Dict[str, Any]] = []
|
| 11 |
+
|
| 12 |
+
# Manual QA managed by teacher
|
| 13 |
+
MANUAL_QA_LIST: List[Dict[str, Any]] = []
|
| 14 |
+
MANUAL_QA_INDEX: Dict[str, Dict[str, Any]] = {}
|
| 15 |
+
|
| 16 |
+
# Combined index for fast lookup
|
| 17 |
+
QA_INDEX: Dict[str, str] = {}
|
| 18 |
+
ALL_QA_KNOWLEDGE: List[Dict[str, Any]] = []
|
| 19 |
+
|
| 20 |
+
# Counter for new manual IDs
|
| 21 |
+
NEXT_MANUAL_ID: int = 1
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
def normalize_question(q: str) -> str:
|
| 25 |
+
"""
|
| 26 |
+
Normalize Lao/English question text for matching.
|
| 27 |
+
Lowercase + remove punctuation + collapse spaces.
|
| 28 |
+
"""
|
| 29 |
+
q = (q or "").lower()
|
| 30 |
+
q = re.sub(r"[?!οΌοΌ\.\,\:\;\"ββ'ββ]", " ", q)
|
| 31 |
+
q = re.sub(r"\s+", " ", q)
|
| 32 |
+
return q.strip()
|