Spaces:
Running
Running
Commit ·
37aecec
1
Parent(s): 7392299
wiki: add ai-popup-and-rag-status concept, update log/index/overview for 2026-05-01 changes
Browse files- analyze_headings.py +60 -0
- webapp/_inspect_db.py +39 -0
analyze_headings.py
ADDED
|
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from tipitaka_query import TipitakaDB, clean_text
|
| 2 |
+
import re
|
| 3 |
+
from collections import Counter
|
| 4 |
+
|
| 5 |
+
def analyze_subheadings():
|
| 6 |
+
print("กำลังสแกนวิเคราะห์เนื้อหาจากทั้ง 45 เล่ม... (อาจใช้เวลา 1-2 นาที)")
|
| 7 |
+
heading_counter = Counter()
|
| 8 |
+
last_word_counter = Counter()
|
| 9 |
+
|
| 10 |
+
# Pattern ดักจับเลขข้อ (เหมือนใน build_rag)
|
| 11 |
+
ref_re = re.compile(r'\[[\u0E50-\u0E590-9]+\]')
|
| 12 |
+
|
| 13 |
+
# ใช้ TipitakaDB ดึงข้อมูลแบบเดียวกับระบบ RAG ของจริง
|
| 14 |
+
with TipitakaDB("tipitaka_mcu.db") as db:
|
| 15 |
+
for vol in range(1, 46):
|
| 16 |
+
toc = db.get_toc(vol)
|
| 17 |
+
if not toc: continue
|
| 18 |
+
|
| 19 |
+
vol_info = db.get_volume_info(vol)
|
| 20 |
+
max_page = vol_info["total_pages"]
|
| 21 |
+
|
| 22 |
+
for i, section in enumerate(toc):
|
| 23 |
+
start_page = section.page_number
|
| 24 |
+
end_page = toc[i + 1].page_number - 1 if i + 1 < len(toc) else max_page
|
| 25 |
+
|
| 26 |
+
try:
|
| 27 |
+
# ดึงข้อความมา Clean เพื่อขจัด Noise แบบเดียวกับ Data Pipeline
|
| 28 |
+
raw = db.get_section_text(vol, start_page, end_page, for_tts=False)
|
| 29 |
+
text = clean_text(raw, remove_line_numbers=True, remove_footnotes=True)
|
| 30 |
+
except Exception:
|
| 31 |
+
continue
|
| 32 |
+
|
| 33 |
+
if not text: continue
|
| 34 |
+
|
| 35 |
+
for line in text.split('\n'):
|
| 36 |
+
l = line.strip()
|
| 37 |
+
|
| 38 |
+
# กรอง: ความยาว 2-40 และใช้ Regex เช็คอักษรไทย ป้องกันอักขระล่องหน
|
| 39 |
+
if 2 <= len(l) <= 40 and re.search(r'[ก-๙]', l):
|
| 40 |
+
# ต้องไม่มีเลขข้อในบรรทัด และไม่ลงท้ายด้วยเครื่องหมายจบประโยค
|
| 41 |
+
if not ref_re.search(l) and not l.endswith(('”', '"', '.', '!', '?', 'ๆ', 'ฯ')):
|
| 42 |
+
heading_counter[l] += 1
|
| 43 |
+
|
| 44 |
+
# หาคำลงท้าย (Suffix) ที่เป็นภาษาไทยล้วน ตัดตัวเลขทิ้ง (เช่น "วิชชา ๓" -> "วิชชา")
|
| 45 |
+
thai_words = [re.sub(r'[^ก-๙]', '', w) for w in l.split()]
|
| 46 |
+
thai_words = [w for w in thai_words if len(w) >= 2]
|
| 47 |
+
|
| 48 |
+
if thai_words:
|
| 49 |
+
last_word_counter[thai_words[-1]] += 1
|
| 50 |
+
|
| 51 |
+
print("\n=== Top 50 ประโยคหัวข้อที่พบบ่อย ===")
|
| 52 |
+
for h, c in heading_counter.most_common(50):
|
| 53 |
+
print(f"{c:5d} ครั้ง : {h}")
|
| 54 |
+
|
| 55 |
+
print("\n=== Top 50 คำลงท้าย (Suffix) ที่พบบ่อย ===")
|
| 56 |
+
for w, c in last_word_counter.most_common(50):
|
| 57 |
+
print(f"{c:5d} ครั้ง : {w}")
|
| 58 |
+
|
| 59 |
+
if __name__ == "__main__":
|
| 60 |
+
analyze_subheadings()
|
webapp/_inspect_db.py
ADDED
|
@@ -0,0 +1,39 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import sqlite3
|
| 2 |
+
|
| 3 |
+
conn = sqlite3.connect(r"f:\_Ai\Tipitaka-AI-Expert\RAG\tipitaka_mcu.db")
|
| 4 |
+
cursor = conn.cursor()
|
| 5 |
+
|
| 6 |
+
# Volumes
|
| 7 |
+
print("=== VOLUMES ===")
|
| 8 |
+
cursor.execute("SELECT * FROM volumes ORDER BY volume_number LIMIT 5")
|
| 9 |
+
for row in cursor.fetchall():
|
| 10 |
+
print(f" id={row[0]}, vol={row[1]}, title={row[2]}, pitaka={row[3]}, pages={row[4]}")
|
| 11 |
+
|
| 12 |
+
cursor.execute("SELECT COUNT(*) FROM volumes")
|
| 13 |
+
print(f"\nTotal volumes: {cursor.fetchone()[0]}")
|
| 14 |
+
|
| 15 |
+
# Pages sample
|
| 16 |
+
print("\n=== PAGES (sample vol_id=1) ===")
|
| 17 |
+
cursor.execute("SELECT id, volume_id, page_number, LENGTH(content_text), LENGTH(content_html) FROM pages WHERE volume_id=1 LIMIT 3")
|
| 18 |
+
for row in cursor.fetchall():
|
| 19 |
+
print(f" id={row[0]}, vol_id={row[1]}, page={row[2]}, text_len={row[3]}, html_len={row[4]}")
|
| 20 |
+
|
| 21 |
+
cursor.execute("SELECT COUNT(*) FROM pages")
|
| 22 |
+
print(f"Total pages: {cursor.fetchone()[0]}")
|
| 23 |
+
|
| 24 |
+
# Contents (TOC)
|
| 25 |
+
print("\n=== CONTENTS/TOC (sample vol_id=1) ===")
|
| 26 |
+
cursor.execute("SELECT * FROM contents WHERE volume_id=1 ORDER BY page_number LIMIT 5")
|
| 27 |
+
for row in cursor.fetchall():
|
| 28 |
+
print(f" id={row[0]}, vol_id={row[1]}, title={row[2][:60]}, page={row[3]}, level={row[4]}")
|
| 29 |
+
|
| 30 |
+
cursor.execute("SELECT COUNT(*) FROM contents")
|
| 31 |
+
print(f"Total TOC entries: {cursor.fetchone()[0]}")
|
| 32 |
+
|
| 33 |
+
# Pitaka distribution
|
| 34 |
+
print("\n=== PITAKA DISTRIBUTION ===")
|
| 35 |
+
cursor.execute("SELECT pitaka, COUNT(*), SUM(total_pages) FROM volumes GROUP BY pitaka")
|
| 36 |
+
for row in cursor.fetchall():
|
| 37 |
+
print(f" {row[0]}: {row[1]} volumes, {row[2]} pages")
|
| 38 |
+
|
| 39 |
+
conn.close()
|