Files
Aislo/resources/tester/scratch/search_pdf_flexible.py
T

27 lines
1.1 KiB
Python

import pymupdf
import sys
import re
sys.stdout.reconfigure(encoding='utf-8')
doc = pymupdf.open("resources/knowledge/original/행정규칙/임도 품셈 적용기준 (현 산림사업 표준품셈)/첨부/(산림청고시 제2025-82호) 산림사업 표준품셈.pdf")
# Search by regex or partial
print("=== Search for 천공기 휘발유 (F0046) ===")
for pno in range(30, 60):
text = doc[pno].get_text()
if "천공기" in text and ("휘발유" in text or "주연료" in text or "95%" in text):
print(f"Found F0046 on PDF Page {pno + 1}:")
for line in text.splitlines():
if any(k in line for k in ["천공", "주연료", "휘발유", "잡품"]):
print(" ", line)
print("\n=== Search for 작업로 선정 (F0076) & 작업로 설치 (F0077) ===")
for pno in range(45, 65):
text = doc[pno].get_text()
if "작업로" in text and ("선정" in text or "설치" in text or "1km" in text):
print(f"Found on PDF Page {pno + 1}:")
for line in text.splitlines():
if any(k in line for k in ["작업로", "선정", "설치", "km", "소요인력"]):
print(" ", line)