import fitz, base64, os, textwrap, json, math
from pathlib import Path
pdfs = {
"en": "/mnt/data/ENG VOCO MENU.pdf",
"ko": "/mnt/data/KOR VOCO MENU.pdf",
"ru": "/mnt/data/RUS VOCO MENU.pdf",
}
# render pages to jpeg base64 with moderate quality
data = {}
for lang, path in pdfs.items():
doc = fitz.open(path)
imgs=[]
for i, page in enumerate(doc):
pix = page.get_pixmap(matrix=fitz.Matrix(1.5,1.5), alpha=False)
img_bytes = pix.tobytes("jpeg", jpg_quality=78)
b64 = base64.b64encode(img_bytes).decode("ascii")
imgs.append(f"data:image/jpeg;base64,{b64}")
data[lang]=imgs
len(data["en
