import os
import shutil
import re

ROOT = "/opt/ai-os/products/ceo/company_kb"
INGEST = os.path.join(ROOT, "notebooklm_ingest")
KNOWLEDGE = os.path.join(ROOT, "knowledge")

# Create structure
os.makedirs(os.path.join(KNOWLEDGE, "tourism"), exist_ok=True)
os.makedirs(os.path.join(KNOWLEDGE, "policy"), exist_ok=True)

# Find directories
tourism_dir = None
policy_dir = None
for d in os.listdir(INGEST):
    if "Báo cáo Du Lịch" in d:
        tourism_dir = os.path.join(INGEST, d)
    elif "Thạc sĩ Phân tích Chính sách" in d:
        policy_dir = os.path.join(INGEST, d)

def clean_and_copy(src_dir, dest_dir):
    if not src_dir or not os.path.isdir(src_dir):
        print(f"Skipping {src_dir}")
        return
    count = 0
    for fn in os.listdir(src_dir):
        if not fn.endswith(".md") or fn == "_notebook.json":
            continue
        fp = os.path.join(src_dir, fn)
        with open(fp, "r", encoding="utf-8", errors="replace") as f:
            content = f.read()
            
        # Optional: remove NotebookLM metadata block
        # metadata is usually between "- **Notebook:**" and "---"
        # We can just keep it for now but organize them cleanly.
        
        # We might want to shorten filename because some are too long
        new_fn = fn[:100] + ".md" if len(fn) > 100 else fn
        dest_fp = os.path.join(dest_dir, new_fn)
        
        with open(dest_fp, "w", encoding="utf-8") as f:
            f.write(content)
        count += 1
    print(f"Ingested {count} files to {dest_dir}")

clean_and_copy(tourism_dir, os.path.join(KNOWLEDGE, "tourism"))
clean_and_copy(policy_dir, os.path.join(KNOWLEDGE, "policy"))