"""Fetch licensed nature field recordings; no login or API key required."""
from concurrent.futures import ThreadPoolExecutor
from pathlib import Path
import re
import subprocess
import urllib.parse
import urllib.request

ROOT = Path(__file__).resolve().parents[2]
DEST = ROOT / "tmp/surface_audio_sources"
SOURCES = {
    "wind.wav": ("https://opengameart.org/content/park-ambiences", "park_ambience_wind.wav"),
    "leaves.7z": ("https://opengameart.org/content/20-rustles-dry-leaves", "qubodup-rustle.7z"),
    "birds.ogg": ("https://opengameart.org/content/ambient-bird-sounds", "birds-isaiah658.ogg"),
    "fauna.7z": ("https://opengameart.org/content/ambient-bird-cricket-and-frog", "amb_bird_cricket_frog.7z"),
}

def get(url):
    request = urllib.request.Request(url,headers={"User-Agent":"Randoloot nature audio preparation"})
    with urllib.request.urlopen(request,timeout=120) as response: return response.read()

def fetch(entry):
    name,(page,filename) = entry
    path = DEST/name
    if not path.exists():
        html = get(page).decode("utf-8")
        links = re.findall(r'href=[\"\x27]([^\"\x27]+)[\"\x27]',html)
        matches = [urllib.parse.urljoin(page,url) for url in links if urllib.parse.unquote(url).endswith(filename)]
        if name == "birds.ogg" and not matches:
            matches = [urllib.parse.urljoin(page,url) for url in links if "birds-isaiah658" in url and url.endswith(".ogg")]
        if not matches: raise ValueError("Missing published file: "+filename)
        path.write_bytes(get(matches[0]))
    if name.endswith(".7z"):
        folder = DEST/path.stem
        folder.mkdir(exist_ok=True)
        names = subprocess.check_output(["tar","-tf",str(path)],text=True).splitlines()
        for member in names:
            if not (folder/member).resolve().is_relative_to(folder.resolve()): raise ValueError("Unsafe archive path")
        subprocess.run(["tar","-xf",str(path),"-C",str(folder)],check=True)
        return name,names
    return name,path.stat().st_size

if __name__ == "__main__":
    DEST.mkdir(parents=True,exist_ok=True)
    (DEST/".gdignore").write_text("")
    with ThreadPoolExecutor(max_workers=4) as pool:
        for result in pool.map(fetch,SOURCES.items()): print(result)
    wood = DEST/"wood_creak.mp3"
    if not wood.exists():
        page = get("https://freesound.org/people/Rudmer_Rotteveel/sounds/506662/").decode()
        previews = re.findall(r'https://cdn\.freesound\.org/previews/[^\s\"<>]+?-hq\.mp3',page)
        if not previews: raise ValueError("Missing public wood-creak preview")
        wood.write_bytes(get(previews[0]))
    print("wood_creak.mp3",wood.stat().st_size)