#!/usr/bin/env python3
"""Sanity-check a rekordbox XML before importing it.

Verifies the file is well-formed, that every Location resolves to a file that
actually exists, that TrackIDs are unique, and that no playlist references a
track missing from the collection.

Usage:
    python validate_rekordbox_xml.py rekordbox.xml
"""
import argparse
import os
import random
import sys
import xml.etree.ElementTree as ET
from urllib.parse import unquote


def to_path(loc):
    p = unquote(loc.replace("file://localhost", ""))
    if os.name == "nt" and p.startswith("/"):
        p = p[1:]
    return p.replace("/", os.sep)


def main():
    ap = argparse.ArgumentParser(description="Validate a rekordbox XML.")
    ap.add_argument("xml")
    ap.add_argument("--samples", type=int, default=5,
                    help="how many random tracks to print (default: 5)")
    args = ap.parse_args()

    try:
        root = ET.parse(args.xml).getroot()
    except ET.ParseError as e:
        sys.exit(f"FAIL - not well-formed XML: {e}")
    print("well-formed XML          : OK")

    coll = root.find("COLLECTION")
    if coll is None:
        sys.exit("FAIL - no COLLECTION element")
    tracks = coll.findall("TRACK")
    declared = int(coll.get("Entries", 0))
    print(f"collection entries       : {len(tracks)} "
          f"(attribute says {declared}) "
          f"{'OK' if declared == len(tracks) else '<-- MISMATCH'}")

    ids = {t.get("TrackID") for t in tracks}
    print(f"unique TrackIDs          : {len(ids)} "
          f"{'OK' if len(ids) == len(tracks) else '<-- DUPLICATES'}")

    missing = [to_path(t.get("Location")) for t in tracks
               if not os.path.exists(to_path(t.get("Location")))]
    print(f"files not found on disk  : {len(missing)}")
    for m in missing[:10]:
        print("   MISSING:", m)

    print(f"\nwithout title            : {sum(1 for t in tracks if not t.get('Name'))}")
    print(f"without artist           : {sum(1 for t in tracks if not t.get('Artist'))}")
    print(f"without BPM              : {sum(1 for t in tracks if not t.get('AverageBpm'))}")
    print(f"zero length              : "
          f"{sum(1 for t in tracks if t.get('TotalTime') in ('0', '', None))}")

    pl = root.find("PLAYLISTS")
    stats = {"pl": 0, "folder": 0, "refs": 0, "bad": 0, "mismatch": []}

    def walk(n):
        if n.get("Type") == "1":
            stats["pl"] += 1
            kids = n.findall("TRACK")
            stats["refs"] += len(kids)
            if int(n.get("Entries", -1)) != len(kids):
                stats["mismatch"].append(n.get("Name"))
            for k in kids:
                if k.get("Key") not in ids:
                    stats["bad"] += 1
        else:
            stats["folder"] += 1
            for c in n.findall("NODE"):
                walk(c)

    if pl is not None and pl.find("NODE") is not None:
        walk(pl.find("NODE"))
    print(f"\nplaylists                : {stats['pl']}")
    print(f"folder nodes             : {stats['folder']}")
    print(f"track references         : {stats['refs']}")
    print(f"broken references        : {stats['bad']}")
    for m in stats["mismatch"][:10]:
        print(f"   Entries mismatch in playlist {m!r}")

    if tracks and args.samples:
        print("\n--- random sample ---")
        for t in random.sample(tracks, min(args.samples, len(tracks))):
            print(f"  {(t.get('Name') or '')[:40]:42} | "
                  f"{(t.get('Artist') or '')[:20]:22} | "
                  f"{t.get('TotalTime'):>5}s | "
                  f"BPM {t.get('AverageBpm') or '-':>6} | {t.get('Kind')}")

    ok = (not missing and not stats["bad"] and declared == len(tracks)
          and len(ids) == len(tracks))
    print(f"\nRESULT: {'PASS' if ok else 'PROBLEMS FOUND'}")
    return 0 if ok else 1


if __name__ == "__main__":
    sys.exit(main())