Music_Metadata_Enricher/scanner.py

from __future__ import annotations

import re
import sys
from pathlib import Path
from typing import List

from models import AlbumScan, AUDIO_EXTENSIONS, IMAGE_EXTENSIONS, TRACKLIST_EXTENSIONS, PLAYLIST_EXTENSIONS

_DISC_DIR_RE = re.compile(r"(?i)^(?:cd|disc|disk|side)[_ \-]*\d{1,2}$")


def _is_hidden(name: str) -> bool:
    return name.startswith(".") or name.startswith("_")


def _is_disc_dir(name: str) -> bool:
    """True für Ordner wie 'CD1', 'Disc 2', 'Side A', 'Disk_1'."""
    return bool(_DISC_DIR_RE.match(name))


def scan_album(album_dir: Path) -> AlbumScan:
    """
    Scannt ein Album-Verzeichnis.

    Rekursions-Regel:
    - Hat das Album-Verzeichnis selbst Audio-Dateien → kein Abstieg in Unterordner
      (Einzelscheibe; Sub-Ordner wie Artworks, Scans, irrtümliche Kopien werden ignoriert).
    - Hat der Root KEINE Audio-Dateien → Abstieg nur in Disc-Unterordner (CD1, Disc 2 …).
    """
    result = AlbumScan(album_dir=album_dir)

    # Erst nur die Wurzel-Ebene scannen, um zu entscheiden ob rekursiert wird
    root_has_audio = any(
        (album_dir / name).suffix.lower() in AUDIO_EXTENSIONS
        for name in _listdir(album_dir)
        if not _is_hidden(name)
    )

    if root_has_audio:
        # Nur Root-Ebene — keine Unterordner
        _scan_dir(album_dir, album_dir, result, recurse=False)
    else:
        # Kein Audio an der Wurzel → Multi-CD: nur Disc-Unterordner
        _scan_dir(album_dir, album_dir, result, recurse=True)

    result.audio_files.sort()
    result.image_files.sort()
    result.tracklist_files.sort()
    result.playlist_files.sort()
    return result


def _listdir(path: Path) -> List[str]:
    try:
        return [e.name for e in path.iterdir()]
    except (PermissionError, OSError) as e:
        print(f"⚠️ Scan-Fehler: {e}", file=sys.stderr)
        return []


def _scan_dir(current: Path, album_dir: Path, result: AlbumScan, recurse: bool) -> None:
    try:
        entries = sorted(current.iterdir())
    except (PermissionError, OSError) as e:
        print(f"⚠️ Scan-Fehler {current}: {e}", file=sys.stderr)
        return

    for entry in entries:
        name = entry.name
        if _is_hidden(name):
            continue
        if entry.is_dir():
            if recurse and _is_disc_dir(name):
                _scan_dir(entry, album_dir, result, recurse=True)
            # Andere Unterordner (Artworks, irrtümliche Kopien…) werden übersprungen
        elif entry.is_file():
            ext = entry.suffix.lower()
            if ext in AUDIO_EXTENSIONS:
                result.audio_files.append(entry)
            elif ext in IMAGE_EXTENSIONS:
                result.image_files.append(entry)
            elif ext in TRACKLIST_EXTENSIONS:
                result.tracklist_files.append(entry)
            elif ext in PLAYLIST_EXTENSIONS:
                result.playlist_files.append(entry)
            else:
                result.other_files.append(entry)


def collect_album_dirs(root: Path) -> List[Path]:
    """
    Findet rekursiv alle Album-Verzeichnisse unterhalb von root.
    Ein Verzeichnis gilt als Album wenn:
    - es direkt Audio-Dateien enthält, ODER
    - es keine direkten Audio-Dateien hat aber Disc-Unterordner mit Audio (Multi-CD).
    Container-Verzeichnisse ohne Audio werden rekursiv durchsucht.
    """
    result: List[Path] = []
    _find_albums(root, result)
    return result


def _has_local_audio(path: Path) -> bool:
    try:
        return any(
            e.suffix.lower() in AUDIO_EXTENSIONS
            for e in path.iterdir()
            if e.is_file() and not _is_hidden(e.name)
        )
    except (PermissionError, OSError):
        return False


def _find_albums(current: Path, result: List[Path]) -> None:
    try:
        entries = sorted(current.iterdir())
        subdirs = [e for e in entries if e.is_dir() and not _is_hidden(e.name)]
    except (PermissionError, OSError) as e:
        print(f"⚠️ Lesefehler {current}: {e}", file=sys.stderr)
        return

    # Verzeichnis enthält direkt Audio → Album
    if any(e.is_file() and not _is_hidden(e.name) and e.suffix.lower() in AUDIO_EXTENSIONS
           for e in entries):
        result.append(current)
        return

    # Disc-Unterordner mit Audio → Multi-CD-Album (scan_album übernimmt die Disc-Logik)
    disc_dirs = [d for d in subdirs if _is_disc_dir(d.name)]
    if disc_dirs and any(_has_local_audio(d) for d in disc_dirs):
        result.append(current)
        return

    # Container-Verzeichnis → rekursiv weiter suchen
    for subdir in subdirs:
        _find_albums(subdir, result)