vibesonic/internal/coverart/coverart.go

// Package coverart discovers cover art (embedded artwork or folder
// images) and caches resized copies on disk, so the same image is never
// re-resized for every request.
package coverart

import (
	"context"
	"crypto/sha256"
	"encoding/hex"
	"fmt"
	"image"
	_ "image/gif"
	"image/jpeg"
	_ "image/png"
	"io"
	"os"
	"os/exec"
	"path/filepath"

	"golang.org/x/image/draw"
)

// maxDecodePixels bounds the decoded pixel count of a source image so a
// small, highly-compressed file (a "decompression bomb") can't force an
// enormous in-memory pixel buffer allocation. 40 megapixels comfortably
// covers any legitimate cover art.
const maxDecodePixels = 40_000_000

// commonFolderImageNames are checked, in priority order, for a
// folder-level cover image when a file has no embedded artwork.
var commonFolderImageNames = []string{
	"cover.jpg", "cover.jpeg", "cover.png",
	"folder.jpg", "folder.jpeg", "folder.png",
	"front.jpg", "front.jpeg", "front.png",
	"albumart.jpg", "albumart.png",
}

// FindFolderImage returns the first common cover image filename found
// directly inside dir, or "" if none exists.
func FindFolderImage(dir string) string {
	for _, name := range commonFolderImageNames {
		path := filepath.Join(dir, name)
		if info, err := os.Stat(path); err == nil && !info.IsDir() {
			return path
		}
	}
	return ""
}

// Store manages the on-disk resized-image cache under cacheDir.
type Store struct {
	CacheDir       string
	FFmpegPath     string
	MaxSourceBytes int64 // guard against absurd source images before decoding
}

func New(cacheDir, ffmpegPath string) *Store {
	if ffmpegPath == "" {
		ffmpegPath = "ffmpeg"
	}
	return &Store{CacheDir: cacheDir, FFmpegPath: ffmpegPath, MaxSourceBytes: 32 * 1024 * 1024}
}

// ExtractEmbedded pulls the attached-picture stream out of a media file
// into a temporary file using ffmpeg, returning its path. The caller owns
// cleanup.
func (s *Store) ExtractEmbedded(ctx context.Context, mediaPath string, streamIndex int) (string, error) {
	tmp, err := os.CreateTemp("", "vibesonic-art-*.img")
	if err != nil {
		return "", err
	}
	tmp.Close()

	cmd := exec.CommandContext(ctx, s.FFmpegPath,
		"-y", "-v", "error",
		"-i", mediaPath,
		"-map", fmt.Sprintf("0:%d", streamIndex),
		"-frames:v", "1",
		"-c:v", "copy",
		// Temporary files deliberately have a neutral .img suffix. Tell
		// ffmpeg which muxer to use instead of asking it to infer one from
		// that suffix. ffmpeg does not accept `--` as an end-of-options
		// marker, so including it here made every embedded-art request fail.
		"-f", "image2", tmp.Name(),
	)
	if out, err := cmd.CombinedOutput(); err != nil {
		os.Remove(tmp.Name())
		return "", fmt.Errorf("ffmpeg extract art: %w: %s", err, string(out))
	}
	return tmp.Name(), nil
}

// CachedPath returns the on-disk path for a resized variant of the image
// identified by contentHash at the given target size, without generating
// it.
func (s *Store) CachedPath(contentHash string, size int) string {
	return filepath.Join(s.CacheDir, fmt.Sprintf("%s_%d.jpg", contentHash, size))
}

// GetOrCreate returns a reader-ready path to a JPEG resized so its
// longest side is at most `size` pixels, generating and caching it from
// sourcePath on first request. size <= 0 returns the original image
// unresized (still copied through the cache so repeat requests are
// cheap).
func (s *Store) GetOrCreate(sourcePath, contentHash string, size int) (string, error) {
	if err := os.MkdirAll(s.CacheDir, 0o755); err != nil {
		return "", err
	}
	cached := s.CachedPath(contentHash, size)
	if _, err := os.Stat(cached); err == nil {
		return cached, nil
	}

	if info, err := os.Stat(sourcePath); err != nil {
		return "", err
	} else if s.MaxSourceBytes > 0 && info.Size() > s.MaxSourceBytes {
		return "", fmt.Errorf("source image %s exceeds %d byte limit", sourcePath, s.MaxSourceBytes)
	}

	src, err := os.Open(sourcePath)
	if err != nil {
		return "", err
	}
	defer src.Close()

	if cfg, _, err := image.DecodeConfig(src); err == nil {
		if cfg.Width*cfg.Height > maxDecodePixels {
			return "", fmt.Errorf("source image %s exceeds %d pixel limit (%dx%d)", sourcePath, maxDecodePixels, cfg.Width, cfg.Height)
		}
	}
	if _, err := src.Seek(0, io.SeekStart); err != nil {
		return "", err
	}

	img, _, err := image.Decode(src)
	if err != nil {
		return "", fmt.Errorf("decoding image: %w", err)
	}

	resized := img
	if size > 0 {
		resized = resize(img, size)
	}

	tmpPath := cached + ".tmp"
	out, err := os.Create(tmpPath)
	if err != nil {
		return "", err
	}
	if err := jpeg.Encode(out, resized, &jpeg.Options{Quality: 85}); err != nil {
		out.Close()
		os.Remove(tmpPath)
		return "", err
	}
	if err := out.Close(); err != nil {
		return "", err
	}
	if err := os.Rename(tmpPath, cached); err != nil {
		return "", err
	}
	return cached, nil
}

// resize scales img so its longest side is at most maxSize pixels,
// preserving aspect ratio. Images already smaller are left unscaled.
func resize(img image.Image, maxSize int) image.Image {
	b := img.Bounds()
	w, h := b.Dx(), b.Dy()
	if w <= maxSize && h <= maxSize {
		return img
	}
	var newW, newH int
	if w >= h {
		newW = maxSize
		newH = h * maxSize / w
	} else {
		newH = maxSize
		newW = w * maxSize / h
	}
	if newW < 1 {
		newW = 1
	}
	if newH < 1 {
		newH = 1
	}
	dst := image.NewRGBA(image.Rect(0, 0, newW, newH))
	draw.CatmullRom.Scale(dst, dst.Bounds(), img, b, draw.Over, nil)
	return dst
}

// ContentHashOf returns a stable cache key for a file's current content,
// cheap enough to compute per-request by hashing size+mtime rather than
// the full file body.
func ContentHashOf(path string) (string, error) {
	f, err := os.Open(path)
	if err != nil {
		return "", err
	}
	defer f.Close()
	info, err := f.Stat()
	if err != nil {
		return "", err
	}
	h := sha256.New()
	fmt.Fprintf(h, "%s-%d-%d", path, info.Size(), info.ModTime().Unix())
	return hex.EncodeToString(h.Sum(nil))[:16], nil
}