velody/apps/apple/VelodyMac/Sources/LocalMusicScanner.swift

import AVFoundation
import CryptoKit
import Foundation
import VelodyPersistence

final class AVFoundationMetadataReader: MetadataReader {
    func readMetadata(for fileURL: URL) async throws -> LocalTrackMetadata {
        let asset = AVURLAsset(url: fileURL)
        let commonMetadata = try await asset.load(.commonMetadata)
        let duration = try? await asset.load(.duration)

        return LocalTrackMetadata(
            title: commonMetadata.firstStringValue(for: .commonIdentifierTitle),
            artist: commonMetadata.firstStringValue(for: .commonIdentifierArtist),
            album: commonMetadata.firstStringValue(for: .commonIdentifierAlbumName),
            durationSeconds: duration?.seconds.validDurationSeconds
        )
    }
}

final class FileSystemLocalMusicScanner: LocalMusicScanner {
    private let metadataReader: any MetadataReader
    private let fileHasher: SHA256FileHasher
    private let fileManager: FileManager

    init(
        metadataReader: any MetadataReader,
        fileHasher: SHA256FileHasher = SHA256FileHasher(),
        fileManager: FileManager = .default
    ) {
        self.metadataReader = metadataReader
        self.fileHasher = fileHasher
        self.fileManager = fileManager
    }

    func scanFolder(at folderURL: URL) async throws -> [ScannedLocalTrack] {
        let hasScopedAccess = folderURL.startAccessingSecurityScopedResource()
        defer {
            if hasScopedAccess {
                folderURL.stopAccessingSecurityScopedResource()
            }
        }

        guard let enumerator = fileManager.enumerator(
            at: folderURL,
            includingPropertiesForKeys: [
                .isRegularFileKey,
                .contentModificationDateKey,
            ],
            options: [.skipsHiddenFiles, .skipsPackageDescendants]
        ) else {
            throw CocoaError(.fileReadUnknown)
        }

        var discoveredTracks: [ScannedLocalTrack] = []

        for case let fileURL as URL in enumerator {
            guard shouldScan(fileURL) else {
                continue
            }

            let fallbackTitle = fileURL.deletingPathExtension().lastPathComponent
            let sha256 = try fileHasher.hashFile(at: fileURL)
            let fileModifiedAt = modificationDate(for: fileURL)

            do {
                let metadata = try await metadataReader.readMetadata(for: fileURL)
                discoveredTracks.append(
                    ScannedLocalTrack(
                        title: metadata.title?.trimmedNonEmpty ?? fallbackTitle,
                        artist: metadata.artist?.trimmedNonEmpty ?? "Unknown Artist",
                        album: metadata.album?.trimmedNonEmpty,
                        durationSeconds: metadata.durationSeconds,
                        localFilePath: fileURL.path,
                        sha256: sha256,
                        fileModifiedAt: fileModifiedAt
                    )
                )
            } catch {
                discoveredTracks.append(
                    ScannedLocalTrack(
                        title: fallbackTitle,
                        artist: "Unknown Artist",
                        album: nil,
                        durationSeconds: nil,
                        localFilePath: fileURL.path,
                        sha256: sha256,
                        fileModifiedAt: fileModifiedAt
                    )
                )
            }
        }

        return discoveredTracks.sorted { lhs, rhs in
            let titleOrder = lhs.title.localizedCaseInsensitiveCompare(rhs.title)
            if titleOrder == .orderedSame {
                return lhs.localFilePath.localizedCaseInsensitiveCompare(rhs.localFilePath) == .orderedAscending
            }

            return titleOrder == .orderedAscending
        }
    }

    private func shouldScan(_ fileURL: URL) -> Bool {
        guard fileURL.pathExtension.lowercased() == "mp3" else {
            return false
        }

        let resourceValues = try? fileURL.resourceValues(forKeys: [.isRegularFileKey])
        return resourceValues?.isRegularFile == true
    }

    private func modificationDate(for fileURL: URL) -> Date {
        if let resourceValues = try? fileURL.resourceValues(forKeys: [.contentModificationDateKey]),
           let contentModificationDate = resourceValues.contentModificationDate
        {
            return contentModificationDate
        }

        if let attributes = try? fileManager.attributesOfItem(atPath: fileURL.path),
           let modificationDate = attributes[.modificationDate] as? Date
        {
            return modificationDate
        }

        return .distantPast
    }
}

final class SHA256FileHasher {
    private let bufferSize = 1_048_576

    func hashFile(at fileURL: URL) throws -> String {
        let fileHandle = try FileHandle(forReadingFrom: fileURL)
        defer {
            try? fileHandle.close()
        }

        var hasher = SHA256()

        while true {
            let data = try fileHandle.read(upToCount: bufferSize) ?? Data()
            if data.isEmpty {
                break
            }

            hasher.update(data: data)
        }

        return hasher.finalize().map { String(format: "%02x", $0) }.joined()
    }
}

private extension Array where Element == AVMetadataItem {
    func firstStringValue(for identifier: AVMetadataIdentifier) -> String? {
        AVMetadataItem
            .metadataItems(from: self, filteredByIdentifier: identifier)
            .first?
            .stringValue?
            .trimmedNonEmpty
    }
}

private extension Double {
    var validDurationSeconds: Double? {
        guard isFinite, self > 0 else {
            return nil
        }

        return self
    }
}

private extension String {
    var trimmedNonEmpty: String? {
        let trimmed = trimmingCharacters(in: .whitespacesAndNewlines)
        return trimmed.isEmpty ? nil : trimmed
    }
}