refactor(libvideo2x): convert the video processor into a class (#1246)

Signed-off-by: k4yt3x <i@k4yt3x.com>

include/libvideo2x/libvideo2x.h

@@ -1,17 +1,20 @@
 #ifndef LIBVIDEO2X_H
 #define LIBVIDEO2X_H
 
-#include <filesystem>
-#include <variant>
-#include <vector>
+#include <atomic>
+#include <cstdint>
+#include <memory>
 
 extern "C" {
 #include <libavcodec/avcodec.h>
 #include <libavformat/avformat.h>
 }
 
-#include "fsutils.h"
+#include "avutils.h"
+#include "decoder.h"
+#include "encoder.h"
 #include "logging.h"
+#include "processor.h"
 
 #ifdef _WIN32
 #ifdef LIBVIDEO2X_EXPORTS
@@ -23,105 +26,75 @@ extern "C" {
 #define LIBVIDEO2X_API
 #endif
 
-enum class ProcessingMode {
-    Filter,
-    Interpolate,
-};
-
-enum class ProcessorType {
-    Libplacebo,
-    RealESRGAN,
-    RIFE,
-};
-
-struct LibplaceboConfig {
-    StringType shader_path;
-};
-
-struct RealESRGANConfig {
-    bool tta_mode;
-    StringType model_name;
-};
-
-struct RIFEConfig {
-    bool tta_mode;
-    bool tta_temporal_mode;
-    bool uhd_mode;
-    int num_threads;
-    StringType model_name;
-};
-
-// Unified filter configuration
-struct ProcessorConfig {
-    ProcessorType processor_type;
-    int width;
-    int height;
-    int scaling_factor;
-    int frm_rate_mul;
-    float scn_det_thresh;
-    std::variant<LibplaceboConfig, RealESRGANConfig, RIFEConfig> config;
-};
-
-// Encoder configurations
-struct EncoderConfig {
-    // Non-AVCodecContext options
-    AVCodecID codec;
-    bool copy_streams;
-
-    // Basic video options
-    int width;
-    int height;
-    AVPixelFormat pix_fmt;
-
-    // Rate control and compression
-    int64_t bit_rate;
-    int rc_buffer_size;
-    int rc_min_rate;
-    int rc_max_rate;
-    int qmin;
-    int qmax;
-
-    // GOP and frame structure
-    int gop_size;
-    int max_b_frames;
-    int keyint_min;
-    int refs;
-
-    // Performance and threading
-    int thread_count;
-
-    // Latency and buffering
-    int delay;
-
-    // Extra AVOptions
-    std::vector<std::pair<StringType, StringType>> extra_opts;
-};
-
 struct HardwareConfig {
     uint32_t vk_device_index;
     AVHWDeviceType hw_device_type;
 };
 
-// Video processing context
-struct VideoProcessingContext {
-    int64_t processed_frames;
-    int64_t total_frames;
-    std::time_t start_time;
-    bool pause;
-    bool abort;
-    bool completed;
+class LIBVIDEO2X_API VideoProcessor {
+   public:
+    VideoProcessor(
+        const HardwareConfig hw_cfg,
+        const ProcessorConfig proc_cfg,
+        EncoderConfig enc_cfg,
+        Video2xLogLevel = Video2xLogLevel::Info,
+        bool benchmark = false
+    );
+
+    virtual ~VideoProcessor() = default;
+
+    [[nodiscard]] int
+    process(const std::filesystem::path in_fname, const std::filesystem::path out_fname);
+
+    void pause() { paused_.store(true); }
+    void resume() { paused_.store(false); }
+    void abort() { aborted_.store(true); }
+
+    int64_t get_processed_frames() const { return frame_index_.load(); }
+    int64_t get_total_frames() const { return total_frames_.load(); }
+
+    bool is_paused() const { return paused_.load(); }
+    bool is_aborted() const { return aborted_.load(); }
+    bool is_completed() const { return completed_.load(); }
+
+   private:
+    [[nodiscard]] int
+    process_frames(Decoder &decoder, Encoder &encoder, std::unique_ptr<Processor> &processor);
+
+    [[nodiscard]] int write_frame(AVFrame *frame, Encoder &encoder);
+
+    [[nodiscard]] inline int write_raw_packet(
+        AVPacket *packet,
+        AVFormatContext *ifmt_ctx,
+        AVFormatContext *ofmt_ctx,
+        int *stream_map
+    );
+
+    [[nodiscard]] inline int process_filtering(
+        std::unique_ptr<Processor> &processor,
+        Encoder &encoder,
+        AVFrame *frame,
+        AVFrame *proc_frame
+    );
+
+    [[nodiscard]] inline int process_interpolation(
+        std::unique_ptr<Processor> &processor,
+        Encoder &encoder,
+        std::unique_ptr<AVFrame, decltype(&av_frame_deleter)> &prev_frame,
+        AVFrame *frame,
+        AVFrame *proc_frame
+    );
+
+    HardwareConfig hw_cfg_;
+    ProcessorConfig proc_cfg_;
+    EncoderConfig enc_cfg_;
+    bool benchmark_ = false;
+
+    std::atomic<int64_t> frame_index_ = 0;
+    std::atomic<int64_t> total_frames_ = 0;
+    std::atomic<bool> paused_ = false;
+    std::atomic<bool> aborted_ = false;
+    std::atomic<bool> completed_ = false;
 };
 
-// Process a video file using the specified configurations
-[[nodiscard]] LIBVIDEO2X_API int process_video(
-    const std::filesystem::path in_fname,
-    const std::filesystem::path out_fname,
-    const HardwareConfig hw_cfg,
-    const ProcessorConfig proc_cfg,
-    EncoderConfig enc_cfg,
-    VideoProcessingContext *proc_ctx,
-    Libvideo2xLogLevel log_level,
-    bool benchmark
-);
-
 #endif  // LIBVIDEO2X_H