Extend structures to store updated version of the dependency descriptor

Rename structures to match terminology in the spec

Bug: webrtc:10342
Change-Id: I1329abaca98ae7f82307451032d5ce1533e80772
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/143960
Reviewed-by: Philip Eliasson <philipel@webrtc.org>
Commit-Queue: Danil Chapovalov <danilchap@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#28402}

common_video/generic_frame_descriptor/BUILD.gn

@@ -20,5 +20,6 @@ rtc_source_set("generic_frame_descriptor") {
     "../../rtc_base:checks",
     "//third_party/abseil-cpp/absl/container:inlined_vector",
     "//third_party/abseil-cpp/absl/strings",
+    "//third_party/abseil-cpp/absl/types:optional",
   ]
 }

common_video/generic_frame_descriptor/generic_frame_info.cc

@@ -15,7 +15,7 @@
 
 namespace webrtc {
 
-absl::InlinedVector<GenericFrameInfo::DecodeTargetIndication, 10>
+absl::InlinedVector<DecodeTargetIndication, 10>
 GenericFrameInfo::DecodeTargetInfo(absl::string_view indication_symbols) {
   absl::InlinedVector<DecodeTargetIndication, 10> decode_targets;
   for (char symbol : indication_symbols) {
@@ -67,10 +67,4 @@ GenericFrameInfo::Builder& GenericFrameInfo::Builder::Fdiffs(
   return *this;
 }
 
-TemplateStructure::TemplateStructure() = default;
-TemplateStructure::TemplateStructure(const TemplateStructure&) = default;
-TemplateStructure::TemplateStructure(TemplateStructure&&) = default;
-TemplateStructure& TemplateStructure::operator=(const TemplateStructure&) =
-    default;
-TemplateStructure::~TemplateStructure() = default;
 }  // namespace webrtc

common_video/generic_frame_descriptor/generic_frame_info.h

@@ -16,10 +16,85 @@
 
 #include "absl/container/inlined_vector.h"
 #include "absl/strings/string_view.h"
+#include "absl/types/optional.h"
 #include "api/array_view.h"
 #include "api/video/video_codec_constants.h"
 
 namespace webrtc {
+// Structures to build and parse dependency descriptor as described in
+// https://aomediacodec.github.io/av1-rtp-spec/#dependency-descriptor-rtp-header-extension
+class RenderResolution {
+ public:
+  constexpr RenderResolution() = default;
+  constexpr RenderResolution(int width, int height)
+      : width_(width), height_(height) {}
+  RenderResolution(const RenderResolution&) = default;
+  RenderResolution& operator=(const RenderResolution&) = default;
+
+  friend bool operator==(const RenderResolution& lhs,
+                         const RenderResolution& rhs) {
+    return lhs.width_ == rhs.width_ && lhs.height_ == rhs.height_;
+  }
+
+  constexpr int Width() const { return width_; }
+  constexpr int Height() const { return height_; }
+
+ private:
+  int width_ = 0;
+  int height_ = 0;
+};
+
+// Relationship of a frame to a Decode target.
+enum class DecodeTargetIndication {
+  kNotPresent = 0,   // DecodeTargetInfo symbol '-'
+  kDiscardable = 1,  // DecodeTargetInfo symbol 'D'
+  kSwitch = 2,       // DecodeTargetInfo symbol 'S'
+  kRequired = 3      // DecodeTargetInfo symbol 'R'
+};
+
+struct FrameDependencyTemplate {
+  friend bool operator==(const FrameDependencyTemplate& lhs,
+                         const FrameDependencyTemplate& rhs) {
+    return lhs.spatial_id == rhs.spatial_id &&
+           lhs.temporal_id == rhs.temporal_id &&
+           lhs.decode_target_indications == rhs.decode_target_indications &&
+           lhs.frame_diffs == rhs.frame_diffs &&
+           lhs.chain_diffs == rhs.chain_diffs;
+  }
+
+  int spatial_id = 0;
+  int temporal_id = 0;
+  absl::InlinedVector<DecodeTargetIndication, 10> decode_target_indications;
+  absl::InlinedVector<int, 4> frame_diffs;
+  absl::InlinedVector<int, 4> chain_diffs;
+};
+
+struct FrameDependencyStructure {
+  friend bool operator==(const FrameDependencyStructure& lhs,
+                         const FrameDependencyStructure& rhs) {
+    return lhs.num_decode_targets == rhs.num_decode_targets &&
+           lhs.num_chains == rhs.num_chains &&
+           lhs.decode_target_protected_by_chain ==
+               rhs.decode_target_protected_by_chain &&
+           lhs.resolutions == rhs.resolutions && lhs.templates == rhs.templates;
+  }
+
+  int structure_id = 0;
+  int num_decode_targets = 0;
+  int num_chains = 0;
+  absl::InlinedVector<int, 10> decode_target_protected_by_chain;
+  absl::InlinedVector<RenderResolution, 4> resolutions;
+  std::vector<FrameDependencyTemplate> templates;
+};
+
+struct DependencyDescriptor {
+  bool first_packet_in_frame = true;
+  bool last_packet_in_frame = true;
+  bool has_structure_attached = false;
+  int frame_number = 0;
+  FrameDependencyTemplate frame_dependencies;
+  absl::optional<RenderResolution> resolution;
+};
 
 // Describes how a certain encoder buffer was used when encoding a frame.
 struct CodecBufferUsage {
@@ -31,14 +106,7 @@ struct CodecBufferUsage {
   bool updated = false;
 };
 
-struct GenericFrameInfo {
-  enum class DecodeTargetIndication {
-    kNotPresent,   // DecodeTargetInfo symbol '-'
-    kDiscardable,  // DecodeTargetInfo symbol 'D'
-    kSwitch,       // DecodeTargetInfo symbol 'S'
-    kRequired      // DecodeTargetInfo symbol 'R'
-  };
-
+struct GenericFrameInfo : public FrameDependencyTemplate {
   static absl::InlinedVector<DecodeTargetIndication, 10> DecodeTargetInfo(
       absl::string_view indication_symbols);
 
@@ -49,10 +117,6 @@ struct GenericFrameInfo {
   ~GenericFrameInfo();
 
   int64_t frame_id = 0;
-  int temporal_id = 0;
-  int spatial_id = 0;
-  absl::InlinedVector<int, 10> frame_diffs;
-  absl::InlinedVector<DecodeTargetIndication, 10> decode_target_indications;
   absl::InlinedVector<CodecBufferUsage, kMaxEncoderBuffers> encoder_buffers;
 };
 
@@ -71,16 +135,6 @@ class GenericFrameInfo::Builder {
   GenericFrameInfo info_;
 };
 
-struct TemplateStructure {
-  TemplateStructure();
-  TemplateStructure(const TemplateStructure&);
-  TemplateStructure(TemplateStructure&&);
-  TemplateStructure& operator=(const TemplateStructure&);
-  ~TemplateStructure();
-
-  int num_decode_targets = 0;
-  std::vector<GenericFrameInfo> templates;
-};
 }  // namespace webrtc
 
 #endif  // COMMON_VIDEO_GENERIC_FRAME_DESCRIPTOR_GENERIC_FRAME_INFO_H_

modules/video_coding/codecs/vp8/default_temporal_layers.cc

@@ -585,12 +585,12 @@ void DefaultTemporalLayers::OnRttUpdate(int64_t rtt_ms) {}
 void DefaultTemporalLayers::OnLossNotification(
     const VideoEncoder::LossNotification& loss_notification) {}
 
-TemplateStructure DefaultTemporalLayers::GetTemplateStructure(
+FrameDependencyStructure DefaultTemporalLayers::GetTemplateStructure(
     int num_layers) const {
   RTC_CHECK_LT(num_layers, 5);
   RTC_CHECK_GT(num_layers, 0);
 
-  TemplateStructure template_structure;
+  FrameDependencyStructure template_structure;
   template_structure.num_decode_targets = num_layers;
 
   using Builder = GenericFrameInfo::Builder;

modules/video_coding/codecs/vp8/default_temporal_layers.h

@@ -77,8 +77,7 @@ class DefaultTemporalLayers final : public Vp8FrameBufferController {
               GenericFrameInfo::DecodeTargetInfo(indication_symbols)),
           frame_config(frame_config) {}
 
-    absl::InlinedVector<GenericFrameInfo::DecodeTargetIndication, 10>
-        decode_target_indications;
+    absl::InlinedVector<DecodeTargetIndication, 10> decode_target_indications;
     Vp8FrameConfig frame_config;
   };
 
@@ -93,7 +92,7 @@ class DefaultTemporalLayers final : public Vp8FrameBufferController {
   const std::vector<DependencyInfo> temporal_pattern_;
   // Set of buffers that are never updated except by keyframes.
   std::set<Vp8FrameConfig::Vp8BufferReference> kf_buffers_;
-  TemplateStructure GetTemplateStructure(int num_layers) const;
+  FrameDependencyStructure GetTemplateStructure(int num_layers) const;
 
   uint8_t pattern_idx_;
   // Updated cumulative bitrates, per temporal layer.

modules/video_coding/codecs/vp8/screenshare_layers.cc

@@ -421,12 +421,12 @@ void ScreenshareLayers::OnRttUpdate(int64_t rtt_ms) {}
 void ScreenshareLayers::OnLossNotification(
     const VideoEncoder::LossNotification& loss_notification) {}
 
-TemplateStructure ScreenshareLayers::GetTemplateStructure(
+FrameDependencyStructure ScreenshareLayers::GetTemplateStructure(
     int num_layers) const {
   RTC_CHECK_LT(num_layers, 3);
   RTC_CHECK_GT(num_layers, 0);
 
-  TemplateStructure template_structure;
+  FrameDependencyStructure template_structure;
   template_structure.num_decode_targets = num_layers;
 
   using Builder = GenericFrameInfo::Builder;

modules/video_coding/codecs/vp8/screenshare_layers.h

@@ -81,8 +81,7 @@ class ScreenshareLayers final : public Vp8FrameBufferController {
               GenericFrameInfo::DecodeTargetInfo(indication_symbols)),
           frame_config(frame_config) {}
 
-    absl::InlinedVector<GenericFrameInfo::DecodeTargetIndication, 10>
-        decode_target_indications;
+    absl::InlinedVector<DecodeTargetIndication, 10> decode_target_indications;
     Vp8FrameConfig frame_config;
   };
 
@@ -140,7 +139,7 @@ class ScreenshareLayers final : public Vp8FrameBufferController {
   } layers_[kMaxNumTemporalLayers];
 
   void UpdateHistograms();
-  TemplateStructure GetTemplateStructure(int num_layers) const;
+  FrameDependencyStructure GetTemplateStructure(int num_layers) const;
 
   // Data for histogram statistics.
   struct Stats {

modules/video_coding/include/video_codec_interface.h

@@ -107,7 +107,7 @@ struct RTC_EXPORT CodecSpecificInfo {
   VideoCodecType codecType;
   CodecSpecificInfoUnion codecSpecific;
   absl::optional<GenericFrameInfo> generic_frame_info;
-  absl::optional<TemplateStructure> template_structure;
+  absl::optional<FrameDependencyStructure> template_structure;
 };
 
 }  // namespace webrtc