// Copyright 2015 Google LLC. All rights reserved. // // Use of this source code is governed by a BSD-style // license that can be found in the LICENSE file or at // https://developers.google.com/open-source/licenses/bsd #include "packager/media/codecs/vp8_parser.h" #include "packager/base/logging.h" #include "packager/media/base/bit_reader.h" #include "packager/media/base/rcheck.h" namespace shaka { namespace media { namespace { const uint32_t MB_FEATURE_TREE_PROBS = 3; const uint32_t MAX_MB_SEGMENTS = 4; const uint32_t MAX_REF_LF_DELTAS = 4; const uint32_t MAX_MODE_LF_DELTAS = 4; const uint32_t MB_LVL_MAX = 2; const uint32_t MB_FEATURE_DATA_BITS[MB_LVL_MAX] = {7, 6}; bool VerifySyncCode(const uint8_t* data) { return data[0] == 0x9d && data[1] == 0x01 && data[2] == 0x2a; } bool ReadSegmentation(BitReader* reader) { bool enabled; RCHECK(reader->ReadBits(1, &enabled)); if (!enabled) return true; bool update_map; RCHECK(reader->ReadBits(1, &update_map)); bool update_data; RCHECK(reader->ReadBits(1, &update_data)); if (update_data) { RCHECK(reader->SkipBits(1)); // abs_delta for (uint32_t i = 0; i < MAX_MB_SEGMENTS; ++i) for (uint32_t j = 0; j < MB_LVL_MAX; ++j) { RCHECK(reader->SkipBitsConditional(true, MB_FEATURE_DATA_BITS[j] + 1)); } } if (update_map) { for (uint32_t i = 0; i < MB_FEATURE_TREE_PROBS; ++i) RCHECK(reader->SkipBitsConditional(true, 8)); } return true; } bool ReadLoopFilter(BitReader* reader) { RCHECK(reader->SkipBits(10)); // filter_type, filter_evel, sharness_level bool mode_ref_delta_enabled; RCHECK(reader->ReadBits(1, &mode_ref_delta_enabled)); if (!mode_ref_delta_enabled) return true; bool mode_ref_delta_update; RCHECK(reader->ReadBits(1, &mode_ref_delta_update)); if (!mode_ref_delta_update) return true; for (uint32_t i = 0; i < MAX_REF_LF_DELTAS + MAX_MODE_LF_DELTAS; ++i) RCHECK(reader->SkipBitsConditional(true, 6 + 1)); return true; } bool ReadQuantization(BitReader* reader) { uint32_t yac_index; RCHECK(reader->ReadBits(7, &yac_index)); VLOG(4) << "yac_index: " << yac_index; RCHECK(reader->SkipBitsConditional(true, 4 + 1)); // y dc delta RCHECK(reader->SkipBitsConditional(true, 4 + 1)); // y2 dc delta RCHECK(reader->SkipBitsConditional(true, 4 + 1)); // y2 ac delta RCHECK(reader->SkipBitsConditional(true, 4 + 1)); // chroma dc delta RCHECK(reader->SkipBitsConditional(true, 4 + 1)); // chroma ac delta return true; } bool ReadRefreshFrame(BitReader* reader) { bool refresh_golden_frame; RCHECK(reader->ReadBits(1, &refresh_golden_frame)); bool refresh_altref_frame; RCHECK(reader->ReadBits(1, &refresh_altref_frame)); if (!refresh_golden_frame) RCHECK(reader->SkipBits(2)); // buffer copy flag if (!refresh_altref_frame) RCHECK(reader->SkipBits(2)); // buffer copy flag RCHECK(reader->SkipBits(2)); // sign bias flags return true; } } // namespace VP8Parser::VP8Parser() : width_(0), height_(0) {} VP8Parser::~VP8Parser() {} bool VP8Parser::Parse(const uint8_t* data, size_t data_size, std::vector* vpx_frames) { DCHECK(data); DCHECK(vpx_frames); BitReader reader(data, data_size); // The following 3 bytes are read directly from |data|. RCHECK(reader.SkipBytes(3)); // One bit for frame type. bool is_interframe = data[0] & 1; // 3-bit version number with 2 bits for profile and the other bit reserved for // future variants. uint8_t profile = (data[0] >> 1) & 3; // One bit for show frame flag. // Then 19 bits (the remaining 3 bits in the first byte + next two bytes) for // header size. uint32_t header_size = (data[0] | (data[1] << 8) | (data[2] << 16)) >> 5; RCHECK(header_size <= data_size); if (!is_interframe) { // The following 7 bytes are read directly from |data|. RCHECK(reader.SkipBytes(7)); RCHECK(VerifySyncCode(&data[3])); // Bits 0b11000000 for data[7] and data[9] are scaling. width_ = data[6] | ((data[7] & 0x3f) << 8); height_ = data[8] | ((data[9] & 0x3f) << 8); RCHECK(reader.SkipBits(2)); // colorspace and pixel value clamping. } RCHECK(ReadSegmentation(&reader)); RCHECK(ReadLoopFilter(&reader)); RCHECK(reader.SkipBits(2)); // partitions bits RCHECK(ReadQuantization(&reader)); if (is_interframe) { RCHECK(ReadRefreshFrame(&reader)); RCHECK(reader.SkipBits(1)); // refresh_entropy_probs RCHECK(reader.SkipBits(1)); // refresh last frame flag } else { RCHECK(reader.SkipBits(1)); // refresh_entropy_probs } // The next field is entropy header (coef probability tree), which is encoded // using bool entropy encoder, i.e. compressed. We don't consider it as part // of uncompressed header. writable_codec_config()->set_profile(profile); // VP8 uses an 8-bit YUV 4:2:0 format. // http://tools.ietf.org/html/rfc6386 Section 2. writable_codec_config()->set_bit_depth(8); writable_codec_config()->SetChromaSubsampling( VPCodecConfigurationRecord::CHROMA_420_COLLOCATED_WITH_LUMA); VPxFrameInfo vpx_frame; vpx_frame.frame_size = data_size; vpx_frame.uncompressed_header_size = vpx_frame.frame_size - reader.bits_available() / 8; vpx_frame.is_keyframe = !is_interframe; vpx_frame.width = width_; vpx_frame.height = height_; vpx_frames->clear(); vpx_frames->push_back(vpx_frame); VLOG(3) << "\n frame_size: " << vpx_frame.frame_size << "\n uncompressed_header_size: " << vpx_frame.uncompressed_header_size << "\n bits read: " << reader.bit_position() << "\n header_size: " << header_size << "\n width: " << vpx_frame.width << "\n height: " << vpx_frame.height; return true; } bool VP8Parser::IsKeyframe(const uint8_t* data, size_t data_size) { // Make sure the block is big enough for the minimal keyframe header size. if (data_size < 10) return false; // The LSb of the first byte must be a 0 for a keyframe. if ((data[0] & 0x01) != 0) return false; return VerifySyncCode(&data[3]); } } // namespace media } // namespace shaka