diff mbox series

[FFmpeg-devel,v2,3/3] avcodec/avs2_parser: parse more info

Message ID tencent_D71DC155B6D86B3E007569BA7CA2AE32380A@qq.com
State New
Headers show
Series None | expand

Commit Message

zhilizhao(赵志立) June 12, 2022, 4:42 p.m. UTC
Including video resolution, framerate and picture type, etc.

Signed-off-by: Zhao Zhili <quinkblack@foxmail.com>
---
v2: fix unused variable warning

 libavcodec/avs2.h        |  27 ++++++++++
 libavcodec/avs2_parser.c | 103 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 130 insertions(+)

Comments

Andreas Rheinhardt June 12, 2022, 5:07 p.m. UTC | #1
Zhao Zhili:
> Including video resolution, framerate and picture type, etc.
> 
> Signed-off-by: Zhao Zhili <quinkblack@foxmail.com>
> ---
> v2: fix unused variable warning
> 
>  libavcodec/avs2.h        |  27 ++++++++++
>  libavcodec/avs2_parser.c | 103 +++++++++++++++++++++++++++++++++++++++
>  2 files changed, 130 insertions(+)
> 
> diff --git a/libavcodec/avs2.h b/libavcodec/avs2.h
> index f342ba52a0..e961c7d517 100644
> --- a/libavcodec/avs2.h
> +++ b/libavcodec/avs2.h
> @@ -23,6 +23,8 @@
>  #ifndef AVCODEC_AVS2_H
>  #define AVCODEC_AVS2_Hto 
>  
> +#include "libavutil/rational.h"
> +
>  #define AVS2_SLICE_MAX_START_CODE    0x000001AF
>  
>  enum {
> @@ -38,4 +40,29 @@ enum {
>  #define AVS2_ISPIC(x)  ((x) == AVS2_INTRA_PIC_START_CODE || (x) == AVS2_INTER_PIC_START_CODE)
>  #define AVS2_ISUNIT(x) ((x) == AVS2_SEQ_START_CODE || AVS2_ISPIC(x))
>  
> +enum AVS2Profile {
> +    AVS2_PROFILE_MAIN_PIC   = 0x12,
> +    AVS2_PROFILE_MAIN       = 0x20,
> +    AVS2_PROFILE_MAIN10     = 0x22,
> +};
> +
> +static const AVRational ff_avs2_frame_rate_tab[16] = {

If this is only used at one place (as it is), it should be moved to that
place; if it is to be used in several places, it should be moved to a
single file instead of a header.

> +    { 0    , 0   }, // forbid
> +    { 24000, 1001},
> +    { 24   , 1   },
> +    { 25   , 1   },
> +    { 30000, 1001},
> +    { 30   , 1   },
> +    { 50   , 1   },
> +    { 60000, 1001},
> +    { 60   , 1   },
> +    { 100  , 1   },
> +    { 120  , 1   },
> +    { 200  , 1   },
> +    { 240  , 1   },
> +    { 300  , 1   },
> +    { 0    , 0   }, // reserved
> +    { 0    , 0   }  // reserved
> +};
> +
>  #endif
> diff --git a/libavcodec/avs2_parser.c b/libavcodec/avs2_parser.c
> index 71cf442903..18d7729768 100644
> --- a/libavcodec/avs2_parser.c
> +++ b/libavcodec/avs2_parser.c
> @@ -19,7 +19,9 @@
>   * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
>   */
>  
> +#include "libavutil/avutil.h"
>  #include "avs2.h"
> +#include "get_bits.h"
>  #include "parser.h"
>  
>  static int avs2_find_frame_end(ParseContext *pc, const uint8_t *buf, int buf_size)
> @@ -58,6 +60,105 @@ static int avs2_find_frame_end(ParseContext *pc, const uint8_t *buf, int buf_siz
>      return END_NOT_FOUND;
>  }
>  
> +static void parse_avs2_seq_header(AVCodecParserContext *s, const uint8_t *buf,
> +                                 int buf_size, AVCodecContext *avctx)
> +{
> +    GetBitContext gb;
> +    int profile, level;
> +    int width, height;
> +    int chroma, sample_precision, encoding_precision = 1;
> +    // sample_precision and encoding_precision is 3 bits
> +    static uint8_t precision[8] = { 0, 8, 10 };

Missing const.

> +    unsigned aspect_ratio;
> +    unsigned frame_rate_code;
> +    int low_delay;
> +
> +    if (buf_size < 15)
> +        return;
> +
> +    init_get_bits8(&gb, buf, buf_size);

If buf_size is gigantic, init_get_bits8 can fail. Given that you only
ever read the first 15 bytes of the header, you can avoid this by
initializing the GetBitContext with a size of 15.

> +
> +    s->key_frame = 1;
> +    s->pict_type = AV_PICTURE_TYPE_I;
> +
> +    profile = get_bits(&gb, 8);
> +    level = get_bits(&gb, 8);
> +
> +    // progressive_sequence     u(1)
> +    // field_coded_sequence     u(1)
> +    skip_bits(&gb, 2);
> +
> +    width = get_bits(&gb, 14);
> +    height = get_bits(&gb, 14);
> +
> +    chroma = get_bits(&gb, 2);
> +    sample_precision = get_bits(&gb, 3);
> +    if (profile == AVS2_PROFILE_MAIN10)
> +        encoding_precision = get_bits(&gb, 3);
> +
> +    aspect_ratio = get_bits(&gb, 4);
> +    frame_rate_code = get_bits(&gb, 4);
> +
> +    // bit_rate_lower       u(18)
> +    // marker_bit           f(1)
> +    // bit_rate_upper       u(12)
> +    skip_bits(&gb, 31);
> +
> +    low_delay = get_bits(&gb, 1);
> +
> +    s->width = width;
> +    s->height = height;
> +    s->coded_width = FFALIGN(width, 8);
> +    s->coded_height = FFALIGN(height, 8);
> +    avctx->framerate.num = avctx->time_base.den =
> +        ff_avs2_frame_rate_tab[frame_rate_code].num;
> +    avctx->framerate.den = avctx->time_base.num =
> +        ff_avs2_frame_rate_tab[frame_rate_code].den;
> +    avctx->has_b_frames = FFMAX(avctx->has_b_frames, !low_delay);
> +
> +    av_log(avctx, AV_LOG_DEBUG,
> +           "AVS2 parse seq HDR: profile %x, level %x, "
> +           "width %d, height %d, "
> +           "chroma %d, sample_precision %d bits, encoding_precision %d bits, "
> +           "aspect_ratio 0x%x, framerate %d/%d, low_delay %d\n",
> +           profile, level,
> +           width, height,
> +           chroma, precision[sample_precision], precision[encoding_precision],
> +           aspect_ratio, avctx->framerate.num, avctx->framerate.den, low_delay);
> +}
> +
> +static void parse_avs2_units(AVCodecParserContext *s, const uint8_t *buf,
> +                             int buf_size, AVCodecContext *avctx)
> +{
> +    if (buf_size < 5)
> +        return;
> +
> +    if (!(buf[0] == 0x0 && buf[1] == 0x0 && buf[2] == 0x1))
> +        return;
> +
> +    switch (buf[3]) {
> +    case AVS2_SEQ_START_CODE:
> +        parse_avs2_seq_header(s, buf + 4, buf_size - 4, avctx);
> +        return;
> +    case AVS2_INTRA_PIC_START_CODE:
> +        s->key_frame = 1;
> +        s->pict_type = AV_PICTURE_TYPE_I;
> +        return;
> +    case AVS2_INTER_PIC_START_CODE:
> +        s->key_frame = 0;
> +        if (buf_size > 9) {
> +            int pic_code_type = buf[8] & 0x3;
> +            if (pic_code_type == 1)
> +                s->pict_type = AV_PICTURE_TYPE_P;
> +            else if (pic_code_type == 3)
> +                s->pict_type = AV_PICTURE_TYPE_S;
> +            else
> +                s->pict_type = AV_PICTURE_TYPE_B;
> +        }
> +        return;
> +    }
> +}
> +
>  static int avs2_parse(AVCodecParserContext *s, AVCodecContext *avctx,
>                        const uint8_t **poutbuf, int *poutbuf_size,
>                        const uint8_t *buf, int buf_size)
> @@ -76,6 +177,8 @@ static int avs2_parse(AVCodecParserContext *s, AVCodecContext *avctx,
>          }
>      }
>  
> +    parse_avs2_units(s, buf, buf_size, avctx);
> +
>      *poutbuf = buf;
>      *poutbuf_size = buf_size;
>
zhilizhao(赵志立) June 13, 2022, 3:42 a.m. UTC | #2
> On Jun 13, 2022, at 1:07 AM, Andreas Rheinhardt <andreas.rheinhardt@outlook.com> wrote:
> 
> Zhao Zhili:
>> Including video resolution, framerate and picture type, etc.
>> 
>> Signed-off-by: Zhao Zhili <quinkblack@foxmail.com>
>> ---
>> v2: fix unused variable warning
>> 
>> libavcodec/avs2.h        |  27 ++++++++++
>> libavcodec/avs2_parser.c | 103 +++++++++++++++++++++++++++++++++++++++
>> 2 files changed, 130 insertions(+)
>> 
>> diff --git a/libavcodec/avs2.h b/libavcodec/avs2.h
>> index f342ba52a0..e961c7d517 100644
>> --- a/libavcodec/avs2.h
>> +++ b/libavcodec/avs2.h
>> @@ -23,6 +23,8 @@
>> #ifndef AVCODEC_AVS2_H
>> #define AVCODEC_AVS2_Hto 
>> 
>> +#include "libavutil/rational.h"
>> +
>> #define AVS2_SLICE_MAX_START_CODE    0x000001AF
>> 
>> enum {
>> @@ -38,4 +40,29 @@ enum {
>> #define AVS2_ISPIC(x)  ((x) == AVS2_INTRA_PIC_START_CODE || (x) == AVS2_INTER_PIC_START_CODE)
>> #define AVS2_ISUNIT(x) ((x) == AVS2_SEQ_START_CODE || AVS2_ISPIC(x))
>> 
>> +enum AVS2Profile {
>> +    AVS2_PROFILE_MAIN_PIC   = 0x12,
>> +    AVS2_PROFILE_MAIN       = 0x20,
>> +    AVS2_PROFILE_MAIN10     = 0x22,
>> +};
>> +
>> +static const AVRational ff_avs2_frame_rate_tab[16] = {
> 
> If this is only used at one place (as it is), it should be moved to that
> place; if it is to be used in several places, it should be moved to a
> single file instead of a header.

Thanks for the review, fixed by v3:

1. Move ff_avs2_frame_rate_tab from avs2.h to avs2.c
2. Consitify `precision` in patch 3/5
3. init_get_bits8() with size limites to buf_size_min
4. Select avs2_parser for libdavs2
5. Use ff_avs2_frame_rate_tab in libdavs2

http://ffmpeg.org/pipermail/ffmpeg-devel/2022-June/297476.html


> 
>> +    { 0    , 0   }, // forbid
>> +    { 24000, 1001},
>> +    { 24   , 1   },
>> +    { 25   , 1   },
>> +    { 30000, 1001},
>> +    { 30   , 1   },
>> +    { 50   , 1   },
>> +    { 60000, 1001},
>> +    { 60   , 1   },
>> +    { 100  , 1   },
>> +    { 120  , 1   },
>> +    { 200  , 1   },
>> +    { 240  , 1   },
>> +    { 300  , 1   },
>> +    { 0    , 0   }, // reserved
>> +    { 0    , 0   }  // reserved
>> +};
>> +
>> #endif
>> diff --git a/libavcodec/avs2_parser.c b/libavcodec/avs2_parser.c
>> index 71cf442903..18d7729768 100644
>> --- a/libavcodec/avs2_parser.c
>> +++ b/libavcodec/avs2_parser.c
>> @@ -19,7 +19,9 @@
>>  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
>>  */
>> 
>> +#include "libavutil/avutil.h"
>> #include "avs2.h"
>> +#include "get_bits.h"
>> #include "parser.h"
>> 
>> static int avs2_find_frame_end(ParseContext *pc, const uint8_t *buf, int buf_size)
>> @@ -58,6 +60,105 @@ static int avs2_find_frame_end(ParseContext *pc, const uint8_t *buf, int buf_siz
>>     return END_NOT_FOUND;
>> }
>> 
>> +static void parse_avs2_seq_header(AVCodecParserContext *s, const uint8_t *buf,
>> +                                 int buf_size, AVCodecContext *avctx)
>> +{
>> +    GetBitContext gb;
>> +    int profile, level;
>> +    int width, height;
>> +    int chroma, sample_precision, encoding_precision = 1;
>> +    // sample_precision and encoding_precision is 3 bits
>> +    static uint8_t precision[8] = { 0, 8, 10 };
> 
> Missing const.
> 
>> +    unsigned aspect_ratio;
>> +    unsigned frame_rate_code;
>> +    int low_delay;
>> +
>> +    if (buf_size < 15)
>> +        return;
>> +
>> +    init_get_bits8(&gb, buf, buf_size);
> 
> If buf_size is gigantic, init_get_bits8 can fail. Given that you only
> ever read the first 15 bytes of the header, you can avoid this by
> initializing the GetBitContext with a size of 15.
> 
>> +
>> +    s->key_frame = 1;
>> +    s->pict_type = AV_PICTURE_TYPE_I;
>> +
>> +    profile = get_bits(&gb, 8);
>> +    level = get_bits(&gb, 8);
>> +
>> +    // progressive_sequence     u(1)
>> +    // field_coded_sequence     u(1)
>> +    skip_bits(&gb, 2);
>> +
>> +    width = get_bits(&gb, 14);
>> +    height = get_bits(&gb, 14);
>> +
>> +    chroma = get_bits(&gb, 2);
>> +    sample_precision = get_bits(&gb, 3);
>> +    if (profile == AVS2_PROFILE_MAIN10)
>> +        encoding_precision = get_bits(&gb, 3);
>> +
>> +    aspect_ratio = get_bits(&gb, 4);
>> +    frame_rate_code = get_bits(&gb, 4);
>> +
>> +    // bit_rate_lower       u(18)
>> +    // marker_bit           f(1)
>> +    // bit_rate_upper       u(12)
>> +    skip_bits(&gb, 31);
>> +
>> +    low_delay = get_bits(&gb, 1);
>> +
>> +    s->width = width;
>> +    s->height = height;
>> +    s->coded_width = FFALIGN(width, 8);
>> +    s->coded_height = FFALIGN(height, 8);
>> +    avctx->framerate.num = avctx->time_base.den =
>> +        ff_avs2_frame_rate_tab[frame_rate_code].num;
>> +    avctx->framerate.den = avctx->time_base.num =
>> +        ff_avs2_frame_rate_tab[frame_rate_code].den;
>> +    avctx->has_b_frames = FFMAX(avctx->has_b_frames, !low_delay);
>> +
>> +    av_log(avctx, AV_LOG_DEBUG,
>> +           "AVS2 parse seq HDR: profile %x, level %x, "
>> +           "width %d, height %d, "
>> +           "chroma %d, sample_precision %d bits, encoding_precision %d bits, "
>> +           "aspect_ratio 0x%x, framerate %d/%d, low_delay %d\n",
>> +           profile, level,
>> +           width, height,
>> +           chroma, precision[sample_precision], precision[encoding_precision],
>> +           aspect_ratio, avctx->framerate.num, avctx->framerate.den, low_delay);
>> +}
>> +
>> +static void parse_avs2_units(AVCodecParserContext *s, const uint8_t *buf,
>> +                             int buf_size, AVCodecContext *avctx)
>> +{
>> +    if (buf_size < 5)
>> +        return;
>> +
>> +    if (!(buf[0] == 0x0 && buf[1] == 0x0 && buf[2] == 0x1))
>> +        return;
>> +
>> +    switch (buf[3]) {
>> +    case AVS2_SEQ_START_CODE:
>> +        parse_avs2_seq_header(s, buf + 4, buf_size - 4, avctx);
>> +        return;
>> +    case AVS2_INTRA_PIC_START_CODE:
>> +        s->key_frame = 1;
>> +        s->pict_type = AV_PICTURE_TYPE_I;
>> +        return;
>> +    case AVS2_INTER_PIC_START_CODE:
>> +        s->key_frame = 0;
>> +        if (buf_size > 9) {
>> +            int pic_code_type = buf[8] & 0x3;
>> +            if (pic_code_type == 1)
>> +                s->pict_type = AV_PICTURE_TYPE_P;
>> +            else if (pic_code_type == 3)
>> +                s->pict_type = AV_PICTURE_TYPE_S;
>> +            else
>> +                s->pict_type = AV_PICTURE_TYPE_B;
>> +        }
>> +        return;
>> +    }
>> +}
>> +
>> static int avs2_parse(AVCodecParserContext *s, AVCodecContext *avctx,
>>                       const uint8_t **poutbuf, int *poutbuf_size,
>>                       const uint8_t *buf, int buf_size)
>> @@ -76,6 +177,8 @@ static int avs2_parse(AVCodecParserContext *s, AVCodecContext *avctx,
>>         }
>>     }
>> 
>> +    parse_avs2_units(s, buf, buf_size, avctx);
>> +
>>     *poutbuf = buf;
>>     *poutbuf_size = buf_size;
>> 
> 
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
> 
> To unsubscribe, visit link above, or email
> ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".
diff mbox series

Patch

diff --git a/libavcodec/avs2.h b/libavcodec/avs2.h
index f342ba52a0..e961c7d517 100644
--- a/libavcodec/avs2.h
+++ b/libavcodec/avs2.h
@@ -23,6 +23,8 @@ 
 #ifndef AVCODEC_AVS2_H
 #define AVCODEC_AVS2_H
 
+#include "libavutil/rational.h"
+
 #define AVS2_SLICE_MAX_START_CODE    0x000001AF
 
 enum {
@@ -38,4 +40,29 @@  enum {
 #define AVS2_ISPIC(x)  ((x) == AVS2_INTRA_PIC_START_CODE || (x) == AVS2_INTER_PIC_START_CODE)
 #define AVS2_ISUNIT(x) ((x) == AVS2_SEQ_START_CODE || AVS2_ISPIC(x))
 
+enum AVS2Profile {
+    AVS2_PROFILE_MAIN_PIC   = 0x12,
+    AVS2_PROFILE_MAIN       = 0x20,
+    AVS2_PROFILE_MAIN10     = 0x22,
+};
+
+static const AVRational ff_avs2_frame_rate_tab[16] = {
+    { 0    , 0   }, // forbid
+    { 24000, 1001},
+    { 24   , 1   },
+    { 25   , 1   },
+    { 30000, 1001},
+    { 30   , 1   },
+    { 50   , 1   },
+    { 60000, 1001},
+    { 60   , 1   },
+    { 100  , 1   },
+    { 120  , 1   },
+    { 200  , 1   },
+    { 240  , 1   },
+    { 300  , 1   },
+    { 0    , 0   }, // reserved
+    { 0    , 0   }  // reserved
+};
+
 #endif
diff --git a/libavcodec/avs2_parser.c b/libavcodec/avs2_parser.c
index 71cf442903..18d7729768 100644
--- a/libavcodec/avs2_parser.c
+++ b/libavcodec/avs2_parser.c
@@ -19,7 +19,9 @@ 
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
+#include "libavutil/avutil.h"
 #include "avs2.h"
+#include "get_bits.h"
 #include "parser.h"
 
 static int avs2_find_frame_end(ParseContext *pc, const uint8_t *buf, int buf_size)
@@ -58,6 +60,105 @@  static int avs2_find_frame_end(ParseContext *pc, const uint8_t *buf, int buf_siz
     return END_NOT_FOUND;
 }
 
+static void parse_avs2_seq_header(AVCodecParserContext *s, const uint8_t *buf,
+                                 int buf_size, AVCodecContext *avctx)
+{
+    GetBitContext gb;
+    int profile, level;
+    int width, height;
+    int chroma, sample_precision, encoding_precision = 1;
+    // sample_precision and encoding_precision is 3 bits
+    static uint8_t precision[8] = { 0, 8, 10 };
+    unsigned aspect_ratio;
+    unsigned frame_rate_code;
+    int low_delay;
+
+    if (buf_size < 15)
+        return;
+
+    init_get_bits8(&gb, buf, buf_size);
+
+    s->key_frame = 1;
+    s->pict_type = AV_PICTURE_TYPE_I;
+
+    profile = get_bits(&gb, 8);
+    level = get_bits(&gb, 8);
+
+    // progressive_sequence     u(1)
+    // field_coded_sequence     u(1)
+    skip_bits(&gb, 2);
+
+    width = get_bits(&gb, 14);
+    height = get_bits(&gb, 14);
+
+    chroma = get_bits(&gb, 2);
+    sample_precision = get_bits(&gb, 3);
+    if (profile == AVS2_PROFILE_MAIN10)
+        encoding_precision = get_bits(&gb, 3);
+
+    aspect_ratio = get_bits(&gb, 4);
+    frame_rate_code = get_bits(&gb, 4);
+
+    // bit_rate_lower       u(18)
+    // marker_bit           f(1)
+    // bit_rate_upper       u(12)
+    skip_bits(&gb, 31);
+
+    low_delay = get_bits(&gb, 1);
+
+    s->width = width;
+    s->height = height;
+    s->coded_width = FFALIGN(width, 8);
+    s->coded_height = FFALIGN(height, 8);
+    avctx->framerate.num = avctx->time_base.den =
+        ff_avs2_frame_rate_tab[frame_rate_code].num;
+    avctx->framerate.den = avctx->time_base.num =
+        ff_avs2_frame_rate_tab[frame_rate_code].den;
+    avctx->has_b_frames = FFMAX(avctx->has_b_frames, !low_delay);
+
+    av_log(avctx, AV_LOG_DEBUG,
+           "AVS2 parse seq HDR: profile %x, level %x, "
+           "width %d, height %d, "
+           "chroma %d, sample_precision %d bits, encoding_precision %d bits, "
+           "aspect_ratio 0x%x, framerate %d/%d, low_delay %d\n",
+           profile, level,
+           width, height,
+           chroma, precision[sample_precision], precision[encoding_precision],
+           aspect_ratio, avctx->framerate.num, avctx->framerate.den, low_delay);
+}
+
+static void parse_avs2_units(AVCodecParserContext *s, const uint8_t *buf,
+                             int buf_size, AVCodecContext *avctx)
+{
+    if (buf_size < 5)
+        return;
+
+    if (!(buf[0] == 0x0 && buf[1] == 0x0 && buf[2] == 0x1))
+        return;
+
+    switch (buf[3]) {
+    case AVS2_SEQ_START_CODE:
+        parse_avs2_seq_header(s, buf + 4, buf_size - 4, avctx);
+        return;
+    case AVS2_INTRA_PIC_START_CODE:
+        s->key_frame = 1;
+        s->pict_type = AV_PICTURE_TYPE_I;
+        return;
+    case AVS2_INTER_PIC_START_CODE:
+        s->key_frame = 0;
+        if (buf_size > 9) {
+            int pic_code_type = buf[8] & 0x3;
+            if (pic_code_type == 1)
+                s->pict_type = AV_PICTURE_TYPE_P;
+            else if (pic_code_type == 3)
+                s->pict_type = AV_PICTURE_TYPE_S;
+            else
+                s->pict_type = AV_PICTURE_TYPE_B;
+        }
+        return;
+    }
+}
+
 static int avs2_parse(AVCodecParserContext *s, AVCodecContext *avctx,
                       const uint8_t **poutbuf, int *poutbuf_size,
                       const uint8_t *buf, int buf_size)
@@ -76,6 +177,8 @@  static int avs2_parse(AVCodecParserContext *s, AVCodecContext *avctx,
         }
     }
 
+    parse_avs2_units(s, buf, buf_size, avctx);
+
     *poutbuf = buf;
     *poutbuf_size = buf_size;