//
// MPEG Closed Caption Extractor (v2.2)
//
// Version History:
// 1.0: Initial release
// 1.1: Handle special case of DVD format with ff xx xx ff yy yy pattern
// 1.2: Swapped identification of ReplayTV 4000 and 5000 formats
//       (no visible effect),
//      Swapped field output for both ReplayTV formats
// 1.3: improved faulty DVD format handling (thanks to Keith Hui)
// 1.4: Added ATSC format generated by Pinnacle MediaStream
// 2.0: Re-wrote using General Parser sample code so that caption code
//       is aware of picture type (required by ATSC format),
//      Added DSS-SD captions (they were included in video.bdl)
// 2.1: Extraction of CC from ATSC streams fixed (thanks to Ken Schultz)
// 2.2: Fixed the omission of some CC bytes from ATSC streams (by Ken Schultz)

enum {
  START_CODE                = 0x00000100,
  PICTURE_START_CODE        = 0x00000100,
  USER_DATA_START_CODE      = 0x000001b2,
  SEQUENCE_START_CODE       = 0x000001b3,
  SEQUENCE_ERROR_START_CODE = 0x000001b4,
  EXTENSION_START_CODE      = 0x000001b5,
  SEQUENCE_END_CODE         = 0x000001b7,
  GROUP_START_CODE          = 0x000001b8
};

enum {
  DVD_CLOSED_CAPTION           = 0x434301f8,
  ATSC_CLOSED_CAPTION          = 0x47413934,
  REPLAYTV_4000_CLOSED_CAPTION = 0xbb02,
  REPLAYTV_5000_CLOSED_CAPTION = 0x9902,
  DISH_NETWORK_CLOSED_CAPTION  = 0x0502
};

enum {
  MAX_CLOSED_CAPTION_DATA_PER_PICTURE = 32
};

int sequence_counter;

int DVD_FIELD[2];    // EIA 608 format captions (analog TV)
DVD_FIELD[0] = 0xff;
DVD_FIELD[1] = 0xfe;

int HDCC_FIELD[2];   // EIA 708 format captions (HDTV)
HDCC_FIELD[0] = 0xfc;
HDCC_FIELD[1] = 0xfd;

enum {
  DN_2_BYTE   = 0x02,
  DN_4_BYTE   = 0x04,
  DN_BUFFERED = 0x05
};

int REPEAT = 0x04;

int  ccPick = -1;
while ((ccPick != 0) && (ccPick !=1)) {
  input "Closed Captioning Field 1 or 2" int f = 1;
  ccPick = --f;
}

int current_picture_coding_type;

// ATSC transmits captions with every frame, so I & P frame data has to be held
//  until after B frame data or it would end up out of order
int cc_buffer[MAX_CLOSED_CAPTION_DATA_PER_PICTURE];

// Dish Network transmits captions with every frame, so B frame data has to be held
//  or it would end up out of order
int DN_buffer = -1;

quiet void main(void)
{
  int dummy;
  dummy = 0xffffffff;
  output dummy:32;
  int i;
  for (i=0; i < MAX_CLOSED_CAPTION_DATA_PER_PICTURE; i++) {
    cc_buffer[i] = -1;
  }
  sequence_counter = 0;
  while (1) {
    switch(search_word(START_CODE, 0xff)) {
    case SEQUENCE_START_CODE:
      {field int remove_start_code:32;}
      sequence();
      continue;
    case SEQUENCE_END_CODE:
      {field int remove_start_code:32;}
      sequence_end();
      continue;
    case SEQUENCE_ERROR_START_CODE:
      {field int remove_start_code:32;}
      sequence_error();
      continue;
    default:
      {field int remove_start_code:32;}
      continue;
    }
  }
}
      
void sequence(void)
{
  sequence_counter++;

  while (1) {
    switch(search_word(START_CODE, 0xff)) {
    case EXTENSION_START_CODE:
      {field int remove_start_code:32;}
      continue;
    case USER_DATA_START_CODE:
      {field int remove_start_code:32;}
      user_data(SEQUENCE_START_CODE);
      continue;
    case GROUP_START_CODE:
      {field int remove_start_code:32;}
      group_of_pictures_header();
      continue;
    case PICTURE_START_CODE:
      {field int remove_start_code:32;}
      picture_header();
      continue;
    default:
      return;
    }
  }
}

void sequence_end(void)
{
}

void sequence_error(void)
{
}

void group_of_pictures_header(void)
{
  while(1) {
    switch(search_word(START_CODE, 0xff)) {
    case EXTENSION_START_CODE:
      {field int remove_start_code:32;}
      continue;
    case USER_DATA_START_CODE:
      {field int remove_start_code:32;}
      user_data(GROUP_START_CODE);
      continue;
    case PICTURE_START_CODE:
      {field int remove_start_code:32;}
      picture_header();
      continue;
    default:
      return;
    }
  }
}

void user_data(int start_code)
{
  int cc_data, cc_count, skip;
  switch(start_code) {
  // no sequence-based caption systems
  // GOP-based caption systems: DVD
  case GROUP_START_CODE:
    if (looking_at_bits(DVD_CLOSED_CAPTION, 32)) {
      field "0x%08x" int DVD_identifier:32;
      field "%d" cc_count:8;
      // patterns: 0x00 = Field 2 then Field 1, 0x80 = Field 1 then Field 2
      int pattern = cc_count & 0x80;
      int desired_packet = 0;
      if (ccPick == 0 && pattern == 0x00) {
        desired_packet = 1;
      }
      if (ccPick == 1 && pattern == 0x80) {
        desired_packet = 1;
      }
      cc_count &= 0x1e;
      cc_count /= 2;
      // printf ("\n%d: ", cc_count);
      int i, j;
      for (i=0; i < cc_count; i++) {
        for (j=0; j < 2; j++) {
          field int caption_field:8;
          if ((caption_field != DVD_FIELD[ccPick]) || (j != desired_packet)) {
            field skip:16;
            // printf ("%x/%x ", caption_field, skip);
          } else {
            field output cc_data:16;
            // printf ("%x/%x ", caption_field, cc_data);
          }
        }
      }
      return;
    }
  // frame-based caption systems: DSS-SD, ATSC, ReplayTV (4000 & 5000 series), DishNetwork
  case PICTURE_START_CODE:
    if(looking_at_bits(ATSC_CLOSED_CAPTION, 32)) {
      field "0x%08x" int ATSC_identifier:32;
      field "%d" int user_data_type_code:8;
      if(user_data_type_code == 0x03) {
        field "%d" int process_em_data_flag:1;
        field "%d" int process_cc_data_flag:1;
        field "%d" int additional_data_flag:1;
        field "%d" cc_count:5;
        field "%d" int em_data:8;
        int i, j;
        if(current_picture_coding_type != 3) {
          // I or P picture
          for(i=0; i < MAX_CLOSED_CAPTION_DATA_PER_PICTURE; i++) {
            if(cc_buffer[i] == -1) {
              break;
            }
            output cc_buffer[i]:16;
            cc_buffer[i] = -1;
          }
        }
        j=0;
        for(i=0; i < cc_count; i++) {
          field "0x%02x" int marker_bits:5;
          field "%d" int cc_valid:1;
          field "%d" int cc_type:2;
          field "0x%04x" cc_data:16;
          if(cc_valid && ccPick == cc_type) {
            if(current_picture_coding_type == 3) {
              // B picture
              output cc_data:16;
            } else {
              // I or P picture
              cc_buffer[j++] = cc_data;
            }
          }
        }
        field "0x%02x" int marker_bits:8;
        if(additional_data_flag) {
          while(!looking_at_bits(0x000001, 24)) {
            field "0x%02x" int additional_usr_data:8;
          }
        }
      } else {
        // user_data_type_code != 0x03
        while(!looking_at_bits(0x000001, 24)) {
          field "0x%02x" int byte_data:8;
        }
      }
      return;
    } else if(looking_at_bits(REPLAYTV_4000_CLOSED_CAPTION, 16)) {
      field "0x%04x" int ReplayTV4000_identifier:16;
      // Field 2 data
      if (ccPick) {
        field output cc_data:16;
      } else {
        field skip:16;
      }
      field skip:16;    // 0xcc02
      // Field 1 data
      if (!ccPick) {
        field output cc_data:16;
      } else {
        field skip:16;
      }
      return;
    } else if(looking_at_bits(REPLAYTV_5000_CLOSED_CAPTION, 16)) {
      field "0x%04x" int ReplayTV5000_identifier:16;
      // Field 1 data
      if (!ccPick) {
        field output cc_data:16;
      } else {
        field skip:16;
      }
      field skip:16;    // 0xaa02
      // Field 2 data
      if (ccPick) {
        field output cc_data:16;
      } else {
        field skip:16;
      }
      return;
    } else if(looking_at_bits(DISH_NETWORK_CLOSED_CAPTION, 16)) {
      field "0x%04x" int DishNetwork_identifier:16;
      // Dish Network does not transmit Field 2
      if (ccPick) {
        return;
      }
      int repeater;
      field skip:8;  // 0x04
      field skip:16; // 2-byte counter
      field skip:16; // 2-byte checksum?
      field int DN_pattern:8;
      switch(DN_pattern) {
      case DN_2_BYTE:
        field skip:8;  // 0x09
        field output cc_data:16;
        field repeater:8;
        if (repeater == REPEAT) {
          output cc_data:16;
        }
        break;
      case DN_4_BYTE:
        field skip:8;  // 0x09
        field output cc_data:32;
        field skip:32; // 0x020a, followed by 2-byte checksum?
        break;
      case DN_BUFFERED:
        // play the previous buffered caption first
        if (DN_buffer > 0) {
          output DN_buffer:32;
        }
        field skip:8;  // 0x04
        field skip:8;  // 0x09
        field skip:32; // last buffered caption
        field int count:8;
        field skip:8;  // 0x09
        switch (count) {
        case DN_2_BYTE:
          field cc_data:16;
          DN_buffer = cc_data * 256;
          field repeater:8;
          if (repeater == REPEAT) {
            DN_buffer += cc_data;
          }
          break;
        case DN_4_BYTE:
          field cc_data:32;
          DN_buffer = cc_data;
          field skip:8;  // 0x02
          break;
        }
        break;
      }
      return;
    } else {
      // This stream seems to be a DSS-SD video stream
      while(!looking_at_bits(0x000001, 24)) {
        field "%d" int user_data_length:8;
        if(user_data_length == 0)
          return;
        field "====== <%s> (0x%02x)" enum {
          Forbidden = 0x00,
          Presentation_Time_Stamp = 0x02,
          Decode_Time_Stamp = 0x04,
          Chroma_Flags = 0x05,
          Pan_and_Scan = 0x06,
          Field_Display_Flags = 0x07,
          No_Burst = 0x08,
          Closed_Caption = 0x09,
          Extended_Data_Services = 0x0a,
          Reserved = 0x0b..0xfe,
          Escape_to_ext_user_data_type = 0xff
        } user_data_type:8;
        if (user_data_type == Closed_Caption) {
          field cc_data:16;
          if(current_picture_coding_type != 3) {
            // I or P picture
            if(cc_buffer[0] != -1) {
              output cc_buffer[0]:16;
            }
          } else {
            cc_buffer[0] = cc_data;
          }
        }
      }
      return;
    } 
  }  // end of "switch(start_code)"
  while(!looking_at_bits(0x000001, 24)) {
    field "0x%02x" int byte_data:8;
  }
}

void picture_header(void)
{
  sequence_counter = 0;
  field "%d" int temporal_reference:10;
  field "<%s> (%d)" enum {
    forbidden,
    I_Picture,
    P_Picture,
    B_Picture,
    D_Picture_shall_not_be_used
  } picture_coding_type:3;
  current_picture_coding_type = picture_coding_type;
  while(1) {
    int start_code = search_word(START_CODE, 0xff);
    if(start_code >= 0x00000101 && start_code < 0x000001af) {
      {field int remove_start_code:32;}
      continue;
    }
    switch(start_code) {
    case PICTURE_START_CODE:
    case SEQUENCE_START_CODE:
    case SEQUENCE_END_CODE:
    case SEQUENCE_ERROR_START_CODE:
    case GROUP_START_CODE:
      return;
    case USER_DATA_START_CODE:
      {field int remove_start_code:32;}
      user_data(PICTURE_START_CODE);
      continue;
    default:
      {field int remove_start_code:32;}
      continue;
    }
  }
}