class Parser {}

Source

class Parser {
  private $handle; // Input stream.
  private $num_parsed_boxes = 0;
  private $data_was_skipped = false;
  public $features;

  function __construct( $handle ) {
    $this->handle   = $handle;
    $this->features = new Features();
  }

  /**
   * Parses an "ipco" box.
   *
   * "ispe" is used for width and height, "pixi" and "av1C" are used for bit depth
   * and number of channels, and "auxC" is used for alpha.
   *
   * @param stream  $handle              The resource the box will be parsed from.
   * @param int     $num_remaining_bytes The number of bytes that should be available from the resource.
   * @return Status                      FOUND on success or an error on failure.
   */
  private function parse_ipco( $num_remaining_bytes ) {
    $box_index = 1; // 1-based index. Used for iterating over properties.
    do {
      $box    = new Box();
      $status = $box->parse( $this->handle, $this->num_parsed_boxes, $num_remaining_bytes );
      if ( $status != FOUND ) {
        return $status;
      }

      if ( $box->type == 'ispe' ) {
        // See ISO/IEC 23008-12:2017(E) 6.5.3.2
        if ( $box->content_size < 8 ) {
          return INVALID;
        }
        if ( !( $data = read( $this->handle, 8 ) ) ) {
          return TRUNCATED;
        }
        $width  = read_big_endian( substr( $data, 0, 4 ), 4 );
        $height = read_big_endian( substr( $data, 4, 4 ), 4 );
        if ( $width == 0 || $height == 0 ) {
          return INVALID;
        }
        if ( count( $this->features->dim_props ) <= MAX_FEATURES &&
             $box_index <= MAX_VALUE ) {
          $dim_prop_count = count( $this->features->dim_props );
          $this->features->dim_props[$dim_prop_count]                 = new Dim_Prop();
          $this->features->dim_props[$dim_prop_count]->property_index = $box_index;
          $this->features->dim_props[$dim_prop_count]->width          = $width;
          $this->features->dim_props[$dim_prop_count]->height         = $height;
        } else {
          $this->data_was_skipped = true;
        }
        if ( !skip( $this->handle, $box->content_size - 8 ) ) {
          return TRUNCATED;
        }
      } else if ( $box->type == 'pixi' ) {
        // See ISO/IEC 23008-12:2017(E) 6.5.6.2
        if ( $box->content_size < 1 ) {
          return INVALID;
        }
        if ( !( $data = read( $this->handle, 1 ) ) ) {
          return TRUNCATED;
        }
        $num_channels = read_big_endian( $data, 1 );
        if ( $num_channels < 1 ) {
          return INVALID;
        }
        if ( $box->content_size < 1 + $num_channels ) {
          return INVALID;
        }
        if ( !( $data = read( $this->handle, 1 ) ) ) {
          return TRUNCATED;
        }
        $bit_depth = read_big_endian( $data, 1 );
        if ( $bit_depth < 1 ) {
          return INVALID;
        }
        for ( $i = 1; $i < $num_channels; ++$i ) {
          if ( !( $data = read( $this->handle, 1 ) ) ) {
            return TRUNCATED;
          }
          // Bit depth should be the same for all channels.
          if ( read_big_endian( $data, 1 ) != $bit_depth ) {
            return INVALID;
          }
          if ( $i > 32 ) {
            return ABORTED; // Be reasonable.
          }
        }
        if ( count( $this->features->chan_props ) <= MAX_FEATURES &&
             $box_index <= MAX_VALUE && $bit_depth <= MAX_VALUE &&
             $num_channels <= MAX_VALUE ) {
          $chan_prop_count = count( $this->features->chan_props );
          $this->features->chan_props[$chan_prop_count]                 = new Chan_Prop();
          $this->features->chan_props[$chan_prop_count]->property_index = $box_index;
          $this->features->chan_props[$chan_prop_count]->bit_depth      = $bit_depth;
          $this->features->chan_props[$chan_prop_count]->num_channels   = $num_channels;
        } else {
          $this->data_was_skipped = true;
        }
        if ( !skip( $this->handle, $box->content_size - ( 1 + $num_channels ) ) ) {
          return TRUNCATED;
        }
      } else if ( $box->type == 'av1C' ) {
        // See AV1 Codec ISO Media File Format Binding 2.3.1
        // at https://aomediacodec.github.io/av1-isobmff/#av1c
        // Only parse the necessary third byte. Assume that the others are valid.
        if ( $box->content_size < 3 ) {
          return INVALID;
        }
        if ( !( $data = read( $this->handle, 3 ) ) ) {
          return TRUNCATED;
        }
        $byte          = read_big_endian( substr( $data, 2, 1 ), 1 );
        $high_bitdepth = ( $byte & 0x40 ) != 0;
        $twelve_bit    = ( $byte & 0x20 ) != 0;
        $monochrome    = ( $byte & 0x10 ) != 0;
        if ( $twelve_bit && !$high_bitdepth ) {
            return INVALID;
        }
        if ( count( $this->features->chan_props ) <= MAX_FEATURES &&
             $box_index <= MAX_VALUE ) {
          $chan_prop_count = count( $this->features->chan_props );
          $this->features->chan_props[$chan_prop_count]                 = new Chan_Prop();
          $this->features->chan_props[$chan_prop_count]->property_index = $box_index;
          $this->features->chan_props[$chan_prop_count]->bit_depth      =
              $high_bitdepth ? $twelve_bit ? 12 : 10 : 8;
          $this->features->chan_props[$chan_prop_count]->num_channels   = $monochrome ? 1 : 3;
        } else {
          $this->data_was_skipped = true;
        }
        if ( !skip( $this->handle, $box->content_size - 3 ) ) {
          return TRUNCATED;
        }
      } else if ( $box->type == 'auxC' ) {
        // See AV1 Image File Format (AVIF) 4
        // at https://aomediacodec.github.io/av1-avif/#auxiliary-images
        $kAlphaStr       = "urn:mpeg:mpegB:cicp:systems:auxiliary:alpha\0";
        $kAlphaStrLength = 44; // Includes terminating character.
        if ( $box->content_size >= $kAlphaStrLength ) {
          if ( !( $data = read( $this->handle, $kAlphaStrLength ) ) ) {
            return TRUNCATED;
          }
          if ( substr( $data, 0, $kAlphaStrLength ) == $kAlphaStr ) {
            // Note: It is unlikely but it is possible that this alpha plane does
            //       not belong to the primary item or a tile. Ignore this issue.
            $this->features->has_alpha = true;
          }
          if ( !skip( $this->handle, $box->content_size - $kAlphaStrLength ) ) {
            return TRUNCATED;
          }
        } else {
          if ( !skip( $this->handle, $box->content_size ) ) {
            return TRUNCATED;
          }
        }
      } else {
        if ( !skip( $this->handle, $box->content_size ) ) {
          return TRUNCATED;
        }
      }
      ++$box_index;
      $num_remaining_bytes -= $box->size;
    } while ( $num_remaining_bytes > 0 );
    return NOT_FOUND;
  }

  /**
   * Parses an "iprp" box.
   *
   * The "ipco" box contain the properties which are linked to items by the "ipma" box.
   *
   * @param stream  $handle              The resource the box will be parsed from.
   * @param int     $num_remaining_bytes The number of bytes that should be available from the resource.
   * @return Status                      FOUND on success or an error on failure.
   */
  private function parse_iprp( $num_remaining_bytes ) {
    do {
      $box    = new Box();
      $status = $box->parse( $this->handle, $this->num_parsed_boxes, $num_remaining_bytes );
      if ( $status != FOUND ) {
        return $status;
      }

      if ( $box->type == 'ipco' ) {
        $status = $this->parse_ipco( $box->content_size );
        if ( $status != NOT_FOUND ) {
          return $status;
        }
      } else if ( $box->type == 'ipma' ) {
        // See ISO/IEC 23008-12:2017(E) 9.3.2
        $num_read_bytes = 4;
        if ( $box->content_size < $num_read_bytes ) {
          return INVALID;
        }
        if ( !( $data = read( $this->handle, $num_read_bytes ) ) ) {
          return TRUNCATED;
        }
        $entry_count        = read_big_endian( $data, 4 );
        $id_num_bytes       = ( $box->version < 1 ) ? 2 : 4;
        $index_num_bytes    = ( $box->flags & 1 ) ? 2 : 1;
        $essential_bit_mask = ( $box->flags & 1 ) ? 0x8000 : 0x80;

        for ( $entry = 0; $entry < $entry_count; ++$entry ) {
          if ( $entry >= MAX_PROPS ||
               count( $this->features->props ) >= MAX_PROPS ) {
            $this->data_was_skipped = true;
            break;
          }
          $num_read_bytes += $id_num_bytes + 1;
          if ( $box->content_size < $num_read_bytes ) {
            return INVALID;
          }
          if ( !( $data = read( $this->handle, $id_num_bytes + 1 ) ) ) {
            return TRUNCATED;
          }
          $item_id           = read_big_endian(
              substr( $data, 0, $id_num_bytes ), $id_num_bytes );
          $association_count = read_big_endian(
              substr( $data, $id_num_bytes, 1 ), 1 );

          for ( $property = 0; $property < $association_count; ++$property ) {
            if ( $property >= MAX_PROPS ||
                 count( $this->features->props ) >= MAX_PROPS ) {
              $this->data_was_skipped = true;
              break;
            }
            $num_read_bytes += $index_num_bytes;
            if ( $box->content_size < $num_read_bytes ) {
              return INVALID;
            }
            if ( !( $data = read( $this->handle, $index_num_bytes ) ) ) {
              return TRUNCATED;
            }
            $value          = read_big_endian( $data, $index_num_bytes );
            // $essential = ($value & $essential_bit_mask);  // Unused.
            $property_index = ( $value & ~$essential_bit_mask );
            if ( $property_index <= MAX_VALUE && $item_id <= MAX_VALUE ) {
              $prop_count = count( $this->features->props );
              $this->features->props[$prop_count]                 = new Prop();
              $this->features->props[$prop_count]->property_index = $property_index;
              $this->features->props[$prop_count]->item_id        = $item_id;
            } else {
              $this->data_was_skipped = true;
            }
          }
          if ( $property < $association_count ) {
            break; // Do not read garbage.
          }
        }

        // If all features are available now, do not look further.
        $status = $this->features->get_primary_item_features();
        if ( $status != NOT_FOUND ) {
          return $status;
        }

        // Mostly if 'data_was_skipped'.
        if ( !skip( $this->handle, $box->content_size - $num_read_bytes ) ) {
          return TRUNCATED;
        }
      } else {
        if ( !skip( $this->handle, $box->content_size ) ) {
          return TRUNCATED;
        }
      }
      $num_remaining_bytes -= $box->size;
    } while ( $num_remaining_bytes > 0 );
    return NOT_FOUND;
  }

  /**
   * Parses an "iref" box.
   *
   * The "dimg" boxes contain links between tiles and their parent items, which
   * can be used to infer bit depth and number of channels for the primary item
   * when the latter does not have these properties.
   *
   * @param stream  $handle              The resource the box will be parsed from.
   * @param int     $num_remaining_bytes The number of bytes that should be available from the resource.
   * @return Status                      FOUND on success or an error on failure.
   */
  private function parse_iref( $num_remaining_bytes ) {
    do {
      $box    = new Box();
      $status = $box->parse( $this->handle, $this->num_parsed_boxes, $num_remaining_bytes );
      if ( $status != FOUND ) {
        return $status;
      }

      if ( $box->type == 'dimg' ) {
        // See ISO/IEC 14496-12:2015(E) 8.11.12.2
        $num_bytes_per_id = ( $box->version == 0 ) ? 2 : 4;
        $num_read_bytes   = $num_bytes_per_id + 2;
        if ( $box->content_size < $num_read_bytes ) {
          return INVALID;
        }
        if ( !( $data = read( $this->handle, $num_read_bytes ) ) ) {
          return TRUNCATED;
        }
        $from_item_id    = read_big_endian( $data, $num_bytes_per_id );
        $reference_count = read_big_endian( substr( $data, $num_bytes_per_id, 2 ), 2 );

        for ( $i = 0; $i < $reference_count; ++$i ) {
          if ( $i >= MAX_TILES ) {
            $this->data_was_skipped = true;
            break;
          }
          $num_read_bytes += $num_bytes_per_id;
          if ( $box->content_size < $num_read_bytes ) {
            return INVALID;
          }
          if ( !( $data = read( $this->handle, $num_bytes_per_id ) ) ) {
            return TRUNCATED;
          }
          $to_item_id = read_big_endian( $data, $num_bytes_per_id );
          $tile_count = count( $this->features->tiles );
          if ( $from_item_id <= MAX_VALUE && $to_item_id <= MAX_VALUE &&
               $tile_count < MAX_TILES ) {
            $this->features->tiles[$tile_count]                 = new Tile();
            $this->features->tiles[$tile_count]->tile_item_id   = $to_item_id;
            $this->features->tiles[$tile_count]->parent_item_id = $from_item_id;
          } else {
            $this->data_was_skipped = true;
          }
        }

        // If all features are available now, do not look further.
        $status = $this->features->get_primary_item_features();
        if ( $status != NOT_FOUND ) {
          return $status;
        }

        // Mostly if 'data_was_skipped'.
        if ( !skip( $this->handle, $box->content_size - $num_read_bytes ) ) {
          return TRUNCATED;
        }
      } else {
        if ( !skip( $this->handle, $box->content_size ) ) {
          return TRUNCATED;
        }
      }
      $num_remaining_bytes -= $box->size;
    } while ( $num_remaining_bytes > 0 );
    return NOT_FOUND;
  }

  /**
   * Parses a "meta" box.
   *
   * It looks for the primary item ID in the "pitm" box and recurses into other boxes
   * to find its features.
   *
   * @param stream  $handle              The resource the box will be parsed from.
   * @param int     $num_remaining_bytes The number of bytes that should be available from the resource.
   * @return Status                      FOUND on success or an error on failure.
   */
  private function parse_meta( $num_remaining_bytes ) {
    do {
      $box    = new Box();
      $status = $box->parse( $this->handle, $this->num_parsed_boxes, $num_remaining_bytes );
      if ( $status != FOUND ) {
        return $status;
      }

      if ( $box->type == 'pitm' ) {
        // See ISO/IEC 14496-12:2015(E) 8.11.4.2
        $num_bytes_per_id = ( $box->version == 0 ) ? 2 : 4;
        if ( $num_bytes_per_id > $num_remaining_bytes ) {
          return INVALID;
        }
        if ( !( $data = read( $this->handle, $num_bytes_per_id ) ) ) {
          return TRUNCATED;
        }
        $primary_item_id = read_big_endian( $data, $num_bytes_per_id );
        if ( $primary_item_id > MAX_VALUE ) {
          return ABORTED;
        }
        $this->features->has_primary_item = true;
        $this->features->primary_item_id  = $primary_item_id;
        if ( !skip( $this->handle, $box->content_size - $num_bytes_per_id ) ) {
          return TRUNCATED;
        }
      } else if ( $box->type == 'iprp' ) {
        $status = $this->parse_iprp( $box->content_size );
        if ( $status != NOT_FOUND ) {
          return $status;
        }
      } else if ( $box->type == 'iref' ) {
        $status = $this->parse_iref( $box->content_size );
        if ( $status != NOT_FOUND ) {
          return $status;
        }
      } else {
        if ( !skip( $this->handle, $box->content_size ) ) {
          return TRUNCATED;
        }
      }
      $num_remaining_bytes -= $box->size;
    } while ( $num_remaining_bytes != 0 );
    // According to ISO/IEC 14496-12:2012(E) 8.11.1.1 there is at most one "meta".
    return INVALID;
  }

  /**
   * Parses a file stream.
   *
   * The file type is checked through the "ftyp" box.
   *
   * @return bool True if the input stream is an AVIF bitstream or false.
   */
  public function parse_ftyp() {
    $box    = new Box();
    $status = $box->parse( $this->handle, $this->num_parsed_boxes );
    if ( $status != FOUND ) {
      return false;
    }

    if ( $box->type != 'ftyp' ) {
      return false;
    }
    // Iterate over brands. See ISO/IEC 14496-12:2012(E) 4.3.1
    if ( $box->content_size < 8 ) {
      return false;
    }
    for ( $i = 0; $i + 4 <= $box->content_size; $i += 4 ) {
      if ( !( $data = read( $this->handle, 4 ) ) ) {
        return false;
      }
      if ( $i == 4 ) {
        continue; // Skip minor_version.
      }
      if ( substr( $data, 0, 4 ) == 'avif' || substr( $data, 0, 4 ) == 'avis' ) {
        return skip( $this->handle, $box->content_size - ( $i + 4 ) );
      }
      if ( $i > 32 * 4 ) {
        return false; // Be reasonable.
      }

    }
    return false; // No AVIF brand no good.
  }

  /**
   * Parses a file stream.
   *
   * Features are extracted from the "meta" box.
   *
   * @return bool True if the main features of the primary item were parsed or false.
   */
  public function parse_file() {
    $box = new Box();
    while ( $box->parse( $this->handle, $this->num_parsed_boxes ) == FOUND ) {
      if ( $box->type === 'meta' ) {
        if ( $this->parse_meta( $box->content_size ) != FOUND ) {
          return false;
        }
        return true;
      }
      if ( !skip( $this->handle, $box->content_size ) ) {
        return false;
      }
    }
    return false; // No "meta" no good.
  }
}

View all references View on Trac View on GitHub