Source: lib/cea/cea_decoder.js

  1. /*! @license
  2.  * Shaka Player
  3.  * Copyright 2016 Google LLC
  4.  * SPDX-License-Identifier: Apache-2.0
  5.  */
  7. goog.provide('shaka.cea.CeaDecoder');
  9. goog.require('shaka.cea.Cea608DataChannel');
  10. goog.require('shaka.cea.Cea708Service');
  11. goog.require('shaka.cea.DtvccPacketBuilder');
  12. goog.require('shaka.cea.ICaptionDecoder');
  13. goog.require('shaka.log');
  14. goog.require('shaka.util.DataViewReader');
  15. goog.require('shaka.util.Error');
  16. goog.requireType('shaka.cea.DtvccPacket');
  19. /**
  20.  * CEA-X08 captions decoder. Currently only CEA-608 supported.
  21.  * @implements {shaka.cea.ICaptionDecoder}
  22.  */
  23. shaka.cea.CeaDecoder = class {
  24.   /** */
  25.   constructor() {
  26.     /**
  27.      * An array of CEA-608 closed caption data extracted for decoding.
  28.      * @private {!Array<!shaka.cea.Cea608DataChannel.Cea608Packet>}
  29.      */
  30.     this.cea608DataArray_ = [];
  32.     /**
  33.      * An array of CEA-708 closed caption data extracted for decoding.
  34.      * @private {!Array<!shaka.cea.Cea708Service.Cea708Byte>}
  35.      */
  36.     this.cea708DataArray_ = [];
  38.     /**
  39.      * A DTVCC Packet builder for CEA-708 data.
  40.      * @private {!shaka.cea.DtvccPacketBuilder}
  41.      */
  42.     this.dtvccPacketBuilder_ = new shaka.cea.DtvccPacketBuilder();
  44.     /**
  45.      * Number of consecutive bad frames decoded on CEA-608.
  46.      * @private {number}
  47.      */
  48.     this.badFrames_ = 0;
  50.     /**
  51.      * A map containing the stream for each mode.
  52.      * @private {!Map<string, !shaka.cea.Cea608DataChannel>}
  53.      */
  54.     this.cea608ModeToStream_ = new Map([
  55.       ['CC1', new shaka.cea.Cea608DataChannel(0, 0)], // F1 + C1 -> CC1
  56.       ['CC2', new shaka.cea.Cea608DataChannel(0, 1)], // F1 + C2 -> CC2
  57.       ['CC3', new shaka.cea.Cea608DataChannel(1, 0)], // F2 + C1 -> CC3
  58.       ['CC4', new shaka.cea.Cea608DataChannel(1, 1)], // F2 + C2 -> CC4
  59.     ]);
  61.     /**
  62.      * The current channel that is active on CEA-608 field 1.
  63.      * @private {number}
  64.      */
  65.     this.currentField1Channel_ = 0;
  67.     /**
  68.      * The current channel that is active on CEA-608 field 2.
  69.      * @private {number}
  70.      */
  71.     this.currentField2Channel_ = 0;
  73.     /**
  74.      * Map of service number to CEA-708 services, initially empty. Since there
  75.      * can be up to 63 services, they are created dynamically only when needed.
  76.      * @private {!Map<number, shaka.cea.Cea708Service>}
  77.      */
  78.     this.serviceNumberToService_ = new Map();
  80.     this.reset();
  81.   }
  83.   /**
  84.    * Clears the decoder.
  85.    * @override
  86.    */
  87.   clear() {
  88.     this.badFrames_ = 0;
  89.     this.cea608DataArray_ = [];
  90.     this.cea708DataArray_ = [];
  91.     this.dtvccPacketBuilder_.clear();
  92.     this.reset();
  94.     // Clear all the CEA-708 services.
  95.     for (const service of this.serviceNumberToService_.values()) {
  96.       service.clear();
  97.     }
  98.   }
  100.   /**
  101.    * Resets the decoder.
  102.    */
  103.   reset() {
  104.     this.currentField1Channel_ = 0;
  105.     this.currentField2Channel_ = 0;
  106.     for (const stream of this.cea608ModeToStream_.values()) {
  107.       stream.reset();
  108.     }
  109.   }
  111.   /**
  112.    * Extracts closed caption bytes from CEA-X08 packets from the stream based on
  113.    * ANSI/SCTE 128 and A/53, Part 4.
  114.    * @override
  115.    */
  116.   extract(userDataSeiMessage, pts) {
  117.     const reader = new shaka.util.DataViewReader(
  118.         userDataSeiMessage, shaka.util.DataViewReader.Endianness.BIG_ENDIAN);
  120.     if (reader.getLength() < shaka.cea.CeaDecoder.MIN_LENGTH) {
  121.       return;
  122.     }
  123.     if (reader.readUint8() !== shaka.cea.CeaDecoder.USA_COUNTRY_CODE) {
  124.       return;
  125.     }
  126.     if (reader.readUint16() !== shaka.cea.CeaDecoder.ATSC_PROVIDER_CODE) {
  127.       return;
  128.     }
  129.     if (reader.readUint32() !== shaka.cea.CeaDecoder.ATSC1_USER_IDENTIFIER) {
  130.       return;
  131.     }
  133.     // user_data_type_code: 0x03 - cc_data()
  134.     if (reader.readUint8() !== 0x03) {
  135.       return;
  136.     }
  138.     // 1 bit reserved
  139.     // 1 bit process_cc_data_flag
  140.     // 1 bit zero_bit
  141.     // 5 bits cc_count
  142.     const captionData = reader.readUint8();
  143.     // If process_cc_data_flag is not set, do not process this data.
  144.     if ((captionData & 0x40) === 0) {
  145.       return;
  146.     }
  148.     const count = captionData & 0x1f;
  150.     // 8 bits reserved
  151.     reader.skip(1);
  153.     for (let i = 0; i < count; i++) {
  154.       const cc = reader.readUint8();
  155.       // When ccValid is 0, the next two bytes should be discarded.
  156.       const ccValid = (cc & 0x04) >> 2;
  157.       const ccData1 = reader.readUint8();
  158.       const ccData2 = reader.readUint8();
  160.       if (ccValid) {
  161.         const ccType = cc & 0x03;
  162.         // Send the packet to the appropriate data array (CEA-608 or CEA-708).
  163.         if (ccType === shaka.cea.CeaDecoder.NTSC_CC_FIELD_1 ||
  164.             ccType === shaka.cea.CeaDecoder.NTSC_CC_FIELD_2) {
  165.           // CEA-608 NTSC (Line 21) Data.
  166.           this.cea608DataArray_.push({
  167.             pts,
  168.             type: ccType,
  169.             ccData1,
  170.             ccData2,
  171.             order: this.cea608DataArray_.length,
  172.           });
  173.         } else {
  174.           // CEA-708 DTVCC Data.
  175.           this.cea708DataArray_.push({
  176.             pts,
  177.             type: ccType,
  178.             value: ccData1,
  179.             order: this.cea708DataArray_.length,
  180.           });
  182.           // The second byte should always be labelled as DTVCC packet data.
  183.           // Even if this pair was a DTVCC packet start, only the first byte
  184.           // contains header info, and the second byte is just packet data.
  185.           this.cea708DataArray_.push({
  186.             pts,
  187.             type: shaka.cea.DtvccPacketBuilder.DTVCC_PACKET_DATA,
  188.             value: ccData2,
  189.             order: this.cea708DataArray_.length,
  190.           });
  191.         }
  192.       }
  193.     }
  194.   }
  196.   /**
  197.    * Decodes extracted closed caption data.
  198.    * @override
  199.    */
  200.   decode() {
  201.     /** @type {!Array.<!shaka.cea.ICaptionDecoder.ClosedCaption>} */
  202.     const parsedClosedCaptions = [];
  204.     // In some versions of Chrome, and other browsers, the default sorting
  205.     // algorithm isn't stable. This comparator sorts on presentation
  206.     // timestamp, and breaks ties on receive order (position in array).
  207.     const stableComparator =
  208.       (p1, p2) => (p1.pts - p2.pts) || (p1.order - p2.order);
  210.     this.cea608DataArray_.sort(stableComparator);
  211.     this.cea708DataArray_.sort(stableComparator);
  213.     // CEA-608 packets are just byte pairs. Decode all of them.
  214.     for (const cea608Packet of this.cea608DataArray_) {
  215.       const parsedClosedCaption = this.decodeCea608_(cea608Packet);
  216.       if (parsedClosedCaption) {
  217.         parsedClosedCaptions.push(parsedClosedCaption);
  218.       }
  219.     }
  221.     // CEA-708 packets are DTVCC packets composed of many byte pairs. Add all
  222.     // byte pairs to the packet builder, and process + clear any ready packets.
  223.     for (const cea708Byte of this.cea708DataArray_) {
  224.       this.dtvccPacketBuilder_.addByte(cea708Byte);
  225.     }
  226.     const dtvccPackets = this.dtvccPacketBuilder_.getBuiltPackets();
  227.     for (const dtvccPacket of dtvccPackets) {
  228.       const closedCaptions = this.decodeCea708_(dtvccPacket);
  229.       parsedClosedCaptions.push(...closedCaptions);
  230.     }
  232.     // Clear all processed data.
  233.     this.dtvccPacketBuilder_.clearBuiltPackets();
  234.     this.cea608DataArray_ = [];
  235.     this.cea708DataArray_ = [];
  237.     return parsedClosedCaptions;
  238.   }
  240.   /**
  241.    * Decodes a CEA-608 closed caption packet based on ANSI/CEA-608.
  242.    * @param {shaka.cea.Cea608DataChannel.Cea608Packet} ccPacket
  243.    * @return {?shaka.cea.ICaptionDecoder.ClosedCaption}
  244.    * @private
  245.    */
  246.   decodeCea608_(ccPacket) {
  247.     const fieldNum = ccPacket.type;
  249.     // If this packet is a control code, then it also sets the channel.
  250.     // For control codes, cc_data_1 has the form |P|0|0|1|C|X|X|X|.
  251.     // "C" is the channel bit. It indicates whether to set C2 active.
  252.     if (shaka.cea.Cea608DataChannel.isControlCode(ccPacket.ccData1)) {
  253.       const channelNum = (ccPacket.ccData1 >> 3) & 0x01; // Get channel bit.
  255.       // Change the stream based on the field, and the new channel
  256.       if (fieldNum === 0) {
  257.         this.currentField1Channel_ = channelNum;
  258.       } else {
  259.         this.currentField2Channel_ = channelNum;
  260.       }
  261.     }
  263.     // Get the correct stream for this caption packet (CC1, ..., CC4)
  264.     const selectedChannel = fieldNum ?
  265.         this.currentField2Channel_ : this.currentField1Channel_;
  266.     const selectedMode = `CC${((fieldNum << 1) | selectedChannel) + 1}`;
  267.     const selectedStream = this.cea608ModeToStream_.get(selectedMode);
  269.     // Check for bad frames (bad pairs). This can be two 0xff, two 0x00, or any
  270.     // byte of even parity. ccData1 and ccData2 should be uint8 of odd parity.
  271.     if ((ccPacket.ccData1 === 0xff && ccPacket.ccData2 === 0xff) ||
  272.         (!ccPacket.ccData1 && !ccPacket.ccData2) ||
  273.         !this.isOddParity_(ccPacket.ccData1) ||
  274.         !this.isOddParity_(ccPacket.ccData2)) {
  275.       // Per CEA-608-B C.21, reset the memory after 45 consecutive bad frames.
  276.       if (++this.badFrames_ >= 45) {
  277.         this.reset();
  278.       }
  279.       return null;
  280.     }
  281.     this.badFrames_ = 0;
  283.     // Remove the MSB (parity bit).
  284.     ccPacket.ccData1 &= 0x7f;
  285.     ccPacket.ccData2 &= 0x7f;
  287.     // Check for empty captions and skip them.
  288.     if (!ccPacket.ccData1 && !ccPacket.ccData2) {
  289.       return null;
  290.     }
  292.     // Process the clean CC data pair.
  293.     let parsedClosedCaption = null;
  294.     if (shaka.cea.Cea608DataChannel.isControlCode(ccPacket.ccData1)) {
  295.       parsedClosedCaption = selectedStream.handleControlCode(ccPacket);
  296.     } else {
  297.       // Handle as a Basic North American Character.
  298.       selectedStream.handleBasicNorthAmericanChar(
  299.           ccPacket.ccData1, ccPacket.ccData2);
  300.     }
  302.     return parsedClosedCaption;
  303.   }
  305.   /**
  306.    * Decodes a CEA-708 DTVCC packet based on ANSI/CTA-708-E.
  307.    * @param {shaka.cea.DtvccPacket} dtvccPacket
  308.    * @return {!Array<!shaka.cea.ICaptionDecoder.ClosedCaption>}
  309.    * @private
  310.    */
  311.   decodeCea708_(dtvccPacket) {
  312.     const parsedClosedCaptions = [];
  313.     try {
  314.       while (dtvccPacket.hasMoreData()) {
  315.         // Process a service block.
  316.         const serviceBlockHeader = dtvccPacket.readByte().value;
  318.         // First 3 bits are service number, next 5 are block size,
  319.         // representing the number of bytes coming in this block
  320.         // (discluding a possible extended service block header byte)
  321.         let serviceNumber = (serviceBlockHeader & 0xe0) >> 5;
  322.         const blockSize = serviceBlockHeader & 0x1f;
  324.         if (serviceNumber === /* 0b111 */ 0x07 && blockSize != 0) {
  325.           // 2 bits null padding, 6 bits extended service number
  326.           const extendedServiceBlockHeader = dtvccPacket.readByte().value;
  327.           serviceNumber = extendedServiceBlockHeader & 0x3f;
  328.         }
  330.         // As per CEA-708-E, service number 0 is invalid, and should be ignored.
  331.         if (serviceNumber != 0) {
  332.           // If the service doesn't already exist, create it.
  333.           if (!this.serviceNumberToService_.has(serviceNumber)) {
  334.             const service = new shaka.cea.Cea708Service(serviceNumber);
  335.             this.serviceNumberToService_.set(serviceNumber, service);
  336.           }
  337.           const service = this.serviceNumberToService_.get(serviceNumber);
  339.           // Process all control codes.
  340.           const startPos = dtvccPacket.getPosition();
  342.           // Execute this loop `blockSize` times, to decode the control codes.
  343.           while (dtvccPacket.getPosition() - startPos < blockSize) {
  344.             const closedCaption = service.handleCea708ControlCode(dtvccPacket);
  345.             if (closedCaption) {
  346.               parsedClosedCaptions.push(closedCaption);
  347.             }
  348.           }  // position < end of block
  349.         }  // serviceNumber != 0
  350.       }  // hasMoreData
  351.     } catch (error) {
  352.       if (error instanceof shaka.util.Error &&
  353.       error.code === shaka.util.Error.Code.BUFFER_READ_OUT_OF_BOUNDS) {
  354.         shaka.log.warnOnce('CEA708_INVALID_DATA',
  355.             'Buffer read out of bounds / invalid CEA-708 Data.');
  356.       } else {
  357.         // This is an unexpected error, and should be rethrown.
  358.         throw error;
  359.       }
  360.     }
  361.     return parsedClosedCaptions;
  362.   }
  364.   /**
  365.    * Checks if a byte has odd parity (Odd number of 1s in binary).
  366.    * @param {number} byte
  367.    * @return {boolean} True if the byte has odd parity.
  368.    * @private
  369.    */
  370.   isOddParity_(byte) {
  371.     let parity = 0;
  372.     while (byte) {
  373.       parity ^= (byte & 1); // toggle parity if low bit is 1
  374.       byte >>= 1; // shift away the low bit
  375.     }
  376.     return parity === 1;
  377.   }
  378. };
  380. /**
  381.  * itu_t_35_provider_code for ATSC user_data
  382.  * @private @const {number}
  383.  */
  384. shaka.cea.CeaDecoder.ATSC_PROVIDER_CODE = 0x0031;
  386. /**
  387.  * When provider is ATSC user data, the ATSC_user_identifier code
  388.  * for ATSC1_data is "GA94" (0x47413934)
  389.  * @private @const {number}
  390.  */
  391. shaka.cea.CeaDecoder.ATSC1_USER_IDENTIFIER = 0x47413934;
  393. /**
  394.  * @private @const {number}
  395.  */
  396. shaka.cea.CeaDecoder.NTSC_CC_FIELD_1 = 0;
  398. /**
  399.  * @private @const {number}
  400.  */
  401. shaka.cea.CeaDecoder.NTSC_CC_FIELD_2 = 1;
  403. /**
  404.  * 0xB5 is USA's code (Rec. ITU-T T.35)
  405.  * @private @const {number}
  406.  */
  407. shaka.cea.CeaDecoder.USA_COUNTRY_CODE = 0xb5;
  409. /**
  410.  * Caption packet min length
  411.  * Country Code + ATSC_PROVIDER_CODE + ATSC_1_USER_IDENTIFIER + USER_DATA_TYPE
  412.  * @private @const {number}
  413.  */
  414. shaka.cea.CeaDecoder.MIN_LENGTH = 8;