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.readUint8() !== shaka.cea.CeaDecoder.USA_COUNTRY_CODE) {
  121.       return;
  122.     }
  123.     if (reader.readUint16() !== shaka.cea.CeaDecoder.ATSC_PROVIDER_CODE) {
  124.       return;
  125.     }
  126.     if (reader.readUint32() !== shaka.cea.CeaDecoder.ATSC1_USER_IDENTIFIER) {
  127.       return;
  128.     }
  130.     // user_data_type_code: 0x03 - cc_data()
  131.     if (reader.readUint8() !== 0x03) {
  132.       return;
  133.     }
  135.     // 1 bit reserved
  136.     // 1 bit process_cc_data_flag
  137.     // 1 bit zero_bit
  138.     // 5 bits cc_count
  139.     const captionData = reader.readUint8();
  140.     // If process_cc_data_flag is not set, do not process this data.
  141.     if ((captionData & 0x40) === 0) {
  142.       return;
  143.     }
  145.     const count = captionData & 0x1f;
  147.     // 8 bits reserved
  148.     reader.skip(1);
  150.     for (let i = 0; i < count; i++) {
  151.       const cc = reader.readUint8();
  152.       // When ccValid is 0, the next two bytes should be discarded.
  153.       const ccValid = (cc & 0x04) >> 2;
  154.       const ccData1 = reader.readUint8();
  155.       const ccData2 = reader.readUint8();
  157.       if (ccValid) {
  158.         const ccType = cc & 0x03;
  159.         // Send the packet to the appropriate data array (CEA-608 or CEA-708).
  160.         if (ccType === shaka.cea.CeaDecoder.NTSC_CC_FIELD_1 ||
  161.             ccType === shaka.cea.CeaDecoder.NTSC_CC_FIELD_2) {
  162.           // CEA-608 NTSC (Line 21) Data.
  163.           this.cea608DataArray_.push({
  164.             pts,
  165.             type: ccType,
  166.             ccData1,
  167.             ccData2,
  168.             order: this.cea608DataArray_.length,
  169.           });
  170.         } else {
  171.           // CEA-708 DTVCC Data.
  172.           this.cea708DataArray_.push({
  173.             pts,
  174.             type: ccType,
  175.             value: ccData1,
  176.             order: this.cea708DataArray_.length,
  177.           });
  179.           // The second byte should always be labelled as DTVCC packet data.
  180.           // Even if this pair was a DTVCC packet start, only the first byte
  181.           // contains header info, and the second byte is just packet data.
  182.           this.cea708DataArray_.push({
  183.             pts,
  184.             type: shaka.cea.DtvccPacketBuilder.DTVCC_PACKET_DATA,
  185.             value: ccData2,
  186.             order: this.cea708DataArray_.length,
  187.           });
  188.         }
  189.       }
  190.     }
  191.   }
  193.   /**
  194.    * Decodes extracted closed caption data.
  195.    * @override
  196.    */
  197.   decode() {
  198.     /** @type {!Array.<!shaka.cea.ICaptionDecoder.ClosedCaption>} */
  199.     const parsedClosedCaptions = [];
  201.     // In some versions of Chrome, and other browsers, the default sorting
  202.     // algorithm isn't stable. This comparator sorts on presentation
  203.     // timestamp, and breaks ties on receive order (position in array).
  204.     const stableComparator =
  205.       (p1, p2) => (p1.pts - p2.pts) || (p1.order - p2.order);
  207.     this.cea608DataArray_.sort(stableComparator);
  208.     this.cea708DataArray_.sort(stableComparator);
  210.     // CEA-608 packets are just byte pairs. Decode all of them.
  211.     for (const cea608Packet of this.cea608DataArray_) {
  212.       const parsedClosedCaption = this.decodeCea608_(cea608Packet);
  213.       if (parsedClosedCaption) {
  214.         parsedClosedCaptions.push(parsedClosedCaption);
  215.       }
  216.     }
  218.     // CEA-708 packets are DTVCC packets composed of many byte pairs. Add all
  219.     // byte pairs to the packet builder, and process + clear any ready packets.
  220.     for (const cea708Byte of this.cea708DataArray_) {
  221.       this.dtvccPacketBuilder_.addByte(cea708Byte);
  222.     }
  223.     const dtvccPackets = this.dtvccPacketBuilder_.getBuiltPackets();
  224.     for (const dtvccPacket of dtvccPackets) {
  225.       const closedCaptions = this.decodeCea708_(dtvccPacket);
  226.       parsedClosedCaptions.push(...closedCaptions);
  227.     }
  229.     // Clear all processed data.
  230.     this.dtvccPacketBuilder_.clearBuiltPackets();
  231.     this.cea608DataArray_ = [];
  232.     this.cea708DataArray_ = [];
  234.     return parsedClosedCaptions;
  235.   }
  237.   /**
  238.    * Decodes a CEA-608 closed caption packet based on ANSI/CEA-608.
  239.    * @param {shaka.cea.Cea608DataChannel.Cea608Packet} ccPacket
  240.    * @return {?shaka.cea.ICaptionDecoder.ClosedCaption}
  241.    * @private
  242.    */
  243.   decodeCea608_(ccPacket) {
  244.     const fieldNum = ccPacket.type;
  246.     // If this packet is a control code, then it also sets the channel.
  247.     // For control codes, cc_data_1 has the form |P|0|0|1|C|X|X|X|.
  248.     // "C" is the channel bit. It indicates whether to set C2 active.
  249.     if (shaka.cea.Cea608DataChannel.isControlCode(ccPacket.ccData1)) {
  250.       const channelNum = (ccPacket.ccData1 >> 3) & 0x01; // Get channel bit.
  252.       // Change the stream based on the field, and the new channel
  253.       if (fieldNum === 0) {
  254.         this.currentField1Channel_ = channelNum;
  255.       } else {
  256.         this.currentField2Channel_ = channelNum;
  257.       }
  258.     }
  260.     // Get the correct stream for this caption packet (CC1, ..., CC4)
  261.     const selectedChannel = fieldNum ?
  262.         this.currentField2Channel_ : this.currentField1Channel_;
  263.     const selectedMode = `CC${(fieldNum << 1) | selectedChannel + 1}`;
  264.     const selectedStream = this.cea608ModeToStream_.get(selectedMode);
  266.     // Check for bad frames (bad pairs). This can be two 0xff, two 0x00, or any
  267.     // byte of even parity. ccData1 and ccData2 should be uint8 of odd parity.
  268.     if ((ccPacket.ccData1 === 0xff && ccPacket.ccData2 === 0xff) ||
  269.         (!ccPacket.ccData1 && !ccPacket.ccData2) ||
  270.         !this.isOddParity_(ccPacket.ccData1) ||
  271.         !this.isOddParity_(ccPacket.ccData2)) {
  272.       // Per CEA-608-B C.21, reset the memory after 45 consecutive bad frames.
  273.       if (++this.badFrames_ >= 45) {
  274.         this.reset();
  275.       }
  276.       return null;
  277.     }
  278.     this.badFrames_ = 0;
  280.     // Remove the MSB (parity bit).
  281.     ccPacket.ccData1 &= 0x7f;
  282.     ccPacket.ccData2 &= 0x7f;
  284.     // Check for empty captions and skip them.
  285.     if (!ccPacket.ccData1 && !ccPacket.ccData2) {
  286.       return null;
  287.     }
  289.     // Process the clean CC data pair.
  290.     let parsedClosedCaption = null;
  291.     if (shaka.cea.Cea608DataChannel.isControlCode(ccPacket.ccData1)) {
  292.       parsedClosedCaption = selectedStream.handleControlCode(ccPacket);
  293.     } else {
  294.       // Handle as a Basic North American Character.
  295.       selectedStream.handleBasicNorthAmericanChar(
  296.           ccPacket.ccData1, ccPacket.ccData2);
  297.     }
  299.     return parsedClosedCaption;
  300.   }
  302.   /**
  303.    * Decodes a CEA-708 DTVCC packet based on ANSI/CTA-708-E.
  304.    * @param {shaka.cea.DtvccPacket} dtvccPacket
  305.    * @return {!Array<!shaka.cea.ICaptionDecoder.ClosedCaption>}
  306.    * @private
  307.    */
  308.   decodeCea708_(dtvccPacket) {
  309.     const parsedClosedCaptions = [];
  310.     try {
  311.       while (dtvccPacket.hasMoreData()) {
  312.         // Process a service block.
  313.         const serviceBlockHeader = dtvccPacket.readByte().value;
  315.         // First 3 bits are service number, next 5 are block size,
  316.         // representing the number of bytes coming in this block
  317.         // (discluding a possible extended service block header byte)
  318.         let serviceNumber = (serviceBlockHeader & 0xe0) >> 5;
  319.         const blockSize = serviceBlockHeader & 0x1f;
  321.         if (serviceNumber === /* 0b111 */ 0x07 && blockSize != 0) {
  322.           // 2 bits null padding, 6 bits extended service number
  323.           const extendedServiceBlockHeader = dtvccPacket.readByte().value;
  324.           serviceNumber = extendedServiceBlockHeader & 0x3f;
  325.         }
  327.         // As per CEA-708-E, service number 0 is invalid, and should be ignored.
  328.         if (serviceNumber != 0) {
  329.           // If the service doesn't already exist, create it.
  330.           if (!this.serviceNumberToService_.has(serviceNumber)) {
  331.             const service = new shaka.cea.Cea708Service(serviceNumber);
  332.             this.serviceNumberToService_.set(serviceNumber, service);
  333.           }
  334.           const service = this.serviceNumberToService_.get(serviceNumber);
  336.           // Process all control codes.
  337.           const startPos = dtvccPacket.getPosition();
  339.           // Execute this loop `blockSize` times, to decode the control codes.
  340.           while (dtvccPacket.getPosition() - startPos < blockSize) {
  341.             const closedCaption = service.handleCea708ControlCode(dtvccPacket);
  342.             if (closedCaption) {
  343.               parsedClosedCaptions.push(closedCaption);
  344.             }
  345.           }  // position < end of block
  346.         }  // serviceNumber != 0
  347.       }  // hasMoreData
  348.     } catch (error) {
  349.       if (error instanceof shaka.util.Error &&
  350.       error.code === shaka.util.Error.Code.BUFFER_READ_OUT_OF_BOUNDS) {
  351.         shaka.log.warnOnce('CEA708_INVALID_DATA',
  352.             'Buffer read out of bounds / invalid CEA-708 Data.');
  353.       } else {
  354.         // This is an unexpected error, and should be rethrown.
  355.         throw error;
  356.       }
  357.     }
  358.     return parsedClosedCaptions;
  359.   }
  361.   /**
  362.    * Checks if a byte has odd parity (Odd number of 1s in binary).
  363.    * @param {number} byte
  364.    * @return {boolean} True if the byte has odd parity.
  365.    * @private
  366.    */
  367.   isOddParity_(byte) {
  368.     let parity = 0;
  369.     while (byte) {
  370.       parity ^= (byte & 1); // toggle parity if low bit is 1
  371.       byte >>= 1; // shift away the low bit
  372.     }
  373.     return parity === 1;
  374.   }
  375. };
  377. /**
  378.  * itu_t_35_provider_code for ATSC user_data
  379.  * @private @const {number}
  380.  */
  381. shaka.cea.CeaDecoder.ATSC_PROVIDER_CODE = 0x0031;
  383. /**
  384.  * When provider is ATSC user data, the ATSC_user_identifier code
  385.  * for ATSC1_data is "GA94" (0x47413934)
  386.  * @private @const {number}
  387.  */
  388. shaka.cea.CeaDecoder.ATSC1_USER_IDENTIFIER = 0x47413934;
  390. /**
  391.  * @private @const {number}
  392.  */
  393. shaka.cea.CeaDecoder.NTSC_CC_FIELD_1 = 0;
  395. /**
  396.  * @private @const {number}
  397.  */
  398. shaka.cea.CeaDecoder.NTSC_CC_FIELD_2 = 1;
  400. /**
  401.  * 0xB5 is USA's code (Rec. ITU-T T.35)
  402.  * @private @const {number}
  403.  */
  404. shaka.cea.CeaDecoder.USA_COUNTRY_CODE = 0xb5;