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.Cea708Service');
  10. goog.require('shaka.cea.Cea608DataChannel');
  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');
  17. goog.requireType('shaka.cea.ICaptionDecoder');
  20. /**
  21.  * CEA-X08 captions decoder. Currently only CEA-608 supported.
  22.  * @implements {shaka.cea.ICaptionDecoder}
  23.  */
  24. shaka.cea.CeaDecoder = class {
  25.   /** */
  26.   constructor() {
  27.     /**
  28.      * An array of CEA-608 closed caption data extracted for decoding.
  29.      * @private {!Array<!shaka.cea.Cea608DataChannel.Cea608Packet>}
  30.      */
  31.     this.cea608DataArray_ = [];
  33.     /**
  34.      * An array of CEA-708 closed caption data extracted for decoding.
  35.      * @private {!Array<!shaka.cea.Cea708Service.Cea708Byte>}
  36.      */
  37.     this.cea708DataArray_ = [];
  39.     /**
  40.      * A DTVCC Packet builder for CEA-708 data.
  41.      * @private {!shaka.cea.DtvccPacketBuilder}
  42.      */
  43.     this.dtvccPacketBuilder_ = new shaka.cea.DtvccPacketBuilder();
  45.     /**
  46.      * Number of consecutive bad frames decoded on CEA-608.
  47.      * @private {!number}
  48.      */
  49.     this.badFrames_ = 0;
  51.     /**
  52.      * A map containing the stream for each mode.
  53.      * @private {!Map<!string, !shaka.cea.Cea608DataChannel>}
  54.      */
  55.     this.cea608ModeToStream_ = new Map([
  56.       ['CC1', new shaka.cea.Cea608DataChannel(0, 0)], // F1 + C1 -> CC1
  57.       ['CC2', new shaka.cea.Cea608DataChannel(0, 1)], // F1 + C2 -> CC2
  58.       ['CC3', new shaka.cea.Cea608DataChannel(1, 0)], // F2 + C1 -> CC3
  59.       ['CC4', new shaka.cea.Cea608DataChannel(1, 1)], // F2 + C2 -> CC4
  60.     ]);
  62.     /**
  63.      * The current channel that is active on CEA-608 field 1.
  64.      * @private {!number}
  65.      */
  66.     this.currentField1Channel_ = 0;
  68.     /**
  69.      * The current channel that is active on CEA-608 field 2.
  70.      * @private {!number}
  71.      */
  72.     this.currentField2Channel_ = 0;
  74.     /**
  75.      * Map of service number to CEA-708 services, initially empty. Since there
  76.      * can be up to 63 services, they are created dynamically only when needed.
  77.      * @private {!Map<!number, shaka.cea.Cea708Service>}
  78.      */
  79.     this.serviceNumberToService_ = new Map();
  81.     this.reset();
  82.   }
  84.   /**
  85.    * Clears the decoder.
  86.    * @override
  87.    */
  88.   clear() {
  89.     this.badFrames_ = 0;
  90.     this.cea608DataArray_ = [];
  91.     this.cea708DataArray_ = [];
  92.     this.dtvccPacketBuilder_.clear();
  93.     this.reset();
  95.     // Clear all the CEA-708 services.
  96.     for (const service of this.serviceNumberToService_.values()) {
  97.       service.clear();
  98.     }
  99.   }
  101.   /**
  102.    * Resets the decoder.
  103.    */
  104.   reset() {
  105.     this.currentField1Channel_ = 0;
  106.     this.currentField2Channel_ = 0;
  107.     for (const stream of this.cea608ModeToStream_.values()) {
  108.       stream.reset();
  109.     }
  110.   }
  112.   /**
  113.    * Extracts closed caption bytes from CEA-X08 packets from the stream based on
  114.    * ANSI/SCTE 128 and A/53, Part 4.
  115.    * @override
  116.    */
  117.   extract(userDataSeiMessage, pts) {
  118.     const reader = new shaka.util.DataViewReader(
  119.         userDataSeiMessage, shaka.util.DataViewReader.Endianness.BIG_ENDIAN);
  121.     if (reader.readUint8() !== shaka.cea.CeaDecoder.USA_COUNTRY_CODE) {
  122.       return;
  123.     }
  124.     if (reader.readUint16() !== shaka.cea.CeaDecoder.ATSC_PROVIDER_CODE) {
  125.       return;
  126.     }
  127.     if (reader.readUint32() !== shaka.cea.CeaDecoder.ATSC1_USER_IDENTIFIER) {
  128.       return;
  129.     }
  131.     // user_data_type_code: 0x03 - cc_data()
  132.     if (reader.readUint8() !== 0x03) {
  133.       return;
  134.     }
  136.     // 1 bit reserved
  137.     // 1 bit process_cc_data_flag
  138.     // 1 bit zero_bit
  139.     // 5 bits cc_count
  140.     const captionData = reader.readUint8();
  141.     // If process_cc_data_flag is not set, do not process this data.
  142.     if ((captionData & 0x40) === 0) {
  143.       return;
  144.     }
  146.     const count = captionData & 0x1f;
  148.     // 8 bits reserved
  149.     reader.skip(1);
  151.     for (let i = 0; i < count; i++) {
  152.       const cc = reader.readUint8();
  153.       // When ccValid is 0, the next two bytes should be discarded.
  154.       const ccValid = (cc & 0x04) >> 2;
  155.       const ccData1 = reader.readUint8();
  156.       const ccData2 = reader.readUint8();
  158.       if (ccValid) {
  159.         const ccType = cc & 0x03;
  160.         // Send the packet to the appropriate data array (CEA-608 or CEA-708).
  161.         if (ccType === shaka.cea.CeaDecoder.NTSC_CC_FIELD_1 ||
  162.             ccType === shaka.cea.CeaDecoder.NTSC_CC_FIELD_2) {
  163.           // CEA-608 NTSC (Line 21) Data.
  164.           this.cea608DataArray_.push({
  165.             pts,
  166.             type: ccType,
  167.             ccData1,
  168.             ccData2,
  169.             order: this.cea608DataArray_.length,
  170.           });
  171.         } else {
  172.           // CEA-708 DTVCC Data.
  173.           this.cea708DataArray_.push({
  174.             pts,
  175.             type: ccType,
  176.             value: ccData1,
  177.             order: this.cea708DataArray_.length,
  178.           });
  180.           // The second byte should always be labelled as DTVCC packet data.
  181.           // Even if this pair was a DTVCC packet start, only the first byte
  182.           // contains header info, and the second byte is just packet data.
  183.           this.cea708DataArray_.push({
  184.             pts,
  185.             type: shaka.cea.DtvccPacketBuilder.DTVCC_PACKET_DATA,
  186.             value: ccData2,
  187.             order: this.cea708DataArray_.length,
  188.           });
  189.         }
  190.       }
  191.     }
  192.   }
  194.   /**
  195.    * Decodes extracted closed caption data.
  196.    * @override
  197.    */
  198.   decode() {
  199.     /** @type {!Array.<!shaka.cea.ICaptionDecoder.ClosedCaption>} */
  200.     const parsedClosedCaptions = [];
  202.     // In some versions of Chrome, and other browsers, the default sorting
  203.     // algorithm isn't stable. This comparator sorts on presentation
  204.     // timestamp, and breaks ties on receive order (position in array).
  205.     const stableComparator =
  206.       (p1, p2) => (p1.pts - p2.pts) || (p1.order - p2.order);
  208.     this.cea608DataArray_.sort(stableComparator);
  209.     this.cea708DataArray_.sort(stableComparator);
  211.     // CEA-608 packets are just byte pairs. Decode all of them.
  212.     for (const cea608Packet of this.cea608DataArray_) {
  213.       const parsedClosedCaption = this.decodeCea608_(cea608Packet);
  214.       if (parsedClosedCaption) {
  215.         parsedClosedCaptions.push(parsedClosedCaption);
  216.       }
  217.     }
  219.     // CEA-708 packets are DTVCC packets composed of many byte pairs. Add all
  220.     // byte pairs to the packet builder, and process + clear any ready packets.
  221.     for (const cea708Byte of this.cea708DataArray_) {
  222.       this.dtvccPacketBuilder_.addByte(cea708Byte);
  223.     }
  224.     const dtvccPackets = this.dtvccPacketBuilder_.getBuiltPackets();
  225.     for (const dtvccPacket of dtvccPackets) {
  226.       const closedCaptions = this.decodeCea708_(dtvccPacket);
  227.       parsedClosedCaptions.push(...closedCaptions);
  228.     }
  230.     // Clear all processed data.
  231.     this.dtvccPacketBuilder_.clearBuiltPackets();
  232.     this.cea608DataArray_ = [];
  233.     this.cea708DataArray_ = [];
  235.     return parsedClosedCaptions;
  236.   }
  238.   /**
  239.    * Decodes a CEA-608 closed caption packet based on ANSI/CEA-608.
  240.    * @param {shaka.cea.Cea608DataChannel.Cea608Packet} ccPacket
  241.    * @return {?shaka.cea.ICaptionDecoder.ClosedCaption}
  242.    * @private
  243.    */
  244.   decodeCea608_(ccPacket) {
  245.     const fieldNum = ccPacket.type;
  247.     // If this packet is a control code, then it also sets the channel.
  248.     // For control codes, cc_data_1 has the form |P|0|0|1|C|X|X|X|.
  249.     // "C" is the channel bit. It indicates whether to set C2 active.
  250.     if (shaka.cea.Cea608DataChannel.isControlCode(ccPacket.ccData1)) {
  251.       const channelNum = (ccPacket.ccData1 >> 3) & 0x01; // Get channel bit.
  253.       // Change the stream based on the field, and the new channel
  254.       if (fieldNum === 0) {
  255.         this.currentField1Channel_ = channelNum;
  256.       } else {
  257.         this.currentField2Channel_ = channelNum;
  258.       }
  259.     }
  261.     // Get the correct stream for this caption packet (CC1, ..., CC4)
  262.     const selectedChannel = fieldNum ?
  263.         this.currentField2Channel_ : this.currentField1Channel_;
  264.     const selectedMode = `CC${(fieldNum << 1) | selectedChannel + 1}`;
  265.     const selectedStream = this.cea608ModeToStream_.get(selectedMode);
  267.     // Check for bad frames (bad pairs). This can be two 0xff, two 0x00, or any
  268.     // byte of even parity. ccData1 and ccData2 should be uint8 of odd parity.
  269.     if ((ccPacket.ccData1 === 0xff && ccPacket.ccData2 === 0xff) ||
  270.         (!ccPacket.ccData1 && !ccPacket.ccData2) ||
  271.         !this.isOddParity_(ccPacket.ccData1) ||
  272.         !this.isOddParity_(ccPacket.ccData2)) {
  273.       // Per CEA-608-B C.21, reset the memory after 45 consecutive bad frames.
  274.       if (++this.badFrames_ >= 45) {
  275.         this.reset();
  276.       }
  277.       return null;
  278.     }
  279.     this.badFrames_ = 0;
  281.     // Remove the MSB (parity bit).
  282.     ccPacket.ccData1 &= 0x7f;
  283.     ccPacket.ccData2 &= 0x7f;
  285.     // Check for empty captions and skip them.
  286.     if (!ccPacket.ccData1 && !ccPacket.ccData2) {
  287.       return null;
  288.     }
  290.     // Process the clean CC data pair.
  291.     let parsedClosedCaption = null;
  292.     if (shaka.cea.Cea608DataChannel.isControlCode(ccPacket.ccData1)) {
  293.       parsedClosedCaption = selectedStream.handleControlCode(ccPacket);
  294.     } else {
  295.       // Handle as a Basic North American Character.
  296.       selectedStream.handleBasicNorthAmericanChar(
  297.           ccPacket.ccData1, ccPacket.ccData2);
  298.     }
  300.     return parsedClosedCaption;
  301.   }
  303.   /**
  304.    * Decodes a CEA-708 DTVCC packet based on ANSI/CTA-708-E.
  305.    * @param {shaka.cea.DtvccPacket} dtvccPacket
  306.    * @return {!Array<!shaka.cea.ICaptionDecoder.ClosedCaption>}
  307.    * @private
  308.    */
  309.   decodeCea708_(dtvccPacket) {
  310.     const parsedClosedCaptions = [];
  311.     try {
  312.       while (dtvccPacket.hasMoreData()) {
  313.         // Process a service block.
  314.         const serviceBlockHeader = dtvccPacket.readByte().value;
  316.         // First 3 bits are service number, next 5 are block size,
  317.         // representing the number of bytes coming in this block
  318.         // (discluding a possible extended service block header byte)
  319.         let serviceNumber = (serviceBlockHeader & 0xe0) >> 5;
  320.         const blockSize = serviceBlockHeader & 0x1f;
  322.         if (serviceNumber === /* 0b111 */ 0x07 && blockSize != 0) {
  323.           // 2 bits null padding, 6 bits extended service number
  324.           const extendedServiceBlockHeader = dtvccPacket.readByte().value;
  325.           serviceNumber = extendedServiceBlockHeader & 0x3f;
  326.         }
  328.         // As per CEA-708-E, service number 0 is invalid, and should be ignored.
  329.         if (serviceNumber != 0) {
  330.           // If the service doesn't already exist, create it.
  331.           if (!this.serviceNumberToService_.has(serviceNumber)) {
  332.             const service = new shaka.cea.Cea708Service(serviceNumber);
  333.             this.serviceNumberToService_.set(serviceNumber, service);
  334.           }
  335.           const service = this.serviceNumberToService_.get(serviceNumber);
  337.           // Process all control codes.
  338.           const startPos = dtvccPacket.getPosition();
  340.           // Execute this loop `blockSize` times, to decode the control codes.
  341.           while (dtvccPacket.getPosition() - startPos < blockSize) {
  342.             const closedCaption = service.handleCea708ControlCode(dtvccPacket);
  343.             if (closedCaption) {
  344.               parsedClosedCaptions.push(closedCaption);
  345.             }
  346.           }  // position < end of block
  347.         }  // serviceNumber != 0
  348.       }  // hasMoreData
  349.     } catch (error) {
  350.       if (error instanceof shaka.util.Error &&
  351.       error.code === shaka.util.Error.Code.BUFFER_READ_OUT_OF_BOUNDS) {
  352.         shaka.log.warnOnce('CEA708_INVALID_DATA',
  353.             'Buffer read out of bounds / invalid CEA-708 Data.');
  354.       } else {
  355.         // This is an unexpected error, and should be rethrown.
  356.         throw error;
  357.       }
  358.     }
  359.     return parsedClosedCaptions;
  360.   }
  362.   /**
  363.    * Checks if a byte has odd parity (Odd number of 1s in binary).
  364.    * @param {!number} byte
  365.    * @return {!boolean} True if the byte has odd parity.
  366.    * @private
  367.    */
  368.   isOddParity_(byte) {
  369.     let parity = 0;
  370.     while (byte) {
  371.       parity ^= (byte & 1); // toggle parity if low bit is 1
  372.       byte >>= 1; // shift away the low bit
  373.     }
  374.     return parity === 1;
  375.   }
  376. };
  378. /**
  379.  * itu_t_35_provider_code for ATSC user_data
  380.  * @private @const {!number}
  381.  */
  382. shaka.cea.CeaDecoder.ATSC_PROVIDER_CODE = 0x0031;
  384. /**
  385.  * When provider is ATSC user data, the ATSC_user_identifier code
  386.  * for ATSC1_data is "GA94" (0x47413934)
  387.  * @private @const {!number}
  388.  */
  389. shaka.cea.CeaDecoder.ATSC1_USER_IDENTIFIER = 0x47413934;
  391. /**
  392.  * @private @const {!number}
  393.  */
  394. shaka.cea.CeaDecoder.NTSC_CC_FIELD_1 = 0;
  396. /**
  397.  * @private @const {!number}
  398.  */
  399. shaka.cea.CeaDecoder.NTSC_CC_FIELD_2 = 1;
  401. /**
  402.  * 0xB5 is USA's code (Rec. ITU-T T.35)
  403.  * @private @const {!number}
  404.  */
  405. shaka.cea.CeaDecoder.USA_COUNTRY_CODE = 0xb5;