music-metadata/lib/wav/WaveParser.ts

import * as strtok3 from 'strtok3';
import * as Token from 'token-types';
import initDebug from 'debug';

import * as riff from '../riff/RiffChunk.js';
import * as WaveChunk from './WaveChunk.js';
import { ID3v2Parser } from '../id3v2/ID3v2Parser.js';
import * as util from '../common/Util.js';
import { FourCcToken } from '../common/FourCC.js';
import { BasicParser } from '../common/BasicParser.js';
import { BroadcastAudioExtensionChunk, type IBroadcastAudioExtensionChunk } from './BwfChunk.js';
import type { AnyTagValue } from '../type.js';
import { WaveContentError } from './WaveChunk.js';

const debug = initDebug('music-metadata:parser:RIFF');

/**
 * Resource Interchange File Format (RIFF) Parser
 *
 * WAVE PCM soundfile format
 *
 * Ref:
 * - http://www.johnloomis.org/cpe102/asgn/asgn1/riff.html
 * - http://soundfile.sapp.org/doc/WaveFormat
 *
 * ToDo: Split WAVE part from RIFF parser
 */
export class WaveParser extends BasicParser {

  private fact: WaveChunk.IFactChunk | undefined;
  private blockAlign = 0;
  private header: riff.IChunkHeader | undefined;

  public async parse(): Promise<void> {

    const riffHeader = await this.tokenizer.readToken<riff.IChunkHeader>(riff.Header);
    debug(`pos=${this.tokenizer.position}, parse: chunkID=${riffHeader.chunkID}`);
    if (riffHeader.chunkID !== 'RIFF')
      return; // Not RIFF format
    this.metadata.setAudioOnly();
    return this.parseRiffChunk(riffHeader.chunkSize).catch(err => {
      if (!(err instanceof strtok3.EndOfStreamError)) {
        throw err;
      }
    });
  }

  public async parseRiffChunk(chunkSize: number): Promise<void> {
    const type = await this.tokenizer.readToken<string>(FourCcToken);
    this.metadata.setFormat('container', type);
    switch (type) {
      case 'WAVE':
        return this.readWaveChunk(chunkSize - FourCcToken.len);
      default:
        throw new WaveContentError(`Unsupported RIFF format: RIFF/${type}`);
    }
  }

  public async readWaveChunk(remaining: number): Promise<void> {

    while (remaining >= riff.Header.len) {
      const header = await this.tokenizer.readToken<riff.IChunkHeader>(riff.Header);
      remaining -= riff.Header.len + header.chunkSize;
      if (header.chunkSize > remaining) {
        this.metadata.addWarning('Data chunk size exceeds file size');
      }

      this.header = header;
      debug(`pos=${this.tokenizer.position}, readChunk: chunkID=RIFF/WAVE/${header.chunkID}`);
      switch (header.chunkID) {

        case 'LIST':
          await this.parseListTag(header);
          break;

        case 'fact': // extended Format chunk,
          this.metadata.setFormat('lossless', false);
          this.fact = await this.tokenizer.readToken(new WaveChunk.FactChunk(header));
          break;

        case 'fmt ': { // The Util Chunk, non-PCM Formats
          const fmt = await this.tokenizer.readToken<WaveChunk.IWaveFormat>(new WaveChunk.Format(header));

          let subFormat = WaveChunk.WaveFormatNameMap[fmt.wFormatTag];
          if (!subFormat) {
            debug(`WAVE/non-PCM format=${fmt.wFormatTag}`);
            subFormat = `non-PCM (${fmt.wFormatTag})`;
          }
          this.metadata.setFormat('codec', subFormat);
          this.metadata.setFormat('bitsPerSample', fmt.wBitsPerSample);
          this.metadata.setFormat('sampleRate', fmt.nSamplesPerSec);
          this.metadata.setFormat('numberOfChannels', fmt.nChannels);
          this.metadata.setFormat('bitrate', fmt.nBlockAlign * fmt.nSamplesPerSec * 8);
          this.blockAlign = fmt.nBlockAlign;
          break;
        }

        case 'id3 ': // The way Picard, FooBar currently stores, ID3 meta-data
        case 'ID3 ': { // The way Mp3Tags stores ID3 meta-data
          const id3_data = await this.tokenizer.readToken<Uint8Array>(new Token.Uint8ArrayType(header.chunkSize));
          const rst = strtok3.fromBuffer(id3_data);
          await new ID3v2Parser().parse(this.metadata, rst, this.options);
          break;
        }

        case 'data': { // PCM-data
          if (this.metadata.format.lossless !== false) {
            this.metadata.setFormat('lossless', true);
          }

          let chunkSize = header.chunkSize;
          if (this.tokenizer.fileInfo.size) {
            const calcRemaining = this.tokenizer.fileInfo.size - this.tokenizer.position;
            if (calcRemaining < chunkSize) {
              this.metadata.addWarning('data chunk length exceeding file length');
              chunkSize = calcRemaining;
            }
          }

          const numberOfSamples = this.fact ? this.fact.dwSampleLength : (chunkSize === 0xffffffff ? undefined : chunkSize / this.blockAlign);
          if (numberOfSamples) {
            this.metadata.setFormat('numberOfSamples', numberOfSamples);
            if (this.metadata.format.sampleRate) {
              this.metadata.setFormat('duration', numberOfSamples / this.metadata.format.sampleRate);
            }
          }

          if (this.metadata.format.codec === 'ADPCM') { // ADPCM is 4 bits lossy encoding resulting in 352kbps
            this.metadata.setFormat('bitrate', 352000);
          } else if (this.metadata.format.sampleRate) {
            this.metadata.setFormat('bitrate', this.blockAlign * this.metadata.format.sampleRate * 8);
          }
          await this.tokenizer.ignore(header.chunkSize);
          break;
        }

        case 'bext': { // Broadcast Audio Extension chunk	https://tech.ebu.ch/docs/tech/tech3285.pdf
          const bext = await this.tokenizer.readToken(BroadcastAudioExtensionChunk);
          Object.keys(bext).forEach(key => {
            this.metadata.addTag('exif', `bext.${key}`, bext[key as keyof IBroadcastAudioExtensionChunk]);
          });
          const bextRemaining = header.chunkSize - BroadcastAudioExtensionChunk.len;
          await this.tokenizer.ignore(bextRemaining);
          break;
        }

        case '\x00\x00\x00\x00': // padding ??
          debug(`Ignore padding chunk: RIFF/${header.chunkID} of ${header.chunkSize} bytes`);
          this.metadata.addWarning(`Ignore chunk: RIFF/${header.chunkID}`);
          await this.tokenizer.ignore(header.chunkSize);
          break;

        default:
          debug(`Ignore chunk: RIFF/${header.chunkID} of ${header.chunkSize} bytes`);
          this.metadata.addWarning(`Ignore chunk: RIFF/${header.chunkID}`);
          await this.tokenizer.ignore(header.chunkSize);
      }

      if ((this.header as riff.IChunkHeader).chunkSize % 2 === 1) {
        debug('Read odd padding byte'); // https://wiki.multimedia.cx/index.php/RIFF
        await this.tokenizer.ignore(1);
      }
    }
  }

  public async parseListTag(listHeader: riff.IChunkHeader): Promise<void> {
    const listType = await this.tokenizer.readToken(new Token.StringType(4, 'latin1'));
    debug('pos=%s, parseListTag: chunkID=RIFF/WAVE/LIST/%s', this.tokenizer.position, listType);
    switch (listType) {
      case 'INFO':
        return this.parseRiffInfoTags(listHeader.chunkSize - 4);
      default:
        this.metadata.addWarning(`Ignore chunk: RIFF/WAVE/LIST/${listType}`);
        debug(`Ignoring chunkID=RIFF/WAVE/LIST/${listType}`);
        return this.tokenizer.ignore(listHeader.chunkSize - 4).then();
    }
  }

  private async parseRiffInfoTags(chunkSize: number): Promise<void> {
    while (chunkSize >= 8) {
      const header = await this.tokenizer.readToken<riff.IChunkHeader>(riff.Header);
      const valueToken = new riff.ListInfoTagValue(header);
      const value = await this.tokenizer.readToken(valueToken);
      this.addTag(header.chunkID, util.stripNulls(value));
      chunkSize -= (8 + valueToken.len);
    }

    if (chunkSize !== 0) {
      throw new WaveContentError(`Illegal remaining size: ${chunkSize}`);
    }
  }

  private addTag(id: string, value: AnyTagValue) {
    this.metadata.addTag('exif', id, value);
  }

}