werk85/node-html-to-text

// eslint-disable-next-line no-unused-vars
import { InlineTextBuilder } from './inline-text-builder';

// eslint-disable-next-line import/no-unassigned-import
import './typedefs';


function charactersToCodes (str) {
  return [...str]
    .map(c => '\\u' + c.charCodeAt(0).toString(16).padStart(4, '0'))
    .join('');
}

/**
 * Helps to handle HTML whitespaces.
 *
 * @class WhitespaceProcessor
 */
class WhitespaceProcessor {

  /**
   * Creates an instance of WhitespaceProcessor.
   *
   * @param { Options } options    HtmlToText options.
   * @memberof WhitespaceProcessor
   */
  constructor (options) {
    this.whitespaceChars = (options.preserveNewlines)
      ? options.whitespaceCharacters.replace(/\n/g, '')
      : options.whitespaceCharacters;
    const whitespaceCodes = charactersToCodes(this.whitespaceChars);
    this.leadingWhitespaceRe = new RegExp(`^[${whitespaceCodes}]`);
    this.trailingWhitespaceRe = new RegExp(`[${whitespaceCodes}]$`);
    this.allWhitespaceOrEmptyRe = new RegExp(`^[${whitespaceCodes}]*$`);
    this.newlineOrNonWhitespaceRe = new RegExp(`(\\n|[^\\n${whitespaceCodes}])`, 'g');
    this.newlineOrNonNewlineStringRe = new RegExp(`(\\n|[^\\n]+)`, 'g');

    if (options.preserveNewlines) {

      const wordOrNewlineRe = new RegExp(`\\n|[^\\n${whitespaceCodes}]+`, 'gm');

      /**
       * Shrink whitespaces and wrap text, add to the builder.
       *
       * @param { string }                  text              Input text.
       * @param { InlineTextBuilder }       inlineTextBuilder A builder to receive processed text.
       * @param { (str: string) => string } [ transform ]     A transform to be applied to words.
       * @param { boolean }                 [noWrap] Don't wrap text even if the line is too long.
       */
      this.shrinkWrapAdd = function (text, inlineTextBuilder, transform = (str => str), noWrap = false) {
        if (!text) { return; }
        const previouslyStashedSpace = inlineTextBuilder.stashedSpace;
        let anyMatch = false;
        let m = wordOrNewlineRe.exec(text);
        if (m) {
          anyMatch = true;
          if (m[0] === '\n') {
            inlineTextBuilder.startNewLine();
          } else if (previouslyStashedSpace || this.testLeadingWhitespace(text)) {
            inlineTextBuilder.pushWord(transform(m[0]), noWrap);
          } else {
            inlineTextBuilder.concatWord(transform(m[0]), noWrap);
          }
          while ((m = wordOrNewlineRe.exec(text)) !== null) {
            if (m[0] === '\n') {
              inlineTextBuilder.startNewLine();
            } else {
              inlineTextBuilder.pushWord(transform(m[0]), noWrap);
            }
          }
        }
        inlineTextBuilder.stashedSpace = (previouslyStashedSpace && !anyMatch) || (this.testTrailingWhitespace(text));
        // No need to stash a space in case last added item was a new line,
        // but that won't affect anything later anyway.
      };

    } else {

      const wordRe = new RegExp(`[^${whitespaceCodes}]+`, 'g');

      this.shrinkWrapAdd = function (text, inlineTextBuilder, transform = (str => str), noWrap = false) {
        if (!text) { return; }
        const previouslyStashedSpace = inlineTextBuilder.stashedSpace;
        let anyMatch = false;
        let m = wordRe.exec(text);
        if (m) {
          anyMatch = true;
          if (previouslyStashedSpace || this.testLeadingWhitespace(text)) {
            inlineTextBuilder.pushWord(transform(m[0]), noWrap);
          } else {
            inlineTextBuilder.concatWord(transform(m[0]), noWrap);
          }
          while ((m = wordRe.exec(text)) !== null) {
            inlineTextBuilder.pushWord(transform(m[0]), noWrap);
          }
        }
        inlineTextBuilder.stashedSpace = (previouslyStashedSpace && !anyMatch) || this.testTrailingWhitespace(text);
      };

    }
  }

  /**
   * Add text with only minimal processing.
   * Everything between newlines considered a single word.
   * No whitespace is trimmed.
   * Not affected by preserveNewlines option - `\n` always starts a new line.
   *
   * `noWrap` argument is `true` by default - this won't start a new line
   * even if there is not enough space left in the current line.
   *
   * @param { string }            text              Input text.
   * @param { InlineTextBuilder } inlineTextBuilder A builder to receive processed text.
   * @param { boolean }           [noWrap] Don't wrap text even if the line is too long.
   */
  addLiteral (text, inlineTextBuilder, noWrap = true) {
    if (!text) { return; }
    const previouslyStashedSpace = inlineTextBuilder.stashedSpace;
    let anyMatch = false;
    let m = this.newlineOrNonNewlineStringRe.exec(text);
    if (m) {
      anyMatch = true;
      if (m[0] === '\n') {
        inlineTextBuilder.startNewLine();
      } else if (previouslyStashedSpace) {
        inlineTextBuilder.pushWord(m[0], noWrap);
      } else {
        inlineTextBuilder.concatWord(m[0], noWrap);
      }
      while ((m = this.newlineOrNonNewlineStringRe.exec(text)) !== null) {
        if (m[0] === '\n') {
          inlineTextBuilder.startNewLine();
        } else {
          inlineTextBuilder.pushWord(m[0], noWrap);
        }
      }
    }
    inlineTextBuilder.stashedSpace = (previouslyStashedSpace && !anyMatch);
  }

  /**
   * Test whether the given text starts with HTML whitespace character.
   *
   * @param   { string }  text  The string to test.
   * @returns { boolean }
   */
  testLeadingWhitespace (text) {
    return this.leadingWhitespaceRe.test(text);
  }

  /**
   * Test whether the given text ends with HTML whitespace character.
   *
   * @param   { string }  text  The string to test.
   * @returns { boolean }
   */
  testTrailingWhitespace (text) {
    return this.trailingWhitespaceRe.test(text);
  }

  /**
   * Test whether the given text contains any non-whitespace characters.
   *
   * @param   { string }  text  The string to test.
   * @returns { boolean }
   */
  testContainsWords (text) {
    return !this.allWhitespaceOrEmptyRe.test(text);
  }

  /**
   * Return the number of newlines if there are no words.
   *
   * If any word is found then return zero regardless of the actual number of newlines.
   *
   * @param   { string }  text  Input string.
   * @returns { number }
   */
  countNewlinesNoWords (text) {
    this.newlineOrNonWhitespaceRe.lastIndex = 0;
    let counter = 0;
    let match;
    while ((match = this.newlineOrNonWhitespaceRe.exec(text)) !== null) {
      if (match[0] === '\n') {
        counter++;
      } else {
        return 0;
      }
    }
    return counter;
  }

}

export { WhitespaceProcessor };