merge-prefix-exceptions.js

/**
 * @typedef {import('nlcst').Paragraph} Paragraph
 */

import {toString} from 'nlcst-to-string'
import {modifyChildren} from 'unist-util-modify-children'

// Full stop characters that should not be treated as terminal sentence markers:
// A case-insensitive abbreviation.
const abbreviationPrefix = new RegExp(
  '^(' +
    '[0-9]{1,3}|' +
    '[a-z]|' +
    // Common Latin Abbreviations:
    // Based on: <https://en.wikipedia.org/wiki/List_of_Latin_abbreviations>.
    // Where only the abbreviations written without joining full stops,
    // but with a final full stop, were extracted.
    //
    // circa, capitulus, confer, compare, centum weight, eadem, (et) alii,
    // et cetera, floruit, foliis, ibidem, idem, nemine && contradicente,
    // opere && citato, (per) cent, (per) procurationem, (pro) tempore,
    // sic erat scriptum, (et) sequentia, statim, videlicet. */
    'al|ca|cap|cca|cent|cf|cit|con|cp|cwt|ead|etc|ff|' +
    'fl|ibid|id|nem|op|pro|seq|sic|stat|tem|viz' +
    ')$'
)

// Merge a sentence into its next sentence, when the sentence ends with a
// certain word.
export const mergePrefixExceptions = modifyChildren(
  /**
   * @type {import('unist-util-modify-children').Modifier<Paragraph>}
   */
  function (child, index, parent) {
    if ('children' in child && child.children.length > 1) {
      const period = child.children[child.children.length - 1]

      if (
        period &&
        (period.type === 'PunctuationNode' || period.type === 'SymbolNode') &&
        toString(period) === '.'
      ) {
        const node = child.children[child.children.length - 2]

        if (
          node &&
          node.type === 'WordNode' &&
          abbreviationPrefix.test(toString(node).toLowerCase())
        ) {
          // Merge period into abbreviation.
          node.children.push(period)
          child.children.pop()

          // Update position.
          if (period.position && node.position) {
            node.position.end = period.position.end
          }

          // Merge sentences.
          const next = parent.children[index + 1]

          if (next && next.type === 'SentenceNode') {
            child.children.push(...next.children)
            parent.children.splice(index + 1, 1)

            // Update position.
            if (next.position && child.position) {
              child.position.end = next.position.end
            }

            // Next, iterate over the current node again.
            return index - 1
          }
        }
      }
    }
  }
)