use the unified / rehype / retext ecosystem to extract enriched keywords from each article in a corpus of JATS XML articles

index.js

var fs = require('fs')

var unified = require('unified')
var parseHTML = require('rehype-parse')
var visit = require('unist-util-visit')
var findAfter = require('unist-util-find-after')

var file = '/Users/rds45/.sciencefair/data/elife_dws2/articles/01096/elife-01096-v1.xml'

var getBody = function () {
  return function (ast, file) {
    var article = findAfter(ast, ast.children[0], function (node) {
      return node.type === 'element' && node.tagName === 'article'
    })
    var body = findAfter(article, article.children[0], function (node) {
      return node.type === 'element' && node.tagName === 'body'
    })
    // console.log(body)
    return body
  }
}

var toPlaintext = function (processor, config) {
  function Compiler (file, options, processor) {
    this.options = options
    this.data = processor.data
    this.file = file
  }

  function compile (tree) {
    var parts = []
    visit(tree, 'text', function (node) {
      parts.push(node.value)
    })
    return parts.join('')
  }

  /* Expose methods. */
  Compiler.prototype.compile = compile

  processor.Compiler = Compiler
}

fs.createReadStream(file, 'utf8')
  .pipe(unified())
  .use(parseHTML)
  .use(getBody)
  .use(toPlaintext)
  .pipe(process.stdout())