Process editors extracted from a custom reffy run

process-extracted-editors.js

const {JSDOM} = require("jsdom");
const {Parser} = require("json2csv");

const fs = require("fs");

const specData = require(process.argv[2]);

const affiliations = {};

const sortKeys = (o) =>
  Object (o) !== o || Array .isArray (o)
    ? o
      : Object .keys (o) .sort () .reduce ((a, k) => ({...a, [k]: sortKeys (o [k])}), {});

const domainToAffiliation = {
  "adobe.com": "Adobe",
  "adboe.com": "Adobe",
  "google.com": "Google",
  "mozilla.com": "Mozilla",
  "mozilla.org": "Mozilla",
  "cs.stanford.edu": "Mozilla",
  "mcc.id.au": "Mozilla",
  "apple.com": "Apple",
  "w3.org": "W3C",
  "microsoft.com": "Microsoft",
  "chromium.org": "Google",
  "disruptive-innovations.com": "Disruptive Innovations",
  "compuware.com": "Compuware",
  "gmail.com": "undetermined"
};

specData.filter(s => s.editors).forEach(s => {
  s.editors.forEach(e => {
    let affiliation, name, m;
    const edNode = JSDOM.fragment(e.markup);
    if (edNode.querySelector(".org")) {
      affiliation = edNode.querySelector(".org").textContent;
      name = edNode.querySelector(".p-name").textContent;
    } else if (m = edNode.querySelector('a[href^="mailto:"]')) {
      const domain = m.href.split('@')[1];
      affiliation = domainToAffiliation[domain];
      if (!affiliation) {
        affiliation = "undetermined";
      }
      if (!m.textContent.match(/@/)) {
        name = m.textContent;
      } else {
        name = edNode.textContent.split(',')[0];
      }
    }
    if (!affiliation || affiliation === "undetermined") {
      if (m = e.text.match(/^([^\(]*) \(([^\)]*)\)/)) {
        affiliation = m[2];
        name = m[1];
      } else if (m = e.text.match(/^([^,]*), ([a-zA-Z ]*)/)) {
        affiliation = m[2];
        name = m[1];
      } else {
        affiliation = "undetermined";
        name = e.text;
      }
    }
    affiliation = affiliation.trim().replace(/ Systems/, '').replace(/ LLC/, '').replace(/ inc\.?/i, '').replace(/ Corp.*/, '').replace(/ \[[^\]]*\]/, '').replace(/ \([^\)]*\)/, '').replace(/,.*/, '').replace(/ Foundation/, '').replace(/ ASA/, '').replace("W3C Invited Expert", "Invited Expert").toLowerCase();
    if (affiliation.match(/^until /) || affiliation.match(/former /)) return;
    name = name.replace(/’/, "'");
    if (!name || name.match(/see contributors/i)) return;
    if (!affiliations[affiliation]) {
      affiliations[affiliation] = {};
    }
    if (!affiliations[affiliation][name]) {
      affiliations[affiliation][name] = [];
    }
    affiliations[affiliation][name].push({url: s.nightly.url, title: s.title});
  });
});


fs.writeFileSync("editors-affiliations.json", JSON.stringify(sortKeys(affiliations), null, 2));
const json2csv = new Parser();
const flat = Object.keys(affiliations).sort().map(aff => Object.keys(affiliations[aff]).sort().map(ed => affiliations[aff][ed].map(spec => { spec.editor = ed; spec.affiliation = aff; return spec ;}))).flat().flat();
const csv = json2csv.parse(flat);
fs.writeFileSync("editors-affiliations.csv", csv);