find-bad-notes.js

"use strict";

const fs = require("fs")
const path = require("path")

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

/* get all ".html" files under the given directory into an array */
function getAllFiles(dirPath, files) {
  const dir = fs.readdirSync(dirPath);

  files = files || [];

  for (const file of dir) {
    if (fs.statSync(dirPath + "/" + file).isDirectory()) {
      files = getAllFiles(dirPath + "/" + file, files)
    } else {
      if (file.endsWith(".html")) {
        files.push(path.join(__dirname, dirPath, "/", file))
      }
    }
  }

  return files
}

const allDocPaths = getAllFiles(process.argv[2]);
const query = "div.note";

/* scrape off the front matter.
We're considering this very roughly as the second occurrence of "---" in the file.*/
function removeFrontMatter(doc) {
  const regexp = /---/g;
  const matchArray = [...doc.matchAll(regexp)];
  let end = matchArray[1].index + 3;
  return doc.substr(end);
}

/*
Matches the correct format for a note
*/
function isPNote(note) {
  let first = note.firstElementChild;
  if (first && first.tagName === "P") {
    let second = first.firstElementChild;
    if (second && second.tagName === "STRONG" && second.textContent === "Note:")
    {
      return true;
    }
  }
  return false;
}

/*
Matches any note whose first child is a paragraph
*/
function isP(note) {
  let first = note.firstElementChild;
  if (first && first.tagName === "P") {
      return true;
    }
  return false;
}

/*
Matches the format for a note which uses H4
*/
function isHNote(note) {
  let first = note.firstElementChild;
  if (first && first.tagName === "H4" && first.textContent === "Note") {
    return true;
  }
  return false;
}

/*
Matches the format for a note which uses `<strong>` without a wrapping paragraph
*/
function isStrong(note) {
  let first = note.firstElementChild;
  if (first && first.tagName === "STRONG" && first.textContent === "Note:") {
    return true;
  }
  return false;
}

for (const docPath of allDocPaths) {
  let docHTML = fs.readFileSync(docPath, {encoding: "utf-8"});
  docHTML = removeFrontMatter(docHTML);
  const { window } = new JSDOM(docHTML);
  const found = window.document.querySelectorAll(query);
  if (found.length > 0) {
    let log = false;
    for (let item of found) {
        // obviously, change this condition to log different sorts of notes,
        // such as all notes which use the heading style
        if (!isPNote(item)) {
          // this is weird but gives me a quick way to review the note contents,
          // to decide if it really fits a particular pattern
          console.log(`${item.innerHTML.substring(0, 50)}            ${docPath}`);
          log = true;
          }
      }
      if (log) {
        //console.log(docPath)
      }
  }
  window.close();
}