mdx-link-checker/index.js

"use strict";

const fs = require('fs');
const path = require('path');
const fsWalk = require('fs-walk');
const mdx = require('@mdx-js/mdx');
const slugPlugin = require('remark-slug');

var cache = {};
var exitCode = 0;

fsWalk.walkSync(process.argv[2], function(basedir, filename, stat, next) {
  if (stat.isFile()) {
    let filePath = basedir + "/" + filename
    let filePathAbs = path.resolve(filePath)
    let fileExt = filename.split('.').pop()

    if(["mdx", "md", "html", "htm"].indexOf(fileExt) == -1){
      return
    }

    let markdown = fs.readFileSync(filePathAbs).toString();
    let html = ""

    try {
      html = mdx.sync(markdown, {
        remarkPlugins: [
          slugPlugin
        ]
      })
    } catch(e) {
      // Fail if there was an error parsing a mdx/md file
      if (fileExt === "mdx" || fileExt == "md") {
        console.error("Unable to parse mdx to html: " + filename)
        throw e
      }
    }

    if (!cache[filePathAbs]) {
      cache[filePathAbs] = {
        filePath,
        filePathAbs,
        externalLinks: [],
        internalLinks: [],
        ids: {}
      }
    }


    if (fileExt === "md" || fileExt == "mdx") {
      markdown.replace(/\[[^\]]*\]\(([^)]+)\)/g, (_, match) => {
        if (!match.match(/(^https?:\/\/)|(^#)|(^[^:]+:.*)|(\.mdx?(#[a-zA-Z0-9._,-]*)?$)/)) {
          console.warn(`Missing .md/.mdx suffix for internal link: '${match}' in file ${filePathAbs}`)
        }
      });
    }

    let fillCache = function(html) {
      html.replace(/\s+(?:(?:"id":\s*)|(?:id=))"([^"]+)"/g, (_, match) => {
        if (match && match.match) {
          cache[filePathAbs].ids[match] = true
        }
      });

      html.replace(/\s+(?:(?:"href":\s*)|(?:href=))"([^"]+)"/g, (_, match) => {
        if (match && match.match) {
          if (match.match(/^https?:\/\//)) {
            cache[filePathAbs].externalLinks.push(match)
          } else if (match.match(/^[^:]+:.*/)) {
            // ignore links such as "mailto:" or "javascript:"
          } else {
            if (match.match(/^#/)) {
              match = filename + match;
            }

            cache[filePathAbs].internalLinks.push({
              original: match,
              absolute: path.resolve(basedir + "/" + match),
            })
          }
        }
      });
    }

    fillCache(html)
    fillCache(markdown)
  }
}, function(err) {
  if (err) console.log(err);
});

for (let file in cache) {
  let data = cache[file];

  data.internalLinks.map((link) => {
    let [targetFile, targetId] = link.absolute.split("#")

    if (!cache[targetFile]) {
      targetFile = targetFile + ".md"

      if (!cache[targetFile]) {
        targetFile = targetFile + "x"

        if (!cache[targetFile]) {
          targetFile = targetFile.trimEnd(path.sep).replace(/\.md(x)?$/i, "") + path.sep + "index.html"

          if (!cache[targetFile]) {
            // make sure linked file does not exist
            if (fs.existsSync(path.join(path.dirname(file), link.original)) === false) {
              exitCode = 1;
              console.error(`Link is broken: '${link.original}' in file ${data.filePathAbs}`)
              return;
            }
          }
        }
      }
    }

    if (targetId && !cache[targetFile].ids[targetId]) {
      console.warn(`Anchor of link is broken: '${link.original}' in file ${data.filePathAbs}`)
    }
  })
}

process.exit(exitCode);