Fix legacy build

The build process failed because webpack does not allow failed imports
when it is run as a module.
This detects modules with default exports and only generates import
statements for default exports where they are available.

config/jsdoc/plugins/markdown.cjs

@@ -0,0 +1,115 @@
+/* eslint-disable import/no-commonjs */
+
+/**
+ * Modified from JSDoc's plugins/markdown and lib/jsdoc/util/markdown modules
+ * (see https://github.com/jsdoc3/jsdoc/), which are licensed under the Apache 2
+ * license (see https://www.apache.org/licenses/LICENSE-2.0).
+ *
+ * This version does not protect http(s) urls from being turned into links, and
+ * works around an issue with `~` characters in module paths by escaping them.
+ */
+
+const marked = require('marked');
+const format = require('util').format;
+
+const tags = [
+  'author',
+  'classdesc',
+  'description',
+  'exceptions',
+  'params',
+  'properties',
+  'returns',
+  'see',
+  'summary',
+];
+
+const hasOwnProp = Object.prototype.hasOwnProperty;
+
+const markedRenderer = new marked.Renderer();
+
+// Allow prettyprint to work on inline code samples
+markedRenderer.code = function (code, language) {
+  const langClass = language ? ' lang-' + language : '';
+
+  return format(
+    '<pre class="prettyprint source%s"><code>%s</code></pre>',
+    langClass,
+    escapeCode(code)
+  );
+};
+
+function escapeCode(source) {
+  return source
+    .replace(/</g, '&lt;')
+    .replace(/"/g, '&quot;')
+    .replace(/'/g, '&#39;');
+}
+
+function escapeUnderscoresAndTildes(source) {
+  return source.replace(/\{@[^}\r\n]+\}/g, function (wholeMatch) {
+    return wholeMatch.replace(/(^|[^\\])_/g, '$1\\_').replace('~', '&tilde;');
+  });
+}
+
+function unencodeQuotesAndTildes(source) {
+  return source.replace(/\{@[^}\r\n]+\}/g, function (wholeMatch) {
+    return wholeMatch.replace(/&quot;/g, '"').replace(/&tilde;/g, '~');
+  });
+}
+
+function parse(source) {
+  let result;
+
+  source = escapeUnderscoresAndTildes(source);
+
+  result = marked(source, {renderer: markedRenderer})
+    .replace(/\s+$/, '')
+    .replace(/&#39;/g, "'");
+
+  result = unencodeQuotesAndTildes(result);
+
+  return result;
+}
+
+function shouldProcessString(tagName, text) {
+  let shouldProcess = true;
+
+  // we only want to process `@author` and `@see` tags that contain Markdown links
+  if ((tagName === 'author' || tagName === 'see') && text.indexOf('[') === -1) {
+    shouldProcess = false;
+  }
+
+  return shouldProcess;
+}
+
+function process(doclet) {
+  tags.forEach(function (tag) {
+    if (!hasOwnProp.call(doclet, tag)) {
+      return;
+    }
+
+    if (
+      typeof doclet[tag] === 'string' &&
+      shouldProcessString(tag, doclet[tag])
+    ) {
+      doclet[tag] = parse(doclet[tag]);
+    } else if (Array.isArray(doclet[tag])) {
+      doclet[tag].forEach(function (value, index, original) {
+        const inner = {};
+
+        inner[tag] = value;
+        process(inner);
+        original[index] = inner[tag];
+      });
+    } else if (doclet[tag]) {
+      process(doclet[tag]);
+    }
+  });
+}
+
+exports.handlers = {
+  newDoclet: function (e) {
+    process(e.doclet);
+  },
+};