Replace helpers.urlRe with helpers.urlFe to address "innefficient regular expression" CodeQL alert, introduce function expressions as an alternative, test more inputs.

helpers/helpers.js

@@ -23,9 +23,6 @@ const htmlElementRe = /<(([A-Za-z][A-Za-z0-9-]*)(?:\s[^`>]*)?)\/?>/g;
 module.exports.htmlElementRe = htmlElementRe;
 
 // Regular expressions for range matching
-module.exports.urlRe =
-  // eslint-disable-next-line max-len
-  /(?:http|ftp)s?:\/\/(?:[^\s()<>\]"'`]|\([^\s<>\]"'`]*\))*\b(?:[-#/]|\([^\s<>\]"'`]*\))*/ig;
 module.exports.listItemMarkerRe = /^([\s>]*)(?:[*+-]|\d+[.)])\s+/;
 module.exports.orderedListItemMarkerRe = /^[\s>]*0*(\d+)[.)]/;
 
@@ -1187,3 +1184,110 @@ function expandTildePath(file, os) {
   return homedir ? file.replace(/^~($|\/|\\)/, `${homedir}$1`) : file;
 }
 module.exports.expandTildePath = expandTildePath;
+
+/**
+ * RegExp.exec-style implementation of function expressions.
+ *
+ * @param {Function} funcExp Function that takes string and returns
+ * [index, length] or null.
+ * @param {string} input String to search.
+ * @returns {string[] | null} RegExp.exec-style [match] with an index property.
+ */
+function funcExpExec(funcExp, input) {
+  // Start or resume match
+  // @ts-ignore
+  const lastIndex = funcExp.lastIndex || 0;
+  const result = funcExp(input.slice(lastIndex));
+  if (result) {
+    // Update lastIndex and return match
+    const [ subIndex, length ] = result;
+    const index = lastIndex + subIndex;
+    // @ts-ignore
+    funcExp.lastIndex = index + length;
+    const match = [ input.slice(index, index + length) ];
+    // @ts-ignore
+    match.index = index;
+    return match;
+  }
+  // Reset lastIndex and return no match
+  // @ts-ignore
+  funcExp.lastIndex = 0;
+  return null;
+}
+module.exports.funcExpExec = funcExpExec;
+
+const urlFeProtocolRe = /(?:http|ftp)s?:\/\//i;
+const urlFeAutolinkTerminalsRe = / |$/;
+const urlFeBareTerminalsRe = /[ ,!`'"\]]|$/;
+const urlFeNonTerminalsRe = "-#/";
+const urlFePunctuationRe = /\p{Punctuation}/u;
+const urlFePrefixToPostfix = new Map([
+  [ " ", " " ],
+  [ "`", "`" ],
+  [ "'", "'" ],
+  [ "\"", "\"" ],
+  [ "‘", "’" ],
+  [ "“", "”" ],
+  [ "«", "»" ],
+  [ "*", "*" ],
+  [ "_", "_" ],
+  [ "(", ")" ],
+  [ "[", "]" ],
+  [ "{", "}" ],
+  [ "<", ">" ],
+  [ ">", "<" ]
+]);
+
+/**
+ * Function expression that matches URLs.
+ *
+ * @param {string} input Substring to search for a URL.
+ * @returns {Array | null} [index, length] of URL or null.
+ */
+function urlFe(input) {
+  // Find start of URL by searching for protocol
+  const match = input.match(urlFeProtocolRe);
+  if (match) {
+    // Look for matching pre/postfix characters (ex: <...>)
+    const start = match.index || 0;
+    const length = match[0].length;
+    const prefix = input[start - 1] || " ";
+    const postfix = urlFePrefixToPostfix.get(prefix);
+    // @ts-ignore
+    let endPostfix = input.indexOf(postfix, start + length);
+    if (endPostfix === -1) {
+      endPostfix = input.length;
+    }
+    // Look for characters that terminate a URL
+    const terminalsRe =
+      (prefix === "<") ? urlFeAutolinkTerminalsRe : urlFeBareTerminalsRe;
+    const endTerminal = start + input.slice(start).search(terminalsRe);
+    // Determine tentative end of URL
+    let end = Math.min(endPostfix, endTerminal);
+    if (prefix === " ") {
+      // If the URL used " " as pre/postfix characters, trim the end
+      if (input[end - 1] === ")") {
+        // Trim any ")" beyond the last "(...)" pair
+        const lastOpenParen = input.lastIndexOf("(", end - 2);
+        if (lastOpenParen <= start) {
+          end--;
+        } else {
+          const nextCloseParen = input.indexOf(")", lastOpenParen + 1);
+          end = nextCloseParen + 1;
+        }
+      } else {
+        // Trim unwanted punctuation
+        while (
+          !urlFeNonTerminalsRe.includes(input[end - 1]) &&
+          urlFePunctuationRe.test(input[end - 1])
+        ) {
+          end--;
+        }
+      }
+    }
+    return [ start, end - start ];
+  }
+  // No match
+  return null;
+}
+module.exports.urlFe = urlFe;