Fix HTML file detection when dots are in path

Author: jotto

lib/util.js

@@ -33,7 +33,20 @@ const isHtml = urlStr => {
 
   if (endsInHtml) return true;
 
+  // hack to detect extensions that are not HTML so we can handle
+  // paths with dots in them
+  const endsInOtherExtension = basename.match(/\.[a-zA-Z]{1,5}$/);
+  if (!endsInOtherExtension) return true;
+
   return false;
 };
 
-module.exports = { USER_AGENT_PLACEHOLDER, toBase64, fromBase64, createUri, getHeader, parseUriField, isHtml };
+module.exports = {
+  USER_AGENT_PLACEHOLDER,
+  toBase64,
+  fromBase64,
+  createUri,
+  getHeader,
+  parseUriField,
+  isHtml
+};

spec/utilSpec.js

@@ -0,0 +1,28 @@
+const util = require("../lib/util");
+
+describe("util", function() {
+  describe("isHtml", function() {
+    it("detects no extension", function() {
+      expect(util.isHtml("/")).toBe(true);
+    });
+    it("detects html", function() {
+      expect(util.isHtml("index.html")).toBe(true);
+    });
+    it("detects htm", function() {
+      expect(util.isHtml("index.htm")).toBe(true);
+    });
+    it("detects double dot html", function() {
+      expect(util.isHtml("index.bak.html")).toBe(true);
+    });
+    it("does not detect js", function() {
+      expect(util.isHtml("index.js")).toBe(false);
+    });
+    it("handles miscellaneous dots", function() {
+      expect(
+        util.isHtml(
+          "categories/1234;lat=-999999.8888888;lng=12341234.13371337;location=SanFrancisco"
+        )
+      ).toBe(true);
+    });
+  });
+});