From bb1edf671731343d684ad01c3019127cf440dd90 Mon Sep 17 00:00:00 2001
From: Aidan Feldman <aidan.feldman@gmail.com>
Date: Tue, 8 Jan 2019 22:33:51 -0500
Subject: [PATCH] support unicode characters

---
 bootstrap-toc.js | 25 ++++++++++++++++++++-----
 test/toc-test.js |  8 +++++++-
 2 files changed, 27 insertions(+), 6 deletions(-)

diff --git a/bootstrap-toc.js b/bootstrap-toc.js
index 4bb1c1c..534f100 100644
--- a/bootstrap-toc.js
+++ b/bootstrap-toc.js
@@ -20,11 +20,26 @@
 
       generateUniqueIdBase: function(el) {
         var text = $(el).text();
-        var anchor = text
-          .trim()
-          .toLowerCase()
-          .replace(/[^A-Za-z0-9]+/g, "-");
-        return anchor || el.tagName.toLowerCase();
+
+        // adapted from
+        // https://github.com/bryanbraun/anchorjs/blob/65fede08d0e4a705f72f1e7e6284f643d5ad3cf3/anchor.js#L237-L257
+
+        // Regex for finding the non-safe URL characters (many need escaping): & +$,:;=?@"#{}|^~[`%!'<>]./()*\ (newlines, tabs, backspace, & vertical tabs)
+        var nonsafeChars = /[& +$,:;=?@"#{}|^~[`%!'<>\]\.\/\(\)\*\\\n\t\b\v]/g,
+          urlText;
+
+        // Note: we trim hyphens after truncating because truncating can cause dangling hyphens.
+        // Example string:                      // " ⚡⚡ Don't forget: URL fragments should be i18n-friendly, hyphenated, short, and clean."
+        urlText = text
+          .trim() // "⚡⚡ Don't forget: URL fragments should be i18n-friendly, hyphenated, short, and clean."
+          .replace(/\'/gi, "") // "⚡⚡ Dont forget: URL fragments should be i18n-friendly, hyphenated, short, and clean."
+          .replace(nonsafeChars, "-") // "⚡⚡-Dont-forget--URL-fragments-should-be-i18n-friendly--hyphenated--short--and-clean-"
+          .replace(/-{2,}/g, "-") // "⚡⚡-Dont-forget-URL-fragments-should-be-i18n-friendly-hyphenated-short-and-clean-"
+          .substring(0, 64) // "⚡⚡-Dont-forget-URL-fragments-should-be-i18n-friendly-hyphenated-"
+          .replace(/^-+|-+$/gm, "") // "⚡⚡-Dont-forget-URL-fragments-should-be-i18n-friendly-hyphenated"
+          .toLowerCase(); // "⚡⚡-dont-forget-url-fragments-should-be-i18n-friendly-hyphenated"
+
+        return urlText || el.tagName.toLowerCase();
       },
 
       generateUniqueId: function(el) {
diff --git a/test/toc-test.js b/test/toc-test.js
index abaa256..73e3c9d 100644
--- a/test/toc-test.js
+++ b/test/toc-test.js
@@ -43,7 +43,7 @@ describe("Toc", function() {
         var el = document.createElement("h1");
         el.innerHTML = "Some  tExt- with aidan's /. stuff   ";
         var base = Toc.helpers.generateUniqueIdBase(el);
-        expect(base).to.eql("some-text-with-aidan-s-stuff");
+        expect(base).to.eql("some-text-with-aidans-stuff");
       });
 
       it("uses the tag name of the element if there's no text", function() {
@@ -51,6 +51,12 @@ describe("Toc", function() {
         var base = Toc.helpers.generateUniqueIdBase(el);
         expect(base).to.eql("h1");
       });
+
+      it("handles unicode", function() {
+        var el = $("<h1>💃 🕺</h1>")[0];
+        var base = Toc.helpers.generateUniqueIdBase(el);
+        expect(base).to.eql("💃-🕺");
+      });
     });
 
     describe(".generateUniqueId()", function() {