From 957bb694c6561a61f3126a606d4b52b8c2bb8e87 Mon Sep 17 00:00:00 2001
From: "Erik C. Thauvin" <erik@thauvin.net>
Date: Wed, 31 Jul 2024 17:10:00 -0700
Subject: [PATCH] Improved normalizing with support for common separators

---
 src/main/java/rife/render/RenderUtils.java    | 27 +++++++++----------
 .../java/rife/render/TestRenderUtils.java     | 10 +++++--
 2 files changed, 21 insertions(+), 16 deletions(-)

diff --git a/src/main/java/rife/render/RenderUtils.java b/src/main/java/rife/render/RenderUtils.java
index 43c92c1..f2f0859 100644
--- a/src/main/java/rife/render/RenderUtils.java
+++ b/src/main/java/rife/render/RenderUtils.java
@@ -334,21 +334,20 @@ public final class RenderUtils {
             return src;
         }
 
-        var normalized = Normalizer.normalize(src.trim(), Normalizer.Form.NFD);
-        var sb = new StringBuilder(normalized.length());
-        boolean space = false;
-        for (var c : normalized.toCharArray()) {
-            if (c <= '\u007F') { // ascii only
-                if (!space && c == ' ') {
-                    space = true;
-                    sb.append('-');
-                } else {
-                    space = false;
-                    if (c >= '0' && c <= '9' || c >= 'a' && c <= 'z') {
-                        sb.append(c);
-                    } else if (c >= 'A' && c <= 'Z') {
-                        sb.append((char) (c + 32)); // lowercase
+        var normalized = Normalizer.normalize(src.trim(), Normalizer.Form.NFD).toCharArray();
+
+        var sb = new StringBuilder(normalized.length);
+        for (var i = 0; i < normalized.length; i++) {
+            var c = normalized[i];
+            if (c <= '\u007F') { // ASCII only
+                if (" &()-_=[{]}\\|;:,<.>/".indexOf(c) != -1) { // common separators
+                    if (!sb.isEmpty() && i != normalized.length - 1 && sb.charAt(sb.length() - 1) != '-') {
+                        sb.append('-');
                     }
+                } else if (c >= '0' && c <= '9' || c >= 'a' && c <= 'z') { // letters & digits
+                    sb.append(c);
+                } else if (c >= 'A' && c <= 'Z') { // uppercase letters
+                    sb.append((char) (c + 32)); // make lowercase
                 }
             }
         }
diff --git a/src/test/java/rife/render/TestRenderUtils.java b/src/test/java/rife/render/TestRenderUtils.java
index 6ab1a22..fcee416 100644
--- a/src/test/java/rife/render/TestRenderUtils.java
+++ b/src/test/java/rife/render/TestRenderUtils.java
@@ -96,8 +96,14 @@ class TestRenderUtils {
 
     @Test
     void testNormalize() {
-        assertThat(RenderUtils.normalize("")).isEmpty();
-        assertThat(RenderUtils.normalize(SAMPLE_GERMAN)).isEqualTo("mochten-sie-ein-paar-apfel");
+        assertThat(RenderUtils.normalize("")).as("empty").isEmpty();
+        assertThat(RenderUtils.normalize(" &()-_=[{]}\\|;:,<.>/")).as("blank").isEmpty();
+        assertThat(RenderUtils.normalize(SAMPLE_GERMAN)).as("greman").isEqualTo("mochten-sie-ein-paar-apfel");
+        assertThat(RenderUtils.normalize("foo  bar, <foo-bar>,foo:bar,foo;(bar), {foo} & bar=foo.bar[foo|bar]"))
+                .as("foo-bar")
+                .isEqualTo("foo-bar-foo-bar-foo-bar-foo-bar-foo-bar-foo-bar-foo-bar");
+        assertThat(RenderUtils.normalize("News for January 6, 2023 (Paris)")).as("docs example")
+                .isEqualTo("news-for-january-6-2023-paris");
     }
 
     @Test