Decode HTML entities when reading PuzzleMe text.

A recent indie puzzle has instances of entities like `"` and `'`. These are valid HTML, but unnecessary. So the simplest way to handle these is to decode any valid HTML entities before re-encoding the essential ones, & and <.
jpd236 · Dec 8, 2023 · d72f05f · d72f05f
1 parent 2ae6be2
commit d72f05f
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 4 deletions.
diff --git a/src/commonMain/kotlin/com/jeffpdavidson/kotwords/formats/PuzzleMe.kt b/src/commonMain/kotlin/com/jeffpdavidson/kotwords/formats/PuzzleMe.kt
@@ -335,12 +335,12 @@ class PuzzleMe(val json: String) : DelegatingPuzzleable() {
         /**
          * Convert a PuzzleMe JSON string to HTML.
          *
-         * PuzzleMe mixes unescaped special XML characters (&, <) with HTML tags. This method escapes the special
-         * characters while leaving supported HTML tags untouched. <br> tags are replaced with newlines. Attributes
-         * are stripped out.
+         * PuzzleMe mixes unescaped special XML characters (&, <) with HTML tags. Sometimes there are real HTML escape
+         * sequences as well. This method escapes the special characters while leaving supported HTML tags untouched.
+         * <br> tags are replaced with newlines. Attributes are stripped out.
          */
         internal fun toHtml(clue: String): String {
-            return clue
+            return Encodings.decodeHtmlEntities(clue)
                 .replace("&", "&amp;")
                 .replace("\\s*<br/?>\\s*".toRegex(RegexOption.IGNORE_CASE), "\n")
                 // Strip other unsupported tags.

diff --git a/src/commonTest/kotlin/com/jeffpdavidson/kotwords/formats/PuzzleMeTest.kt b/src/commonTest/kotlin/com/jeffpdavidson/kotwords/formats/PuzzleMeTest.kt
@@ -10,6 +10,7 @@ import kotlin.test.assertTrue
 import kotlin.test.fail
 
 class PuzzleMeTest {
+
     @Test
     fun extractPuzzleJson() = runTest {
         assertEquals(
@@ -169,4 +170,11 @@ class PuzzleMeTest {
             "test link and <b>test bold link</b>",
             PuzzleMe.toHtml("<a href=\"abc\">test link</a> and <a href=\"def\"><b>test bold link</b></a>")
         )
+
+    @Test
+    fun toHtml_htmlEntities() =
+        assertEquals(
+            "Test with \"Quotes\" ' &lt; &amp; &amp;",
+            PuzzleMe.toHtml("Test with &#34;Quotes&#34; &#39; < & &amp;")
+        )
 }