From adc6f3a26f55eb696071b459452326a5ef40fef5 Mon Sep 17 00:00:00 2001
From: Koji Ishii <kojii@chromium.org>
Date: Sat, 11 Nov 2023 15:58:14 +0900
Subject: [PATCH] Java: Stop emitting close tags if self-closing

This patch changes Java `HTMLProcessor` not to emit close tags if the tag is self-closing.

Also adds tests for:
* Unpaired close tags.
* Self-closing tags don't affect skip nodes (e.g., `<nobr>`.)
These test cases are from #355.

Fixes #361.
---
 .../java/com/google/budoux/HTMLProcessor.java |  4 ++
 .../com/google/budoux/HTMLProcessorTest.java  | 58 ++++++++++++++-----
 2 files changed, 49 insertions(+), 13 deletions(-)

diff --git a/java/src/main/java/com/google/budoux/HTMLProcessor.java b/java/src/main/java/com/google/budoux/HTMLProcessor.java
index 5797967d..b63a6db3 100644
--- a/java/src/main/java/com/google/budoux/HTMLProcessor.java
+++ b/java/src/main/java/com/google/budoux/HTMLProcessor.java
@@ -128,6 +128,10 @@ public void tail(Node node, int depth) {
       }
       // assume node instanceof Element;
       toSkip = elementStack.pop();
+      Element element = (Element)node;
+      if (element.tag().isSelfClosing()) {
+        return;
+      }
       output.append(String.format("</%s>", node.nodeName()));
     }
   }
diff --git a/java/src/test/java/com/google/budoux/HTMLProcessorTest.java b/java/src/test/java/com/google/budoux/HTMLProcessorTest.java
index 123ba018..fe5cd39d 100644
--- a/java/src/test/java/com/google/budoux/HTMLProcessorTest.java
+++ b/java/src/test/java/com/google/budoux/HTMLProcessorTest.java
@@ -27,15 +27,19 @@
 /** Unit tests for {@link HTMLProcessor}. */
 @RunWith(JUnit4.class)
 public class HTMLProcessorTest {
+  String pre = "<span style=\"word-break: keep-all; overflow-wrap: anywhere;\">";
+  String post = "</span>";
+
+  private String wrap(String input) {
+    return this.pre + input + this.post;
+  }
 
   @Test
   public void testResolveWithSimpleTextInput() {
     List<String> phrases = Arrays.asList("abc", "def");
     String html = "abcdef";
     String result = HTMLProcessor.resolve(phrases, html, "<wbr>");
-    assertEquals(
-        "<span style=\"word-break: keep-all; overflow-wrap: anywhere;\">abc<wbr>def</span>",
-        result);
+    assertEquals(this.wrap("abc<wbr>def"), result);
   }
 
   @Test
@@ -44,20 +48,32 @@ public void testResolveWithStandardHTMLInput() {
     String html = "ab<a href=\"http://example.com\">cd</a>ef";
     String result = HTMLProcessor.resolve(phrases, html, "<wbr>");
     assertEquals(
-        "<span style=\"word-break: keep-all; overflow-wrap: anywhere;\">ab<a"
-            + " href=\"http://example.com\">c<wbr>d</a>ef</span>",
+        this.wrap("ab<a href=\"http://example.com\">c<wbr>d</a>ef"),
         result);
   }
 
+  @Test
+  public void testResolveWithImg() {
+    List<String> phrases = Arrays.asList("abc", "def");
+    String html = "<img>abcdef";
+    String result = HTMLProcessor.resolve(phrases, html, "<wbr>");
+    assertEquals(this.wrap("<img>abc<wbr>def"), result);
+  }
+
+  @Test
+  public void testResolveWithUnpairedClose() {
+    List<String> phrases = Arrays.asList("abc", "def");
+    String html = "abcdef</p>";
+    String result = HTMLProcessor.resolve(phrases, html, "<wbr>");
+    assertEquals(this.wrap("abc<wbr>def<p></p>"), result);
+  }
+
   @Test
   public void testResolveWithNodesToSkip() {
     List<String> phrases = Arrays.asList("abc", "def", "ghi");
     String html = "a<button>bcde</button>fghi";
     String result = HTMLProcessor.resolve(phrases, html, "<wbr>");
-    assertEquals(
-        "<span style=\"word-break: keep-all; overflow-wrap:"
-            + " anywhere;\">a<button>bcde</button>f<wbr>ghi</span>",
-        result);
+    assertEquals(this.wrap("a<button>bcde</button>f<wbr>ghi"), result);
   }
 
   @Test
@@ -65,9 +81,26 @@ public void testResolveWithNodesBreakBeforeSkip() {
     List<String> phrases = Arrays.asList("abc", "def", "ghi", "jkl");
     String html = "abc<nobr>defghi</nobr>jkl";
     String result = HTMLProcessor.resolve(phrases, html, "<wbr>");
+    assertEquals(this.wrap("abc<wbr><nobr>defghi</nobr><wbr>jkl"), result);
+  }
+
+  @Test
+  public void testResolveWithAfterSkip() {
+    List<String> phrases = Arrays.asList("abc", "def", "ghi", "jkl");
+    String html = "abc<nobr>def</nobr>ghijkl";
+    String result = HTMLProcessor.resolve(phrases, html, "<wbr>");
+    assertEquals(
+        this.wrap("abc<wbr><nobr>def</nobr><wbr>ghi<wbr>jkl"),
+        result);
+  }
+
+  @Test
+  public void testResolveWithAfterSkipWithImg() {
+    List<String> phrases = Arrays.asList("abc", "def", "ghi", "jkl");
+    String html = "abc<nobr>d<img>ef</nobr>ghijkl";
+    String result = HTMLProcessor.resolve(phrases, html, "<wbr>");
     assertEquals(
-        "<span style=\"word-break: keep-all; overflow-wrap:"
-            + " anywhere;\">abc<wbr><nobr>defghi</nobr><wbr>jkl</span>",
+        this.wrap("abc<wbr><nobr>d<img>ef</nobr><wbr>ghi<wbr>jkl"),
         result);
   }
 
@@ -76,8 +109,7 @@ public void testResolveWithNothingToSplit() {
     List<String> phrases = Arrays.asList("abcdef");
     String html = "abcdef";
     String result = HTMLProcessor.resolve(phrases, html, "<wbr>");
-    assertEquals(
-        "<span style=\"word-break: keep-all; overflow-wrap: anywhere;\">abcdef</span>", result);
+    assertEquals(this.wrap("abcdef"), result);
   }
 
   @Test