Support non-breaking content in java

This patch supports non-breaking content in Java.

In Java and Python implementations, the "Skip" operation includes the
skipped content to the BudouX parser, so no changes to the text for the
parser is needed.

This patch changes following items:
1. Add `NOBR` to the "skip" element.
2. Fix "skip" is applied only to its descendants. Before this patch, all
   content following "skip" elements are skipped.
3. When there's a phrase boundary right before the "skip" element,
   insert a break before the "skip" element.

budoux/skip_nodes.json

@@ -5,6 +5,7 @@
   "IFRAME",
   "INPUT",
   "META",
+  "NOBR",
   "SCRIPT",
   "STYLE",
   "TEXTAREA",

java/src/main/java/com/google/budoux/HTMLProcessor.java

@@ -29,6 +29,7 @@
 import java.util.List;
 import java.util.Locale;
 import java.util.Set;
+import java.util.Stack;
 import java.util.stream.Collectors;
 import org.jsoup.Jsoup;
 import org.jsoup.nodes.Document;
@@ -61,6 +62,7 @@ private static class PhraseResolvingNodeVisitor implements NodeVisitor {
     private final StringBuilder output = new StringBuilder();
     private Integer scanIndex = 0;
     private boolean toSkip = false;
+    private Stack<Boolean> elementStack = new Stack<Boolean>();
 
     PhraseResolvingNodeVisitor(List<String> phrases) {
       this.phrasesJoined = String.join(Character.toString(SEP), phrases);
@@ -76,14 +78,20 @@ public void head(Node node, int depth) {
         return;
       }
       if (node instanceof Element) {
+        elementStack.push(toSkip);
         String attributesEncoded =
             node.attributes().asList().stream()
                 .map(attribute -> " " + attribute)
                 .collect(Collectors.joining(""));
-        output.append(String.format("<%s%s>", node.nodeName(), attributesEncoded));
-        if (skipNodes.contains(node.nodeName().toUpperCase(Locale.ENGLISH))) {
+        final String nodeName = node.nodeName();
+        if (skipNodes.contains(nodeName.toUpperCase(Locale.ENGLISH))) {
+          if (!toSkip && phrasesJoined.charAt(scanIndex) == SEP) {
+            output.append("<wbr>");
+            scanIndex++;
+          }
           toSkip = true;
         }
+        output.append(String.format("<%s%s>", nodeName, attributesEncoded));
       } else if (node instanceof TextNode) {
         String data = ((TextNode) node).getWholeText();
         for (int i = 0; i < data.length(); i++) {
@@ -105,6 +113,8 @@ public void tail(Node node, int depth) {
       if (node.nodeName().equals("body") || node instanceof TextNode) {
         return;
       }
+      assert node instanceof Element;
+      toSkip = elementStack.pop();
       output.append(String.format("</%s>", node.nodeName()));
     }
   }

java/src/test/java/com/google/budoux/HTMLProcessorTest.java

@@ -51,12 +51,23 @@ public void testResolveWithStandardHTMLInput() {
 
   @Test
   public void testResolveWithNodesToSkip() {
-    List<String> phrases = Arrays.asList("abc", "def");
-    String html = "a<button>bcde</button>f";
+    List<String> phrases = Arrays.asList("abc", "def", "ghi");
+    String html = "a<button>bcde</button>fghi";
+    String result = HTMLProcessor.resolve(phrases, html);
+    assertEquals(
+        "<span style=\"word-break: keep-all; overflow-wrap:"
+            + " anywhere;\">a<button>bcde</button>f<wbr>ghi</span>",
+        result);
+  }
+
+  @Test
+  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);
     assertEquals(
         "<span style=\"word-break: keep-all; overflow-wrap:"
-            + " anywhere;\">a<button>bcde</button>f</span>",
+            + " anywhere;\">abc<wbr><nobr>defghi</nobr><wbr>jkl</span>",
         result);
   }