Read page breaks

7 files changed, 59 insertions, 22 deletions

diff --git a/src/main/java/org/zwobble/mammoth/internal/conversion/DocumentToHtml.java b/src/main/java/org/zwobble/mammoth/internal/conversion/DocumentToHtml.java
index b22742c..e88688a 100644
--- a/src/main/java/org/zwobble/mammoth/internal/conversion/DocumentToHtml.java
+++ b/src/main/java/org/zwobble/mammoth/internal/conversion/DocumentToHtml.java
@@ -186,8 +186,12 @@ public class DocumentToHtml {
             }
 
             @Override
-            public List<HtmlNode> visit(LineBreak lineBreak) {
-                return list(Html.selfClosingElement("br"));
+            public List<HtmlNode> visit(Break breakElement) {
+                if (breakElement.getType() == Break.Type.LINE) {
+                    return list(Html.selfClosingElement("br"));
+                } else {
+                    return list();
+                }
             }
 
             @Override
diff --git a/src/main/java/org/zwobble/mammoth/internal/documents/Break.java b/src/main/java/org/zwobble/mammoth/internal/documents/Break.java
new file mode 100644
index 0000000..9076d21
--- /dev/null
+++ b/src/main/java/org/zwobble/mammoth/internal/documents/Break.java
@@ -0,0 +1,26 @@
+package org.zwobble.mammoth.internal.documents;
+
+public class Break implements DocumentElement {
+    public enum Type {
+        LINE,
+        PAGE
+    }
+
+    public static final Break LINE_BREAK = new Break(Type.LINE);
+    public static final Break PAGE_BREAK = new Break(Type.PAGE);
+
+    private final Type type;
+
+    private Break(Type type) {
+        this.type = type;
+    }
+
+    public Type getType() {
+        return type;
+    }
+
+    @Override
+    public <T> T accept(DocumentElementVisitor<T> visitor) {
+        return visitor.visit(this);
+    }
+}
diff --git a/src/main/java/org/zwobble/mammoth/internal/documents/DocumentElementVisitor.java b/src/main/java/org/zwobble/mammoth/internal/documents/DocumentElementVisitor.java
index e087746..c83e3c5 100644
--- a/src/main/java/org/zwobble/mammoth/internal/documents/DocumentElementVisitor.java
+++ b/src/main/java/org/zwobble/mammoth/internal/documents/DocumentElementVisitor.java
@@ -6,7 +6,7 @@ public interface DocumentElementVisitor<T> {
     T visit(Text text);
 
     T visit(Tab tab);
-    T visit(LineBreak lineBreak);
+    T visit(Break lineBreak);
 
     T visit(Table table);
     T visit(TableRow tableRow);
diff --git a/src/main/java/org/zwobble/mammoth/internal/documents/LineBreak.java b/src/main/java/org/zwobble/mammoth/internal/documents/LineBreak.java
deleted file mode 100644
index 00b2a06..0000000
--- a/src/main/java/org/zwobble/mammoth/internal/documents/LineBreak.java
+++ /dev/null
@@ -1,13 +0,0 @@
-package org.zwobble.mammoth.internal.documents;
-
-public class LineBreak implements DocumentElement {
-    public static final LineBreak LINE_BREAK = new LineBreak();
-
-    private LineBreak() {
-    }
-
-    @Override
-    public <T> T accept(DocumentElementVisitor<T> visitor) {
-        return visitor.visit(this);
-    }
-}
diff --git a/src/main/java/org/zwobble/mammoth/internal/docx/BodyXmlReader.java b/src/main/java/org/zwobble/mammoth/internal/docx/BodyXmlReader.java
index ba09182..35f4aa1 100644
--- a/src/main/java/org/zwobble/mammoth/internal/docx/BodyXmlReader.java
+++ b/src/main/java/org/zwobble/mammoth/internal/docx/BodyXmlReader.java
@@ -243,7 +243,9 @@ public class BodyXmlReader {
     private ReadResult readBreak(XmlElement element) {
         String breakType = element.getAttributeOrNone("w:type").orElse("");
         if (breakType.equals("")) {
-            return success(LineBreak.LINE_BREAK);
+            return success(Break.LINE_BREAK);
+        } else if (breakType.equals("page")) {
+            return success(Break.PAGE_BREAK);
         } else {
             return ReadResult.emptyWithWarning("Unsupported break type: " + breakType);
         }
diff --git a/src/test/java/org/zwobble/mammoth/tests/conversion/DocumentToHtmlTests.java b/src/test/java/org/zwobble/mammoth/tests/conversion/DocumentToHtmlTests.java
index b144b78..c93460e 100644
--- a/src/test/java/org/zwobble/mammoth/tests/conversion/DocumentToHtmlTests.java
+++ b/src/test/java/org/zwobble/mammoth/tests/conversion/DocumentToHtmlTests.java
@@ -197,11 +197,19 @@ public class DocumentToHtmlTests {
     @Test
     public void lineBreakIsConvertedToBreakElement() {
         assertThat(
-            convertToHtml(LineBreak.LINE_BREAK),
+            convertToHtml(Break.LINE_BREAK),
             deepEquals(list(Html.selfClosingElement("br"))));
     }
 
     @Test
+    public void breaksThatAreNotLineBreaksAreIgnored() {
+        assertThat(
+            convertToHtml(Break.PAGE_BREAK),
+            deepEquals(list())
+        );
+    }
+
+    @Test
     public void tableIsConvertedToHtmlTable() {
         assertThat(
             convertToHtml(new Table(list(
diff --git a/src/test/java/org/zwobble/mammoth/tests/docx/BodyXmlTests.java b/src/test/java/org/zwobble/mammoth/tests/docx/BodyXmlTests.java
index 05ea9f0..1206c41 100644
--- a/src/test/java/org/zwobble/mammoth/tests/docx/BodyXmlTests.java
+++ b/src/test/java/org/zwobble/mammoth/tests/docx/BodyXmlTests.java
@@ -348,21 +348,31 @@ public class BodyXmlTests {
     }
 
     @Test
-    public void brIsReadAsLineBreak() {
+    public void brWithoutExplicitTypeIsReadAsLineBreak() {
         XmlElement element = element("w:br");
 
         assertThat(
             readSuccess(bodyReader(), element),
-            equalTo(LineBreak.LINE_BREAK));
+            equalTo(Break.LINE_BREAK));
     }
 
     @Test
-    public void warningOnBreaksThatArentLineBreaks() {
+    public void brWithPageTypeIsReadAsPageBreak() {
         XmlElement element = element("w:br", map("w:type", "page"));
 
         assertThat(
+            readSuccess(bodyReader(), element),
+            equalTo(Break.PAGE_BREAK)
+        );
+    }
+
+    @Test
+    public void warningOnBreaksThatArentLineBreaks() {
+        XmlElement element = element("w:br", map("w:type", "column"));
+
+        assertThat(
             readAll(bodyReader(), element),
-            isInternalResult(equalTo(list()), list("Unsupported break type: page")));
+            isInternalResult(equalTo(list()), list("Unsupported break type: column")));
     }
 
     @Test