Generate HTML using HTML AST

6 files changed, 105 insertions, 25 deletions

diff --git a/src/main/java/org/zwobble/mammoth/Mammoth.java b/src/main/java/org/zwobble/mammoth/Mammoth.java
index 63a313f..40f2a6e 100644
--- a/src/main/java/org/zwobble/mammoth/Mammoth.java
+++ b/src/main/java/org/zwobble/mammoth/Mammoth.java
@@ -1,17 +1,20 @@
 package org.zwobble.mammoth;
 
-import com.google.common.base.Joiner;
 import org.zwobble.mammoth.documents.*;
 import org.zwobble.mammoth.docx.*;
+import org.zwobble.mammoth.html.Html;
+import org.zwobble.mammoth.html.HtmlNode;
 import org.zwobble.mammoth.results.Result;
 import org.zwobble.mammoth.xml.XmlElement;
 
 import java.io.File;
 import java.io.IOException;
 import java.io.InputStream;
+import java.util.List;
 import java.util.zip.ZipFile;
 
-import static com.google.common.collect.Iterables.transform;
+import static org.zwobble.mammoth.util.MammothLists.eagerFlatMap;
+import static org.zwobble.mammoth.util.MammothLists.list;
 
 public class Mammoth {
     public static Result<String> convertToHtml(File file) {
@@ -36,86 +39,91 @@ public class Mammoth {
                 zipFile,
                 fileReader));
             return reader.readElement(documentXml)
-                .map(Mammoth::convertToHtml);
+                .map(Mammoth::convertToHtml)
+                .map(Html::write);
         } catch (IOException e) {
             throw new UnsupportedOperationException("Should return a result of failure");   
         }
     }
 
-    private static String convertToHtml(Document document) {
+    private static List<HtmlNode> convertToHtml(Document document) {
         return convertChildrenToHtml(document);
     }
 
-    private static String convertChildrenToHtml(HasChildren element) {
-        return Joiner.on("").join(transform(
+    private static List<HtmlNode> convertChildrenToHtml(HasChildren element) {
+        return eagerFlatMap(
             element.getChildren(),
-            Mammoth::convertToHtml));
+            Mammoth::convertToHtml);
     }
 
-    private static String convertToHtml(DocumentElement element) {
-        return element.accept(new DocumentElementVisitor<String>() {
+    private static List<HtmlNode> convertToHtml(DocumentElement element) {
+        return element.accept(new DocumentElementVisitor<List<HtmlNode>>() {
             @Override
-            public String visit(Paragraph paragraph) {
-                String content = convertChildrenToHtml(paragraph);
+            public List<HtmlNode> visit(Paragraph paragraph) {
+                List<HtmlNode> content = convertChildrenToHtml(paragraph);
                 if (content.isEmpty()) {
-                    return "";
+                    return list();
                 } else {
-                    return "<p>" + content + "</p>";
+                    return list(Html.element("p", content));
                 }
             }
 
             @Override
-            public String visit(Run run) {
+            public List<HtmlNode> visit(Run run) {
                 return convertChildrenToHtml(run);
             }
 
             @Override
-            public String visit(Text text) {
-                return text.getValue();
+            public List<HtmlNode> visit(Text text) {
+                if (text.getValue().isEmpty()) {
+                    return list();
+                } else {
+                    return list(Html.text(text.getValue()));
+                }
             }
 
             @Override
-            public String visit(Tab tab) {
+            public List<HtmlNode> visit(Tab tab) {
                 throw new UnsupportedOperationException();
             }
 
             @Override
-            public String visit(LineBreak lineBreak) {
+            public List<HtmlNode> visit(LineBreak lineBreak) {
                 throw new UnsupportedOperationException();
             }
 
             @Override
-            public String visit(Table table) {
+            public List<HtmlNode> visit(Table table) {
                 throw new UnsupportedOperationException();
             }
 
             @Override
-            public String visit(TableRow tableRow) {
+            public List<HtmlNode> visit(TableRow tableRow) {
                 throw new UnsupportedOperationException();
             }
 
             @Override
-            public String visit(TableCell tableCell) {
+            public List<HtmlNode> visit(TableCell tableCell) {
                 throw new UnsupportedOperationException();
             }
 
             @Override
-            public String visit(Hyperlink hyperlink) {
+            public List<HtmlNode> visit(Hyperlink hyperlink) {
                 throw new UnsupportedOperationException();
             }
 
             @Override
-            public String visit(Bookmark bookmark) {
+            public List<HtmlNode> visit(Bookmark bookmark) {
                 throw new UnsupportedOperationException();
             }
 
             @Override
-            public String visit(NoteReference noteReference) {
+            public List<HtmlNode> visit(NoteReference noteReference) {
                 throw new UnsupportedOperationException();
             }
 
             @Override
-            public String visit(Image image) {
+            public List<HtmlNode> visit(Image image) {
                 throw new UnsupportedOperationException();
             }
         });
diff --git a/src/main/java/org/zwobble/mammoth/html/Html.java b/src/main/java/org/zwobble/mammoth/html/Html.java
new file mode 100644
index 0000000..9b6ccbb
--- /dev/null
+++ b/src/main/java/org/zwobble/mammoth/html/Html.java
@@ -0,0 +1,19 @@
+package org.zwobble.mammoth.html;
+
+import java.util.List;
+
+public class Html {
+    public static String write(List<HtmlNode> nodes) {
+        StringBuilder builder = new StringBuilder();
+        nodes.forEach(node -> node.write(builder));
+        return builder.toString();
+    }
+
+    public static HtmlNode text(String value) {
+        return new HtmlTextNode(value);
+    }
+
+    public static HtmlNode element(String tagName, List<HtmlNode> children) {
+        return new HtmlElementNode(tagName, children);
+    }
+}
diff --git a/src/main/java/org/zwobble/mammoth/html/HtmlElementNode.java b/src/main/java/org/zwobble/mammoth/html/HtmlElementNode.java
new file mode 100644
index 0000000..34de63e
--- /dev/null
+++ b/src/main/java/org/zwobble/mammoth/html/HtmlElementNode.java
@@ -0,0 +1,28 @@
+package org.zwobble.mammoth.html;
+
+import java.util.List;
+
+public class HtmlElementNode implements HtmlNode {
+    private final String tagName;
+    private final List<HtmlNode> children;
+
+    public HtmlElementNode(String tagName, List<HtmlNode> children) {
+        this.tagName = tagName;
+        this.children = children;
+    }
+
+    @Override
+    public void write(StringBuilder builder) {
+        builder
+            .append("<")
+            .append(tagName)
+            .append(">");
+
+        children.forEach(node -> node.write(builder));
+
+        builder
+            .append("</")
+            .append(tagName)
+            .append(">");
+    }
+}
diff --git a/src/main/java/org/zwobble/mammoth/html/HtmlNode.java b/src/main/java/org/zwobble/mammoth/html/HtmlNode.java
new file mode 100644
index 0000000..9977c66
--- /dev/null
+++ b/src/main/java/org/zwobble/mammoth/html/HtmlNode.java
@@ -0,0 +1,5 @@
+package org.zwobble.mammoth.html;
+
+public interface HtmlNode {
+    void write(StringBuilder builder);
+}
diff --git a/src/main/java/org/zwobble/mammoth/html/HtmlTextNode.java b/src/main/java/org/zwobble/mammoth/html/HtmlTextNode.java
new file mode 100644
index 0000000..2c3a7c2
--- /dev/null
+++ b/src/main/java/org/zwobble/mammoth/html/HtmlTextNode.java
@@ -0,0 +1,14 @@
+package org.zwobble.mammoth.html;
+
+public class HtmlTextNode implements HtmlNode {
+    private final String value;
+
+    public HtmlTextNode(String  value) {
+        this.value = value;
+    }
+
+    @Override
+    public void write(StringBuilder builder) {
+        builder.append(value);
+    }
+}
diff --git a/src/main/java/org/zwobble/mammoth/util/MammothLists.java b/src/main/java/org/zwobble/mammoth/util/MammothLists.java
index e038ec7..59f23f8 100644
--- a/src/main/java/org/zwobble/mammoth/util/MammothLists.java
+++ b/src/main/java/org/zwobble/mammoth/util/MammothLists.java
@@ -1,6 +1,8 @@
 package org.zwobble.mammoth.util;
 
+import com.google.common.base.Function;
 import com.google.common.collect.ImmutableList;
+import com.google.common.collect.Iterables;
 
 import java.util.List;
 
@@ -16,4 +18,8 @@ public class MammothLists {
     public static <T> List<T> list(T value1, T value2) {
         return ImmutableList.of(value1, value2);
     }
+
+    public static <T, R> List<R> eagerFlatMap(Iterable<T> iterable, Function<T, Iterable<R>> function) {
+        return ImmutableList.copyOf(Iterables.concat(Iterables.transform(iterable, function)));
+    }
 }