Read text boxes

3 files changed, 75 insertions, 18 deletions

diff --git a/src/main/java/org/zwobble/mammoth/docx/BodyXmlReader.java b/src/main/java/org/zwobble/mammoth/docx/BodyXmlReader.java
index c8e1785..19f649a 100644
--- a/src/main/java/org/zwobble/mammoth/docx/BodyXmlReader.java
+++ b/src/main/java/org/zwobble/mammoth/docx/BodyXmlReader.java
@@ -59,12 +59,15 @@ public class BodyXmlReader {
             case "w:endnoteReference":
                 return readNoteReference(NoteType.ENDNOTE, element);
 
+            case "w:pict":
+                return readPict(element);
+
             case "w:ins":
             case "w:smartTag":
             case "w:drawing":
-            case "w:roundrect":
-            case "w:shape":
-            case "w:textbox":
+            case "v:roundrect":
+            case "v:shape":
+            case "v:textbox":
             case "w:txbxContent":
                 return readElements(element.children());
 
@@ -90,7 +93,7 @@ public class BodyXmlReader {
 
             default:
                 Warning warning = warning("An unrecognised element was ignored: " + element.getName());
-                return new ReadResult(list(), list(warning));
+                return ReadResult.emptyWithWarning(warning);
         }
     }
 
@@ -141,7 +144,7 @@ public class BodyXmlReader {
         return ReadResult.map(
             readParagraphStyle(properties),
             readElements(element.children()),
-            (style, children) -> new Paragraph(style, numbering, children));
+            (style, children) -> new Paragraph(style, numbering, children)).appendExtra();
     }
 
     private Result<Optional<Style>> readParagraphStyle(XmlElementLike properties) {
@@ -188,7 +191,7 @@ public class BodyXmlReader {
         if (breakType.equals("")) {
             return success(LineBreak.LINE_BREAK);
         } else {
-            return new ReadResult(list(), list(warning("Unsupported break type: " + breakType)));
+            return ReadResult.emptyWithWarning(warning("Unsupported break type: " + breakType));
         }
     }
 
@@ -221,6 +224,10 @@ public class BodyXmlReader {
         return success(new NoteReference(noteType, noteId));
     }
 
+    private ReadResult readPict(XmlElement element) {
+        return readElements(element.children()).toExtra();
+    }
+
     private Optional<String> readVal(XmlElementLike element, String name) {
         return element.findChildOrEmpty(name).getAttributeOrNone("w:val");
     }
diff --git a/src/main/java/org/zwobble/mammoth/docx/ReadResult.java b/src/main/java/org/zwobble/mammoth/docx/ReadResult.java
index 7c672b2..31bbf2a 100644
--- a/src/main/java/org/zwobble/mammoth/docx/ReadResult.java
+++ b/src/main/java/org/zwobble/mammoth/docx/ReadResult.java
@@ -16,12 +16,14 @@ public class ReadResult {
 
     public static ReadResult concat(Iterable<ReadResult> results) {
         ImmutableList.Builder<DocumentElement> elements = ImmutableList.builder();
+        ImmutableList.Builder<DocumentElement> extra = ImmutableList.builder();
         ImmutableList.Builder<Warning> warnings = ImmutableList.builder();
         for (ReadResult result : results) {
             elements.addAll(result.elements);
+            extra.addAll(result.extra);
             warnings.addAll(result.warnings);
         }
-        return new ReadResult(elements.build(), warnings.build());
+        return new ReadResult(elements.build(), extra.build(), warnings.build());
     }
 
     public static <T> ReadResult map(
@@ -29,13 +31,10 @@ public class ReadResult {
         ReadResult second,
         BiFunction<T, List<DocumentElement>, DocumentElement> function)
     {
-        ImmutableList.Builder<Warning> warnings = ImmutableList.builder();
-        warnings.addAll(first.getWarnings());
-        warnings.addAll(second.warnings);
-
         return new ReadResult(
             list(function.apply(first.getValue(), second.elements)),
-            warnings.build());
+            second.extra,
+            concat(first.getWarnings(), second.warnings));
     }
 
     public static ReadResult success(DocumentElement element) {
@@ -43,22 +42,46 @@ public class ReadResult {
     }
 
     public static ReadResult success(List<DocumentElement> elements) {
-        return new ReadResult(elements, list());
+        return new ReadResult(elements, list(), list());
+    }
+
+    public static ReadResult emptyWithWarning(Warning warning) {
+        return new ReadResult(list(), list(), list(warning));
     }
 
     private final List<DocumentElement> elements;
+    private final List<DocumentElement> extra;
     private final List<Warning> warnings;
 
-    public ReadResult(List<DocumentElement> elements, List<Warning> warnings) {
+    public ReadResult(List<DocumentElement> elements, List<DocumentElement> extra, List<Warning> warnings) {
         this.elements = elements;
+        this.extra = extra;
         this.warnings = warnings;
     }
 
     public ReadResult map(Function<List<DocumentElement>, DocumentElement> function) {
-        return new ReadResult(list(function.apply(elements)), warnings);
+        return new ReadResult(
+            list(function.apply(elements)),
+            extra,
+            warnings);
+    }
+
+    public ReadResult toExtra() {
+        return new ReadResult(list(), concat(extra, elements), warnings);
+    }
+
+    public ReadResult appendExtra() {
+        return new ReadResult(concat(elements, extra), list(), warnings);
     }
 
     public Result<List<DocumentElement>> toResult() {
         return new Result<>(elements, warnings);
     }
+
+    private static <T> List<T> concat(List<T> first, List<T> second) {
+        ImmutableList.Builder<T> builder = ImmutableList.builder();
+        builder.addAll(first);
+        builder.addAll(second);
+        return builder.build();
+    }
 }
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 d4fa358..c8f8887 100644
--- a/src/test/java/org/zwobble/mammoth/tests/docx/BodyXmlTests.java
+++ b/src/test/java/org/zwobble/mammoth/tests/docx/BodyXmlTests.java
@@ -354,13 +354,40 @@ public class BodyXmlTests {
     }
 
     @Test
+    public void textBoxesHaveContentAppendedAfterContainingParagraph() {
+        XmlElement textBox = element("w:pict", list(
+            element("v:shape", list(
+                element("v:textbox", list(
+                    element("w:txbxContent", list(
+                        paragraphXml(list(
+                            runXml(list(textXml("[textbox-content]")))))))))))));
+        XmlElement paragraph = paragraphXml(list(
+            runXml(list(textXml("[paragragh start]"))),
+            runXml(list(textBox, textXml("[paragragh end]")))));
+
+        List<DocumentElement> expected = list(
+            make(a(PARAGRAPH, with(CHILDREN, list(
+                make(a(RUN, with(CHILDREN, list(
+                    new Text("[paragragh start]"))))),
+                make(a(RUN, with(CHILDREN, list(
+                    new Text("[paragragh end]"))))))))),
+            make(a(PARAGRAPH, with(CHILDREN, list(
+                make(a(RUN, with(CHILDREN, list(
+                    new Text("[textbox-content]"))))))))));
+
+        assertThat(
+            readAll(a(bodyReader), paragraph),
+            isResult(deepEquals(expected), list()));
+    }
+
+    @Test
     public void appropriateElementsHaveTheirChildrenReadNormally() {
         assertChildrenAreReadNormally("w:ins");
         assertChildrenAreReadNormally("w:smartTag");
         assertChildrenAreReadNormally("w:drawing");
-        assertChildrenAreReadNormally("w:roundrect");
-        assertChildrenAreReadNormally("w:shape");
-        assertChildrenAreReadNormally("w:textbox");
+        assertChildrenAreReadNormally("v:roundrect");
+        assertChildrenAreReadNormally("v:shape");
+        assertChildrenAreReadNormally("v:textbox");
         assertChildrenAreReadNormally("w:txbxContent");
     }