Support merged paragraphs when revisions are tracked

4 files changed, 86 insertions, 7 deletions

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 ace6628..4282bf6 100644
--- a/src/main/java/org/zwobble/mammoth/internal/docx/BodyXmlReader.java
+++ b/src/main/java/org/zwobble/mammoth/internal/docx/BodyXmlReader.java
@@ -29,7 +29,7 @@ public class BodyXmlReader {
         this.fileReader = fileReader;
     }
 
-    ReadResult readElements(Iterable<XmlNode> nodes) {
+    public ReadResult readElements(Iterable<? extends XmlNode> nodes) {
         return new StatefulBodyXmlReader(
             styles,
             numbering,
diff --git a/src/main/java/org/zwobble/mammoth/internal/docx/StatefulBodyXmlReader.java b/src/main/java/org/zwobble/mammoth/internal/docx/StatefulBodyXmlReader.java
index b549387..3be34ad 100644
--- a/src/main/java/org/zwobble/mammoth/internal/docx/StatefulBodyXmlReader.java
+++ b/src/main/java/org/zwobble/mammoth/internal/docx/StatefulBodyXmlReader.java
@@ -19,6 +19,7 @@ import static org.zwobble.mammoth.internal.docx.ReadResult.*;
 import static org.zwobble.mammoth.internal.docx.Uris.uriToZipEntryName;
 import static org.zwobble.mammoth.internal.util.Iterables.lazyFilter;
 import static org.zwobble.mammoth.internal.util.Iterables.tryGetLast;
+import static org.zwobble.mammoth.internal.util.Lists.eagerConcat;
 import static org.zwobble.mammoth.internal.util.Lists.list;
 import static org.zwobble.mammoth.internal.util.Maps.entry;
 import static org.zwobble.mammoth.internal.util.Maps.lookup;
@@ -36,6 +37,7 @@ class StatefulBodyXmlReader {
     private final FileReader fileReader;
     private final StringBuilder currentInstrText;
     private final Queue<ComplexField> complexFieldStack;
+    private List<XmlNode> deletedParagraphContents;
 
     private interface ComplexField {
         ComplexField UNKNOWN = new ComplexField() {};
@@ -70,6 +72,7 @@ class StatefulBodyXmlReader {
         this.fileReader = fileReader;
         this.currentInstrText = new StringBuilder();
         this.complexFieldStack = Queues.stack();
+        this.deletedParagraphContents = new ArrayList<>();
     }
 
     ReadResult readElement(XmlElement element) {
@@ -249,17 +252,33 @@ class StatefulBodyXmlReader {
         return readStyle(properties, "w:rStyle", "Run", styles::findCharacterStyleById);
     }
 
-    ReadResult readElements(Iterable<XmlNode> nodes) {
+    ReadResult readElements(Iterable<? extends XmlNode> nodes) {
         return ReadResult.flatMap(lazyFilter(nodes, XmlElement.class), this::readElement);
     }
 
     private ReadResult readParagraph(XmlElement element) {
         XmlElementLike properties = element.findChildOrEmpty("w:pPr");
-        ParagraphIndent indent = readParagraphIndent(properties);
-        return ReadResult.map(
-            readParagraphStyle(properties),
-            readElements(element.getChildren()),
-            (style, children) -> new Paragraph(style, readNumbering(style, properties), indent, children)).appendExtra();
+
+        boolean isDeleted = properties
+            .findChildOrEmpty("w:rPr")
+            .findChild("w:del")
+            .isPresent();
+
+        if (isDeleted) {
+            deletedParagraphContents.addAll(element.getChildren());
+            return ReadResult.success(list());
+        } else {
+            ParagraphIndent indent = readParagraphIndent(properties);
+            List<XmlNode> childrenXml = element.getChildren();
+            if (!deletedParagraphContents.isEmpty()) {
+                childrenXml = eagerConcat(deletedParagraphContents, childrenXml);
+                deletedParagraphContents = new ArrayList<>();
+            }
+            return ReadResult.map(
+                readParagraphStyle(properties),
+                readElements(childrenXml),
+                (style, children) -> new Paragraph(style, readNumbering(style, properties), indent, children)).appendExtra();
+        }
     }
 
     private ReadResult readFieldChar(XmlElement element) {
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 1898897..c8b6c7c 100644
--- a/src/test/java/org/zwobble/mammoth/tests/docx/BodyXmlTests.java
+++ b/src/test/java/org/zwobble/mammoth/tests/docx/BodyXmlTests.java
@@ -264,6 +264,62 @@ public class BodyXmlTests {
             hasNumbering(Optional.empty()));
     }
 
+    @Test
+    public void contentOfDeletedParagraphIsPreprendedToNextParagraph() {
+        Style heading1 = new Style("Heading1", Optional.of("Heading 1"));
+        Style heading2 = new Style("Heading2", Optional.of("Heading 2"));
+        Styles styles = new Styles(
+            map(
+                "Heading1", heading1,
+                "Heading2", heading2
+            ),
+            map(),
+            map(),
+            map()
+        );
+        List<XmlElement> bodyXml = list(
+            element("w:p", list(
+                element("w:pPr", list(
+                    element("w:pStyle", map("w:val", "Heading1")),
+                    element("w:rPr", list(
+                        element("w:del")
+                    ))
+                )),
+                runXml("One")
+            )),
+            element("w:p", list(
+                element("w:pPr", list(
+                    element("w:pStyle", map("w:val", "Heading2"))
+                )),
+                runXml("Two")
+            )),
+            // Include a second paragraph that isn't deleted to ensure we only add
+            // the deleted paragraph contents once.
+            element("w:p", list(
+                runXml("Three")
+            ))
+        );
+
+        BodyXmlReader reader = bodyReader(styles);
+        InternalResult<List<DocumentElement>> result = reader.readElements(bodyXml).toResult();
+
+        assertThat(result.getWarnings(), emptyIterable());
+        assertThat(result.getValue(), contains(
+            isParagraph(
+                hasParagraphStyle(equalTo(Optional.of(heading2))),
+                hasChildren(
+                    isRun(run(withChildren(text("One")))),
+                    isRun(run(withChildren(text("Two"))))
+                )
+            ),
+            isParagraph(
+                hasChildren(
+                    isRun(run(withChildren(text("Three"))))
+                )
+            )
+        ));
+    }
+
     @Nested
     public class ComplexFieldsTests {
         private final String URI = "http://example.com";
diff --git a/src/test/java/org/zwobble/mammoth/tests/docx/DocumentMatchers.java b/src/test/java/org/zwobble/mammoth/tests/docx/DocumentMatchers.java
index 53632c1..4f3628e 100644
--- a/src/test/java/org/zwobble/mammoth/tests/docx/DocumentMatchers.java
+++ b/src/test/java/org/zwobble/mammoth/tests/docx/DocumentMatchers.java
@@ -37,6 +37,10 @@ public class DocumentMatchers {
         return cast(Paragraph.class, allOf(matchers));
     }
 
+    static Matcher<Paragraph> hasParagraphStyle(Matcher<Optional<Style>> style) {
+        return hasProperty("style", style);
+    }
+
     static Matcher<DocumentElement> isEmptyRun() {
         return isRun(hasChildren());
     }