Handle hyperlinked wp:anchor and wp:inline elements

Author: mwilliamson

NEWS

@@ -1,3 +1,7 @@
+# 1.12.0
+
+* Handle hyperlinked wp:anchor and wp:inline elements.
+
 # 1.11.0
 
 * Ignore style definitions using a style ID that has already been used.

README.md

@@ -330,6 +330,7 @@ For instance:
   automatically convert the document into HTML on the server,
   and embed the HTML into your website,
   this may allow arbitrary files on the server to be read and exfiltrated.
+
   To avoid this issue, access to any such external files is disabled by default.
   To enable access when converting trusted source documents,
   call `enableExternalFileAccess()`.

src/main/java/org/zwobble/mammoth/internal/docx/StatefulBodyXmlReader.java

@@ -694,34 +694,52 @@ private ReadResult readImagedata(XmlElement element) {
 
     private ReadResult readInline(XmlElement element) {
         XmlElementLike properties = element.findChildOrEmpty("wp:docPr");
+
         Optional<String> altText = Optionals.first(
             properties.getAttributeOrNone("descr").filter(description -> !description.trim().isEmpty()),
             properties.getAttributeOrNone("title")
         );
+
+        XmlElementLike hlinkClickElement = properties.findChildOrEmpty("a:hlinkClick");
+        Optional<String> href = hlinkClickElement.getAttributeOrNone("r:id")
+            .map(relationships::findTargetByRelationshipId);
+
         XmlElementList blips = element.findChildren("a:graphic")
             .findChildren("a:graphicData")
             .findChildren("pic:pic")
             .findChildren("pic:blipFill")
             .findChildren("a:blip");
-        return readBlips(blips, altText);
+        return readBlips(blips, altText, href);
     }
 
-    private ReadResult readBlips(XmlElementList blips, Optional<String> altText) {
-        return ReadResult.flatMap(blips, blip -> readBlip(blip, altText));
+    private ReadResult readBlips(
+        XmlElementList blips,
+        Optional<String> altText,
+        Optional<String> href
+    ) {
+        return ReadResult.flatMap(blips, blip -> readBlip(blip, altText, href));
     }
 
-    private ReadResult readBlip(XmlElement blip, Optional<String> altText) {
-        Optional<String> embedRelationshipId = blip.getAttributeOrNone("r:embed");
-        Optional<String> linkRelationshipId = blip.getAttributeOrNone("r:link");
-        if (embedRelationshipId.isPresent()) {
-            String imagePath = relationshipIdToDocxPath(embedRelationshipId.get());
-            return readImage(imagePath, altText, () -> Archives.getInputStream(file, imagePath));
-        } else if (linkRelationshipId.isPresent()) {
-            String imagePath = relationships.findTargetByRelationshipId(linkRelationshipId.get());
-            return readImage(imagePath, altText, () -> fileReader.getInputStream(imagePath));
-        } else {
+    private ReadResult readBlip(
+        XmlElement blip,
+        Optional<String> altText,
+        Optional<String> href
+    ) {
+        Optional<BlipImage> blipImage = findBlipImage(blip);
+
+        if (!blipImage.isPresent()) {
             return ReadResult.emptyWithWarning("Could not find image file for a:blip element");
         }
+
+        ReadResult result = readImage(blipImage.get().path, altText, blipImage.get().open);
+
+        if (!href.isPresent()) {
+            return result;
+        }
+
+        return result.map(imageElements -> list(
+            new Hyperlink(href, Optional.empty(), Optional.empty(), imageElements)
+        ));
     }
 
     private ReadResult readImage(String imagePath, Optional<String> altText, InputStreamSupplier open) {
@@ -736,6 +754,36 @@ private ReadResult readImage(String imagePath, Optional<String> altText, InputSt
         }
     }
 
+    private static class BlipImage {
+        private final String path;
+        private final InputStreamSupplier open;
+
+        public BlipImage(String path, InputStreamSupplier open) {
+            this.path = path;
+            this.open = open;
+        }
+    }
+
+    private Optional<BlipImage> findBlipImage(XmlElement blip) {
+        Optional<String> embedRelationshipId = blip.getAttributeOrNone("r:embed");
+        Optional<String> linkRelationshipId = blip.getAttributeOrNone("r:link");
+        if (embedRelationshipId.isPresent()) {
+            String imagePath = relationshipIdToDocxPath(embedRelationshipId.get());
+            return Optional.of(new BlipImage(
+                imagePath,
+                () -> Archives.getInputStream(file, imagePath)
+            ));
+        } else if (linkRelationshipId.isPresent()) {
+            String imagePath = relationships.findTargetByRelationshipId(linkRelationshipId.get());
+            return Optional.of(new BlipImage(
+                imagePath,
+                () -> fileReader.getInputStream(imagePath)
+            ));
+        } else {
+            return Optional.empty();
+        }
+    }
+
     private ReadResult readSdt(XmlElement element) {
         ReadResult contentResult = readElements(
             element.findChildOrEmpty("w:sdtContent").getChildren()

src/test/java/org/zwobble/mammoth/tests/docx/BodyXmlTests.java

@@ -1944,6 +1944,42 @@ public void noElementsCreatedIfImageCannotBeFoundInWpInline() {
         assertThat(result, isInternalSuccess(empty()));
     }
 
+    @Test
+    public void canReadPicturesWithHyperlinkSpecifiedInDocumentProperties() throws IOException {
+        XmlElement element = element("w:drawing", list(
+            element("wp:inline", list(
+                element("wp:docPr", map(), list(
+                    element("a:hlinkClick", map("r:id", "rId42"))
+                )),
+                graphicXml(embeddedBlipXml(IMAGE_RELATIONSHIP_ID))
+            ))
+        ));
+        Relationships relationships = new Relationships(list(
+            imageRelationship(IMAGE_RELATIONSHIP_ID, "media/hat.png"),
+            hyperlinkRelationship("rId42", "http://example.com")
+        ));
+        Archive file = InMemoryArchive.fromStrings(map("word/media/hat.png", IMAGE_BYTES));
+
+        DocumentElement result = readSuccess(
+            bodyReader(relationships, file),
+            element
+        );
+
+        assertThat(result, isHyperlink(
+            hasHref("http://example.com"),
+            hasChildren(
+                isImage(
+                    hasProperty("contentType", deepEquals(Optional.of("image/png")))
+                )
+            )
+        ));
+        Image image = (Image) ((Hyperlink) result).getChildren().get(0);
+        assertThat(
+            toString(image.open()),
+            equalTo(IMAGE_BYTES)
+        );
+    }
+
     private XmlElement inlineImageXml(XmlElement blip, String description) {
         return inlineImageXml(blip, Optional.of(description), Optional.empty());
     }
@@ -1965,10 +2001,18 @@ private List<XmlNode> imageXml(XmlElement blip, Optional<String> description, Op
 
         return list(
             element("wp:docPr", properties),
-            element("a:graphic", list(
-                element("a:graphicData", list(
-                    element("pic:pic", list(
-                        element("pic:blipFill", list(blip)))))))));
+            graphicXml(blip)
+        );
+    }
+
+    private XmlNode graphicXml(XmlElement blip) {
+        return element("a:graphic", list(
+            element("a:graphicData", list(
+                element("pic:pic", list(
+                    element("pic:blipFill", list(blip))
+                ))
+            ))
+        ));
     }
 
     private XmlElement embeddedBlipXml(String relationshipId) {

src/test/java/org/zwobble/mammoth/tests/docx/DocumentMatchers.java

@@ -122,4 +122,9 @@ static Matcher<Hyperlink> hasTargetFrame(String targetFrame) {
     static Matcher<Hyperlink> hasNoTargetFrame() {
         return hasProperty("targetFrame", equalTo(Optional.empty()));
     }
+
+    @SafeVarargs
+    static Matcher<DocumentElement> isImage(Matcher<? super Image>... matchers) {
+        return cast(Image.class, allOf(matchers));
+    }
 }