Expose marked content IDs for images and alt text.

 - Adds two new public bits of API:
    - StructElement_GetMCID(STRUCTELEMENT) to return the marked content
      ID of the struct element.
    - IMAGEOBJ_METADATA.mcid to contain the marked content ID of the
      image.
 - Restores ContentMark::GetMCID, which was removed with other dead
   code in https://pdfium-review.googlesource.com/c/pdfium/+/19170.
 - Adds a couple calls to tests, including a simple new PDF to test
   the struct tree marked content ID.

Bug:pdfium:957
Change-Id: I92856e43d741df989e53a575a08258da19a39f22
Reviewed-on: https://pdfium-review.googlesource.com/20632
Reviewed-by: Lei Zhang <thestig@chromium.org>
Commit-Queue: Lei Zhang <thestig@chromium.org>

9 files changed, 63 insertions, 0 deletions

diff --git a/core/fpdfapi/page/cpdf_contentmark.cpp b/core/fpdfapi/page/cpdf_contentmark.cpp
index 1d70edeb51..c54ef55b1f 100644
--- a/core/fpdfapi/page/cpdf_contentmark.cpp
+++ b/core/fpdfapi/page/cpdf_contentmark.cpp
@@ -25,6 +25,11 @@ const CPDF_ContentMarkItem& CPDF_ContentMark::GetItem(size_t i) const {
   return m_Ref.GetObject()->GetItem(i);
 }
 
+int CPDF_ContentMark::GetMarkedContentID() const {
+  const MarkData* pData = m_Ref.GetObject();
+  return pData ? pData->GetMarkedContentID() : -1;
+}
+
 void CPDF_ContentMark::AddMark(const ByteString& name,
                                CPDF_Dictionary* pDict,
                                bool bDirect) {
@@ -53,6 +58,15 @@ const CPDF_ContentMarkItem& CPDF_ContentMark::MarkData::GetItem(
   return m_Marks[index];
 }
 
+int CPDF_ContentMark::MarkData::GetMarkedContentID() const {
+  for (const auto& mark : m_Marks) {
+    const CPDF_Dictionary* pDict = mark.GetParam();
+    if (pDict && pDict->KeyExist("MCID"))
+      return pDict->GetIntegerFor("MCID");
+  }
+  return -1;
+}
+
 void CPDF_ContentMark::MarkData::AddMark(const ByteString& name,
                                          CPDF_Dictionary* pDict,
                                          bool bDirect) {
diff --git a/core/fpdfapi/page/cpdf_contentmark.h b/core/fpdfapi/page/cpdf_contentmark.h
index d9bc35a7fb..78a948697e 100644
--- a/core/fpdfapi/page/cpdf_contentmark.h
+++ b/core/fpdfapi/page/cpdf_contentmark.h
@@ -21,6 +21,7 @@ class CPDF_ContentMark {
   CPDF_ContentMark(const CPDF_ContentMark& that);
   ~CPDF_ContentMark();
 
+  int GetMarkedContentID() const;
   size_t CountItems() const;
   const CPDF_ContentMarkItem& GetItem(size_t i) const;
 
@@ -39,6 +40,7 @@ class CPDF_ContentMark {
     size_t CountItems() const;
     const CPDF_ContentMarkItem& GetItem(size_t index) const;
 
+    int GetMarkedContentID() const;
     void AddMark(const ByteString& name,
                  CPDF_Dictionary* pDict,
                  bool bDictNeedClone);
diff --git a/fpdfsdk/fpdf_structtree.cpp b/fpdfsdk/fpdf_structtree.cpp
index 30b7c3fce7..676824ef99 100644
--- a/fpdfsdk/fpdf_structtree.cpp
+++ b/fpdfsdk/fpdf_structtree.cpp
@@ -83,6 +83,14 @@ FPDF_StructElement_GetAltText(FPDF_STRUCTELEMENT struct_element,
              : 0;
 }
 
+FPDF_EXPORT int FPDF_CALLCONV
+FPDF_StructElement_GetMarkedContentID(FPDF_STRUCTELEMENT struct_element) {
+  CPDF_StructElement* elem = ToStructTreeElement(struct_element);
+  CPDF_Object* p =
+      (elem && elem->GetDict()) ? elem->GetDict()->GetObjectFor("K") : nullptr;
+  return p && p->IsNumber() ? p->GetInteger() : -1;
+}
+
 FPDF_EXPORT unsigned long FPDF_CALLCONV
 FPDF_StructElement_GetType(FPDF_STRUCTELEMENT struct_element,
                            void* buffer,
diff --git a/fpdfsdk/fpdf_structtree_embeddertest.cpp b/fpdfsdk/fpdf_structtree_embeddertest.cpp
index 05b77fc0f7..7ca81f7b2e 100644
--- a/fpdfsdk/fpdf_structtree_embeddertest.cpp
+++ b/fpdfsdk/fpdf_structtree_embeddertest.cpp
@@ -6,6 +6,7 @@
 #include "public/fpdf_structtree.h"
 #include "testing/embedder_test.h"
 #include "testing/test_support.h"
+#include "third_party/base/optional.h"
 
 class FPDFStructTreeEmbeddertest : public EmbedderTest {};
 
@@ -24,6 +25,7 @@ TEST_F(FPDFStructTreeEmbeddertest, GetAltText) {
   EXPECT_EQ(nullptr, element);
   element = FPDF_StructTree_GetChildAtIndex(struct_tree, 0);
   ASSERT_NE(nullptr, element);
+  EXPECT_EQ(-1, FPDF_StructElement_GetMarkedContentID(element));
   EXPECT_EQ(0U, FPDF_StructElement_GetAltText(element, nullptr, 0));
 
   ASSERT_EQ(1, FPDF_StructElement_CountChildren(element));
@@ -34,6 +36,7 @@ TEST_F(FPDFStructTreeEmbeddertest, GetAltText) {
   EXPECT_EQ(nullptr, child_element);
   child_element = FPDF_StructElement_GetChildAtIndex(element, 0);
   ASSERT_NE(nullptr, child_element);
+  EXPECT_EQ(-1, FPDF_StructElement_GetMarkedContentID(child_element));
   EXPECT_EQ(0U, FPDF_StructElement_GetAltText(child_element, nullptr, 0));
 
   ASSERT_EQ(1, FPDF_StructElement_CountChildren(child_element));
@@ -44,6 +47,7 @@ TEST_F(FPDFStructTreeEmbeddertest, GetAltText) {
   EXPECT_EQ(nullptr, gchild_element);
   gchild_element = FPDF_StructElement_GetChildAtIndex(child_element, 0);
   ASSERT_NE(nullptr, gchild_element);
+  EXPECT_EQ(-1, FPDF_StructElement_GetMarkedContentID(gchild_element));
   ASSERT_EQ(24U, FPDF_StructElement_GetAltText(gchild_element, nullptr, 0));
 
   unsigned short buffer[12];
@@ -54,6 +58,7 @@ TEST_F(FPDFStructTreeEmbeddertest, GetAltText) {
   for (size_t i = 0; i < FX_ArraySize(buffer); ++i)
     EXPECT_EQ(0U, buffer[i]);
 
+  EXPECT_EQ(-1, FPDF_StructElement_GetMarkedContentID(gchild_element));
   ASSERT_EQ(24U, FPDF_StructElement_GetAltText(gchild_element, buffer,
                                                sizeof(buffer)));
   const wchar_t kExpected[] = L"Black Image";
@@ -69,6 +74,22 @@ TEST_F(FPDFStructTreeEmbeddertest, GetAltText) {
   FPDF_ClosePage(page);
 }
 
+TEST_F(FPDFStructTreeEmbeddertest, GetMarkedContentID) {
+  ASSERT_TRUE(OpenDocument("marked_content_id.pdf"));
+  FPDF_PAGE page = LoadPage(0);
+  ASSERT_TRUE(page);
+
+  FPDF_STRUCTTREE struct_tree = FPDF_StructTree_GetForPage(page);
+  ASSERT_TRUE(struct_tree);
+  ASSERT_EQ(1, FPDF_StructTree_CountChildren(struct_tree));
+
+  FPDF_STRUCTELEMENT element = FPDF_StructTree_GetChildAtIndex(struct_tree, 0);
+  EXPECT_EQ(0, FPDF_StructElement_GetMarkedContentID(element));
+
+  FPDF_StructTree_Close(struct_tree);
+  FPDF_ClosePage(page);
+}
+
 TEST_F(FPDFStructTreeEmbeddertest, GetType) {
   ASSERT_TRUE(OpenDocument("tagged_alt_text.pdf"));
   FPDF_PAGE page = LoadPage(0);
diff --git a/fpdfsdk/fpdfedit_embeddertest.cpp b/fpdfsdk/fpdfedit_embeddertest.cpp
index 6cb2771154..ee2fc7eb85 100644
--- a/fpdfsdk/fpdfedit_embeddertest.cpp
+++ b/fpdfsdk/fpdfedit_embeddertest.cpp
@@ -1205,6 +1205,7 @@ TEST_F(FPDFEditEmbeddertest, GetImageMetadata) {
   // |page|, all values are correct, with the last two being default values.
   ASSERT_EQ(FPDF_PAGEOBJ_IMAGE, FPDFPageObj_GetType(obj));
   ASSERT_TRUE(FPDFImageObj_GetImageMetadata(obj, nullptr, &metadata));
+  EXPECT_EQ(7, metadata.marked_content_id);
   EXPECT_EQ(92u, metadata.width);
   EXPECT_EQ(68u, metadata.height);
   EXPECT_NEAR(96.000000, metadata.horizontal_dpi, 0.001);
@@ -1214,6 +1215,7 @@ TEST_F(FPDFEditEmbeddertest, GetImageMetadata) {
 
   // Verify the metadata of a bitmap image with indexed colorspace.
   ASSERT_TRUE(FPDFImageObj_GetImageMetadata(obj, page, &metadata));
+  EXPECT_EQ(7, metadata.marked_content_id);
   EXPECT_EQ(92u, metadata.width);
   EXPECT_EQ(68u, metadata.height);
   EXPECT_NEAR(96.000000, metadata.horizontal_dpi, 0.001);
@@ -1225,6 +1227,7 @@ TEST_F(FPDFEditEmbeddertest, GetImageMetadata) {
   obj = FPDFPage_GetObject(page, 37);
   ASSERT_EQ(FPDF_PAGEOBJ_IMAGE, FPDFPageObj_GetType(obj));
   ASSERT_TRUE(FPDFImageObj_GetImageMetadata(obj, page, &metadata));
+  EXPECT_EQ(9, metadata.marked_content_id);
   EXPECT_EQ(126u, metadata.width);
   EXPECT_EQ(106u, metadata.height);
   EXPECT_NEAR(162.173752, metadata.horizontal_dpi, 0.001);
diff --git a/fpdfsdk/fpdfeditimg.cpp b/fpdfsdk/fpdfeditimg.cpp
index 4842e35c76..0d7ba56559 100644
--- a/fpdfsdk/fpdfeditimg.cpp
+++ b/fpdfsdk/fpdfeditimg.cpp
@@ -267,6 +267,8 @@ FPDFImageObj_GetImageMetadata(FPDF_PAGEOBJECT image_object,
   if (!pImg)
     return false;
 
+  metadata->marked_content_id = pObj->m_ContentMark.GetMarkedContentID();
+
   const int nPixelWidth = pImg->GetPixelWidth();
   const int nPixelHeight = pImg->GetPixelHeight();
   metadata->width = nPixelWidth;
diff --git a/public/fpdf_edit.h b/public/fpdf_edit.h
index 3bd246b0ad..685fbdd029 100644
--- a/public/fpdf_edit.h
+++ b/public/fpdf_edit.h
@@ -80,6 +80,9 @@ typedef struct FPDF_IMAGEOBJ_METADATA {
   unsigned int bits_per_pixel;
   // The image's colorspace. See above for the list of FPDF_COLORSPACE_*.
   int colorspace;
+  // The image's marked content ID. Useful for pairing with associated alt-text.
+  // A value of -1 indicates no ID.
+  int marked_content_id;
 } FPDF_IMAGEOBJ_METADATA;
 
 #ifdef __cplusplus
diff --git a/public/fpdf_structtree.h b/public/fpdf_structtree.h
index 688471f28d..2e86ae65b3 100644
--- a/public/fpdf_structtree.h
+++ b/public/fpdf_structtree.h
@@ -75,6 +75,16 @@ FPDF_StructElement_GetAltText(FPDF_STRUCTELEMENT struct_element,
                               void* buffer,
                               unsigned long buflen);
 
+// Function: FPDF_StructElement_GetMarkedContentID
+//          Get the marked content ID for a given element.
+// Parameters:
+//          struct_element -   Handle to the struct element.
+// Return value:
+//          The marked content ID of the element. If no ID exists, returns
+//          -1.
+FPDF_EXPORT int FPDF_CALLCONV
+FPDF_StructElement_GetMarkedContentID(FPDF_STRUCTELEMENT struct_element);
+
 // Function: FPDF_StructElement_GetType
 //           Get the type (/S) for a given element.
 // Parameters:
diff --git a/testing/resources/marked_content_id.pdf b/testing/resources/marked_content_id.pdf
new file mode 100644
index 0000000000..ddbf11f319
--- /dev/null
+++ b/testing/resources/marked_content_id.pdf