Open FPDF_PAGEOBJECTMARK API.

Initially it contains:
- FPDFPageObj_CountMarks
- FPDFPageObj_GetMark
- FPDFPageObjMark_GetName

This allows reading the content marks from a page object.

Bug: pdfium:1037
Change-Id: I3d7554b71f938778890fdb44088e0d45dd2718d9
Reviewed-on: https://pdfium-review.googlesource.com/28710
Commit-Queue: Henrique Nakashima <hnakashima@chromium.org>
Reviewed-by: Lei Zhang <thestig@chromium.org>

5 files changed, 141 insertions, 4 deletions

diff --git a/fpdfsdk/fpdfedit_embeddertest.cpp b/fpdfsdk/fpdfedit_embeddertest.cpp
index ec403c74b3..65e790cc32 100644
--- a/fpdfsdk/fpdfedit_embeddertest.cpp
+++ b/fpdfsdk/fpdfedit_embeddertest.cpp
@@ -387,7 +387,8 @@ TEST_F(FPDFEditEmbeddertest, RemovePageObject) {
   FPDF_PAGE page = LoadPage(0);
   ASSERT_TRUE(page);
 
-// Show how the original file looks like.
+  // Show what the original file looks like.
+  {
 #if _FX_PLATFORM_ == _FX_PLATFORM_APPLE_
   const char kOriginalMD5[] = "b90475ca64d1348c3bf5e2b77ad9187a";
 #elif _FX_PLATFORM_ == _FX_PLATFORM_WINDOWS_
@@ -395,7 +396,6 @@ TEST_F(FPDFEditEmbeddertest, RemovePageObject) {
 #else
   const char kOriginalMD5[] = "2baa4c0e1758deba1b9c908e1fbd04ed";
 #endif
-  {
     std::unique_ptr<void, FPDFBitmapDeleter> page_bitmap =
         RenderPageWithFlags(page, nullptr, 0);
     CompareBitmap(page_bitmap.get(), 200, 200, kOriginalMD5);
@@ -407,7 +407,8 @@ TEST_F(FPDFEditEmbeddertest, RemovePageObject) {
   ASSERT_TRUE(page_object);
   EXPECT_TRUE(FPDFPage_RemoveObject(page, page_object));
 
-// Verify the "Hello, world!" text is gone.
+  // Verify the "Hello, world!" text is gone.
+  {
 #if _FX_PLATFORM_ == _FX_PLATFORM_APPLE_
   const char kRemovedMD5[] = "af760c4702467cb1492a57fb8215efaa";
 #elif _FX_PLATFORM_ == _FX_PLATFORM_WINDOWS_
@@ -415,7 +416,6 @@ TEST_F(FPDFEditEmbeddertest, RemovePageObject) {
 #else
   const char kRemovedMD5[] = "b76df015fe88009c3c342395df96abf1";
 #endif
-  {
     std::unique_ptr<void, FPDFBitmapDeleter> page_bitmap =
         RenderPageWithFlags(page, nullptr, 0);
     CompareBitmap(page_bitmap.get(), 200, 200, kRemovedMD5);
@@ -426,6 +426,89 @@ TEST_F(FPDFEditEmbeddertest, RemovePageObject) {
   FPDFPageObj_Destroy(page_object);
 }
 
+TEST_F(FPDFEditEmbeddertest, RemoveMarkedObjectsPrime) {
+  // Load document with some text.
+  EXPECT_TRUE(OpenDocument("text_in_page_marked.pdf"));
+  FPDF_PAGE page = LoadPage(0);
+  ASSERT_TRUE(page);
+
+  // Show what the original file looks like.
+  {
+#if _FX_PLATFORM_ == _FX_PLATFORM_APPLE_
+    const char kOriginalMD5[] = "5a5eb63cb21cc15084fea1f14284b8df";
+#elif _FX_PLATFORM_ == _FX_PLATFORM_WINDOWS_
+    const char kOriginalMD5[] = "587c507a40f613f9c530b2ce2d58d655";
+#else
+    const char kOriginalMD5[] = "2edc6e70d54889aa0c0b7bdf3e168f86";
+#endif
+    std::unique_ptr<void, FPDFBitmapDeleter> page_bitmap =
+        RenderPageWithFlags(page, nullptr, 0);
+    CompareBitmap(page_bitmap.get(), 200, 200, kOriginalMD5);
+  }
+
+  // Iterate over all objects, counting the number of times each content mark
+  // name appears.
+  int object_count = FPDFPage_CountObjects(page);
+  ASSERT_EQ(19, object_count);
+
+  unsigned long prime_count = 0;
+  unsigned long square_count = 0;
+  unsigned long greater_than_ten_count = 0;
+  std::vector<FPDF_PAGEOBJECT> primes;
+  for (int i = 0; i < object_count; ++i) {
+    FPDF_PAGEOBJECT page_object = FPDFPage_GetObject(page, i);
+
+    int mark_count = FPDFPageObj_CountMarks(page_object);
+    for (int j = 0; j < mark_count; ++j) {
+      FPDF_PAGEOBJECTMARK mark = FPDFPageObj_GetMark(page_object, j);
+
+      char buffer[256];
+      ASSERT_GT(FPDFPageObjMark_GetName(mark, buffer, 256), 0u);
+      std::wstring name =
+          GetPlatformWString(reinterpret_cast<unsigned short*>(buffer));
+      if (name == L"Prime") {
+        prime_count++;
+        primes.push_back(page_object);
+      } else if (name == L"Square") {
+        square_count++;
+      } else if (name == L"GreaterThanTen") {
+        greater_than_ten_count++;
+      } else {
+        FAIL();
+      }
+    }
+  }
+
+  // Expect certain number of tagged objects. The test file contains strings
+  // from 1 to 19.
+  EXPECT_EQ(8u, prime_count);
+  EXPECT_EQ(4u, square_count);
+  EXPECT_EQ(9u, greater_than_ten_count);
+
+  // Remove all objects marked with "Prime".
+  for (FPDF_PAGEOBJECT page_object : primes) {
+    EXPECT_TRUE(FPDFPage_RemoveObject(page, page_object));
+    FPDFPageObj_Destroy(page_object);
+  }
+
+  EXPECT_EQ(11, FPDFPage_CountObjects(page));
+
+  {
+#if _FX_PLATFORM_ == _FX_PLATFORM_APPLE_
+    const char kNonPrimesMD5[] = "57e76dc7375d896704f0fd6d6d1b9e65";
+#elif _FX_PLATFORM_ == _FX_PLATFORM_WINDOWS_
+    const char kNonPrimesMD5[] = "4d906b57fba36c70c600cf50d60f508c";
+#else
+    const char kNonPrimesMD5[] = "33d9c45bec41ead92a295e252f6b7922";
+#endif
+    std::unique_ptr<void, FPDFBitmapDeleter> page_bitmap =
+        RenderPageWithFlags(page, nullptr, 0);
+    CompareBitmap(page_bitmap.get(), 200, 200, kNonPrimesMD5);
+  }
+
+  UnloadPage(page);
+}
+
 TEST_F(FPDFEditEmbeddertest, AddAndRemovePaths) {
   // Start with a blank page.
   FPDF_PAGE page = FPDFPage_New(CreateNewDocument(), 0, 612, 792);
diff --git a/fpdfsdk/fpdfeditpage.cpp b/fpdfsdk/fpdfeditpage.cpp
index 902aab4dd1..d0724d5359 100644
--- a/fpdfsdk/fpdfeditpage.cpp
+++ b/fpdfsdk/fpdfeditpage.cpp
@@ -222,6 +222,47 @@ FPDF_EXPORT void FPDF_CALLCONV FPDFPageObj_Destroy(FPDF_PAGEOBJECT page_obj) {
   delete CPDFPageObjectFromFPDFPageObject(page_obj);
 }
 
+FPDF_EXPORT int FPDF_CALLCONV
+FPDFPageObj_CountMarks(FPDF_PAGEOBJECT page_object) {
+  if (!page_object)
+    return -1;
+
+  const auto& mark =
+      CPDFPageObjectFromFPDFPageObject(page_object)->m_ContentMark;
+  return mark.HasRef() ? mark.CountItems() : 0;
+}
+
+FPDF_EXPORT FPDF_PAGEOBJECTMARK FPDF_CALLCONV
+FPDFPageObj_GetMark(FPDF_PAGEOBJECT page_object, unsigned long index) {
+  if (!page_object)
+    return nullptr;
+
+  const auto& mark =
+      CPDFPageObjectFromFPDFPageObject(page_object)->m_ContentMark;
+  if (!mark.HasRef())
+    return nullptr;
+
+  if (index >= mark.CountItems())
+    return nullptr;
+
+  return static_cast<FPDF_PAGEOBJECTMARK>(&mark.GetItem(index));
+}
+
+FPDF_EXPORT unsigned long FPDF_CALLCONV
+FPDFPageObjMark_GetName(FPDF_PAGEOBJECTMARK mark,
+                        void* buffer,
+                        unsigned long buflen) {
+  if (!mark)
+    return 0;
+
+  const CPDF_ContentMarkItem* pMarkItem =
+      CPDFContentMarkItemFromFPDFPageObjectMark(mark);
+
+  return Utf16EncodeMaybeCopyAndReturnLength(
+      WideString::FromUTF8(pMarkItem->GetName().AsStringView()), buffer,
+      buflen);
+}
+
 FPDF_EXPORT FPDF_BOOL FPDF_CALLCONV
 FPDFPageObj_HasTransparency(FPDF_PAGEOBJECT pageObject) {
   if (!pageObject)
diff --git a/fpdfsdk/fpdfview.cpp b/fpdfsdk/fpdfview.cpp
index e890aa0090..a9078102e3 100644
--- a/fpdfsdk/fpdfview.cpp
+++ b/fpdfsdk/fpdfview.cpp
@@ -12,6 +12,7 @@
 
 #include "core/fpdfapi/cpdf_modulemgr.h"
 #include "core/fpdfapi/cpdf_pagerendercontext.h"
+#include "core/fpdfapi/page/cpdf_contentmarkitem.h"
 #include "core/fpdfapi/page/cpdf_image.h"
 #include "core/fpdfapi/page/cpdf_imageobject.h"
 #include "core/fpdfapi/page/cpdf_page.h"
@@ -345,6 +346,11 @@ CPDF_PageObject* CPDFPageObjectFromFPDFPageObject(FPDF_PAGEOBJECT page_object) {
   return static_cast<CPDF_PageObject*>(page_object);
 }
 
+const CPDF_ContentMarkItem* CPDFContentMarkItemFromFPDFPageObjectMark(
+    FPDF_PAGEOBJECTMARK mark) {
+  return static_cast<const CPDF_ContentMarkItem*>(mark);
+}
+
 CPDF_Object* CPDFObjectFromFPDFAttachment(FPDF_ATTACHMENT attachment) {
   return static_cast<CPDF_Object*>(attachment);
 }
diff --git a/fpdfsdk/fpdfview_c_api_test.c b/fpdfsdk/fpdfview_c_api_test.c
index 201fd90669..a0bcc13ff3 100644
--- a/fpdfsdk/fpdfview_c_api_test.c
+++ b/fpdfsdk/fpdfview_c_api_test.c
@@ -144,6 +144,9 @@ int CheckPDFiumCApi() {
     CHK(FPDFPageObj_Transform);
     CHK(FPDFPage_TransformAnnots);
     CHK(FPDFPageObj_NewImageObj);
+    CHK(FPDFPageObj_CountMarks);
+    CHK(FPDFPageObj_GetMark);
+    CHK(FPDFPageObjMark_GetName);
     CHK(FPDFImageObj_LoadJpegFile);
     CHK(FPDFImageObj_LoadJpegFileInline);
     CHK(FPDFImageObj_SetMatrix);
diff --git a/fpdfsdk/fsdk_define.h b/fpdfsdk/fsdk_define.h
index acd8437169..94497bb237 100644
--- a/fpdfsdk/fsdk_define.h
+++ b/fpdfsdk/fsdk_define.h
@@ -23,6 +23,7 @@
 #endif
 
 class CPDF_Annot;
+class CPDF_ContentMarkItem;
 class CPDF_Page;
 class CPDF_PageObject;
 class CPDF_PageRenderContext;
@@ -70,6 +71,9 @@ CPDF_PathObject* CPDFPathObjectFromFPDFPageObject(FPDF_PAGEOBJECT page_object);
 
 CPDF_PageObject* CPDFPageObjectFromFPDFPageObject(FPDF_PAGEOBJECT page_object);
 
+const CPDF_ContentMarkItem* CPDFContentMarkItemFromFPDFPageObjectMark(
+    FPDF_PAGEOBJECTMARK mark);
+
 CPDF_Object* CPDFObjectFromFPDFAttachment(FPDF_ATTACHMENT attachment);
 
 ByteString CFXByteStringFromFPDFWideString(FPDF_WIDESTRING wide_string);