APIs and tests for extracting bitmaps from image objects

Added FPDFImageObj_GetBitmap() that returns the bitmap of an image
object, and a FPDFBitmap_GetFormat() that returns the format of a
bitmap.
    * Fixed a small bitmap conversion bug in cfx_dibsource.cpp.
    * Enabled EmbedderTest::CompareBitmap() to support different formats
      of bitmaps.
    * Added an embedder test and a test PDF file with images of many
      different formats.

Bug=pdfium:677

Change-Id: I6a72f9d969cf5f3577db9400ca33197c213622ed
Reviewed-on: https://pdfium-review.googlesource.com/9690
Reviewed-by: Lei Zhang <thestig@chromium.org>
Commit-Queue: Jane Liu <janeliulwq@google.com>

4 files changed, 103 insertions, 0 deletions

diff --git a/fpdfsdk/fpdfedit_embeddertest.cpp b/fpdfsdk/fpdfedit_embeddertest.cpp
index cc3e50b362..dcaeb945d7 100644
--- a/fpdfsdk/fpdfedit_embeddertest.cpp
+++ b/fpdfsdk/fpdfedit_embeddertest.cpp
@@ -925,3 +925,57 @@ TEST_F(FPDFEditEmbeddertest, SaveAndRender) {
   }
   TestAndCloseSaved(612, 792, md5);
 }
+
+TEST_F(FPDFEditEmbeddertest, ExtractImageBitmap) {
+  ASSERT_TRUE(OpenDocument("embedded_images.pdf"));
+  FPDF_PAGE page = LoadPage(0);
+  ASSERT_TRUE(page);
+  ASSERT_EQ(39, FPDFPage_CountObject(page));
+
+  FPDF_PAGEOBJECT obj = FPDFPage_GetObject(page, 32);
+  EXPECT_NE(FPDF_PAGEOBJ_IMAGE, FPDFPageObj_GetType(obj));
+  EXPECT_FALSE(FPDFImageObj_GetBitmap(obj));
+
+  obj = FPDFPage_GetObject(page, 33);
+  ASSERT_EQ(FPDF_PAGEOBJ_IMAGE, FPDFPageObj_GetType(obj));
+  FPDF_BITMAP bitmap = FPDFImageObj_GetBitmap(obj);
+  EXPECT_EQ(FPDFBitmap_BGR, FPDFBitmap_GetFormat(bitmap));
+  CompareBitmap(bitmap, 109, 88, "d65e98d968d196abf13f78aec655ffae");
+  FPDFBitmap_Destroy(bitmap);
+
+  obj = FPDFPage_GetObject(page, 34);
+  ASSERT_EQ(FPDF_PAGEOBJ_IMAGE, FPDFPageObj_GetType(obj));
+  bitmap = FPDFImageObj_GetBitmap(obj);
+  EXPECT_EQ(FPDFBitmap_BGR, FPDFBitmap_GetFormat(bitmap));
+  CompareBitmap(bitmap, 103, 75, "1287711c84dbef767c435d11697661d6");
+  FPDFBitmap_Destroy(bitmap);
+
+  obj = FPDFPage_GetObject(page, 35);
+  ASSERT_EQ(FPDF_PAGEOBJ_IMAGE, FPDFPageObj_GetType(obj));
+  bitmap = FPDFImageObj_GetBitmap(obj);
+  EXPECT_EQ(FPDFBitmap_Gray, FPDFBitmap_GetFormat(bitmap));
+  CompareBitmap(bitmap, 92, 68, "9c6d76cb1e37ef8514f9455d759391f3");
+  FPDFBitmap_Destroy(bitmap);
+
+  obj = FPDFPage_GetObject(page, 36);
+  ASSERT_EQ(FPDF_PAGEOBJ_IMAGE, FPDFPageObj_GetType(obj));
+  bitmap = FPDFImageObj_GetBitmap(obj);
+  EXPECT_EQ(FPDFBitmap_BGR, FPDFBitmap_GetFormat(bitmap));
+  CompareBitmap(bitmap, 79, 60, "15cb6a49a2e354ed0e9f45dd34e3da1a");
+  FPDFBitmap_Destroy(bitmap);
+
+  obj = FPDFPage_GetObject(page, 37);
+  ASSERT_EQ(FPDF_PAGEOBJ_IMAGE, FPDFPageObj_GetType(obj));
+  bitmap = FPDFImageObj_GetBitmap(obj);
+  EXPECT_EQ(FPDFBitmap_BGR, FPDFBitmap_GetFormat(bitmap));
+  CompareBitmap(bitmap, 126, 106, "be5a64ba7890d2657522af6524118534");
+  FPDFBitmap_Destroy(bitmap);
+
+  obj = FPDFPage_GetObject(page, 38);
+  ASSERT_EQ(FPDF_PAGEOBJ_IMAGE, FPDFPageObj_GetType(obj));
+  bitmap = FPDFImageObj_GetBitmap(obj);
+  EXPECT_EQ(FPDFBitmap_BGR, FPDFBitmap_GetFormat(bitmap));
+  CompareBitmap(bitmap, 194, 119, "f9e24207ee1bc0db6c543d33a5f12ec5");
+  FPDFBitmap_Destroy(bitmap);
+  UnloadPage(page);
+}
diff --git a/fpdfsdk/fpdfeditimg.cpp b/fpdfsdk/fpdfeditimg.cpp
index fdc98e06e9..bfd12b2441 100644
--- a/fpdfsdk/fpdfeditimg.cpp
+++ b/fpdfsdk/fpdfeditimg.cpp
@@ -110,3 +110,30 @@ DLLEXPORT FPDF_BOOL STDCALL FPDFImageObj_SetBitmap(FPDF_PAGE* pages,
   pImgObj->SetDirty(true);
   return true;
 }
+
+DLLEXPORT FPDF_BITMAP STDCALL
+FPDFImageObj_GetBitmap(FPDF_PAGEOBJECT image_object) {
+  CPDF_PageObject* pObj = CPDFPageObjectFromFPDFPageObject(image_object);
+  if (!pObj || !pObj->IsImage())
+    return nullptr;
+
+  CFX_RetainPtr<CPDF_Image> pImg = pObj->AsImage()->GetImage();
+  if (!pImg)
+    return nullptr;
+
+  CFX_RetainPtr<CFX_DIBSource> pSource = pImg->LoadDIBSource();
+  if (!pSource)
+    return nullptr;
+
+  CFX_RetainPtr<CFX_DIBitmap> pBitmap;
+  // If the source image has a representation of 1 bit per pixel, then convert
+  // it to a grayscale bitmap having 1 byte per pixel, since bitmaps have no
+  // concept of bits. Otherwise, convert the source image to a bitmap directly,
+  // retaining its color representation.
+  if (pSource->GetBPP() == 1)
+    pBitmap = pSource->CloneConvert(FXDIB_8bppRgb);
+  else
+    pBitmap = pSource->Clone(nullptr);
+
+  return pBitmap.Leak();
+}
diff --git a/fpdfsdk/fpdfview.cpp b/fpdfsdk/fpdfview.cpp
index 355fcb3f8f..5aa80139ae 100644
--- a/fpdfsdk/fpdfview.cpp
+++ b/fpdfsdk/fpdfview.cpp
@@ -1126,6 +1126,26 @@ DLLEXPORT FPDF_BITMAP STDCALL FPDFBitmap_CreateEx(int width,
   return pBitmap.Leak();
 }
 
+DLLEXPORT int STDCALL FPDFBitmap_GetFormat(FPDF_BITMAP bitmap) {
+  if (!bitmap)
+    return FPDFBitmap_Unknown;
+
+  FXDIB_Format format = CFXBitmapFromFPDFBitmap(bitmap)->GetFormat();
+  switch (format) {
+    case FXDIB_8bppRgb:
+    case FXDIB_8bppMask:
+      return FPDFBitmap_Gray;
+    case FXDIB_Rgb:
+      return FPDFBitmap_BGR;
+    case FXDIB_Rgb32:
+      return FPDFBitmap_BGRx;
+    case FXDIB_Argb:
+      return FPDFBitmap_BGRA;
+    default:
+      return FPDFBitmap_Unknown;
+  }
+}
+
 DLLEXPORT void STDCALL FPDFBitmap_FillRect(FPDF_BITMAP bitmap,
                                            int left,
                                            int top,
diff --git a/fpdfsdk/fpdfview_c_api_test.c b/fpdfsdk/fpdfview_c_api_test.c
index 7d42be8792..e47f4d172c 100644
--- a/fpdfsdk/fpdfview_c_api_test.c
+++ b/fpdfsdk/fpdfview_c_api_test.c
@@ -132,6 +132,7 @@ int CheckPDFiumCApi() {
     CHK(FPDFImageObj_LoadJpegFileInline);
     CHK(FPDFImageObj_SetMatrix);
     CHK(FPDFImageObj_SetBitmap);
+    CHK(FPDFImageObj_GetBitmap);
     CHK(FPDFPageObj_CreateNewPath);
     CHK(FPDFPageObj_CreateNewRect);
     CHK(FPDFPath_SetStrokeColor);
@@ -302,6 +303,7 @@ int CheckPDFiumCApi() {
     CHK(FPDF_PageToDevice);
     CHK(FPDFBitmap_Create);
     CHK(FPDFBitmap_CreateEx);
+    CHK(FPDFBitmap_GetFormat);
     CHK(FPDFBitmap_FillRect);
     CHK(FPDFBitmap_GetBuffer);
     CHK(FPDFBitmap_GetWidth);