Add APIs for limited use of document tagged code.

BUG=pdfium:568

Review-Url: https://codereview.chromium.org/2519343002

5 files changed, 172 insertions, 4 deletions

diff --git a/fpdfsdk/fpdf_structtree.cpp b/fpdfsdk/fpdf_structtree.cpp
new file mode 100644
index 0000000000..541c46b378
--- /dev/null
+++ b/fpdfsdk/fpdf_structtree.cpp
@@ -0,0 +1,88 @@
+// Copyright 2016 PDFium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "public/fpdf_structtree.h"
+
+#include "core/fpdfapi/page/cpdf_page.h"
+#include "core/fpdfapi/parser/cpdf_dictionary.h"
+#include "core/fpdfdoc/fpdf_tagged.h"
+#include "fpdfsdk/fsdk_define.h"
+
+namespace {
+
+IPDF_StructTree* ToStructTree(FPDF_STRUCTTREE struct_tree) {
+  return reinterpret_cast<IPDF_StructTree*>(struct_tree);
+}
+
+IPDF_StructElement* ToStructTreeElement(FPDF_STRUCTELEMENT struct_element) {
+  return reinterpret_cast<IPDF_StructElement*>(struct_element);
+}
+
+}  // namespace
+
+DLLEXPORT FPDF_STRUCTTREE STDCALL FPDF_StructTree_GetForPage(FPDF_PAGE page) {
+  CPDF_Page* pPage = CPDFPageFromFPDFPage(page);
+  if (!pPage)
+    return nullptr;
+  return IPDF_StructTree::LoadPage(pPage->m_pDocument, pPage->m_pFormDict);
+}
+
+DLLEXPORT void STDCALL FPDF_StructTree_Close(FPDF_STRUCTTREE struct_tree) {
+  delete ToStructTree(struct_tree);
+}
+
+DLLEXPORT int STDCALL
+FPDF_StructTree_CountChildren(FPDF_STRUCTTREE struct_tree) {
+  IPDF_StructTree* tree = ToStructTree(struct_tree);
+  return tree ? tree->CountTopElements() : -1;
+}
+
+DLLEXPORT FPDF_STRUCTELEMENT STDCALL
+FPDF_StructTree_GetChildAtIndex(FPDF_STRUCTTREE struct_tree, int index) {
+  IPDF_StructTree* tree = ToStructTree(struct_tree);
+  if (!tree || index < 0 || index >= tree->CountTopElements())
+    return nullptr;
+  return tree->GetTopElement(index);
+}
+
+DLLEXPORT unsigned long STDCALL
+FPDF_StructElement_GetAltText(FPDF_STRUCTELEMENT struct_element,
+                              void* buffer,
+                              unsigned long buflen) {
+  IPDF_StructElement* elem = ToStructTreeElement(struct_element);
+  if (!elem)
+    return 0;
+
+  CPDF_Dictionary* dict = elem->GetDict();
+  if (!dict)
+    return 0;
+
+  CFX_WideString str = elem->GetDict()->GetUnicodeTextFor("Alt");
+  if (str.IsEmpty())
+    return 0;
+
+  CFX_ByteString encodedStr = str.UTF16LE_Encode();
+  const unsigned long len = encodedStr.GetLength();
+  if (buffer && len <= buflen)
+    FXSYS_memcpy(buffer, encodedStr.c_str(), len);
+  return len;
+}
+
+DLLEXPORT int STDCALL
+FPDF_StructElement_CountChildren(FPDF_STRUCTELEMENT struct_element) {
+  IPDF_StructElement* elem = ToStructTreeElement(struct_element);
+  return elem ? elem->CountKids() : -1;
+}
+
+DLLEXPORT FPDF_STRUCTELEMENT STDCALL
+FPDF_StructElement_GetChildAtIndex(FPDF_STRUCTELEMENT struct_element,
+                                   int index) {
+  IPDF_StructElement* elem = ToStructTreeElement(struct_element);
+  if (!elem || index < 0 || index >= elem->CountKids())
+    return nullptr;
+
+  CPDF_StructKid kid = elem->GetKid(index);
+  return kid.m_Type == CPDF_StructKid::Element ? kid.m_Element.m_pElement
+                                               : nullptr;
+}
diff --git a/fpdfsdk/fpdf_structtree_embeddertest.cpp b/fpdfsdk/fpdf_structtree_embeddertest.cpp
new file mode 100644
index 0000000000..58b3172057
--- /dev/null
+++ b/fpdfsdk/fpdf_structtree_embeddertest.cpp
@@ -0,0 +1,70 @@
+// Copyright 2016 PDFium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "core/fxcrt/fx_string.h"
+#include "public/fpdf_structtree.h"
+#include "testing/embedder_test.h"
+#include "testing/test_support.h"
+
+class FPDFStructTreeEmbeddertest : public EmbedderTest, public TestSaver {};
+
+TEST_F(FPDFStructTreeEmbeddertest, GetAltText) {
+  ASSERT_TRUE(OpenDocument("tagged_alt_text.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, -1);
+  EXPECT_EQ(nullptr, element);
+  element = FPDF_StructTree_GetChildAtIndex(struct_tree, 1);
+  EXPECT_EQ(nullptr, element);
+  element = FPDF_StructTree_GetChildAtIndex(struct_tree, 0);
+  ASSERT_NE(nullptr, element);
+  EXPECT_EQ(0U, FPDF_StructElement_GetAltText(element, nullptr, 0));
+
+  ASSERT_EQ(1, FPDF_StructElement_CountChildren(element));
+  FPDF_STRUCTELEMENT child_element =
+      FPDF_StructElement_GetChildAtIndex(element, -1);
+  EXPECT_EQ(nullptr, child_element);
+  child_element = FPDF_StructElement_GetChildAtIndex(element, 1);
+  EXPECT_EQ(nullptr, child_element);
+  child_element = FPDF_StructElement_GetChildAtIndex(element, 0);
+  ASSERT_NE(nullptr, child_element);
+  EXPECT_EQ(0U, FPDF_StructElement_GetAltText(child_element, nullptr, 0));
+
+  ASSERT_EQ(1, FPDF_StructElement_CountChildren(child_element));
+  FPDF_STRUCTELEMENT gchild_element =
+      FPDF_StructElement_GetChildAtIndex(child_element, -1);
+  EXPECT_EQ(nullptr, gchild_element);
+  gchild_element = FPDF_StructElement_GetChildAtIndex(child_element, 1);
+  EXPECT_EQ(nullptr, gchild_element);
+  gchild_element = FPDF_StructElement_GetChildAtIndex(child_element, 0);
+  ASSERT_NE(nullptr, gchild_element);
+  ASSERT_EQ(24U, FPDF_StructElement_GetAltText(gchild_element, nullptr, 0));
+
+  unsigned short buffer[12];
+  memset(buffer, 0, sizeof(buffer));
+  // Deliberately pass in a small buffer size to make sure |buffer| remains
+  // untouched.
+  ASSERT_EQ(24U, FPDF_StructElement_GetAltText(gchild_element, buffer, 1));
+  for (size_t i = 0; i < FX_ArraySize(buffer); ++i)
+    EXPECT_EQ(0U, buffer[i]);
+
+  ASSERT_EQ(24U, FPDF_StructElement_GetAltText(gchild_element, buffer,
+                                               sizeof(buffer)));
+  const FX_WCHAR kExpected[] = L"Black Image";
+  EXPECT_EQ(CFX_WideString(kExpected),
+            CFX_WideString::FromUTF16LE(buffer, FXSYS_len(kExpected)));
+
+  ASSERT_EQ(1, FPDF_StructElement_CountChildren(gchild_element));
+  FPDF_STRUCTELEMENT ggchild_element =
+      FPDF_StructElement_GetChildAtIndex(gchild_element, 0);
+  EXPECT_EQ(nullptr, ggchild_element);
+
+  FPDF_StructTree_Close(struct_tree);
+  FPDF_ClosePage(page);
+}
diff --git a/fpdfsdk/fpdfdoc.cpp b/fpdfsdk/fpdfdoc.cpp
index 254be3f883..2dcf606a7c 100644
--- a/fpdfsdk/fpdfdoc.cpp
+++ b/fpdfsdk/fpdfdoc.cpp
@@ -64,7 +64,7 @@ unsigned long Utf16EncodeMaybeCopyAndReturnLength(const CFX_WideString& text,
                                                   unsigned long buflen) {
   CFX_ByteString encodedText = text.UTF16LE_Encode();
   unsigned long len = encodedText.GetLength();
-  if (buffer && buflen >= len)
+  if (buffer && len <= buflen)
     FXSYS_memcpy(buffer, encodedText.c_str(), len);
   return len;
 }
@@ -186,7 +186,7 @@ DLLEXPORT unsigned long STDCALL FPDFAction_GetFilePath(FPDF_ACTION pDict,
   CPDF_Action action(ToDictionary(static_cast<CPDF_Object*>(pDict)));
   CFX_ByteString path = action.GetFilePath().UTF8Encode();
   unsigned long len = path.GetLength() + 1;
-  if (buffer && buflen >= len)
+  if (buffer && len <= buflen)
     FXSYS_memcpy(buffer, path.c_str(), len);
   return len;
 }
@@ -203,7 +203,7 @@ DLLEXPORT unsigned long STDCALL FPDFAction_GetURIPath(FPDF_DOCUMENT document,
   CPDF_Action action(ToDictionary(static_cast<CPDF_Object*>(pDict)));
   CFX_ByteString path = action.GetURI(pDoc);
   unsigned long len = path.GetLength() + 1;
-  if (buffer && buflen >= len)
+  if (buffer && len <= buflen)
     FXSYS_memcpy(buffer, path.c_str(), len);
   return len;
 }
diff --git a/fpdfsdk/fpdfview.cpp b/fpdfsdk/fpdfview.cpp
index 959bf14390..3f5115afd6 100644
--- a/fpdfsdk/fpdfview.cpp
+++ b/fpdfsdk/fpdfview.cpp
@@ -1110,7 +1110,7 @@ DLLEXPORT FPDF_DEST STDCALL FPDF_GetNamedDest(FPDF_DOCUMENT document,
   int len = utf16Name.GetLength();
   if (!buffer) {
     *buflen = len;
-  } else if (*buflen >= len) {
+  } else if (len <= *buflen) {
     memcpy(buffer, utf16Name.c_str(), len);
     *buflen = len;
   } else {
diff --git a/fpdfsdk/fpdfview_c_api_test.c b/fpdfsdk/fpdfview_c_api_test.c
index 5e6c36f2b1..ed9a3fafe2 100644
--- a/fpdfsdk/fpdfview_c_api_test.c
+++ b/fpdfsdk/fpdfview_c_api_test.c
@@ -20,6 +20,7 @@
 #include "public/fpdf_progressive.h"
 #include "public/fpdf_save.h"
 #include "public/fpdf_searchex.h"
+#include "public/fpdf_structtree.h"
 #include "public/fpdf_sysfontinfo.h"
 #include "public/fpdf_text.h"
 #include "public/fpdf_transformpage.h"
@@ -154,6 +155,15 @@ int CheckPDFiumCApi() {
     // fpdf_searchex.h
     CHK(FPDFText_GetCharIndexFromTextIndex);
 
+    // fpdf_structtree.h
+    CHK(FPDF_StructTree_GetForPage);
+    CHK(FPDF_StructTree_Close);
+    CHK(FPDF_StructTree_CountChildren);
+    CHK(FPDF_StructTree_GetChildAtIndex);
+    CHK(FPDF_StructElement_GetAltText);
+    CHK(FPDF_StructElement_CountChildren);
+    CHK(FPDF_StructElement_GetChildAtIndex);
+
     // fpdf_sysfontinfo.h
     CHK(FPDF_GetDefaultTTFMap);
     CHK(FPDF_AddInstalledFont);