Add API for getting page labels.

BUG=pdfium:479

Review-Url: https://codereview.chromium.org/2521843003

7 files changed, 149 insertions, 52 deletions

diff --git a/core/fpdfdoc/cpdf_pagelabel.cpp b/core/fpdfdoc/cpdf_pagelabel.cpp
index 64075ef55a..2a79d77ca7 100644
--- a/core/fpdfdoc/cpdf_pagelabel.cpp
+++ b/core/fpdfdoc/cpdf_pagelabel.cpp
@@ -75,16 +75,21 @@ CFX_WideString GetLabelNumPortion(int num, const CFX_ByteString& bsStyle) {
 CPDF_PageLabel::CPDF_PageLabel(CPDF_Document* pDocument)
     : m_pDocument(pDocument) {}
 
-CFX_WideString CPDF_PageLabel::GetLabel(int nPage) const {
-  CFX_WideString wsLabel;
+bool CPDF_PageLabel::GetLabel(int nPage, CFX_WideString* wsLabel) const {
   if (!m_pDocument)
-    return wsLabel;
+    return false;
+
+  if (nPage < 0 || nPage >= m_pDocument->GetPageCount())
+    return false;
 
   CPDF_Dictionary* pPDFRoot = m_pDocument->GetRoot();
   if (!pPDFRoot)
-    return wsLabel;
+    return false;
 
   CPDF_Dictionary* pLabels = pPDFRoot->GetDictFor("PageLabels");
+  if (!pLabels)
+    return false;
+
   CPDF_NumberTree numberTree(pLabels);
   CPDF_Object* pValue = nullptr;
   int n = nPage;
@@ -99,18 +104,18 @@ CFX_WideString CPDF_PageLabel::GetLabel(int nPage) const {
     pValue = pValue->GetDirect();
     if (CPDF_Dictionary* pLabel = pValue->AsDictionary()) {
       if (pLabel->KeyExist("P"))
-        wsLabel += pLabel->GetUnicodeTextFor("P");
+        *wsLabel += pLabel->GetUnicodeTextFor("P");
 
       CFX_ByteString bsNumberingStyle = pLabel->GetStringFor("S", "");
       int nLabelNum = nPage - n + pLabel->GetIntegerFor("St", 1);
       CFX_WideString wsNumPortion =
           GetLabelNumPortion(nLabelNum, bsNumberingStyle);
-      wsLabel += wsNumPortion;
-      return wsLabel;
+      *wsLabel += wsNumPortion;
+      return true;
     }
   }
-  wsLabel.Format(L"%d", nPage + 1);
-  return wsLabel;
+  wsLabel->Format(L"%d", nPage + 1);
+  return true;
 }
 
 int32_t CPDF_PageLabel::GetPageByLabel(const CFX_ByteStringC& bsLabel) const {
@@ -123,7 +128,10 @@ int32_t CPDF_PageLabel::GetPageByLabel(const CFX_ByteStringC& bsLabel) const {
 
   int nPages = m_pDocument->GetPageCount();
   for (int i = 0; i < nPages; i++) {
-    if (PDF_EncodeText(GetLabel(i)).Compare(bsLabel))
+    CFX_WideString str;
+    if (!GetLabel(i, &str))
+      continue;
+    if (PDF_EncodeText(str).Compare(bsLabel))
       return i;
   }
 
diff --git a/core/fpdfdoc/cpdf_pagelabel.h b/core/fpdfdoc/cpdf_pagelabel.h
index 6a0664bfca..0f91f614d9 100644
--- a/core/fpdfdoc/cpdf_pagelabel.h
+++ b/core/fpdfdoc/cpdf_pagelabel.h
@@ -15,7 +15,7 @@ class CPDF_PageLabel {
  public:
   explicit CPDF_PageLabel(CPDF_Document* pDocument);
 
-  CFX_WideString GetLabel(int nPage) const;
+  bool GetLabel(int nPage, CFX_WideString* wsLabel) const;
   int32_t GetPageByLabel(const CFX_ByteStringC& bsLabel) const;
   int32_t GetPageByLabel(const CFX_WideStringC& wsLabel) const;
 
diff --git a/fpdfsdk/fpdfdoc.cpp b/fpdfsdk/fpdfdoc.cpp
index 01d91242f4..254be3f883 100644
--- a/fpdfsdk/fpdfdoc.cpp
+++ b/fpdfsdk/fpdfdoc.cpp
@@ -15,6 +15,7 @@
 #include "core/fpdfdoc/cpdf_bookmark.h"
 #include "core/fpdfdoc/cpdf_bookmarktree.h"
 #include "core/fpdfdoc/cpdf_dest.h"
+#include "core/fpdfdoc/cpdf_pagelabel.h"
 #include "fpdfsdk/fsdk_define.h"
 #include "third_party/base/stl_util.h"
 
@@ -58,6 +59,16 @@ CPDF_LinkList* GetLinkList(CPDF_Page* page) {
   return pHolder->get();
 }
 
+unsigned long Utf16EncodeMaybeCopyAndReturnLength(const CFX_WideString& text,
+                                                  void* buffer,
+                                                  unsigned long buflen) {
+  CFX_ByteString encodedText = text.UTF16LE_Encode();
+  unsigned long len = encodedText.GetLength();
+  if (buffer && buflen >= len)
+    FXSYS_memcpy(buffer, encodedText.c_str(), len);
+  return len;
+}
+
 }  // namespace
 
 DLLEXPORT FPDF_BOOKMARK STDCALL
@@ -91,12 +102,7 @@ DLLEXPORT unsigned long STDCALL FPDFBookmark_GetTitle(FPDF_BOOKMARK pDict,
     return 0;
   CPDF_Bookmark bookmark(ToDictionary(static_cast<CPDF_Object*>(pDict)));
   CFX_WideString title = bookmark.GetTitle();
-  CFX_ByteString encodedTitle = title.UTF16LE_Encode();
-  unsigned long len = encodedTitle.GetLength();
-  if (buffer && buflen >= len) {
-    FXSYS_memcpy(buffer, encodedTitle.c_str(), len);
-  }
-  return len;
+  return Utf16EncodeMaybeCopyAndReturnLength(title, buffer, buflen);
 }
 
 DLLEXPORT FPDF_BOOKMARK STDCALL FPDFBookmark_Find(FPDF_DOCUMENT document,
@@ -354,42 +360,53 @@ DLLEXPORT FPDF_BOOL STDCALL FPDFLink_GetQuadPoints(FPDF_LINK linkAnnot,
   CPDF_Dictionary* pAnnotDict =
       ToDictionary(static_cast<CPDF_Object*>(linkAnnot));
   CPDF_Array* pArray = pAnnotDict->GetArrayFor("QuadPoints");
-  if (pArray) {
-    if (quadIndex < 0 ||
-        static_cast<size_t>(quadIndex) >= pArray->GetCount() / 8 ||
-        (static_cast<size_t>(quadIndex * 8 + 7) >= pArray->GetCount()))
-      return false;
-    quadPoints->x1 = pArray->GetNumberAt(quadIndex * 8);
-    quadPoints->y1 = pArray->GetNumberAt(quadIndex * 8 + 1);
-    quadPoints->x2 = pArray->GetNumberAt(quadIndex * 8 + 2);
-    quadPoints->y2 = pArray->GetNumberAt(quadIndex * 8 + 3);
-    quadPoints->x3 = pArray->GetNumberAt(quadIndex * 8 + 4);
-    quadPoints->y3 = pArray->GetNumberAt(quadIndex * 8 + 5);
-    quadPoints->x4 = pArray->GetNumberAt(quadIndex * 8 + 6);
-    quadPoints->y4 = pArray->GetNumberAt(quadIndex * 8 + 7);
-    return true;
+  if (!pArray)
+    return false;
+
+  if (quadIndex < 0 ||
+      static_cast<size_t>(quadIndex) >= pArray->GetCount() / 8 ||
+      (static_cast<size_t>(quadIndex * 8 + 7) >= pArray->GetCount())) {
+    return false;
   }
-  return false;
+
+  quadPoints->x1 = pArray->GetNumberAt(quadIndex * 8);
+  quadPoints->y1 = pArray->GetNumberAt(quadIndex * 8 + 1);
+  quadPoints->x2 = pArray->GetNumberAt(quadIndex * 8 + 2);
+  quadPoints->y2 = pArray->GetNumberAt(quadIndex * 8 + 3);
+  quadPoints->x3 = pArray->GetNumberAt(quadIndex * 8 + 4);
+  quadPoints->y3 = pArray->GetNumberAt(quadIndex * 8 + 5);
+  quadPoints->x4 = pArray->GetNumberAt(quadIndex * 8 + 6);
+  quadPoints->y4 = pArray->GetNumberAt(quadIndex * 8 + 7);
+  return true;
 }
 
-DLLEXPORT unsigned long STDCALL FPDF_GetMetaText(FPDF_DOCUMENT doc,
+DLLEXPORT unsigned long STDCALL FPDF_GetMetaText(FPDF_DOCUMENT document,
                                                  FPDF_BYTESTRING tag,
                                                  void* buffer,
                                                  unsigned long buflen) {
   if (!tag)
     return 0;
-  CPDF_Document* pDoc = CPDFDocumentFromFPDFDocument(doc);
+  CPDF_Document* pDoc = CPDFDocumentFromFPDFDocument(document);
   if (!pDoc)
     return 0;
   CPDF_Dictionary* pInfo = pDoc->GetInfo();
   if (!pInfo)
     return 0;
   CFX_WideString text = pInfo->GetUnicodeTextFor(tag);
-  // Use UTF-16LE encoding
-  CFX_ByteString encodedText = text.UTF16LE_Encode();
-  unsigned long len = encodedText.GetLength();
-  if (buffer && buflen >= len) {
-    FXSYS_memcpy(buffer, encodedText.c_str(), len);
-  }
-  return len;
+  return Utf16EncodeMaybeCopyAndReturnLength(text, buffer, buflen);
+}
+
+DLLEXPORT unsigned long STDCALL FPDF_GetPagelLabel(FPDF_DOCUMENT document,
+                                                   int page_index,
+                                                   void* buffer,
+                                                   unsigned long buflen) {
+  if (page_index < 0)
+    return 0;
+
+  // CPDF_PageLabel can deal with NULL |document|.
+  CPDF_PageLabel label(CPDFDocumentFromFPDFDocument(document));
+  CFX_WideString str;
+  if (!label.GetLabel(page_index, &str))
+    return 0;
+  return Utf16EncodeMaybeCopyAndReturnLength(str, buffer, buflen);
 }
diff --git a/fpdfsdk/fpdfdoc_embeddertest.cpp b/fpdfsdk/fpdfdoc_embeddertest.cpp
index d7f1f97117..67934c5626 100644
--- a/fpdfsdk/fpdfdoc_embeddertest.cpp
+++ b/fpdfsdk/fpdfdoc_embeddertest.cpp
@@ -172,3 +172,57 @@ TEST_F(FPDFDocEmbeddertest, DeletePage) {
   FPDFPage_Delete(document(), 0);
   EXPECT_EQ(0, FPDF_GetPageCount(document()));
 }
+
+TEST_F(FPDFDocEmbeddertest, NoPageLabels) {
+  EXPECT_TRUE(OpenDocument("about_blank.pdf"));
+  EXPECT_EQ(1, FPDF_GetPageCount(document()));
+
+  ASSERT_EQ(0u, FPDF_GetPagelLabel(document(), 0, nullptr, 0));
+}
+
+TEST_F(FPDFDocEmbeddertest, GetPageLabels) {
+  EXPECT_TRUE(OpenDocument("page_labels.pdf"));
+  EXPECT_EQ(7, FPDF_GetPageCount(document()));
+
+  unsigned short buf[128];
+  EXPECT_EQ(0u, FPDF_GetPagelLabel(document(), -2, buf, sizeof(buf)));
+  EXPECT_EQ(0u, FPDF_GetPagelLabel(document(), -1, buf, sizeof(buf)));
+
+  const FX_WCHAR kExpectedPageLabel0[] = L"i";
+  ASSERT_EQ(4u, FPDF_GetPagelLabel(document(), 0, buf, sizeof(buf)));
+  EXPECT_EQ(CFX_WideString(kExpectedPageLabel0),
+            CFX_WideString::FromUTF16LE(buf, FXSYS_len(kExpectedPageLabel0)));
+
+  const FX_WCHAR kExpectedPageLabel1[] = L"ii";
+  ASSERT_EQ(6u, FPDF_GetPagelLabel(document(), 1, buf, sizeof(buf)));
+  EXPECT_EQ(CFX_WideString(kExpectedPageLabel1),
+            CFX_WideString::FromUTF16LE(buf, FXSYS_len(kExpectedPageLabel1)));
+
+  const FX_WCHAR kExpectedPageLabel2[] = L"1";
+  ASSERT_EQ(4u, FPDF_GetPagelLabel(document(), 2, buf, sizeof(buf)));
+  EXPECT_EQ(CFX_WideString(kExpectedPageLabel2),
+            CFX_WideString::FromUTF16LE(buf, FXSYS_len(kExpectedPageLabel2)));
+
+  const FX_WCHAR kExpectedPageLabel3[] = L"2";
+  ASSERT_EQ(4u, FPDF_GetPagelLabel(document(), 3, buf, sizeof(buf)));
+  EXPECT_EQ(CFX_WideString(kExpectedPageLabel3),
+            CFX_WideString::FromUTF16LE(buf, FXSYS_len(kExpectedPageLabel3)));
+
+  const FX_WCHAR kExpectedPageLabel4[] = L"zzA";
+  ASSERT_EQ(8u, FPDF_GetPagelLabel(document(), 4, buf, sizeof(buf)));
+  EXPECT_EQ(CFX_WideString(kExpectedPageLabel4),
+            CFX_WideString::FromUTF16LE(buf, FXSYS_len(kExpectedPageLabel4)));
+
+  const FX_WCHAR kExpectedPageLabel5[] = L"zzB";
+  ASSERT_EQ(8u, FPDF_GetPagelLabel(document(), 5, buf, sizeof(buf)));
+  EXPECT_EQ(CFX_WideString(kExpectedPageLabel5),
+            CFX_WideString::FromUTF16LE(buf, FXSYS_len(kExpectedPageLabel5)));
+
+  const FX_WCHAR kExpectedPageLabel6[] = L"";
+  ASSERT_EQ(2u, FPDF_GetPagelLabel(document(), 6, buf, sizeof(buf)));
+  EXPECT_EQ(CFX_WideString(kExpectedPageLabel6),
+            CFX_WideString::FromUTF16LE(buf, FXSYS_len(kExpectedPageLabel6)));
+
+  ASSERT_EQ(0u, FPDF_GetPagelLabel(document(), 7, buf, sizeof(buf)));
+  ASSERT_EQ(0u, FPDF_GetPagelLabel(document(), 8, buf, sizeof(buf)));
+}
diff --git a/fpdfsdk/fpdfview_c_api_test.c b/fpdfsdk/fpdfview_c_api_test.c
index afc357f1e3..5e6c36f2b1 100644
--- a/fpdfsdk/fpdfview_c_api_test.c
+++ b/fpdfsdk/fpdfview_c_api_test.c
@@ -64,6 +64,7 @@ int CheckPDFiumCApi() {
     CHK(FPDFLink_CountQuadPoints);
     CHK(FPDFLink_GetQuadPoints);
     CHK(FPDF_GetMetaText);
+    CHK(FPDF_GetPagelLabel);
 
     // fpdf_edit.h
     CHK(FPDF_CreateNewDocument);
diff --git a/public/fpdf_doc.h b/public/fpdf_doc.h
index 9d2229c0c1..b245d46900 100644
--- a/public/fpdf_doc.h
+++ b/public/fpdf_doc.h
@@ -281,26 +281,43 @@ DLLEXPORT FPDF_BOOL STDCALL FPDFLink_GetQuadPoints(FPDF_LINK linkAnnot,
 
 // Get meta-data |tag| content from |document|.
 //
-//   doc    - handle to the document
-//   tag    - the tag to retrieve. The tag can be one of:
-//              Title, Author, Subject, Keywords, Creator, Producer,
-//              CreationDate, or ModDate.
-//            For detailed explanations of these tags and their respective
-//            values, please refer to PDF Reference 1.6, section 10.2.1,
-//            'Document Information Dictionary'.
-//   buffer - a buffer for the title. May be NULL.
-//   buflen - the length of the buffer, in bytes. May be 0.
+//   document - handle to the document.
+//   tag      - the tag to retrieve. The tag can be one of:
+//                Title, Author, Subject, Keywords, Creator, Producer,
+//                CreationDate, or ModDate.
+//              For detailed explanations of these tags and their respective
+//              values, please refer to PDF Reference 1.6, section 10.2.1,
+//              'Document Information Dictionary'.
+//   buffer   - a buffer for the tag. May be NULL.
+//   buflen   - the length of the buffer, in bytes. May be 0.
 //
-// Returns the number of bytes in the title, including trailing zeros.
+// Returns the number of bytes in the tag, including trailing zeros.
 //
 // The |buffer| is always encoded in UTF-16LE. The |buffer| is followed by two
 // bytes of zeros indicating the end of the string.  If |buflen| is less than
 // the returned length, or |buffer| is NULL, |buffer| will not be modified.
-DLLEXPORT unsigned long STDCALL FPDF_GetMetaText(FPDF_DOCUMENT doc,
+DLLEXPORT unsigned long STDCALL FPDF_GetMetaText(FPDF_DOCUMENT document,
                                                  FPDF_BYTESTRING tag,
                                                  void* buffer,
                                                  unsigned long buflen);
 
+// Get the page label for |page_index| from |document|.
+//
+//   document    - handle to the document.
+//   page_index  - the 0-based index of the page.
+//   buffer      - a buffer for the page label. May be NULL.
+//   buflen      - the length of the buffer, in bytes. May be 0.
+//
+// Returns the number of bytes in the page label, including trailing zeros.
+//
+// The |buffer| is always encoded in UTF-16LE. The |buffer| is followed by two
+// bytes of zeros indicating the end of the string.  If |buflen| is less than
+// the returned length, or |buffer| is NULL, |buffer| will not be modified.
+DLLEXPORT unsigned long STDCALL FPDF_GetPagelLabel(FPDF_DOCUMENT document,
+                                                   int page_index,
+                                                   void* buffer,
+                                                   unsigned long buflen);
+
 #ifdef __cplusplus
 }  // extern "C"
 #endif  // __cplusplus
diff --git a/testing/resources/page_labels.pdf b/testing/resources/page_labels.pdf
new file mode 100644
index 0000000000..fd1a1e55e9
--- /dev/null
+++ b/testing/resources/page_labels.pdf