Added helper functions in the class CPDF_FileSpec

1. Added two helper functions in CPDF_FileSpec, which will be useful
when adding support for embedded files:
CPDF_FileSpec::GetFileStream() - useful because the stream object
contains the file's data
CPDF_FileSpec::GetParamsDict() - useful because the params dictionary
contains parameters of the file, such as creation date, size, etc.
    * Added two unit tests testing both functions.

Bug=pdfium:174

Change-Id: I33ea21ddb621434007f94767f281ead0b00ecb8a
Reviewed-on: https://pdfium-review.googlesource.com/7355
Reviewed-by: dsinclair <dsinclair@chromium.org>
Reviewed-by: Lei Zhang <thestig@chromium.org>
Commit-Queue: Jane Liu <janeliulwq@google.com>

3 files changed, 146 insertions, 5 deletions

diff --git a/core/fpdfdoc/cpdf_filespec.cpp b/core/fpdfdoc/cpdf_filespec.cpp
index 59915eaa05..7d1f0e68b9 100644
--- a/core/fpdfdoc/cpdf_filespec.cpp
+++ b/core/fpdfdoc/cpdf_filespec.cpp
@@ -6,9 +6,12 @@
 
 #include "core/fpdfdoc/cpdf_filespec.h"
 
+#include <vector>
+
 #include "core/fpdfapi/parser/cpdf_dictionary.h"
 #include "core/fpdfapi/parser/cpdf_name.h"
 #include "core/fpdfapi/parser/cpdf_object.h"
+#include "core/fpdfapi/parser/cpdf_stream.h"
 #include "core/fpdfapi/parser/cpdf_string.h"
 #include "core/fpdfapi/parser/fpdf_parser_decode.h"
 #include "core/fxcrt/fx_system.h"
@@ -50,6 +53,12 @@ CFX_WideString ChangeSlashToPDF(const wchar_t* str) {
 
 }  // namespace
 
+CPDF_FileSpec::CPDF_FileSpec(CPDF_Object* pObj) : m_pObj(pObj) {
+  ASSERT(m_pObj);
+}
+
+CPDF_FileSpec::~CPDF_FileSpec() {}
+
 CFX_WideString CPDF_FileSpec::DecodeFileName(const CFX_WideString& filepath) {
   if (filepath.GetLength() <= 1)
     return CFX_WideString();
@@ -112,9 +121,42 @@ bool CPDF_FileSpec::GetFileName(CFX_WideString* csFileName) const {
   return true;
 }
 
-CPDF_FileSpec::CPDF_FileSpec(CPDF_Object* pObj) : m_pObj(pObj) {}
+CPDF_Stream* CPDF_FileSpec::GetFileStream() const {
+  CPDF_Dictionary* pDict = m_pObj->AsDictionary();
+  if (!pDict)
+    return nullptr;
+
+  // Get the embedded files dictionary.
+  CPDF_Dictionary* pFiles = pDict->GetDictFor("EF");
+  if (!pFiles)
+    return nullptr;
+
+  // Get the file stream of the highest precedence with its file specification
+  // string available. Follows the same precedence order as GetFileName().
+  constexpr const char* keys[] = {"UF", "F", "DOS", "Mac", "Unix"};
+  size_t end = pDict->GetStringFor("FS") == "URL" ? 2 : FX_ArraySize(keys);
+  for (size_t i = 0; i < end; ++i) {
+    const CFX_ByteString& key = keys[i];
+    if (!pDict->GetUnicodeTextFor(key).IsEmpty()) {
+      CPDF_Stream* pStream = pFiles->GetStreamFor(key);
+      if (pStream)
+        return pStream;
+    }
+  }
+  return nullptr;
+}
 
-CPDF_FileSpec::~CPDF_FileSpec() {}
+CPDF_Dictionary* CPDF_FileSpec::GetParamsDict() const {
+  CPDF_Stream* pStream = GetFileStream();
+  if (!pStream)
+    return nullptr;
+
+  CPDF_Dictionary* pDict = pStream->GetDict();
+  if (!pDict)
+    return nullptr;
+
+  return pDict->GetDictFor("Params");
+}
 
 CFX_WideString CPDF_FileSpec::EncodeFileName(const CFX_WideString& filepath) {
   if (filepath.GetLength() <= 1)
@@ -146,9 +188,6 @@ CFX_WideString CPDF_FileSpec::EncodeFileName(const CFX_WideString& filepath) {
 }
 
 void CPDF_FileSpec::SetFileName(const CFX_WideString& wsFileName) {
-  if (!m_pObj)
-    return;
-
   CFX_WideString wsStr = EncodeFileName(wsFileName);
   if (m_pObj->IsString()) {
     m_pObj->SetString(CFX_ByteString::FromUnicode(wsStr));
diff --git a/core/fpdfdoc/cpdf_filespec.h b/core/fpdfdoc/cpdf_filespec.h
index 04baf2e63f..2cef20b48a 100644
--- a/core/fpdfdoc/cpdf_filespec.h
+++ b/core/fpdfdoc/cpdf_filespec.h
@@ -12,7 +12,9 @@
 #include "core/fxcrt/cfx_weak_ptr.h"
 #include "core/fxcrt/fx_string.h"
 
+class CPDF_Dictionary;
 class CPDF_Object;
+class CPDF_Stream;
 
 class CPDF_FileSpec {
  public:
@@ -27,6 +29,8 @@ class CPDF_FileSpec {
 
   CPDF_Object* GetObj() const { return m_pObj.Get(); }
   bool GetFileName(CFX_WideString* wsFileName) const;
+  CPDF_Stream* GetFileStream() const;
+  CPDF_Dictionary* GetParamsDict() const;
 
   // Set this file spec to refer to a file name (not a url).
   void SetFileName(const CFX_WideString& wsFileName);
diff --git a/core/fpdfdoc/cpdf_filespec_unittest.cpp b/core/fpdfdoc/cpdf_filespec_unittest.cpp
index e237aa5b6b..7e2975c5f0 100644
--- a/core/fpdfdoc/cpdf_filespec_unittest.cpp
+++ b/core/fpdfdoc/cpdf_filespec_unittest.cpp
@@ -3,10 +3,13 @@
 // found in the LICENSE file.
 
 #include <memory>
+#include <utility>
 #include <vector>
 
 #include "core/fpdfapi/parser/cpdf_dictionary.h"
 #include "core/fpdfapi/parser/cpdf_name.h"
+#include "core/fpdfapi/parser/cpdf_number.h"
+#include "core/fpdfapi/parser/cpdf_stream.h"
 #include "core/fpdfapi/parser/cpdf_string.h"
 #include "core/fpdfdoc/cpdf_filespec.h"
 #include "testing/gtest/include/gtest/gtest.h"
@@ -161,3 +164,98 @@ TEST(cpdf_filespec, SetFileName) {
   EXPECT_TRUE(file_spec2.GetFileName(&file_name));
   EXPECT_TRUE(file_name == test_data.input);
 }
+
+TEST(cpdf_filespec, GetFileStream) {
+  {
+    // Invalid object.
+    auto name_obj = pdfium::MakeUnique<CPDF_Name>(nullptr, "test.pdf");
+    CPDF_FileSpec file_spec(name_obj.get());
+    EXPECT_FALSE(file_spec.GetFileStream());
+  }
+  {
+    // Dictionary object missing its embedded files dictionary.
+    auto dict_obj = pdfium::MakeUnique<CPDF_Dictionary>();
+    CPDF_FileSpec file_spec(dict_obj.get());
+    EXPECT_FALSE(file_spec.GetFileStream());
+  }
+  {
+    // Dictionary object with an empty embedded files dictionary.
+    auto dict_obj = pdfium::MakeUnique<CPDF_Dictionary>();
+    dict_obj->SetNewFor<CPDF_Dictionary>("EF");
+    CPDF_FileSpec file_spec(dict_obj.get());
+    EXPECT_FALSE(file_spec.GetFileStream());
+  }
+  {
+    // Dictionary object with a non-empty embedded files dictionary.
+    auto dict_obj = pdfium::MakeUnique<CPDF_Dictionary>();
+    dict_obj->SetNewFor<CPDF_Dictionary>("EF");
+    CPDF_FileSpec file_spec(dict_obj.get());
+
+    const char* const keys[5] = {"Unix", "Mac", "DOS", "F", "UF"};
+    const wchar_t file_name[] = L"test.pdf";
+    const char* const stream[5] = {"test1", "test2", "test3", "test4", "test5"};
+    CPDF_Dictionary* file_dict =
+        file_spec.GetObj()->AsDictionary()->GetDictFor("EF");
+
+    // Keys in reverse order of precedence to retrieve the file content stream.
+    for (size_t i = 0; i < FX_ArraySize(keys); ++i) {
+      // Set the file name.
+      dict_obj->SetNewFor<CPDF_String>(keys[i], file_name);
+
+      // Set the file stream.
+      auto pDict = pdfium::MakeUnique<CPDF_Dictionary>();
+      size_t buf_len = strlen(stream[i]) + 1;
+      std::unique_ptr<uint8_t, FxFreeDeleter> buf(FX_Alloc(uint8_t, buf_len));
+      memcpy(buf.get(), stream[i], buf_len);
+      file_dict->SetNewFor<CPDF_Stream>(keys[i], std::move(buf), buf_len,
+                                        std::move(pDict));
+
+      // Check that the file content stream is as expected.
+      EXPECT_STREQ(
+          stream[i],
+          file_spec.GetFileStream()->GetUnicodeText().UTF8Encode().c_str());
+
+      if (i == 2) {
+        dict_obj->SetNewFor<CPDF_String>("FS", "URL", false);
+        EXPECT_FALSE(file_spec.GetFileStream());
+      }
+    }
+  }
+}
+
+TEST(cpdf_filespec, GetParamsDict) {
+  {
+    // Invalid object.
+    auto name_obj = pdfium::MakeUnique<CPDF_Name>(nullptr, "test.pdf");
+    CPDF_FileSpec file_spec(name_obj.get());
+    EXPECT_FALSE(file_spec.GetParamsDict());
+  }
+  {
+    // Dictionary object.
+    auto dict_obj = pdfium::MakeUnique<CPDF_Dictionary>();
+    dict_obj->SetNewFor<CPDF_Dictionary>("EF");
+    dict_obj->SetNewFor<CPDF_String>("UF", L"test.pdf");
+    CPDF_FileSpec file_spec(dict_obj.get());
+    EXPECT_FALSE(file_spec.GetParamsDict());
+
+    // Add a file stream to the embedded files dictionary.
+    CPDF_Dictionary* file_dict =
+        file_spec.GetObj()->AsDictionary()->GetDictFor("EF");
+    auto pDict = pdfium::MakeUnique<CPDF_Dictionary>();
+    std::unique_ptr<uint8_t, FxFreeDeleter> buf(FX_Alloc(uint8_t, 6));
+    memcpy(buf.get(), "hello", 6);
+    file_dict->SetNewFor<CPDF_Stream>("UF", std::move(buf), 6,
+                                      std::move(pDict));
+
+    // Add a params dictionary to the file stream.
+    CPDF_Stream* stream = file_dict->GetStreamFor("UF");
+    CPDF_Dictionary* stream_dict = stream->GetDict();
+    stream_dict->SetNewFor<CPDF_Dictionary>("Params");
+    EXPECT_TRUE(file_spec.GetParamsDict());
+
+    // Add a parameter to the params dictionary.
+    CPDF_Dictionary* params_dict = stream_dict->GetDictFor("Params");
+    params_dict->SetNewFor<CPDF_Number>("Size", 6);
+    EXPECT_EQ(6, file_spec.GetParamsDict()->GetIntegerFor("Size"));
+  }
+}