Merge pull request #534 from miloyip/issue325

Add encoding validation option for Writer/PrettyWriter

include/rapidjson/fwd.h

@@ -91,9 +91,14 @@ typedef GenericReader<UTF8<char>, UTF8<char>, CrtAllocator> Reader;
 
 // writer.h
 
-template<typename OutputStream, typename SourceEncoding, typename TargetEncoding, typename StackAllocator>
+template<typename OutputStream, typename SourceEncoding, typename TargetEncoding, typename StackAllocator, unsigned writeFlags>
 class Writer;
 
+// prettywriter.h
+
+template<typename OutputStream, typename SourceEncoding, typename TargetEncoding, typename StackAllocator, unsigned writeFlags>
+class PrettyWriter;
+
 // document.h
 
 template <typename Encoding, typename Allocator> 

include/rapidjson/prettywriter.h

@@ -31,8 +31,8 @@ RAPIDJSON_NAMESPACE_BEGIN
     \tparam TargetEncoding Encoding of output stream.
     \tparam StackAllocator Type of allocator for allocating memory of stack.
 */
-template<typename OutputStream, typename SourceEncoding = UTF8<>, typename TargetEncoding = UTF8<>, typename StackAllocator = CrtAllocator>
-class PrettyWriter : public Writer<OutputStream, SourceEncoding, TargetEncoding, StackAllocator> {
+template<typename OutputStream, typename SourceEncoding = UTF8<>, typename TargetEncoding = UTF8<>, typename StackAllocator = CrtAllocator, unsigned writeFlags = kWriteDefaultFlags>
+class PrettyWriter : public Writer<OutputStream, SourceEncoding, TargetEncoding, StackAllocator, writeFlags> {
 public:
     typedef Writer<OutputStream, SourceEncoding, TargetEncoding, StackAllocator> Base;
     typedef typename Base::Ch Ch;
@@ -42,9 +42,13 @@ public:
         \param allocator User supplied allocator. If it is null, it will create a private one.
         \param levelDepth Initial capacity of stack.
     */
-    PrettyWriter(OutputStream& os, StackAllocator* allocator = 0, size_t levelDepth = Base::kDefaultLevelDepth) : 
+    explicit PrettyWriter(OutputStream& os, StackAllocator* allocator = 0, size_t levelDepth = Base::kDefaultLevelDepth) : 
         Base(os, allocator, levelDepth), indentChar_(' '), indentCharCount_(4) {}
 
+
+    explicit PrettyWriter(StackAllocator* allocator = 0, size_t levelDepth = Base::kDefaultLevelDepth) : 
+        Base(allocator, levelDepth), indentChar_(' '), indentCharCount_(4) {}
+
     //! Set custom indentation.
     /*! \param indentChar       Character for indentation. Must be whitespace character (' ', '\\t', '\\n', '\\r').
         \param indentCharCount  Number of indent characters for each indentation level.

include/rapidjson/writer.h

@@ -35,6 +35,26 @@ RAPIDJSON_DIAG_OFF(padded)
 
 RAPIDJSON_NAMESPACE_BEGIN
 
+///////////////////////////////////////////////////////////////////////////////
+// WriteFlag
+
+/*! \def RAPIDJSON_WRITE_DEFAULT_FLAGS 
+    \ingroup RAPIDJSON_CONFIG
+    \brief User-defined kWriteDefaultFlags definition.
+
+    User can define this as any \c WriteFlag combinations.
+*/
+#ifndef RAPIDJSON_WRITE_DEFAULT_FLAGS
+#define RAPIDJSON_WRITE_DEFAULT_FLAGS kWriteNoFlags
+#endif
+
+//! Combination of writeFlags
+enum WriteFlag {
+    kWriteNoFlags = 0,              //!< No flags are set.
+    kWriteValidateEncodingFlag = 1, //!< Validate encoding of JSON strings.
+    kWriteDefaultFlags = RAPIDJSON_WRITE_DEFAULT_FLAGS  //!< Default write flags. Can be customized by defining RAPIDJSON_WRITE_DEFAULT_FLAGS
+};
+
 //! JSON writer
 /*! Writer implements the concept Handler.
     It generates JSON text by events to an output os.
@@ -51,7 +71,7 @@ RAPIDJSON_NAMESPACE_BEGIN
     \tparam StackAllocator Type of allocator for allocating memory of stack.
     \note implements Handler concept
 */
-template<typename OutputStream, typename SourceEncoding = UTF8<>, typename TargetEncoding = UTF8<>, typename StackAllocator = CrtAllocator>
+template<typename OutputStream, typename SourceEncoding = UTF8<>, typename TargetEncoding = UTF8<>, typename StackAllocator = CrtAllocator, unsigned writeFlags = kWriteDefaultFlags>
 class Writer {
 public:
     typedef typename SourceEncoding::Ch Ch;
@@ -318,9 +338,10 @@ protected:
                     PutUnsafe(*os_, hexDigits[static_cast<unsigned char>(c) & 0xF]);
                 }
             }
-            else
-                if (RAPIDJSON_UNLIKELY(!(Transcoder<SourceEncoding, TargetEncoding>::TranscodeUnsafe(is, *os_))))
-                    return false;
+            else if (RAPIDJSON_UNLIKELY(!(writeFlags & kWriteValidateEncodingFlag ? 
+                Transcoder<SourceEncoding, TargetEncoding>::Validate(is, *os_) :
+                Transcoder<SourceEncoding, TargetEncoding>::TranscodeUnsafe(is, *os_))))
+                return false;
         }
         PutUnsafe(*os_, '\"');
         return true;

test/unittest/fwdtest.cpp

@@ -69,7 +69,10 @@ struct Foo {
     Reader* reader;
 
     // writer.h
-    Writer<StringBuffer, UTF8<char>, UTF8<char>, CrtAllocator>* writer;
+    Writer<StringBuffer, UTF8<char>, UTF8<char>, CrtAllocator, 0>* writer;
+
+    // prettywriter.h
+    PrettyWriter<StringBuffer, UTF8<char>, UTF8<char>, CrtAllocator, 0>* prettywriter;
 
     // document.h
     Value* value;
@@ -94,6 +97,7 @@ struct Foo {
 #include "rapidjson/memorystream.h"
 #include "rapidjson/document.h" // -> reader.h
 #include "rapidjson/writer.h"
+#include "rapidjson/prettywriter.h"
 #include "rapidjson/schema.h"   // -> pointer.h
 
 Foo::Foo() : 
@@ -139,6 +143,9 @@ Foo::Foo() :
     // writer.h
     writer(RAPIDJSON_NEW((Writer<StringBuffer>))),
 
+    // prettywriter.h
+    prettywriter(RAPIDJSON_NEW((PrettyWriter<StringBuffer>))),
+
     // document.h
     value(RAPIDJSON_NEW(Value)),
     document(RAPIDJSON_NEW(Document)),
@@ -196,6 +203,9 @@ Foo::~Foo() {
     // writer.h
     RAPIDJSON_DELETE(writer);
 
+    // prettywriter.h
+    RAPIDJSON_DELETE(prettywriter);
+
     // document.h
     RAPIDJSON_DELETE(value);
     RAPIDJSON_DELETE(document);

test/unittest/writertest.cpp

@@ -347,6 +347,31 @@ TEST(Writer, InvalidEncoding) {
     }
 }
 
+TEST(Writer, ValidateEncoding) {
+    {
+        StringBuffer buffer;
+        Writer<StringBuffer, UTF8<>, UTF8<>, CrtAllocator, kWriteValidateEncodingFlag> writer(buffer);
+        writer.StartArray();
+        EXPECT_TRUE(writer.String("\x24"));             // Dollar sign U+0024
+        EXPECT_TRUE(writer.String("\xC2\xA2"));         // Cents sign U+00A2
+        EXPECT_TRUE(writer.String("\xE2\x82\xAC"));     // Euro sign U+20AC
+        EXPECT_TRUE(writer.String("\xF0\x9D\x84\x9E")); // G clef sign U+1D11E
+        writer.EndArray();
+        EXPECT_STREQ("[\"\x24\",\"\xC2\xA2\",\"\xE2\x82\xAC\",\"\xF0\x9D\x84\x9E\"]", buffer.GetString());
+    }
+
+    // Fail in decoding invalid UTF-8 sequence http://www.cl.cam.ac.uk/~mgk25/ucs/examples/UTF-8-test.txt
+    {
+        StringBuffer buffer;
+        Writer<StringBuffer, UTF8<>, UTF8<>, CrtAllocator, kWriteValidateEncodingFlag> writer(buffer);
+        writer.StartArray();
+        EXPECT_FALSE(writer.String("\xfe"));
+        EXPECT_FALSE(writer.String("\xff"));
+        EXPECT_FALSE(writer.String("\xfe\xfe\xff\xff"));
+        writer.EndArray();
+    }
+}
+
 TEST(Writer, InvalidEventSequence) {
     // {]
     {