corelib/src/SRDBinary.cc

/******************************************************************************/
/* SRD PARSER AND PREPROCESSOR - BINARY FORM **********************************/
/******************************************************************************/

#include <lw/lib/SRDBinary.hh>
using namespace lw;


/*= MAGIC NUMBERS, VERSIONS AND TAGS =========================================*/

namespace {

// Magic number for binary SRD
static constexpr uint32_t C_MAGIC_ = 0xea7ca1ce;

// E_Versions_ - Binary SRD version numbers
enum class E_Version_ : uint32_t {
	V1 = 0xc0ffee00 ,
};

// E_V1Tag_ - Version 1 SRD tags
enum class E_V1Tag_ : uint8_t {
	END = 0x00 ,
	LIST = 0x01 ,
	WORD_NEW = 0x02 ,
	WORD_KNOWN = 0x03 ,
	VAR_WORD = 0x04 ,
	STRING = 0x05 ,
	INT = 0x06 ,
	LONG = 0x07 ,
	FLOAT = 0x08 ,
	COMMENT = 0x09 ,
	BINARY = 0x0a ,
};

} // namespace


namespace lw {

// Writer for version numbers
template< >
struct T_ObjectWriter< E_Version_ >
{
	static void write( T_BinaryWriter const& writer , E_Version_ const& v )
	{
		writer.write( ( uint32_t ) v );
	}
};

// Writer for version 1 tags
template< >
struct T_ObjectWriter< E_V1Tag_ >
{
	static void write( T_BinaryWriter const& writer , E_V1Tag_ const& v )
	{
		writer.write( ( uint8_t ) v );
	}
};

// Reader for version 1 tags
template< >
struct T_ObjectReader< E_V1Tag_ >
{
	static E_V1Tag_ read( T_BinaryReader const& reader )
	{
		return ( E_V1Tag_ ) reader.read< uint8_t >( );
	}
};

} // namespace


/*= VERSION 1 BINARY READER ==================================================*/

namespace {

class T_SRDBRVersion1_
{
   private:
	T_BinaryReader& reader_;
	T_String const& name_;
	T_SRDErrors& errors_;
	A_SRDReaderTarget& target_;

	T_HashIndex wordsIndex_;
	T_Array< T_String > words_;
	uint32_t lastTagPos_;

	typedef std::function< T_SRDToken( T_String const& ) > F_StringTok;

	size_t curPos( ) const
	{
		return reader_.stream( ).position( );
	}

	void push( T_SRDToken&& token )
	{
		token.location( name_ , lastTagPos_ );
		target_.push( errors_ , std::move( token ) );
	}

	void readNewWord( F_StringTok mt );
	void readKnownWord( F_StringTok mt );
	void readString( F_StringTok mt );

	void readLoop( );

   public:
	T_SRDBRVersion1_( T_BinaryReader& reader , T_String const& name , T_SRDErrors& errors ,
			 A_SRDReaderTarget& target )
		: reader_( reader ) , name_( name ) , errors_( errors ) ,
		target_( target )
	{ }

	void read( );
};

/*----------------------------------------------------------------------------*/

void T_SRDBRVersion1_::readNewWord( F_StringTok mt )
{
	const auto wordPos( curPos( ) );
	const T_String word( reader_.read< T_String >( ).usePool( ) );
	if ( word.size( ) == 0 ) {
		errors_.add( "empty word" , name_ , uint32_t( wordPos ) );
		return;
	}

	const bool valid = T_SRDToken::IsWord( word );
	if ( !valid ) {
		errors_.add( "invalid word" , name_ , wordPos );
	}

	const auto hash( ComputeHash( word ) );
	uint32_t index( wordsIndex_.first( hash ) );
	while ( index != T_HashIndex::INVALID_INDEX ) {
		if ( words_[ index ] == word ) {
			errors_.add( "duplicate word" , name_ , wordPos );
			return;
		}
		index = wordsIndex_.next( index );
	}

	wordsIndex_.add( hash );
	words_.add( word );
	if ( valid ) {
		push( mt( word ) );
	}
}

void T_SRDBRVersion1_::readKnownWord( F_StringTok mt )
{
	const auto pos( curPos( ) );
	const auto widx( reader_.read< uint32_t >( ) );

	if ( widx >= words_.size( ) ) {
		errors_.add( "unregistered word" , name_ , pos );
		return;
	}

	T_String const& word( words_[ widx ] );
	if ( T_SRDToken::IsWord( word ) ) {
		push( mt( word ) );
	}
}

void T_SRDBRVersion1_::readString( F_StringTok mt )
{
	const auto str( reader_.read< T_String >( ).usePool( ) );
	push( mt( str ) );
}

/*----------------------------------------------------------------------------*/

void T_SRDBRVersion1_::readLoop( )
{
	bool pendingVariable = false;
	uint32_t depth = 1;
	while ( depth > 0 ) {
		lastTagPos_ = reader_.stream( ).position( );
		const auto tag( reader_.read< E_V1Tag_ >( ) );

		if ( pendingVariable && tag != E_V1Tag_::WORD_NEW && tag != E_V1Tag_::WORD_KNOWN ) {
			pendingVariable = false;
			errors_.add( "spurious variable tag" , name_ , lastTagPos_ - 1 );
		}

		switch ( tag ) {

		    case E_V1Tag_::LIST:
			depth ++;
			push( T_SRDToken::ListStart( ) );
			break;

		    case E_V1Tag_::END:
			depth --;
			if ( depth > 0 ) {
				push( T_SRDToken::ListEnd( ) );
			}
			break;

		    case E_V1Tag_::WORD_NEW:
			if ( pendingVariable ) {
				lastTagPos_ --;
				readNewWord( T_SRDToken::Variable );
				pendingVariable = false;
			} else {
				readNewWord( T_SRDToken::Word );
			}
			break;

		    case E_V1Tag_::WORD_KNOWN:
			if ( pendingVariable ) {
				lastTagPos_ --;
				readKnownWord( T_SRDToken::Variable );
				pendingVariable = false;
			} else {
				readKnownWord( T_SRDToken::Word );
			}
			break;

		    case E_V1Tag_::VAR_WORD:
			pendingVariable = true;
			break;

		    case E_V1Tag_::STRING:
			readString( T_SRDToken::String );
			break;

		    case E_V1Tag_::COMMENT:
			readString( T_SRDToken::Comment );
			break;

		    case E_V1Tag_::INT:
			push( T_SRDToken::Integer( reader_.read< int32_t >( ) ) );
			break;

		    case E_V1Tag_::LONG:
			push( T_SRDToken::Long( reader_.read< int64_t >( ) ) );
			break;

		    case E_V1Tag_::FLOAT:
			push( T_SRDToken::Float( reader_.read< double >( ) ) );
			break;

		    case E_V1Tag_::BINARY:
			{
				const uint32_t size( reader_.read< uint32_t >( ) );
				T_SharedPtr< T_Buffer< uint8_t > > buffer(
						NewShared< T_Buffer< uint8_t > >( size ) );
				if ( size ) {
					reader_.stream( ).read( buffer->data( ) , size );
				}
				push( T_SRDToken::Binary( buffer ) );
			}
			break;
		}
	}
}

void T_SRDBRVersion1_::read( )
{
	T_SRDReaderTargetHelper rth( target_ , errors_ );
	try {
		readLoop( );
	} catch ( X_StreamError const& e ) {
		switch ( e.code( ) ) {

		    case E_StreamError::BAD_DATA:
			errors_.add( "invalid data" , name_ , reader_.stream( ).position( ) );
			break;

		    case E_StreamError::END:
			errors_.add( "unexpected end of file" , name_ , reader_.stream( ).position( ) );
			break;

		    default:
			errors_.add( "read error" , name_ , reader_.stream( ).position( ) );
			break;
		}
	}
}


} // namespace


/*= T_SRDBinaryWriter ========================================================*/

T_SRDBinaryWriter::T_SRDBinaryWriter( A_OutputStream& output )
	: writer_( output , E_Endian::LITTLE ) , depth_( 0 )
{ }

/*----------------------------------------------------------------------------*/

void T_SRDBinaryWriter::flushComment( )
{
	if ( comment_.size( ) != 0 ) {
		writer_.write( E_V1Tag_::COMMENT );
		writer_.write( comment_ );
		comment_.clear( );
	}
}

void T_SRDBinaryWriter::writeWord( T_String const& word )
{
	// Try to find the word in the index
	const uint32_t hash = ComputeHash( word );
	uint32_t index = wordsIndex_.first( hash );
	while ( index != T_HashIndex::INVALID_INDEX ) {
		if ( word == words_[ index ] ) {
			break;
		}
		index = wordsIndex_.next( index );
	}

	if ( index == T_HashIndex::INVALID_INDEX ) {
		// New word
		wordsIndex_.add( hash );
		words_.add( word );
		writer_.write( E_V1Tag_::WORD_NEW );
		writer_.write( word );
	} else {
		// Known word, use the index
		writer_.write( E_V1Tag_::WORD_KNOWN );
		writer_.write( index );
	}
}

/*----------------------------------------------------------------------------*/

T_SRDBinaryWriter& T_SRDBinaryWriter::start( )
{
	if ( depth_ != 0 ) {
		throw X_SRDWriterError( "already started" );
	}
	writer_.write( C_MAGIC_ );
	writer_.write( E_Version_::V1 );
	depth_ = 1;
	return *this;
}

T_SRDBinaryWriter& T_SRDBinaryWriter::end( )
{
	if ( depth_ != 1 ) {
		throw X_SRDWriterError( depth_ == 0 ? "already ended" : "unterminated lists" );
	}

	flushComment( );
	writer_.write( E_V1Tag_::END );
	depth_ = 0;

	comment_.free( );
	wordsIndex_.free( );
	words_.free( );

	return *this;
}

/*----------------------------------------------------------------------------*/

T_SRDBinaryWriter& T_SRDBinaryWriter::startList( )
{
	if ( depth_ == 0 ) {
		throw X_SRDWriterError( "not started" );
	}
	flushComment( );
	writer_.write( E_V1Tag_::LIST );
	depth_ ++;
	return *this;
}

T_SRDBinaryWriter& T_SRDBinaryWriter::endList( )
{
	if ( depth_ == 0 ) {
		throw X_SRDWriterError( "not started" );
	}
	flushComment( );
	writer_.write( E_V1Tag_::END );
	depth_ --;
	return *this;
}

/*----------------------------------------------------------------------------*/

T_SRDBinaryWriter& T_SRDBinaryWriter::putText( T_String const& text )
{
	if ( T_SRDToken::IsWord( text ) ) {
		return putWord( text );
	} else {
		return putString( text );
	}
}

T_SRDBinaryWriter& T_SRDBinaryWriter::putWord( T_String const& word )
{
	if ( depth_ == 0 ) {
		throw X_SRDWriterError( "not started" );
	}
	flushComment( );
	writeWord( word );
	return *this;
}

T_SRDBinaryWriter& T_SRDBinaryWriter::putString( T_String const& string )
{
	if ( depth_ == 0 ) {
		throw X_SRDWriterError( "not started" );
	}
	flushComment( );
	writer_.write( E_V1Tag_::STRING );
	writer_.write( string );
	return *this;
}

T_SRDBinaryWriter& T_SRDBinaryWriter::putComment( T_String const& comment )
{
	if ( depth_ == 0 ) {
		throw X_SRDWriterError( "not started" );
	}
	comment_ << comment;
	return *this;
}

T_SRDBinaryWriter& T_SRDBinaryWriter::putVariable( T_String const& variable )
{
	if ( depth_ == 0 ) {
		throw X_SRDWriterError( "not started" );
	}
	flushComment( );
	writer_.write( E_V1Tag_::VAR_WORD );
	writeWord( variable );
	return *this;
}

/*----------------------------------------------------------------------------*/

T_SRDBinaryWriter& T_SRDBinaryWriter::putBinary(
		T_Buffer< uint8_t > const& binary )
{
	if ( depth_ == 0 ) {
		throw X_SRDWriterError( "not started" );
	}
	flushComment( );
	writer_.write( E_V1Tag_::BINARY );
	writer_.write( uint32_t( binary.bytes( ) ) );
	if ( binary.size( ) != 0 ) {
		writer_.stream( ).write( binary.data( ) , binary.bytes( ) );
	}
	return *this;
}

/*----------------------------------------------------------------------------*/

T_SRDBinaryWriter& T_SRDBinaryWriter::putInteger( int64_t value )
{
	if ( value >= INT32_MIN && value <= INT32_MAX ) {
		return putInt32( int32_t( value ) );
	} else {
		return putInt64( int64_t( value ) );
	}
}

T_SRDBinaryWriter& T_SRDBinaryWriter::putInt32( int32_t value )
{
	if ( depth_ == 0 ) {
		throw X_SRDWriterError( "not started" );
	}
	flushComment( );
	writer_.write( E_V1Tag_::INT );
	writer_.write( value );
	return *this;
}

T_SRDBinaryWriter& T_SRDBinaryWriter::putInt64( int64_t value )
{
	if ( depth_ == 0 ) {
		throw X_SRDWriterError( "not started" );
	}
	flushComment( );
	writer_.write( E_V1Tag_::LONG );
	writer_.write( value );
	return *this;
}

/*----------------------------------------------------------------------------*/

T_SRDBinaryWriter& T_SRDBinaryWriter::putFloat( double value )
{
	if ( depth_ == 0 ) {
		throw X_SRDWriterError( "not started" );
	}
	flushComment( );
	writer_.write( E_V1Tag_::FLOAT );
	writer_.write( value );
	return *this;
}


/*= T_SRDBinaryReader ========================================================*/

void T_SRDBinaryReader::read( T_String const& name , A_InputStream& input )
{
	T_BinaryReader rd( input , E_Endian::LITTLE );
	T_SRDErrors errors;

	// Read/check magic number
	const auto magicPos( input.position( ) );
	const uint32_t magic( ([&]( ) {
		try {
			return rd.read< uint32_t >( );
		} catch ( X_StreamError& e ) {
			switch ( e.code( ) ) {

			    case E_StreamError::BAD_DATA:
			    case E_StreamError::END:
				errors.add( "missing magic" , name , input.position( ) );
				break;

			    default:
				errors.add( "read error" , name , input.position( ) );
				break;
			}
			throw X_SRDErrors( errors );
		}
	})( ) );
	if ( magic != C_MAGIC_ ) {
		errors.add( "invalid magic number" , name , magicPos );
		throw X_SRDErrors( errors );
	}

	// Read version number
	const uint32_t versionPos( input.position( ) );
	const uint32_t version( ([&]( ) {
		try {
			return rd.read< uint32_t >( );
		} catch ( X_StreamError& e ) {
			switch ( e.code( ) ) {

			    case E_StreamError::BAD_DATA:
			    case E_StreamError::END:
				errors.add( "missing version ID" , name , input.position( ) );
				break;

			    default:
				errors.add( "read error" , name , input.position( ) );
				break;
			}
			throw X_SRDErrors( errors );
		}
	})( ) );
	switch ( version ) {

	    case uint32_t( E_Version_::V1 ):
		T_SRDBRVersion1_( rd , name , errors , target_ ).read( );
		break;

	    default:
		errors.add( "invalid version ID" , name , versionPos );
		break;
	}

	if ( errors.size( ) != 0 ) {
		throw X_SRDErrors( errors );
	}
}