xmpscanner.cpp

flash xmp sdk,flash官方SDK

// =================================================================================================
// Copyright 2002-2007 Adobe Systems Incorporated
// All Rights Reserved.
//
// NOTICE:  Adobe permits you to use, modify, and distribute this file in accordance with the terms
// of the Adobe license agreement accompanying it.
//
// Adobe patent application tracking #P435, entitled 'Unique markers to simplify embedding data of
// one format in a file with a different format', inventors: Sean Parent, Greg Gilley.
// =================================================================================================

#if WIN32
	// The VC++ debugger can't handle long symbol names.
	#pragma warning ( disable : 4786 )
#endif


#include "XMPScanner.hpp"

#include <cassert>
#include <string>
#include <cstdlib>

#if DEBUG
	#include <iostream>
	#include <iomanip>
	#include <fstream>
#endif


#ifndef UseStringPushBack	// VC++ 6.x does not provide push_back for strings!
	#define UseStringPushBack	0
#endif


using namespace std;


// *** Consider Boyer-Moore style search for "<?xpacket begin=".  It isn't an obvious win, the
// *** additional code might be slower than scanning every character.  Especially if we will
// *** read every cache line anyway.


// =================================================================================================
// =================================================================================================
// class PacketMachine
// ===================
//
// This is the packet recognizer state machine.  The top of the machine is FindNextPacket, this
// calls the specific state components and handles transitions.  The states are described by an
// array of RecognizerInfo records, indexed by the RecognizerKind enumeration.  Each RecognizerInfo
// record has a function that does that state's work, the success and failure transition states,
// and a string literal that is passed to the state function.  The literal lets a common MatchChar
// or MatchString function be used in several places.
//
// The state functions are responsible for consuming input to recognize their particular state.
// This includes intervening nulls for 16 and 32 bit character forms.  For the simplicity, things
// are treated as essentially little endian and the nulls are not actually checked.  The opening
// '<' is found with a byte-by-byte search, then the number of bytes per character is determined
// by counting the following nulls.  From then on, consuming a character means incrementing the
// buffer pointer by the number of bytes per character.  Thus the buffer pointer only points to
// the "real" bytes.  This also means that the pointer can go off the end of the buffer by a
// variable amount.  The amount of overrun is saved so that the pointer can be positioned at the
// right byte to start the next buffer.
//
// The state functions return a TriState value, eTriYes means the pattern was found, eTriNo means
// the pattern was definitely not found, eTriMaybe means that the end of the buffer was reached
// while working through the pattern.
//
// When eTriYes is returned, the fBufferPtr data member is left pointing to the "real" byte
// following the last actual byte.  Which might not be addressable memory!  This also means that
// a state function can be entered with nothing available in the buffer.  When eTriNo is returned,
// the fBufferPtr data member is left pointing to the byte that caused the failure.  The state 
// machine starts over from the failure byte.
//
// The state functions must preserve their internal micro-state before returning eTriMaybe, and
// resume processing when called with the next buffer.  The fPosition data member is used to denote
// how many actual characters have been consumed.  The fNullCount data member is used to denote how
// many nulls are left before the next actual character.


// =================================================================================================
// PacketMachine
// =============

XMPScanner::PacketMachine::PacketMachine ( XMP_Int64 bufferOffset, const void * bufferOrigin, XMP_Int64 bufferLength ) :

	// Public members
	fPacketStart ( 0 ),
	fPacketLength ( 0 ),
	fBytesAttr ( -1 ),
	fCharForm ( eChar8Bit ),
	fAccess ( ' ' ),
	fBogusPacket ( false ),
	
	// Private members
	fBufferOffset ( bufferOffset ),
	fBufferOrigin ( (const char *) bufferOrigin ),
	fBufferPtr ( fBufferOrigin ),
	fBufferLimit ( fBufferOrigin + bufferLength ),
	fRecognizer ( eLeadInRecognizer ),
	fPosition ( 0 ),
	fBytesPerChar ( 1 ),
	fBufferOverrun ( 0 ),
	fQuoteChar ( ' ' )

{
	/*
	REVIEW NOTES : Should the buffer stuff be in a class?
	*/
	
	assert ( bufferOrigin != NULL );
	assert ( bufferLength != 0 );
	
}	// PacketMachine


// =================================================================================================
// ~PacketMachine
// ==============

XMPScanner::PacketMachine::~PacketMachine ()
{

	// An empty placeholder.

}	// ~PacketMachine


// =================================================================================================
// AssociateBuffer
// ===============

void
XMPScanner::PacketMachine::AssociateBuffer ( XMP_Int64 bufferOffset, const void * bufferOrigin, XMP_Int64 bufferLength )
{

	fBufferOffset = bufferOffset;
	fBufferOrigin = (const char *) bufferOrigin;
	fBufferPtr = fBufferOrigin + fBufferOverrun;
	fBufferLimit = fBufferOrigin + bufferLength;
		
}	// AssociateBuffer


// =================================================================================================
// ResetMachine
// ============

void
XMPScanner::PacketMachine::ResetMachine ()
{

	fRecognizer = eLeadInRecognizer;
	fPosition = 0;
	fBufferOverrun = 0;
	fCharForm = eChar8Bit;
	fBytesPerChar = 1;
	fAccess = ' ';
	fBytesAttr = -1;
	fBogusPacket = false;
	
	fAttrName.erase ( fAttrName.begin(), fAttrName.end() );
	fAttrValue.erase ( fAttrValue.begin(), fAttrValue.end() );
	fEncodingAttr.erase ( fEncodingAttr.begin(), fEncodingAttr.end() );
		
}	// ResetMachine


// =================================================================================================
// FindLessThan
// ============

XMPScanner::PacketMachine::TriState
XMPScanner::PacketMachine::FindLessThan ( PacketMachine * ths, const char * which )
{
	
	if ( *which == 'H' ) {
	
		// --------------------------------------------------------------------------------
		// We're looking for the '<' of the header.  If we fail there is no packet in this
		// part of the input, so return eTriNo.
	
		ths->fCharForm = eChar8Bit;	// We might have just failed from a bogus 16 or 32 bit case.
		ths->fBytesPerChar = 1;

		while ( ths->fBufferPtr < ths->fBufferLimit ) {	// Don't skip nulls for the header's '<'!
			if ( *ths->fBufferPtr == '<' ) break;
			ths->fBufferPtr++;
		}
		
		if ( ths->fBufferPtr >= ths->fBufferLimit ) return eTriNo;
		ths->fBufferPtr++;
		return eTriYes;

	} else {
	
		// --------------------------------------------------------------------------------
		// We're looking for the '<' of the trailer.  We're already inside the packet body,
		// looking for the trailer.  So here if we fail we must return eTriMaybe so that we
		// keep looking for the trailer in the next buffer.

		const int bytesPerChar = ths->fBytesPerChar;

		while ( ths->fBufferPtr < ths->fBufferLimit ) {
			if ( *ths->fBufferPtr == '<' ) break;
			ths->fBufferPtr += bytesPerChar;
		}
		
		if ( ths->fBufferPtr >= ths->fBufferLimit ) return eTriMaybe;
		ths->fBufferPtr += bytesPerChar;
		return eTriYes;

	}

}	// FindLessThan


// =================================================================================================
// MatchString
// ===========

XMPScanner::PacketMachine::TriState
XMPScanner::PacketMachine::MatchString ( PacketMachine * ths, const char * literal )
{
	const int		bytesPerChar	= ths->fBytesPerChar;
	const char *	litPtr			= literal + ths->fPosition;
	const int		charsToGo		= strlen ( literal ) - ths->fPosition;
	int				charsDone		= 0;
	
	while ( (charsDone < charsToGo) && (ths->fBufferPtr < ths->fBufferLimit) ) {
		if ( *litPtr != *ths->fBufferPtr ) return eTriNo;
		charsDone++;
		litPtr++;
		ths->fBufferPtr += bytesPerChar;
	}
	
	if ( charsDone == charsToGo ) return eTriYes;
	ths->fPosition += charsDone;
	return eTriMaybe;

}	// MatchString


// =================================================================================================
// MatchChar
// =========

XMPScanner::PacketMachine::TriState
XMPScanner::PacketMachine::MatchChar ( PacketMachine * ths, const char * literal )
{
	const int	bytesPerChar	= ths->fBytesPerChar;

	if ( ths->fBufferPtr >= ths->fBufferLimit ) return eTriMaybe;
	
	const char currChar = *ths->fBufferPtr;
	if ( currChar != *literal ) return eTriNo;
	ths->fBufferPtr += bytesPerChar;	
	return eTriYes;

}	// MatchChar


// =================================================================================================
// MatchOpenQuote
// ==============

XMPScanner::PacketMachine::TriState
XMPScanner::PacketMachine::MatchOpenQuote ( PacketMachine * ths, const char * /* unused */ )
{
	const int	bytesPerChar	= ths->fBytesPerChar;

	if ( ths->fBufferPtr >= ths->fBufferLimit ) return eTriMaybe;
	
	const char currChar = *ths->fBufferPtr;
	if ( (currChar != '\'') && (currChar != '"') ) return eTriNo;
	ths->fQuoteChar = currChar;
	ths->fBufferPtr += bytesPerChar;	
	return eTriYes;

}	// MatchOpenQuote


// =================================================================================================
// MatchCloseQuote
// ===============

XMPScanner::PacketMachine::TriState
XMPScanner::PacketMachine::MatchCloseQuote ( PacketMachine * ths, const char * /* unused */ )
{

	return MatchChar ( ths, &ths->fQuoteChar );

}	// MatchCloseQuote


// =================================================================================================
// CaptureAttrName
// ===============

XMPScanner::PacketMachine::TriState
XMPScanner::PacketMachine::CaptureAttrName ( PacketMachine * ths, const char * /* unused */ )
{
	const int	bytesPerChar	= ths->fBytesPerChar;
	char		currChar;

	if ( ths->fPosition == 0 ) {	// Get the first character in the name.
		
		if ( ths->fBufferPtr >= ths->fBufferLimit ) return eTriMaybe;

		currChar = *ths->fBufferPtr;
		if ( ths->fAttrName.size() == 0 ) {
			if ( ! ( ( ('a' <= currChar) && (currChar <= 'z') ) ||
					 ( ('A' <= currChar) && (currChar <= 'Z') ) ||
					 (currChar == '_') || (currChar == ':') ) ) {
				return eTriNo;
			}
		}

		ths->fAttrName.erase ( ths->fAttrName.begin(), ths->fAttrName.end() );
		#if UseStringPushBack
			ths->fAttrName.push_back ( currChar );
		#else
			ths->fAttrName.insert ( ths->fAttrName.end(), currChar );
		#endif
		ths->fBufferPtr += bytesPerChar;
	
	}
	
	while ( ths->fBufferPtr < ths->fBufferLimit ) {	// Get the remainder of the name.

		currChar = *ths->fBufferPtr;
		if ( ! ( ( ('a' <= currChar) && (currChar <= 'z') ) ||
				 ( ('A' <= currChar) && (currChar <= 'Z') ) ||
				 ( ('0' <= currChar) && (currChar <= '9') ) ||
				 (currChar == '-') || (currChar == '.') || (currChar == '_') || (currChar == ':') ) ) {
			break;
		}

		#if UseStringPushBack
			ths->fAttrName.push_back ( currChar );
		#else
			ths->fAttrName.insert ( ths->fAttrName.end(), currChar );
		#endif
		ths->fBufferPtr += bytesPerChar;

	}
	
	if ( ths->fBufferPtr < ths->fBufferLimit ) return eTriYes;
	ths->fPosition = ths->fAttrName.size();	// The name might span into the next buffer.
	return eTriMaybe;

}	// CaptureAttrName


// =================================================================================================
// CaptureAttrValue
// ================
//
// Recognize the equal sign and the quoted string value, capture the value along the way.

XMPScanner::PacketMachine::TriState
XMPScanner::PacketMachine::CaptureAttrValue ( PacketMachine * ths, const char * /* unused */ )
{
	const int	bytesPerChar	= ths->fBytesPerChar;
	char		currChar		= 0;
	TriState	result			= eTriMaybe;
	
	if ( ths->fBufferPtr >= ths->fBufferLimit ) return eTriMaybe;