Blame xmpsdk/src/UnicodeInlines.incl_cpp

Packit 01d647
#ifndef __UnicodeInlines_incl_cpp__
Packit 01d647
#define __UnicodeInlines_incl_cpp__
Packit 01d647
Packit 01d647
// =================================================================================================
Packit 01d647
// Copyright 2002-2007 Adobe Systems Incorporated
Packit 01d647
// All Rights Reserved.
Packit 01d647
//
Packit 01d647
// NOTICE:  Adobe permits you to use, modify, and distribute this file in accordance with the terms
Packit 01d647
// of the Adobe license agreement accompanying it.
Packit 01d647
// =================================================================================================
Packit 01d647
Packit 01d647
#include "UnicodeConversions.hpp"
Packit 01d647
Packit 01d647
// =================================================================================================
Packit 01d647
// Inner loop utilities that need to be inlined.
Packit 01d647
// =================================================================================================
Packit 01d647
Packit 01d647
static inline XMP_Uns32 GetCodePoint ( const XMP_Uns8 ** utf8Str_io )
Packit 01d647
{
Packit 01d647
	const XMP_Uns8 * u8Ptr = *utf8Str_io;
Packit 01d647
	XMP_Uns32 cp;
Packit 01d647
	size_t u8Len;
Packit 01d647
	CodePoint_from_UTF8 ( u8Ptr, 4, &cp, &u8Len );	// Throws an exception for errors.
Packit 01d647
	*utf8Str_io = u8Ptr + u8Len;
Packit 01d647
	return cp;
Packit 01d647
}
Packit 01d647
Packit 01d647
// =================================================================================================
Packit 01d647
Packit 01d647
static inline bool IsStartChar_ASCII ( XMP_Uns32 cp )
Packit 01d647
{
Packit 01d647
	// ASCII starting characters for an XML name.
Packit 01d647
	if ( (('a' <= cp) && (cp <= 'z')) || (('A' <= cp) && (cp <= 'Z')) || (cp == '_') ) return true;
Packit 01d647
	return false;
Packit 01d647
}
Packit 01d647
Packit 01d647
// -------------------------------------------------------------------------------------------------
Packit 01d647
Packit 01d647
static inline bool IsStartChar_NonASCII ( XMP_Uns32 cp )
Packit 01d647
{
Packit 01d647
	// Non-ASCII starting characters for an XML name.
Packit 01d647
	
Packit 01d647
	if ( ((0xC0 <= cp) && (cp <= 0xD6))  || ((0xD8 <= cp) && (cp <= 0xF6)) ) return true;
Packit 01d647
	if ( ((0xF8 <= cp) && (cp <= 0x2FF)) || ((0x370 <= cp) && (cp <= 0x37D)) ) return true;
Packit 01d647
Packit 01d647
	if ( ((0x37F <= cp) && (cp <= 0x1FFF))  || ((0x200C <= cp) && (cp <= 0x200D)) ) return true;
Packit 01d647
	if ( ((0x2070 <= cp) && (cp <= 0x218F)) || ((0x2C00 <= cp) && (cp <= 0x2FEF)) ) return true;
Packit 01d647
	if ( ((0x3001 <= cp) && (cp <= 0xD7FF)) || ((0xF900 <= cp) && (cp <= 0xFDCF)) ) return true;
Packit 01d647
	if ( ((0xFDF0 <= cp) && (cp <= 0xFFFD)) || ((0x10000 <= cp) && (cp <= 0xEFFFF)) ) return true;
Packit 01d647
Packit 01d647
	return false;
Packit 01d647
Packit 01d647
}
Packit 01d647
Packit 01d647
// -------------------------------------------------------------------------------------------------
Packit 01d647
Packit 01d647
static inline bool IsOtherChar_ASCII ( XMP_Uns32 cp )
Packit 01d647
{
Packit 01d647
	// ASCII following characters for an XML name.
Packit 01d647
	if ( (('0' <= cp) && (cp <= '9')) || (cp == '-') || (cp == '.') ) return true;
Packit 01d647
	return false;
Packit 01d647
}
Packit 01d647
Packit 01d647
// -------------------------------------------------------------------------------------------------
Packit 01d647
Packit 01d647
static inline bool IsOtherChar_NonASCII ( XMP_Uns32 cp )
Packit 01d647
{
Packit 01d647
	// Non-ASCII following characters for an XML name.
Packit 01d647
	if ( (cp == 0xB7) || ((0x300 <= cp) && (cp <= 0x36F))  || ((0x203F <= cp) && (cp <= 0x2040)) ) return true;
Packit 01d647
	return false;
Packit 01d647
}
Packit 01d647
Packit 01d647
// -------------------------------------------------------------------------------------------------
Packit 01d647
Packit 01d647
static inline void VerifyUTF8 ( XMP_StringPtr str )
Packit 01d647
{
Packit 01d647
	const XMP_Uns8 * utf8Str = (XMP_Uns8*)str;
Packit 01d647
	while ( *utf8Str != 0 ) {
Packit 01d647
		while ( (*utf8Str != 0) && (*utf8Str < 0x80) ) ++utf8Str;
Packit 01d647
		if ( *utf8Str >= 0x80 ) (void) GetCodePoint ( &utf8Str );	// Throws for bad UTF-8.
Packit 01d647
	}
Packit 01d647
}
Packit 01d647
Packit 01d647
// -------------------------------------------------------------------------------------------------
Packit 01d647
Packit 01d647
static inline void VerifySimpleXMLName ( XMP_StringPtr _nameStart, XMP_StringPtr _nameEnd )
Packit 01d647
{
Packit 01d647
Packit 01d647
	const XMP_Uns8 * nameStart = (const XMP_Uns8 *) _nameStart;
Packit 01d647
	const XMP_Uns8 * nameEnd   = (const XMP_Uns8 *) _nameEnd;
Packit 01d647
	const XMP_Uns8 * namePos   = nameStart;
Packit 01d647
	XMP_Uns32 cp;
Packit 01d647
	
Packit 01d647
	// The first character is more restricted.
Packit 01d647
	
Packit 01d647
	if ( nameStart >= nameEnd ) XMP_Throw ( "Empty XML name", kXMPErr_BadXPath );
Packit 01d647
Packit 01d647
	cp = *namePos;
Packit 01d647
	if ( cp < 0x80 ) {
Packit 01d647
		++namePos;
Packit 01d647
		if ( ! IsStartChar_ASCII(cp) ) goto NameError;
Packit 01d647
	} else {
Packit 01d647
		cp = GetCodePoint ( &namePos );
Packit 01d647
		if ( ! IsStartChar_NonASCII(cp) ) goto NameError;
Packit 01d647
	}
Packit 01d647
Packit 01d647
	// Check the rest of the name.
Packit 01d647
	
Packit 01d647
	while ( namePos < nameEnd ) {
Packit 01d647
		cp = *namePos;
Packit 01d647
		if ( cp < 0x80 ) {
Packit 01d647
			++namePos;
Packit 01d647
			if ( (! IsStartChar_ASCII(cp)) && (! IsOtherChar_ASCII(cp)) ) goto NameError;
Packit 01d647
		} else {
Packit 01d647
			cp = GetCodePoint ( &namePos );
Packit 01d647
			if ( (! IsStartChar_NonASCII(cp)) && (! IsOtherChar_NonASCII(cp)) ) goto NameError;
Packit 01d647
		}
Packit 01d647
	}
Packit 01d647
Packit 01d647
	return;
Packit 01d647
Packit 01d647
NameError:
Packit 01d647
	XMP_Throw ( "Bad XML name", kXMPErr_BadXPath );
Packit 01d647
	
Packit 01d647
}	// VerifySimpleXMLName
Packit 01d647
Packit 01d647
// =================================================================================================
Packit 01d647
Packit 01d647
#endif	// __UnicodeInlines_incl_cpp__