antlr3string.c

ANTLR(ANother Tool for Language Recognition)它是这样的一种工具

/** \file
 * Implementation of the ANTLR3 string and string factory classes
 */
#include    <antlr3string.h>

/* Factory API
 */
static    pANTLR3_STRING    newRaw8	(pANTLR3_STRING_FACTORY factory);
static    pANTLR3_STRING    newRaw16	(pANTLR3_STRING_FACTORY factory);
static    pANTLR3_STRING    newSize8	(pANTLR3_STRING_FACTORY factory, ANTLR3_UINT32 size);
static    pANTLR3_STRING    newSize16	(pANTLR3_STRING_FACTORY factory, ANTLR3_UINT32 size);
static    pANTLR3_STRING    newPtr8	(pANTLR3_STRING_FACTORY factory, pANTLR3_UINT8 string, ANTLR3_UINT32 size);
static    pANTLR3_STRING    newPtr16_8	(pANTLR3_STRING_FACTORY factory, pANTLR3_UINT8 string, ANTLR3_UINT32 size);
static    pANTLR3_STRING    newPtr16_16	(pANTLR3_STRING_FACTORY factory, pANTLR3_UINT8 string, ANTLR3_UINT32 size);
static    pANTLR3_STRING    newStr8	(pANTLR3_STRING_FACTORY factory, pANTLR3_UINT8 string);
static    pANTLR3_STRING    newStr16_8	(pANTLR3_STRING_FACTORY factory, pANTLR3_UINT8 string);
static    pANTLR3_STRING    newStr16_16	(pANTLR3_STRING_FACTORY factory, pANTLR3_UINT8 string);
static    void		    destroy	(pANTLR3_STRING_FACTORY factory, pANTLR3_STRING string);
static    pANTLR3_STRING    printable8	(pANTLR3_STRING_FACTORY factory, pANTLR3_STRING string);
static    pANTLR3_STRING    printable16	(pANTLR3_STRING_FACTORY factory, pANTLR3_STRING string);
static    void		    closeFactory(pANTLR3_STRING_FACTORY factory);

/* String API
 */
static    pANTLR3_UINT8	    set8	(pANTLR3_STRING string, const char * chars);
static    pANTLR3_UINT8	    set16_8	(pANTLR3_STRING string, const char * chars);
static    pANTLR3_UINT8	    set16_16	(pANTLR3_STRING string, const char * chars);
static    pANTLR3_UINT8	    append8	(pANTLR3_STRING string, const char * newbit);
static    pANTLR3_UINT8	    append16_8	(pANTLR3_STRING string, const char * newbit);
static    pANTLR3_UINT8	    append16_16	(pANTLR3_STRING string, const char * newbit);
static	  pANTLR3_UINT8	    insert8	(pANTLR3_STRING string, ANTLR3_UINT32 point, const char * newbit);
static	  pANTLR3_UINT8	    insert16_8	(pANTLR3_STRING string, ANTLR3_UINT32 point, const char * newbit);
static	  pANTLR3_UINT8	    insert16_16	(pANTLR3_STRING string, ANTLR3_UINT32 point, const char * newbit);

static    pANTLR3_UINT8	    setS	(pANTLR3_STRING string, pANTLR3_STRING chars);
static    pANTLR3_UINT8	    appendS	(pANTLR3_STRING string, pANTLR3_STRING newbit);
static	  pANTLR3_UINT8	    insertS	(pANTLR3_STRING string, ANTLR3_UINT32 point, pANTLR3_STRING newbit);

static    pANTLR3_UINT8	    addc8	(pANTLR3_STRING string, ANTLR3_UINT32 c);
static    pANTLR3_UINT8	    addc16	(pANTLR3_STRING string, ANTLR3_UINT32 c);
static    pANTLR3_UINT8	    addi8	(pANTLR3_STRING string, ANTLR3_INT32 i);
static    pANTLR3_UINT8	    addi16	(pANTLR3_STRING string, ANTLR3_INT32 i);
static	  pANTLR3_UINT8	    inserti8	(pANTLR3_STRING string, ANTLR3_UINT32 point, ANTLR3_INT32 i);
static	  pANTLR3_UINT8	    inserti16	(pANTLR3_STRING string, ANTLR3_UINT32 point, ANTLR3_INT32 i);

static    ANTLR3_UINT32     compare8	(pANTLR3_STRING string, const char * compStr);
static    ANTLR3_UINT32     compare16_8	(pANTLR3_STRING string, const char * compStr);
static    ANTLR3_UINT32     compare16_16(pANTLR3_STRING string, const char * compStr);
static    ANTLR3_UINT32     compareS	(pANTLR3_STRING string, pANTLR3_STRING compStr);
static    ANTLR3_UCHAR      charAt8	(pANTLR3_STRING string, ANTLR3_UINT32 offset);
static    ANTLR3_UCHAR      charAt16	(pANTLR3_STRING string, ANTLR3_UINT32 offset);
static    pANTLR3_STRING    subString8	(pANTLR3_STRING string, ANTLR3_UINT32 startIndex, ANTLR3_UINT32 endIndex);
static    pANTLR3_STRING    subString16	(pANTLR3_STRING string, ANTLR3_UINT32 startIndex, ANTLR3_UINT32 endIndex);
static	  ANTLR3_INT32	    toInt32_8	(pANTLR3_STRING string);
static	  ANTLR3_INT32	    toInt32_16  (pANTLR3_STRING string);
static	  pANTLR3_STRING    to8_8	(pANTLR3_STRING string);
static	  pANTLR3_STRING    to8_16	(pANTLR3_STRING string);

/* Local helpers
 */
static	void	stringInit8	(pANTLR3_STRING string);
static	void	stringInit16	(pANTLR3_STRING string);
static	void	stringFree	(pANTLR3_STRING string);

ANTLR3_API pANTLR3_STRING_FACTORY 
antlr3StringFactoryNew()
{
    pANTLR3_STRING_FACTORY  factory;

    /* Allocate memory
     */
    factory	= (pANTLR3_STRING_FACTORY) ANTLR3_MALLOC(sizeof(ANTLR3_STRING_FACTORY));

    if	(factory == NULL)
    {
	return	(pANTLR3_STRING_FACTORY)(ANTLR3_ERR_NOMEM);
    }

    /* Now we make a new list to track the strings, 256 will allow thousands
     * before there is any real performance degradation (until free() ;-).
     */
    factory->strings	= antlr3ListNew(256);
    factory->index	= 0;

    if	(factory->strings == (pANTLR3_LIST)(ANTLR3_ERR_NOMEM))
    {
	ANTLR3_FREE(factory);
	return	(pANTLR3_STRING_FACTORY)(ANTLR3_ERR_NOMEM);
    }

    /* Install the API (8 bit assumed)
     */
    factory->newRaw	=  newRaw8;
    factory->newSize	=  newSize8;

    factory->newPtr	=  newPtr8;
    factory->newPtr8	=  newPtr8;
    factory->newStr	=  newStr8;
    factory->newStr8	=  newStr8;
    factory->destroy	=  destroy;
    factory->printable	=  printable8;
    factory->destroy	=  destroy;
    factory->close	=  closeFactory;

    return  factory;
}

/** Create a string factory that is UCS2 (16 bit) encodingn based
 */
ANTLR3_API pANTLR3_STRING_FACTORY 
antlr3UCS2StringFactoryNew()
{
     pANTLR3_STRING_FACTORY  factory;

    /* Allocate an 8 bit factory, then override with 16 bit UCS2 functions where we 
     * need to.
     */
    factory	= antlr3StringFactoryNew();

    if	(factory == NULL)
    {
	return	(pANTLR3_STRING_FACTORY)(ANTLR3_ERR_NOMEM);
    }

    /* Override the 8 bit API with the UCS2 (mostly just 16 bit) API
     */
    factory->newRaw	=  newRaw16;
    factory->newSize	=  newSize16;

    factory->newPtr	=  newPtr16_16;
    factory->newPtr8	=  newPtr16_8;
    factory->newStr	=  newStr16_16;
    factory->newStr8	=  newStr16_8;
    factory->printable	=  printable16;

    factory->destroy	=  destroy;
    factory->destroy	=  destroy;
    factory->close	=  closeFactory;

    return  factory;
}

/**
 *
 * \param factory 
 * \return 
 */
static    pANTLR3_STRING    
newRaw8	(pANTLR3_STRING_FACTORY factory)
{
    pANTLR3_STRING  string;

    string  = (pANTLR3_STRING) ANTLR3_MALLOC(sizeof(ANTLR3_STRING));

    if	(string == NULL)
    {
	return	(pANTLR3_STRING)(ANTLR3_ERR_NOMEM);
    }

    /* Structure is allocated, now fill in the API etc.
     */
    stringInit8(string);
    string->factory = factory;

    /* Add the string into the allocated list
     */
    factory->strings->put(factory->strings, factory->index, (void *) string, (void (*)(void *))(stringFree));
    string->index   = factory->index++;

    return string;
}
/**
 *
 * \param factory 
 * \return 
 */
static    pANTLR3_STRING    
newRaw16	(pANTLR3_STRING_FACTORY factory)
{
    pANTLR3_STRING  string;

    string  = (pANTLR3_STRING) ANTLR3_MALLOC(sizeof(ANTLR3_STRING));

    if	(string == NULL)
    {
	return	(pANTLR3_STRING)(ANTLR3_ERR_NOMEM);
    }

    /* Structure is allocated, now fill in the API etc.
     */
    stringInit16(string);
    string->factory = factory;

    /* Add the string into the allocated list
     */
    factory->strings->put(factory->strings, factory->index, (void *) string, (void (*)(void *))(stringFree));
    string->index   = factory->index++;

    return string;
}
static	void	stringFree  (pANTLR3_STRING string)
{
    /* First free the string itself if there was anything in it
     */
    if	(string->chars)
    {
	ANTLR3_FREE(string->chars);
    }

    /* Now free the space for this string
     */
    ANTLR3_FREE(string);

    return;
}
/**
 *
 * \param string 
 * \return 
 */
static	void
stringInit8  (pANTLR3_STRING string)
{
    string->len		= 0;
    string->size	= 0;
    string->chars	= NULL;
    string->encoding	= ANTLR3_ENCODING_LATIN1;

    /* API for 8 bit strings*/

    string->set		= set8;
    string->set8	= set8;
    string->append	= append8;
    string->append8	= append8;
    string->insert	= insert8;
    string->insert8	= insert8;
    string->addi	= addi8;
    string->inserti	= inserti8;
    string->addc	= addc8;
    string->charAt	= charAt8;
    string->compare	= compare8;
    string->compare8	= compare8;
    string->subString	= subString8;
    string->toInt32	= toInt32_8;
    string->to8		= to8_8;

    string->compareS	= compareS;
    string->setS	= setS;
    string->appendS	= appendS;
    string->insertS	= insertS;

}
/**
 *
 * \param string 
 * \return 
 */
static	void
stringInit16  (pANTLR3_STRING string)
{
    string->len		= 0;
    string->size	= 0;
    string->chars	= NULL;
    string->encoding	= ANTLR3_ENCODING_UCS2;

    /* API for 16 bit strings */

    string->set		= set16_16;
    string->set8	= set16_8;
    string->append	= append16_16;
    string->append8	= append16_8;
    string->insert	= insert16_16;
    string->insert8	= insert16_8;
    string->addi	= addi16;
    string->inserti	= inserti16;
    string->addc	= addc16;
    string->charAt	= charAt16;
    string->compare	= compare16_16;
    string->compare8	= compare16_8;
    string->subString	= subString16;
    string->toInt32	= toInt32_16;
    string->to8		= to8_16;

    string->compareS	= compareS;
    string->setS	= setS;
    string->appendS	= appendS;
    string->insertS	= insertS;
}
/**
 *
 * \param string 
 * \return 
 * TODO: Implement UTF-8
 */
static	void
stringInitUTF8  (pANTLR3_STRING string)
{
    string->len	    = 0;
    string->size    = 0;
    string->chars   = NULL;

    /* API */

}

/**
 * Creates a new string with enough capacity for size 8 bit characters plus a terminator.
 *
 * \param[in] factory - Poitner to the string factory that owns strings
 * \param[in] size - In characters
 * \return pointer to the new string.
 */
static    pANTLR3_STRING    
newSize8	(pANTLR3_STRING_FACTORY factory, ANTLR3_UINT32 size)
{
    pANTLR3_STRING  string;

    string  = factory->newRaw(factory);

    if	(string == (pANTLR3_STRING)(ANTLR3_ERR_NOMEM))
    {
	return	string;
    }

    /* Always add one more byte for a terminator ;-)
     */
    string->chars   = (pANTLR3_UINT8) ANTLR3_MALLOC((size_t)(sizeof(ANTLR3_UINT8) * (size+1)));
    string->size    = size + 1;

    return string;
}
/**
 * Creates a new string with enough capacity for size 16 bit characters plus a terminator.
 *
 * \param[in] factory - POitner to the string factory that owns strings
 * \param[in] size - In characters
 * \return pointer to the new string.
 */
static    pANTLR3_STRING    
newSize16	(pANTLR3_STRING_FACTORY factory, ANTLR3_UINT32 size)
{
    pANTLR3_STRING  string;

    string  = factory->newRaw(factory);

    if	(string == (pANTLR3_STRING)(ANTLR3_ERR_NOMEM))
    {
	return	string;
    }

    /* Always add one more byte for a terminator ;-)
     */
    string->chars   = (pANTLR3_UINT8) ANTLR3_MALLOC((size_t)(sizeof(ANTLR3_UINT16) * (size+1)));
    string->size    = size+1;	/* Size is always in characters, as is len */

    return string;
}

/** Creates a new 8 bit string initialized with the 8 bit characters at the 
 *  supplied ptr, of pre-determined size.
 * \param[in] factory - Pointer to the string factory that owns the strings
 * \param[in] ptr - Pointer to 8 bit encoded characters
 * \return pointer to the new string
 */
static    pANTLR3_STRING    
newPtr8	(pANTLR3_STRING_FACTORY factory, pANTLR3_UINT8 ptr, ANTLR3_UINT32 size)
{
    pANTLR3_STRING  string;

    string  = factory->newSize(factory, size);

    if	(string == (pANTLR3_STRING)(ANTLR3_ERR_NOMEM))
    {
	return	string;
    }

    if	(size <= 0)
    {
	return	string;
    }

    if	(ptr != NULL)
    {
	ANTLR3_MEMMOVE(string->chars, (const void *)ptr, size);
	*(string->chars + size) = '\0';	    /* Terminate, these strings are usually used for Token streams and printing etc.	*/
	string->len = size;
    }

    return  string;
}

/** Creates a new 16 bit string initialized with the 8 bit characters at the 
 *  supplied 8 bit character ptr, of pre-determined size.
 * \param[in] factory - Pointer to the string factory that owns the strings
 * \param[in] ptr - Pointer to 8 bit encoded characters
 * \return pointer to the new string
 */
static    pANTLR3_STRING    
newPtr16_8	(pANTLR3_STRING_FACTORY factory, pANTLR3_UINT8 ptr, ANTLR3_UINT32 size)
{
    pANTLR3_STRING  string;

    /* newSize accepts size in characters, not bytes
     */
    string  = factory->newSize(factory, size);

    if	(string == (pANTLR3_STRING)(ANTLR3_ERR_NOMEM))
    {
	return	string;
    }

    if	(size <= 0)
    {
	return	string;
    }

    if	(ptr != NULL)