src/jpeg/jdct.h

/*
 * jdct.h
 *
 * Copyright (C) 1994-1996, Thomas G. Lane.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
 * This include file contains common declarations for the forward and
 * inverse DCT modules.  These declarations are private to the DCT managers
 * (jcdctmgr.c, jddctmgr.c) and the individual DCT algorithms.
 * The individual DCT algorithms are kept in separate files to ease
 * machine-dependent tuning (e.g., assembly coding).
 */


/*
 * A forward DCT routine is given a pointer to a work area of type DCTELEM[];
 * the DCT is to be performed in-place in that buffer.  Type DCTELEM is int
 * for 8-bit samples, INT32 for 12-bit samples.  (NOTE: Floating-point DCT
 * implementations use an array of type FAST_FLOAT, instead.)
 * The DCT inputs are expected to be signed (range +-CENTERJSAMPLE).
 * The DCT outputs are returned scaled up by a factor of 8; they therefore
 * have a range of +-8K for 8-bit data, +-128K for 12-bit data.  This
 * convention improves accuracy in integer implementations and saves some
 * work in floating-point ones.
 * Quantization of the output coefficients is done by jcdctmgr.c.
 */

#if BITS_IN_JSAMPLE == 8
typedef int DCTELEM;		/* 16 or 32 bits is fine */
#else
typedef JPEG_INT32 DCTELEM;		/* must have 32 bits */
#endif

typedef JMETHOD(void, forward_DCT_method_ptr, (DCTELEM * data));
typedef JMETHOD(void, float_DCT_method_ptr, (FAST_FLOAT * data));


/*
 * An inverse DCT routine is given a pointer to the input JBLOCK and a pointer
 * to an output sample array.  The routine must dequantize the input data as
 * well as perform the IDCT; for dequantization, it uses the multiplier table
 * pointed to by compptr->dct_table.  The output data is to be placed into the
 * sample array starting at a specified column.  (Any row offset needed will
 * be applied to the array pointer before it is passed to the IDCT code.)
 * Note that the number of samples emitted by the IDCT routine is
 * DCT_scaled_size * DCT_scaled_size.
 */

/* typedef inverse_DCT_method_ptr is declared in jpegint.h */

/*
 * Each IDCT routine has its own ideas about the best dct_table element type.
 */

typedef MULTIPLIER ISLOW_MULT_TYPE; /* short or int, whichever is faster */
#if BITS_IN_JSAMPLE == 8
typedef MULTIPLIER IFAST_MULT_TYPE; /* 16 bits is OK, use short if faster */
#define IFAST_SCALE_BITS  2	/* fractional bits in scale factors */
#else
typedef JPEG_INT32 IFAST_MULT_TYPE;	/* need 32 bits for scaled quantizers */
#define IFAST_SCALE_BITS  13	/* fractional bits in scale factors */
#endif
typedef FAST_FLOAT FLOAT_MULT_TYPE; /* preferred floating type */


/*
 * Each IDCT routine is responsible for range-limiting its results and
 * converting them to unsigned form (0..MAXJSAMPLE).  The raw outputs could
 * be quite far out of range if the input data is corrupt, so a bulletproof
 * range-limiting step is required.  We use a mask-and-table-lookup method
 * to do the combined operations quickly.  See the comments with
 * prepare_range_limit_table (in jdmaster.c) for more info.
 */

#define IDCT_range_limit(cinfo)  ((cinfo)->sample_range_limit + CENTERJSAMPLE)

#define RANGE_MASK  (MAXJSAMPLE * 4 + 3) /* 2 bits wider than legal samples */


/* Short forms of external names for systems with brain-damaged linkers. */

#ifdef NEED_SHORT_EXTERNAL_NAMES
#define jpeg_fdct_islow		jFDislow
#define jpeg_fdct_ifast		jFDifast
#define jpeg_fdct_float		jFDfloat
#define jpeg_idct_islow		jRDislow
#define jpeg_idct_ifast		jRDifast
#define jpeg_idct_float		jRDfloat
#define jpeg_idct_4x4		jRD4x4
#define jpeg_idct_2x2		jRD2x2
#define jpeg_idct_1x1		jRD1x1
#endif /* NEED_SHORT_EXTERNAL_NAMES */

/* Extern declarations for the forward and inverse DCT routines. */

EXTERN(void) jpeg_fdct_islow JPP((DCTELEM * data));
EXTERN(void) jpeg_fdct_ifast JPP((DCTELEM * data));
EXTERN(void) jpeg_fdct_float JPP((FAST_FLOAT * data));

EXTERN(void) jpeg_idct_islow
    JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
	 JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
EXTERN(void) jpeg_idct_ifast
    JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
	 JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
EXTERN(void) jpeg_idct_float
    JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
	 JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
EXTERN(void) jpeg_idct_4x4
    JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
	 JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
EXTERN(void) jpeg_idct_2x2
    JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
	 JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
EXTERN(void) jpeg_idct_1x1
    JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
	 JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));


/*
 * Macros for handling fixed-point arithmetic; these are used by many
 * but not all of the DCT/IDCT modules.
 *
 * All values are expected to be of type INT32.
 * Fractional constants are scaled left by CONST_BITS bits.
 * CONST_BITS is defined within each module using these macros,
 * and may differ from one module to the next.
 */

#define ONE	((JPEG_INT32) 1)
#define CONST_SCALE (ONE << CONST_BITS)

/* Convert a positive real constant to an integer scaled by CONST_SCALE.
 * Caution: some C compilers fail to reduce "FIX(constant)" at compile time,
 * thus causing a lot of useless floating-point operations at run time.
 */

#define FIX(x)	((JPEG_INT32) ((x) * CONST_SCALE + 0.5))

/* Descale and correctly round an INT32 value that's scaled by N bits.
 * We assume RIGHT_SHIFT rounds towards minus infinity, so adding
 * the fudge factor is correct for either sign of X.
 */

#define DESCALE(x,n)  RIGHT_SHIFT((x) + (ONE << ((n)-1)), n)

/* Multiply an INT32 variable by an INT32 constant to yield an INT32 result.
 * This macro is used only when the two inputs will actually be no more than
 * 16 bits wide, so that a 16x16->32 bit multiply can be used instead of a
 * full 32x32 multiply.  This provides a useful speedup on many machines.
 * Unfortunately there is no way to specify a 16x16->32 multiply portably
 * in C, but some C compilers will do the right thing if you provide the
 * correct combination of casts.
 */

#ifdef SHORTxSHORT_32		/* may work if 'int' is 32 bits */
#define MULTIPLY16C16(var,const)  (((INT16) (var)) * ((INT16) (const)))
#endif
#ifdef SHORTxLCONST_32		/* known to work with Microsoft C 6.0 */
#define MULTIPLY16C16(var,const)  (((INT16) (var)) * ((JPEG_INT32) (const)))
#endif

#ifndef MULTIPLY16C16		/* default definition */
#define MULTIPLY16C16(var,const)  ((var) * (const))
#endif

/* Same except both inputs are variables. */

#ifdef SHORTxSHORT_32		/* may work if 'int' is 32 bits */
#define MULTIPLY16V16(var1,var2)  (((INT16) (var1)) * ((INT16) (var2)))
#endif

#ifndef MULTIPLY16V16		/* default definition */
#define MULTIPLY16V16(var1,var2)  ((var1) * (var2))
#endif
Commit	Line	Data
	1	/*
	2	* jdct.h
	3	*
	4	* Copyright (C) 1994-1996, Thomas G. Lane.
	5	* This file is part of the Independent JPEG Group's software.
	6	* For conditions of distribution and use, see the accompanying README file.
	7	*
	8	* This include file contains common declarations for the forward and
	9	* inverse DCT modules. These declarations are private to the DCT managers
	10	* (jcdctmgr.c, jddctmgr.c) and the individual DCT algorithms.
	11	* The individual DCT algorithms are kept in separate files to ease
	12	* machine-dependent tuning (e.g., assembly coding).
	13	*/
	14
	15
	16	/*
	17	* A forward DCT routine is given a pointer to a work area of type DCTELEM[];
	18	* the DCT is to be performed in-place in that buffer. Type DCTELEM is int
	19	* for 8-bit samples, INT32 for 12-bit samples. (NOTE: Floating-point DCT
	20	* implementations use an array of type FAST_FLOAT, instead.)
	21	* The DCT inputs are expected to be signed (range +-CENTERJSAMPLE).
	22	* The DCT outputs are returned scaled up by a factor of 8; they therefore
	23	* have a range of +-8K for 8-bit data, +-128K for 12-bit data. This
	24	* convention improves accuracy in integer implementations and saves some
	25	* work in floating-point ones.
	26	* Quantization of the output coefficients is done by jcdctmgr.c.
	27	*/
	28
	29	#if BITS_IN_JSAMPLE == 8
	30	typedef int DCTELEM; /* 16 or 32 bits is fine */
	31	#else
	32	typedef JPEG_INT32 DCTELEM; /* must have 32 bits */
	33	#endif
	34
	35	typedef JMETHOD(void, forward_DCT_method_ptr, (DCTELEM * data));
	36	typedef JMETHOD(void, float_DCT_method_ptr, (FAST_FLOAT * data));
	37
	38
	39	/*
	40	* An inverse DCT routine is given a pointer to the input JBLOCK and a pointer
	41	* to an output sample array. The routine must dequantize the input data as
	42	* well as perform the IDCT; for dequantization, it uses the multiplier table
	43	* pointed to by compptr->dct_table. The output data is to be placed into the
	44	* sample array starting at a specified column. (Any row offset needed will
	45	* be applied to the array pointer before it is passed to the IDCT code.)
	46	* Note that the number of samples emitted by the IDCT routine is
	47	* DCT_scaled_size * DCT_scaled_size.
	48	*/
	49
	50	/* typedef inverse_DCT_method_ptr is declared in jpegint.h */
	51
	52	/*
	53	* Each IDCT routine has its own ideas about the best dct_table element type.
	54	*/
	55
	56	typedef MULTIPLIER ISLOW_MULT_TYPE; /* short or int, whichever is faster */
	57	#if BITS_IN_JSAMPLE == 8
	58	typedef MULTIPLIER IFAST_MULT_TYPE; /* 16 bits is OK, use short if faster */
	59	#define IFAST_SCALE_BITS 2 /* fractional bits in scale factors */
	60	#else
	61	typedef JPEG_INT32 IFAST_MULT_TYPE; /* need 32 bits for scaled quantizers */
	62	#define IFAST_SCALE_BITS 13 /* fractional bits in scale factors */
	63	#endif
	64	typedef FAST_FLOAT FLOAT_MULT_TYPE; /* preferred floating type */
	65
	66
	67	/*
	68	* Each IDCT routine is responsible for range-limiting its results and
	69	* converting them to unsigned form (0..MAXJSAMPLE). The raw outputs could
	70	* be quite far out of range if the input data is corrupt, so a bulletproof
	71	* range-limiting step is required. We use a mask-and-table-lookup method
	72	* to do the combined operations quickly. See the comments with
	73	* prepare_range_limit_table (in jdmaster.c) for more info.
	74	*/
	75
	76	#define IDCT_range_limit(cinfo) ((cinfo)->sample_range_limit + CENTERJSAMPLE)
	77
	78	#define RANGE_MASK (MAXJSAMPLE * 4 + 3) /* 2 bits wider than legal samples */
	79
	80
	81	/* Short forms of external names for systems with brain-damaged linkers. */
	82
	83	#ifdef NEED_SHORT_EXTERNAL_NAMES
	84	#define jpeg_fdct_islow jFDislow
	85	#define jpeg_fdct_ifast jFDifast
	86	#define jpeg_fdct_float jFDfloat
	87	#define jpeg_idct_islow jRDislow
	88	#define jpeg_idct_ifast jRDifast
	89	#define jpeg_idct_float jRDfloat
	90	#define jpeg_idct_4x4 jRD4x4
	91	#define jpeg_idct_2x2 jRD2x2
	92	#define jpeg_idct_1x1 jRD1x1
	93	#endif /* NEED_SHORT_EXTERNAL_NAMES */
	94
	95	/* Extern declarations for the forward and inverse DCT routines. */
	96
	97	EXTERN(void) jpeg_fdct_islow JPP((DCTELEM * data));
	98	EXTERN(void) jpeg_fdct_ifast JPP((DCTELEM * data));
	99	EXTERN(void) jpeg_fdct_float JPP((FAST_FLOAT * data));
	100
	101	EXTERN(void) jpeg_idct_islow
	102	JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
	103	JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
	104	EXTERN(void) jpeg_idct_ifast
	105	JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
	106	JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
	107	EXTERN(void) jpeg_idct_float
	108	JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
	109	JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
	110	EXTERN(void) jpeg_idct_4x4
	111	JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
	112	JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
	113	EXTERN(void) jpeg_idct_2x2
	114	JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
	115	JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
	116	EXTERN(void) jpeg_idct_1x1
	117	JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
	118	JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
	119
	120
	121	/*
	122	* Macros for handling fixed-point arithmetic; these are used by many
	123	* but not all of the DCT/IDCT modules.
	124	*
	125	* All values are expected to be of type INT32.
	126	* Fractional constants are scaled left by CONST_BITS bits.
	127	* CONST_BITS is defined within each module using these macros,
	128	* and may differ from one module to the next.
	129	*/
	130
	131	#define ONE ((JPEG_INT32) 1)
	132	#define CONST_SCALE (ONE << CONST_BITS)
	133
	134	/* Convert a positive real constant to an integer scaled by CONST_SCALE.
	135	* Caution: some C compilers fail to reduce "FIX(constant)" at compile time,
	136	* thus causing a lot of useless floating-point operations at run time.
	137	*/
	138
	139	#define FIX(x) ((JPEG_INT32) ((x) * CONST_SCALE + 0.5))
	140
	141	/* Descale and correctly round an INT32 value that's scaled by N bits.
	142	* We assume RIGHT_SHIFT rounds towards minus infinity, so adding
	143	* the fudge factor is correct for either sign of X.
	144	*/
	145
	146	#define DESCALE(x,n) RIGHT_SHIFT((x) + (ONE << ((n)-1)), n)
	147
	148	/* Multiply an INT32 variable by an INT32 constant to yield an INT32 result.
	149	* This macro is used only when the two inputs will actually be no more than
	150	* 16 bits wide, so that a 16x16->32 bit multiply can be used instead of a
	151	* full 32x32 multiply. This provides a useful speedup on many machines.
	152	* Unfortunately there is no way to specify a 16x16->32 multiply portably
	153	* in C, but some C compilers will do the right thing if you provide the
	154	* correct combination of casts.
	155	*/
	156
	157	#ifdef SHORTxSHORT_32 /* may work if 'int' is 32 bits */
	158	#define MULTIPLY16C16(var,const) (((INT16) (var)) * ((INT16) (const)))
	159	#endif
	160	#ifdef SHORTxLCONST_32 /* known to work with Microsoft C 6.0 */
	161	#define MULTIPLY16C16(var,const) (((INT16) (var)) * ((JPEG_INT32) (const)))
	162	#endif
	163
	164	#ifndef MULTIPLY16C16 /* default definition */
	165	#define MULTIPLY16C16(var,const) ((var) * (const))
	166	#endif
	167
	168	/* Same except both inputs are variables. */
	169
	170	#ifdef SHORTxSHORT_32 /* may work if 'int' is 32 bits */
	171	#define MULTIPLY16V16(var1,var2) (((INT16) (var1)) * ((INT16) (var2)))
	172	#endif
	173
	174	#ifndef MULTIPLY16V16 /* default definition */
	175	#define MULTIPLY16V16(var1,var2) ((var1) * (var2))
	176	#endif