add VIPS_IMAGE_N_PELS()

a macro to calculate the number of pels in an image, in 64 bits

TODO

@@ -1,13 +1,19 @@
+- why is the cache operation only? it'd be useful to cache things like
+  interpolate as well, wouldn't it?
+
+- perhaps Xsize and Ysize should become gint64
+
+  eg. easy to make an image that's 3gb and has xsize == 1 ... death!
+
+  also bands I guess
+
+  now everything that iterates over a dimension must be gint64 too
+
+
+
 - Vips.Image has members like chain, __subclasshook__ etc etc, are we
   really subclassing it correctly?
 
-- have this warning:
-
-im_histspec.c: In function 'im_histspec':
-im_histspec.c:101:6: warning: 'px' may be used uninitialized in this function
-[-Wuninitialized]
-im_histspec.c:79:6: note: 'px' was declared here:n ../his	
-
 - add support for constants
 
   how do boxed types work? confusing

libvips/convolution/im_spcor.c

@@ -99,7 +99,7 @@ typedef struct {
 			sum1 += a1[i]; \
 		a1 += in_lsk;  \
 	} \
-	imean = sum1 / (ref->Xsize * ref->Ysize); \
+	imean = sum1 / VIPS_IMAGE_N_PELS( ref ); \
  	\
 	/* Loop over ir again, this time calculating  \
 	 * sum-of-squares-of-differences for this window on \
@@ -202,7 +202,7 @@ static Spcor *
 spcor_new( IMAGE *out, IMAGE *ref )
 {
 	Spcor *spcor;
-	size_t sz = ref->Xsize * ref->Ysize;
+	guint64 sz = VIPS_IMAGE_N_PELS( ref );
 	VipsPel *p = ref->data;
 	double s;
 	size_t i;

libvips/foreign/matlab.c

@@ -229,7 +229,7 @@ mat2vips_get_data( mat_t *mat, matvar_t *var, VipsImage *im )
 	/* Matlab images are plane-separate, so we have to assemble bands in
 	 * image-size chunks.
 	 */
-	const int is = es * im->Xsize * im->Ysize;
+	const guint64 is = es * VIPS_IMAGE_N_PELS( im );
 
 	if( Mat_VarReadDataAll( mat, var ) ) {
 		vips_error( "mat2vips", "%s", 

libvips/freq_filt/im_fwfft.c

@@ -85,7 +85,7 @@
 static int 
 rfwfft1( IMAGE *dummy, IMAGE *in, IMAGE *out )
 {
-	const int size = in->Xsize * in->Ysize;
+	const guint64 size = VIPS_IMAGE_N_PELS( in ); 
 	const int half_width = in->Xsize / 2 + 1;
 
 	/* Pack to double real here.
@@ -241,7 +241,7 @@ cfwfft1( IMAGE *dummy, IMAGE *in, IMAGE *out )
 	/* Copy to out, normalise.
 	 */
 	for( p = (double *) cmplx->data, y = 0; y < out->Ysize; y++ ) {
-		int size = out->Xsize * out->Ysize;
+		guint64 size = VIPS_IMAGE_N_PELS( out );
 
 		q = buf;
 
@@ -274,7 +274,7 @@ fwfft1( IMAGE *dummy, IMAGE *in, IMAGE *out )
 static int 
 rfwfft1( IMAGE *dummy, IMAGE *in, IMAGE *out )
 {
-	const int size = in->Xsize * in->Ysize;
+	const guint64 size = VIPS_IMAGE_N_PELS( in );
 	const int half_width = in->Xsize / 2 + 1;
 
 	/* Pack to double real here.
@@ -289,7 +289,7 @@ rfwfft1( IMAGE *dummy, IMAGE *in, IMAGE *out )
 	/* We have to have a separate real buffer for the planner to work on.
 	 */
 	double *planner_scratch = IM_ARRAY( dummy, 
-		in->Xsize * in->Ysize, double );
+		VIPS_IMAGE_N_PELS( in ), double );
 
 	fftw_plan plan;
 	double *buf, *q, *p;
@@ -403,7 +403,7 @@ cfwfft1( IMAGE *dummy, IMAGE *in, IMAGE *out )
 	/* We have to have a separate buffer for the planner to work on.
 	 */
 	double *planner_scratch = IM_ARRAY( dummy, 
-		in->Xsize * in->Ysize * 2, double );
+		VIPS_IMAGE_N_PELS( in ) * 2, double );
 
 	/* Make dp complex image.
 	 */
@@ -448,7 +448,7 @@ cfwfft1( IMAGE *dummy, IMAGE *in, IMAGE *out )
 	/* Copy to out, normalise.
 	 */
 	for( p = (double *) cmplx->data, y = 0; y < out->Ysize; y++ ) {
-		int size = out->Xsize * out->Ysize;
+		guint64 size = VIPS_IMAGE_N_PELS( out );
 
 		q = buf;
 
@@ -480,7 +480,7 @@ fwfft1( IMAGE *dummy, IMAGE *in, IMAGE *out )
 static int 
 fwfft1( IMAGE *dummy, IMAGE *in, IMAGE *out )
 {
-	int size = in->Xsize * in->Ysize;
+	int size = VIPS_IMAGE_N_PELS( in );
 	int bpx = im_ispoweroftwo( in->Xsize );
 	int bpy = im_ispoweroftwo( in->Ysize );
 	float *buf, *q, *p1, *p2;

libvips/freq_filt/im_invfft.c

@@ -125,7 +125,7 @@ invfft1( IMAGE *dummy, IMAGE *in, IMAGE *out )
 	/* We have to have a separate buffer for the planner to work on.
 	 */
 	double *planner_scratch = IM_ARRAY( dummy, 
-		in->Xsize * in->Ysize * 2, double );
+		VIPS_IMAGE_N_PELS( in ) * 2, double );
 
 	/* Make dp complex image.
 	 */

libvips/freq_filt/im_invfftr.c

@@ -100,7 +100,7 @@ invfft1( IMAGE *dummy, IMAGE *in, IMAGE *out )
 	 */
 	q = half_complex;
 	for( y = 0; y < cmplx->Ysize; y++ ) {
-		p = ((double *) cmplx->data) + y * in->Xsize * 2; 
+		p = ((double *) cmplx->data) + (guint64) y * in->Xsize * 2; 
 
 		for( x = 0; x < half_width; x++ ) {
 			q[0] = p[0];
@@ -182,7 +182,7 @@ invfft1( IMAGE *dummy, IMAGE *in, IMAGE *out )
 	 */
 	q = half_complex;
 	for( y = 0; y < cmplx->Ysize; y++ ) {
-		p = ((double *) cmplx->data) + y * in->Xsize * 2; 
+		p = ((double *) cmplx->data) + (guint64) y * in->Xsize * 2; 
 
 		for( x = 0; x < half_width; x++ ) {
 			q[0] = p[0];

libvips/histograms_lut/im_histeq.c

@@ -85,13 +85,13 @@
 int 
 im_histcum( IMAGE *in, IMAGE *out )
 {
-	const int px = in->Xsize * in->Ysize;
+	const guint64 px = VIPS_IMAGE_N_PELS( in );
 	const int nb = vips_bandfmt_iscomplex( in->BandFmt ) ? 
 		in->Bands * 2 : in->Bands;
-	const int mx = px * nb;
+	const guint64 mx = px * nb;
 
 	VipsPel *outbuf;		
-	int b, x;
+	guint64 b, x;
 
 	if( im_check_uncoded( "im_histcum", in ) ||
 		im_check_hist( "im_histcum", in ) ||
@@ -159,7 +159,7 @@ im_histcum( IMAGE *in, IMAGE *out )
 int 
 im_histnorm( IMAGE *in, IMAGE *out )
 {
-	const int px = in->Xsize * in->Ysize;
+	const guint64 px = VIPS_IMAGE_N_PELS( in );
 	DOUBLEMASK *stats;
 	double *a, *b;
 	int i;

libvips/histograms_lut/im_histspec.c

@@ -64,8 +64,8 @@
 static int
 match( IMAGE *in, IMAGE *ref, IMAGE *out )
 {
-	const guint64 inpx = (guint64) in->Xsize * in->Ysize;
+	const guint64 inpx = VIPS_IMAGE_N_PELS( in );
-	const guint64 refpx = (guint64) ref->Xsize * ref->Ysize;
+	const guint64 refpx = VIPS_IMAGE_N_PELS( ref );
 	const int bands = in->Bands;	
 
 	unsigned int *inbuf;		/* in and ref, padded to same size */
@@ -175,7 +175,7 @@ int
 im_histspec( IMAGE *in, IMAGE *ref, IMAGE *out )
 {
 	IMAGE *t[5];
-	int px;
+	guint64 px;
 	int fmt;
 
 	if( im_check_uint( "im_histspec", in ) ||
@@ -194,7 +194,7 @@ im_histspec( IMAGE *in, IMAGE *ref, IMAGE *out )
 
 	/* Clip type down.
 	 */
-	px = t[4]->Xsize * t[4]->Ysize;
+	px = VIPS_IMAGE_N_PELS( t[4] );
 	if( px <= 256 ) 
 		fmt = IM_BANDFMT_UCHAR;
 	else if( px <= 65536 ) 

libvips/include/vips/image.h

@@ -429,6 +429,8 @@ extern const guint64 vips__image_sizeof_bandformat[];
 	(VIPS_IMAGE_SIZEOF_LINE( I ) * (I)->Ysize)
 #define VIPS_IMAGE_N_ELEMENTS( I ) \
 	((I)->Bands * (I)->Xsize)
+#define VIPS_IMAGE_N_PELS( I ) \
+	((guint64) (I)->Xsize * (I)->Ysize)
 
 /* If VIPS_DEBUG is defined, add bounds checking.
  */

libvips/iofuncs/error.c

@@ -1089,7 +1089,7 @@ vips_check_hist( const char *domain, VipsImage *im )
 			_( "histograms must have width or height 1" ) );
 		return( -1 );
 	}
-	if( im->Xsize * im->Ysize > 65536 ) {
+	if( VIPS_IMAGE_N_PELS( im ) > 65536 ) {
 		vips_error( domain, "%s", 
 			_( "histograms must have not have more than "
 				"65536 elements" ) );

libvips/iofuncs/image.c

@@ -126,6 +126,13 @@
  * Returns: The number of band elements in a scanline.
  */
 
+/**
+ * VIPS_IMAGE_N_PELS:
+ * @I: a #VipsImage
+ *
+ * Returns: The number of pels in an image. A 64-bit unsigned int.
+ */
+
 /**
  * VIPS_IMAGE_ADDR:
  * @I: a #VipsImage
@@ -1086,7 +1093,7 @@ vips_progress_add( VipsImage *image )
 	g_timer_start( progress->start );
 	progress->run = 0;
 	progress->eta = 0;
-	progress->tpels = (gint64) image->Xsize * image->Ysize;
+	progress->tpels = VIPS_IMAGE_N_PELS( image );
 	progress->npels = 0;
 	progress->percent = 0;
 

libvips/mosaicing/global_balance.c

@@ -1005,7 +1005,7 @@ count_nonzero( IMAGE *in, gint64 *count )
 
 	if( im_avg( in, &avg ) )
 		return( -1 );
-	*count = (avg * in->Xsize * in->Ysize ) / 255.0;
+	*count = (avg * VIPS_IMAGE_N_PELS( in )) / 255.0;
 	
 	return( 0 );
 }
@@ -1041,8 +1041,7 @@ find_image_stats( IMAGE *in, IMAGE *mask, Rect *area )
 
 	/* And scale masked average to match.
 	 */
-	stats->coeff[4] *= (double) count / 
+	stats->coeff[4] *= (double) count / VIPS_IMAGE_N_PELS( mask );
-		((double) mask->Xsize * mask->Ysize);
 
 	/* Yuk! Zap the deviation column with the pixel count. Used later to
 	 * determine if this is likely to be a significant overlap.