[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[gnuastro-commits] master b0f1ab6c: Library (arithmetic): correctly acco
|
From: |
Mohammad Akhlaghi |
|
Subject: |
[gnuastro-commits] master b0f1ab6c: Library (arithmetic): correctly accounting for all-NaN pixels |
|
Date: |
Fri, 12 Apr 2024 13:48:43 -0400 (EDT) |
branch: master
commit b0f1ab6c704e1ef71b21ecd76435bbce2e779027
Author: Mohammad Akhlaghi <mohammad@akhlaghi.org>
Commit: Mohammad Akhlaghi <mohammad@akhlaghi.org>
Library (arithmetic): correctly accounting for all-NaN pixels
Until now, when a pixel was NaN in all the inputs of the stacking
operators, they would crash with a segmentation fault because of the newly
changed feature in the statistics operators (that would free the input
array's pointer and was necessary for many low-level checks).
With this commit, before giving the pixel value in all the inputs to the
respective statistics operator, we count how many are not blank and only
call the statisitics operators when we actually have data.
---
lib/arithmetic.c | 79 +++++++++++++++++++++++++++++++++-----------------------
1 file changed, 46 insertions(+), 33 deletions(-)
diff --git a/lib/arithmetic.c b/lib/arithmetic.c
index 02e6b6c2..b07dc049 100644
--- a/lib/arithmetic.c
+++ b/lib/arithmetic.c
@@ -1685,6 +1685,42 @@ struct multioperandparams
+#define MULTIOPERAND_PIXS_USE { \
+ int use; \
+ \
+ /* If there is the 'cont->array' is empty, it has been */ \
+ /* mistakenly free'd by the in-place statistics operators! */ \
+ if(cont->array==NULL) \
+ error(EXIT_FAILURE, 0, "%s: a bug! Please contact us at " \
+ "'%s' to fix the problem. The 'cont' array should " \
+ "not be NULL", __func__, PACKAGE_BUGREPORT); \
+ \
+ /* Loop over each array, 'i' is input dataset's index, 'n' is */ \
+ /* the number of useful values to use. */ \
+ n=0; \
+ for(i=0;i<p->dnum;++i) \
+ { \
+ /* Only integers and non-NaN floats: v==v is 1. */ \
+ if(p->hasblank[i]) \
+ use = ( b==b \
+ ? ( a[i][j]!=b ? 1 : 0 ) /* Integer */ \
+ : ( a[i][j]==a[i][j] ? 1 : 0 ) ); /* Float */ \
+ else use=1; \
+ \
+ /* Put the value into the array of values. */ \
+ if(use) pixs[n++]=a[i][j]; \
+ } \
+ \
+ /* Since later operations are done in place, the size and flags */ \
+ /* need to be reset. */ \
+ cont->flag=0; \
+ cont->size=cont->dsize[0]=n; \
+ }
+
+
+
+
+
#define MULTIOPERAND_MIN(TYPE) { \
size_t n, j=0; \
TYPE t, max, *o=p->out->array; \
@@ -1942,12 +1978,9 @@ struct multioperandparams
/* Go over all the pixels assigned to this thread. */ \
for(tind=0; tprm->indexs[tind] != GAL_BLANK_SIZE_T; ++tind) \
{ \
- /* Initialize, 'j' is desired pixel's index. */ \
- n=0; \
- j=tprm->indexs[tind]; \
- \
- /* Read the necessay values from each input. */ \
- for(i=0;i<p->dnum;++i) pixs[n++]=a[i][j]; \
+ /* Initialize and set the pixels to use in the macro. */ \
+ j=tprm->indexs[tind]; /* desired pixel's index */ \
+ MULTIOPERAND_PIXS_USE; \
\
/* If there are any elements, do the measurement. */ \
if(n) \
@@ -1957,11 +1990,6 @@ struct multioperandparams
memcpy(&o[j], quantile->array, \
gal_type_sizeof(p->list->type)); \
gal_data_free(quantile); \
- \
- /* Since we are doing sigma-clipping in place, the size, */ \
- /* and flags need to be reset. */ \
- cont->flag=0; \
- cont->size=cont->dsize[0]=p->dnum; \
} \
else \
o[j]=b; \
@@ -1987,12 +2015,9 @@ struct multioperandparams
/* Go over all the pixels assigned to this thread. */ \
for(tind=0; tprm->indexs[tind] != GAL_BLANK_SIZE_T; ++tind) \
{ \
- /* Initialize, 'j' is desired pixel's index. */ \
- n=0; \
- j=tprm->indexs[tind]; \
- \
- /* Read the necessay values from each input. */ \
- for(i=0;i<p->dnum;++i) pixs[n++]=a[i][j]; \
+ /* Initialize and set the pixels to use in the macro. */ \
+ j=tprm->indexs[tind]; /* desired pixel's index */ \
+ MULTIOPERAND_PIXS_USE; \
\
/* If there are any elements, do the measurement. */ \
if(n) \
@@ -2002,11 +2027,6 @@ struct multioperandparams
mad=gal_data_copy_to_new_type_free(mad, GAL_TYPE_FLOAT32); \
memcpy(&o[j], mad->array, gal_type_sizeof(GAL_TYPE_FLOAT32)); \
gal_data_free(mad); \
- \
- /* Since we are sorting in place, the size, and flags */ \
- /* need to be reset. */ \
- cont->flag=0; \
- cont->size=cont->dsize[0]=p->dnum; \
} \
else \
o[j]=NAN; \
@@ -2058,6 +2078,7 @@ arithmetic_multioperand_number_write(struct
multioperandparams *p,
+
#define MULTIOPERAND_CLIP(TYPE, SIG1_MAD0) { \
size_t n, j; \
gal_data_t *clip; \
@@ -2072,12 +2093,9 @@ arithmetic_multioperand_number_write(struct
multioperandparams *p,
/* Go over all the pixels assigned to this thread. */ \
for(tind=0; tprm->indexs[tind] != GAL_BLANK_SIZE_T; ++tind) \
{ \
- /* Initialize, 'j' is desired pixel's index. */ \
- n=0; \
- j=tprm->indexs[tind]; \
- \
- /* Put the values from each input into a single array. */ \
- for(i=0;i<p->dnum;++i) pixs[n++]=a[i][j]; \
+ /* Initialize and set the pixels to use in the macro. */ \
+ j=tprm->indexs[tind]; /* desired pixel's index */ \
+ MULTIOPERAND_PIXS_USE; \
\
/* If there are any usable elements, do the measurement. */ \
if(n) \
@@ -2127,11 +2145,6 @@ arithmetic_multioperand_number_write(struct
multioperandparams *p,
\
/* Free the output of the clipping. */ \
gal_data_free(clip); \
- \
- /* Since we are doing sigma-clipping in place, the size, */ \
- /* and flags need to be reset. */ \
- cont->flag=0; \
- cont->size=cont->dsize[0]=p->dnum; \
} \
\
/* No usable elements. */ \
| [Prev in Thread] |
Current Thread |
[Next in Thread] |
- [gnuastro-commits] master b0f1ab6c: Library (arithmetic): correctly accounting for all-NaN pixels,
Mohammad Akhlaghi <=