- cffi-devel - mailman3.common-lisp.net

[cffi-devel] Help with defcallback
by Joeish W 11 Nov '13

11 Nov '13

I think the issue is with my callback function but not 100% when i run the c code and print with these 2 lines(from c callback): cout << "a = " << "(" << "" << "," << _a << ")" << endl ; cout << "b = " << "(" << "" << "," << _b << ")" << endl; i get 53 elements printed not icluding the other printf output and in the lisp code i converterd when i print with this line(from lisp defcallback): (format t "~%x~a~%b~a" _a _b) in the defcallback i get 114 elements Hoping someone can look at my code and tell me if my coding is off ....the functions you wont recognize are opencv functions with correctly lispified names i/e cvSeqSort = seq-sort ... If requested will post the wrappers .... the -> is a macro for with-foreign-slots (and if not a pointer just a getf for a plist) ;;;;;;;;C CODE;;;;;;;;;;;;;;;;;;; #include <cv.h> #include <highgui.h> using namespace std; static int cmp_func( const void* _a, const void* _b, void* userdata ) { //cout << "test"; //cout <<_a << _b; CvPoint* a = (CvPoint*)_a; CvPoint* b = (CvPoint*)_b; cout << "a = " << "(" << "" << "," << _a << ")" << endl ; cout << "b = " << "(" << "" << "," << _b << ")" << endl; int y_diff = a->y - b->y; int x_diff = a->x - b->x; return y_diff ? y_diff : x_diff; } int main() { CvMemStorage* storage = cvCreateMemStorage(0); CvSeq* seq = cvCreateSeq( CV_32SC2, sizeof(CvSeq), sizeof(CvPoint), storage ); int i; printf("\n=== Test sequence sorting ==="); for( i= 0; i < 10; i++ ) { CvPoint pt; pt.x = rand() % 1000; // 1000 以内的随机数 pt.y = rand() % 1000; cvSeqPush( seq, &pt ); } printf("\nOriginal point set:\n"); for( i = 0; i < seq->total; i++ ) { CvPoint* pt = (CvPoint*)cvGetSeqElem( seq, i ); printf( "(%d,%d)\n", pt->x, pt->y ); } cvSeqSort( seq, cmp_func, 0 /* userdata is not used here */ ); /* print out the sorted sequence */ printf("\nAfter sorting:\n"); for( i = 0; i < seq->total; i++ ) { CvPoint* pt = (CvPoint*)cvGetSeqElem( seq, i ); printf( "(%d,%d)\n", pt->x, pt->y ); } cvClearSeq( seq ); // Sequence clearing should be done before storage clearing cvReleaseMemStorage( &storage ); } ;;;;;;;;;;;;;;;;;;;;;;;;;LISP CODE;;;;;;;;;;;;;;;;;;;; (defcallback cmp-func :void ((_a :pointer) (_b :pointer) (user-data :pointer) (a :int) (b :int) (y-diff :int) (x-diff :int)) (format t "~%x~a~%b~a" _a _b) (setf a (mem-aref _a '(:struct cv-point))) (setf b (mem-aref _b '(:struct cv-point))) ;(format t "~%a~a~%b~a" a b) (setf y-diff (- (-> cv-point a y) (-> cv-point b y))) (setf x-diff (- (-> cv-point a x) (-> cv-point b x))) (if y-diff y-diff x-diff)) (defun seq-sort-example () (let* ((storage (create-mem-storage 0)) (seq (create-seq +32sc2+ (size-of cv-seq) (size-of cv-point) storage))) (format t "~%=== Test sequence sorting ===") (with-foreign-object (pt '(:struct cv-point)) (dotimes (i 10) (setf (foreign-slot-value pt '(:struct cv-point) 'x) (random 1000) (foreign-slot-value pt '(:struct cv-point) 'y) (random 1000)) (seq-push seq pt))) (with-foreign-object (pt '(:pointer (:struct cv-point))) (format t "~%Original point set:~%") (dotimes (i (-> cv-seq seq total)) (setf pt (get-seq-elem seq i)) (format t "(~a, ~a)~%" (-> cv-point pt x) (-> cv-point pt y) ))) (seq-sort seq (callback cmp-func) (null-pointer)) (with-foreign-object (pt '(:pointer (:struct cv-point))) (format t "~%After sorting:~%") (dotimes (i (-> cv-seq seq total)) (setf pt (get-seq-elem seq i)) (format t "(~a, ~a)~%" (-> cv-point pt x) (-> cv-point pt y) ))) (clear-seq seq) (release-mem-storage storage)))

1 0

[cffi-devel] Fw: Struct member changes when its assigned back to variable it was dereferenced from...do same in lisp fails
by Joeish W 03 Nov '13

03 Nov '13

On Saturday, November 2, 2013 10:44 PM, Joeish W <joeish80829(a)yahoo.com> wrote: here is the c code with explanation didn't add all code to keep it short but i know it is still long...if this seems daunting because all the code I added....pls know 1. I worked on this a long time and added the minimal code i could think to explain the issue 2. if it does seem daunting...if you would be so kind as to look through this and tell me how i could edit my ? to make it easier to answer...with that being said... though this is a very simple post .... Im hoping at the end of the explanation you might be able to help me re-wrap my CV-SEQ struct so i get the same resultt as i do in c > > >CvSeq* contours; //hold the pointer to a contour in the memory block > > > >// adds data to CvSeq* contours with &contours call >cvFindContours(imgGrayScale, storage, &contours, sizeof(CvContour), CV_RETR_LIST, CV_CHAIN_APPROX_SIMPLE, cvPoint(0,0)); > > >// This is how its supposed to work ....no matter how many times you dereference contours to get the h_next struct member >// the value of contours->h_next is always the same > > >cout << "c1 = " << endl << " " << contours->h_next << endl << endl; //output same as next line - see output below >cout << "c2 = " << endl << " " << contours->h_next << endl << endl; //output same as above line see output below > > >// If you assign contours->h_next to a new variable "c" > >CvSeq* c = contours->h_next; > > >// then cout contours->h_next the output of the below 2 cout's is still the same as the output of the above 2 - see output below > >cout << "c3 = " << endl << " " << contours->h_next << endl << endl; >cout << "c4 = " << endl << " " << contours->h_next << endl << endl; > > >// but if you assign contours->h_next back to contours the output of the next 2 couts is different than above 4 > >// see output below > > contours = contours->h_next; >cout << "c5 = " << endl << " " << contours->h_next << endl << endl; >cout << "c6 = " << endl << " " << contours->h_next << endl << endl; > > > > >here is the cout output > > >c1 = > 0x182f730 > > >c2 = > 0x182f730 > >c3 = > 0x182f730 > >c4 = > 0x182f730 > >c5 = > 0x182f560 > >c6 = > 0x182f560 > > > > >The struct CvSeq is defined as this > >#define CV_TREE_NODE_FIELDS(node_type) \ > int flags; /* Miscellaneous flags. */ \ > int header_size; /* Size of sequence header. */ \ > struct node_type* h_prev; /* Previous sequence. */ \ > struct node_type* h_next; /* Next sequence. */ \ > struct node_type* v_prev; /* 2nd previous sequence. */ \ > struct node_type* v_next /* 2nd next sequence. */ > >/* > Read/Write sequence. > Elements can be dynamically inserted to or deleted from the sequence. >*/ >#define CV_SEQUENCE_FIELDS() \ > CV_TREE_NODE_FIELDS(CvSeq); \ > int total; /* Total number of elements. */ \ > int elem_size; /* Size of sequence element in bytes. */ \ > schar* block_max; /* Maximal bound of the last block. */ \ > schar* ptr; /* Current write pointer. */ \ > int delta_elems; /* Grow seq this many at a time. */ \ > CvMemStorage* storage; /* Where the seq is stored. */ \ > CvSeqBlock* free_blocks; /* Free blocks list. */ \ > CvSeqBlock* first; /* Pointer to the first sequence block. */ > >typedef struct CvSeq >{ > CV_SEQUENCE_FIELDS() >} >CvSeq; > > > >Im not getting the same thing as in lisp with my struct like this - SWIG wrapped > > >(cffi:defcstruct cv-seq > (flags :int) > (header-size :int) > (h-prev :pointer) > (h-next :pointer) > (v-prev :pointer) > (v-next :pointer) > (total :int) > (elem-size :int) > (block-max :pointer) > (ptr :pointer) > (delta-elems :int) > (storage :pointer) > (free-blocks :pointer) > (first :pointer)) > > > > >with the below code my output is the same for both princ...not newline of course =) > > > >(with-foreign-object (contours '(:pointer (:struct cv-seq))) > (find-contours img-grayscale storage contours > (size-of cv-contour) +retr-list+ > +chain-approx-simple+ (point 0 0)) > > (with-foreign-slots ((h-next) contours (:struct cv-seq)) > (setf contours h-next) > (princ contours) > (princ #\newline) > (setf contours h-next) > (princ contours))) > > >any help on fixing my struct, my coding or editing this post =) is much appreciated > > >Cheers! > > >

1 0

RE: [cffi-devel] There doesn't seem to be anything in CFFI doc. about wrapping macros
by Joeish W 03 Nov '13

03 Nov '13

ok i do appreciate you help so far but 1 more thing....i get everything else you said what do i cast the return value to for CV_GET_SEQ_ELEM because the documentation for it says...."The macro checks first whether the desired element belongs to the first block of the sequence and returns it if it does; otherwise the macro calls the main function GetSeqElem" tthe documentation for cvGetSeqElem says "Returns a pointer to a sequence element according to its index." so the macro seems to have ability to return 2 separate values but the function cvGetseqElem does not....cvGetSeqElems return is usually cast to other types when its used though....I was hoping you could help me figure out how to cast its return and how to get by the parameters not being declared here i/e to int or double etc #define CV_GET_SEQ_ELEM( elem_type, seq, index ) CV_SEQ_ELEM( (seq), elem_type, (index) ) I know what these variables are i/e elem_type is an int and sq is a CvSeq*, index is an int....but i have alot glue code to write for macros like this....so if i didnt know what the variables should be declared as ....what do i do.....this will help me master this part of writing my library(gloe code for macros) if you could answer this for me On Saturday, November 2, 2013 1:14 PM, Frank Gönninger <frank.goenninger(a)consequor.de> wrote: A C macro is merely a text replacement, not a function >definition/declaration. So you can not "call" it - neither from C nor from >Lisp. > >In every place in C code where CV_SEQ_ELEM() appears the C preprocessor >replaces this *text* with what the macro generates. > >Not knowing what this macro really does it will replace > >CV_SEQ_ELEM(myseq,int,1) > >with > >( assert(sizeof((mytype)->first[0]) == sizeof(CvSeqBlock) && > (mytype)->elem_size == sizeof(int)), (int*)((mytype)->first && >(unsigned)1 < (unsigned)((mytype)->first->count) ? (mytype)->first->data + >(1) * sizeof(int) : cvGetSeqElem( (CvSeq*)(mytype), (1) ))) > >You see, a simple text replacement. > >So - all you need to do is to write a C layer that has a function around >this. If you want to do this in Lisp, then you need to manually translate >this into Lisp. You need to have a solid understanding of C and a solid >understanding of Lisp to do that. > >Cheers > Frank > >-- > C o n s e q u o r C o n s u l t i n g A G > Frank Gönninger > > Mobil: +49 175 43 21 058 > E-Mail: frank.goenninger(a)consequor.de > Telefon: +49 711 781181 10 > Fax: +49 711 781181 11 > > Consequor Consulting AG > Liebknechtstr. 33 > D-70565 Stuttgart, Germany > > Vorstand: Frank Gönninger > Aufsichtsratsvorsitzender: Dipl.-Kfm. Matthias Filbinger > Sitz der Gesellschaft: 70565 Stuttgart, Deutschland > Registergericht Amtsgericht Stuttgart HRB 727446 > > > > > > > > >Am 02.11.13 19:54 schrieb "Joeish W" unter <joeish80829(a)yahoo.com>: > >>well I have to do a glue layer for it any way because the macro gets an >>unidentified alien function error when I try to call it from a defcfun >>in lisp...I write alot of glue code for unidentified alien function >>errors but those are for functions not macros. could you help get me >> started writing a glue layer for this macro i posted so i can call it in >>a defcfun for c....i have no idea how to type cast the parameters or how >>to do a return on them...btw .sorry if i sent a reply to your mailbox >>Frank =) >> >> >> >> >> On Saturday, November 2, 2013 9:23 AM, Frank Gönninger >><frank.goenninger(a)consequor.de> wrote: >> >> >> This is a simple code generation C macro, so this is simple to be >>transcripted into Common Lisp. There are several alternatives to do this. >> >>Either you write the code explicitly in every use occasion or you write a >>Common Lisp macro that does the same. >> >>(defmacro cv-seq-elem( seq elem_type index ) >> `( >> Š >> >>But why do you what to do this in >> Lisp? I would rather write some glue >>layer in C and have much simpler calls from Lisp to C. >> >>Good luck. >> >>Frank >> >> >>Am 02.11.13 16:43 schrieb "Joeish W" unter <joeish80829(a)yahoo.com>: >> >>> >>>I'm trying to wrap the GET_SEQ_ELEM macro below can some one give me >>>clues or example how i would go about this easily >>> >>>with a cffi function like defcfun...If there is no clear way i definately >>>propose an addendum to cffi for wrapping macros >>> >>> >>>#define CV_SEQ_ELEM( seq, elem_type, index ) >> \ >>> >>>/* assert gives some guarantee that <seq> parameter is valid */ \ >>> >>>( assert(sizeof((seq)->first[0]) == sizeof(CvSeqBlock) && \ >>> >>> (seq)->elem_size == sizeof(elem_type)), \ >>> >>> (elem_type*)((seq)->first && (unsigned)index >>> < \ >>> >>> (unsigned)((seq)->first->count) ? \ >>> >>> (seq)->first->data + (index) * >> sizeof(elem_type) : \ >>> >>> cvGetSeqElem( (CvSeq*)(seq), (index) ))) >>> >>>#define CV_GET_SEQ_ELEM( elem_type, seq, index ) CV_SEQ_ELEM( (seq), >>>elem_type, (index) ) >> >> >> >> >> >> >> >> >> >> >> >> > > >

1 0

Re: [cffi-devel] There doesn't seem to be anything in CFFI doc. about wrapping macros
by Joeish W 02 Nov '13

02 Nov '13

well I have to do a glue layer for it any way because the macro gets an unidentified alien function error when I try to call it from a defcfun in lisp...I write alot of glue code for unidentified alien function errors but those are for functions not macros. could you help get me started writing a glue layer for this macro i posted so i can call it in a defcfun for c....i have no idea how to type cast the parameters or how to do a return on them...btw .sorry if i sent a reply to your mailbox Frank =) On Saturday, November 2, 2013 9:23 AM, Frank Gönninger <frank.goenninger(a)consequor.de> wrote: This is a simple code generation C macro, so this is simple to be >transcripted into Common Lisp. There are several alternatives to do this. > >Either you write the code explicitly in every use occasion or you write a >Common Lisp macro that does the same. > >(defmacro cv-seq-elem( seq elem_type index ) > `( Š > >But why do you what to do this in Lisp? I would rather write some glue >layer in C and have much simpler calls from Lisp to C. > >Good luck. > >Frank > > > >Am 02.11.13 16:43 schrieb "Joeish W" unter <joeish80829(a)yahoo.com>: > >> >>I'm trying to wrap the GET_SEQ_ELEM macro below can some one give me >>clues or example how i would go about this easily >> >>with a cffi function like defcfun...If there is no clear way i definately >>propose an addendum to cffi for wrapping macros >> >> >>#define CV_SEQ_ELEM( seq, elem_type, index ) \ >> >>/* assert gives some guarantee that <seq> parameter is valid */ \ >> >>( assert(sizeof((seq)->first[0]) == sizeof(CvSeqBlock) && \ >> >> (seq)->elem_size == sizeof(elem_type)), \ >> >> (elem_type*)((seq)->first && (unsigned)index >> < \ >> >> (unsigned)((seq)->first->count) ? \ >> >> (seq)->first->data + (index) * sizeof(elem_type) : \ >> >> cvGetSeqElem( (CvSeq*)(seq), (index) ))) >> >>#define CV_GET_SEQ_ELEM( elem_type, seq, index ) CV_SEQ_ELEM( (seq), >>elem_type, (index) ) > > >

1 0

[cffi-devel] There doesn't seem to be anything in CFFI doc. about wrapping macros
by Joeish W 02 Nov '13

02 Nov '13

I'm trying to wrap the GET_SEQ_ELEM macro below can some one give me clues or example how i would go about this easily with a cffi function like defcfun...If there is no clear way i definately propose an addendum to cffi for wrapping macros #define CV_SEQ_ELEM( seq, elem_type, index ) \ /* assert gives some guarantee that <seq> parameter is valid */ \ ( assert(sizeof((seq)->first[0]) == sizeof(CvSeqBlock) && \ (seq)->elem_size == sizeof(elem_type)), \ (elem_type*)((seq)->first && (unsigned)index < \ (unsigned)((seq)->first->count) ? \ (seq)->first->data + (index) * sizeof(elem_type) : \ cvGetSeqElem( (CvSeq*)(seq), (index) ))) #define CV_GET_SEQ_ELEM( elem_type, seq, index ) CV_SEQ_ELEM( (seq), elem_type, (index) )

1 0

[cffi-devel] There doesn't seem to be anything in CFFI doc. about wrapping macros
by Joeish W 02 Nov '13

02 Nov '13

I'm trying to wrap the GET_SEQ_ELEM macro below can some one give me clues or example how i would go about this easily with a cffi function like defcfun...If there is no clear way i definately propose an addendum to cffi for wrapping macros #define CV_SEQ_ELEM( seq, elem_type, index ) \ /* assert gives some guarantee that <seq> parameter is valid */ \ ( assert(sizeof((seq)->first[0]) == sizeof(CvSeqBlock) && \ (seq)->elem_size == sizeof(elem_type)), \ (elem_type*)((seq)->first && (unsigned)index < \ (unsigned)((seq)->first->count) ? \ (seq)->first->data + (index) * sizeof(elem_type) : \ cvGetSeqElem( (CvSeq*)(seq), (index) ))) #define CV_GET_SEQ_ELEM( elem_type, seq, index ) CV_SEQ_ELEM( (seq), elem_type, (index) )

1 0

[cffi-devel] Help with recursive struct member
by Joeish W 01 Nov '13

01 Nov '13

here is the way CvMemStorage is defined typedef struct CvMemStorage { int signature; CvMemBlock* bottom; /* First allocated block. */ CvMemBlock* top; /* Current memory block - top of the stack. */ struct CvMemStorage* parent; /* We get new blocks from parent as needed. */ int block_size; /* Block size. */ int free_space; /* Remaining free space in current block. */ } CvMemStorage; bu i get unknown CFFI type: (:STRUCT CV-MEM-STORAGE). [Condition of type SIMPLE-ERROR] when i wrap like this....i know lisp is known for recursion how do i bypass this (defcstruct cv-mem-storage (signature :int) (bottom (:pointer (:struct cv-mem-block))) (top (:pointer (:struct cv-mem-block))) (parent (:pointer (:struct cv-mem-storage))) (block-size :int) (free-space :int))

2 2

[cffi-devel] help writing macro for with-foreign-slots aka help with the ULTIMATE derefrencing macro
by Joeish W 01 Nov '13

01 Nov '13

I'm trying to write an all purpose dereferencing macro...ive done pretty good so far I have (defmacro -> (var1 var2 var3 &optional var4 var5) ;; if var4 and var5 arent supplied give value of struct member (if (not (and var4 var5)) `(cffi:with-foreign-slots ((,var3) ,var2 (:struct ,var1)) (list ,var3)) ;; if var4 and var5 are supplied give value of nested struct member (if (and var4 var5) `(cffi:with-foreign-slots ((,var3) ,var2 (:struct ,var1)) (cffi:with-foreign-slots ((,var5) ,var3 (:struct ,var4)) ,var5) )))) the roi struct member of ipl-image is defined as this (roi (:pointer (:struct ipl-roi))) ;this gets the roi slot value of ipl-image correctly CL-OPENCV> (-> ipl-image a roi ) (#.(SB-SYS:INT-SAP #X7FFFDC000E80)) ;this gets the nested struct member "width" of ipl-roi which is just a an :int correctly CL-OPENCV> (-> ipl-image a roi ipl-roi width) 100 both work correctly but i could use help getting rid of some of the typing to access these members in this (-> ipl-image a roi ipl-roi width) i have to type "ipl-image"(the name of the struct) "a"( the name of the variable i'm dereferencing) "roi" (the name of the ipl-image struct member) "ipl-roi" (the foreign-type of roi) and "width" (the ipl-roi struct member) . if there was a way i can get the foreign type of the roi struct member like i can call foreign-slot-names to get all the slot names of a struct i could not have to use the ipl-roi here (-> ipl-image a roi ipl-roi width) but i looked in the manual and nothing....I couldnt think of any other way to shorten this (-> ipl-image a roi ipl-roi width) and make it a little closer to c so if someone could give me advice, or better yet their awesome dereferencing macro=), i named this post so everyone....including me=) would have access to this great information

1 0

[cffi-devel] Need some one to show me how to write this defcstruct correctly
by Joeish W 31 Oct '13

31 Oct '13

it is the CvSeq struct from opencv and it is defined a wee bit different than the others here it is as defined #define CV_TREE_NODE_FIELDS(node_type) \ int flags; /* Miscellaneous flags. */ \ int header_size; /* Size of sequence header. */ \ struct node_type* h_prev; /* Previous sequence. */ \ struct node_type* h_next; /* Next sequence. */ \ struct node_type* v_prev; /* 2nd previous sequence. */ \ struct node_type* v_next /* 2nd next sequence. */ /* Read/Write sequence. Elements can be dynamically inserted to or deleted from the sequence. */ #define CV_SEQUENCE_FIELDS() \ CV_TREE_NODE_FIELDS(CvSeq); \ int total; /* Total number of elements. */ \ int elem_size; /* Size of sequence element in bytes. */ \ schar* block_max; /* Maximal bound of the last block. */ \ schar* ptr; /* Current write pointer. */ \ int delta_elems; /* Grow seq this many at a time. */ \ CvMemStorage* storage; /* Where the seq is stored. */ \ CvSeqBlock* free_blocks; /* Free blocks list. */ \ CvSeqBlock* first; /* Pointer to the first sequence block. */ typedef struct CvSeq { CV_SEQUENCE_FIELDS() } the CV_SEQUENCE_FIELDS() part of the typedef struct CvSeq threw me I dont really know what to do with can someone just show me how to write it so I can then wrap all other structs like this....I could show things i tried but t hat would take up so much space it would be exhausting to look at...below is the swig wrapped struct i'm working with but it hasn't been working right so if you can show me how to edit it i would be very grateful =) (defcstruct cv-seq (flags :int) (header-size :int) (h-prev :pointer) (h-next :pointer) (v-prev :pointer) (v-next :pointer) (total :int) (elem-size :int) (block-max :pointer) (ptr :pointer) (delta-elems :int) (storage :pointer) (free-blocks :pointer) (first :pointer)) note normally when i add members like this CvSeqBlock* first; i do it like this (first (:pointer (:struct cv-block))) the new cffi style so if you can show me the new way on this that would be perfect =)

1 0

[cffi-devel] What foreign type would this be...I could use help writing a function definition for documentation
by Joeish W 30 Oct '13

30 Oct '13

the question is about CvArr* in OpenCV it is defined in the source like this /* CvArr* is used to pass arbitrary * array-like data structures * into functions where the particular * array type is recognized at runtime: */ typedef void CvArr; and documentation like this CvArr struct CvArr This is the “metatype” used only as a function parameter. It denotes that the function accepts arrays of multiple types, such as IplImage*, CvMat* or even CvSeq* sometimes. The particular array type is determined at runtime by analyzing the first 4 bytes of the header. In C++ interface the role of CvArr is played by InputArray and OutputArray. I'm writing my libraries documentation so i was wondering when i show my function definition as in c below CvMat* cvGetCol(const CvArr* arr, CvMat* submat, int col) How i would do it ? since CvMat is a struct i so far have this but is it the best it could be (get-cols (arr cv-arr) (:pointer (:struct submat)) (col :int)) but how would someone more skilled in cffi write the int part(col) and the CvArr part. In my library CvArr* is defined as this (defctype cv-arr :pointer)....I would appreciate any help on this

1 0