Use another language to call a function
You are encouraged to solve this task according to the task description, using any language you may know.
This task is inverse to the task Call foreign language function. Consider the following C program:
#include <stdio.h>
extern int Query (char * Data, size_t * Length);
int main (int argc, char * argv [])
{
char Buffer [1024];
size_t Size = sizeof (Buffer);
if (0 == Query (Buffer, &Size))
{
printf ("failed to call Query\n");
}
else
{
char * Ptr = Buffer;
while (Size-- > 0) putchar (*Ptr++);
putchar ('\n');
}
}
Implement the missing Query
function in your language, and let this C program call it. The function should place the string Here am I into the buffer which is passed to it as the parameter Data
. The buffer size in bytes is passed as the parameter Length
. When there is no room in the buffer, Query
shall return 0. Otherwise it overwrites the beginning of Buffer
, sets the number of overwritten bytes into Length
and returns 1.
Ada
The interface package Exported specification:
with Interfaces.C; use Interfaces.C;
with Interfaces.C.Strings; use Interfaces.C.Strings;
package Exported is
function Query (Data : chars_ptr; Size : access size_t)
return int;
pragma Export (C, Query, "Query");
end Exported;
The package implementation:
package body Exported is
function Query (Data : chars_ptr; Size : access size_t)
return int is
Result : char_array := "Here am I";
begin
if Size.all < Result'Length then
return 0;
else
Update (Data, 0, Result);
Size.all := Result'Length;
return 1;
end if;
end Query;
end Exported;
With GNAT it can be built as follows:
gcc -c main.c
gnatmake -c exported.adb
gnatbind -n exported.ali
gnatlink exported.ali main.o -o main
Sample output:
Here am I
AutoHotkey
It is possible to register an autohotkey function as a callback and get a pointer to it using the builtin registercallback function. Care should be taken that the external language code is running in the same thread as autohotkey. This is not a problem when using dllcall to use the external language. To run an autohotkey function from an external program running in a different thread, you can use ahkFunction in AutoHotkey.dll From the documentation on registercallback:
; Example: The following is a working script that displays a summary of all top-level windows.
; For performance and memory conservation, call RegisterCallback() only once for a given callback:
if not EnumAddress ; Fast-mode is okay because it will be called only from this thread:
EnumAddress := RegisterCallback("EnumWindowsProc", "Fast")
DetectHiddenWindows On ; Due to fast-mode, this setting will go into effect for the callback too.
; Pass control to EnumWindows(), which calls the callback repeatedly:
DllCall("EnumWindows", UInt, EnumAddress, UInt, 0)
MsgBox %Output% ; Display the information accumulated by the callback.
EnumWindowsProc(hwnd, lParam)
{
global Output
WinGetTitle, title, ahk_id %hwnd%
WinGetClass, class, ahk_id %hwnd%
if title
Output .= "HWND: " . hwnd . "`tTitle: " . title . "`tClass: " . class . "`n"
return true ; Tell EnumWindows() to continue until all windows have been enumerated.
}
C
I rewrote the driver as
#if 0
I rewrote the driver according to good sense, my style,
and discussion.
This is file main.c on Autumn 2011 ubuntu linux release.
The emacs compile command output:
-*- mode: compilation; default-directory: "/tmp/" -*-
Compilation started at Mon Mar 12 20:25:27
make -k CFLAGS=-Wall main.o
cc -Wall -c -o main.o main.c
Compilation finished at Mon Mar 12 20:25:27
#endif
#include <stdio.h>
#include <stdlib.h>
extern int Query(char *Data, unsigned *Length);
int main(int argc, char *argv[]) {
char Buffer[1024], *pc;
unsigned Size = sizeof(Buffer);
if (!Query(Buffer, &Size))
fputs("failed to call Query", stdout);
else
for (pc = Buffer; Size--; ++pc)
putchar(*pc);
putchar('\n');
return EXIT_SUCCESS;
}
With solution
#if 0
This is file query.c
-*- mode: compilation; default-directory: "/tmp/" -*-
Compilation started at Mon Mar 12 20:36:25
make -k CFLAGS=-Wall query.o
cc -Wall -c -o query.o query.c
Compilation finished at Mon Mar 12 20:36:26
#endif
#include<string.h>
int Query(char *Data, unsigned *Length) {
const char *message = "Here am I";
unsigned n = strlen(message);
if (n <= *Length)
return strncpy(Data, message, (size_t)n), *Length = n, 1;
return 0;
}
And finally, excitement!
$ gcc main.c query.o -o main && ./main
Here am I
$
C++
#include <string>
using std::string;
// C++ functions with extern "C" can get called from C.
extern "C" int
Query (char *Data, size_t *Length)
{
const string Message = "Here am I";
// Check that Message fits in Data.
if (*Length < Message.length())
return false; // C++ converts bool to int.
*Length = Message.length();
Message.copy(Data, *Length);
return true;
}
We must compile main() with a C compiler and Query() with a C++ compiler. One can use gcc and g++ (or clang and clang++).
$ gcc -c main.c
$ g++ -c query.cpp
$ g++ -o main main.o query.o
$ ./main
Here am I
COBOL
GnuCOBOL uses C intermediates and blends well with C programming. GnuCOBOL is also a fixed length data item language, so this Query routine has to set some limits on passed in external value lengths. 8K in this example, defined using OCCURS DEPENDING ON the input Length.
Instead of C being the master builder, cobc is used to combine the .c source and .cob source into the executable simplifying the tectonic for this example (cobc can generate and use .o object code, but that is all hidden here). GnuCOBOL also requires a COBOL runtime system, implicitly initialized with the -fimplicit-init compiler switch here. This emits code to ensure libcob is properly setup for calling from foreign languages (that would otherwise have to call cob_init() before invoking COBOL modules). The source code in the task description was saved to disk as call-query.c
, and the listing below was saved as query.cob
(with the internal subprogram named Query
). A special call-convention
is also used so the GnuCOBOL module does not make any assumptions about how the Query module is invoked (normal COBOL programs set some control fields in the libcob runtime space when calling modules, which won't be set when called from a foreign C ABI program). GnuCOBOL also sets RETURN-CODE
to zero unless told otherwise (or some error occurs).
identification division.
program-id. Query.
environment division.
configuration section.
special-names.
call-convention 0 is extern.
repository.
function all intrinsic.
data division.
working-storage section.
01 query-result.
05 filler value "Here I am".
linkage section.
01 data-reference.
05 data-buffer pic x occurs 0 to 8192 times
depending on length-reference.
01 length-reference usage binary-long.
procedure division extern using data-reference length-reference.
if length(query-result) less than or equal to length-reference
and length-reference less than 8193 then
move query-result to data-reference
move length(query-result) to length-reference
move 1 to return-code
end-if
goback.
end program Query.
- Output:
prompt$ cobc -x -fimplicit-init call-query.c query.cob prompt$ ./call-query Here I am
D
This shows how to perform the task on Windows. Elsewhere the procedure is very similar.
First write a D module like this, named "query_func.d":
import core.stdc.string;
extern(C) bool query(char *data, size_t *length) pure nothrow {
immutable text = "Here am I";
if (*length < text.length) {
*length = 0; // Also clears length.
return false;
} else {
memcpy(data, text.ptr, text.length);
*length = text.length;
return true;
}
}
Generate a library file with:
dmd -lib query_func.d
This generates a query_func.lib
file.
Then create a C file named "mainc.c", given in the task description and here improved a little:
#include <stdio.h>
#include <stdbool.h>
extern bool query(char *data, size_t *length);
int main() {
char buffer[1024];
size_t size = sizeof(buffer);
if (query(buffer, &size))
printf("%.*s\n", size, buffer);
else
puts("The call to query has failed.");
return 0;
}
Then you can compile and link all with the DMC C compiler(on Linux you can use GCC):
dmc query_func.lib mainc.c
It generates the "mainc.exe" binary, that prints the desired output:
Here am I
Delphi
function Query(Buffer: PChar; var Size: Int64): LongBool;
const
Text = 'Hello World!';
begin
If not Assigned(Buffer) Then
begin
Size := 0;
Result := False;
Exit;
end;
If Size < Length(Text) Then
begin
Size := 0;
Result := False;
Exit;
end;
Size := Length(Text);
Move(Text[1], Buffer^, Size);
Result := True;
end;
To use this function from C you have to export this as a DLL and bind your C program to this function.
Fortran
Simple task because interoperability with C is in Fortran language since F2003 standard.
!-----------------------------------------------------------------------
!Function
!-----------------------------------------------------------------------
function fortran_query(data, length) result(answer) bind(c, name='Query')
use, intrinsic :: iso_c_binding, only: c_char, c_int, c_size_t, c_null_char
implicit none
character(len=1,kind=c_char), dimension(length), intent(inout) :: data
integer(c_size_t), intent(inout) :: length
integer(c_int) :: answer
answer = 0
if(length<10) return
data = transfer("Here I am"//c_null_char, data)
length = 10_c_size_t
answer = 1
end function fortran_query
compile it: gfortran main.c query.f90 -o main.x
Go
Possible—if you allow a small stretch of the task specification.
Cgo, Go's interface to C, allows calls from C to Go, but only if it gets to start Go first. That is, it doesn't work with a program started with C startup code and C main(), but only with a program started with Go startup code and Go main().
Thus, I changed the specified C code to begin as follows,
#include <stdio.h>
#include "_cgo_export.h"
void Run()
{
char Buffer [1024];
size_t Size = sizeof (Buffer);
if (0 == Query (Buffer, &Size))
...
The biggest change is that I renamed main, since it is no longer a C main function. Another small change is that the extern declaration is replaced by an include. The included file is generated by cgo and contains an equivalent extern declaration.
In the Go code, below, you see that all main does is call C.Run. The C code is then in the driver's seat.
package main
// #include <stdlib.h>
// extern void Run();
import "C"
import "unsafe"
func main() {
C.Run()
}
const msg = "Here am I"
//export Query
func Query(cbuf *C.char, csiz *C.size_t) C.int {
if int(*csiz) <= len(msg) {
return 0
}
pbuf := uintptr(unsafe.Pointer(cbuf))
for i := 0; i < len(msg); i++ {
*((*byte)(unsafe.Pointer(pbuf))) = msg[i]
pbuf++
}
*((*byte)(unsafe.Pointer(pbuf))) = 0
*csiz = C.size_t(len(msg) + 1)
return 1
}
Output:
Here am I
Alternative Method
As of Go 1.5, this is now possible without modifying the C code thanks to the addition of buildmodes. Buildmodes allow Go code to be compiled to standard C libraries (both dynamic and static).
The Go code for this task is as follows:
// This buildmode requires the package to be main
package main
// Import C so we can export the function to C and use C types
//#include <stdlib.h> // for size_t
import "C"
// Import reflect and unsafe so we can wrap the C array in a Go slice
import "reflect"
import "unsafe"
// This buildmode also requires a main function, but it is never actually called
func main() {}
// The message to copy into the buffer
const msg = "Here am I"
// Here we declare the Query function using C types and export it to C
//export Query
func Query(buffer *C.char, length *C.size_t) C.int {
// Check there is enough space in the buffer
if int(*length) < len(msg) {
return 0
}
// Wrap the buffer in a slice to make it easier to copy into
sliceHeader := reflect.SliceHeader {
Data: uintptr(unsafe.Pointer(buffer)),
Len: len(msg),
Cap: len(msg),
}
bufferSlice := *(*[]byte)(unsafe.Pointer(&sliceHeader))
// Iterate through the message and copy it to the buffer, byte by byte
for i:=0;i<len(msg);i++ {
bufferSlice[i] = msg[i]
}
// Set length to the amount of bytes we copied
(*length) = C.size_t(len(msg))
return 1
}
Assuming this is saved to query.go (and that the C code is saved as main.c) it can be compiled with:
$ go build -buildmode=c-shared query.go $ gcc main.c -L. -lquery -o main
This creates a library file, a header file for the library (not used) and an executable dynamically linked to the library file.
The executable can be run with:
$ LD_LIBRARY_PATH=. ./main Here am I
Haskell
I modified the C source to include Haskell-specific headers and to init the Haskell environment. I also changed "Query" to "query_hs" due to capitalization issues:
#ifdef __GLASGOW_HASKELL__
#include "Called_stub.h"
extern void __stginit_Called(void);
#endif
#include <stdio.h>
#include <HsFFI.h>
int main (int argc, char * argv [])
{
char Buffer [1024];
size_t Size = sizeof (Buffer);
hs_init(&argc, &argv);
#ifdef __GLASGOW_HASKELL__
hs_add_root(__stginit_Called);
#endif
if (0 == query_hs (Buffer, &Size))
{
printf ("failed to call Query\n");
}
else
{
char * Ptr = Buffer;
while (Size-- > 0) putchar (*Ptr++);
putchar ('\n');
}
hs_exit();
return 0;
}
The Haskell code then is:
{-# LANGUAGE ForeignFunctionInterface #-}
module Called where
import Foreign
import Foreign.C.String (CString, withCStringLen)
import Foreign.C.Types
-- place a string into the buffer pointed to by ptrBuff (with size
-- pointed to by ptrSize). If successful, sets number of overwritten
-- bytes in ptrSize and returns 1, otherwise, it does nothing and
-- returns 0
query_hs :: CString -> Ptr CSize -> IO CInt
query_hs ptrBuff ptrSize = withCStringLen "Here I am"
(\(str, len) -> do
buffSize <- peek ptrSize
if sizeOf str > (fromIntegral buffSize)
then do
poke ptrSize 0
return 0
else do
poke ptrSize (fromIntegral len)
copyArray ptrBuff str len
return 1)
foreign export ccall query_hs :: CString -> Ptr CSize -> IO CInt
Compile the Haskell code with:
ghc -c -O Called.hs
Then compile the C code together with the generated Haskell files (using GHC):
ghc -optc-O calling.c Called.o Called_stub.o -o calling
Output:
Here I am
Haxe
PHP
untyped __call__("functionName", args);
J
Install an input and an output routine to use the J engine externally. These pass character strings arguments. To complete the task, I made these two routines communicate using compilation-unit scope variables (static). I tested the program on a 64 bit Ubuntu linux 2011 Autumn release with j versions 602 and 701. Comment of asterisks marks the input and output routines.
The J verb evaluates to the string unless there is no space. File rc_embed.ijs
query=:3 :'0&#^:(y < #)''Here am I'''
main.c
#include<stdio.h>
#include<stdlib.h>
#include<string.h>
int Query(char*,unsigned*);
int main(int argc,char*argv[]) {
char Buffer[1024], *pc;
unsigned Size = (unsigned)sizeof(Buffer);
if (!Query(Buffer,&Size))
fputs("Failed to call Query",stdout);
else
for (pc = Buffer; Size--; ++pc)
putchar(*pc);
putchar('\n');
return EXIT_SUCCESS;
}
Query.c
// J Front End Example
// define _WIN32 for Windows, __MACH__ for MAC, J64 for 64-bit
// JE is loaded from current working directory
//make jfex && LD_LIBRARY_PATH=/usr/local/j64-701/bin ./jfex
#ifdef _WIN32
#define _CRT_SECURE_NO_WARNINGS
#include <windows.h>
#include <direct.h>
#define GETPROCADDRESS(h,p) GetProcAddress(h,p)
#define JDLLNAME "\\j.dll"
#else
#define _stdcall
#include <dlfcn.h>
#define GETPROCADDRESS(h,p) dlsym(h,p)
#ifdef __MACH__
#define JDLLNAME "/libj.dylib"
#else
#define JDLLNAME "/libj.so"
#endif
#define _getcwd getcwd
#endif
#include<stdio.h>
#include<signal.h>
#include<stdlib.h>
#include<string.h>
#include"jfex.h"
#include"jlib.h"
static JDoType jdo;
static JFreeType jfree;
static JgaType jga;
static JGetLocaleType jgetlocale;
static J jt;
static void* hjdll;
static char **adadbreak;
static void sigint(int k){**adadbreak+=1;signal(SIGINT,sigint);}
static char input[1000];
// J calls for input (debug suspension and 1!:1[1) and we call for input
char* _stdcall Jinput(J jt,char* prompt)
{
fputs(prompt,stdout);
if(fgets(input, sizeof(input), stdin))
{
fputs("\n",stdout);
**adadbreak+=1;
}
return input;
}
static char*buffer = NULL; /**************************************/
static unsigned length = 0; /**************************************/
static int Jouts = 0; /**************************************/
// J calls for output
#define LINEFEED 10 /**************************************/
void _stdcall Joutput(J jt,int type, char* s) /********************/
{
size_t L;
if(MTYOEXIT==type) exit((int)(I)s);
L = strlen(s);
L -= (L && (LINEFEED==s[L-1])); /* CRLF not handled. */
if (L && (!Jouts)) {
length = L;
strncpy(buffer,s,L);
Jouts = 1;
}
}
int Query(char*Data,unsigned*Length)
{
void* callbacks[] = {Joutput,NULL,Jinput,0,(void*)SMCON};
char pathdll[1000];
_getcwd(pathdll,sizeof(pathdll));
strcat(pathdll,JDLLNAME);
#ifdef _WIN32
hjdll=LoadLibraryA(pathdll);
#else
hjdll=dlopen(pathdll,RTLD_LAZY);
if (NULL == hjdll)
hjdll=dlopen(JDLLNAME+1,RTLD_LAZY); /* use LD_LIBRARY_PATH */
#endif
if(NULL == hjdll)
{
fprintf(stderr,"Unix use: $ LD_LIBRARY_PATH=path/to/libj.so %s\n","programName");//*argv);
fputs("Load library failed: ",stderr);
fputs(pathdll,stderr);
fputs("\n",stderr);
return 0; // load library failed
}
jt=((JInitType)GETPROCADDRESS(hjdll,"JInit"))();
if(!jt) return 0; // JE init failed
((JSMType)GETPROCADDRESS(hjdll,"JSM"))(jt,callbacks);
jdo=(JDoType)GETPROCADDRESS(hjdll,"JDo");
jfree=(JFreeType)GETPROCADDRESS(hjdll,"JFree");
jga=(JgaType)GETPROCADDRESS(hjdll,"Jga");
jgetlocale=(JGetLocaleType)GETPROCADDRESS(hjdll,"JGetLocale");
adadbreak=(char**)jt; // first address in jt is address of breakdata
signal(SIGINT,sigint);
{
char input[999];
//memset(input,0,sizeof input);
buffer = Data;
sprintf(input,"query %u [ 0!:110<'rc_embed.ijs'\n",*Length); /***deceptive input routine, a hard coded string*********/
jdo(jt,input);
if (!Jouts)
return 0;
*Length = length;
}
jfree(jt);
return 1;
}
makefile, adjust for your j installation.
# jfe makefile info
# customize to create makefile suitable for your platform
# 32bit builds on 64bit systems require -m32 in CFLAGS and FLAGS
# Unix requires -ldl in FLAGS and Windows does not
CPPFLAGS= -I/usr/local/j64-602/system/examples/jfe
CFLAGS= -O0 -g
LOADLIBES= -ldl
main: main.o Query.o
Finally, build and execution. Again, adjust LD_LIBRARY_PATH to the directory of libj.so .
$ make main && LD_LIBRARY_PATH=~/Downloads/jgplsrc/j/bin ./main
Here am I
$
Java
We write a Java method, then write a C function Query()
to use the Java Native Interface (JNI) to call our Java method. The C compiler must find jni.h and jni_md.h and link with libjvm.
/* Query.java */
public class Query {
public static boolean call(byte[] data, int[] length)
throws java.io.UnsupportedEncodingException
{
String message = "Here am I";
byte[] mb = message.getBytes("utf-8");
if (length[0] < mb.length)
return false;
length[0] = mb.length;
System.arraycopy(mb, 0, data, 0, mb.length);
return true;
}
}
/* query-jni.c */
#include <stdio.h>
#include <stdlib.h>
#include <jni.h>
static JavaVM *jvm = NULL;
static JNIEnv *jenv = NULL;
static void die(const char *message) {
fprintf(stderr, "%s\n", message);
exit(1);
}
static void oom(void) {
die("Query: out of memory");
}
static void except(void) {
if ((*jenv)->ExceptionCheck(jenv))
die("Query: unexpected Java exception");
}
static void do_at_exit(void) {
(*jvm)->DestroyJavaVM(jvm);
}
static void require_jvm(void) {
JavaVMInitArgs args;
if (jvm)
return;
args.version = JNI_VERSION_1_4;
args.nOptions = 0;
args.options = NULL;
args.ignoreUnrecognized = JNI_FALSE;
if (JNI_CreateJavaVM(&jvm, (void **)&jenv, &args) != JNI_OK)
die("Query: can't create Java VM");
atexit(do_at_exit);
}
int Query(char *data, size_t *length) {
jclass cQuery;
jmethodID mcall;
jintArray jlength;
jint jlength0;
jbyteArray jdata;
jboolean result;
jlength0 = (jint)length[0];
if ((size_t)jlength0 != length[0])
die("Query: length is too large for Java array");
require_jvm();
/* Create a local frame for references to Java objects. */
if ((*jenv)->PushLocalFrame(jenv, 16))
oom();
/* Look for class Query, static boolean call(byte[], int[]) */
cQuery = (*jenv)->FindClass(jenv, "Query");
if (cQuery == NULL)
die("Query: can't find Query.class");
mcall = (*jenv)->GetStaticMethodID(jenv, cQuery, "call", "([B[I)Z");
if (mcall == NULL)
die("Query: missing call() method");
/*
* Make arguments to Query.call(). We can't pass data[] and
* length[] to Java, so we make new Java arrays jdata[] and
* jlength[].
*/
jdata = (*jenv)->NewByteArray(jenv, (jsize)jlength0);
if (jdata == NULL)
oom();
jlength = (*jenv)->NewIntArray(jenv, 1);
if (jlength == NULL)
oom();
/* Set jlength[0] = length[0]. */
(*jenv)->SetIntArrayRegion(jenv, jlength, 0, 1, &jlength0);
except();
/*
* Call our Java method.
*/
result = (*jenv)->CallStaticBooleanMethod
(jenv, cQuery, mcall, jdata, jlength);
except();
/*
* Set length[0] = jlength[0].
* Copy length[0] bytes from jdata[] to data[].
*/
(*jenv)->GetIntArrayRegion(jenv, jlength, 0, 1, &jlength0);
except();
length[0] = (size_t)jlength0;
(*jenv)->GetByteArrayRegion
(jenv, jdata, 0, (jsize)jlength0, (jbyte *)data);
/* Drop our local frame and its references. */
(*jenv)->PopLocalFrame(jenv, NULL);
return (int)result;
}
# Makefile
# Edit these lines to match your JDK.
JAVA_HOME = /Library/Java/Home
CPPFLAGS = -I$(JAVA_HOME)/include
LIBS = -framework JavaVM
JAVAC = $(JAVA_HOME)/bin/javac
CC = cc
all: calljava Query.class
calljava: main.o query-jni.o
$(CC) -o calljava main.o query-jni.o $(LIBS)
.SUFFIXES: .c .class .java .o
.c.o:
$(CC) $(CPPFLAGS) -c $<
.java.class:
$(JAVAC) $<
clean:
rm -f calljava main.o query-jni.o Query.class
Kotlin
Reverse interop (calling Kotlin from C) was added to Kotlin Native in version 0.5 and the following shows how to perform this task on Ubuntu Linux.
First we compile the following Kotlin source file (Query.kt) using the '-platform dynamic' flag:
// Kotlin Native v0.6
import kotlinx.cinterop.*
import platform.posix.*
fun query(data: CPointer<ByteVar>, length: CPointer<size_tVar>): Int {
val s = "Here am I"
val strLen = s.length
val bufferSize = length.pointed.value
if (strLen > bufferSize) return 0 // buffer not large enough
for (i in 0 until strLen) data[i] = s[i].toByte()
length.pointed.value = strLen.signExtend<size_t>()
return 1
}
This produces the dynamic library, libQuery.so, and the C header file libQuery_api.h:
#ifndef KONAN_LIBQUERY_H
#define KONAN_LIBQUERY_H
#ifdef __cplusplus
extern "C" {
#endif
typedef unsigned char libQuery_KBoolean;
typedef char libQuery_KByte;
typedef unsigned short libQuery_KChar;
typedef short libQuery_KShort;
typedef int libQuery_KInt;
typedef long long libQuery_KLong;
typedef float libQuery_KFloat;
typedef double libQuery_KDouble;
typedef void* libQuery_KNativePtr;
struct libQuery_KType;
typedef struct libQuery_KType libQuery_KType;
typedef struct {
/* Service functions. */
void (*DisposeStablePointer)(libQuery_KNativePtr ptr);
void (*DisposeString)(const char* string);
libQuery_KBoolean (*IsInstance)(libQuery_KNativePtr ref, const libQuery_KType* type);
/* User functions. */
struct {
struct {
libQuery_KInt (*query)(void* data, void* length);
} root;
} kotlin;
} libQuery_ExportedSymbols;
extern libQuery_ExportedSymbols* libQuery_symbols(void);
#ifdef __cplusplus
} /* extern "C" */
#endif
#endif /* KONAN_LIBQUERY_H */
We now compile a slightly modified version of the C program required for this task, linking to the above library, and 'including' the header file:
#include <stdio.h>
#include <stdlib.h>
#include "libQuery_api.h"
static int Query (char * Data, size_t * Length)
{
return libQuery_symbols() -> kotlin.root.query(Data, Length);
}
int main (int argc, char * argv [])
{
char Buffer [1024];
size_t Size = sizeof (Buffer);
if (0 == Query (Buffer, &Size))
{
printf ("failed to call Query\n");
}
else
{
char * Ptr = Buffer;
while (Size-- > 0) putchar (*Ptr++);
putchar ('\n');
}
}
which when executed produces the expected output:
Here am I
Lisaac
query.li
Section Header
+ name := QUERY;
- external := `#define main _query_main`;
- external := `#define query Query`;
Section External
- query(buffer : NATIVE_ARRAY[CHARACTER], size : NATIVE_ARRAY[INTEGER]) : INTEGER <- (
+ s : STRING_CONSTANT;
+ len, result : INTEGER;
s := "Here am I";
len := s.count;
(len > size.item(0)).if {
result := 0;
} else {
1.to len do { i : INTEGER;
buffer.put (s @ i) to (i - 1);
};
size.put len to 0;
result := 1;
};
result
);
Section Public
- main <- (
+ buffer : NATIVE_ARRAY[CHARACTER];
+ size : NATIVE_ARRAY[INTEGER];
query(buffer, size); // need this to pull the query() method
);
Makefile
TARGET=test_query
all: $(TARGET)
$(TARGET): main.o query.o
gcc -o $@ main.o query.o
.c.o:
gcc -c $<
query.c: query.li
-lisaac $<
clean:
rm -f $(TARGET) *.o query.c
Mercury
The code as written is horrible for Mercury, so some additional C is added as a shim that actually calls the Mercury predicate. Although no changes are required to the C code in this simple example, in a larger project, with modules that need initialization, there are some additional compilation steps needed to get that initialization code in.
:- module query.
:- interface.
:- pred query(string::in, string::out) is det.
:- implementation.
query(_, "Hello, world!").
:- pragma foreign_export("C", query(in, out), "query").
:- pragma foreign_decl("C",
"
#include <string.h>
int Query (char * Data, size_t * Length);
").
:- pragma foreign_code("C",
"
int Query (char *Data, size_t *Length) {
MR_String input, result;
MR_allocate_aligned_string_msg(input, *Length, MR_ALLOC_ID);
memmove(input, Data, *Length);
query(input, &result);
*Length = strlen(result);
memmove(Data, result, *Length);
return 1;
}
").
Building with the unchanged C in useanother.c:
$ mmc -c query $ gcc -Wall -c useanother.c $ ml -o useanother useanother.o query.o
- Output:
Hello, world!
Nim
proc Query*(data: var array[1024, char], length: var cint): cint {.exportc.} =
const text = "Here am I"
if length < text.len:
return 0
for i in 0 .. text.high:
data[i] = text[i]
length = text.len
return 1
Compile the above with nim c --app:staticlib --no_main query.nim
.
#include <stdio.h>
extern int Query (char * Data, size_t * Length);
int main (int argc, char * argv [])
{
char Buffer [1024];
size_t Size = sizeof (Buffer);
if (0 == Query (Buffer, &Size))
{
printf ("failed to call Query\n");
}
else
{
char * Ptr = Buffer;
while (Size-- > 0) putchar (*Ptr++);
putchar ('\n');
}
}
Compile the above with gcc -ldl -o main main.c libquery.a
, then execute the resulting main
binary:
./main Here am I
OCaml
#include <stdio.h>
#include <string.h>
#include <caml/mlvalues.h>
#include <caml/callback.h>
extern int Query (char * Data, size_t * Length)
{
static value * closure_f = NULL;
if (closure_f == NULL) {
closure_f = caml_named_value("Query function cb");
}
value ret = caml_callback(*closure_f, Val_unit);
*Length = Int_val(Field(ret, 1));
strncpy(Data, String_val(Field(ret, 0)), *Length);
return 1;
}
int main (int argc, char * argv [])
{
char Buffer [1024];
unsigned Size = 0;
caml_main(argv); /* added from the original main */
if (0 == Query (Buffer, &Size))
{
printf ("failed to call Query\n");
}
else
{
char * Ptr = Buffer;
printf("size: %d\n", Size);
while (Size-- > 0) putchar (*Ptr++);
putchar ('\n');
}
}
let caml_query () =
let s = "Here am I" in
(s, String.length s)
;;
let () =
Callback.register "Query function cb" caml_query;
;;
compile with:
ocamlopt -output-obj caml_part.ml -o caml_part_obj.o gcc -c main.c -I"`ocamlc -where`" gcc -o prog.opt main.o caml_part_obj.o \ -L"`ocamlc -where`" \ -lm -ldl -lasmrun
Ol
Simpler solution: just return from otus-lisp a string.
#include <extensions/embed.h>
#define min(x,y) (x < y ? x : y)
extern unsigned char repl[];
int Query(char *Data, size_t *Length) {
ol_t ol;
embed_new(&ol, repl, 0);
word s = embed_eval(&ol, new_string(&ol,
"(define sample \"Here am I\")"
"sample"
), 0);
if (!is_string(s))
goto fail;
int i = *Length = min(string_length(s), *Length);
memcpy(Data, string_value(s), i);
*Length = i;
OL_free(ol.vm);
return 1;
fail:
OL_free(ol.vm);
return 0;
}
Better solution: do a copy of string directly in otus-lisp.
#include <extensions/embed.h>
extern unsigned char repl[];
int Query(char *Data, size_t *Length) {
ol_t ol;
embed_new(&ol, repl, 0);
embed_eval(&ol, new_string(&ol,
"(import (otus ffi))"
"(define lib (load-dynamic-library #f))"
"(define memcpy (lib fft-void* \"memcpy\" fft-void* type-string fft-int))"
"(define (Query Data Length)"
" (define sample (c-string \"Here am I\"))"
" (when (memcpy Data sample (min (string-length sample) Length))"
" (min (string-length sample) Length)))"
), 0);
word r =
embed_eval(&ol, new_string(&ol, "Query"), new_vptr(&ol, Data), make_integer(*Length), 0);
if (!is_number(r))
goto fail;
*Length = ol2int(r);
OL_free(ol.vm);
return 1;
fail:
OL_free(ol.vm);
return 0;
}
PARI/GP
This is a Linux solution. Message "Here I am" is encrypted with ROT13: "Urer V nz".
ROT13() is implemented as a PARI one-liner:
Strchr(Vecsmall(apply(k->if(k>96&&k<123,(k-84)%26+97,if(k>64&&k<91,(k-52)%26+65,k)),Vec(Vecsmall(s)))))
PARI's interface for Query()... query.c:
#include <pari/pari.h>
#define PARI_SECRET "s=\"Urer V nz\";Strchr(Vecsmall(apply(k->if(k>96&&k<123,(k-84)%26+97,if(k>64&&k<91,(k-52)%26+65,k)),Vec(Vecsmall(s)))))"
int Query(char *Data, size_t *Length)
{
int rc = 0;
GEN result;
pari_init(1000000, 2);
result = geval(strtoGENstr(PARI_SECRET)); /* solve the secret */
if (result) {
strncpy(Data, GSTR(result), *Length); /* return secret */
rc = 1;
}
pari_close();
return rc;
}
Compile interface to a library: gcc -O2 -Wall -fPIC -shared query.c -o libquery.so -lpari
Compile main() C code from above and link against this library: gcc -O2 -Wall main.c -o main -L. libquery.so
Start main(): LD_LIBRARY_PATH=. ./main
PARI solves the ROT13 encrypted message and returns result to caller.
Output:
Here I am
NB. It's also possible to compile both files together without building an interface: gcc -O2 -Wall main.c query.c -o main2 -lpari
./main2 yields same output as stated above.
Pascal
See Delphi
Phix
The following code declares a callback for the C code (which I'd expect to be in a .dll or .so) to invoke.
A 32-bit-only or a 64-bit-only version would of course be slightly shorter.
without js -- (peek/poke, call_back) constant Here_am_I = "Here am I" function Query(atom pData, pLength) integer len = peekNS(pLength,machine_word(),0) if poke_string(pData,len,Here_am_I) then return 0 end if pokeN(pLength,length(Here_am_I)+1,machine_word()) return 1 end function constant Query_cb = call_back(Query)
PicoLisp
Calling a PicoLisp function from another program requires a running interpreter. There are several possibilities, like IPC via fifo's or sockets using the PLIO (PicoLisp-I/O) protocol, but the easiest is calling the interpreter in a pipe. This is relatively efficient, as the interpreter's startup time is quite short.
If there is a file "query.l"
(let (Str "Here am I" Len (format (opt))) # Get length from command line
(unless (>= (size Str) Len) # Check buffer size
(prinl Str) ) ) # Return string if OK
then the C function 'Query' could be
int Query(char *Data, size_t *Length) {
FILE *fp;
char buf[64];
sprintf(buf, "/usr/bin/picolisp query.l %d -bye", *Length);
if (!(fp = popen(buf, "r")))
return 0;
fgets(Data, *Length, fp);
*Length = strlen(Data);
return pclose(fp) >= 0 && *Length != 0;
}
Python
Our embedded python function a) uses information from the main routine in c, and b) determines the information to populate the result returned to the main routine. This, I believe, fulfills the task requirement. The modifications and compilation are shown for Ubuntu linux Autumn 2011 version, with python3. It's easier to call a dynamic library from python using the ctypes module. Consider using PyRun_SimpleString to have main.c call python calling back to c.
# store this in file rc_embed.py
# store this in file rc_embed.py
def query(buffer_length):
message = b'Here am I'
L = len(message)
return message[0:L*(L <= buffer_length)]
main.c
#if 0
//I rewrote the driver according to good sense, my style,
//and discussion --Kernigh 15:45, 12 February 2011 (UTC).
#endif
#include<stdio.h>
#include<stdlib.h>
#include<string.h>
extern int Query(char*,unsigned*);
int main(int argc,char*argv[]) {
char Buffer[1024], *pc;
unsigned Size = sizeof(Buffer);
if (!Query(Buffer,&Size))
fputs("Failed to call Query",stdout);
else
for (pc = Buffer; Size--; ++pc)
putchar(*pc);
putchar('\n');
return EXIT_SUCCESS;
}
In Query.c I don't promise to have tested every case with missing module, missing function, or to have used Py_DECREF correctly.
#include<stdio.h>
#include<stdlib.h>
#include<string.h>
#include<Python.h>
int Query(char*Data,unsigned*Length) {
char *module = "rc_embed", *function = "query";
PyObject *pName, *pModule, *pFunc, *pResult, *pArgs, *pLength;
long result = 0;
if (!Py_IsInitialized())
Py_Initialize();
pName = PyUnicode_FromString(module);
pModule = PyImport_Import(pName);
Py_DECREF(pName);
if (NULL == pModule) {
PyErr_Print();
fprintf(stderr,"Failed to load \"%s\"\n",module);
return 0;
}
pFunc = PyObject_GetAttrString(pModule,function);
if ((NULL == pFunc) || (!PyCallable_Check(pFunc))) {
if (PyErr_Occurred())
PyErr_Print();
fprintf(stderr,"Cannot find function \"%s\"\n",function);
if (NULL != pFunc)
Py_DECREF(pFunc);
Py_DECREF(pModule);
return 0;
}
pArgs = PyTuple_New(1);
pLength = PyLong_FromUnsignedLong((unsigned long)(*Length));
if (NULL == pLength) {
Py_DECREF(pArgs);
Py_DECREF(pFunc);
Py_DECREF(pModule);
return 0;
}
PyTuple_SetItem(pArgs,0,pLength);
pResult = PyObject_CallObject(pFunc, pArgs);
if (NULL == pResult)
result = 0;
else if (!PyBytes_Check(pResult)) {
result = 0;
Py_DECREF(pResult);
} else {
if (! PyBytes_Size(pResult))
result = 0;
else {
*Length = (unsigned)PyBytes_Size(pResult);
strncpy(Data,PyBytes_AsString(pResult),*Length);
Py_DECREF(pResult);
result = 1;
}
}
Py_DECREF(pArgs);
Py_DECREF(pFunc);
Py_DECREF(pModule);
Py_Finalize();
return result;
}
Compilation, linkage, execution. Note the python tools used to extract the correct flags.
$ make main.o
cc -c -o main.o main.c
$ D=$( dirname $( which python3 ) )
$ gcc $( $D/python3.2-config --cflags ) -c Query.c
In file included from /usr/include/python3.2mu/Python.h:8:0,
from Q.c:18:
/usr/include/python3.2mu/pyconfig.h:1173:0: warning: "_POSIX_C_SOURCE" redefined [enabled by default]
/usr/include/features.h:214:0: note: this is the location of the previous definition
$ gcc -o main main.o Query.o $( $D/python3.2-config --ldflags )
$ ./main
Here am I
$
Racket
Since this problem is presented as the inverse to Call foreign language function, I've focused on just demonstrating a callback from C into Racket, instead of showing how to embed the whole Racket runtime into C.
Starting with the given C code, modify it so that Query is a variable instead of an external:
typedef int strfun (char * Data, size_t * Length);
strfun *Query = NULL;
The rest of the C code is left as-is. Compile it into a dynamic library, then run the following Racket code:
#lang racket
(require ffi/unsafe)
(define xlib (ffi-lib "./x.so"))
(set-ffi-obj! "Query" xlib (_fun _pointer _pointer -> _bool)
(λ(bs len)
(define out #"Here I am")
(let ([bs (make-sized-byte-string bs (ptr-ref len _int))])
(and ((bytes-length out) . <= . (bytes-length bs))
(begin (bytes-copy! bs 0 out)
(ptr-set! len _int (bytes-length out))
#t)))))
((get-ffi-obj "main" xlib (_fun _int (_list i _bytes) -> _void))
0 '())
Note that this code is intentionally in a simple low-level form, for example, it sets the pointer directly instead of using a C function.
The output is the expected “Here I am” line.
Raku
(formerly Perl 6) This SO answer, by JJ Merelo, explained the difficulty on providing a simple solution. So this attempt takes the same approach as PicoLisp by summoning the interpreter at run time.
query.raku
#!/usr/bin/env raku
sub MAIN (Int :l(:len(:$length))) {
my Str $String = "Here am I";
$*OUT.print: $String if $String.codes ≤ $length
}
query.c
#include<stdio.h>
#include<stddef.h>
#include<string.h>
int Query(char *Data, size_t *Length) {
FILE *fp;
char buf[64];
sprintf(buf, "/home/user/query.raku --len=%zu", *Length);
if (!(fp = popen(buf, "r")))
return 0;
fgets(Data, *Length, fp);
*Length = strlen(Data);
return pclose(fp) >= 0 && *Length != 0;
}
- Output:
gcc -Wall -o main main.c query.c ./main Here am I
Ruby
We have four files. First, main.c has the main()
from the top of this page. Second, query.rb uses Fiddle to create a C function at run time. It sets the C variable QueryPointer
to this function. It needs at least Ruby 2.3 for Array#unpack('J')
. Our main()
can't call QueryPointer()
, because it expects Query()
to exist at link time, and it would crash if Ruby raised an error. The third file query-rb.c provides Query()
as a wrapper around QueryPointer()
. The wrapper embeds the Ruby interpreter and protects against Ruby errors. The fourth file Rakefile builds the program.
# query.rb
require 'fiddle'
# Look for a C variable named QueryPointer.
# Raise an error if it is missing.
c_var = Fiddle.dlopen(nil)['QueryPointer']
int = Fiddle::TYPE_INT
voidp = Fiddle::TYPE_VOIDP
sz_voidp = Fiddle::SIZEOF_VOIDP
# Implement the C function
# int Query(void *data, size_t *length)
# in Ruby code. Store it in a global constant in Ruby (named Query)
# to protect it from Ruby's garbage collector.
#
Query = Fiddle::Closure::BlockCaller
.new(int, [voidp, voidp]) do |datap, lengthp|
message = "Here am I"
# We got datap and lengthp as Fiddle::Pointer objects.
# Read length, assuming sizeof(size_t) == sizeof(void *).
length = lengthp[0, sz_voidp].unpack('J').first
# Does the message fit in length bytes?
if length < message.bytesize
0 # failure
else
length = message.bytesize
datap[0, length] = message # Copy the message.
lengthp[0, sz_voidp] = [length].pack('J') # Update the length.
1 # success
end
end
# Set the C variable to our Query.
Fiddle::Pointer.new(c_var)[0, sz_voidp] = [Query.to_i].pack('J')
/* query-rb.c */
#include <stdlib.h>
#include <ruby.h>
/*
* QueryPointer() uses Ruby and may raise a Ruby error. Query() is a
* C wrapper around QueryPointer() that loads Ruby, sets QueryPointer,
* and protects against Ruby errors.
*/
int (*QueryPointer)(char *, size_t *) = NULL;
static int in_bad_exit = 0;
static void
do_at_exit(void)
{
RUBY_INIT_STACK;
if (!in_bad_exit)
ruby_cleanup(0);
}
static void
bad_exit(int state)
{
in_bad_exit = 1;
ruby_stop(state); /* Clean up Ruby and exit the process. */
}
static void
require_query(void)
{
static int done = 0;
int state;
if (done)
return;
done = 1;
ruby_init();
atexit(do_at_exit);
ruby_init_loadpath(); /* needed to require 'fiddle' */
/* Require query.rb in current directory. */
rb_eval_string_protect("require_relative 'query'", &state);
if (!state && !QueryPointer)
rb_eval_string_protect("fail 'missing QueryPointer'", &state);
if (state)
bad_exit(state); /* Ruby will report the error. */
}
struct args {
char *data;
size_t *length;
int result;
};
static VALUE
Query1(VALUE v) {
struct args *a = (struct args *)v;
a->result = QueryPointer(a->data, a->length);
return Qnil;
}
int
Query(char *data, size_t *length)
{
struct args a;
int state;
RUBY_INIT_STACK;
require_query();
/* Call QueryPointer(), protect against errors. */
a.data = data;
a.length = length;
rb_protect(Query1, (VALUE)&a, &state);
if (state)
bad_exit(state);
return a.result;
}
# Rakefile
# To build and run:
# $ rake
# $ ./callruby
# Must link with cc -Wl,-E so query.c exports QueryPointer.
CC = ENV.fetch('CC', 'cc')
LDFLAGS = '-Wl,-E'
CPPFLAGS = RbConfig.expand('-I$(rubyarchhdrdir) -I$(rubyhdrdir)')
LIBS = RbConfig.expand('$(LIBRUBYARG) $(LIBS)')
task 'default' => 'callruby'
desc 'compiles callruby'
file 'callruby' => %w[main.o query-rb.o] do |t|
sh "#{CC} #{LDFLAGS} -o #{t.name} #{t.sources.join(' ')} #{LIBS}"
end
rule '.o' => %w[.c] do |t|
sh "#{CC} #{CPPFLAGS} -o #{t.name} -c #{t.source}"
end
desc 'removes callruby and .o files'
task 'clean' do
rm_f %w[callruby main.o query-rb.o]
end
Rust
//! In order to run this task, you will need to compile the C program locating in the task linked
//! above. The C program will need to be linked with the library produced by this file.
//!
//! 1. Compile this library:
//!
//! ```bash
//! $ cargo build --release
//! ```
//!
//! 2. Copy the C program into query.c.
//! 3. Compile and link the C program with the produced library:
//!
//! ```bash
//! $ LD_LIBRARY_PATH=/path/to/library gcc query.c -o query -Wall -Werror libquery
//! ```
//! 4. Run the resulting binary.
//!
//! ```bash
//! $ LD_LIBRARY_PATH=/path/to/library ./query
//! Here am I
//! ```
#![crate_type = "cdylib"]
extern crate libc;
use std::ffi::CString;
use libc::{c_char, c_int, size_t};
#[no_mangle]
#[allow(non_snake_case)]
#[allow(clippy::missing_safety_doc)]
pub unsafe extern "C" fn Query(data: *mut c_char, length: *mut size_t) -> c_int {
let string = "Here am I";
if *length + 1 < string.len() {
0
} else {
let c_string = CString::new(string).unwrap();
libc::strcpy(data, c_string.as_ptr());
*length = string.len();
1
}
}
Scala
Using the JVM
We write a Scala function, then write a C function Query()
to use the Java Native Interface (JNI) to call our Scala method. The C compiler must find jni.h and jni_md.h and link with libjvm.
/* Query.scala */
object Query {
def call(data: Array[Byte], length: Array[Int]): Boolean = {
val message = "Here am I"
val mb = message.getBytes("utf-8")
if (length(0) >= mb.length) {
length(0) = mb.length
System.arraycopy(mb, 0, data, 0, mb.length)
true
} else false
}
}
/* query-jni.c */
#include <stdio.h>
#include <stdlib.h>
#include <jni.h>
static JavaVM *jvm = NULL;
static JNIEnv *jenv = NULL;
static void die(const char *message) {
fprintf(stderr, "%s\n", message);
exit(1);
}
static void oom(void) {
die("Query: out of memory");
}
static void except(void) {
if ((*jenv)->ExceptionCheck(jenv))
die("Query: unexpected Java exception");
}
static void do_at_exit(void) {
(*jvm)->DestroyJavaVM(jvm);
}
static void require_jvm(void) {
JavaVMInitArgs args;
if (jvm)
return;
args.version = JNI_VERSION_1_4;
args.nOptions = 0;
args.options = NULL;
args.ignoreUnrecognized = JNI_FALSE;
if (JNI_CreateJavaVM(&jvm, (void **)&jenv, &args) != JNI_OK)
die("Query: can't create Java VM");
atexit(do_at_exit);
}
int Query(char *data, size_t *length) {
jclass cQuery;
jmethodID mcall;
jintArray jlength;
jint jlength0;
jbyteArray jdata;
jboolean result;
jlength0 = (jint)length[0];
if ((size_t)jlength0 != length[0])
die("Query: length is too large for Scala array");
require_jvm();
/* Create a local frame for references to Scala objects. */
if ((*jenv)->PushLocalFrame(jenv, 16))
oom();
/* Look for class Query, static boolean call(byte[], int[]) */
cQuery = (*jenv)->FindClass(jenv, "Query");
if (cQuery == NULL)
die("Query: can't find Query.class");
mcall = (*jenv)->GetStaticMethodID(jenv, cQuery, "call", "([B[I)Z");
if (mcall == NULL)
die("Query: missing call() method");
/*
* Make arguments to Query.call(). We can't pass data[] and
* length[] to Scala, so we make new Scala arrays jdata[] and
* jlength[].
*/
jdata = (*jenv)->NewByteArray(jenv, (jsize)jlength0);
if (jdata == NULL)
oom();
jlength = (*jenv)->NewIntArray(jenv, 1);
if (jlength == NULL)
oom();
/* Set jlength[0] = length[0]. */
(*jenv)->SetIntArrayRegion(jenv, jlength, 0, 1, &jlength0);
except();
/*
* Call our Scala method.
*/
result = (*jenv)->CallStaticBooleanMethod
(jenv, cQuery, mcall, jdata, jlength);
except();
/*
* Set length[0] = jlength[0].
* Copy length[0] bytes from jdata[] to data[].
*/
(*jenv)->GetIntArrayRegion(jenv, jlength, 0, 1, &jlength0);
except();
length[0] = (size_t)jlength0;
(*jenv)->GetByteArrayRegion
(jenv, jdata, 0, (jsize)jlength0, (jbyte *)data);
/* Drop our local frame and its references. */
(*jenv)->PopLocalFrame(jenv, NULL);
return (int)result;
}
# Makefile
# Edit these lines to match your JDK.
JAVA_HOME = /Library/Java/Home
CPPFLAGS = -I$(JAVA_HOME)/include
LIBS = -framework JavaVM
JAVAC = $(JAVA_HOME)/bin/javac
CC = cc
all: calljava Query.class
calljava: main.o query-jni.o
$(CC) -o calljava main.o query-jni.o $(LIBS)
.SUFFIXES: .c .class .java .o
.c.o:
$(CC) $(CPPFLAGS) -c $<
.java.class:
$(JAVAC) $<
clean:
rm -f calljava main.o query-jni.o Query.class
Tcl
The way you would tackle this problem depends on whether you are working with ‘In’ or ‘Out’ parameters. (It is normal model ‘inout’ parameters as Tcl variables; omitted for brevity.)
‘In’ Parameters
To connect a function to Tcl that passes an arbitrary C string as input, you'd use a short C thunk, like this:
int Query (char * Data, size_t * Length) {
Tcl_Obj *arguments[2];
int code;
arguments[0] = Tcl_NewStringObj("Query", -1); /* -1 for "use up to zero byte" */
arguments[1] = Tcl_NewStringObj(Data, Length);
Tcl_IncrRefCount(arguments[0]);
Tcl_IncrRefCount(arguments[1]);
if (Tcl_EvalObjv(interp, 2, arguments, 0) != TCL_OK) {
/* Was an error or other exception; report here... */
Tcl_DecrRefCount(arguments[0]);
Tcl_DecrRefCount(arguments[1]);
return 0;
}
Tcl_DecrRefCount(arguments[0]);
Tcl_DecrRefCount(arguments[1]);
if (Tcl_GetObjResult(NULL, Tcl_GetObjResult(interp), &code) != TCL_OK) {
/* Not an integer result */
return 0;
}
return code;
}
Which would lead to a Query
implementation like this:
proc Query data {
puts "Query was $data"
return 1;
}
‘Out’ Parameters
However, in the specific case of writing to a user-specified buffer (an “out” parameter) the thunk code would instead manage copying the result from the interpreter back to the buffer:
int Query (char * Data, size_t * Length) {
const char *str;
int len;
if (Tcl_Eval(interp, "Query") != TCL_OK) {
return 0;
}
str = Tcl_GetStringFromObj(Tcl_GetObjResult(interp), &len);
if (len+1 > Length) {
return 0;
}
memcpy(Data, str, len+1);
return 1;
}
And the implementation of Query
would be just:
proc Query {} {
return "Here am I"
}
(Since this is working with a literal, this would actually be efficient and just result in references being passed.)
Connecting up the pieces
You would also need a short piece of code in main()
to initialize the Tcl library and create an interpreter instance, and you would need to build and link against libtcl.
#include <tcl.h>
Tcl_Interp *interp;
int main(int argc, char **argv) {
Tcl_FindExecutable(argv[0]); /* Initializes library */
interp = Tcl_CreateInterp(); /* Make an interpreter */
/* Rest of contents of main() from task header... */
}
TXR
This is really two tasks: how to accept foreign callbacks, and how to link code to a C program which controls the main
startup function.
The TXR run-time is not available as a library that can be linked to a C program. Instead, we can put the C driver into a small library and call out to it from TXR, then accept its callback. Here is that library:
#include <stdio.h>
int query(int (*callback)(char *, size_t *))
{
char buffer[1024];
size_t size = sizeof buffer;
if (callback(buffer, &size) == 0) {
puts("query: callback failed");
} else {
char *ptr = buffer;
while (size-- > 0)
putchar (*ptr++);
putchar('\n');
}
}
Here are the build steps to produce a `query.so` object from it on GNU/Linux:
gcc -g -fPIC query.c -c
gcc -g --shared query.c -o query.c
Using carray
In this situation, the most appropriate FFI type to use for the foreign buffer is the carray
type. This type allows TXR Lisp code to manipulate a foreign array while retaining its identity, so that it is able to pass the same pointer to the foreign code that it received from that code. carray
also solves the problem of dealing with the common representational approach in C when arrays are represented by pointers, and do not include their size as part of their type information. A carray
object can be constructed with an zero size, which can be adjusted when the size is known, using carray-set-length
.
Like the array
type, carray
has specialized behaviors when its element type is char
, bchar
or wchar
. The carray-get
function will decode a string from the underlying array, and carray-put
will encode a string into the array. In the case of the char
type, this involves UTF-8 coding.
Callbacks are modeled as "FFI closures". The macro deffi-cb
defines a function which itself isn't a callback, but is rather a combinator which converts a Lisp function into a FFI callback.
(with-dyn-lib "./query.so"
(deffi query "query" void (closure)))
(deffi-cb query-cb int ((carray char) (ptr (array 1 size-t))))
(query (query-cb (lambda (buf sizeptr)
(symacrolet ((size [sizeptr 0]))
(let* ((s "Here am I")
(l (length s)))
(cond
((> l size) 0)
(t (carray-set-length buf size)
(carray-put buf s)
(set size l))))))))
- Output:
Here am I
Note that the obvious way of passing a size_t
value by pointer, namely (ptr size-t)
doesn't work. While the callback will receive the size (FFI will decode the pointer type's semantics and get the size value), updating the size will not propagate back to the caller, because it becomes, effectively, a by-value parameter. A (ptr size-t)
object has to be embedded in an aggregate that is passed by reference, in order to have two-way semantics. Here we use the trick of treating the size_t *
as an array of 1, which it de facto is. In the callback, we establish local symbol macro which lets us just refer to [sizeptr 0]
it as size
.
Using cptr
and memcpy
An alternative approach is possible if we avail ourselves of the memcpy
function via FFI. We can receive the data as an opaque foreign pointer represented by the cptr
type. We can set up memcpy
so that its destination argument and return value is a cptr
, but the source argument is a string:
(with-dyn-lib "./query.so"
(deffi query "query" void (closure)))
(with-dyn-lib nil
(deffi memcpy "memcpy" cptr (cptr str size-t)))
(deffi-cb query-cb int (cptr (ptr (array 1 size-t))))
(query (query-cb (lambda (buf sizeptr) ; int lambda(void *buf, siz
(symacrolet ((size [sizeptr 0])) ; { #define size sizeptr[0]
(let* ((s "Here am I") ; char *s = "Here am I";
(l (length s))) ; size_t l = strlen(s);
(cond ; if (length > size)
((> l size) 0) ; { return 0; } else
(t (memcpy buf s l) ; { memcpy(buf, s, l);
(set size l)))))))) ; return size = l; } }
Here, the use of the str
type in the memcpy
interface means that FFI automatically produces a UTF-8 encoding of the string in a temporary buffer. The pointer to that temporary buffer is what is passed into memcpy
. The temporary buffer is released after memcpy
returns.
To reveal the similarity between the Lisp logic and how a C function might be written, the corresponding C code is shown. However, that C code's semantics is, of course, devoid of any hidden UTF-8 conversion.
Exceptions from Callback
If the callback throws an exception or performs any other non-local return, it will return a default return value of all zero bits in the given return type. This value can be specified, but the zero default suits our particular situation, because the problem task defines the return value of zero as an error indicator.
We can explore this interactively:
$ txr This is the TXR Lisp interactive listener of TXR 177. Use the :quit command or type Ctrl-D on empty line to exit. 1> (with-dyn-lib "./query.so" (deffi query "query" void (closure))) #:lib-0177 2> (deffi-cb query-cb int ((ptr (array 1024 char)) (ptr size-t))) query-cb 3> (query (query-cb (lambda (x y) (error "oops")))) query: callback failed ** oops ** during evaluation at expr-3:1 of form (error "oops") 4>
Here we can see that when the callback throws the error
exception, the C code prints query: callback failed
, due to receiving the default abort return value of zero. Then, the exception continues up to the interactive prompt.
If a return value other than zero indicates that the callback failed, that can be arranged with an additional argument in deffi-cb
:
(deffi-cb query-cb int (cptr (ptr (array 1 size-t))) -1)
Now the query-cb
function generates callbacks that return -1 to the caller, rather than zero, if aborted by a non-local control transfer such as an exception.
Wren
Currently, the only way to call a Wren function from another language is to embed it in a program written in that language. Therefore, we embed this Wren script:
/* Use_another_language_to_call_a_function.wren */
class RCQuery {
// Both arguments are lists as we need pass by reference here
static query(Data, Length) {
var s = "Here am I"
var sc = s.count
if (sc > Length[0]) return 0 // buffer too small
for (i in 0...sc) Data[i] = s[i].bytes[0]
Length[0] = sc
return 1
}
}
in the following C program, compile and run:
/* gcc Use_another_language_to_call_a_function.c -o Use_another_language_to_call_a_function -lwren -lm */
#include <stdio.h>
#include "wren.h"
char *script;
WrenVM * vm;
WrenForeignMethodFn bindForeignMethod(
WrenVM* vm,
const char* module,
const char* className,
bool isStatic,
const char* signature) {
return NULL; // nothing needed here
}
static void writeFn(WrenVM* vm, const char* text) {
printf("%s", text);
}
void errorFn(WrenVM* vm, WrenErrorType errorType, const char* module, const int line, const char* msg) {
switch (errorType) {
case WREN_ERROR_COMPILE:
printf("[%s line %d] [Error] %s\n", module, line, msg);
break;
case WREN_ERROR_STACK_TRACE:
printf("[%s line %d] in %s\n", module, line, msg);
break;
case WREN_ERROR_RUNTIME:
printf("[Runtime Error] %s\n", msg);
break;
}
}
char *readFile(const char *fileName) {
FILE *f = fopen(fileName, "r");
fseek(f, 0, SEEK_END);
long fsize = ftell(f);
rewind(f);
char *script = malloc(fsize + 1);
fread(script, 1, fsize, f);
fclose(f);
script[fsize] = 0;
return script;
}
int configWrenVM() {
WrenConfiguration config;
wrenInitConfiguration(&config);
config.writeFn = &writeFn;
config.errorFn = &errorFn;
config.bindForeignMethodFn = &bindForeignMethod;
vm = wrenNewVM(&config);
const char* module = "main";
const char* fileName = "Use_another_language_to_call_a_function.wren";
script = readFile(fileName);
WrenInterpretResult result = wrenInterpret(vm, module, script);
switch (result) {
case WREN_RESULT_COMPILE_ERROR:
printf("Compile Error!\n");
return -1;
case WREN_RESULT_RUNTIME_ERROR:
printf("Runtime Error!\n");
return -1;
case WREN_RESULT_SUCCESS:
break;
}
return 0;
}
int Query(char *Data, size_t *Length) {
int i, r;
wrenEnsureSlots(vm, 4);
// create list for Data, fill with zeros and put in slot 1
wrenSetSlotNewList(vm, 1);
wrenSetSlotDouble(vm, 2, 0.0);
for (i = 0; i < *Length; ++i) wrenInsertInList(vm, 1, i, 2);
// create list for Length and put in slot 2
wrenSetSlotNewList(vm, 2);
wrenSetSlotDouble(vm, 3, (double)*Length);
wrenInsertInList(vm, 2, 0, 3);
// get handle to Wren's query method
WrenHandle* method = wrenMakeCallHandle(vm, "query(_,_)");
// get its class and put in slot 0
wrenGetVariable(vm, "main", "RCQuery", 0);
// call the Wren method
wrenCall(vm, method);
// get the result and check it's 1
r = (int)wrenGetSlotDouble(vm, 0);
if (r) {
// get the length of the string from slot 2
wrenGetListElement(vm, 2, 0, 3);
*Length = (int)wrenGetSlotDouble(vm, 3);
// copy the bytes from the list in slot 1 to the C buffer
for (i = 0; i < *Length; ++i) {
wrenGetListElement(vm, 1, i, 3);
Data[i] = (char)wrenGetSlotDouble(vm, 3);
}
}
return r;
}
int main() {
int e = configWrenVM();
if (!e) {
char Buffer [1024];
size_t Size = sizeof(Buffer);
if (0 == Query(Buffer, &Size)) {
printf ("failed to call Query\n");
e = 1;
} else {
char * Ptr = Buffer;
while (Size-- > 0) putchar (*Ptr++);
putchar ('\n');
}
}
wrenFreeVM(vm);
free(script);
return e;
}
- Output:
Here am I
X86-64 Assembly
UASM 2.52
option casemap:none
strlen proto :qword
strncpy proto :qword, :qword, :dword
Query proto :qword, :qword
.data
szstr db "Here am I",0
.code
Query proc Data:qword, len:qword
local d:qword, l:qword, s:dword
mov d, Data
mov l, len
invoke strlen, addr szstr
.if rax <= l
mov s, eax
invoke strncpy, d, addr szstr, s
mov eax, s
mov rax, l
mov dword ptr [rax], ecx
mov rax, 1
ret
.endif
mov rax, 0
ret
Query endp
end
NASM
section .data
szmsg db "Here I am",0
section .text
global Query
strlen:
push rbp
mov rbp, rsp
mov rsi, rdi
mov rcx, -1
_1:
inc rcx
cmp byte [rsi+rcx], 0
jne _1
mov rax, rcx
pop rbp
ret
Query:
push rbp
mov rbp, rsp
;;mov r9, rcx ;;Arg 1, windows
;;mov r8, rdx ;;Arg 2, windows
mov r9, rdi ;;Arg 1, Linux
mov r8, rsi ;;Arg 2, Linux
lea rdi, szmsg
call strlen
cmp rax, r8
jg _err
mov r10d, eax
mov rdi, r9
lea rsi, szmsg
rep movsb
mov rax, r8
mov dword [rax], r10d
jmp _exit
_err:
mov rax, 0
_exit:
pop rbp
ret
Zig
const std = @import("std");
export fn Query(Data: [*c]u8, Length: *usize) callconv(.C) c_int {
const value = "Here I am";
if (Length.* >= value.len) {
@memcpy(@ptrCast([*]u8, Data), value, value.len);
Length.* = value.len;
return 1;
}
return 0;
}
zkl
To make this as simple as possible, the [zkl] query program sets a variable and main.c runs query.zkl and extracts the variable. A more realistic scenario (which several of the extension libraries utilize) is to compile the zkl code, wad it into C code (a byte stream of the compiled code) and link that with main. Not hard but messy (the source of a suitable extension gives you something to copy). Also, this solution uses the shared library version of zkl (you could use the all in one version but you would go about it in a [slightly] different way).
Modified main.c:
// query.c
// export zklRoot=/home/ZKL
// clang query.c -I $zklRoot/VM -L $zklRoot/Lib -lzkl -pthread -lncurses -o query
// LD_LIBRARY_PATH=$zklRoot/Lib ./query
#include <stdio.h>
#include <string.h>
#include "zklObject.h"
#include "zklImports.h"
#include "zklClass.h"
#include "zklFcn.h"
#include "zklString.h"
int query(char *buf, size_t *sz)
{
Instance *r;
pVM vm;
MLIST(mlist,10);
// Bad practice: not protecting things from the garbage collector
// build the call parameters: ("query.zkl",False,False,True)
mlistBuild(mlist,stringCreate("query.zkl",I_OWNED,NoVM),
BoolFalse,BoolFalse,BoolTrue,ZNIL);
// Import is in the Vault, a store of useful stuff
// We want to call TheVault.Import.import("query.zkl",False,False,True)
// which will load/compile/run query.zkl
r = fcnRunith("Import","import",(Instance *)mlist,NoVM);
// query.zkl is a class with a var that has the query result
r = classFindVar(r,"query",0,NoVM); // -->the var contents
strcpy(buf,stringText(r)); // decode the string into a char *
*sz = strlen(buf); // screw overflow checking
return 1;
}
int main(int argc, char* argv[])
{
char buf[100];
size_t sz = sizeof(buf);
zklConstruct(argc,argv); // initialize the zkl shared library
query(buf,&sz);
printf("Query() --> \"%s\"\n",buf);
return 0;
}
Our query program:
// query.zkl
var query="Here am I";
On Linux:
- Output:
$ clang query.c -I $zklRoot/VM -L $zklRoot/Lib -lzkl -pthread -lncurses -o query $ LD_LIBRARY_PATH=$zklRoot/Lib ./query Query() --> "Here am I"
- Programming Tasks
- Programming environment operations
- Clarified and Needing Review
- Ada
- AutoHotkey
- C
- C++
- COBOL
- D
- Delphi
- Fortran
- Go
- Haskell
- Haxe
- J
- Java
- Kotlin
- Lisaac
- Mercury
- Nim
- OCaml
- Ol
- PARI/GP
- Pascal
- Phix
- PicoLisp
- Python
- Racket
- Raku
- Ruby
- Rust
- Scala
- Tcl
- TXR
- Wren
- X86-64 Assembly
- Zig
- Zkl
- Batch File/Omit
- GUISS/Omit
- Lotus 123 Macro Scripting/Omit
- Retro/Omit
- TI-83 BASIC/Omit
- TI-89 BASIC/Omit