CFFI User Manual

Table of Contents


Next: , Up: (dir)

cffi

Copyright © 2005 James Bielman <jamesjb at jamesjb.com>
Copyright © 2005, 2006 Luís Oliveira <loliveira at common-lisp.net>
Copyright © 2006 Stephen Compall <s11 at member.fsf.org>

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

The software is provided “as is”, without warranty of any kind, express or implied, including but not limited to the warranties of merchantability, fitness for a particular purpose and noninfringement. In no event shall the authors or copyright holders be liable for any claim, damages or other liability, whether in an action of contract, tort or otherwise, arising from, out of or in connection with the software or the use or other dealings in the software.

--- Dictionary ---

Foreign Types

Pointers

Strings

Variables

Functions

Libraries

Callbacks


Next: , Previous: Top, Up: Top

1 Introduction

CFFI is the Common Foreign Function Interface for ANSI Common Lisp systems. By foreign function we mean a function written in another programming language and having different data and calling conventions than Common Lisp, namely, C. CFFI allows you to call foreign functions and access foreign variables, all without leaving the Lisp image.

We consider this manual ever a work in progress. If you have difficulty with anything CFFI-specific presented in the manual, please contact the developers with details.

Motivation

See What makes Lisp different, for an argument in favor of FFI in general.

CFFI's primary role in any image is to mediate between Lisp developers and the widely varying FFIs present in the various Lisp implementations it supports. With CFFI, you can define foreign function interfaces while still maintaining portability between implementations. It is not the first Common Lisp package with this objective; however, it is meant to be a more malleable framework than similar packages.

Design Philosophy


Next: , Previous: Introduction, Up: Top

2 Implementation Support

CFFI supports various free and commercial Lisp implementations: Allegro CL, Corman CL, clisp, CMUCL, ECL, LispWorks, OpenMCL, SBCL and the Scieneer CL.

There are also plans to support Digitool MCL, and GCL.

2.1 Allegro CL

Tested platforms: linux/x86, linux/ppc, win32/x86, darwin/ppc.

Version 7.0 is supported. The 8.0 beta is also known to work. Earlier versions are untested and unsupported but patches to support them are welcome.

Limitations

2.2 Corman CL

Tested platforms: win32/x86.

Versions prior to 2.51 are untested and unsupported. Also, you will need to avoid Corman's buggy COMPILE-FILE and fasl loader. Please follow these instructions by Edi Weitz to setup ASDF for Corman CL in a way that works around these issues.

Limitations

2.3 clisp

Tested platforms: linux/x86, linux/ppc, win32/x86, darwin/ppc.

Version is 2.34 or newer is required on win32/x86. For other platforms version 2.35 or newer is required.

2.4 CMUCL

Tested platforms: linux/x86, darwin/ppc.

Versions prior to 19B are untested. For darwin/ppc, the 2006-02 (19C) snapshot or later is recommended.

2.5 ECL

Tested platforms: needs testing...

As of November 2005, the CVS version of ECL is required. It is reported to pass all tests.

Limitations

2.6 Lispworks

Tested platforms: linux/x86, win32/x86, darwin/ppc.

Versions prior to 4.4 are untested.

Limitations

2.7 OpenMCL

Tested platforms: darwin/ppc, linux/ppc.

OpenMCL 1.0 or newer is recommended.

2.8 SBCL

Tested platforms: linux/x86, linux/ppc, darwin/ppc.

Version 0.9.6 or newer is recommended.

Limitations

2.9 Scieneer CL

Tested platforms: linux/x86, linux/amd64.

Version 1.2.10 or newer is recommended. Passes all tests. The x86 and AMD64 ports feature long-double support.


Next: , Previous: Implementation Support, Up: Top

3 An Introduction to Foreign Interfaces and CFFI

Users of many popular languages bearing semantic similarity to Lisp, such as Perl and Python, are accustomed to having access to popular C libraries, such as GTK, by way of “bindings”. In Lisp, we do something similar, but take a fundamentally different approach. This tutorial first explains this difference, then explains how you can use CFFI, a powerful system for calling out to C and C++ and access C data from many Common Lisp implementations.

The concept can be generalized to other languages; at the time of writing, only CFFI's C support is fairly complete, but C++ support is being worked on. Therefore, we will interchangeably refer to foreign functions and foreign data, and “C functions” and “C data”. At no time will the word “foreign” carry its usual, non-programming meaning.

This tutorial expects you to have a working understanding of both Common Lisp and C, including the Common Lisp macro system.


Next: , Up: Tutorial

3.1 What makes Lisp different

The following sums up how bindings to foreign libraries are usually implemented in other languages, then in Common Lisp:

Perl, Python, Java, other one-implementation languages
Bindings are implemented as shared objects written in C. In some cases, the C code is generated by a tool, such as SWIG, but the result is the same: a new C library that manually translates between the language implementation's objects, such as PyObject in Python, and whatever C object is called for, often using C functions provided by the implementation. It also translates between the calling conventions of the language and C.
Common Lisp
Bindings are written in Lisp. They can be created at-will by Lisp programs. Lisp programmers can write new bindings and add them to the image, using a listener such as SLIME, as easily as with regular Lisp definitions. The only foreign library to load is the one being wrapped—the one with the pure C interface; no C or other non-Lisp compilation is required.

We believe the advantages of the Common Lisp approach far outweigh any disadvantages. Incremental development with a listener can be as productive for C binding development as it is with other Lisp development. Keeping it “in the [Lisp] family”, as it were, makes it much easier for you and other Lisp programmers to load and use the bindings. Common Lisp implementations such as CMUCL, freed from having to provide a C interface to their own objects, are thus freed to be implemented in another language (as CMUCL is) while still allowing programmers to call foreign functions.

Perhaps the greatest advantage is that using an FFI doesn't obligate you to become a professional binding developer. Writers of bindings for other languages usually end up maintaining or failing to maintain complete bindings to the foreign library. Using an FFI, however, means if you only need one or two functions, you can write bindings for only those functions, and be assured that you can just as easily add to the bindings if need be.

The removal of the C compiler, or C interpretation of any kind, creates the main disadvantage: some of C's “abstractions” are not available, violating information encapsulation. For example, structs that must be passed on the stack, or used as return values, without corresponding functional abstractions to create and manage the structs, must be declared explicitly in Lisp. This is fine for structs whose contents are “public”, but is not so pleasant when a struct is supposed to be “opaque” by convention, even though it is not so defined.1

Without an abstraction to create the struct, Lisp needs to be able to lay out the struct in memory, so must know its internal details.

In these cases, you can create a minimal C library to provide the missing abstractions, without destroying all the advantages of the Common Lisp approach discussed above. In the case of structs, you can write simple, pure C functions that tell you how many bytes a struct requires or allocate new structs, read and write fields of the struct, or whatever operations are supposed to be public.2

Implementor's note: cffi-grovel, a project not yet part of CFFI, automates this and other processes.

Another disadvantage appears when you would rather use the foreign language than Lisp. However, someone who prefers C to Lisp is not a likely candidate for developing a Lisp interface to a C library.


Next: , Previous: Tutorial-Comparison, Up: Tutorial

3.2 Getting a URL

The widely available libcurl is a library for downloading files over protocols like HTTP. We will use libcurl with CFFI to download a web page.

Please note that there are many other ways to download files from the web, not least the cl-curl project to provide bindings to libcurl via a similar FFI.3

libcurl-tutorial(3) is a tutorial for libcurl programming in C. We will follow that to develop a binding to download a file. We will also use curl.h, easy.h, and the man pages for the libcurl function, all available in the `curl-dev' package or equivalent for your system, or in the cURL source code package. If you have the development package, the headers should be installed in /usr/include/curl/, and the man pages may be accessed through your favorite man facility.


Next: , Previous: Tutorial-Getting a URL, Up: Tutorial

3.3 Loading foreign libraries

First of all, we will create a package to work in. You can save these forms in a file, or just send them to the listener as they are. If creating bindings for an ASDF package of yours, you will want to add :cffi to the :depends-on list in your .asd file. Otherwise, just use the asdf:oos function to load CFFI.

  (asdf:oos 'asdf:load-op :cffi)
   
  ;;; Nothing special about the "CFFI-USER" package.  We're just
  ;;; using it as a substitute for your own CL package.
  (defpackage :cffi-user
    (:use :common-lisp :cffi))
   
  (in-package :cffi-user)
   
  (define-foreign-library libcurl
    (:unix (:or "libcurl.so.3" "libcurl.so"))
    (t (:default "libcurl")))
   
  (use-foreign-library libcurl)

Using define-foreign-library and use-foreign-library, we have loaded libcurl into Lisp, much as the linker does when you start a C program, or common-lisp:load does with a Lisp source file or FASL file. We special-cased for unix machines to always load a particular version, the one this tutorial was tested with; for those who don't care, the define-foreign-library clause (t (:default "libcurl")) should be satisfactory, and will adapt to various operating systems.


Next: , Previous: Tutorial-Loading, Up: Tutorial

3.4 Initializing libcurl

After the introductory matter, the tutorial goes on to present the first function you should use.

  CURLcode curl_global_init(long flags);

Let's pick this apart into appropriate Lisp code:

  ;;; A CURLcode is the universal error code.  curl/curl.h says
  ;;; no return code will ever be removed, and new ones will be
  ;;; added to the end.
  (defctype curl-code :int)
   
  ;;; Initialize libcurl with FLAGS.
  (defcfun "curl_global_init" curl-code
    (flags :long))
Implementor's note: CFFI currently assumes the UNIX viewpoint that there is one C symbol namespace, containing all symbols in all loaded objects. This is not so on Windows and Darwin. The interface may be changed to deal with this.

Note the parallels with the original C declaration. We've defined curl-code as a wrapping type for :int; right now, it only marks it as special, but later we will do something more interesting with it. The point is that we don't have to do it yet.

Looking at curl.h, CURL_GLOBAL_NOTHING, a possible value for flags above, is defined as `0'. So we can now call the function:

  cffi-user> (curl-global-init 0)
  => 0

Looking at curl.h again, 0 means CURLE_OK, so it looks like the call succeeded. Note that CFFI converted the function name to a Lisp-friendly name. You can specify your own name if you want; use ("curl_global_init" your-name-here) as the name argument to defcfun.

The tutorial goes on to have us allocate a handle. For good measure, we should also include the deallocator. Let's look at these functions:

  CURL *curl_easy_init( );
  void curl_easy_cleanup(CURL *handle);

Advanced users may want to define special pointer types; we will explore this possibility later. For now, just treat every pointer as the same:

  (defcfun "curl_easy_init" :pointer)
   
  (defcfun "curl_easy_cleanup" :void
    (easy-handle :pointer))

Now we can continue with the tutorial:

  cffi-user> (defparameter *easy-handle* (curl-easy-init))
  => *EASY-HANDLE*
  cffi-user> *easy-handle*
  => #<FOREIGN-ADDRESS #x09844EE0>

Note the print representation of a pointer. It changes depending on what Lisp you are using, but that doesn't make any difference to CFFI.


Next: , Previous: Tutorial-Initializing, Up: Tutorial

3.5 Setting download options

The libcurl tutorial says we'll want to set many options before performing any download actions. This is done through curl_easy_setopt:

  CURLcode curl_easy_setopt(CURL *curl, CURLoption option, ...);

We've introduced a new twist: variable arguments. There is no obvious translation to the defcfun form, particularly as there are four possible argument types. Because of the way C works, we could define four wrappers around curl_easy_setopt, one for each type; in this case, however, we'll use the general-purpose macro foreign-funcall to call this function.

To make things easier on ourselves, we'll create an enumeration of the kinds of options we want to set. The enum CURLoption isn't the most straightforward, but reading the CINIT C macro definition should be enlightening.

  (defmacro define-curl-options (name type-offsets &rest enum-args)
    "As with CFFI:DEFCENUM, except each of ENUM-ARGS is as follows:
   
      (NAME TYPE NUMBER)
   
  Where the arguments are as they are with the CINIT macro defined
  in curl.h, except NAME is a keyword.
   
  TYPE-OFFSETS is a plist of TYPEs to their integer offsets, as
  defined by the CURLOPTTYPE_LONG et al constants in curl.h."
    (flet ((enumerated-value (type offset)
             (+ (getf type-offsets type) offset)))
      `(progn
         (defcenum ,name
           ,@(loop for (name type number) in enum-args
                collect (list name (enumerated-value type number))))
         ',name)))                ;for REPL users' sanity
   
  (define-curl-options curl-option
      (long 0 objectpoint 10000 functionpoint 20000 off-t 30000)
    (:noprogress long 43)
    (:nosignal long 99)
    (:errorbuffer objectpoint 10)
    (:url objectpoint 2))

With some well-placed Emacs query-replace-regexps, you could probably similarly define the entire CURLoption enumeration. I have selected to transcribe a few that we will use in this tutorial.

If you're having trouble following the macrology, just macroexpand the curl-option definition, or see the following macroexpansion, conveniently downcased and reformatted:

  (progn
    (defcenum curl-option
      (:noprogress 43)
      (:nosignal 99)
      (:errorbuffer 10010)
      (:url 10002))
    'curl-option)

That seems more than reasonable. You may notice that we only use the type to compute the real enumeration offset; we will also need the type information later.

First, however, let's make sure a simple call to the foreign function works:

  cffi-user> (foreign-funcall "curl_easy_setopt"
                 :pointer *easy-handle*
                 curl-option :nosignal :long 1 curl-code)
  => 0

foreign-funcall, despite its surface simplicity, can be used to call any C function. Its first argument is a string, naming the function to be called. Next, for each argument, we pass the name of the C type, which is the same as in defcfun, followed by a Lisp object representing the data to be passed as the argument. The final argument is the return type, for which we use the curl-code type defined earlier.

defcfun just puts a convenient façade on foreign-funcall.4 Our earlier call to curl-global-init could have been written as follows:

  cffi-user> (foreign-funcall "curl_global_init" :long 0
                              curl-code)
  => 0

Before we continue, we will take a look at what CFFI can and can't do, and why this is so.


Next: , Previous: Tutorial-easy_setopt, Up: Tutorial

3.6 Breaking the abstraction

In What makes Lisp different, we mentioned that writing an FFI sometimes requires depending on information not provided as part of the interface. The easy option CURLOPT_WRITEDATA, which we will not provide as part of the Lisp interface, illustrates this issue.

Strictly speaking, the curl-option enumeration is not necessary; we could have used :int 99 instead of curl-option :nosignal in our call to curl_easy_setopt above. We defined it anyway, in part to hide the fact that we are breaking the abstraction that the C enum provides. If the cURL developers decide to change those numbers later, we must change the Lisp enumeration, because enumeration values are not provided in the compiled C library, libcurl.so.3.

CFFI works because the most useful things in C libraries — non-static functions and non-static variables — are included accessibly in libcurl.so.3. A C compiler that violated this would be considered a worthless compiler.

The other thing define-curl-options does is give the “type” of the third argument passed to curl_easy_setopt. Using this information, we can tell that the :nosignal option should accept a long integer argument. We can implicitly assume t == 1 and nil == 0, as it is in C, which takes care of the fact that CURLOPT_NOSIGNAL is really asking for a boolean.

The “type” of CURLOPT_WRITEDATA is objectpoint. However, it is really looking for a FILE*. CURLOPT_ERRORBUFFER is looking for a char*, so there is no obvious CFFI type but :pointer.

The first thing to note is that nowhere in the C interface includes this information; it can only be found in the manual. We could disjoin these clearly different types ourselves, by splitting objectpoint into filepoint and charpoint, but we are still breaking the abstraction, because we have to augment the entire enumeration form with this additional information.5

The second is that the CURLOPT_WRITEDATA argument is completely incompatible with the desired Lisp data, a stream.6 It is probably acceptable if we are controlling every file we might want to use as this argument, in which case we can just call the foreign function fopen. Regardless, though, we can't write to arbitrary streams, which is exactly what we want to do for this application.

Finally, note that the curl_easy_setopt interface itself is a hack, intended to work around some of the drawbacks of C. The definition of Curl_setopt, while long, is far less cluttered than the equivalent disjoint-function set would be; in addition, setting a new option in an old libcurl can generate a run-time error rather than breaking the compile. Lisp can just as concisely generate functions as compare values, and the “undefined function” error is just as useful as any explicit error we could define here might be.


Next: , Previous: Tutorial-Abstraction, Up: Tutorial

3.7 Option functions in Lisp

We could use foreign-funcall directly every time we wanted to call curl_easy_setopt. However, we can encapsulate some of the necessary information with the following.

  ;;; We will use this type later in a more creative way.  For
  ;;; now, just consider it a marker that this isn't just any
  ;;; pointer.
  (defctype easy-handle :pointer)
   
  (defmacro curl-easy-setopt (easy-handle enumerated-name
                              value-type new-value)
    "Call `curl_easy_setopt' on EASY-HANDLE, using ENUMERATED-NAME
  as the OPTION.  VALUE-TYPE is the CFFI foreign type of the third
  argument, and NEW-VALUE is the Lisp data to be translated to the
  third argument.  VALUE-TYPE is not evaluated."
    `(foreign-funcall "curl_easy_setopt" easy-handle ,easy-handle
                      curl-option ,enumerated-name
                      ,value-type ,new-value curl-code))

Now we define a function for each kind of argument that encodes the correct value-type in the above. This can be done reasonably in the define-curl-options macroexpansion; after all, that is where the different options are listed!

We could make cl:defun forms in the expansion that simply call curl-easy-setopt; however, it is probably easier and clearer to use defcfun. define-curl-options was becoming unwieldy, so I defined some helpers in this new definition.

  (defun curry-curl-option-setter (function-name option-keyword)
    "Wrap the function named by FUNCTION-NAME with a version that
  curries the second argument as OPTION-KEYWORD.
   
  This function is intended for use in DEFINE-CURL-OPTION-SETTER."
    (setf (symbol-function function-name)
            (let ((c-function (symbol-function function-name)))
              (lambda (easy-handle new-value)
                (funcall c-function easy-handle option-keyword
                         new-value)))))
   
  (defmacro define-curl-option-setter (name option-type
                                       option-value foreign-type)
    "Define (with DEFCFUN) a function NAME that calls
  curl_easy_setopt.  OPTION-TYPE and OPTION-VALUE are the CFFI
  foreign type and value to be passed as the second argument to
  easy_setopt, and FOREIGN-TYPE is the CFFI foreign type to be used
  for the resultant function's third argument.
   
  This macro is intended for use in DEFINE-CURL-OPTIONS."
    `(progn
       (defcfun ("curl_easy_setopt" ,name) curl-code
         (easy-handle easy-handle)
         (option ,option-type)
         (new-value ,foreign-type))
       (curry-curl-option-setter ',name ',option-value)))
   
  (defmacro define-curl-options (type-name type-offsets &rest enum-args)
    "As with CFFI:DEFCENUM, except each of ENUM-ARGS is as follows:
   
      (NAME TYPE NUMBER)
   
  Where the arguments are as they are with the CINIT macro defined
  in curl.h, except NAME is a keyword.
   
  TYPE-OFFSETS is a plist of TYPEs to their integer offsets, as
  defined by the CURLOPTTYPE_LONG et al constants in curl.h.
   
  Also, define functions for each option named
  set-`TYPE-NAME'-`OPTION-NAME', where OPTION-NAME is the NAME from
  the above destructuring."
    (flet ((enumerated-value (type offset)
             (+ (getf type-offsets type) offset))
           ;;map PROCEDURE, destructuring each of ENUM-ARGS
           (map-enum-args (procedure)
             (mapcar (lambda (arg) (apply procedure arg)) enum-args))
           ;;build a name like SET-CURL-OPTION-NOSIGNAL
           (make-setter-name (option-name)
             (intern (concatenate
                      'string "SET-" (symbol-name type-name)
                      "-" (symbol-name option-name)))))
      `(progn
         (defcenum ,type-name
           ,@(map-enum-args
              (lambda (name type number)
                (list name (enumerated-value type number)))))
         ,@(map-enum-args
            (lambda (name type number)
              (declare (ignore number))
              `(define-curl-option-setter ,(make-setter-name name)
                 ,type-name ,name ,(ecase type
                                     (long :long)
                                     (objectpoint :pointer)
                                     (functionpoint :pointer)
                                     (off-t :long)))))
         ',type-name)))

Macroexpanding our define-curl-options form once more, we see something different:

  (progn
    (defcenum curl-option
      (:noprogress 43)
      (:nosignal 99)
      (:errorbuffer 10010)
      (:url 10002))
    (define-curl-option-setter set-curl-option-noprogress
      curl-option :noprogress :long)
    (define-curl-option-setter set-curl-option-nosignal
      curl-option :nosignal :long)
    (define-curl-option-setter set-curl-option-errorbuffer
      curl-option :errorbuffer :pointer)
    (define-curl-option-setter set-curl-option-url
      curl-option :url :pointer)
    'curl-option)

Macroexpanding one of the new define-curl-option-setter forms yields the following:

  (progn
    (defcfun ("curl_easy_setopt" set-curl-option-nosignal) curl-code
      (easy-handle easy-handle)
      (option curl-option)
      (new-value :long))
    (curry-curl-option-setter 'set-curl-option-nosignal ':nosignal))

Finally, let's try this out:

  cffi-user> (set-curl-option-nosignal *easy-handle* 1)
  => 0

Looks like it works just as well. This interface is now reasonably high-level to wash out some of the ugliness of the thinnest possible curl_easy_setopt FFI, without obscuring the remaining C bookkeeping details we will explore.


Next: , Previous: Tutorial-Lisp easy_setopt, Up: Tutorial

3.8 Memory management

According to the documentation for curl_easy_setopt, the type of the third argument when option is CURLOPT_ERRORBUFFER is char*. Above, we've defined set-curl-option-errorbuffer to accept a :pointer as the new option value. However, there is a CFFI type :string, which translates Lisp strings to C strings when passed as arguments to foreign function calls. Why not, then, use :string as the CFFI type of the third argument? There are two reasons, both related to the necessity of breaking abstraction described in Breaking the abstraction.

The first reason also applies to CURLOPT_URL, which we will use to illustrate the point. Assuming we have changed the type of the third argument underlying set-curl-option-url to :string, look at these two equivalent forms.

  (set-curl-option-url *easy-handle* "http://www.cliki.net/CFFI")
   
  == (with-foreign-string (url "http://www.cliki.net/CFFI")
       (foreign-funcall "curl_easy_setopt" easy-handle *easy-handle*
                        curl-option :url :pointer url curl-code))

The latter, in fact, is mostly equivalent to what a foreign function call's macroexpansion actually does. As you can see, the Lisp string "http://www.cliki.net/CFFI" is copied into a char array and null-terminated; the pointer to beginning of this array, now a C string, is passed as a CFFI :pointer to the foreign function.

Unfortunately, the C abstraction has failed us, and we must break it. While :string works well for many char* arguments, it does not for cases like this. As the curl_easy_setopt documentation explains, “The string must remain present until curl no longer needs it, as it doesn't copy the string.” The C string created by with-foreign-string, however, only has dynamic extent: it is “deallocated” when the body (above containing the foreign-funcall form) exits.

If we are supposed to keep the C string around, but it goes away, what happens when some libcurl function tries to access the URL string? We have reentered the dreaded world of C “undefined behavior”. In some Lisps, it will probably get a chunk of the Lisp/C stack. You may segfault. You may get some random piece of other data from the heap. Maybe, in a world where “dynamic extent” is defined to be “infinite extent”, everything will turn out fine. Regardless, results are likely to be almost universally unpleasant.7

Returning to the current set-curl-option-url interface, here is what we must do:

  (let (easy-handle)
    (unwind-protect
      (with-foreign-string (url "http://www.cliki.net/CFFI")
        (setf easy-handle (curl-easy-init))
        (set-curl-option-url easy-handle url)
        #|do more with the easy-handle, like actually get the URL|#)
      (when easy-handle
        (curl-easy-cleanup easy-handle))))

That is fine for the single string defined here, but for every string option we want to pass, we have to surround the body of with-foreign-string with another with-foreign-string wrapper, or else do some extremely error-prone pointer manipulation and size calculation in advance. We could alleviate some of the pain with a recursively expanding macro, but this would not remove the need to modify the block every time we want to add an option, anathema as it is to a modular interface.

Before modifying the code to account for this case, consider the other reason we can't simply use :string as the foreign type. In C, a char * is a char *, not necessarily a string. The option CURLOPT_ERRORBUFFER accepts a char *, but does not expect anything about the data there. However, it does expect that some libcurl function we call later can write a C string of up to 255 characters there. We, the callers of the function, are expected to read the C string at a later time, exactly the opposite of what :string implies.

With the semantics for an input string in mind — namely, that the string should be kept around until we curl_easy_cleanup the easy handle — we are ready to extend the Lisp interface:

  (defvar *easy-handle-cstrings* (make-hash-table)
    "Hashtable of easy handles to lists of C strings that may be
  safely freed after the handle is freed.")
   
  (defun make-easy-handle ()
    "Answer a new CURL easy interface handle, to which the lifetime
  of C strings may be tied.  See `add-curl-handle-cstring'."
    (let ((easy-handle (curl-easy-init)))
      (setf (gethash easy-handle *easy-handle-cstrings*) '())
      easy-handle))
   
  (defun free-easy-handle (handle)
    "Free CURL easy interface HANDLE and any C strings created to
  be its options."
    (curl-easy-cleanup handle)
    (mapc #'foreign-string-free
          (gethash handle *easy-handle-cstrings*))
    (remhash handle *easy-handle-cstrings*))
   
  (defun add-curl-handle-cstring (handle cstring)
    "Add CSTRING to be freed when HANDLE is, answering CSTRING."
    (car (push cstring (gethash handle *easy-handle-cstrings*))))

Here we have redefined the interface to create and free handles, to associate a list of allocated C strings with each handle while it exists. The strategy of using different function names to wrap around simple foreign functions is more common than the solution implemented earlier with curry-curl-option-setter, which was to modify the function name's function slot.8

Incidentally, the next step is to redefine curry-curl-option-setter to allocate C strings for the appropriate length of time, given a Lisp string as the new-value argument:

  (defun curry-curl-option-setter (function-name option-keyword)
    "Wrap the function named by FUNCTION-NAME with a version that
  curries the second argument as OPTION-KEYWORD.
   
  This function is intended for use in DEFINE-CURL-OPTION-SETTER."
    (setf (symbol-function function-name)
            (let ((c-function (symbol-function function-name)))
              (lambda (easy-handle new-value)
                (funcall c-function easy-handle option-keyword
                         (if (stringp new-value)
                           (add-curl-handle-cstring
                            easy-handle
                            (foreign-string-alloc new-value))
                           new-value))))))

A quick analysis of the code shows that you need only reevaluate the curl-option enumeration definition to take advantage of these new semantics. Now, for good measure, let's reallocate the handle with the new functions we just defined, and set its URL:

  cffi-user> (curl-easy-cleanup *easy-handle*)
  => NIL
  cffi-user> (setf *easy-handle* (make-easy-handle))
  => #<FOREIGN-ADDRESS #x09844EE0>
  cffi-user> (set-curl-option-nosignal *easy-handle* 1)
  => 0
  cffi-user> (set-curl-option-url *easy-handle*
                                  "http://www.cliki.net/CFFI")
  => 0

For fun, let's inspect the Lisp value of the C string that was created to hold "http://www.cliki.net/CFFI". By virtue of the implementation of add-curl-handle-cstring, it should be accessible through the hash table defined:

  cffi-user> (foreign-string-to-lisp
              (car (gethash *easy-handle* *easy-handle-cstrings*)))
  => "http://www.cliki.net/CFFI"

Looks like that worked, and libcurl now knows what URL we want to retrieve.

Finally, we turn back to the :errorbuffer option mentioned at the beginning of this section. Whereas the abstraction added to support string inputs works fine for cases like CURLOPT_URL, it hides the detail of keeping the C string; for :errorbuffer, however, we need that C string.

In a moment, we'll define something slightly cleaner, but for now, remember that you can always hack around anything. We're modifying handle creation, so make sure you free the old handle before redefining free-easy-handle.

  (defvar *easy-handle-errorbuffers* (make-hash-table)
    "Hashtable of easy handles to C strings serving as error
  writeback buffers.")
   
  ;;; An extra byte is very little to pay for peace of mind.
  (defparameter *curl-error-size* 257
    "Minimum char[] size used by cURL to report errors.")
   
  (defun make-easy-handle ()
    "Answer a new CURL easy interface handle, to which the lifetime
  of C strings may be tied.  See `add-curl-handle-cstring'."
    (let ((easy-handle (curl-easy-init)))
      (setf (gethash easy-handle *easy-handle-cstrings*) '())
      (setf (gethash easy-handle *easy-handle-errorbuffers*)
              (foreign-alloc :char :count *curl-error-size*
                             :initial-element 0))
      easy-handle))
   
  (defun free-easy-handle (handle)
    "Free CURL easy interface HANDLE and any C strings created to
  be its options."
    (curl-easy-cleanup handle)
    (foreign-free (gethash handle *easy-handle-errorbuffers*))
    (remhash handle *easy-handle-errorbuffers*)
    (mapc #'foreign-string-free
          (gethash handle *easy-handle-cstrings*))
    (remhash handle *easy-handle-cstrings*))
   
  (defun get-easy-handle-error (handle)
    "Answer a string containing HANDLE's current error message."
    (foreign-string-to-lisp
     (gethash handle *easy-handle-errorbuffers*)))

Be sure to once again set the options we've set thus far. You may wish to define yet another wrapper function to do this.


Next: , Previous: Tutorial-Memory, Up: Tutorial

3.9 Calling Lisp from C

If you have been reading curl_easy_setopt(3), you should have noticed that some options accept a function pointer. In particular, we need one function pointer to set as CURLOPT_WRITEFUNCTION, to be called by libcurl rather than the reverse, in order to receive data as it is downloaded.

A binding writer without the aid of FFI usually approaches this problem by writing a C function that accepts C data, converts to the language's internal objects, and calls the callback provided by the user, again in a reverse of usual practices.

The CFFI approach to callbacks precisely mirrors its differences with the non-FFI approach on the “calling C from Lisp” side, which we have dealt with exclusively up to now. That is, you define a callback function in Lisp using defcallback, and CFFI effectively creates a C function to be passed as a function pointer.

Implementor's note: This is much trickier than calling C functions from Lisp, as it literally involves somehow generating a new C function that is as good as any created by the compiler. Therefore, not all Lisps support them. See Implementation Support, for information about CFFI support issues in this and other areas. You may want to consider changing to a Lisp that supports callbacks in order to continue with this tutorial.

Defining a callback is very similar to defining a callout; the main difference is that we must provide some Lisp forms to be evaluated as part of the callback. Here is the signature for the function the :writefunction option takes:

  size_t
  function(void *ptr, size_t size, size_t nmemb, void *stream);
Implementor's note: size_t is almost always an unsigned int. You can get this and many other types using feature tests for your system by using cffi-grovel.

The above signature trivially translates into a CFFI defcallback form, as follows.

  ;;; Alias in case size_t changes.
  (defctype size :unsigned-int)
   
  ;;; To be set as the CURLOPT_WRITEFUNCTION of every easy handle.
  (defcallback easy-write size ((ptr :pointer) (size size)
                                (nmemb size) (stream :pointer))
    (let ((data-size (* size nmemb)))
      (handler-case
        ;; We use the dynamically-bound *easy-write-procedure* to
        ;; call a closure with useful lexical context.
        (progn (funcall (symbol-value '*easy-write-procedure*)
                        (foreign-string-to-lisp ptr data-size nil))
               data-size)         ;indicates success
        ;; The WRITEFUNCTION should return something other than the
        ;; #bytes available to signal an error.
        (error () (if (zerop data-size) 1 0)))))

First, note the correlation of the first few forms, used to declare the C function's signature, with the signature in C syntax. We provide a Lisp name for the function, its return type, and a name and type for each argument.

In the body, we call the dynamically-bound *easy-write-procedure* with a “finished” translation, of pulling together the raw data and size into a Lisp string, rather than deal with the data directly. As part of calling curl_easy_perform later, we'll bind that variable to a closure with more useful lexical bindings than the top-level defcallback form.

Finally, we make a halfhearted effort to prevent non-local exits from unwinding the C stack, covering the most likely case with an error handler, which is usually triggered unexpectedly.9 The reason is that most C code is written to understand its own idiosyncratic error condition, implemented above in the case of curl_easy_perform, and more “undefined behavior” can result if we just wipe C stack frames without allowing them to execute whatever cleanup actions as they like.

Using the CURLoption enumeration in curl.h once more, we can describe the new option by modifying and reevaluating define-curl-options.

  (define-curl-options curl-option
      (long 0 objectpoint 10000 functionpoint 20000 off-t 30000)
    (:noprogress long 43)
    (:nosignal long 99)
    (:errorbuffer objectpoint 10)
    (:url objectpoint 2)
    (:writefunction functionpoint 11)) ;new item here

Finally, we can use the defined callback and the new set-curl-option-writefunction to finish configuring the easy handle, using the callback macro to retrieve a CFFI :pointer, which works like a function pointer in C code.

  cffi-user> (set-curl-option-writefunction
              *easy-handle* (callback easy-write))
  => 0


Next: , Previous: Tutorial-Callbacks, Up: Tutorial

3.10 A complete FFI?

With all options finally set and a medium-level interface developed, we can finish the definition and retrieve http://www.cliki.net/CFFI, as is done in the tutorial.

  (defcfun "curl_easy_perform" curl-code
    (handle easy-handle))
  cffi-user> (with-output-to-string (contents)
               (let ((*easy-write-procedure*
                       (lambda (string)
                         (write-string string contents))))
                 (declare (special *easy-write-procedure*))
                 (curl-easy-perform *easy-handle*)))
  => "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01//EN\"
  ....
  Now fear, comprehensively</P>
  "

Of course, that itself is slightly unwieldy, so you may want to define a function around it that simply retrieves a URL. I will leave synthesis of all the relevant REPL forms presented thus far into a single function as an exercise for the reader.

The remaining sections of this tutorial explore some advanced features of CFFI; the definition of new types will receive special attention. Some of these features are essential for particular foreign function calls; some are very helpful when trying to develop a Lispy interface to C.


Next: , Previous: Tutorial-Completion, Up: Tutorial

3.11 Defining new types

We've occasionally used the defctype macro in previous sections as a kind of documentation, much what you'd use typedef for in C. We also tried one special kind of type definition, the defcenum type. See defcstruct, for a definition macro that may come in handy if you need to use C structs as data.

However, all of these are mostly sugar for the powerful underlying foreign type interface called type translators. You can easily define new translators for any simple named foreign type. Since we've defined the new type curl-code to use as the return type for various libcurl functions, we can use that to directly convert cURL errors to Lisp errors.

defctype's purpose is to define simple typedef-like aliases. In order to use type translators we must use the define-foreign-type macro. So let's redefine curl-code using it.

  (define-foreign-type curl-code-type ()
    ()
    (:actual-type :int)
    (:simple-parser curl-code))

define-foreign-type is a thin wrapper around defclass. For now, all you need to know in the context of this example is that it does what (defctype curl-code :int) would do and, additionally, defines a new class curl-code-type which we will take advantage of shortly.

The CURLcode enumeration seems to follow the typical error code convention of `0' meaning all is well, and each non-zero integer indicating a different kind of error. We can apply that trivially to differentiate between normal exits and error exits.

  (define-condition curl-code-error (error)
    (($code :initarg :curl-code :reader curl-error-code))
    (:report (lambda (c stream)
               (format stream "libcurl function returned error ~A"
                              (curl-error-code c))))
    (:documentation "Signalled when a libcurl function answers
  a code other than CURLE_OK."))
   
  (defmethod translate-from-foreign (value (type curl-code-type))
    "Raise a CURL-CODE-ERROR if VALUE, a curl-code, is non-zero."
    (if (zerop value)
        :curle-ok
        (error 'curl-code-error :curl-code value)))

The heart of this translator is new method translate-from-foreign. By specializing the type parameter on curl-code-type, we immediately modify the behavior of every function that returns a curl-code to pass the result through this new method.

To see the translator in action, try invoking a function that returns a curl-code. You need to reevaluate the respective defcfun form so that it picks up the new curl-code definition.

  cffi-user> (set-curl-option-nosignal *easy-handle* 1)
  => :CURLE-OK

As the result was `0', the new method returned :curle-ok, just as specified.10 I will leave disjoining the separate CURLcodes into condition types and improving the :report function as an exercise for you.

The creation of *easy-handle-cstrings* and *easy-handle-errorbuffers* as properties of easy-handles is a kluge. What we really want is a Lisp structure that stores these properties along with the C pointer. Unfortunately, easy-handle is currently just a fancy name for the foreign type :pointer; the actual pointer object varies from Common Lisp implementation to implementation, needing only to satisfy pointerp and be returned from make-pointer and friends.

One solution that would allow us to define a new Lisp structure to represent easy-handles would be to write a wrapper around every function that currently takes an easy-handle; the wrapper would extract the pointer and pass it to the foreign function. However, we can use type translators to more elegantly integrate this “translation” into the foreign function calling framework, using translate-to-foreign.

  (defclass easy-handle ()
    ((pointer :initform (curl-easy-init)
              :documentation "Foreign pointer from curl_easy_init")
     (error-buffer
      :initform (foreign-alloc :char :count *curl-error-size*
                               :initial-element 0)
      :documentation "C string describing last error")
     (c-strings :initform '()
                :documentation "C strings set as options"))
    (:documentation "I am a parameterization you may pass to
  curl-easy-perform to perform a cURL network protocol request."))
   
  (defmethod initialize-instance :after ((self easy-handle) &key)
    (set-curl-option-errorbuffer self (slot-value self 'error-buffer)))
   
  (defun add-curl-handle-cstring (handle cstring)
    "Add CSTRING to be freed when HANDLE is, answering CSTRING."
    (car (push cstring (slot-value handle 'c-strings))))
   
  (defun get-easy-handle-error (handle)
    "Answer a string containing HANDLE's current error message."
    (foreign-string-to-lisp
     (slot-value handle 'error-buffer)))
   
  (defun free-easy-handle (handle)
    "Free CURL easy interface HANDLE and any C strings created to
  be its options."
    (with-slots (pointer error-buffer c-strings) handle
      (curl-easy-cleanup pointer)
      (foreign-free error-buffer)
      (mapc #'foreign-string-free c-strings)))
   
  (define-foreign-type easy-handle-type ()
    ()
    (:actual-type :pointer)
    (:simple-parser easy-handle))
   
  (defmethod translate-to-foreign (handle (type easy-handle-type))
    "Extract the pointer from an easy-HANDLE."
    (slot-value handle 'pointer))

While we changed some of the Lisp functions defined earlier to use CLOS slots rather than hash tables, the foreign functions work just as well as they did before.

The greatest strength, and the greatest limitation, of the type translator comes from its generalized interface. As stated previously, we could define all foreign function calls in terms of the primitive foreign types provided by CFFI. The type translator interface allows us to cleanly specify the relationship between Lisp and C data, independent of where it appears in a function call. This independence comes at a price; for example, it cannot be used to modify translation semantics based on other arguments to a function call. In these cases, you should rely on other features of Lisp, rather than the powerful, yet domain-specific, type translator interface.


Previous: Tutorial-Types, Up: Tutorial

3.12 What's next?

CFFI provides a rich and powerful foundation for communicating with foreign libraries; as we have seen, it is up to you to make that experience a pleasantly Lispy one. This tutorial does not cover all the features of CFFI; please see the rest of the manual for details. In particular, if something seems obviously missing, it is likely that either code or a good reason for lack of code is already present.

Implementor's note: There are some other things in CFFI that might deserve tutorial sections, such as define-foreign-type, free-translated-object, or structs. Let us know which ones you care about.


Next: , Previous: Tutorial, Up: Top

4 Wrapper generators

CFFI's interface is designed for human programmers, being aimed at aesthetic as well as technical sophistication. However, there are a few programs aimed at translating C and C++ header files, or approximations thereof, into CFFI forms constituting a foreign interface to the symbols in those files.

These wrapper generators are known to support output of CFFI forms.

Verrazano
Designed specifically for Common Lisp. Uses GCC's parser output in XML format to discover functions, variables, and other header file data. This means you need GCC to generate forms; on the other hand, the parser employed is mostly compliant with ANSI C.
SWIG
A foreign interface generator originally designed to generate Python bindings, it has been ported to many other systems, including CFFI in version 1.3.28. Includes its own C declaration munger, not intended to be fully-compliant with ANSI C.

First, this manual does not describe use of these other programs; they have documentation of their own. If you have problems using a generated interface, please look at the output CFFI forms and verify that they are a correct CFFI interface to the library in question; if they are correct, contact CFFI developers with details, keeping in mind that they communicate in terms of those forms rather than any particular wrapper generator. Otherwise, contact the maintainers of the wrapper generator you are using, provided you can reasonably expect more accuracy from the generator.

When is more accuracy an unreasonable expectation? As described in the tutorial (see Breaking the abstraction), the information in C declarations is insufficient to completely describe every interface. In fact, it is quite common to run into an interface that cannot be handled automatically, and generators should be excused from generating a complete interface in these cases.

As further described in the tutorial, the thinnest Lisp interface to a C function is not always the most pleasant one. In many cases, you will want to manually write a Lispier interface to the C functions that interest you.

Wrapper generators should be treated as time-savers, not complete automation of the full foreign interface writing job. Reports of the amount of work done by generators vary from 30% to 90%. The incremental development style enabled by CFFI generally reduces this proportion below that for languages like Python.


Next: , Previous: Wrapper generators, Up: Top

5 Foreign Types

Foreign types describe how data is translated back and forth between C and Lisp. CFFI provides various built-in types and allows the user to define new types.

Dictionary


Next: , Up: Foreign Types

5.1 Built-In Types

— Foreign Type: :char
— Foreign Type: :unsigned-char
— Foreign Type: :short
— Foreign Type: :unsigned-short
— Foreign Type: :int
— Foreign Type: :unsigned-int
— Foreign Type: :long
— Foreign Type: :unsigned-long
— Foreign Type: :long-long
— Foreign Type: :unsigned-long-long

These types correspond to the native C integer types according to the ABI of the Lisp implementation's host system.

— Foreign Type: :uchar
— Foreign Type: :ushort
— Foreign Type: :uint
— Foreign Type: :ulong
— Foreign Type: :llong
— Foreign Type: :ullong

For convenience, the above types are provided as shortcuts for unsigned-char, unsigned-short, unsigned-int, unsigned-long, long-long and unsigned-long-long, respectively.

:long-long and :unsigned-long-long are not supported on all implementations. When those types are not available, the symbol cffi-features:no-long-long is pushed into *features*.

— Foreign Type: :int8
— Foreign Type: :uint8
— Foreign Type: :int16
— Foreign Type: :uint16
— Foreign Type: :int32
— Foreign Type: :uint32
— Foreign Type: :int64
— Foreign Type: :uint64

Foreign integer types of specific sizes, corresponding to the C types defined in stdint.h.

— Foreign Type: :float
— Foreign Type: :double

On all systems, the :float and :double types represent a C float and double, respectively. On most but not all systems, :float and :double represent a Lisp single-float and double-float, respectively. It is not so useful to consider the relationship between Lisp types and C types as isomorphic, as simply to recognize the relationship, and relative precision, among each respective category.

— Foreign Type: :long-double

This type is only supported on SCL.

— Foreign Type: :pointer
— Foreign Type: (:pointer &optional type)

A foreign pointer to an object of any type, corresponding to void *. You can optionally specify type of pointer (e.g. (:pointer :char)). Although CFFI won't do anything with that information yet, it is useful for documentation purposes.

— Foreign Type: :void

No type at all. Only valid as the return type of a function.


Next: , Previous: Built-In Types, Up: Foreign Types

5.2 Other Types

CFFI also provides a few useful types that aren't built-in C types.

— Foreign Type: :string

The :string type performs automatic conversion between Lisp and C strings. Note that, in the case of functions the converted C string will have dynamic extent (i.e. it will be automatically freed after the foreign function returns).

In addition to Lisp strings, this type will also convert Lisp arrays of element type (unsigned-byte 8) and will pass foreign pointers unmodified.

A method for free-translated-object is specialized for this type. So, for example, foreign strings allocated by this type and passed to a foreign function will be freed after the function returns.

  CFFI> (foreign-funcall "getenv" :string "SHELL" :string)
  => "/bin/bash"
   
  CFFI> (with-foreign-string (str "abcdef")
          (foreign-funcall "strlen" :string str :int))
  => 6
   
  CFFI> (let ((str (make-array 4 :element-type '(unsigned-byte 8)
                               :initial-element 65)))
          (foreign-funcall "strlen" :string str :int))
  => 4
— Foreign Type: :string+ptr

Like :string but returns a list with two values when convert from C to Lisp: a Lisp string and the C string's foreign pointer.

  CFFI> (foreign-funcall "getenv" :string "SHELL" :string+ptr)
  => ("/bin/bash" #.(SB-SYS:INT-SAP #XBFFFFC6F))
— Foreign Type: :boolean &optional (base-type :int)

The :boolean type converts between a Lisp boolean and a C boolean. It canonicalizes to base-type which is :int by default.

  (convert-to-foreign nil :boolean) => 0
  (convert-to-foreign t :boolean) => 1
  (convert-from-foreign 0 :boolean) => nil
  (convert-from-foreign 1 :boolean) => t
— Foreign Type: :wrapper base-type &key to-c from-c

The :wrapper type stores two symbols passed to the to-c and from-c arguments. When a value is being translated to or from C, this type funcalls the respective symbol.

:wrapper types will be typedefs for base-type and will inherit its translators, if any.

Here's an example of how the :boolean type could be defined in terms of :wrapper.

  (defun bool-c-to-lisp (value)
    (not (zerop value)))
   
  (defun bool-lisp-to-c (value)
    (if value 1 0))
   
  (defctype my-bool (:wrapper :int :from-c bool-c-to-lisp
                                   :to-c bool-lisp-to-c))
   
  (convert-to-foreign nil 'my-bool) => 0
  (convert-from-foreign 1 'my-bool) => t


Next: , Previous: Other Types, Up: Foreign Types

5.3 Defining Foreign Types

You can define simple C-like typedefs through the defctype macro. Defining a typedef is as simple as giving defctype a new name and the name of the type to be wrapped.

  ;;; Define MY-INT as an alias for the built-in type :INT.
  (defctype my-int :int)

With this type definition, one can, for instance, declare arguments to foreign functions as having the type my-int, and they will be passed as integers.

More complex types

CFFI offers another way to define types through define-foreign-type, a thin wrapper macro around defclass. As an example, let's go through the steps needed to define a (my-string &key encoding) type. First, we need to define our type class:

  (define-foreign-type my-string-type ()
    ((encoding :reader string-type-encoding :initarg :encoding))
    (:actual-type :pointer))

The :actual-type class option tells CFFI that this type will ultimately be passed to and received from foreign code as a :pointer. Now you need to tell CFFI how to parse a type specification such as (my-string :encoding :utf8) into an instance of my-string-type. We do that with define-parse-method:

  (define-parse-method my-string (&key (encoding :utf-8))
    (make-instance 'my-string-type :encoding encoding))

The next section describes how make this type actually translate between C and Lisp strings.


Next: , Previous: Defining Foreign Types, Up: Foreign Types

5.4 Foreign Type Translators

Type translators are used to automatically convert Lisp values to or from foreign values. For example, using type translators, one can take the my-string type defined in the previous section and specify that it should:

In order to tell CFFI how to automatically convert Lisp values to foreign values, define a specialized method for the translate-to-foreign generic function:

  ;;; Define a method that converts Lisp strings to C strings.
  (defmethod translate-to-foreign (string (type my-string-type))
    (foreign-string-alloc string :encoding (string-type-encoding type)))

From now on, whenever an object is passed as a my-string to a foreign function, this method will be invoked to convert the Lisp value. To perform the inverse operation, which is needed for functions that return a my-string, specialize the translate-from-foreign generic function in the same manner:

  ;;; Define a method that converts C strings to Lisp strings.
  (defmethod translate-from-foreign (pointer (type my-string-type))
    (foreign-string-to-lisp pointer :encoding (string-type-encoding type)))

When a translate-to-foreign method requires allocation of foreign memory, you must also define a free-translated-object method to free the memory once the foreign object is no longer needed, otherwise you'll be faced with memory leaks. This generic function is called automatically by CFFI when passing objects to foreign functions. Let's do that:

  ;;; Free strings allocated by translate-to-foreign.
  (defmethod free-translated-object (pointer (type my-string-type) param)
    (declare (ignore param))
    (foreign-string-free pointer))

In this specific example, we don't need the param argument, so we ignore it. See free-translated-object, for an explanation of its purpose and how you can use it.

A type translator does not necessarily need to convert the value. For example, one could define a typedef for :pointer that ensures, in the translate-to-foreign method, that the value is not a null pointer, signalling an error if a null pointer is passed. This would prevent some pointer errors when calling foreign functions that cannot handle null pointers.

Please note: these methods are meant as extensible hooks only, and you should not call them directly. Use convert-to-foreign, convert-from-foreign and free-converted-object instead.

See Defining new types, for another example of type translators.


Next: , Previous: Foreign Type Translators, Up: Foreign Types

5.5 Optimizing Type Translators

Being based on generic functions, the type translation mechanism described above can add a bit of overhead. This is usually not significant, but we nevertheless provide a way of getting rid of the overhead for the cases where it matters.

A good way to understand this issue is to look at the code generated by defcfun. Consider the following example using the previously defined my-string type:

  CFFI> (macroexpand-1 '(defcfun foo my-string (x my-string)))
  ;; (simplified, downcased, etc...)
  (defun foo (x)
    (multiple-value-bind (#:G2019 #:PARAM3149)
        (translate-to-foreign x #<MY-STRING-TYPE {11ED5A79}>)
      (unwind-protect
          (translate-from-foreign
           (foreign-funcall "foo" :pointer #:G2019 :pointer)
           #<MY-STRING-TYPE {11ED5659}>)
        (free-translated-object #:G2019 #<MY-STRING-TYPE {11ED51A79}>
                                #:PARAM3149))))

In order to get rid of those generic function calls, CFFI has another set of extensible generic functions that provide functionality similar to CL's compiler macros: expand-to-foreign-dyn, expand-to-foreign and expand-from-foreign. Here's how one could define a my-boolean with them:

  (define-foreign-type my-boolean-type ()
    ()
    (:actual-type :int)
    (:simple-parser my-boolean))
   
  (defmethod expand-to-foreign (value (type my-boolean-type))
    `(if ,value 1 0))
   
  (defmethod expand-from-foreign (value (type my-boolean-type))
    `(not (zerop ,value)))

And here's what the macroexpansion of a function using this type would look like:

  CFFI> (macroexpand-1 '(defcfun bar my-boolean (x my-boolean)))
  ;; (simplified, downcased, etc...)
  (defun bar (x)
    (let ((#:g3182 (if x 1 0)))
      (not (zerop (foreign-funcall "bar" :int #:g3182 :int)))))

No generic function overhead.

Let's go back to our my-string type. The expansion interface has no equivalent of free-translated-object; you must instead define a method on expand-to-foreign-dyn, the third generic function in this interface. This is especially useful when you can allocate something much more efficiently if you know the object has dynamic extent, as is the case with function calls that don't save the relevant allocated arguments.

This exactly what we need for the my-string type:

  (defmethod expand-from-foreign (form (type my-string-type))
    `(foreign-string-to-lisp ,form))
   
  (defmethod expand-to-foreign-dyn (value var body (type my-string-type))
    (let ((encoding (string-type-encoding type)))
      `(with-foreign-string (,var ,value :encoding ',encoding)
         ,@body)))

So let's look at the macro expansion:

  CFFI> (macroexpand-1 '(defcfun foo my-string (x my-string)))
  ;; (simplified, downcased, etc...)
  (defun foo (x)
    (with-foreign-string (#:G2021 X :encoding ':utf-8)
      (foreign-string-to-lisp
       (foreign-funcall "foo" :pointer #:g2021 :pointer))))

Again, no generic function overhead.

Other details

To short-circuit expansion and use the translate-* functions instead, simply call the next method. Return its result in cases where your method cannot generate an appropriate replacement for it. This analogous to the &whole form mechanism compiler macros provide.

The expand-* methods have precedence over their translate-* counterparts and are guaranteed to be used in defcfun, foreign-funcall, defcvar and defcallback. If you define a method on each of the expand-* generic functions, you are guaranteed to have full control over the expressions generated for type translation in these macros.

They may or may not be used in other CFFI operators that need to translate between Lisp and C data; you may only assume that expand-* methods will probably only be called during Lisp compilation.

expand-to-foreign-dyn has precedence over expand-to-foreign and is only used in defcfun and foreign-funcall, only making sense in those contexts.

Important note: this set of generic functions is called at macroexpansion time. Methods are defined when loaded or evaluated, not compiled. You are responsible for ensuring that your expand-* methods are defined when the foreign-funcall or other forms that use them are compiled. One way to do this is to put the method definitions earlier in the file and inside an appropriate eval-when form; another way is to always load a separate Lisp or FASL file containing your expand-* definitions before compiling files with forms that ought to use them. Otherwise, they will not be found and the runtime translators will be used instead.


Next: , Previous: Optimizing Type Translators, Up: Foreign Types

5.6 Foreign Structure Types

For more involved C types than simple aliases to built-in types, such as you can make with defctype, CFFI allows declaration of structures and unions with defcstruct and defcunion.

For example, consider this fictional C structure declaration holding some personal information:

  struct person {
    int number;
    char* reason;
  };

The equivalent defcstruct form follows:

  (defcstruct person
    (number :int)
    (reason :string))

CFFI knows how to align C structs, and how to figure in padding between struct elements.

Please note that this interface is only for those that must know about the values contained in a relevant struct. If the library you are interfacing returns an opaque pointer that needs only be passed to other C library functions, by all means just use :pointer or a type-safe definition munged together with defctype and type translation.


Next: , Previous: Foreign Structure Types, Up: Foreign Types

5.7 Operations on Types

Implementor's note: Which “operations” are worth going over here? –stephen


Next: , Previous: Operations on Types, Up: Foreign Types

5.8 Allocating Foreign Objects

See Allocating Foreign Memory.


Next: , Previous: Allocating Foreign Objects, Up: Foreign Types

convert-from-foreign

Syntax

— Function: convert-from-foreign foreign-value type => value

Arguments and Values

foreign-value
The primitive C value as returned from a primitive foreign function or from convert-to-foreign.
type
A CFFI type specifier.
value
The Lisp value translated from foreign-value.

Description

This is an external interface to the type translation facility. In the implementation, all foreign functions are ultimately defined as type translation wrappers around primitive foreign function invocations.

This function is available mostly for inspection of the type translation process, and possibly optimization of special cases of your foreign function calls.

Its behavior is better described under translate-from-foreign's documentation.

Examples

  CFFI-USER> (convert-to-foreign "a boat" :string)
  => #<FOREIGN-ADDRESS #x097ACDC0>
  => (T)
  CFFI-USER> (convert-from-foreign * :string)
  => "a boat"

See Also

convert-to-foreign
translate-from-foreign


Next: , Previous: convert-from-foreign, Up: Foreign Types

convert-to-foreign

Syntax

— Function: convert-to-foreign value type => foreign-value, alloc-params

Arguments and Values

value
The Lisp object to be translated to a foreign object.
type
A CFFI type specifier.
foreign-value
The primitive C value, ready to be passed to a primitive foreign function.
alloc-params
Something of a translation state; you must pass it to free-converted-object along with the foreign value for that to work.

Description

This is an external interface to the type translation facility. In the implementation, all foreign functions are ultimately defined as type translation wrappers around primitive foreign function invocations.

This function is available mostly for inspection of the type translation process, and possibly optimization of special cases of your foreign function calls.

Its behavior is better described under translate-to-foreign's documentation.

Examples

  CFFI-USER> (convert-to-foreign t :boolean)
  => 1
  => (NIL)
  CFFI-USER> (convert-to-foreign "hello, world" :string)
  => #<FOREIGN-ADDRESS #x097C5F80>
  => (T)
  CFFI-USER> (code-char (mem-aref * :char 5))
  => #\,

See Also

convert-from-foreign
free-converted-object
translate-to-foreign


Next: , Previous: convert-to-foreign, Up: Foreign Types

defbitfield

Syntax

— Macro: defbitfield name-and-options &body masks

masks ::= [docstring] { (symbol value) }*
name-and-options ::= name | (name &optional (base-type :int))

Arguments and Values

name
The name of the new bitfield type.
docstring
A documentation string, ignored.
base-type
A symbol denoting a foreign type.
symbol
A Lisp symbol.
value
An integer representing a bitmask.

Description

The defbitfield macro is used to define foreign types that map lists of symbols to integer values.

If value is omitted, it will be computed as follows: find the greatest value previously used, including those so computed, with only a single 1-bit in its binary representation (that is, powers of two), and left-shift it by one. This rule guarantees that a computed value cannot clash with previous values, but may clash with future explicitly specified values.

Symbol lists will be automatically converted to values and vice versa when being passed as arguments to or returned from foreign functions, respectively. The same applies to any other situations where an object of a bitfield type is expected.

Types defined with defbitfield canonicalize to base-type which is :int by default.

Examples

  (defbitfield open-flags
    (:rdonly #x0000)
    :wronly               ;#x0001
    :rdwr                 ;...
    :nonblock
    :append
    (:creat    #x0200))
    ;; etc...
   
  CFFI> (foreign-bitfield-symbols 'open-flags #b1101)
  => (:RDONLY :WRONLY :NONBLOCK :APPEND)
   
  CFFI> (foreign-bitfield-value 'open-flags '(:rdwr :creat))
  => 514   ; #x0202
   
  (defcfun ("open" unix-open) :int
    (path :string)
    (flags open-flags)
    (mode :uint16)) ; unportable
   
  CFFI> (unix-open "/tmp/foo" '(:wronly :creat) #o644)
  => #<an fd>
   
  ;;; Consider also the following lispier wrapper around open()
  (defun lispier-open (path mode &rest flags)
    (unix-open path flags mode))

See Also

foreign-bitfield-value
foreign-bitfield-symbols


Next: , Previous: defbitfield, Up: Foreign Types

defcstruct

Syntax

— Macro: defcstruct name-and-options &body doc-and-slots => name

name-and-options ::= structure-name | (structure-name &key size)

doc-and-slots ::= [docstring] { (slot-name slot-type &key count offset) }*

Arguments and Values

structure-name
The name of new structure type.
docstring
A documentation string, ignored.
slot-name
A symbol naming the slot. It must be unique among slot names in this structure.
size
Use this option to override the size (in bytes) of the struct.
slot-type
The type specifier for the slot.
count
Used to declare an array of size count inside the structure. Defaults to 1 as such an array and a single element are semantically equivalent.
offset
Overrides the slot's offset. The next slot's offset is calculated based on this one.

Description

This defines a new CFFI aggregate type akin to C structs. In other words, it specifies that foreign objects of the type structure-name are groups of different pieces of data, or “slots”, of the slot-types, distinguished from each other by the slot-names. Each structure is located in memory at a position, and the slots are allocated sequentially beginning at that point in memory (with some padding allowances as defined by the C ABI, unless otherwise requested by specifying an offset from the beginning of the structure (offset 0).

In other words, it is isomorphic to the C struct, giving several extra features.

There are two kinds of slots, for the two kinds of CFFI types:

Simple
Contain a single instance of a type that canonicalizes to a built-in type, such as :long or :pointer. Used for simple CFFI types.
Aggregate
Contain an embedded structure or union, or an array of objects. Used for aggregate CFFI types.

The use of CLOS terminology for the structure-related features is intentional; structure definitions are very much like classes with (far) fewer features.

Examples

  (defcstruct point
    "Pointer structure."
    (x :int)
    (y :int))
   
  CFFI> (with-foreign-object (ptr 'point)
          ;; Initialize the slots
          (setf (foreign-slot-value ptr 'point 'x) 42
                (foreign-slot-value ptr 'point 'y) 42)
          ;; Return a list with the coordinates
          (with-foreign-slots ((x y) ptr point)
            (list x y)))
  => (42 42)
  ;; Using the :size and :offset options to define a partial structure.
  ;; (this is useful when you are interested in only a few slots
  ;; of a big foreign structure)
   
  (defcstruct (foo :size 32)
    "Some struct with 32 bytes."
                          ; <16 bytes we don't care about>
    (x :int :offset 16)   ; an int at offset 16
    (y :int)              ; another int at offset 16+sizeof(int)
                          ; <a couple more bytes we don't care about>
    (z :char :offset 24)  ; a char at offset 24
                          ; <7 more bytes ignored (since size is 32)>
    )
   
  CFFI> (foreign-type-size 'foo)
  => 32

See Also

foreign-slot-pointer
foreign-slot-value
with-foreign-slots


Next: , Previous: defcstruct, Up: Foreign Types

defcunion

Syntax

— Macro: defcunion name &body doc-and-slots => name

doc-and-slots ::= [docstring] { (slot-name slot-type &key count) }*

Arguments and Values

name
The name of new union type.
docstring
A documentation string, ignored.
slot-name
A symbol naming the slot.
slot-type
The type specifier for the slot.
count
Used to declare an array of size count inside the structure.

Description

A union is a structure in which all slots have an offset of zero. It is isomorphic to the C union. Therefore, you should use the usual foreign structure operations for accessing a union's slots.

Examples

  (defcunion uint32-bytes
    (int-value :unsigned-int)
    (bytes :unsigned-char :count 4))

See Also

foreign-slot-pointer
foreign-slot-value


Next: , Previous: defcunion, Up: Foreign Types

defctype

Syntax

— Macro: defctype name base-type &optional documentation

Arguments and Values

name
The name of the new foreign type.
base-type
A symbol or a list defining the new type.
documentation
A documentation string, currently ignored.

Description

The defctype macro provides a mechanism similar to C's typedef to define new types. The new type inherits base-type's translators, if any.

There is no way to define translations for types for types defined with defctype. For that, you should use define-foreign-type.

Examples

  (defctype my-string :string
    "My own string type.")
   
  (defctype long-bools (:boolean :long)
    "Booleans that map to C longs.")

See Also

define-foreign-type


Next: , Previous: defctype, Up: Foreign Types

defcenum

Syntax

— Macro: defcenum name-and-options &body enum-list

enum-list ::= [docstring] { keyword | (keyword value) }* name-and-options ::= name | (name &optional (base-type :int))

Arguments and Values

name
The name of the new enum type.
docstring
A documentation string, ignored.
base-type
A symbol denoting a foreign type.
keyword
A keyword symbol.
value
An index value for a keyword.

Description

The defcenum macro is used to define foreign types that map keyword symbols to integer values, similar to the C enum type.

If value is omitted its value will either be 0, if it's the first entry, or it it will continue the progression from the last specified value.

Keywords will be automatically converted to values and vice-versa when being passed as arguments to or returned from foreign functions, respectively. The same applies to any other situations where an object of an enum type is expected.

Types defined with defcenum canonicalize to base-type which is :int by default.

Examples

  (defcenum boolean
    :no
    :yes)
   
  CFFI> (foreign-enum-value 'boolean :no)
  => 0
  (defcenum numbers
    (:one 1)
    :two
    (:four 4))
   
  CFFI> (foreign-enum-keyword 'numbers 2)
  => :TWO

See Also

foreign-enum-value
foreign-enum-keyword


Next: , Previous: defcenum, Up: Foreign Types

define-foreign-type

Syntax

— Macro: define-foreign-type class-name supers slots &rest options => class-name

options ::= (:actual-type type) | (:simple-parser symbol) | regular defclass option

Arguments and Values

class-name
A symbol naming the new foreign type class.
supers
A list of symbols naming the super classes.
slots
A list of slot definitions, passed to defclass.

Description

The macro define-foreign-type defines a new class class-name. It is a thin wrapper around defclass. Among other things, it ensures that class-name becomes a subclass of foreign-type, what you need to know about that is that there's an initarg :actual-type which serves the same purpose as defctype's base-type argument.

Examples

Taken from CFFI's :boolean type definition:

  (define-foreign-type :boolean (&optional (base-type :int))
    "Boolean type. Maps to an :int by default. Only accepts integer types."
    (ecase base-type
      ((:char
        :unsigned-char
        :int
        :unsigned-int
        :long
        :unsigned-long) base-type)))
   
  CFFI> (canonicalize-foreign-type :boolean)
  => :INT
  CFFI> (canonicalize-foreign-type '(:boolean :long))
  => :LONG
  CFFI> (canonicalize-foreign-type '(:boolean :float))
  ;; error--> signalled by ECASE.
  

See Also

defctype
define-parse-method


Next: , Previous: define-foreign-type, Up: Foreign Types

define-parse-method

Syntax

— Macro: define-parse-method name lambda-list &body body => name

Arguments and Values

type-name
A symbol naming the new foreign type.
lambda-list
A lambda list which is the argument list of the new foreign type.
body
One or more forms that provide a definition of the new foreign type.

Description

Examples

Taken from CFFI's :boolean type definition:

  (define-foreign-type :boolean (&optional (base-type :int))
    "Boolean type. Maps to an :int by default. Only accepts integer types."
    (ecase base-type
      ((:char
        :unsigned-char
        :int
        :unsigned-int
        :long
        :unsigned-long) base-type)))
   
  CFFI> (canonicalize-foreign-type :boolean)
  => :INT
  CFFI> (canonicalize-foreign-type '(:boolean :long))
  => :LONG
  CFFI> (canonicalize-foreign-type '(:boolean :float))
  ;; error--> signalled by ECASE.
  

See Also

define-foreign-type


Next: , Previous: define-parse-method, Up: Foreign Types

foreign-bitfield-symbols

Syntax

— Function: foreign-bitfield-symbols type value => symbols

Arguments and Values

type
A bitfield type.
value
An integer.
symbols
A potentially shared list of symbols. nil.

Description

The function foreign-bitfield-symbols returns a possibly shared list of symbols that correspond to value in type.

Examples

  (defbitfield flags
    (flag-a 1)
    (flag-b 2)
    (flag-c 4))
   
  CFFI> (foreign-bitfield-symbols 'boolean #b101)
  => (FLAG-A FLAG-C)

See Also

defbitfield
foreign-bitfield-value


Next: , Previous: foreign-bitfield-symbols, Up: Foreign Types

foreign-bitfield-value

Syntax

— Function: foreign-bitfield-value type symbols => value

Arguments and Values

type
A bitfield type.
symbol
A Lisp symbol.
value
An integer.

Description

The function foreign-bitfield-value returns the value that corresponds to the symbols in the symbols list.

Examples

  (defbitfield flags
    (flag-a 1)
    (flag-b 2)
    (flag-c 4))
   
  CFFI> (foreign-bitfield-value 'flags '(flag-a flag-c))
  => 5  ; #b101

See Also

defbitfield
foreign-bitfield-symbols


Next: , Previous: foreign-bitfield-value, Up: Foreign Types

foreign-enum-keyword

Syntax

— Function: foreign-enum-keyword type value &key errorp => keyword

Arguments and Values

type
An enum type.
value
An integer.
errorp
If true (the default), signal an error if value is not defined in type. If false, foreign-enum-keyword returns nil.
keyword
A keyword symbol.

Description

The function foreign-enum-keyword returns the keyword symbol that corresponds to value in type.

An error is signaled if type doesn't contain such value and errorp is true.

Examples

  (defcenum boolean
    :no
    :yes)
   
  CFFI> (foreign-enum-keyword 'boolean 1)
  => :YES

See Also

defcenum
foreign-enum-value


Next: , Previous: foreign-enum-keyword, Up: Foreign Types

foreign-enum-value

Syntax

— Function: foreign-enum-value type keyword &key errorp => value

Arguments and Values

type
An enum type.
keyword
A keyword symbol.
errorp
If true (the default), signal an error if keyword is not defined in type. If false, foreign-enum-value returns nil.
value
An integer.

Description

The function foreign-enum-value returns the value that corresponds to keyword in type.

An error is signaled if type doesn't contain such keyword, and errorp is true.

Examples

  (defcenum boolean
    :no
    :yes)
   
  CFFI> (foreign-enum-value 'boolean :yes)
  => 1

See Also

defcenum
foreign-enum-keyword


Next: , Previous: foreign-enum-value, Up: Foreign Types

foreign-slot-names

Syntax

— Function: foreign-slot-names type => names

Arguments and Values

type
A foreign struct type.
names
A list.

Description

The function foreign-slot-names returns a potentially shared list of slot names for the given structure type. This list has no particular order.

Examples

  (defcstruct timeval
    (tv-secs :long)
    (tv-usecs :long))
   
  CFFI> (foreign-slot-names 'timeval)
  => (TV-SECS TV-USECS)

See Also

defcstruct
foreign-slot-offset
foreign-slot-value
foreign-slot-pointer


Next: , Previous: foreign-slot-names, Up: Foreign Types

foreign-slot-offset

Syntax

— Function: foreign-slot-offset type slot-name => offset

Arguments and Values

type
A foreign struct type.
slot-name
A symbol.
offset
An integer.

Description

The function foreign-slot-offset returns the offset in bytes of a slot in a foreign struct type.

Examples

  (defcstruct timeval
    (tv-secs :long)
    (tv-usecs :long))
   
  CFFI> (foreign-slot-offset 'timeval 'tv-secs)
  => 0
  CFFI> (foreign-slot-offset 'timeval 'tv-usecs)
  => 4

See Also

defcstruct
foreign-slot-names
foreign-slot-pointer
foreign-slot-value


Next: , Previous: foreign-slot-offset, Up: Foreign Types

foreign-slot-pointer

Syntax

— Function: foreign-slot-pointer ptr type slot-name => pointer

Arguments and Values

ptr
A pointer to a structure.
type
A foreign structure type.
slot-names
A slot name in the type.
pointer
A pointer to the slot slot-name.

Description

Returns a pointer to the location of the slot slot-name in a foreign object of type type at ptr. The returned pointer points inside the structure. Both the pointer and the memory it points to have the same extent as ptr.

For aggregate slots, this is the same value returned by foreign-slot-value.

Examples

  (defcstruct point
    "Pointer structure."
    (x :int)
    (y :int))
   
  CFFI> (with-foreign-object (ptr 'point)
          (foreign-slot-pointer ptr 'point 'x))
  => #<FOREIGN-ADDRESS #xBFFF6E60>
  ;; Note: the exact pointer representation varies from lisp to lisp.
  

See Also

defcstruct
foreign-slot-value
foreign-slot-names
foreign-slot-offset


Next: , Previous: foreign-slot-pointer, Up: Foreign Types

foreign-slot-value

Syntax

— Accessor: foreign-slot-value ptr type slot-name => object

Arguments and Values

ptr
A pointer to a structure.
type
A foreign structure type.
slot-name
A symbol naming a slot in the structure type.
object
The object contained in the slot specified by slot-name.

Description

For simple slots, foreign-slot-value returns the value of the object, such as a Lisp integer or pointer. In C, this would be expressed as ptr->slot.

For aggregate slots, a pointer inside the structure to the beginning of the slot's data is returned. In C, this would be expressed as &ptr->slot. This pointer and the memory it points to have the same extent as ptr.

There are compiler macros for foreign-slot-value and its setf expansion that open code the memory access when type and slot-names are constant at compile-time.

Examples

  (defcstruct point
    "Pointer structure."
    (x :int)
    (y :int))
   
  CFFI> (with-foreign-object (ptr 'point)
          ;; Initialize the slots
          (setf (foreign-slot-value ptr 'point 'x) 42
                (foreign-slot-value ptr 'point 'y) 42)
          ;; Return a list with the coordinates
          (with-foreign-slots ((x y) ptr point)
            (list x y)))
  => (42 42)

See Also

defcstruct
foreign-slot-names
foreign-slot-offset
foreign-slot-pointer
with-foreign-slots


Next: , Previous: foreign-slot-value, Up: Foreign Types

foreign-type-alignment

Syntax

— Function: foreign-type-alignment type => alignment

Arguments and Values

type
A foreign type.
alignment
An integer.

Description

The function foreign-type-alignment returns the alignment of type in bytes.

Examples

  CFFI> (foreign-type-alignment :char)
  => 1
  CFFI> (foreign-type-alignment :short)
  => 2
  CFFI> (foreign-type-alignment :int)
  => 4
  (defcstruct foo
    (a :char))
   
  CFFI> (foreign-type-alignment 'foo)
  => 1

See Also

foreign-type-size


Next: , Previous: foreign-type-alignment, Up: Foreign Types

foreign-type-size

Syntax

— Function: foreign-type-size type => size

Arguments and Values

type
A foreign type.
size
An integer.

Description

The function foreign-type-size return the size of type in bytes. This includes any padding within and following the in-memory representation as needed to create an array of type objects.

Examples

  (defcstruct foo
    (a :double)
    (c :char))
   
  CFFI> (foreign-type-size :double)
  => 8
  CFFI> (foreign-type-size :char)
  => 1
  CFFI> (foreign-type-size 'foo)
  => 16

See Also

foreign-type-alignment


Next: , Previous: foreign-type-size, Up: Foreign Types

free-converted-object

Syntax

— Function: free-converted-object foreign-value type params

Arguments and Values

foreign-value
The C object to be freed.
type
A CFFI type specifier.
params
The state returned as the second value from convert-to-foreign; used to implement the third argument to free-translated-object.

Description

The return value is unspecified.

This is an external interface to the type translation facility. In the implementation, all foreign functions are ultimately defined as type translation wrappers around primitive foreign function invocations.

This function is available mostly for inspection of the type translation process, and possibly optimization of special cases of your foreign function calls.

Its behavior is better described under free-translated-object's documentation.

Examples

  CFFI-USER> (convert-to-foreign "a boat" :string)
  => #<FOREIGN-ADDRESS #x097ACDC0>
  => (T)
  CFFI-USER> (free-converted-object * :string '(t))
  => NIL

See Also

convert-from-foreign
convert-to-foreign
free-translated-object


Next: , Previous: free-converted-object, Up: Foreign Types

free-translated-object

Syntax

— Generic Function: free-translated-object value type-name param

Arguments and Values

pointer
The foreign value returned by translate-to-foreign.
type-name
A symbol naming a foreign type defined by defctype.
param
The second value, if any, returned by translate-to-foreign.

Description

This generic function may be specialized by user code to perform automatic deallocation of foreign objects as they are passed to C functions.

Any methods defined on this generic function must EQL-specialize the type-name parameter on a symbol defined as a foreign type by the defctype macro.

See Also

Foreign Type Translators
translate-to-foreign


Next: , Previous: free-translated-object, Up: Foreign Types

translate-from-foreign

Syntax

— Generic Function: translate-from-foreign foreign-value type-name => lisp-value

Arguments and Values

foreign-value
The foreign value to convert to a Lisp object.
type-name
A symbol naming a foreign type defined by defctype.
lisp-value
The lisp value to pass in place of foreign-value to Lisp code.

Description

This generic function is invoked by CFFI to convert a foreign value to a Lisp value, such as when returning from a foreign function, passing arguments to a callback function, or accessing a foreign variable.

To extend the CFFI type system by performing custom translations, this method may be specialized by eql-specializing type-name on a symbol naming a foreign type defined with defctype. This method should return the appropriate Lisp value to use in place of the foreign value.

The results are undefined if the type-name parameter is specialized in any way except an eql specializer on a foreign type defined with defctype. Specifically, translations may not be defined for built-in types.

See Also

Foreign Type Translators
translate-to-foreign
free-translated-object


Next: , Previous: translate-from-foreign, Up: Foreign Types

translate-to-foreign

Syntax

— Generic Function: translate-to-foreign lisp-value type-name => foreign-value, alloc-param

Arguments and Values

lisp-value
The Lisp value to convert to foreign representation.
type-name
A symbol naming a foreign type defined by defctype.
foreign-value
The foreign value to pass in place of lisp-value to foreign code.
alloc-param
If present, this value will be passed to free-translated-object.

Description

This generic function is invoked by CFFI to convert a Lisp value to a foreign value, such as when passing arguments to a foreign function, returning a value from a callback, or setting a foreign variable. A “foreign value” is one appropriate for passing to the next-lowest translator, including the low-level translators that are ultimately invoked invisibly with CFFI.

To extend the CFFI type system by performing custom translations, this method may be specialized by eql-specializing type-name on a symbol naming a foreign type defined with defctype. This method should return the appropriate foreign value to use in place of the Lisp value.

In cases where CFFI can determine the lifetime of the foreign object returned by this method, it will invoke free-translated-object on the foreign object at the appropriate time. If translate-to-foreign returns a second value, it will be passed as the param argument to free-translated-object. This can be used to establish communication between the allocation and deallocation methods.

The results are undefined if the type-name parameter is specialized in any way except an eql specializer on a foreign type defined with defctype. Specifically, translations may not be defined for built-in types.

See Also

Foreign Type Translators
translate-from-foreign
free-translated-object


Previous: translate-to-foreign, Up: Foreign Types

with-foreign-slots

Syntax

— Macro: with-foreign-slots (vars ptr type) &body body

Arguments and Values

vars
A list of symbols.
ptr
A foreign pointer to a structure.
type
A structure type.
body
A list of forms to be executed.

Description

The with-foreign-slots macro creates local symbol macros for each var in vars to reference foreign slots in ptr of type. It is similar to Common Lisp's with-slots macro.

Examples

  (defcstruct tm
    (sec :int)
    (min :int)
    (hour :int)
    (mday :int)
    (mon  :int)
    (year :int)
    (wday :int)
    (yday :int)
    (isdst  :boolean)
    (zone   :string)
    (gmtoff :long))
   
  CFFI> (with-foreign-object (time :int)
          (setf (mem-ref time :int)
                (foreign-funcall "time" :pointer (null-pointer) :int))
          (foreign-funcall "gmtime" :pointer time tm))
  => #<A Mac Pointer #x102A30>
  CFFI> (with-foreign-slots ((sec min hour mday mon year) * tm)
          (format nil "~A:~A:~A, ~A/~A/~A"
                  hour min sec (+ 1900 year) mon mday))
  => "7:22:47, 2005/8/2"

See Also

defcstruct
defcunion
foreign-slot-value


Next: , Previous: Foreign Types, Up: Top

6 Pointers

All C data in CFFI is referenced through pointers. This includes defined C variables that hold immediate values, and integers.

To see why this is, consider the case of the C integer. It is not only an arbitrary representation for an integer, congruent to Lisp's fixnums; the C integer has a specific bit pattern in memory defined by the C ABI. Lisp has no such constraint on its fixnums; therefore, it only makes sense to think of fixnums as C integers if you assume that CFFI converts them when necessary, such as when storing one for use in a C function call, or as the value of a C variable. This requires defining an area of memory11, represented through an effective address, and storing it there.

Due to this compartmentalization, it only makes sense to manipulate raw C data in Lisp through pointers to it. For example, while there may be a Lisp representation of a struct that is converted to C at store time, you may only manipulate its raw data through a pointer. The C compiler does this also, albeit informally.

Dictionary


Next: , Up: Pointers

6.1 Basic Pointer Operations

Manipulating pointers proper can be accomplished through most of the other operations defined in the Pointers dictionary, such as make-pointer, pointer-address, and pointer-eq. When using them, keep in mind that they merely manipulate the Lisp representation of pointers, not the values they point to.


Next: , Previous: Basic Pointer Operations, Up: Pointers

6.2 Allocating Foreign Memory

CFFI provides support for stack and heap C memory allocation. Stack allocation, done with with-foreign-object, is sometimes called “dynamic” allocation in Lisp, because memory allocated as such has dynamic extent, much as with let bindings of special variables.

This should not be confused with what C calls “dynamic” allocation, or that done with malloc and friends. This sort of heap allocation is done with foreign-alloc, creating objects that exist until freed with foreign-free.


Next: , Previous: Allocating Foreign Memory, Up: Pointers

6.3 Accessing Foreign Memory

When manipulating raw C data, consider that all pointers are pointing to an array. When you only want one C value, such as a single struct, this array only has one such value. It is worthwhile to remember that everything is an array, though, because this is also the semantic that C imposes natively.

C values are accessed as the setf-able places defined by mem-aref and mem-ref. Given a pointer and a CFFI type (see Foreign Types), either of these will dereference the pointer, translate the C data there back to Lisp, and return the result of said translation, performing the reverse operation when setf-ing. To decide which one to use, consider whether you would use the array index operator [n] or the pointer dereference * in C; use mem-aref for array indexing and mem-ref for pointer dereferencing.


Next: , Previous: Accessing Foreign Memory, Up: Pointers

foreign-free

Syntax

— Function: foreign-free ptr => undefined

Arguments and Values

ptr
A foreign pointer.

Description

The foreign-free function frees a ptr previously allocated by foreign-alloc. The consequences of freeing a given pointer twice are undefined.

Examples

  CFFI> (foreign-alloc :int)
  => #<A Mac Pointer #x1022E0>
  CFFI> (foreign-free *)
  => NIL

See Also

foreign-alloc
with-foreign-pointer


Next: , Previous: foreign-free, Up: Pointers

foreign-alloc

Syntax

— Function: foreign-alloc type &key initial-element initial-contents (count 1) null-terminated-p => pointer

Arguments and Values

type
A foreign type.
initial-element
A Lisp object.
initial-contents
A sequence.
count
An integer. Defaults to 1 or the length of initial-contents if supplied.
null-terminated-p
A boolean, false by default.
pointer
A foreign pointer to the newly allocated memory.

Description

The foreign-alloc function allocates enough memory to hold count objects of type type and returns a pointer. This memory must be explicitly freed using foreign-free once it is no longer needed.

If initial-element is supplied, it is used to initialize the count objects the newly allocated memory holds.

If an initial-contents sequence is supplied, it must have a length less than or equal to count and each of its elements will be used to initialize the contents of the newly allocated memory.

If count is omitted and initial-contents is specified, it will default to (length initial-contents).

initial-element and initial-contents are mutually exclusive.

When null-terminated-p is true, (1+ (max count (length initial-contents))) elements are allocated and the last one is set to NULL. Note that in this case type must be a pointer type (ie. a type that canonicalizes to :pointer), otherwise an error is signaled.

Examples

  CFFI> (foreign-alloc :char)
  => #<A Mac Pointer #x102D80>     ; A pointer to 1 byte of memory.
   
  CFFI> (foreign-alloc :char :count 20)
  => #<A Mac Pointer #x1024A0>     ; A pointer to 20 bytes of memory.
   
  CFFI> (foreign-alloc :int :initial-element 12)
  => #<A Mac Pointer #x1028B0>
  CFFI> (mem-ref * :int)
  => 12
   
  CFFI> (foreign-alloc :int :initial-contents '(1 2 3))
  => #<A Mac Pointer #x102950>
  CFFI> (loop for i from 0 below 3
              collect (mem-aref * :int i))
  => (1 2 3)
   
  CFFI> (foreign-alloc :int :initial-contents #(1 2 3))
  => #<A Mac Pointer #x102960>
  CFFI> (loop for i from 0 below 3
              collect (mem-aref * :int i))
  => (1 2 3)
   
  ;;; Allocate a char** pointer that points to newly allocated memory
  ;;; by the :string type translator for the string "foo".
  CFFI> (foreign-alloc :string :initial-element "foo")
  => #<A Mac Pointer #x102C40>
  ;;; Allocate a null-terminated array of strings.
  ;;; (Note: FOREIGN-STRING-TO-LISP returns NIL when passed a null pointer)
  CFFI> (foreign-alloc :string
                       :initial-contents '("foo" "bar" "baz")
                       :null-terminated-p t)
  => #<A Mac Pointer #x102D20>
  CFFI> (loop for i from 0 below 4
              collect (mem-aref * :string i))
  => ("foo" "bar" "baz" NIL)
  CFFI> (progn
          (dotimes (i 3)
            (foreign-free (mem-aref ** :pointer i)))
          (foreign-free **))
  => nil

See Also

foreign-free
with-foreign-object
with-foreign-pointer


Next: , Previous: foreign-alloc, Up: Pointers

foreign-symbol-pointer

Syntax

— Function: foreign-symbol-pointer foreign-name &key library => pointer

Arguments and Values

foreign-name
A string.
pointer
A foreign pointer, or nil.
library
A Lisp symbol or an instance of foreign-library.

Description

The function foreign-symbol-pointer will return a foreign pointer corresponding to the foreign symbol denoted by the string foreign-name. If a foreign symbol named foreign-name doesn't exist, nil is returned.

ABI name manglings will be performed on foreign-name by foreign-symbol-pointer if necessary. (eg: adding a leading underscore on darwin/ppc)

library should name a foreign library as defined by define-foreign-library, :default (which is the default) or an instance of foreign-library as returned by load-foreign-library.

Important note: do not keep these pointers across saved Lisp cores as the foreign-library may move across sessions.

Examples

  CFFI> (foreign-symbol-pointer "errno")
  => #<A Mac Pointer #xA0008130>
  CFFI> (foreign-symbol-pointer "strerror")
  => #<A Mac Pointer #x9002D0F8>
  CFFI> (foreign-funcall-pointer * () :int (mem-ref ** :int) :string)
  => "No such file or directory"
   
  CFFI> (foreign-symbol-pointer "inexistent symbol")
  => NIL

See Also

defcvar


Next: , Previous: foreign-symbol-pointer, Up: Pointers

inc-pointer

Syntax

— Function: inc-pointer pointer offset => new-pointer

Arguments and Values

pointer
new-pointer
A foreign pointer.
offset
An integer.

Description

The function inc-pointer will return a new-pointer pointing offset bytes past pointer.

Examples

  CFFI> (foreign-string-alloc "Common Lisp")
  => #<A Mac Pointer #x102EA0>
  CFFI> (inc-pointer * 7)
  => #<A Mac Pointer #x102EA7>
  CFFI> (foreign-string-to-lisp *)
  => "Lisp"

See Also

incf-pointer
make-pointer
pointerp
null-pointer
null-pointer-p


Next: , Previous: inc-pointer, Up: Pointers

inc-pointer

Syntax

— Macro: incf-pointer place &optional (offset 1) => new-pointer

Arguments and Values

place
A setf place.
new-pointer
A foreign pointer.
offset
An integer.

Description

The incf-pointer macro takes the foreign pointer from place and creates a new-pointer incremented by offset bytes and which is stored in place.

Examples

  CFFI> (defparameter *two-words* (foreign-string-alloc "Common Lisp"))
  => *TWO-WORDS*
  CFFI> (defparameter *one-word* *two-words*)
  => *ONE-WORD*
  CFFI> (incf-pointer *one-word* 7)
  => #.(SB-SYS:INT-SAP #X00600457)
  CFFI> (foreign-string-to-lisp *one-word*)
  => "Lisp"
  CFFI> (foreign-string-to-lisp *two-words*)
  => "Common Lisp"

See Also

inc-pointer
make-pointer
pointerp
null-pointer
null-pointer-p


Next: , Previous: incf-pointer, Up: Pointers

make-pointer

Syntax

— Function: make-pointer address => ptr

Arguments and Values

address
An integer.
ptr
A foreign pointer.

Description

The function make-pointer will return a foreign pointer pointing to address.

Examples

  CFFI> (make-pointer 42)
  => #<FOREIGN-ADDRESS #x0000002A>
  CFFI> (pointerp *)
  => T
  CFFI> (pointer-address **)
  => 42
  CFFI> (inc-pointer *** -42)
  => #<FOREIGN-ADDRESS #x00000000>
  CFFI> (null-pointer-p *)
  => T

See Also

inc-pointer
null-pointer
null-pointer-p
pointerp
pointer-address
pointer-eq
mem-ref


Next: , Previous: make-pointer, Up: Pointers

mem-aref

Syntax

— Accessor: mem-aref ptr type &optional (index 0)

(setf (mem-aref ptr type &optional (index 0)) new-value)

Arguments and Values

ptr
A foreign pointer.
type
A foreign type.
index
An integer.
new-value
A Lisp value compatible with type.

Description

The mem-aref function is similar to mem-ref but will automatically calculate the offset from an index.

  (mem-aref ptr type n)
   
  ;; is identical to:
   
  (mem-ref ptr type (* n (foreign-type-size type)))

Examples

  CFFI> (with-foreign-string (str "Hello, foreign world!")
          (mem-aref str :char 6))
  => 32
  CFFI> (code-char *)
  => #\Space
   
  CFFI> (with-foreign-object (array :int 10)
          (loop for i below 10
                do (setf (mem-aref array :int i) (random 100)))
          (loop for i below 10 collect (mem-aref array :int i)))
  => (22 7 22 52 69 1 46 93 90 65)

See Also

mem-ref


Next: , Previous: mem-aref, Up: Pointers

mem-ref

Syntax

— Accessor: mem-ref ptr type &optional offset => object

Arguments and Values

ptr
A pointer.
type
A foreign type.
offset
An integer (in byte units).
object
The value ptr points to.

Description

Examples

  CFFI> (with-foreign-string (ptr "Saluton")
          (setf (mem-ref ptr :char 3) (char-code #\a))
          (loop for i from 0 below 8
                collect (code-char (mem-ref ptr :char i))))
  => (#\S #\a #\l #\a #\t #\o #\n #\Null)
  CFFI> (setq ptr-to-int (foreign-alloc :int))
  => #<A Mac Pointer #x1047D0>
  CFFI> (mem-ref ptr-to-int :int)
  => 1054619
  CFFI> (setf (mem-ref ptr-to-int :int) 1984)
  => 1984
  CFFI> (mem-ref ptr-to-int :int)
  => 1984

See Also

mem-aref


Next: , Previous: mem-ref, Up: Pointers

null-pointer

Syntax

— Function: null-pointer => pointer

Arguments and Values

pointer
A NULL pointer.

Description

The function null-pointer returns a null pointer.

Examples

  CFFI> (null-pointer)
  => #<A Null Mac Pointer>
  CFFI> (pointerp *)
  => T

See Also

null-pointer-p
make-pointer


Next: , Previous: null-pointer, Up: Pointers

null-pointer-p

Syntax

— Function: null-pointer-p ptr => boolean

Arguments and Values

ptr
A foreign pointer that may be a null pointer.
boolean
T or NIL.

Description

The function null-pointer-p returns true if ptr is a null pointer and false otherwise.

Examples

  CFFI> (null-pointer-p (null-pointer))
  => T
  (defun contains-str-p (big little)
    (not (null-pointer-p
          (foreign-funcall "strstr" :string big :string little :pointer))))
   
  CFFI> (contains-str-p "Popcorns" "corn")
  => T
  CFFI> (contains-str-p "Popcorns" "salt")
  => NIL

See Also

null-pointer
pointerp


Next: , Previous: null-pointer-p, Up: Pointers

pointerp

Syntax

— Function: pointerp ptr => boolean

Arguments and Values

ptr
An object that may be a foreign pointer.
boolean
T or NIL.

Description

The function pointerp returns true if ptr is a foreign pointer and false otherwise.

Implementation-specific Notes

In Allegro CL, foreign pointers are integers thus in this implementation pointerp will return true for any ordinary integer.

Examples

  CFFI> (foreign-alloc 32)
  => #<A Mac Pointer #x102D20>
  CFFI> (pointerp *)
  => T
  CFFI> (pointerp "this is not a pointer")
  => NIL

See Also

make-pointer null-pointer-p


Next: , Previous: pointerp, Up: Pointers

pointer-address

Syntax

— Function: pointer-address ptr => address

Arguments and Values

ptr
A foreign pointer.
address
An integer.

Description

The function pointer-address will return the address of a foreign pointer ptr.

Examples

  CFFI> (pointer-address (null-pointer))
  => 0
  CFFI> (pointer-address (make-pointer 123))
  => 123

See Also

make-pointer
inc-pointer
null-pointer
null-pointer-p
pointerp
pointer-eq
mem-ref


Next: , Previous: pointer-address, Up: Pointers

pointer-eq

Syntax

— Function: pointer-eq ptr1 ptr2 => boolean

Arguments and Values

ptr1
ptr2
A foreign pointer.
boolean
T or NIL.

Description

The function pointer-eq returns true if ptr1 and ptr2 point to the same memory address and false otherwise.

Implementation-specific Notes

The representation of foreign pointers varies across the various Lisp implementations as does the behaviour of the built-in Common Lisp equality predicates. Comparing two pointers that point to the same address with EQ Lisps will return true on some Lisps, others require more general predicates like EQL or EQUALP and finally some will return false using any of these predicates. Therefore, for portability, you should use POINTER-EQ.

Examples

This is an example using SBCL, see the implementation-specific notes above.

  CFFI> (eql (null-pointer) (null-pointer))
  => NIL
  CFFI> (pointer-eq (null-pointer) (null-pointer))
  => T

See Also

inc-pointer


Next: , Previous: pointer-eq, Up: Pointers

with-foreign-object

Syntax

— Macro: with-foreign-object (var type &optional count) &body body
— Macro: with-foreign-objects (bindings) &body body

bindings ::= {(var type &optional count)}*

Arguments and Values

var
A symbol.
type
A foreign type, evaluated.
count
An integer.

Description

The macros with-foreign-object and with-foreign-objects bind var to a pointer to count newly allocated objects of type type during body. The buffer has dynamic extent and may be stack allocated if supported by the host Lisp.

Examples

  CFFI> (with-foreign-object (array :int 10)
          (dotimes (i 10)
            (setf (mem-aref array :int i) (random 100)))
          (loop for i below 10
                collect (mem-aref array :int i)))
  => (22 7 22 52 69 1 46 93 90 65)

See Also

foreign-alloc


Previous: with-foreign-object, Up: Pointers

with-foreign-pointer

Syntax

— Macro: with-foreign-pointer (var size &optional size-var) &body body

Arguments and Values

var
size-var
A symbol.
size
An integer.
body
A list of forms to be executed.

Description

The with-foreign-pointer macro, binds var to size bytes of foreign memory during body. The pointer in var is invalid beyond the dynamic extend of body and may be stack-allocated if supported by the implementation.

If size-var is supplied, it will be bound to size during body.

Examples

  CFFI> (with-foreign-pointer (string 4 size)
          (setf (mem-ref string :char (1- size)) 0)
          (lisp-string-to-foreign "Popcorns" string size)
          (loop for i from 0 below size
                collect (code-char (mem-ref string :char i))))
  => (#\P #\o #\p #\Null)

See Also

foreign-alloc
foreign-free


Next: , Previous: Pointers, Up: Top

7 Strings

As with many languages, Lisp and C have special support for logical arrays of characters, going so far as to give them a special name, “strings”. In that spirit, CFFI provides special support for translating between Lisp and C strings.

The :string type and the symbols related below also serve as an example of what you can do portably with CFFI; were it not included, you could write an equally functional strings.lisp without referring to any implementation-specific symbols.

Dictionary


Next: , Up: Strings

foreign-string-alloc

Syntax

— Function: foreign-string-alloc string-or-ub8-array => pointer

Arguments and Values

string-or-ub8-array
A Lisp string or a Lisp array with element-type (unsigned-byte 8).
pointer
A pointer to the newly allocated foreign string.

Description

The foreign-string-alloc function allocates a foreign string containing a Lisp string or (unsigned-byte 8) array.

This string must be freed with foreign-string-free.

Examples

  CFFI> (setq str (foreign-string-alloc "Hello, foreign world!"))
  => #<FOREIGN-ADDRESS #x00400560>
  CFFI> (foreign-funcall "strlen" :pointer str :int)
  => 21

See Also

foreign-string-free
with-foreign-string


Next: , Previous: foreign-string-alloc, Up: Strings

foreign-string-free

Syntax

— Function: foreign-string-free pointer

Arguments and Values

pointer
A pointer to a string allocated by foreign-string-alloc.

Description

The foreign-string-free function frees a foreign string allocated by foreign-string-alloc.

Examples

See Also

foreign-string-alloc


Next: , Previous: foreign-string-free, Up: Strings

foreign-string-to-lisp

Syntax

— Function: foreign-string-to-lisp ptr &optional size null-terminated-p => string

Arguments and Values

ptr
A pointer.
size
The maximum string size. array-total-size-limit, by default.
null-terminated-p
Specifies if the string ptr points to is null terminated. True, by default.

Description

The foreign-string-to-lisp function copies at most size characters from ptr into a Lisp string.

When null-terminated-p is true (the default), characters are copied until size is reached or a NULL character is found.

If ptr is a null pointer, returns nil.

Note that the :string type will automatically convert between Lisp strings and foreign strings.

Examples

  CFFI> (foreign-funcall "getenv" :string "HOME" :pointer)
  => #<FOREIGN-ADDRESS #xBFFFFFD5>
  CFFI> (foreign-string-to-lisp *)
  => "/Users/luis"

See Also

lisp-string-to-foreign
foreign-string-alloc


Next: , Previous: foreign-string-to-lisp, Up: Strings

lisp-string-to-foreign

Syntax

— Function: lisp-string-to-foreign string-or-ub8-array ptr size

Arguments and Values

string-or-ub8-array
A Lisp string or a Lisp (unsigned-byte 8) array.
ptr
A foreign pointer.
size
An integer.

Description

The lisp-string-to-foreign function copies at most size-1 characters from a Lisp string or (unsigned-byte 8) arrayto ptr. The foreign string will be null-terminated.

Examples

  CFFI> (with-foreign-pointer-as-string (str 255)
          (lisp-string-to-foreign "Hello, foreign world!" str 6))
  => "Hello"
   
  CFFI> (with-foreign-pointer-as-string (str 255)
          (lisp-string-to-foreign
           (make-array 6 :element-type '(unsigned-byte 8)
                       :initial-contents '(65 66 67 68 69 60))
           str 4))
  => "ABC"

See Also

foreign-string-alloc
foreign-string-to-lisp
with-foreign-pointer-as-string


Next: , Previous: lisp-string-to-foreign, Up: Strings

with-foreign-string

Syntax

— Macro: with-foreign-string (var lisp-string-or-ub8-array) &body body

Arguments and Values

var
A symbol.
lisp-string-or-ub8-array
A Lisp string or a Lisp array with element type (unsigned-byte 8).
body
A list of forms to be executed.

Description

The with-foreign-string macro will bind var to a newly allocated foreign string containing lisp-string-or-ub8-array.

Examples

  CFFI> (with-foreign-string (foo "12345")
          (foreign-funcall "strlen" :pointer foo :int))
  => 5
   
  CFFI> (let ((array (coerce #(84 117 114 97 110 103 97)
                             '(array (unsigned-byte 8)))))
          (with-foreign-string (foreign-string array)
            (foreign-string-to-lisp foreign-string)))
  => "Turanga"

See Also

foreign-string-alloc
with-foreign-pointer-as-string


Previous: with-foreign-string, Up: Strings

with-foreign-pointer-as-string

Syntax

— Macro: with-foreign-pointer-as-string (var size &optional size-var) &body body

Arguments and Values

var
A symbol.
lisp-string
A Lisp string.
body
List of forms to be executed.

Description

The with-foreign-pointer-as-string macro is similar to with-foreign-pointer except that var, as a Lisp string, is used as the returned value of an implicit progn around body.

Examples

  CFFI> (with-foreign-pointer-as-string (str 6 str-size)
          (lisp-string-to-foreign "Hello, foreign world!" str str-size))
  => "Hello"

See Also

foreign-string-alloc
with-foreign-string


Next: , Previous: Strings, Up: Top

8 Variables

Dictionary


Next: , Up: Variables

defcvar

Syntax

— Macro: defcvar name-and-options type documentation => lisp-name

name-and-options ::= name | (name &key read-only (library :default)) name ::= lisp-name [foreign-name] | foreign-name [lisp-name]

Arguments and Values

foreign-name
A string denoting a foreign function.
lisp-name
A symbol naming the Lisp function to be created.
type
A foreign type.
read-only
A boolean.
documentation
A Lisp string; not evaluated.

Description

The defcvar macro defines a symbol macro lisp-name that looks up foreign-name and dereferences it acording to type. It can also be setfed, unless read-only is true, in which case an error will be signaled.

When one of lisp-name or foreign-name is omitted, the other is automatically derived using the following rules:

Examples

  CFFI> (defcvar "errno" :int)
  => *ERRNO*
  CFFI> (foreign-funcall "strerror" :int *errno* :string)
  => "Inappropriate ioctl for device"
  CFFI> (setf *errno* 1)
  => 1
  CFFI> (foreign-funcall "strerror" :int *errno* :string)
  => "Operation not permitted"

Trying to modify a read-only foreign variable:

  CFFI> (defcvar ("errno" +error-number+) :int :read-only t)
  => +ERROR-NUMBER+
  CFFI> (setf +error-number+ 12)
  ;; error--> Trying to modify read-only foreign var: +ERROR-NUMBER+.
  

Note that accessing errno this way won't work with every C standard library.

See Also

get-var-pointer


Previous: defcvar, Up: Variables

get-var-pointer

Syntax

— Function: get-var-pointer symbol => pointer

Arguments and Values

symbol
A symbol denoting a foreign variable defined with defcvar.
pointer
A foreign pointer.

Description

The function get-var-pointer will return a pointer to the foreign global variable symbol previously defined with defcvar.

Examples

  CFFI> (defcvar "errno" :int :read-only t)
  => *ERRNO*
  CFFI> *errno*
  => 25
  CFFI> (get-var-pointer '*errno*)
  => #<A Mac Pointer #xA0008130>
  CFFI> (mem-ref * :int)
  => 25

See Also

defcvar


Next: , Previous: Variables, Up: Top

9 Functions

Dictionary


Next: , Up: Functions

9.1 Calling Foreign Functions


Next: , Previous: Calling Foreign Functions, Up: Functions

9.2 Defining Foreign Functions


Next: , Previous: Defining Foreign Functions, Up: Functions

defcfun

Syntax

— Macro: defcfun name-and-options return-type &body arguments [&rest] => lisp-name
name-and-options name | (name &key library calling-convention cconv)
name ::= lisp-name [foreign-name] | foreign-name [lisp-name]
arguments ::= { (arg-name arg-type) }*

Arguments and Values

foreign-name
A string denoting a foreign function.
lisp-name
A symbol naming the Lisp function to be created.
arg-name
A symbol.
return-type
arg-type
A foreign type.
calling-convention
cconv
One of :cdecl (default) or stdcall.
library
A symbol designating a foreign library.

Description

The defcfun macro provides a declarative interface for defining Lisp functions that call foreign functions.

When one of lisp-name or foreign-name is omitted, the other is automatically derived using the following rules:

If you place the symbol &rest in the end of the argument list after the fixed arguments, defcfun will treat the foreign function as a variadic function. The variadic arguments should be passed in a way similar to what foreign-funcall would expect. Unlike foreign-funcall though, defcfun will take care of doing argument promotion. Note that in this case defcfun will generate a Lisp macro instead of a function and will only work for Lisps that support foreign-funcall.

Examples

  (defcfun "strlen" :int (n :string))
   
  CFFI> (strlen "123")
  => 3
  (defcfun ("abs" c-abs) :int (n :int))
   
  CFFI> (c-abs -42)
  => 42

Variadic function example:

  (defcfun "sprintf" :int
    (str :pointer)
    (control :string)
    &rest)
   
  CFFI> (with-foreign-pointer-as-string (s 100)
          (sprintf s "%c %d %.2f %s" :char 90 :short 42 :float pi
                   :string "super-locrian"))
  => "A 42 3.14 super-locrian"

See Also

foreign-funcall
foreign-funcall-pointer


Next: , Previous: defcfun, Up: Functions

foreign-funcall

Syntax

— Macro: foreign-funcall name-and-options &rest arguments => return-value

arguments ::= { arg-type arg }* [return-type] name-and-options ::= name | ( name &key library calling-convention cconv)

Arguments and Values

name
A Lisp string.
arg-type
A foreign type.
arg
An argument of type arg-type.
return-type
A foreign type, :void by default.
return-value
A lisp object.
library
A lisp symbol; not evaluated.
calling-convention
cconv
One of :cdecl (default) or :stdcall.

Description

The foreign-funcall macro is the main primitive for calling foreign functions.

Note: The return value of foreign-funcall on functions with a :void return type is still undefined.

Implementation-specific Notes

Examples

  CFFI> (foreign-funcall "strlen" :string "foo" :int)
  => 3

Given the C code:

  void print_number(int n)
  {
      printf("N: %d\n", n);
  }
  CFFI> (foreign-funcall "print_number" :int 123456)
  -| N: 123456
  => NIL

Or, equivalently:

  CFFI> (foreign-funcall "print_number" :int 123456 :void)
  -| N: 123456
  => NIL
  CFFI> (foreign-funcall "printf" :string (format nil "%s: %d.~%")
                         :string "So long and thanks for all the fish"
                         :int 42 :int)
  -| So long and thanks for all the fish: 42.
  => 41

See Also

defcfun
foreign-funcall-pointer


Previous: foreign-funcall, Up: Functions

foreign-funcall-pointer

Syntax

— Macro: foreign-funcall pointer options &rest arguments => return-value

arguments ::= { arg-type arg }* [return-type] options ::= ( &key calling-convention cconv )

Arguments and Values

pointer
A foreign pointer.
arg-type
A foreign type.
arg
An argument of type arg-type.
return-type
A foreign type, :void by default.
return-value
A lisp object.
calling-convention
cconv
One of :cdecl (default) or :stdcall.

Description

The foreign-funcall macro is the main primitive for calling foreign functions.

Note: The return value of foreign-funcall on functions with a :void return type is still undefined.

Implementation-specific Notes

Examples

  CFFI> (foreign-funcall-pointer (foreign-symbol-pointer "abs")
                                 :int -42 :int)
  => 42

See Also

defcfun
foreign-funcall


Next: , Previous: Functions, Up: Top

10 Libraries

Dictionary


Next: , Up: Libraries

10.1 Defining a library

Almost all foreign code you might want to access exists in some kind of shared library. The meaning of shared library varies among platforms, but for our purposes, we will consider it to include .so files on unix, frameworks on Darwin (and derivatives like Mac OS X), and .dll files on Windows.

Bringing one of these libraries into the Lisp image is normally a two-step process.

  1. Describe to CFFI how to load the library at some future point, depending on platform and other factors, with a define-foreign-library top-level form.
  2. Load the library so defined with either a top-level use-foreign-library form or by calling the function load-foreign-library.

See Loading foreign libraries, for a working example of the above two steps.


Next: , Previous: Defining a library, Up: Libraries

10.2 Library definition style

Looking at the libcurl library definition presented earlier, you may ask why we did not simply do this:

  (define-foreign-library libcurl
    (t (:default "libcurl")))

Indeed, this would work just as well on the computer on which I tested the tutorial. There are a couple of good reasons to provide the .so's current version number, however. Namely, the versionless .so is not packaged on most unix systems along with the actual, fully-versioned library; instead, it is included in the “development” package along with C headers and static .a libraries.

The reason CFFI does not try to account for this lies in the meaning of the version numbers. A full treatment of shared library versions is beyond this manual's scope; see Library interface versions, for helpful information for the unfamiliar. For our purposes, consider that a mismatch between the library version with which you tested and the installed library version may cause undefined behavior.12

Implementor's note: Maybe some notes should go here about OS X, which I know little about. –stephen


Next: , Previous: Library definition style, Up: Libraries

close-foreign-library

Syntax

— Function: close-foreign-library library => success

Arguments and Values

library
A symbol or an instance of foreign-library.
success
A Lisp boolean.

Description

Closes library which can be a symbol designating a library define through define-foreign-library or an instance of foreign-library as returned by load-foreign-library.

See Also

define-foreign-library
load-foreign-library
use-foreign-library


Next: , Previous: close-foreign-library, Up: Libraries

*darwin-framework-directories*

Syntax

— Special Variable: *darwin-framework-directories*

Value type

A list, in which each element is a string, a pathname, or a simple Lisp expression.

Initial value

A list containing the following, in order: an expression corresponding to Darwin path ~/Library/Frameworks/, #P"/Library/Frameworks/", and #P"/System/Library/Frameworks/".

Description

The meaning of “simple Lisp expression” is explained in *foreign-library-directories*. In contrast to that variable, this is not a fallback search path; the default value described above is intended to be a reasonably complete search path on Darwin systems.

Examples

  CFFI> (load-foreign-library '(:framework "OpenGL"))
  => #P"/System/Library/Frameworks/OpenGL.framework/OpenGL"

See also

*foreign-library-directories*
define-foreign-library


Next: , Previous: *darwin-framework-directories*, Up: Libraries

define-foreign-library

Syntax

— Macro: define-foreign-library name-and-options { load-clause }* => name

name-and-options ::= name | (name &key calling-convention cconv) load-clause ::= (feature library &key calling-convention cconv)

Arguments and Values

name
A symbol.
feature
A feature expression.
library
A library designator.
calling-convention
cconv
One of :cdecl (default) or :stdcall

Description

Creates a new library designator called name. The load-clauses describe how to load that designator when passed to load-foreign-library or use-foreign-library.

When trying to load the library name, the relevant function searches the load-clauses in order for the first one where feature evaluates to true. That happens for any of the following situations:13

  1. If feature is a symbol (idiomatically a keyword), a symbol with the same name, but interned into the cffi-features package, is present in common-lisp:*features*.
  2. If feature is a list, depending on (first feature), a keyword:
    :and
    All of the feature expressions in (rest feature) are true.
    :or
    At least one of the feature expressions in (rest feature) is true.
    :not
    The feature expression (second feature) is not true.
  3. Finally, if feature is t, this load-clause is picked unconditionally.

Upon finding the first true feature, the library loader then loads the library. The meaning of “library designator” is described in load-foreign-library.

Functions associated to a library defined by define-foreign-library (e.g. through defcfun's :library option, will inherit the library's options. The precedence is as follows:

  1. defcfun/foreign-funcall specific options;
  2. load-clause options;
  3. global library options (the name-and-options argument)

Examples

See Loading foreign libraries.

See Also

close-foreign-library
load-foreign-library


Next: , Previous: define-foreign-library, Up: Libraries

*foreign-library-directories*

Syntax

— Special Variable: *foreign-library-directories*

Value type

A list, in which each element is a string, a pathname, or a simple Lisp expression.

Initial value

The empty list.

Description

You should not have to use this variable.

Most, if not all, Lisps supported by CFFI have a reasonable default search algorithm for foreign libraries. For example, Lisps for unix usually call dlopen(3), which in turn looks in the system library directories. Only if that fails does CFFI look for the named library file in these directories, and load it from there if found.

Thus, this is intended to be a CFFI-only fallback to the library search configuration provided by your operating system. For example, if you distribute a foreign library with your Lisp package, you can add the library's containing directory to this list and portably expect CFFI to find it.

A simple Lisp expression is intended to provide functionality commonly used in search paths such as ASDF's14, and is defined recursively as follows:15

  1. A list, whose `first' is a function designator, and whose `rest' is a list of simple Lisp expressions to be evaluated and passed to the so-designated function. The result is the result of the function call.
  2. A symbol, whose result is its symbol value.
  3. Anything else evaluates to itself.

Examples

  $ ls
  -| liblibli.so    libli.lisp

In libli.lisp:

  (pushnew #P"/home/sirian/lisp/libli/" *foreign-library-directories*
           :test #'equal)
   
  (load-foreign-library '(:default "liblibli"))

See also

*darwin-framework-directories*
define-foreign-library


Next: , Previous: *foreign-library-directories*, Up: Libraries

load-foreign-library

Syntax

— Function: load-foreign-library library => handler

Arguments and Values

library
A library designator.
handler
An instance of foreign-library.

Description

Load the library indicated by library. A library designator is defined as follows:

  1. If a symbol, is considered a name previously defined with define-foreign-library.
  2. If a string or pathname, passed as a namestring directly to the implementation's foreign library loader. If that fails, search the directories in *foreign-library-directories* with cl:probe-file; if found, the absolute path is passed to the implementation's loader.
  3. If a list, the meaning depends on (first library):
    :framework
    The second list element is taken to be a Darwin framework name, which is then searched in *darwin-framework-directories*, and loaded when found.
    :or
    Each remaining list element, itself a library designator, is loaded in order, until one succeeds.
    :default
    The name is transformed according to the platform's naming convention to shared libraries, and the resultant string is loaded as a library designator. For example, on unix, the name is suffixed with .so.

If the load fails, signal a load-foreign-library-error.

Please note: For system libraries, you should not need to specify the directory containing the library. Each operating system has its own idea of a default search path, and you should rely on it when it is reasonable.

Implementation-specific Notes

On ECL platforms where its dynamic FFI is not supported (ie. when :dffi is not present in *features*), cffi:load-foreign-library does not work and you must use ECL's own ffi:load-foreign-library with a constant string argument.

Examples

See Loading foreign libraries.

See Also

close-foreign-library
*darwin-framework-directories*
define-foreign-library
*foreign-library-directories*
load-foreign-library-error
use-foreign-library


Next: , Previous: load-foreign-library, Up: Libraries

load-foreign-library-error

Syntax

— Condition Type: load-foreign-library-error

Class precedence list

load-foreign-library-error, error, serious-condition, condition, t

Description

Signalled when a foreign library load completely fails. The exact meaning of this varies depending on the real conditions at work, but almost universally, the implementation's error message is useless. However, CFFI does provide the useful restarts retry and use-value; invoke the retry restart to try loading the foreign library again, or the use-value restart to try loading a different foreign library designator.

See also

load-foreign-library


Previous: load-foreign-library-error, Up: Libraries

use-foreign-library

Syntax

— Macro: use-foreign-library name

Arguments and values

name
A library designator; unevaluated.

Description

See load-foreign-library, for the meaning of “library designator”. This is intended to be the top-level form used idiomatically after a define-foreign-library form to go ahead and load the library. Finally, on implementations where the regular evaluation rule is insufficient for foreign library loading, it loads it at the required time.16

Examples

See Loading foreign libraries.

See also

load-foreign-library


Next: , Previous: Libraries, Up: Top

11 Callbacks

Dictionary


Next: , Up: Callbacks

callback

Syntax

— Macro: callback symbol => pointer

Arguments and Values

symbol
A symbol denoting a callback.
pointer
new-value
A pointer.

Description

The callback macro is analogous to the standard CL special operator function and will return a pointer to the callback denoted by the symbol name.

Examples

  CFFI> (defcallback sum :int ((a :int) (b :int))
          (+ a b))
  => SUM
  CFFI> (callback sum)
  => #<A Mac Pointer #x102350>

See Also

get-callback
defcallback


Next: , Previous: callback, Up: Callbacks

defcallback

Syntax

— Macro: defcallback name-and-options return-type arguments &body body => name

name-and-options ::= name | (name &key calling-convention cconv) arguments ::= ({ (arg-name arg-type) }*)

Arguments and Values

name
A symbol naming the callback created.
return-type
The foreign type for the callback's return value.
arg-name
A symbol.
arg-type
A foreign type.
calling-convention
cconv
One of :cdecl (default) or :stdcall.

Description

The macro defcallback defines a Lisp function the can be called from C (but not from Lisp). The arguments passed to this function will be converted to the appropriate Lisp representation and its return value will be converted to its C representation.

This Lisp function can be accessed by the callback macro or the get-callback function.

Portability note: defcallback will not work correctly on some Lisps if it's not a top-level form.

Examples

  (defcfun "qsort" :void
    (base :pointer)
    (nmemb :int)
    (size :int)
    (fun-compar :pointer))
   
  (defcallback < :int ((a :pointer) (b :pointer))
    (let ((x (mem-ref a :int))
          (y (mem-ref b :int)))
      (cond ((> x y) 1)
            ((< x y) -1)
            (t 0))))
   
  CFFI> (with-foreign-object (array :int 10)
          ;; Initialize array.
          (loop for i from 0 and n in '(7 2 10 4 3 5 1 6 9 8)
                do (setf (mem-aref array :int i) n))
          ;; Sort it.
          (qsort array 10 (foreign-type-size :int) (callback <))
          ;; Return it as a list.
          (loop for i from 0 below 10
                collect (mem-aref array :int i)))
  => (1 2 3 4 5 6 7 8 9 10)

See Also

callback
get-callback


Previous: defcallback, Up: Callbacks

get-callback

Syntax

— Accessor: get-callback symbol => pointer

Arguments and Values

symbol
A symbol denoting a callback.
pointer
A pointer.

Description

This is the functional version of the callback macro. It returns a pointer to the callback named by symbol suitable, for example, to pass as arguments to foreign functions.

Examples

  CFFI> (defcallback sum :int ((a :int) (b :int))
          (+ a b))
  => SUM
  CFFI> (get-callback 'sum)
  => #<A Mac Pointer #x102350>

See Also

callback
defcallback


Next: , Previous: Callbacks, Up: Top

12 Limitations

These are CFFI's limitations across all platforms; for information on the warts on particular Lisp implementations, see Implementation Support.


Next: , Previous: Limitations, Up: Top

Appendix A Platform-specific features

CFFI does some platform tests on loading. The details vary between Lisps; in fact, the purpose is to unify the list of available platform features for use elsewhere in the CFFI code. These features are also part of the public interface; see define-foreign-library.

The exact meanings of the features follow. Though you will usually refer to these symbols as keywords, CFFI internally views them in the package cffi-features.

flat-namespace
This Lisp has a flat namespace for foreign symbols meaning that you won't be able to load two different libraries with homograph functions and successfully differentiate them through the :library option to defcfun, defcvar, etc...
darwin
This operating system is Darwin or a derivative thereof, such as Mac OS X.
no-foreign-funcall
The macro foreign-funcall is not available. On such platforms, the only way to call a foreign function is through defcfun. See foreign-funcall, and defcfun.
no-long-long
The C long long type is not available as a foreign type.
no-stdcall
This Lisp doesn't support the stdcall calling convention. Note that it only makes sense to support stdcall on (32-bit) x86 platforms.
ppc32
The underlying CPU architecture is 32-bit PowerPC.
unix
This operating system is a unix-like, such as GNU/Linux, Darwin, or even Cygwin on Lisps that show the unix-like interface provided by Cygwin to Lisp code.
windows
This operating system is Windows.
x86
The underlying CPU architecture is x86, such as on processors from Intel or AMD.


Next: , Previous: Platform-specific features, Up: Top

Appendix B Glossary

aggregate type
A CFFI type for C data defined as an organization of data of simple type; in structures and unions, which are themselves aggregate types, they are represented by value.
foreign value
This has two meanings; in any context, only one makes sense.

When using type translators, the foreign value is the lower-level Lisp value derived from the object passed to translate-to-foreign (see translate-to-foreign). This value should be a Lisp number or a pointer (satisfies pointerp), and it can be treated like any general Lisp object; it only completes the transformation to a true foreign value when passed through low-level code in the Lisp implementation, such as the foreign function caller or indirect memory addressing combined with a data move.

In other contexts, this refers to a value accessible by C, but which may only be accessed through CFFI functions. The closest you can get to such a foreign value is through a pointer Lisp object, which itself counts as a foreign value in only the previous sense.

simple type
A CFFI type that is ultimately represented as a builtin type; CFFI only provides extra semantics for Lisp that are invisible to C code or data.


Previous: Glossary, Up: Top

Index


Footnotes

[1] Admittedly, this is an advanced issue, and we encourage you to leave this text until you are more familiar with how CFFI works.

[2] This does not apply to structs whose contents are intended to be part of the public library interface. In those cases, a pure Lisp struct definition is always preferred. In fact, many prefer to stay in Lisp and break the encapsulation anyway, placing the burden of correct library interface definition on the library.

[3] Specifically, UFFI, an older FFI that takes a somewhat different approach compared to CFFI. I believe that these days (December 2005) CFFI is more portable and actively developed, though not as mature yet. Consensus in the free unix Common Lisp community seems to be that CFFI is preferred for new development, though UFFI will likely go on for quite some time as many projects already use it. CFFI includes the UFFI-COMPAT package for complete compatibility with UFFI.

[4] This isn't entirely true; some Lisps don't support foreign-funcall, so defcfun is implemented without it. defcfun may also perform optimizations that foreign-funcall cannot.

[5] Another possibility is to allow the caller to specify the desired C type of the third argument. This is essentially what happens in a call to the function written in C.

[6] See Other Kinds of Streams, for a GNU-only way to extend the FILE* type. You could use this to convert Lisp streams to the needed C data. This would be quite involved and far outside the scope of this tutorial.

[7]But I thought Lisp was supposed to protect me from all that buggy C crap!” Before asking a question like that, remember that you are a stranger in a foreign land, whose residents have a completely different set of values.

[8] There are advantages and disadvantages to each approach; I chose to (setf symbol-function) earlier because it entailed generating fewer magic function names.

[9] Unfortunately, we can't protect against all non-local exits, such as returns and throws, because unwind-protect cannot be used to “short-circuit” a non-local exit in Common Lisp, due to proposal minimal in ANSI issue Exit-Extent. Furthermore, binding an error handler prevents higher-up code from invoking restarts that may be provided under the callback's dynamic context. Such is the way of compromise.

[10] It might be better to return (values) than :curle-ok in real code, but this is good for illustration.

[11] The definition of memory includes the CPU registers.

[12] Windows programmers may chafe at adding a unix-specific clause to define-foreign-library. Instead, ask why the Windows solution to library incompatibility is “include your own version of every library you use with every program”.

[13] This is described in cffi-feature-p in libraries.lisp.

[14] See Using asdf to load systems, for information on asdf:*central-registry*.

[15] See mini-eval in libraries.lisp for the source of this definition. As is always the case with a Lisp eval, it's easier to understand the Lisp definition than the english.

[16] Namely, CMUCL. See use-foreign-library in libraries.lisp for details.