2 Funs

2.1  map

The following function, double, doubles every element in a list:

double([H|T]) -> [2*H|double(T)];
double([])    -> [].

Hence, the argument entered as input is doubled as follows:

> double([1,2,3,4]).

The following function, add_one, adds one to every element in a list:

add_one([H|T]) -> [H+1|add_one(T)];
add_one([])    -> [].

The functions double and add_one have a similar structure. This can be used by writing a function map that expresses this similarity:

map(F, [H|T]) -> [F(H)|map(F, T)];
map(F, [])    -> [].

The functions double and add_one can now be expressed in terms of map as follows:

double(L)  -> map(fun(X) -> 2*X end, L).
add_one(L) -> map(fun(X) -> 1 + X end, L).

map(F, List) is a function that takes a function F and a list L as arguments and returns a new list, obtained by applying F to each of the elements in L.

The process of abstracting out the common features of a number of different programs is called procedural abstraction. Procedural abstraction can be used to write several different functions that have a similar structure, but differ in some minor detail. This is done as follows:

  • Step 1. Write one function that represents the common features of these functions.
  • Step 2. Parameterize the difference in terms of functions that are passed as arguments to the common function.

2.2  foreach

This section illustrates procedural abstraction. Initially, the following two examples are written as conventional functions.

This function prints all elements of a list onto a stream:

print_list(Stream, [H|T]) ->
    io:format(Stream, "~p~n", [H]),
    print_list(Stream, T);
print_list(Stream, []) ->

This function broadcasts a message to a list of processes:

broadcast(Msg, [Pid|Pids]) ->
    Pid ! Msg,
    broadcast(Msg, Pids);
broadcast(_, []) ->

These two functions have a similar structure. They both iterate over a list and do something to each element in the list. The "something" is passed on as an extra argument to the function that does this.

The function foreach expresses this similarity:

foreach(F, [H|T]) ->
    foreach(F, T);
foreach(F, []) ->

Using the function foreach, the function print_list becomes:

foreach(fun(H) -> io:format(S, "~p~n",[H]) end, L)

Using the function foreach, the function broadcast becomes:

foreach(fun(Pid) -> Pid ! M end, L)

foreach is evaluated for its side-effect and not its value. foreach(Fun ,L) calls Fun(X) for each element X in L and the processing occurs in the order that the elements were defined in L. map does not define the order in which its elements are processed.

2.3  Syntax of Funs

Funs are written with the following syntax (see Fun Expressions for full description):

F = fun (Arg1, Arg2, ... ArgN) ->

This creates an anonymous function of N arguments and binds it to the variable F.

Another function, FunctionName, written in the same module, can be passed as an argument, using the following syntax:

F = fun FunctionName/Arity

With this form of function reference, the function that is referred to does not need to be exported from the module.

It is also possible to refer to a function defined in a different module, with the following syntax:

F = fun Module:FunctionName/Arity

In this case, the function must be exported from the module in question.

The following program illustrates the different ways of creating funs:

-export([t1/0, t2/0]).
-import(lists, [map/2]).

t1() -> map(fun(X) -> 2 * X end, [1,2,3,4,5]).

t2() -> map(fun double/1, [1,2,3,4,5]).

double(X) -> X * 2.

The fun F can be evaluated with the following syntax:

F(Arg1, Arg2, ..., Argn)

To check whether a term is a fun, use the test is_function/1 in a guard.


f(F, Args) when is_function(F) ->
   apply(F, Args);
f(N, _) when is_integer(N) ->

Funs are a distinct type. The BIFs erlang:fun_info/1,2 can be used to retrieve information about a fun, and the BIF erlang:fun_to_list/1 returns a textual representation of a fun. The check_process_code/2 BIF returns true if the process contains funs that depend on the old version of a module.

2.4  Variable Bindings Within a Fun

The scope rules for variables that occur in funs are as follows:

  • All variables that occur in the head of a fun are assumed to be "fresh" variables.
  • Variables that are defined before the fun, and that occur in function calls or guard tests within the fun, have the values they had outside the fun.
  • Variables cannot be exported from a fun.

The following examples illustrate these rules:

print_list(File, List) ->
    {ok, Stream} = file:open(File, write),
    foreach(fun(X) -> io:format(Stream,"~p~n",[X]) end, List),

Here, the variable X, defined in the head of the fun, is a new variable. The variable Stream, which is used within the fun, gets its value from the file:open line.

As any variable that occurs in the head of a fun is considered a new variable, it is equally valid to write as follows:

print_list(File, List) ->
    {ok, Stream} = file:open(File, write),
    foreach(fun(File) -> 
            end, List),

Here, File is used as the new variable instead of X. This is not so wise because code in the fun body cannot refer to the variable File, which is defined outside of the fun. Compiling this example gives the following diagnostic:

./FileName.erl:Line: Warning: variable 'File' 
      shadowed in 'fun'

This indicates that the variable File, which is defined inside the fun, collides with the variable File, which is defined outside the fun.

The rules for importing variables into a fun has the consequence that certain pattern matching operations must be moved into guard expressions and cannot be written in the head of the fun. For example, you might write the following code if you intend the first clause of F to be evaluated when the value of its argument is Y:

f(...) ->
    Y = ...
    map(fun(X) when X == Y ->
           (_) ->
        end, ...)

instead of writng the following code:

f(...) ->
    Y = ...
    map(fun(Y) ->
           (_) ->
        end, ...)

2.5  Funs and Module Lists

The following examples show a dialogue with the Erlang shell. All the higher order functions discussed are exported from the module lists.


map takes a function of one argument and a list of terms:

map(F, [H|T]) -> [F(H)|map(F, T)];
map(F, [])    -> [].

It returns the list obtained by applying the function to every argument in the list.

When a new fun is defined in the shell, the value of the fun is printed as Fun#<erl_eval>:

> Double = fun(X) -> 2 * X end.
> lists:map(Double, [1,2,3,4,5]).


any takes a predicate P of one argument and a list of terms:

any(Pred, [H|T]) ->
    case Pred(H) of
        true  ->  true;
        false ->  any(Pred, T)
any(Pred, []) ->

A predicate is a function that returns true or false. any is true if there is a term X in the list such that P(X) is true.

A predicate Big(X) is defined, which is true if its argument is greater that 10:

> Big =  fun(X) -> if X > 10 -> true; true -> false end end.
> lists:any(Big, [1,2,3,4]).
> lists:any(Big, [1,2,3,12,5]).


all has the same arguments as any:

all(Pred, [H|T]) ->
    case Pred(H) of
        true  ->  all(Pred, T);
        false ->  false
all(Pred, []) ->

It is true if the predicate applied to all elements in the list is true.

> lists:all(Big, [1,2,3,4,12,6]).   
> lists:all(Big, [12,13,14,15]).       


foreach takes a function of one argument and a list of terms:

foreach(F, [H|T]) ->
    foreach(F, T);
foreach(F, []) ->

The function is applied to each argument in the list. foreach returns ok. It is only used for its side-effect:

> lists:foreach(fun(X) -> io:format("~w~n",[X]) end, [1,2,3,4]). 


foldl takes a function of two arguments, an accumulator and a list:

foldl(F, Accu, [Hd|Tail]) ->
    foldl(F, F(Hd, Accu), Tail);
foldl(F, Accu, []) -> Accu.

The function is called with two arguments. The first argument is the successive elements in the list. The second argument is the accumulator. The function must return a new accumulator, which is used the next time the function is called.

If you have a list of lists L = ["I","like","Erlang"], then you can sum the lengths of all the strings in L as follows:

> L = ["I","like","Erlang"].
10> lists:foldl(fun(X, Sum) -> length(X) + Sum end, 0, L).                    

foldl works like a while loop in an imperative language:

L =  ["I","like","Erlang"],
Sum = 0,
while( L != []){
    Sum += length(head(L)),
    L = tail(L)


mapfoldl simultaneously maps and folds over a list:

mapfoldl(F, Accu0, [Hd|Tail]) ->
    {R,Accu1} = F(Hd, Accu0),
    {Rs,Accu2} = mapfoldl(F, Accu1, Tail),
    {[R|Rs], Accu2};
mapfoldl(F, Accu, []) -> {[], Accu}.

The following example shows how to change all letters in L to upper case and then count them.

First the change to upper case:

> Upcase =  fun(X) when $a =< X,  X =< $z -> X + $A - $a;
(X) -> X 
> Upcase_word = 
fun(X) -> 
lists:map(Upcase, X) 
> Upcase_word("Erlang").
> lists:map(Upcase_word, L).

Now, the fold and the map can be done at the same time:

> lists:mapfoldl(fun(Word, Sum) ->
{Upcase_word(Word), Sum + length(Word)}
end, 0, L).


filter takes a predicate of one argument and a list and returns all elements in the list that satisfy the predicate:

filter(F, [H|T]) ->
    case F(H) of
        true  -> [H|filter(F, T)];
        false -> filter(F, T)
filter(F, []) -> [].
> lists:filter(Big, [500,12,2,45,6,7]).

Combining maps and filters enables writing of very succinct code. For example, to define a set difference function diff(L1, L2) to be the difference between the lists L1 and L2, the code can be written as follows:

diff(L1, L2) -> 
    filter(fun(X) -> not member(X, L2) end, L1).

This gives the list of all elements in L1 that are not contained in L2.

The AND intersection of the list L1 and L2 is also easily defined:

intersection(L1,L2) -> filter(fun(X) -> member(X,L1) end, L2).


takewhile(P, L) takes elements X from a list L as long as the predicate P(X) is true:

takewhile(Pred, [H|T]) ->
    case Pred(H) of
        true  -> [H|takewhile(Pred, T)];
        false -> []
takewhile(Pred, []) ->
> lists:takewhile(Big, [200,500,45,5,3,45,6]).  


dropwhile is the complement of takewhile:

dropwhile(Pred, [H|T]) ->
    case Pred(H) of
        true  -> dropwhile(Pred, T);
        false -> [H|T]
dropwhile(Pred, []) ->
> lists:dropwhile(Big, [200,500,45,5,3,45,6]).


splitwith(P, L) splits the list L into the two sublists {L1, L2}, where L = takewhile(P, L) and L2 = dropwhile(P, L):

splitwith(Pred, L) ->
    splitwith(Pred, L, []).

splitwith(Pred, [H|T], L) ->
    case Pred(H) of 
        true  -> splitwith(Pred, T, [H|L]);
        false -> {reverse(L), [H|T]}
splitwith(Pred, [], L) ->
    {reverse(L), []}.
> lists:splitwith(Big, [200,500,45,5,3,45,6]).

2.6  Funs Returning Funs

So far, only functions that take funs as arguments have been described. More powerful functions, that themselves return funs, can also be written. The following examples illustrate these type of functions.

Simple Higher Order Functions

Adder(X) is a function that given X, returns a new function G such that G(K) returns K + X:

> Adder = fun(X) -> fun(Y) -> X + Y end end.
> Add6 = Adder(6).
> Add6(10).

Infinite Lists

The idea is to write something like:

ints_from(N) ->
    fun() ->

Then proceed as follows:

> XX = lazy:ints_from(1).
> XX().
> hd(XX()).
> Y = tl(XX()).
> hd(Y()).

And so on. This is an example of "lazy embedding".


The following examples show parsers of the following type:

Parser(Toks) -> {ok, Tree, Toks1} | fail

Toks is the list of tokens to be parsed. A successful parse returns {ok, Tree, Toks1}.

  • Tree is a parse tree.
  • Toks1 is a tail of Tree that contains symbols encountered after the structure that was correctly parsed.

An unsuccessful parse returns fail.

The following example illustrates a simple, functional parser that parses the grammar:

(a | b) & (c | d)

The following code defines a function pconst(X) in the module funparse, which returns a fun that parses a list of tokens:

pconst(X) ->
    fun (T) ->
       case T of
           [X|T1] -> {ok, {const, X}, T1};
           _      -> fail

This function can be used as follows:

> P1 = funparse:pconst(a).
> P1([a,b,c]).
> P1([x,y,z]).     

Next, the two higher order functions pand and por are defined. They combine primitive parsers to produce more complex parsers.

First pand:

pand(P1, P2) ->
    fun (T) ->
        case P1(T) of
            {ok, R1, T1} ->
                case P2(T1) of
                    {ok, R2, T2} ->
                        {ok, {'and', R1, R2}};
                    fail ->
            fail ->

Given a parser P1 for grammar G1, and a parser P2 for grammar G2, pand(P1, P2) returns a parser for the grammar, which consists of sequences of tokens that satisfy G1, followed by sequences of tokens that satisfy G2.

por(P1, P2) returns a parser for the language described by the grammar G1 or G2:

por(P1, P2) ->
    fun (T) ->
        case P1(T) of
            {ok, R, T1} -> 
                {ok, {'or',1,R}, T1};
            fail -> 
                case P2(T) of
                    {ok, R1, T1} ->
                        {ok, {'or',2,R1}, T1};
                    fail ->

The original problem was to parse the grammar (a | b) & (c | d). The following code addresses this problem:

grammar() ->
         por(pconst(a), pconst(b)),
         por(pconst(c), pconst(d))).

The following code adds a parser interface to the grammar:

parse(List) ->

The parser can be tested as follows:

> funparse:parse([a,c]).
> funparse:parse([a,d]). 
> funparse:parse([b,c]).   
> funparse:parse([b,d]). 
> funparse:parse([a,b]).