path: root/lib/stdlib/doc/src/io_protocol.xml

<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE chapter SYSTEM "chapter.dtd">

<chapter>
  <header>
    <copyright>
      <year>1999</year>
      <year>2013</year>
      <holder>Ericsson AB. All Rights Reserved.</holder>
    </copyright>
    <legalnotice>
      Licensed under the Apache License, Version 2.0 (the "License");
      you may not use this file except in compliance with the License.
      You may obtain a copy of the License at
 
          http://www.apache.org/licenses/LICENSE-2.0

      Unless required by applicable law or agreed to in writing, software
      distributed under the License is distributed on an "AS IS" BASIS,
      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
      See the License for the specific language governing permissions and
      limitations under the License.
    
    </legalnotice>

    <title>The Erlang I/O-protocol</title>
    <prepared>Patrik Nyblom</prepared>
    <responsible></responsible>
    <docno></docno>
    <approved></approved>
    <checked></checked>
    <date>2009-02-25</date>
    <rev>PA1</rev>
    <file>io_protocol.xml</file>
  </header>


<p>The I/O-protocol in Erlang specifies a way for a client to communicate
with an I/O server and vice versa. The I/O server is a process that handles
the requests and performs the requested task on e.g. an IO device. The
client is any Erlang process wishing to read or write data from/to the
IO device.</p> 

<p>The common I/O-protocol has been present in OTP since the
beginning, but has been fairly undocumented and has also somewhat
evolved over the years. In an addendum to Robert Virdings rationale
the original I/O-protocol is described. This document describes the
current I/O-protocol.</p>

<p>The original I/O-protocol was simple and flexible. Demands for spacial
and execution time efficiency has triggered extensions to the protocol
over the years, making the protocol larger and somewhat less easy to
implement than the original. It can certainly be argued that the
current protocol is too complex, but this text describes how it looks
today, not how it should have looked.</p>

<p>The basic ideas from the original protocol still hold. The I/O server
and client communicate with one single, rather simplistic protocol and
no server state is ever present in the client. Any I/O server can be
used together with any client code and client code need not be aware
of the actual IO device the I/O server communicates with.</p>

<section>
<title>Protocol Basics</title>

<p>As described in Robert's paper, I/O servers and clients communicate using
<c>io_request</c>/<c>io_reply</c> tuples as follows:</p>

<p><em>{io_request, From, ReplyAs, Request}</em><br/>
<em>{io_reply, ReplyAs, Reply}</em></p>

<p>The client sends an <c>io_request</c> tuple to the I/O server and
the server eventually sends a corresponding <c>io_reply</c> tuple.</p>

<list type="bulleted">
<item><c>From</c> is the <c>pid()</c> of the client, the process which
the I/O server sends the IO reply to.</item>

<item><c>ReplyAs</c> can be any datum and is returned in the corresponding
<c>io_reply</c>. The <seealso marker="stdlib:io">io</seealso> module monitors
the I/O server, and uses the monitor reference as the <c>ReplyAs</c> datum.
A more complicated client
could have several outstanding I/O requests to the same I/O server and
would then use different references (or something else) to differentiate among
the incoming IO replies. The <c>ReplyAs</c> element should be considered
opaque by the I/O server. Note that the <c>pid()</c> of the I/O server is not
explicitly present in the <c>io_reply</c> tuple. The reply can be sent from any
process, not necessarily the actual I/O server. The <c>ReplyAs</c> element is
the only thing that connects one I/O request with an I/O-reply.</item>

<item><c>Request</c> and <c>Reply</c> are described below.</item>
</list>

<p>When an I/O server receives an <c>io_request</c> tuple, it acts upon the actual
<c>Request</c> part and eventually sends an <c>io_reply</c> tuple with the corresponding
<c>Reply</c> part.</p> 
</section>
<section>
<title>Output Requests</title>

<p>To output characters on an IO device, the following <c>Request</c>s exist:</p>

<p>
<em>{put_chars, Encoding, Characters}</em><br/>
<em>{put_chars, Encoding, Module, Function, Args}</em>
</p>
<list type="bulleted">
<item><c>Encoding</c> is either <c>unicode</c> or <c>latin1</c>, meaning that the
  characters are (in case of binaries) encoded as either UTF-8 or
  ISO-latin-1 (pure bytes). A well behaved I/O server should also
  return error if list elements contain integers > 255 when
  <c>Encoding</c> is set to <c>latin1</c>. Note that this does not in any way tell
  how characters should be put on the actual IO device or how the
  I/O server should handle them. Different I/O servers may handle the
  characters however they want, this simply tells the I/O server which
  format the data is expected to have. In the <c>Module</c>/<c>Function</c>/<c>Args</c>
  case, <c>Encoding</c> tells which format the designated function
  produces. Note that byte-oriented data is simplest sent using the ISO-latin-1
  encoding.</item>

<item>Characters are the data to be put on the IO device. If <c>Encoding</c> is
  <c>latin1</c>, this is an <c>iolist()</c>. If <c>Encoding</c> is <c>unicode</c>, this is an
  Erlang standard mixed Unicode list (one integer in a list per
  character, characters in binaries represented as UTF-8).</item>

<item><c>Module</c>, <c>Function</c>, and <c>Args</c> denote a function which will be called to
  produce the data (like <c>io_lib:format/2</c>). <c>Args</c> is a list of arguments
  to the function. The function should produce data in the given
  <c>Encoding</c>. The I/O server should call the function as
  <c>apply(Mod, Func, Args)</c> and will put the returned data on the IO device as if it was sent
  in a <c>{put_chars, Encoding, Characters}</c> request. If the function
  returns anything else than a binary or list or throws an exception,
  an error should be sent back to the client.</item>
</list>

<p>The I/O server replies to the client with an <c>io_reply</c> tuple where the <c>Reply</c>
element is one of:</p>
<p>
<em>ok</em><br/>
<em>{error, Error}</em>
</p>

<list type="bulleted">
<item><c>Error</c> describes the error to the client, which may do whatever
  it wants with it. The Erlang <seealso marker="stdlib:io">io</seealso>
  module typically returns it as is.</item>
</list>

<p>For backward compatibility the following <c>Request</c>s should also be
handled by an I/O server (these requests should not be present after
R15B of OTP):</p>
<p>
<em>{put_chars, Characters}</em><br/>
<em>{put_chars, Module, Function, Args}</em>
</p>

<p>These should behave as <c>{put_chars, latin1, Characters}</c> and
<c>{put_chars, latin1, Module, Function, Args}</c> respectively. </p>
</section>
<section>
<title>Input Requests</title>

<p>To read characters from an IO device, the following <c>Request</c>s exist:</p>

<p><em>{get_until, Encoding, Prompt, Module, Function, ExtraArgs}</em></p>

<list type="bulleted">
<item><c>Encoding</c> denotes how data is to be sent back to the client and
  what data is sent to the function denoted by
  <c>Module</c>/<c>Function</c>/<c>ExtraArgs</c>. If the function supplied returns data as a
  list, the data is converted to this encoding. If however the
  function supplied returns data in some other format, no conversion
  can be done and it is up to the client supplied function to return
  data in a proper way. If <c>Encoding</c> is <c>latin1</c>, lists of integers
  0..255 or binaries containing plain bytes are sent back to the
  client when possible; if <c>Encoding</c> is <c>unicode</c>, lists with integers in
  the whole Unicode range or binaries encoded in UTF-8 are sent to the
  client. The user supplied function will always see lists of integers, never
  binaries, but the list may contain numbers > 255 if the <c>Encoding</c> is
  <c>unicode</c>.</item>

<item><c>Prompt</c> is a list of characters (not mixed, no binaries) or an atom
  to be output as a prompt for input on the IO device. <c>Prompt</c> is
  often ignored by the I/O server and if set to <c>''</c> it should always
  be ignored (and result in nothing being written to the IO device).</item>

<item><p><c>Module</c>, <c>Function</c>, and <c>ExtraArgs</c> denote a function and arguments to
  determine when enough data is written. The function should take two
  additional arguments, the last state, and a list of characters. The
  function should return one of:</p>
<p>
<em>{done, Result, RestChars}</em><br/>
<em>{more, Continuation}</em>
</p>
  <p>The <c>Result</c> can be any Erlang term, but if it is a <c>list()</c>, the
  I/O server may convert it to a <c>binary()</c> of appropriate format before
  returning it to the client, if the I/O server is set in binary mode (see
  below).</p>

  <p>The function will be called with the data the I/O server finds on
  its IO device, returning <c>{done, Result, RestChars}</c> when enough data is
  read (in which case <c>Result</c> is sent to the client and <c>RestChars</c> is
  kept in the I/O server as a buffer for subsequent input) or
  <c>{more, Continuation}</c>, indicating that more characters are needed to
  complete the request. The <c>Continuation</c> will be sent as the state in
  subsequent calls to the function when more characters are
  available. When no more characters are available, the function
  shall return <c>{done, eof, Rest}</c>.
  The initial state is the empty list and the data when an
  end of file is reached on the IO device is the atom <c>eof</c>. An emulation
  of the <c>get_line</c> request could be (inefficiently) implemented using
  the following functions:</p>
<code>
-module(demo).
-export([until_newline/3, get_line/1]).

until_newline(_ThisFar,eof,_MyStopCharacter) -&gt;
    {done,eof,[]};
until_newline(ThisFar,CharList,MyStopCharacter) -&gt;
    case
        lists:splitwith(fun(X) -&gt; X =/= MyStopCharacter end,  CharList)
    of
	{L,[]} -&gt;
            {more,ThisFar++L};
	{L2,[MyStopCharacter|Rest]} -&gt;
	    {done,ThisFar++L2++[MyStopCharacter],Rest}
    end.

get_line(IoServer) -&gt;
    IoServer ! {io_request,
                self(),
                IoServer,
                {get_until, unicode, '', ?MODULE, until_newline, [$\n]}},
    receive
        {io_reply, IoServer, Data} -&gt;
	    Data
    end.
</code>
 <p>Note especially that the last element in the <c>Request</c> tuple (<c>[$\n]</c>)
  is appended to the argument list when the function is called. The
  function should be called like 
  <c>apply(Module, Function, [ State, Data | ExtraArgs ])</c> by the I/O server</p>
</item>
</list>

<p>A fixed number of characters is requested using this <c>Request</c>:</p>
<p>
<em>{get_chars, Encoding, Prompt, N}</em>
</p>

<list type="bulleted">
<item><c>Encoding</c> and <c>Prompt</c> as for <c>get_until</c>.</item>

<item><c>N</c> is the number of characters to be read from the IO device.</item>
</list>

<p>A single line (like in the example above) is requested with this <c>Request</c>:</p>
<p>
<em>{get_line, Encoding, Prompt}</em>
</p>

<list type="bulleted">
<item><c>Encoding</c> and <c>Prompt</c> as above.</item>
</list>

<p>Obviously, the <c>get_chars</c> and <c>get_line</c> could be implemented with the
<c>get_until</c> request (and indeed they were originally), but demands for
efficiency has made these additions necessary.</p>

<p>The I/O server replies to the client with an <c>io_reply</c> tuple where the <c>Reply</c>
element is one of:</p>
<p>
<em>Data</em><br/>
<em>eof</em><br/>
<em>{error, Error}</em>
</p>

<list type="bulleted">
<item><c>Data</c> is the characters read, in either list or binary form
  (depending on the I/O server mode, see below).</item>
<item><c>Error</c> describes the error to the client, which may do whatever it
  wants with it. The Erlang <seealso marker="stdlib:io">io</seealso>
  module typically returns it as is.</item>
<item><c>eof</c> is returned when input end is reached and no more data is
available to the client process.</item>
</list>

<p>For backward compatibility the following <c>Request</c>s should also be
handled by an I/O server (these reqeusts should not be present after
R15B of OTP):</p>

<p>
<em>{get_until, Prompt, Module, Function, ExtraArgs}</em><br/>
<em>{get_chars, Prompt, N}</em><br/>
<em>{get_line, Prompt}</em><br/>
</p>

<p>These should behave as <c>{get_until, latin1, Prompt, Module, Function,
ExtraArgs}</c>, <c>{get_chars, latin1, Prompt, N}</c> and <c>{get_line, latin1,
Prompt}</c> respectively.</p>
</section>
<section>
<title>I/O-server Modes</title>

<p>Demands for efficiency when reading data from an I/O server has not
only lead to the addition of the <c>get_line</c> and <c>get_chars</c> requests, but
has also added the concept of I/O server options. No options are
mandatory to implement, but all I/O servers in the Erlang standard
libraries honor the <c>binary</c> option, which allows the <c>Data</c> element of the
<c>io_reply</c> tuple to be a binary instead of a list <em>when possible</em>. 
If the data is sent as a binary, Unicode data will be sent in the 
standard Erlang Unicode
format, i.e. UTF-8 (note that the function of the <c>get_until</c> request still gets
list data regardless of the I/O server mode).</p>

<p>Note that i.e. the <c>get_until</c> request allows for a function with the data specified as always being a list. Also the return value data from such a function can be of any type (as is indeed the case when an <c>io:fread</c> request is sent to an I/O server). The client has to be prepared for data received as answers to those requests to be in a variety of forms, but the I/O server should convert the results to binaries whenever possible (i.e. when the function supplied to <c>get_until</c> actually returns a list). The example shown later in this text does just that.</p>

<p>An I/O-server in binary mode will affect the data sent to the client,
so that it has to be able to handle binary data. For convenience, it
is possible to set and retrieve the modes of an I/O server using the
following I/O requests:</p>

<p>
<em>{setopts, Opts}</em>
</p>


<list type="bulleted">
<item><c>Opts</c> is a list of options in the format recognized by <seealso marker="stdlib:proplists">proplists</seealso> (and
  of course by the I/O server itself).</item>
</list>
<p>As an example, the I/O server for the interactive shell (in <c>group.erl</c>)
understands the following options:</p>
<p>
<em>{binary, boolean()}</em> (or <em>binary</em>/<em>list</em>)<br/>
<em>{echo, boolean()}</em><br/>
<em>{expand_fun, fun()}</em><br/>
<em>{encoding, unicode/latin1}</em> (or <em>unicode</em>/<em>latin1</em>)
</p>

<p>- of which the <c>binary</c> and <c>encoding</c> options are common for all
I/O servers in OTP, while <c>echo</c> and <c>expand</c> are valid only for this
I/O server. It is worth noting that the <c>unicode</c> option notifies how
characters are actually put on the physical IO device, i.e. if the
terminal per se is Unicode aware, it does not affect how characters
are sent in the I/O-protocol, where each request contains encoding
information for the provided or returned data.</p>

<p>The I/O server should send one of the following as <c>Reply</c>:</p>
<p>
<em>ok</em><br/>
<em>{error, Error}</em>
</p>

<p>An error (preferably <c>enotsup</c>) is to be expected if the option is
not supported by the I/O server (like if an <c>echo</c> option is sent in a
<c>setopts</c> request to a plain file).</p>

<p>To retrieve options, this request is used:</p>
<p>
<em>getopts</em>
</p>

<p>The <c>getopts</c> request asks for a complete list of all options
supported by the I/O server as well as their current values.</p>

<p>The I/O server replies:</p>
<p>
<em>OptList</em><br/>
<em>{error, Error}</em>
</p>

<list type="bulleted">
<item><c>OptList</c> is a list of tuples <c>{Option, Value}</c> where <c>Option</c> is always
  an atom.</item>
</list>
</section>
<section>
<title>Multiple I/O Requests</title>

<p>The <c>Request</c> element can in itself contain several <c>Request</c>s by using
the following format:</p>
<p>
<em>{requests, Requests}</em>
</p>
<list type="bulleted">
<item><c>Requests</c> is a list of valid <c>io_request</c> tuples for the protocol, they
  shall be executed in the order in which they appear in the list and
  the execution should continue until one of the requests result in an
  error or the list is consumed. The result of the last request is
  sent back to the client.</item>
</list>

<p>The I/O server can for a list of requests send any of the valid results in
the reply:</p>

<p>
<em>ok</em><br/>
<em>{ok, Data}</em><br/>
<em>{ok, Options}</em><br/>
<em>{error, Error}</em>
</p>
<p>- depending on the actual requests in the list.</p>
</section>
<section>
<title>Optional I/O Requests</title>

<p>The following I/O request is optional to implement and a client
should be prepared for an error return:</p>
<p>
<em>{get_geometry, Geometry}</em>
</p>
<list type="bulleted">
<item><c>Geometry</c> is either the atom <c>rows</c> or the atom <c>columns</c>.</item>
</list>
<p>The I/O server should send the <c>Reply</c> as:</p>
<p>
<em>{ok, N}</em><br/>
<em>{error, Error}</em>
</p>

<list type="bulleted">
<item><c>N</c> is the number of character rows or columns the IO device has, if
  applicable to the IO device the I/O server handles, otherwise <c>{error,
  enotsup}</c> is a good answer.</item>
</list>
</section>
<section>
<title>Unimplemented Request Types</title>

<p>If an I/O server encounters a request it does not recognize (i.e. the
<c>io_request</c> tuple is in the expected format, but the actual <c>Request</c> is
unknown), the I/O server should send a valid reply with the error tuple:</p>
<p>
<em>{error, request}</em>
</p>

<p>This makes it possible to extend the protocol with optional requests
and for the clients to be somewhat backwards compatible.</p>
</section>
<section>
<title>An Annotated and Working Example I/O Server</title>

<p>An I/O server is any process capable of handling the I/O protocol. There is
no generic I/O server behavior, but could well be. The framework is
simple enough, a process handling incoming requests, usually both
I/O-requests and other IO device-specific requests (for i.e. positioning,
closing etc.).</p>

<p>Our example I/O server stores characters in an ETS table, making up a
fairly crude ram-file (it is probably not useful, but working).</p>

<p>The module begins with the usual directives, a function to start the
I/O server and a main loop handling the requests:</p> 

<code>
-module(ets_io_server).

-export([start_link/0, init/0, loop/1, until_newline/3, until_enough/3]).

-define(CHARS_PER_REC, 10).

-record(state, {
	  table,
	  position, % absolute
	  mode % binary | list
	 }).

start_link() -&gt;
    spawn_link(?MODULE,init,[]).

init() -&gt;
    Table = ets:new(noname,[ordered_set]),
    ?MODULE:loop(#state{table = Table, position = 0, mode=list}).

loop(State) -&gt;
    receive
	{io_request, From, ReplyAs, Request} -&gt;
	    case request(Request,State) of
		{Tag, Reply, NewState} when Tag =:= ok; Tag =:= error -&gt;
		    reply(From, ReplyAs, Reply),
		    ?MODULE:loop(NewState);
		{stop, Reply, _NewState} -&gt;
		    reply(From, ReplyAs, Reply),
		    exit(Reply)
	    end;
	%% Private message
	{From, rewind} -&gt;
	    From ! {self(), ok},
	    ?MODULE:loop(State#state{position = 0});
	_Unknown -&gt;
	    ?MODULE:loop(State)
    end.
</code>

<p>The main loop receives messages from the client (which might be using
the <seealso marker="stdlib:io">io</seealso> module to send requests).
For each request the function
<c>request/2</c> is called and a reply is eventually sent using the <c>reply/3</c>
function.</p>

<p>The &quot;private&quot; message <c>{From, rewind}</c> results in the
current position in the pseudo-file to be reset to 0 (the beginning of
the &quot;file&quot;). This is a typical example of IO device-specific
messages not being part of the I/O-protocol. It is usually a bad idea
to embed such private messages in <c>io_request</c> tuples, as that might be
confusing to the reader.</p>

<p>Let us look at the reply function first...</p>

<code>

reply(From, ReplyAs, Reply) -&gt;
    From ! {io_reply, ReplyAs, Reply}.

</code>

<p>Simple enough, it sends the <c>io_reply</c> tuple back to the client,
providing the <c>ReplyAs</c> element received in the request along with the
result of the request, as described above.</p>

<p>Now look at the different requests we need to handle. First the
requests for writing characters:</p>

<code>
request({put_chars, Encoding, Chars}, State) -&gt;
    put_chars(unicode:characters_to_list(Chars,Encoding),State);
request({put_chars, Encoding, Module, Function, Args}, State) -&gt;
    try
	request({put_chars, Encoding, apply(Module, Function, Args)}, State)
    catch
	_:_ -&gt;
	    {error, {error,Function}, State}
    end;
</code>

<p>The <c>Encoding</c> tells us how the characters in the request are
represented. We want to store the characters as lists in the
ETS table, so we convert them to lists using the
<seealso marker="stdlib:unicode#characters_to_list/2"><c>unicode:characters_to_list/2</c></seealso> function. The conversion function
conveniently accepts the encoding types <c>unicode</c> or <c>latin1</c>, so we can
use <c>Encoding</c> directly.</p>

<p>When <c>Module</c>, <c>Function</c> and <c>Arguments</c> are provided, we simply apply it
and do the same thing with the result as if the data was provided
directly.</p>

<p>Let us handle the requests for retrieving data too:</p>

<code>
request({get_until, Encoding, _Prompt, M, F, As}, State) -&gt;
    get_until(Encoding, M, F, As, State);
request({get_chars, Encoding, _Prompt, N}, State) -&gt;
    %% To simplify the code, get_chars is implemented using get_until
    get_until(Encoding, ?MODULE, until_enough, [N], State);
request({get_line, Encoding, _Prompt}, State) -&gt;
    %% To simplify the code, get_line is implemented using get_until
    get_until(Encoding, ?MODULE, until_newline, [$\n], State);
</code>

<p>Here we have cheated a little by more or less only implementing
<c>get_until</c> and using internal helpers to implement <c>get_chars</c> and
<c>get_line</c>. In production code, this might be too inefficient, but that
of course depends on the frequency of the different requests. Before
we start actually implementing the functions <c>put_chars/2</c> and
<c>get_until/5</c>, let us look into the few remaining requests:</p>

<code>
request({get_geometry,_}, State) -&gt;
    {error, {error,enotsup}, State};
request({setopts, Opts}, State) -&gt;
    setopts(Opts, State);
request(getopts, State) -&gt;
    getopts(State);
request({requests, Reqs}, State) -&gt;
     multi_request(Reqs, {ok, ok, State});
</code>

<p>The <c>get_geometry</c> request has no meaning for this I/O server, so the
reply will be <c>{error, enotsup}</c>. The only option we handle is the
<c>binary</c>/<c>list</c> option, which is done in separate functions.</p>

<p>The multi-request tag (<c>requests</c>) is handled in a separate loop
function applying the requests in the list one after another,
returning the last result.</p>

<p>What is left is to handle backward compatibility and the <seealso marker="kernel:file">file</seealso> module
(which uses the old requests until backward compatibility with pre-R13
nodes is no longer needed). Note that the I/O server will not work with
a simple <c>file:write/2</c> if these are not added:</p>

<code>
request({put_chars,Chars}, State) -&gt;
    request({put_chars,latin1,Chars}, State);
request({put_chars,M,F,As}, State) -&gt;
    request({put_chars,latin1,M,F,As}, State);
request({get_chars,Prompt,N}, State) -&gt;
    request({get_chars,latin1,Prompt,N}, State);
request({get_line,Prompt}, State) -&gt;
    request({get_line,latin1,Prompt}, State);
request({get_until, Prompt,M,F,As}, State) -&gt;
    request({get_until,latin1,Prompt,M,F,As}, State);
</code>

<p>OK, what is left now is to return <c>{error, request}</c> if the request is
not recognized:</p>

<code>
request(_Other, State) -&gt;
    {error, {error, request}, State}.
</code>

<p>Let us move further and actually handle the different requests, first
the fairly generic multi-request type:</p>

<code>
multi_request([R|Rs], {ok, _Res, State}) -&gt;
    multi_request(Rs, request(R, State));
multi_request([_|_], Error) -&gt;
    Error;
multi_request([], Result) -&gt;
    Result.
</code>

<p>We loop through the requests one at the time, stopping when we either
encounter an error or the list is exhausted. The last return value is
sent back to the client (it is first returned to the main loop and then
sent back by the function <c>io_reply</c>).</p>

<p>The <c>getopts</c> and <c>setopts</c> requests are also simple to handle, we just
change or read our state record:</p>

<code>
setopts(Opts0,State) -&gt;
    Opts = proplists:unfold(
	     proplists:substitute_negations(
	       [{list,binary}], 
	       Opts0)),
    case check_valid_opts(Opts) of
	true -&gt;
	        case proplists:get_value(binary, Opts) of
		    true -&gt;
			{ok,ok,State#state{mode=binary}};
		    false -&gt;
			{ok,ok,State#state{mode=binary}};
		    _ -&gt;
			{ok,ok,State}
		end;
	false -&gt;
	    {error,{error,enotsup},State}
    end.
check_valid_opts([]) -&gt;
    true;
check_valid_opts([{binary,Bool}|T]) when is_boolean(Bool) -&gt;
    check_valid_opts(T);
check_valid_opts(_) -&gt;
    false.

getopts(#state{mode=M} = S) -&gt;
    {ok,[{binary, case M of
		      binary -&gt;
			  true;
		      _ -&gt;
			  false
		  end}],S}.
</code>

<p>As a convention, all I/O servers handle both <c>{setopts, [binary]}</c>,
<c>{setopts, [list]}</c> and <c>{setopts,[{binary, boolean()}]}</c>, hence the trick
with <c>proplists:substitute_negations/2</c> and <c>proplists:unfold/1</c>. If
invalid options are sent to us, we send <c>{error, enotsup}</c> back to the
client.</p>

<p>The <c>getopts</c> request should return a list of <c>{Option, Value}</c> tuples,
which has the twofold function of providing both the current values
and the available options of this I/O server. We have only one option,
and hence return that.</p>

<p>So far our I/O server has been fairly generic (except for the <c>rewind</c>
request handled in the main loop and the creation of an ETS table). 
Most I/O servers contain code similar to the one above.</p>

<p>To make the example runnable, we now start implementing the actual
reading and writing of the data to/from the ETS table. First the
<c>put_chars/3</c> function:</p>

<code>
put_chars(Chars, #state{table = T, position = P} = State) -&gt;
    R = P div ?CHARS_PER_REC,
    C = P rem ?CHARS_PER_REC,
    [ apply_update(T,U) || U &lt;- split_data(Chars, R, C) ],
    {ok, ok, State#state{position = (P + length(Chars))}}.
</code>

<p>We already have the data as (Unicode) lists and therefore just split
the list in runs of a predefined size and put each run in the
table at the current position (and forward). The functions
<c>split_data/3</c> and <c>apply_update/2</c> are implemented below.</p>

<p>Now we want to read data from the table. The <c>get_until/5</c> function reads
data and applies the function until it says it is done. The result is
sent back to the client:</p>

<code>
get_until(Encoding, Mod, Func, As, 
	  #state{position = P, mode = M, table = T} = State) -&gt;
    case get_loop(Mod,Func,As,T,P,[]) of
	{done,Data,_,NewP} when is_binary(Data); is_list(Data) -&gt;
	    if
		M =:= binary -&gt; 
		    {ok, 
		     unicode:characters_to_binary(Data, unicode, Encoding),
		     State#state{position = NewP}};
		true -&gt;
		    case check(Encoding, 
		               unicode:characters_to_list(Data, unicode))
                    of
			{error, _} = E -&gt;
			    {error, E, State};
			List -&gt;
			    {ok, List,
			     State#state{position = NewP}}
		    end
	    end;
	{done,Data,_,NewP} -&gt;
	    {ok, Data, State#state{position = NewP}};
	Error -&gt;
	    {error, Error, State}
    end.

get_loop(M,F,A,T,P,C) -&gt;
    {NewP,L} = get(P,T),
    case catch apply(M,F,[C,L|A]) of
	{done, List, Rest} -&gt;
	    {done, List, [], NewP - length(Rest)};
	{more, NewC} -&gt;
	    get_loop(M,F,A,T,NewP,NewC);
	_ -&gt;
	    {error,F}
    end.
</code>

<p>Here we also handle the mode (<c>binary</c> or <c>list</c>) that can be set by
the <c>setopts</c> request. By default, all OTP I/O servers send data back to
the client as lists, but switching mode to <c>binary</c> might increase
efficiency if the I/O server handles it in an appropriate way. The
implementation of <c>get_until</c> is hard to get efficient as the supplied
function is defined to take lists as arguments, but <c>get_chars</c> and
<c>get_line</c> can be optimized for binary mode. This example does not
optimize anything however. It is important though that the returned
data is of the right type depending on the options set, so we convert
the lists to binaries in the correct encoding <em>if possible</em>
before returning. The function supplied in the <c>get_until</c> request tuple may,
as its final result return anything, so only functions actually
returning lists can get them converted to binaries. If the request
contained the encoding tag <c>unicode</c>, the lists can contain all Unicode
codepoints and the binaries should be in UTF-8, if the encoding tag
was <c>latin1</c>, the client should only get characters in the range
0..255. The function <c>check/2</c> takes care of not returning arbitrary
Unicode codepoints in lists if the encoding was given as <c>latin1</c>. If
the function did not return a list, the check cannot be performed and
the result will be that of the supplied function untouched.</p>

<p>Now we are more or less done. We implement the utility functions below
to actually manipulate the table:</p>

<code>
check(unicode, List) -&gt;
    List;
check(latin1, List) -&gt;
    try 
	[ throw(not_unicode) || X &lt;- List,
				X > 255 ],
	List
    catch
	throw:_ -&gt;
	    {error,{cannot_convert, unicode, latin1}}
    end.
</code>

<p>The function check takes care of providing an error tuple if Unicode
codepoints above 255 is to be returned if the client requested
latin1.</p>

<p>The two functions <c>until_newline/3</c> and <c>until_enough/3</c> are helpers used 
together with the <c>get_until/5</c> function to implement <c>get_chars</c> and
<c>get_line</c> (inefficiently):</p>
	
<code>
until_newline([],eof,_MyStopCharacter) -&gt;
    {done,eof,[]};
until_newline(ThisFar,eof,_MyStopCharacter) -&gt;
    {done,ThisFar,[]};
until_newline(ThisFar,CharList,MyStopCharacter) -&gt;
    case
        lists:splitwith(fun(X) -&gt; X =/= MyStopCharacter end,  CharList)
    of
	{L,[]} -&gt;
            {more,ThisFar++L};
	{L2,[MyStopCharacter|Rest]} -&gt;
	    {done,ThisFar++L2++[MyStopCharacter],Rest}
    end.

until_enough([],eof,_N) -&gt;
    {done,eof,[]};
until_enough(ThisFar,eof,_N) -&gt;
    {done,ThisFar,[]};
until_enough(ThisFar,CharList,N) 
  when length(ThisFar) + length(CharList) >= N -&gt;
    {Res,Rest} = my_split(N,ThisFar ++ CharList, []),
    {done,Res,Rest};
until_enough(ThisFar,CharList,_N) -&gt;
    {more,ThisFar++CharList}. 
</code>

<p>As can be seen, the functions above are just the type of functions
that should be provided in <c>get_until</c> requests.</p>

<p>Now we only need to read and write the table in an appropriate way to
complete the I/O server:</p>

<code>
get(P,Tab) -&gt;
    R = P div ?CHARS_PER_REC,
    C = P rem ?CHARS_PER_REC,
    case ets:lookup(Tab,R) of
	[] -&gt;
	    {P,eof};
	[{R,List}] -&gt;
	    case my_split(C,List,[]) of
		{_,[]} -&gt;
		    {P+length(List),eof};
		{_,Data} -&gt;
		    {P+length(Data),Data}
	    end
    end.

my_split(0,Left,Acc) -&gt;
    {lists:reverse(Acc),Left};
my_split(_,[],Acc) -&gt;
    {lists:reverse(Acc),[]};
my_split(N,[H|T],Acc) -&gt;
    my_split(N-1,T,[H|Acc]).

split_data([],_,_) -&gt;
    [];
split_data(Chars, Row, Col) -&gt;
    {This,Left} = my_split(?CHARS_PER_REC - Col, Chars, []),
    [ {Row, Col, This} | split_data(Left, Row + 1, 0) ].

apply_update(Table, {Row, Col, List}) -&gt;     
    case ets:lookup(Table,Row) of
	[] -&gt;
	    ets:insert(Table,{Row, lists:duplicate(Col,0) ++ List});
	[{Row, OldData}] -&gt;
	    {Part1,_} = my_split(Col,OldData,[]),
	    {_,Part2} = my_split(Col+length(List),OldData,[]),
	    ets:insert(Table,{Row, Part1 ++ List ++ Part2})
    end.
</code>

<p>The table is read or written in chunks of <c>?CHARS_PER_REC</c>, overwriting
when necessary. The implementation is obviously not efficient, it is
just working.</p>

<p>This concludes the example. It is fully runnable and you can read or
write to the I/O server by using i.e. the <seealso marker="stdlib:io">io</seealso> module or even the <seealso marker="kernel:file">file</seealso>
module. It is as simple as that to implement a fully fledged I/O server
in Erlang.</p>
</section>
</chapter>