path: root/lib/inets/doc/src/httpc.xml

<?xml version="1.0" encoding="iso-8859-1" ?>
<!DOCTYPE erlref SYSTEM "erlref.dtd">

<erlref>
  <header>
    <copyright>
      <year>2004</year><year>2011</year>
      <holder>Ericsson AB. All Rights Reserved.</holder>
    </copyright>
    <legalnotice>
      The contents of this file are subject to the Erlang Public License,
      Version 1.1, (the "License"); you may not use this file except in
      compliance with the License. You should have received a copy of the
      Erlang Public License along with this software. If not, it can be
      retrieved online at http://www.erlang.org/.

      Software distributed under the License is distributed on an "AS IS"
      basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See
      the License for the specific language governing rights and limitations
      under the License.

    </legalnotice>

    <title>httpc</title>
    <prepared>Ingela Anderton Andin</prepared>
    <responsible></responsible>
    <docno></docno>
    <date></date>
    <rev></rev>
  </header>

  <module>httpc</module>
  <modulesummary>An HTTP/1.1 client </modulesummary>

  <description>
    <p>This module provides the API to a HTTP/1.1 compatible client according 
      to RFC 2616, caching is currently not supported.</p>
    <note>
      <p>When starting the Inets application a manager process for the
	default profile will be started. The functions in this API
	that do not explicitly use a profile will access the
	default profile. A profile keeps track of proxy options,
	cookies and other options that can be applied to more than one
	request. </p>

      <p>If the scheme
	https is used the ssl application needs to be started.</p>

      <p>Also note that pipelining will only be used if the pipeline
      timeout is set, otherwise persistent connections without
      pipelining will be used e.i. the  client always waits for
      the previous response before sending the next request.</p>
    </note>
    <p>There are some usage examples in the <seealso
    marker="http_client">Inets User's Guide.</seealso></p>
  </description>
  
  <section>
    <title>COMMON DATA TYPES </title>
    <p>Type definitions that are used more than once in
      this module:</p>
    <code type="none"><![CDATA[
boolean()    = true | false 
string()     = list of ASCII characters 
request_id() = ref() 
profile()    = atom()
path()       = string() representing a file path or directory path 
ip_address() = See inet(3)
socket_opt() = See the Options used by gen_tcp(3) and 
               ssl(3) connect(s)
    ]]></code>

  </section>
  
  <section>
    <title>HTTP DATA TYPES </title>
    <p>Type definitions that are related to HTTP:</p>
    <p>For more information about HTTP see rfc 2616</p>

    <code type="none"><![CDATA[
method()         = head | get | put | post | trace | options | delete
request()        = {url(), headers()} | 
                   {url(), headers(), content_type(), body()}
url()            = string() - Syntax according to the URI definition in rfc 2396, ex: "http://www.erlang.org"
status_line()    = {http_version(), status_code(), reason_phrase()}
http_version()   = string() ex: "HTTP/1.1"
status_code()    = integer()
reason_phrase()  = string()
content_type()   = string()
headers()        = [header()]
header()         = {field(), value()}
field()          = string() 
value()          = string() 
body()           = string() | 
                   binary() |
                   {fun(accumulator()) -> body_processing_result(), 
                    accumulator()} |
                   {chunkify, 
                    fun(accumulator()) -> body_processing_result(), 
                    accumulator()}
body_processing_result() = eof | {ok, iolist(), accumulator()}
accumulator()    = term()
filename()       = string() 
    ]]></code>

  </section>
  
  <section>
    <title>SSL DATA TYPES </title>
    <p>See <seealso marker="ssl:ssl">ssl(3)</seealso> for information
    about ssl options (<c>ssloptions()</c>). </p>
  </section>

  <section>
    <title>HTTP CLIENT SERVICE START/STOP </title>
    
    <p>A HTTP client can be configured to start when starting the inets
      application or started dynamically in runtime by calling the
      inets application API <c>inets:start(httpc, ServiceConfig)</c>, or
      <c>inets:start(httpc,  ServiceConfig, How)</c>
      see <seealso marker="inets">inets(3)</seealso>. Below follows a
      description of the available configuration options.</p>
    <taglist>
      <tag>{profile, profile()}</tag>
      <item>Name of the profile, see
	common data types below, this option is mandatory.</item>
      <tag>{data_dir, path()}</tag>
      <item>Directory where the profile
	may save persistent data, if omitted all cookies will be treated
	as session cookies.</item>
    </taglist>

    <p>The client can be stopped using <c>inets:stop(httpc, Pid)</c> or
    <c>inets:stop(httpc, Profile)</c>.</p>
    
    <marker id="request1"></marker>
  </section>
  
  <funcs>
    <func>
      <name>request(Url) -> </name>
      <name>request(Url, Profile) -> {ok, Result} | {error, Reason}</name>
      <fsummary>Sends a get HTTP-request</fsummary>
      <type>
        <v>Url = url() </v> 
	<v>Result = {status_line(), headers(), Body} | 
                    {status_code(), Body} | request_id() </v>
	<v>Body = string() | binary()</v>
	<v>Profile = profile() | pid() (when started <c>stand_alone</c>)</v>
	<v>Reason = term() </v>
      </type>
      <desc>
        <p>Equivalent to <c>httpc:request(get, {Url, []}, [], [])</c>.</p>

        <marker id="request2"></marker>
      </desc>
    </func>

    <func>
      <name>request(Method, Request, HTTPOptions, Options) -> </name>
      <name>request(Method, Request, HTTPOptions, Options, Profile) -> {ok, Result} | {ok, saved_to_file} | {error, Reason}</name>
      
      <fsummary>Sends a HTTP-request</fsummary>
      <type>
        <v>Method = method() </v>
        <v>Request = request()</v>
        <v>HTTPOptions = http_options()</v>
        <v>http_options() = [http_option()]</v>
        <v>http_option() = {timeout,         timeout()} | 
                           {connect_timeout, timeout()} | 
                           {ssl,             ssloptions()} | 
                           {essl,            ssloptions()} | 
                           {autoredirect,    boolean()} | 
                           {proxy_auth, {userstring(), passwordstring()}} | 
                           {version,         http_version()} | 
                           {relaxed,         boolean()} |
	                   {url_encode,      boolean()}</v>
        <v>timeout() = integer() >= 0 | infinity</v>
        <v>Options = options()</v>
        <v>options() = [option()]</v>
        <v>option() = {sync,                    boolean()} | 
                      {stream,                  stream_to()} | 
                      {body_format,             body_format()} | 
                      {full_result,             boolean()} | 
                      {headers_as_is,           boolean() |
                      {socket_opts,             socket_opts()} | 
                      {receiver,                receiver()},
	              {ipv6_host_with_brackets, boolean()}}</v>
        <v>stream_to() = none | self | {self, once} | filename() </v>
        <v>socket_opts() = [socket_opt()]</v>
        <v>receiver() = pid() | function()/1 | {Module, Function, Args} </v>
        <v>Module = atom() </v>
        <v>Function = atom() </v>
        <v>Args = list() </v>
	<v>body_format() = string | binary </v>
        <v>Result = {status_line(), headers(), Body} | 
                    {status_code(), Body} | request_id() </v>
        <v>Body = string() | binary()</v>
        <v>Profile = profile() | pid() (when started <c>stand_alone</c>)</v>
        <v>Reason = {connect_failed, term()} | 
                    {send_failed, term()} | term() </v>
      </type>

      <desc>
        <p>Sends a HTTP-request. The function can be both synchronous
	and asynchronous. In the later case the function will return
	<c>{ok, RequestId}</c> and later on the information will be delivered
	to the <c>receiver</c> depending on that value. </p>

        <p>Http option (<c>http_option()</c>) details: </p>
        <marker id="request2_http_options"></marker>
        <taglist>
          <tag><c><![CDATA[timeout]]></c></tag>
          <item>
            <p>Timeout time for the request. </p>
            <p>The clock starts ticking as soon as the request has been
	    sent. </p>
            <p>Time is in milliseconds. </p>
            <p>Defaults to <c>infinity</c>. </p>
	  </item>

          <tag><c><![CDATA[connect_timeout]]></c></tag>
          <item>
            <p>Connection timeout time, used during the initial request, 
	    when the client is <em>connecting</em> to the server. </p>
            <p>Time is in milliseconds. </p>
            <p>Defaults to the value of the <c>timeout</c> option. </p>
	  </item>

          <tag><c><![CDATA[ssl]]></c></tag>
          <item>
            <p>This is the default ssl config option, currently defaults to 
	    <c>essl</c>, see below. </p>
            <p>Defaults to <c>[]</c>. </p>
	  </item>

          <tag><c><![CDATA[essl]]></c></tag>
          <item>
            <p>If using the Erlang based implementation of SSL, 
	    these SSL-specific options are used. </p>
            <p>Defaults to <c>[]</c>. </p>
	  </item>

          <tag><c><![CDATA[autoredirect]]></c></tag>
          <item>
	    <p>Should the client automatically retrieve the information
	    from the new URI and return that as the result instead 
	    of a 30X-result code. </p>
	    <p>Note that for some 30X-result codes automatic redirect 
	    is not allowed. In these cases the 30X-result will always
	    be returned. </p>
	    <p>Defaults to <c>true</c>. </p>
	  </item>

          <tag><c><![CDATA[proxy_auth]]></c></tag>
          <item>
            <p>A proxy-authorization header using the provided user name and 
	    password will be added to the request. </p>
	  </item>

          <tag><c><![CDATA[version]]></c></tag>
          <item>
            <p>Can be used to make the client act as an <c>HTTP/1.0</c> or
	    <c>HTTP/0.9</c> client. By default this is an <c>HTTP/1.1</c> 
	    client. When using <c>HTTP/1.0</c> persistent connections will 
	    not be used. </p>
	    <p>Defaults to the string <c>"HTTP/1.1"</c>. </p>
	  </item>

          <tag><c><![CDATA[relaxed]]></c></tag>
          <item>
            <p>If set to <c>true</c> workarounds for known server deviations 
	    from the HTTP-standard are enabled. </p>
            <p>Defaults to <c>false</c>. </p>
	  </item>

	  <tag><c><![CDATA[url_encode]]></c></tag>
          <item>
            <p>Will apply Percent-encoding, also known as URL encoding on the 
	    URL.</p>
	    <p>Defaults to <c>false</c>. </p>
	  </item>
        </taglist>

        <p>Option (<c>option()</c>) details: </p>
        <taglist>
          <tag><c><![CDATA[sync]]></c></tag>
          <item>
            <p>Shall the request be synchronous or asynchronous. </p>
            <p>Defaults to <c>true</c>. </p>
	  </item>

          <tag><c><![CDATA[stream]]></c></tag>
          <item>
            <p>Streams the body of a 200 or 206 response to the calling 
	    process or to a file. When streaming to the calling process 
	    using the option <c>self</c> the following stream messages
	    will be sent to that process: <c>{http, {RequestId,
	    stream_start, Headers}, {http, {RequestId, stream,
	    BinBodyPart}, {http, {RequestId, stream_end, Headers}</c>. When
	    streaming to to the calling processes using the option
	    <c>{self, once}</c> the first message will have an additional
	    element e.i. <c>{http, {RequestId, stream_start, Headers, Pid}</c>,
	    this is the process id that should be used as an argument to
	    <c>http:stream_next/1</c> to trigger the next message to be sent to
	    the calling process. </p>
	    <p>Note that it is possible that chunked encoding will add
	    headers so that there are more headers in the <c>stream_end</c>
	    message than in the <c>stream_start</c>.
	    When streaming to a file and the request is asynchronous the
	    message <c>{http, {RequestId, saved_to_file}}</c> will be sent. </p>
            <p>Defaults to <c>none</c>. </p>
	  </item>

          <tag><c><![CDATA[body_format]]></c></tag>
          <item>
            <p>Defines if the body shall be delivered as a string or as a 
	    binary. This option is only valid for the synchronous 
	    request. </p>
            <p>Defaults to <c>string</c>. </p>
	  </item>

          <tag><c><![CDATA[full_result]]></c></tag>
          <item>
            <p>Should a "full result" be returned to the caller (that is,
	    the body, the headers and the entire status-line) or not 
	    (the body and the status code). </p>
            <p>Defaults to <c>true</c>. </p>
	  </item>

          <tag><c><![CDATA[header_as_is]]></c></tag>
          <item>
            <p>Shall the headers provided by the user be made 
	    lower case or be regarded as case sensitive. </p>
            <p>Note that the http standard requires them to be 
	    case insenstive. This feature should only be used if there is 
	    no other way to communicate with the server or for testing 
	    purpose. Also note that when this option is used no headers 
	    will be automatically added, all necessary headers have to be
	    provided by the user. </p>
	    <p>Defaults to <c>false</c>. </p>
	  </item>

          <tag><c><![CDATA[socket_opts]]></c></tag>
          <item>
            <p>Socket options to be used for this and subsequent 
	    request(s). </p>
	    <p>Overrides any value set by the 
	    <seealso marker="#set_options">set_options</seealso>
	    function. </p>
            <p>Note that the validity of the options are <em>not</em>
	    checked in any way. </p>
            <p>Note that this may change the socket behaviour 
	    (see <seealso marker="kernel:inet#setopts/2">inet:setopts/2</seealso>) 
	    for an already existing one, and therefore an already connected
	    request handler. </p>
            <p>By default the socket options set by the
	    <seealso marker="#set_options">set_options/1,2</seealso> 
	    function are used when establishing a connection. </p>
	  </item>

          <tag><c><![CDATA[receiver]]></c></tag>
          <item>
            <p>Defines how the client will deliver the result of an
	    asynchroneous request (<c>sync</c> has the value 
	    <c>false</c>). </p>

            <taglist>
              <tag><c><![CDATA[pid()]]></c></tag>
              <item>
	        <p>Message(s) will be sent to this process in the format: </p>
<pre>
{http, ReplyInfo}
</pre>
	      </item>

              <tag><c><![CDATA[function/1]]></c></tag>
              <item>
	        <p>Information will be delivered to the receiver via calls 
		to the provided fun: </p>
<pre>
Receiver(ReplyInfo)
</pre>
  	      </item>

              <tag><c><![CDATA[{Module, Funcion, Args}]]></c></tag>
              <item>
	        <p>Information will be delivered to the receiver via calls 
		to the callback function: </p>
<pre>
apply(Module, Function, [ReplyInfo | Args])
</pre>
	      </item>

            </taglist>
            <p>In all of the above cases, <c>ReplyInfo</c> has the following
              structure: </p>

<pre>
{RequestId, saved_to_file}
{RequestId, {error, Reason}}
{RequestId, Result}
{RequestId, stream_start, Headers}
{RequestId, stream_start, Headers, HandlerPid}
{RequestId, stream,       BinBodyPart}
{RequestId, stream_end,   Headers}
</pre>

            <p>Defaults to the <c>pid()</c> of the process calling the request 
	    function (<c>self()</c>). </p>

	    <marker id="ipv6_host_with_brackets"></marker>
	  </item>

          <tag><c><![CDATA[ipv6_host_with_brackets]]></c></tag>
          <item>
            <p>When parsing the Host-Port part of an URI with a IPv6 address 
	    with brackets, shall we retain those brackets (<c>true</c>) or 
	    strip them (<c>false</c>). </p>
            <p>Defaults to <c>false</c>. </p>
	  </item>

        </taglist>

        <marker id="cancel_request"></marker>
      </desc>
    </func>

    <func>
      <name>cancel_request(RequestId) -> </name>
      <name>cancel_request(RequestId, Profile) -> ok</name>
      <fsummary>Cancels an asynchronous HTTP-request.</fsummary>
      <type>
        <v>RequestId = request_id() - A unique identifier as returned
        by request/4</v>
	<v>Profile = profile() | pid() (when started <c>stand_alone</c>)</v>
      </type>
      <desc>
        <p>Cancels an asynchronous HTTP-request. </p>

        <marker id="set_options"></marker>
      </desc>
    </func>

    <func>
      <name>set_options(Options) -> </name>
      <name>set_options(Options, Profile) -> ok | {error, Reason}</name>
      <fsummary>Sets options to be used for subsequent requests.</fsummary>
      <type>
        <v>Options = [Option]</v>
        <v>Option = {proxy,                 {Proxy, NoProxy}} | 
                    {max_sessions,          MaxSessions} |
	            {max_keep_alive_length, MaxKeepAlive} | 
                    {keep_alive_timeout,    KeepAliveTimeout} |
	            {max_pipeline_length,   MaxPipeline} |  
                    {pipeline_timeout,      PipelineTimeout} | 
                    {cookies,               CookieMode} | 
                    {ipfamily,              IpFamily} | 
                    {ip,                    IpAddress} | 
                    {port,                  Port} | 
                    {socket_opts,           socket_opts()} | 
                    {verbose,               VerboseMode} </v>
        <v>Proxy = {Hostname, Port}</v>
        <v>Hostname = string() </v>
        <d>ex: "localhost" or "foo.bar.se"</d>
        <v>Port = integer()</v>
        <d>ex: 8080 </d>
        <v>socket_opts() = [socket_opt()]</v>
        <d>The options are appended to the socket options used by the 
          client. </d>
        <d>These are the default values when a new request handler
          is started (for the initial connect). They are passed directly 
          to the underlying transport (gen_tcp or ssl) <em>without</em> 
          verification! </d>
        <v>NoProxy = [NoProxyDesc]</v>
        <v>NoProxyDesc = DomainDesc | HostName | IPDesc</v>
        <v>DomainDesc = "*.Domain"</v>
        <d>ex: "*.ericsson.se"</d>
        <v>IpDesc = string()</v>
        <d>ex: "134.138" or "[FEDC:BA98" (all IP-addresses starting with 134.138 or FEDC:BA98), "66.35.250.150" or "[2010:836B:4179::836B:4179]" (a complete IP-address).</d>
        <v>MaxSessions = integer() </v>
        <d>Default is <c>2</c>.
          Maximum number of persistent connections to a host.</d>
	<v>MaxKeepAlive = integer() </v>
        <d>Default is <c>5</c>.
          Maximum number of outstanding requests on the same connection to 
          a host.</d>
        <v>KeepAliveTimeout = integer() </v>
        <d>Default is <c>120000</c> (= 2 min).
          If a persistent connection is idle longer than the 
          <c>keep_alive_timeout</c> the client will close the connection.
          The server may also have such a time out but you should
          not count on it!</d>
	<v>MaxPipeline = integer() </v>
        <d>Default is <c>2</c>.
          Maximum number of outstanding requests on a pipelined connection to a host.</d>
        <v>PipelineTimeout = integer() </v>
        <d>Default is <c>0</c>,
          which will result in pipelining not being used. 
          If a persistent connection is idle longer than the 
          <c>pipeline_timeout</c> the client will close the connection. </d>
        <v>CookieMode = enabled | disabled | verify </v>
        <d>Default is <c>disabled</c>.
          If Cookies are enabled all valid cookies will automatically be 
          saved in the client manager's cookie database. 
          If the option <c>verify</c> is used the function <c>store_cookies/2</c>
          has to be called for the cookies to be saved.</d>
        <v>IpFamily = inet | inet6 | inet6fb4 </v>
        <d>By default <c>inet</c>.
          When it is set to <c>inet6fb4</c> you can use both ipv4 and ipv6. 
          It first tries <c>inet6</c> and if that does not works falls back to <c>inet</c>. 
          The option is here to provide a workaround for buggy ipv6 stacks to ensure that 
          ipv4 will always work.</d>
	<v>IpAddress = ip_address() </v>
        <d>If the host has several network interfaces, this option specifies which one to use. 
          See <seealso marker="kernel:gen_tcp#connect">gen_tcp:connect/3,4</seealso> for more info. </d>
	<v>Port = integer() </v>
        <d>Specify which local port number to use. 
          See <seealso marker="kernel:gen_tcp#connect">gen_tcp:connect/3,4</seealso> for more info. </d>
        <v>VerboseMode = false | verbose | debug | trace </v>
        <d>Default is <c>false</c>.
          This option is used to switch on (or off) 
          different levels of erlang trace on the client. 
          It is a debug feature.</d>
	<v>Profile = profile() | pid() (when started <c>stand_alone</c>)</v>
      </type>
      <desc>
        <p>Sets options to be used for subsequent requests.</p>
        <note>
          <p>If possible the client will keep its connections
            alive and use persistent connections
	    with  or without pipeline depending on configuration
	    and current circumstances. The HTTP/1.1 specification does not
            provide a guideline for how many requests would be
            ideal to be sent on a persistent connection,
	    this very much depends on the
            application. Note that a very long queue of requests may cause a
            user perceived delay as earlier requests may take a long time
            to complete. The HTTP/1.1 specification does suggest a
            limit of 2 persistent connections per server, which is the
            default value of the <c>max_sessions</c> option. </p>
        </note>

        <marker id="stream_next"></marker>
      </desc>
    </func>
    
     <func>
      <name>stream_next(Pid) -> ok</name>
      <fsummary> Triggers the next message to be streamed, e.i.
	same behavior as active once for sockets.
      </fsummary>
      <type>
        <v>Pid = pid() - as received in the stream_start message</v>
      </type>
      <desc>
        <p>Triggers the next message to be streamed, e.i.
	same behavior as active once for sockets. </p>

        <marker id="verify_cookies"></marker>
        <marker id="store_cookies"></marker>
      </desc>
    </func>
    
    <func>
      <name>store_cookies(SetCookieHeaders, Url) -> </name>
      <name>store_cookies(SetCookieHeaders, Url, Profile) -> ok | {error, Reason}</name>
      <fsummary>Saves the cookies defined in SetCookieHeaders in the client profile's cookie database.</fsummary>
      <type>
        <v>SetCookieHeaders = headers() - where field = "set-cookie"</v>
        <v>Url = url()</v>
	<v>Profile = profile() | pid() (when started <c>stand_alone</c>)</v>
      </type>
      <desc>
        <p>Saves the cookies defined in SetCookieHeaders
	in the client profile's cookie database. You need to
	call this function if you have set the option <c>cookies</c> 
	to <c>verify</c>.
	If no profile is specified the default profile will be used. </p>

        <marker id="cookie_header"></marker>
      </desc>
    </func>

    <func>
      <name>cookie_header(Url) -> </name>
      <name>cookie_header(Url, Profile | Opts) -> header() | {error, Reason}</name>
      <name>cookie_header(Url, Opts, Profile) -> header() | {error, Reason}</name>
      <fsummary>Returns the cookie header that would be sent when
      making a request to Url using the profile <c>Profile</c>.</fsummary>
      <type>
        <v>Url = url()</v>
        <v>Opts = [cookie_header_opt()]</v>
	<v>Profile = profile() | pid() (when started <c>stand_alone</c>)</v>
        <v>cookie_header_opt() = {ipv6_host_with_brackets, boolean()}</v>
      </type>
      <desc>
        <p>Returns the cookie header that would be sent
	when making a request to <c>Url</c> using the profile <c>Profile</c>.
	If no profile is specified the default profile will be used. </p>
	<p>The option <c>ipv6_host_with_bracket</c> deals with how to 
	parse IPv6 addresses. 
	See the <c>Options</c> argument of the 
	<seealso marker="#request2">request/4,5</seealso> for more info. </p>

        <marker id="reset_cookies"></marker>
      </desc>
    </func>


    <func>
      <name>reset_cookies() -> void()</name>
      <name>reset_cookies(Profile) -> void()</name>
      <fsummary>Reset the cookie database.</fsummary>
      <type>
	<v>Profile = profile() | pid() (when started <c>stand_alone</c>)</v>
      </type>
      <desc>
        <p>Resets (clears) the cookie database for the specified 
	<c>Profile</c>. If no profile is specified the default profile 
	will be used. </p>
      </desc>
    </func>


    <func>
      <name>which_cookies() -> cookies()</name>
      <name>which_cookies(Profile) -> cookies()</name>
      <fsummary>Dumps out the entire cookie database.</fsummary>
      <type>
	<v>Profile = profile() | pid() (when started <c>stand_alone</c>)</v>
	<v>cookies() = [cookie_stores()]</v>
	<v>cookie_stores() = {cookies, cookies()} | {session_cookies, cookies()}</v>
	<v>cookies() = [cookie()]</v>
	<v>cookie() = term()</v>
      </type>
      <desc>
        <p>This function produces a list of the entire cookie database.
	It is intended for debugging/testing purposes. 
	If no profile is specified the default profile will be used. </p>
      </desc>
    </func>
  </funcs>

  <section>
    <title>SEE ALSO</title>
    <p>RFC 2616, <seealso marker="inets">inets(3)</seealso>,
      <seealso marker="kernel:gen_tcp">gen_tcp(3)</seealso>, 
      <seealso marker="ssl:ssl">ssl(3)</seealso>
    </p>
  </section>
  
</erlref>