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

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

<erlref>
  <header>
    <copyright>
      <year>2009</year>
      <year>2021</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.

  The Initial Developer of the Original Code is Ericsson AB.
    </legalnotice>

    <title>binary</title>
    <prepared>Patrik Nyblom</prepared>
    <responsible>Kenneth Lundin</responsible>
    <docno>1</docno>
    <approved></approved>
    <checked></checked>
    <date>2010-05-05</date>
    <rev>A</rev>
    <file>binary.xml</file>
  </header>
  <module since="OTP R14B">binary</module>
  <modulesummary>Library for handling binary data.</modulesummary>
  <description>

    <p>This module contains functions for manipulating byte-oriented
    binaries. Although the majority of functions could be provided
    using bit-syntax, the functions in this library are highly
    optimized and are expected to either execute faster or consume
    less memory, or both, than a counterpart written in pure Erlang.</p>

    <p>The module is provided according to Erlang Enhancement Proposal
      (EEP) 31.</p>

    <note>
      <p>The library handles byte-oriented data. For bitstrings that are not
      binaries (does not contain whole octets of bits) a <c>badarg</c>
      exception is thrown from any of the functions in this module.</p>
    </note>
  </description>

  <datatypes>
    <datatype>
      <name name="cp"/>
      <desc><p>Opaque data type representing a compiled
        search pattern. Guaranteed to be a <c>tuple()</c> to allow programs to
        distinguish it from non-precompiled search patterns.</p>
      </desc>
    </datatype>
    <datatype>
      <name name="part"/>
      <desc><p>A representation of a part (or range) in a binary. <c>Start</c> is
        a zero-based offset into a <c>binary()</c> and <c>Length</c> is the
        length of that part. As input to functions in this module, a reverse
        part specification is allowed, constructed with a negative
        <c>Length</c>, so that the part of the binary begins at <c>Start</c> +
        <c>Length</c> and is -<c>Length</c> long. This is useful for referencing
        the last <c>N</c> bytes of a binary as <c>{size(Binary), -N}</c>. The
        functions in this module always return <c>part()</c>s with positive
        <c>Length</c>.</p>
      </desc>
    </datatype>
  </datatypes>

  <funcs>
    <func>
    <name name="at" arity="2" since="OTP R14B"/>
      <fsummary>Return the byte at a specific position in a binary.</fsummary>
      <desc>
      <p>Returns the byte at position <c><anno>Pos</anno></c> (zero-based) in
      binary <c><anno>Subject</anno></c> as an integer. If
      <c><anno>Pos</anno></c> &gt;= <c>byte_size(<anno>Subject</anno>)</c>,
      a <c>badarg</c> exception is raised.</p>
      </desc>
    </func>

    <func>
      <name name="bin_to_list" arity="1" since="OTP R14B"/>
      <fsummary>Convert a binary to a list of integers.</fsummary>
      <desc>
      <p>Same as <c>bin_to_list(<anno>Subject</anno>, {0,byte_size(<anno>Subject</anno>)})</c>.</p>
      </desc>
    </func>

    <func>
      <name name="bin_to_list" arity="2" since="OTP R14B"/>
      <fsummary>Convert a binary to a list of integers.</fsummary>
      <desc>
      <p>Converts <c><anno>Subject</anno></c> to a list of <c>byte()</c>s, each
      representing the value of one byte. <c>part()</c> denotes which part of
      the <c>binary()</c> to convert.</p>

      <p><em>Example:</em></p>

<code>
1> binary:bin_to_list(&lt;&lt;"erlang"&gt;&gt;, {1,3}).
"rla"
%% or [114,108,97] in list notation.</code>

      <p>If <c><anno>PosLen</anno></c> in any way references outside the binary,
      a <c>badarg</c> exception is raised.</p>
      </desc>
    </func>

    <func>
      <name name="bin_to_list" arity="3" since="OTP R14B"/>
      <fsummary>Convert a binary to a list of integers.</fsummary>
      <desc>
      <p>Same as<c> bin_to_list(<anno>Subject</anno>, {<anno>Pos</anno>, <anno>Len</anno>})</c>.</p>
      </desc>
    </func>

    <func>
      <name name="compile_pattern" arity="1" since="OTP R14B"/>
      <fsummary>Precompile a binary search pattern.</fsummary>
      <desc>
      <p>Builds an internal structure representing a compilation of a
      search pattern, later to be used in functions
      <seemfa marker="#match/3"><c>match/3</c></seemfa>,
      <seemfa marker="#matches/3"><c>matches/3</c></seemfa>,
      <seemfa marker="#split/3"><c>split/3</c></seemfa>, or
      <seemfa marker="#replace/4"><c>replace/4</c></seemfa>.
      The <c>cp()</c> returned is guaranteed to be a
      <c>tuple()</c> to allow programs to distinguish it from
      non-precompiled search patterns.</p>

      <p>When a list of binaries is specified, it denotes a set of
      alternative binaries to search for. For example, if
      <c>[&lt;&lt;"functional"&gt;&gt;,&lt;&lt;"programming"&gt;&gt;]</c>
      is specified as <c><anno>Pattern</anno></c>, this
      means either <c>&lt;&lt;"functional"&gt;&gt;</c> or
      <c>&lt;&lt;"programming"&gt;&gt;</c>". The pattern is a set of
      alternatives; when only a single binary is specified, the set has
      only one element. The order of alternatives in a pattern is
      not significant.</p>

      <p>The list of binaries used for search alternatives must be flat and
      proper.</p>

      <p>If <c><anno>Pattern</anno></c> is not a binary or a flat proper list of
      binaries with length &gt; 0, a <c>badarg</c> exception is raised.</p>
      </desc>
    </func>

    <func>
      <name name="copy" arity="1" since="OTP R14B"/>
      <fsummary>Create a duplicate of a binary.</fsummary>
      <desc>
      <p>Same as <c>copy(<anno>Subject</anno>, 1)</c>.</p>
      </desc>
    </func>

    <func>
      <name name="copy" arity="2" since="OTP R14B"/>
      <fsummary>Duplicate a binary <c>N</c> times and create a new.</fsummary>
      <desc>
      <p>Creates a binary with the content of <c><anno>Subject</anno></c>
      duplicated <c><anno>N</anno></c> times.</p>

      <p>This function always creates a new binary, even if <c><anno>N</anno> =
      1</c>. By using <seemfa marker="#copy/1"><c>copy/1</c></seemfa>
      on a binary referencing a larger binary, one
      can free up the larger binary for garbage collection.</p>

      <note>
      <p>By deliberately copying a single binary to avoid referencing
      a larger binary, one can, instead of freeing up the larger
      binary for later garbage collection, create much more binary
      data than needed. Sharing binary data is usually good. Only in
      special cases, when small parts reference large binaries and the
      large binaries are no longer used in any process, deliberate
      copying can be a good idea.</p>
      </note>

      <p>If <c><anno>N</anno></c> &lt; <c>0</c>, a <c>badarg</c> exception is
      raised.</p>
      </desc>
    </func>

    <func>
      <name name="decode_unsigned" arity="1" since="OTP R14B"/>
      <fsummary>Decode a whole binary into an integer of arbitrary size.
      </fsummary>
      <desc>
      <p>Same as <c>decode_unsigned(<anno>Subject</anno>, big)</c>.</p>
      </desc>
    </func>

    <func>
      <name name="decode_unsigned" arity="2" since="OTP R14B"/>
      <fsummary>Decode a whole binary into an integer of arbitrary size.
      </fsummary>
      <desc>
      <p>Converts the binary digit representation, in big endian or little
      endian, of a positive integer in <c><anno>Subject</anno></c> to an Erlang
      <c>integer()</c>.</p>

      <p><em>Example:</em></p>

      <code>
1> binary:decode_unsigned(&lt;&lt;169,138,199&gt;&gt;,big).
11111111</code>
      </desc>
    </func>

    <func>
      <name name="encode_unsigned" arity="1" since="OTP R14B"/>
      <fsummary>Encode an unsigned integer into the minimal binary.</fsummary>
      <desc>
      <p>Same as <c>encode_unsigned(<anno>Unsigned</anno>, big)</c>.</p>
      </desc>
    </func>

    <func>
      <name name="encode_unsigned" arity="2" since="OTP R14B"/>
      <fsummary>Encode an unsigned integer into the minimal binary.</fsummary>
      <desc>
      <p>Converts a positive integer to the smallest possible
      representation in a binary digit representation, either big endian
      or little endian.</p>

      <p><em>Example:</em></p>

      <code>
1> binary:encode_unsigned(11111111, big).
&lt;&lt;169,138,199&gt;&gt;</code>
      </desc>
    </func>

    <func>
      <name name="encode_hex" arity="1" since="OTP 24.0"/>
      <fsummary>Encodes a binary into a hex encoded binary.</fsummary>
      <desc>
      <p>Encodes a binary into a hex encoded binary.</p>

      <p><em>Example:</em></p>

      <code>
1> binary:encode_hex(&lt;&lt;"f"&gt;&gt;).
&lt;&lt;"66"&gt;&gt;</code>
      </desc>
    </func>

    <func>
      <name name="decode_hex" arity="1" since="OTP 24.0"/>
      <fsummary>Decodes a hex encoded binary into a binary.</fsummary>
      <desc>
      <p>Decodes a hex encoded binary into a binary.</p>

      <p><em>Example</em></p>

      <code>
1> binary:decode_hex(&lt;&lt;"66"&gt;&gt;).
&lt;&lt;"f"&gt;&gt;</code>
      </desc>
    </func>

    <func>
      <name name="first" arity="1" since="OTP R14B"/>
      <fsummary>Return the first byte of a binary.</fsummary>
      <desc>
      <p>Returns the first byte of binary <c><anno>Subject</anno></c> as an
      integer. If the size of <c><anno>Subject</anno></c> is zero, a
      <c>badarg</c> exception is raised.</p>
      </desc>
    </func>

    <func>
      <name name="last" arity="1" since="OTP R14B"/>
      <fsummary>Return the last byte of a binary.</fsummary>
      <desc>
      <p>Returns the last byte of binary <c><anno>Subject</anno></c> as an
      integer. If the size of <c><anno>Subject</anno></c> is zero, a
      <c>badarg</c> exception is raised.</p>
      </desc>
    </func>

    <func>
      <name name="list_to_bin" arity="1" since="OTP R14B"/>
      <fsummary>Convert a list of integers and binaries to a binary.</fsummary>
      <desc>
      <p>Works exactly as
      <seemfa marker="erts:erlang#list_to_binary/1"><c>erlang:list_to_binary/1</c></seemfa>,
      added for completeness.</p>
      </desc>
    </func>

    <func>
      <name name="longest_common_prefix" arity="1" since="OTP R14B"/>
      <fsummary>Return length of longest common prefix for a set of binaries.
      </fsummary>
      <desc>
      <p>Returns the length of the longest common prefix of the
      binaries in list <c><anno>Binaries</anno></c>.</p>

      <p><em>Example:</em></p>

<code>
1> binary:longest_common_prefix([&lt;&lt;"erlang"&gt;&gt;, &lt;&lt;"ergonomy"&gt;&gt;]).
2
2> binary:longest_common_prefix([&lt;&lt;"erlang"&gt;&gt;, &lt;&lt;"perl"&gt;&gt;]).
0</code>

      <p>If <c><anno>Binaries</anno></c> is not a flat list of binaries, a
      <c>badarg</c> exception is raised.</p>
      </desc>
    </func>

    <func>
      <name name="longest_common_suffix" arity="1" since="OTP R14B"/>
      <fsummary>Return length of longest common suffix for a set of binaries.
      </fsummary>
      <desc>
      <p>Returns the length of the longest common suffix of the
      binaries in list <c><anno>Binaries</anno></c>.</p>

      <p><em>Example:</em></p>

<code>
1> binary:longest_common_suffix([&lt;&lt;"erlang"&gt;&gt;, &lt;&lt;"fang"&gt;&gt;]).
3
2> binary:longest_common_suffix([&lt;&lt;"erlang"&gt;&gt;, &lt;&lt;"perl"&gt;&gt;]).
0</code>

      <p>If <c>Binaries</c> is not a flat list of binaries, a <c>badarg</c>
      exception is raised.</p>
      </desc>
    </func>

    <func>
      <name name="match" arity="2" since="OTP R14B"/>
      <fsummary>Search for the first match of a pattern in a binary.</fsummary>
      <desc>
      <p>Same as <c>match(<anno>Subject</anno>, <anno>Pattern</anno>, [])</c>.
      </p>
      </desc>
    </func>

    <func>
      <name name="match" arity="3" since="OTP R14B"/>
      <fsummary>Search for the first match of a pattern in a binary.</fsummary>
      <type name="part"/>
      <desc>
      <p>Searches for the first occurrence of <c><anno>Pattern</anno></c> in
      <c><anno>Subject</anno></c> and returns the position and length.</p>

      <p>The function returns <c>{Pos, Length}</c> for the binary
      in <c><anno>Pattern</anno></c>, starting at the lowest position in
      <c><anno>Subject</anno></c>.</p>

      <p><em>Example:</em></p>

<code>
1> binary:match(&lt;&lt;"abcde"&gt;&gt;, [&lt;&lt;"bcde"&gt;&gt;, &lt;&lt;"cd"&gt;&gt;],[]).
{1,4}</code>

      <p>Even though <c>&lt;&lt;"cd"&gt;&gt;</c> ends before
      <c>&lt;&lt;"bcde"&gt;&gt;</c>, <c>&lt;&lt;"bcde"&gt;&gt;</c>
      begins first and is therefore the first match. If two
      overlapping matches begin at the same position, the longest is
      returned.</p>

      <p>Summary of the options:</p>

      <taglist>
      <tag>{scope, {<anno>Start</anno>, <anno>Length</anno>}}</tag>
      <item><p>Only the specified part is searched. Return values still have
      offsets from the beginning of <c><anno>Subject</anno></c>. A negative
      <c>Length</c> is allowed as described in section Data Types in this
      manual.</p></item>
      </taglist>

      <p>If none of the strings in <c><anno>Pattern</anno></c> is found, the
      atom <c>nomatch</c> is returned.</p>

      <p>For a description of <c><anno>Pattern</anno></c>, see function
      <seemfa marker="#compile_pattern/1"><c>compile_pattern/1</c></seemfa>.
      </p>

      <p>If <c>{scope, {Start,Length}}</c> is specified in the options such
      that <c>Start</c> &gt; size of <c>Subject</c>, <c>Start</c> +
      <c>Length</c> &lt; 0 or <c>Start</c> + <c>Length</c> &gt; size of
      <c>Subject</c>, a <c>badarg</c> exception is raised.</p>
      </desc>
    </func>

    <func>
      <name name="matches" arity="2" since="OTP R14B"/>
      <fsummary>Search for all matches of a pattern in a binary.</fsummary>
      <desc>
      <p>Same as <c>matches(<anno>Subject</anno>, <anno>Pattern</anno>, [])</c>.
      </p>
      </desc>
    </func>

    <func>
      <name name="matches" arity="3" since="OTP R14B"/>
      <fsummary>Search for all matches of a pattern in a binary.</fsummary>
      <type name="part"/>
      <desc>
      <p>As <seemfa marker="#match/2"><c>match/2</c></seemfa>,
      but <c><anno>Subject</anno></c> is searched until
      exhausted and a list of all non-overlapping parts matching
      <c><anno>Pattern</anno></c> is returned (in order).</p>

      <p>The first and longest match is preferred to a shorter,
      which is illustrated by the following example:</p>

<code>
1> binary:matches(&lt;&lt;"abcde"&gt;&gt;,
                  [&lt;&lt;"bcde"&gt;&gt;,&lt;&lt;"bc"&gt;&gt;,&lt;&lt;"de"&gt;&gt;],[]).
[{1,4}]</code>

       <p>The result shows that &lt;&lt;"bcde"&gt;&gt; is selected instead of
       the shorter match &lt;&lt;"bc"&gt;&gt; (which would have given raise to
       one more match, &lt;&lt;"de"&gt;&gt;).
       This corresponds to the behavior of
       POSIX regular expressions (and programs like awk), but is not
       consistent with alternative matches in <c>re</c> (and Perl), where
       instead lexical ordering in the search pattern selects which
       string matches.</p>

       <p>If none of the strings in a pattern is found, an empty list is
       returned.</p>

       <p>For a description of <c><anno>Pattern</anno></c>, see
       <seemfa marker="#compile_pattern/1"><c>compile_pattern/1</c></seemfa>.
       For a description of available options, see
       <seemfa marker="#match/3"><c>match/3</c></seemfa>.</p>

       <p>If <c>{scope, {<anno>Start</anno>,<anno>Length</anno>}}</c> is
       specified in the options such that <c><anno>Start</anno></c> &gt; size
       of <c><anno>Subject</anno></c>, <c><anno>Start</anno> +
       <anno>Length</anno></c> &lt; 0 or <c><anno>Start</anno> +
       <anno>Length</anno></c> is &gt; size of <c><anno>Subject</anno></c>,
       a <c>badarg</c> exception is raised.</p>
     </desc>
    </func>

    <func>
      <name name="part" arity="2" since="OTP R14B"/>
      <fsummary>Extract a part of a binary.</fsummary>
      <desc>
      <p>Extracts the part of binary <c><anno>Subject</anno></c> described by
      <c><anno>PosLen</anno></c>.</p>

      <p>A negative length can be used to extract bytes at the end of a
      binary:</p>

<code>
1> Bin = &lt;&lt;1,2,3,4,5,6,7,8,9,10&gt;&gt;.
2> binary:part(Bin, {byte_size(Bin), -5}).
&lt;&lt;6,7,8,9,10&gt;&gt;</code>

      <note>
      <p><seemfa marker="#part/2">part/2</seemfa> and
      <seemfa marker="#part/3">part/3</seemfa> are also available in the
      <seeerl marker="erts:erlang"><c>erlang</c></seeerl>
      module under the names <c>binary_part/2</c> and
      <c>binary_part/3</c>. Those BIFs are allowed in guard tests.</p>
      </note>

      <p>If <c><anno>PosLen</anno></c> in any way references outside the binary,
      a <c>badarg</c> exception is raised.</p>
      </desc>
    </func>

    <func>
      <name name="part" arity="3" since="OTP R14B"/>
      <fsummary>Extract a part of a binary.</fsummary>
      <desc>
      <p>Same as <c>part(<anno>Subject</anno>, {<anno>Pos</anno>,
      <anno>Len</anno>})</c>.</p>
      </desc>
    </func>

    <func>
      <name name="referenced_byte_size" arity="1" since="OTP R14B"/>
      <fsummary>Determine the size of the binary pointed out by a subbinary.
      </fsummary>
      <desc>
      <p>If a binary references a larger binary (often described as
      being a subbinary), it can be useful to get the size of the
      referenced binary. This function can be used in a program to trigger the
      use of <seemfa marker="#copy/1"><c>copy/1</c></seemfa>. By copying a
      binary, one can dereference the original, possibly large, binary that a
      smaller binary is a reference to.</p>

      <p><em>Example:</em></p>

      <code>
store(Binary, GBSet) ->
  NewBin =
      case binary:referenced_byte_size(Binary) of
          Large when Large > 2 * byte_size(Binary) ->
             binary:copy(Binary);
          _ ->
             Binary
      end,
  gb_sets:insert(NewBin,GBSet).</code>

      <p>In this example, we chose to copy the binary content before
      inserting it in <c>gb_sets:set()</c> if it references a binary more than
      twice the data size we want to keep. Of course,
      different rules apply when copying to different programs.</p>

      <p>Binary sharing occurs whenever binaries are taken apart.
      This is the fundamental reason why binaries are fast,
      decomposition can always be done with O(1) complexity. In rare
      circumstances this data sharing is however undesirable, why this
      function together with <c>copy/1</c> can be useful when optimizing
      for memory use.</p>

      <p>Example of binary sharing:</p>

      <code>
1> A = binary:copy(&lt;&lt;1&gt;&gt;, 100).
&lt;&lt;1,1,1,1,1 ...
2> byte_size(A).
100
3> binary:referenced_byte_size(A).
100
4> &lt;&lt;B:10/binary, C:90/binary&gt;&gt; = A.
&lt;&lt;1,1,1,1,1 ...
5> {byte_size(B), binary:referenced_byte_size(B)}.
{10,10}
6> {byte_size(C), binary:referenced_byte_size(C)}.
{90,100}</code>
      <p>In the above example, the small binary <c>B</c> was copied while the
      larger binary <c>C</c> references binary <c>A</c>.</p>
      <note>
      <p>Binary data is shared among processes. If another process
      still references the larger binary, copying the part this
      process uses only consumes more memory and does not free up the
      larger binary for garbage collection. Use this kind of intrusive
      functions with extreme care and only if a real problem is detected.</p>
      </note>
      </desc>
    </func>

    <func>
      <name name="replace" arity="3" since="OTP R14B"/>
      <fsummary>Replace bytes in a binary according to a pattern.</fsummary>
      <desc>
      <p>Same as <c>replace(<anno>Subject</anno>, <anno>Pattern</anno>, <anno>Replacement</anno>,[])</c>.</p>
      </desc>
    </func>

    <func>
      <name name="replace" arity="4" since="OTP R14B"/>
      <fsummary>Replace bytes in a binary according to a pattern.</fsummary>
      <type_desc variable="OnePos">An integer() =&lt; byte_size(<anno>Replacement</anno>)
      </type_desc>
      <desc>
      <p>Constructs a new binary by replacing the parts in
      <c><anno>Subject</anno></c> matching <c><anno>Pattern</anno></c> with
      the content of <c><anno>Replacement</anno></c>.</p>

      <p>If the matching subpart of <c><anno>Subject</anno></c> giving raise
      to the replacement is to be inserted in the result, option
      <c>{insert_replaced, <anno>InsPos</anno>}</c> inserts the matching part
      into <c><anno>Replacement</anno></c> at the specified position (or
      positions) before inserting <c><anno>Replacement</anno></c> into
      <c><anno>Subject</anno></c>.</p>

      <p><em>Example:</em></p>

<code>
1> binary:replace(&lt;&lt;"abcde"&gt;&gt;,&lt;&lt;"b"&gt;&gt;,&lt;&lt;"[]"&gt;&gt;, [{insert_replaced,1}]).
&lt;&lt;"a[b]cde"&gt;&gt;
2> binary:replace(&lt;&lt;"abcde"&gt;&gt;,[&lt;&lt;"b"&gt;&gt;,&lt;&lt;"d"&gt;&gt;],&lt;&lt;"[]"&gt;&gt;,[global,{insert_replaced,1}]).
&lt;&lt;"a[b]c[d]e"&gt;&gt;
3> binary:replace(&lt;&lt;"abcde"&gt;&gt;,[&lt;&lt;"b"&gt;&gt;,&lt;&lt;"d"&gt;&gt;],&lt;&lt;"[]"&gt;&gt;,[global,{insert_replaced,[1,1]}]).
&lt;&lt;"a[bb]c[dd]e"&gt;&gt;
4> binary:replace(&lt;&lt;"abcde"&gt;&gt;,[&lt;&lt;"b"&gt;&gt;,&lt;&lt;"d"&gt;&gt;],&lt;&lt;"[-]"&gt;&gt;,[global,{insert_replaced,[1,2]}]).
&lt;&lt;"a[b-b]c[d-d]e"&gt;&gt;</code>

        <p>If any position specified in <c><anno>InsPos</anno></c> &gt; size
        of the replacement binary, a <c>badarg</c> exception is raised.</p>

        <p>Options <c>global</c> and <c>{scope, part()}</c> work as for
        <seemfa marker="#split/3"><c>split/3</c></seemfa>.
        The return type is always a <c>binary()</c>.</p>

        <p>For a description of <c><anno>Pattern</anno></c>, see
        <seemfa marker="#compile_pattern/1"><c>compile_pattern/1</c></seemfa>.
        </p>
      </desc>
    </func>

    <func>
      <name name="split" arity="2" since="OTP R14B"/>
      <fsummary>Split a binary according to a pattern.</fsummary>
      <desc>
      <p>Same as <c>split(<anno>Subject</anno>, <anno>Pattern</anno>,
      [])</c>.</p>
      </desc>
    </func>

    <func>
      <name name="split" arity="3" since="OTP R14B"/>
      <fsummary>Split a binary according to a pattern.</fsummary>
      <desc>
      <p>Splits <c><anno>Subject</anno></c> into a list of binaries based on
      <c><anno>Pattern</anno></c>. If option <c>global</c> is not specified,
      only the first occurrence of <c><anno>Pattern</anno></c> in
      <c><anno>Subject</anno></c> gives rise to a split.</p>

      <p>The parts of <c><anno>Pattern</anno></c> found in
      <c><anno>Subject</anno></c> are not included in the result.</p>

      <p><em>Example:</em></p>

<code>
1> binary:split(&lt;&lt;1,255,4,0,0,0,2,3&gt;&gt;, [&lt;&lt;0,0,0&gt;&gt;,&lt;&lt;2&gt;&gt;],[]).
[&lt;&lt;1,255,4&gt;&gt;, &lt;&lt;2,3&gt;&gt;]
2> binary:split(&lt;&lt;0,1,0,0,4,255,255,9&gt;&gt;, [&lt;&lt;0,0&gt;&gt;, &lt;&lt;255,255&gt;&gt;],[global]).
[&lt;&lt;0,1&gt;&gt;,&lt;&lt;4&gt;&gt;,&lt;&lt;9&gt;&gt;]</code>

      <p>Summary of options:</p>

      <taglist>
      <tag>{scope, part()}</tag>
      <item><p>Works as in <seemfa marker="#match/3"><c>match/3</c></seemfa>
      and <seemfa marker="#matches/3"><c>matches/3</c></seemfa>. Notice that
      this only defines the scope of the search for matching strings,
      it does not cut the binary before splitting. The bytes before and after
      the scope are kept in the result. See the example below.</p></item>
      <tag>trim</tag>
      <item><p>Removes trailing empty parts of the result (as does <c>trim</c>
      in <seemfa marker="re#split/3"><c>re:split/3</c></seemfa>.</p></item>
      <tag>trim_all</tag>
      <item><p>Removes all empty parts of the result.</p></item>
      <tag>global</tag>
      <item><p>Repeats the split until <c><anno>Subject</anno></c> is
      exhausted. Conceptually option <c>global</c> makes split work
      on the positions returned by
      <seemfa marker="#matches/3"><c>matches/3</c></seemfa>, while it
      normally works on the position returned by
      <seemfa marker="#match/3"><c>match/3</c></seemfa>.</p></item>
      </taglist>

     <p>Example of the difference between a scope and taking the
     binary apart before splitting:</p>

<code>
1> binary:split(&lt;&lt;"banana"&gt;&gt;, [&lt;&lt;"a"&gt;&gt;],[{scope,{2,3}}]).
[&lt;&lt;"ban"&gt;&gt;,&lt;&lt;"na"&gt;&gt;]
2> binary:split(binary:part(&lt;&lt;"banana"&gt;&gt;,{2,3}), [&lt;&lt;"a"&gt;&gt;],[]).
[&lt;&lt;"n"&gt;&gt;,&lt;&lt;"n"&gt;&gt;]</code>

      <p>The return type is always a list of binaries that are all
      referencing <c><anno>Subject</anno></c>. This means that the data in
      <c><anno>Subject</anno></c> is not copied to new binaries, and that
      <c><anno>Subject</anno></c> cannot be garbage collected until the results
      of the split are no longer referenced.</p>

      <p>For a description of <c><anno>Pattern</anno></c>, see
      <seemfa marker="#compile_pattern/1"><c>compile_pattern/1</c></seemfa>.
      </p>
      </desc>
    </func>
  </funcs>
</erlref>