<?xml version="1.0" encoding="latin1" ?>
<!DOCTYPE erlref SYSTEM "erlref.dtd">
<erlref>
<header>
<copyright>
<year>2000</year><year>2010</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>beam_lib</title>
<prepared>Hans Bolinder</prepared>
<docno></docno>
<date>1999-10-30</date>
<rev>PA1</rev>
</header>
<module>beam_lib</module>
<modulesummary>An Interface To the BEAM File Format</modulesummary>
<description>
<p><c>beam_lib</c> provides an interface to files created by
the BEAM compiler ("BEAM files"). The format used, a variant of
"EA IFF 1985" Standard for Interchange Format Files, divides data
into chunks.</p>
<p>Chunk data can be returned as binaries or as compound terms.
Compound terms are returned when chunks are referenced by names
(atoms) rather than identifiers (strings). The names recognized
and the corresponding identifiers are:</p>
<list type="bulleted">
<item><c>abstract_code ("Abst")</c></item>
<item><c>attributes ("Attr")</c></item>
<item><c>compile_info ("CInf")</c></item>
<item><c>exports ("ExpT")</c></item>
<item><c>labeled_exports ("ExpT")</c></item>
<item><c>imports ("ImpT")</c></item>
<item><c>indexed_imports ("ImpT")</c></item>
<item><c>locals ("LocT")</c></item>
<item><c>labeled_locals ("LocT")</c></item>
<item><c>atoms ("Atom")</c></item>
</list>
</description>
<section>
<marker id="debug_info"></marker>
<title>Debug Information/Abstract Code</title>
<p>The option <c>debug_info</c> can be given to the compiler (see
<seealso marker="compiler:compile#debug_info">compile(3)</seealso>)
in order to have debug information in the form of abstract code
(see <seealso marker="erts:absform">The Abstract Format</seealso>
in ERTS User's Guide) stored in the <c>abstract_code</c> chunk.
Tools such as Debugger and Xref require the debug information to
be included.</p>
<warning>
<p>Source code can be reconstructed from the debug information.
Use encrypted debug information (see below) to prevent this.</p>
</warning>
<p>The debug information can also be removed from BEAM files
using <seealso marker="#strip/1">strip/1</seealso>,
<seealso marker="#strip_files/1">strip_files/1</seealso> and/or
<seealso marker="#strip_release/1">strip_release/1</seealso>.</p>
<section>
<title>Reconstructing source code</title>
<p>Here is an example of how to reconstruct source code from
the debug information in a BEAM file <c>Beam</c>:</p>
<code type="none">
{ok,{_,[{abstract_code,{_,AC}}]}} = beam_lib:chunks(Beam,[abstract_code]).
io:fwrite("~s~n", [erl_prettypr:format(erl_syntax:form_list(AC))]).</code>
</section>
<section>
<title>Encrypted debug information</title>
<p>The debug information can be encrypted in order to keep
the source code secret, but still being able to use tools such as
Xref or Debugger. </p>
<p>To use encrypted debug information, a key must be provided to
the compiler and <c>beam_lib</c>. The key is given as a string and
it is recommended that it contains at least 32 characters and
that both upper and lower case letters as well as digits and
special characters are used.</p>
<p></p>
<p>The default type -- and currently the only type -- of crypto
algorithm is <c>des3_cbc</c>, three rounds of DES. The key string
will be scrambled using <c>erlang:md5/1</c> to generate
the actual keys used for <c>des3_cbc</c>.</p>
<note>
<p>As far as we know by the time of writing, it is
infeasible to break <c>des3_cbc</c> encryption without any
knowledge of the key. Therefore, as long as the key is kept
safe and is unguessable, the encrypted debug information
<em>should</em> be safe from intruders.</p>
</note>
<p>There are two ways to provide the key:</p>
<list type="ordered">
<item>
<p>Use the compiler option <c>{debug_info,Key}</c>, see
<seealso marker="compiler:compile#debug_info_key">compile(3)</seealso>,
and the function
<seealso marker="#crypto_key_fun/1">crypto_key_fun/1</seealso>
to register a fun which returns the key whenever
<c>beam_lib</c> needs to decrypt the debug information.</p>
<p>If no such fun is registered, <c>beam_lib</c> will instead
search for a <c>.erlang.crypt</c> file, see below.</p>
</item>
<item>
<p>Store the key in a text file named <c>.erlang.crypt</c>.</p>
<p>In this case, the compiler option <c>encrypt_debug_info</c>
can be used, see
<seealso marker="compiler:compile#encrypt_debug_info">compile(3)</seealso>.</p>
</item>
</list>
</section>
<section>
<title>.erlang.crypt</title>
<p><c>beam_lib</c> searches for <c>.erlang.crypt</c> in the current
directory and then the home directory for the current user. If
the file is found and contains a key, <c>beam_lib</c> will
implicitly create a crypto key fun and register it.</p>
<p>The <c>.erlang.crypt</c> file should contain a single list of
tuples:</p>
<code type="none">
{debug_info, Mode, Module, Key}</code>
<p><c>Mode</c> is the type of crypto algorithm; currently, the only
allowed value thus is <c>des3_cbc</c>. <c>Module</c> is either an
atom, in which case <c>Key</c> will only be used for the module
<c>Module</c>, or <c>[]</c>, in which case <c>Key</c> will be
used for all modules. <c>Key</c> is the non-empty key string.</p>
<p>The <c>Key</c> in the first tuple where both <c>Mode</c> and
<c>Module</c> matches will be used.</p>
<p>Here is an example of an <c>.erlang.crypt</c> file that returns
the same key for all modules:</p>
<code type="none"><![CDATA[
[{debug_info, des3_cbc, [], "%>7}|pc/DM6Cga*68$Mw]L#&_Gejr]G^"}].]]></code>
<p>And here is a slightly more complicated example of an
<c>.erlang.crypt</c> which provides one key for the module
<c>t</c>, and another key for all other modules:</p>
<code type="none"><![CDATA[
[{debug_info, des3_cbc, t, "My KEY"},
{debug_info, des3_cbc, [], "%>7}|pc/DM6Cga*68$Mw]L#&_Gejr]G^"}].]]></code>
<note>
<p>Do not use any of the keys in these examples. Use your own
keys.</p>
</note>
</section>
</section>
<section>
<title>DATA TYPES</title>
<code type="none">
beam() -> Module | Filename | binary()
Module = atom()
Filename = string() | atom()</code>
<p>Each of the functions described below accept either the module
name, the filename, or a binary containing the beam module.</p>
<code type="none">
chunkdata() = {ChunkId, DataB} | {ChunkName, DataT}
ChunkId = chunkid()
DataB = binary()
{ChunkName, DataT} =
{abstract_code, AbstractCode}
| {attributes, [{Attribute, [AttributeValue]}]}
| {compile_info, [{InfoKey, [InfoValue]}]}
| {exports, [{Function, Arity}]}
| {labeled_exports, [{Function, Arity, Label}]}
| {imports, [{Module, Function, Arity}]}
| {indexed_imports, [{Index, Module, Function, Arity}]}
| {locals, [{Function, Arity}]}]}
| {labeled_locals, [{Function, Arity, Label}]}]}
| {atoms, [{integer(), atom()}]}
AbstractCode = {AbstVersion, Forms} | no_abstract_code
AbstVersion = atom()
Attribute = atom()
AttributeValue = term()
Module = Function = atom()
Arity = int()
Label = int()</code>
<p>It is not checked that the forms conform to the abstract format
indicated by <c>AbstVersion</c>. <c>no_abstract_code</c> means
that the <c>"Abst"</c> chunk is present, but empty.</p>
<p>The list of attributes is sorted on <c>Attribute</c>, and each
attribute name occurs once in the list. The attribute values
occur in the same order as in the file. The lists of functions
are also sorted.</p>
<code type="none">
chunkid() = "Abst" | "Attr" | "CInf"
| "ExpT" | "ImpT" | "LocT"
| "Atom"
chunkname() = abstract_code | attributes | compile_info
| exports | labeled_exports
| imports | indexed_imports
| locals | labeled_locals
| atoms
chunkref() = chunkname() | chunkid()</code>
</section>
<funcs>
<func>
<name>chunks(Beam, [ChunkRef]) -> {ok, {Module, [ChunkData]}} | {error, beam_lib, Reason}</name>
<fsummary>Read selected chunks from a BEAM file or binary</fsummary>
<type>
<v>Beam = beam()</v>
<v>ChunkRef = chunkref()</v>
<v>Module = atom()</v>
<v>ChunkData = chunkdata()</v>
<v>Reason = {unknown_chunk, Filename, atom()}</v>
<v> | {key_missing_or_invalid, Filename, abstract_code}</v>
<v> | Reason1 -- see info/1</v>
<v> Filename = string()</v>
</type>
<desc>
<p>Reads chunk data for selected chunks refs. The order of
the returned list of chunk data is determined by the order
of the list of chunks references.</p>
</desc>
</func>
<func>
<name>chunks(Beam, [ChunkRef], [Option]) -> {ok, {Module, [ChunkResult]}} | {error, beam_lib, Reason}</name>
<fsummary>Read selected chunks from a BEAM file or binary</fsummary>
<type>
<v>Beam = beam()</v>
<v>ChunkRef = chunkref()</v>
<v>Module = atom()</v>
<v>Option = allow_missing_chunks</v>
<v>ChunkResult = {chunkref(), ChunkContents} | {chunkref(), missing_chunk}</v>
<v>Reason = {missing_chunk, Filename, atom()}</v>
<v> | {key_missing_or_invalid, Filename, abstract_code}</v>
<v> | Reason1 -- see info/1</v>
<v> Filename = string()</v>
</type>
<desc>
<p>Reads chunk data for selected chunks refs. The order of
the returned list of chunk data is determined by the order
of the list of chunks references.</p>
<p>By default, if any requested chunk is missing in <c>Beam</c>,
an <c>error</c> tuple is returned.
However, if the option <c>allow_missing_chunks</c> has been given,
a result will be returned even if chunks are missing.
In the result list, any missing chunks will be represented as
<c>{ChunkRef,missing_chunk}</c>.
Note, however, that if the <c>"Atom"</c> chunk if missing, that is
considered a fatal error and the return value will be an <c>error</c>
tuple.</p>
</desc>
</func>
<func>
<name>version(Beam) -> {ok, {Module, [Version]}} | {error, beam_lib, Reason}</name>
<fsummary>Read the BEAM file's module version</fsummary>
<type>
<v>Beam = beam()</v>
<v>Module = atom()</v>
<v>Version = term()</v>
<v>Reason -- see chunks/2</v>
</type>
<desc>
<p>Returns the module version(s). A version is defined by
the module attribute <c>-vsn(Vsn)</c>. If this attribute is
not specified, the version defaults to the checksum of
the module. Note that if the version <c>Vsn</c> is not a list,
it is made into one, that is <c>{ok,{Module,[Vsn]}}</c> is
returned. If there are several <c>-vsn</c> module attributes,
the result is the concatenated list of versions. Examples:</p>
<pre>
1> <input>beam_lib:version(a).</input> % -vsn(1).
{ok,{a,[1]}}
2> <input>beam_lib:version(b).</input> % -vsn([1]).
{ok,{b,[1]}}
3> <input>beam_lib:version(c).</input> % -vsn([1]). -vsn(2).
{ok,{c,[1,2]}}
4> <input>beam_lib:version(d).</input> % no -vsn attribute
{ok,{d,[275613208176997377698094100858909383631]}}</pre>
</desc>
</func>
<func>
<name>md5(Beam) -> {ok, {Module, MD5}} | {error, beam_lib, Reason}</name>
<fsummary>Read the BEAM file's module version</fsummary>
<type>
<v>Beam = beam()</v>
<v>Module = atom()</v>
<v>MD5 = binary()</v>
<v>Reason -- see chunks/2</v>
</type>
<desc>
<p>Calculates an MD5 redundancy check for the code of the module
(compilation date and other attributes are not included).</p>
</desc>
</func>
<func>
<name>info(Beam) -> [{Item, Info}] | {error, beam_lib, Reason1}</name>
<fsummary>Information about a BEAM file</fsummary>
<type>
<v>Beam = beam()</v>
<v>Item, Info -- see below</v>
<v>Reason1 = {chunk_too_big, Filename, ChunkId, ChunkSize, FileSize}</v>
<v> | {invalid_beam_file, Filename, Pos}</v>
<v> | {invalid_chunk, Filename, ChunkId}</v>
<v> | {missing_chunk, Filename, ChunkId}</v>
<v> | {not_a_beam_file, Filename}</v>
<v> | {file_error, Filename, Posix}</v>
<v> Filename = string()</v>
<v> ChunkId = chunkid()</v>
<v> ChunkSize = FileSize = int()</v>
<v> Pos = int()</v>
<v> Posix = posix() -- see file(3)</v>
</type>
<desc>
<p>Returns a list containing some information about a BEAM file
as tuples <c>{Item, Info}</c>:</p>
<taglist>
<tag><c>{file, Filename} | {binary, Binary}</c></tag>
<item>
<p>The name (string) of the BEAM file, or the binary from
which the information was extracted.</p>
</item>
<tag><c>{module, Module}</c></tag>
<item>
<p>The name (atom) of the module.</p>
</item>
<tag><c>{chunks, [{ChunkId, Pos, Size}]}</c></tag>
<item>
<p>For each chunk, the identifier (string) and the position
and size of the chunk data, in bytes.</p>
</item>
</taglist>
</desc>
</func>
<func>
<name>cmp(Beam1, Beam2) -> ok | {error, beam_lib, Reason}</name>
<fsummary>Compare two BEAM files</fsummary>
<type>
<v>Beam1 = Beam2 = beam()</v>
<v>Reason = {modules_different, Module1, Module2}</v>
<v> | {chunks_different, ChunkId}</v>
<v> | Reason1 -- see info/1</v>
<v> Module1 = Module2 = atom()</v>
<v> ChunkId = chunkid()</v>
</type>
<desc>
<p>Compares the contents of two BEAM files. If the module names
are the same, and all chunks except for the <c>"CInf"</c> chunk
(the chunk containing the compilation information which is
returned by <c>Module:module_info(compile)</c>)
have the same contents in both files,
<c>ok</c> is returned. Otherwise an error message is returned.</p>
</desc>
</func>
<func>
<name>cmp_dirs(Dir1, Dir2) -> {Only1, Only2, Different} | {error, beam_lib, Reason1}</name>
<fsummary>Compare the BEAM files in two directories</fsummary>
<type>
<v>Dir1 = Dir2 = string() | atom()</v>
<v>Different = [{Filename1, Filename2}]</v>
<v>Only1 = Only2 = [Filename]</v>
<v>Filename = Filename1 = Filename2 = string()</v>
<v>Reason1 = {not_a_directory, term()} | -- see info/1</v>
</type>
<desc>
<p>The <c>cmp_dirs/2</c> function compares the BEAM files in
two directories. Only files with extension <c>".beam"</c> are
compared. BEAM files that exist in directory <c>Dir1</c>
(<c>Dir2</c>) only are returned in <c>Only1</c>
(<c>Only2</c>). BEAM files that exist on both directories but
are considered different by <c>cmp/2</c> are returned as
pairs {<c>Filename1</c>, <c>Filename2</c>} where
<c>Filename1</c> (<c>Filename2</c>) exists in directory
<c>Dir1</c> (<c>Dir2</c>).</p>
</desc>
</func>
<func>
<name>diff_dirs(Dir1, Dir2) -> ok | {error, beam_lib, Reason1}</name>
<fsummary>Compare the BEAM files in two directories</fsummary>
<type>
<v>Dir1 = Dir2 = string() | atom()</v>
<v>Reason1 = {not_a_directory, term()} | -- see info/1</v>
</type>
<desc>
<p>The <c>diff_dirs/2</c> function compares the BEAM files in
two directories the way <c>cmp_dirs/2</c> does, but names of
files that exist in only one directory or are different are
presented on standard output.</p>
</desc>
</func>
<func>
<name>strip(Beam1) -> {ok, {Module, Beam2}} | {error, beam_lib, Reason1}</name>
<fsummary>Removes chunks not needed by the loader from a BEAM file</fsummary>
<type>
<v>Beam1 = Beam2 = beam()</v>
<v>Module = atom()</v>
<v>Reason1 -- see info/1</v>
</type>
<desc>
<p>The <c>strip/1</c> function removes all chunks from a BEAM
file except those needed by the loader. In particular,
the debug information (<c>abstract_code</c> chunk) is removed.</p>
</desc>
</func>
<func>
<name>strip_files(Files) -> {ok, [{Module, Beam2}]} | {error, beam_lib, Reason1}</name>
<fsummary>Removes chunks not needed by the loader from BEAM files</fsummary>
<type>
<v>Files = [Beam1]</v>
<v> Beam1 = beam()</v>
<v>Module = atom()</v>
<v>Beam2 = beam()</v>
<v>Reason1 -- see info/1</v>
</type>
<desc>
<p>The <c>strip_files/1</c> function removes all chunks except
those needed by the loader from BEAM files. In particular,
the debug information (<c>abstract_code</c> chunk) is removed.
The returned list contains one element for each given file
name, in the same order as in <c>Files</c>.</p>
</desc>
</func>
<func>
<name>strip_release(Dir) -> {ok, [{Module, Filename]}} | {error, beam_lib, Reason1}</name>
<fsummary>Removes chunks not needed by the loader from all BEAM files of a release</fsummary>
<type>
<v>Dir = string() | atom()</v>
<v>Module = atom()</v>
<v>Filename = string()</v>
<v>Reason1 = {not_a_directory, term()} | -- see info/1</v>
</type>
<desc>
<p>The <c>strip_release/1</c> function removes all chunks
except those needed by the loader from the BEAM files of a
release. <c>Dir</c> should be the installation root
directory. For example, the current OTP release can be
stripped with the call
<c>beam_lib:strip_release(code:root_dir())</c>.</p>
</desc>
</func>
<func>
<name>format_error(Reason) -> Chars</name>
<fsummary>Return an English description of a BEAM read error reply</fsummary>
<type>
<v>Reason -- see other functions</v>
<v>Chars = [char() | Chars]</v>
</type>
<desc>
<p>Given the error returned by any function in this module,
the function <c>format_error</c> returns a descriptive string
of the error in English. For file errors, the function
<c>file:format_error(Posix)</c> should be called.</p>
</desc>
</func>
<func>
<name>crypto_key_fun(CryptoKeyFun) -> ok | {error, Reason}</name>
<fsummary>Register a fun that provides a crypto key</fsummary>
<type>
<v>CryptoKeyFun = fun() -- see below</v>
<v>Reason = badfun | exists | term()</v>
</type>
<desc>
<p>The <c>crypto_key_fun/1</c> function registers a unary fun
that will be called if <c>beam_lib</c> needs to read an
<c>abstract_code</c> chunk that has been encrypted. The fun
is held in a process that is started by the function.</p>
<p>If there already is a fun registered when attempting to
register a fun, <c>{error, exists}</c> is returned.</p>
<p>The fun must handle the following arguments:</p>
<code type="none">
CryptoKeyFun(init) -> ok | {ok, NewCryptoKeyFun} | {error, Term}</code>
<p>Called when the fun is registered, in the process that holds
the fun. Here the crypto key fun can do any necessary
initializations. If <c>{ok, NewCryptoKeyFun}</c> is returned
then <c>NewCryptoKeyFun</c> will be registered instead of
<c>CryptoKeyFun</c>. If <c>{error, Term}</c> is returned,
the registration is aborted and <c>crypto_key_fun/1</c>
returns <c>{error, Term}</c> as well.</p>
<code type="none">
CryptoKeyFun({debug_info, Mode, Module, Filename}) -> Key</code>
<p>Called when the key is needed for the module <c>Module</c>
in the file named <c>Filename</c>. <c>Mode</c> is the type of
crypto algorithm; currently, the only possible value thus is
<c>des3_cbc</c>. The call should fail (raise an exception) if
there is no key available.</p>
<code type="none">
CryptoKeyFun(clear) -> term()</code>
<p>Called before the fun is unregistered. Here any cleaning up
can be done. The return value is not important, but is passed
back to the caller of <c>clear_crypto_key_fun/0</c> as part
of its return value.</p>
</desc>
</func>
<func>
<name>clear_crypto_key_fun() -> {ok, Result}</name>
<fsummary>Unregister the current crypto key fun</fsummary>
<type>
<v>Result = undefined | term()</v>
</type>
<desc>
<p>Unregisters the crypto key fun and terminates the process
holding it, started by <c>crypto_key_fun/1</c>.</p>
<p>The <c>clear_crypto_key_fun/1</c> either returns
<c>{ok, undefined}</c> if there was no crypto key fun
registered, or <c>{ok, Term}</c>, where <c>Term</c> is
the return value from <c>CryptoKeyFun(clear)</c>, see
<c>crypto_key_fun/1</c>.</p>
</desc>
</func>
</funcs>
</erlref>
