DSM: State machine notation for VoIP applications
DSM Readme file
The DonkeySM - state machines as SEMS applications
DonkeySM is a state machine interpreter for SEMS. Application
or service logic can comfortably and accurately be defined
as state machine, in a simple textual state machine definition
language, and executed by the dsm module as application in SEMS.
A DSM consists of states and transitions between the states.
One state is the initial state with which the DSM is started.
The transitions have conditions. If something happens,
i.e. the session processes an event, the transitions are checked
in order. If all conditions match, the transition is executed.
Transitions can contain actions, which are run when the transition
is executed. States can have actions that are executed on entering
the state, and on leaving the state. If a transition is executed,
first the leaving actions of the old state, then the transition
actions, and finally the entering actions of the new state are
executed.
DSMs can be defined in a hierarchical manner: Another DSM can be
called as sub-DSM, or one can jump to another DSM, ignoring where
we came from.
A session (call) in the DonkeySM has a set of named (string) variables.
The variables may be used as parameter to most conditions and
actions, by prepending the variable name with a dollar sign. The
parameters of an event (e.g. the key on key press) may be accessed
by prepending the name with a hash (e.g. #key). There are also
'selects' with which a set of dialog properties can be accessed
(e.g. @local_tag).
The DonkeySM can be extended by modules, which add new conditions
and actions to the language. This way, menuing system etc can be
implemented as DSM, while complex logic or processing can efficitently
be implemented in C++. Modules can act on new sessions, and have a
initialization function that is called when the module is loaded.
DonkeySM also has built in actions to call
DI methods from other modules.
Actions (and conditions) can throw exceptions. Once an exception occurs,
execution of the current actions is interrupted. Exceptions are handled
this way that special "exception" transitions are executed. Exception
transitions are marked with "exception" in the conditions list. Once the
FSM is in exception handling, only exception transitions are followed.
DSMs may throw exceptions with the throw(<type>) action or the
throwOnError() action.
DSM can cache a set of prompts, configured at start, in memory
using PromptCollection.
A patch for fmsc 1.0.4 from the graphical FSM editor fsme
(http://fsme.sf.net) is available, so DSMs can be defined in
click-n-drag fashion and compiled to SEMS DSM diagrams.
DI commands
===========
DI commands allow interaction with DSM calls, and DSM script reload:
postDSMEvent(string call_id, [ [[param0,val0],[param1,val1],...] ]
post a DSM event into a call. can be used to interact with running
calls in DSM. See DSM + monitoring + DI example in
examples/dsm_di_monit.
Example:
s.postDSMEvent(call_id, [['action', 'take'],['roomname', 'realworld']])
reloadDSMs()
reload all DSMs from config file (load_diags) from main config
- DSM is loaded with main config
loadDSM(string diag_name)
load DSM with name diag_name, paths are taken from config file
- DSM is loaded with main config
loadDSMWithPaths(string diag_name, string diag_path, string mod_path)
load DSM with specified paths
- DSM is loaded with main config
preloadModules()
preload all modules specified in config file (preload_mods)
preloadModule(string mod_name, string mod_path)
preload module from specific path
hasDSM(string diag_name, [string config])
returns 1 if DSM with diag_name is loaded, 0 if not
if config empty or not given, DSM of main config will be listed
listDSMs([string config])
return list of loaded DSMs
if config empty or not given, DSM of main config will be listed
registerApplication(string diag_name, [string config])
register DSM with name diag_name as application in SEMS
(e.g. to be used with application=$(apphdr), $(ruri.param)
or $(ruri.user)
if config empty or not given, DSM is assumed to be in main config
loadConfig(string conf_file_name, string conf_name)
(re)load application configuration and script
like a file in conf_dir
More info
=========
o doc/dsm_syntax.txt has a quick reference for dsm syntax
o doc/examples/ and lib/ some example DSMs
o doc/examples/dsm_di_monit example on interfacing with DSM
o mods/ (will) have modules
Internals
=========
The DSMStateEngine has a set of DSM diagrams which are loaded by
the DSMStateDiagramCollection from text file and interpreted by
the DSMChartReader, a simple stack based tokenizing compiler.
DSMCall, which implement the DSMSession interface (additionally
to being an AmSession), run DSMStateEngine::runEvent for every event
that occurs that should be processed by the engine (e.g. Audio event,
onBye, ...).
The DSMStateEngine checks every condition of the active state whether
it matches. If all match, the exit actions of the current state, the
transition actions and then the enter actions of the next state are
executed. The DSMCondition::match and DSMAction::execute functions
get the event parameters and the session as parameters, so that they
can operate on variables, implement selects etc.
The DSMCall implementation is very simple, it uses a playlist and
has PromptCollection to simply play prompts etc.
DSMCoreModule is a 'built in' module that implements the basic
conditions (test(), hangup() etc) and actions (set(), playFile(), DI
etc).
Roadmap
=======
On the roadmap is the possibility for modules to be the session factory
used for creating the new session. As the DSMSession is mostly an abstract
interface, other session types can easily be implemented, and their
functionality be exposed to the DSM interpreter by custom actions and
conditions that interact with that specific session type.
A set of modules exposing more of the core functionality.
As the call state representation is nicely encapsulated here, this can
also provide an abstraction layer on which active call replication can
be implemented (rather than re-doing that for every application).
Q&A
===
Why "Donkey" SM?
Thanks Atle for the name: "I dont know why.. but my first thought
when I read good name for FSM interpreter was Donkey. The reason for the
name is that you put alot of things ontop of a donkey, and let it carry
it arround.. and this is the same.. you put loads of stuf ontop.. and it
carries it."
What is repost()?
If an event should be reevaluated, e.g. by the transitions of another
state or the initial state of another DSM which is called with callFSM/
jumpFSM, repost() can be called, which signals the interpreter to
evaluate the current event again.
Is the diagram interpreted or compiled?
DonkeySM reads the DSM script and creates an internal representation
(DSM classes in STL containers). This way it can be executed very
efficiently.
map<string, string> var; - Are you crazy? E stands for Express!
yes, right, there would be more efficient ways to implement
that. Anyway, in my experience one mostly has to manipulate and
check strings, and it is just very comfortable to do it this way.
OTOH, if in a normal call there is a transition maybe on average
every 10 seconds, for which 5 conditions are checked, it is not
so much an impact on performance, considering that we are processing
an audio packet every 20 ms.
You rely too heavily on polymorphism and RTTI - one dynamic_cast for
each condition of each transition is way too heavy!
Sure, but as noted above, there should not be heavy processing done
in the DSM. If you need this, then consider writing your app entirely
in C++.
SEMS has a dynamically typed type (AmArg), why not use that one for
variables? That would also make DI simpler.
a patch is very welcome, best to semsdev list: semsdev@iptel.org or
the tracker: http://tracker.iptel.org.
There is also the avar array ("AmArg-Var"), which can hold AmArg
variables.
some performance numbers?
unfortunately not yet for running DSMs. DSM processing is actually fast:
a (quite simple) 3 state 5 transition DSM compiles in 0.2ms on P-M 2GHz.
So the diagram could actually be read when the call is setup, or DSMs could
load other DSMs (e.g. loadFSM() action)
DSM Syntax
=============================
-- comment
import(mod_name);
[initial] state name
[ enter {
action;
action;
...
} ]
[ exit {
action;
action;
...
} ]
;
transition name s1 - [exception;] [ { [not] condition; [not] condition; ... } ] [/ { action; action; ...} ] -> s2;
or
transition name (s1a, s1b[, s1c, ...]) - [ { [not] condition; [not] condition; ... } ] [/ { action; action; ...} ] -> s2;
=============================
#paramname uses the event parameter 'paramname' (from current event)
$varname uses the variable varname (from session's variable map)
@selectname uses the "select" 'selectname' (from the session's dialog)
=============================
core actions:
-- reprocess the current event after transition:
repost()
-- call/jump/return sub-FSM
jumpFSM(name)
callFSM(name)
returnFSM()
stop(<send bye>)
e.g. stop(false), stop(true)
playPrompt(param)
from promptCollection, e.g. playPrompt("hello");
if $prompts.default_fallback=yes, default prompt set is tried if
prompt not found in current prompt set
Throws "prompt" exeption with #name if prompt not found.
playPromptLooped(param)
setPromptSet(name)
select active prompt set if more prompt sets are loaded
Throws "prompt" exeption with #name if prompt set not found
playFile(filename [, loop=true])
e.g. playFile($myfile, true); will play the file looped.
Throws "file" exeption with #path if file can not be opened
playFileFront(filename [, loop=true])
e.g. playFileFront($myfile, true); will play the file at first
position in the playlist, and looped.
Throws "file" exeption with #path if file can not be opened
recordFile(filename)
Throws "file" exeption with #path if file can not be opened for recording
stopRecord()
getRecordLength([dst_varname]) -- only while recording! default dst var: record_length
getRecordDataSize([dst_varname]) -- only while recording! default dst var: record_data_size
closePlaylist()
addSeparator(id [, bool front])
fires event when playlist hits it ; front=[true|false]
connectMedia()
set playlist as input and output of session, and connect to mediaprocessor
disconnectMedia()
disconnect from mediaprocessor
mute()
set RTP stream to muted (don't send and receive RTP packets)
unmute()
set RTP stream to unmuted (send and receive RTP packets)
postEvent(sess_id[, variable name])
post dsm event to session sess_id; variable copied as event variable
e.g. postEvent(@local_tag, PAI) : post event to ourselves
* sets $errno (arg)
postEvent(sess_id, var)
all local variables copied as event variables
* sets $errno (arg)
enableDTMFDetection()
disableDTMFDetection()
B2B.connectCallee(remote_party, remote_uri)
connect second leg of B2B session (see AmB2BSession)
B2B.terminateOtherLeg
disconnect second leg of B2B session (see AmB2BSession)
B2B.sendReinvite(generate_sdp [, string extra_headers])
send a reinvite in caller leg (first leg), e.g. to
reconnect first leg after B2B.otherBye received.
generate_sdp can be 'true' or 'false'
(B2B.sendReinvite(true) recommended)
B2B.clearHeaders()
clear the headers used for outgoing INVITE on B leg
B2B.addHeader(string header)
add a header for outgoing INVITE on B leg
B2B.setHeaders(string headers [, replace_crlf=true|false])
set headers for outgoing INVITE on B leg
replace_crlf=true for replacing \r\n with CRLF
e.g.
B2B.setHeaders("P-One: value\r\nP-Two: anothervalue", true)
setPrompts(name)
if more than one prompt sets are loaded
set($var=value)
e.g. set($var="text"); set($var=$var2); set($var=#key)
var($dstvar=srcvarname)
e.g. var($dstvar=$var_counter)
eval($var=value)
evaluate expression (only simple binary + and - supported)
e.g. set($var=1+5); set($var=$var2); set($var=#key)
append($var, value)
e.g. append($var, "text"); append($var, #key);
append($var, @select); append($var, $var2);
inc($var)
clear($var)
log(level, text)
e.g. log(1, $var1)
-- log all variables:
logVars(level)
-- log all selects:
logSelects(level)
-- log all Params (only in actions of a transition):
logParams(level)
-- log everything:
logAll(level)
setTimer(timer_id, timeout)
e.g. setTimer(1, $timeout)
* sets $errno (arg,config)
removeTimer(timer_id)
* sets $errno (arg,config)
removeTimers()
* sets $errno (config)
DI(factory, function [, params...])
e.g. DI(factory, function, $var_param, (int)int_param, "$str param", @select_par, ...)
DI(user_timer, setTimer, (int)1, (int)5, @local_tag);
* sets $errno (arg,config)
DIgetResult(factory, function, param,...)
saves result from DI call to DI_res or DI_res0, DI_res1, ...
* sets $errno (arg,config)
throw(<type>[,arg1=val1;arg2=val2;...])
e.g. throw(bad_command), throw(bad_command,cmd=help;reason=whynot)
throwOnError()
=============================
conditions:
test(#key == 1)
test(#key != 1)
test(#key < 1)
test(#key > 1)
test($var == 1)
test($var != 1)
test($var < 1)
test($var < 1)
test(len($var) < len(#key))
test(len($var) < len(@user))
-- like test(expr), but only on key press
keyTest(expr)
timerTest(expr)
noAudioTest(expr)
separatorTest(expr)
e.g. separatorTest(#id == 5)
eventTest(expr)
keyPress(no)
-- bye received:
hangup
-- execution is invite (with run_invite_event):
invite
-- execution is start of session (with run_invite_event, otherwise its always true):
sessionStart
B2B.otherReply
Reply on other leg received (see AmB2BSession)
#code - reply code
#reason - reply reason
#hdrs - headers
B2B.otherBye
BYE on other leg received
#hdrs - headers
=============================
@selects :
local_tag
user
domain
remote_tag
callid
local_uri
remote_uri
=============================
imports:
module imported with import(mod_name); loads mod_name.so in
module load path. modules provide conditions and actions.
modules' actions/conditions-factory is checked first
(modules can override core conditions/actions)
=============================
special variables:
connect_session "0" -> after the start event (initial transition):
do not connect session to
media processor on start
-> after the invite event:
do not reply with 200 OK and do not
connect session to media processor on start
=============================
errors:
actions set $errno
#define DSM_ERRNO_OK ""
#define DSM_ERRNO_FILE "1"
#define DSM_ERRNO_UNKNOWN_ARG "2"
#define DSM_ERRNO_GENERAL "99"
...
Error handling in DSM
Error handling in DSM
---------------------
There are two mechanisms of error handling in DSM:
Reporting an error as $errno, and throwing an exception.
Some actions throw an exception directly if a severe error occurs,
for example the "playFile" action if the file could
not be opened. These functions are marked with
" Throws "<xyz>" exeption with <arguments> if ..."
in the description/documentation.
Other actions just set the error variable $errno. After those
the DSM script may execute the throwOnError(); action, which
throws an exception if the error is set. These actions are
marked in the description as "Sets Errno."
All actions that may set the variable $errno to some value
need to reset $errno if no error occured, otherwise an exception
from an error of a previous action could be thrown.
Exceptions may be caught with exception transitions (see general DSM
description).
For Python (mod_py), exception handling is special - the the mod_py
documentation.
Example 1: Handling file exceptions
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-
state BEGIN
enter {
playFile(maybeexisting.wav);
};
transition "key pressed" BEGIN - keyPress(1) -> OTHER;
state OTHER
enter {
playFile(mayalsobeexisting.wav);
};
transition "file error" (BEGIN, OTHER) - exception; test(#type=="file") /
stop(true) -> end;
transition "hangs up" (BEGIN, OTHER) - hangup / stop(false) -> end;
state end;
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-
Example 2: Handling error with exception
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-
import(mod_mysql);
state BEGIN
enter {
mysql.connect();
throwOnError();
mysql.queryGetResult(select file from prompts where user=@user);
throwOnError();
playFile($file);
};
transition "mysql connection error occured" BEGIN - exception; test(#type=="connection") /
log(1, MySQL connection error); playFile(sorry_error.wav) -> PLAYEND;
transition "mysql query error / no result" BEGIN - exception; test(#type=="noresult") /
playFile(default_prompt.wav) -> PLAYEND;
transition "file opening error" BEGIN - exception; test(#type=="file") /
playFile(default_prompt.wav) -> PLAYEND;
transition "other error occured" BEGIN - exception; /
log(1, error:); log(1, #text); playFile(sorry_error.wav) -> PLAYEND;
state PLAYEND;
transition "finished" PLAYEND - noAudioTest / stop(true) -> end;
state end;
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-
DSM Todo
x run_invite_event per DSM script o explicit condition ordering (explicit priority) o checking of FSM consitency (e.g. target state names) o better error reporting in script compilation (e.g. line number) x better error handling (exceptions) x 'not' operator on conditions o 'or' operator x provide session init function for modules x embed python interpreter in module ? -> py( ... python code ... ) condition and action x correct en/decode DI args (e.g. arrays) o replace DSMSession::map<string, string> var with map<string, AmArg> var -> avar x register scripts as app x mod_mysql o session creation from module
DSM modules
mod_sys - System functions
Actions:
sys.mkdir(string dirname)
sys.mkdirRecursive(string dirname)
sys.rename(string from, string to)
sys.unlink(string filename)
sys.unlinkArray(string filename, string prefix)
Array version of unlink (prefix/filename_0 .. prefix/filename_$filename_size)
sys.tmpnam(string varname)
Conditions:
sys.file_exists(string fname)
sys.file_not_exists(string fname)
mod_dlg - Dialog related functionality
* mod_dlg saves the initial INVITE to DSMSession::last_req * set connect_session to 0 with set(connect_session=0) if you want to reply with other than the standard 200 OK to initial INVITE received. dlg.reply(code,reason); reply to the request in DSMSession::last_req (usually INVITE, if not yet replied) with code and reason * sets $errno (arg,general) dlg.acceptInvite([code, reason]); e.g. dlg.acceptInvite(183, progress); * sets $errno (arg,general) accept audio stream from last_req (INVITE), and reply with 200 OK (default) or code, reason dlg.bye([headers]) send BYE. useful for example for continuing processing after call has ended. * sets $errno (general) dlg.connectCalleeRelayed(string remote_party, string remote_uri) like B2B.connectCallee() but for relayed INVITEs, i.e. for executing in invite run (run_invite_event=yes and transition "on INVITE" START - invite -> runinvite;)
mod_uri - URI parser and related functionality
uri.parse(<uri>, <prefix>)
splits <uri> in
<prefix>display_name
<prefix>user
<prefix>host
<prefix>param
example:
uri.parse(@remote_uri, remote_);
uri.parse($PAI, pai_);
* Sets $errno (general).
uri.getHeader(<header>, <dst>)
get header from initial INVITE into variable <dst>
example:
uri.getHeader(P-Asserted-Identity, PAI);
mod_utils - Utility functions
Actions:
utils.getNewId(string varname)
utils.playCountRight(int cnt [, string basedir])
play count for laguages that have single digits after the 10s (like english)
* Throws "file" exeption with #path if file can not be opened
* sets $errno (arg)
utils.playCountLeft(int cnt [, string basedir])
play count for laguages that have single digits befire the 10s (like german)
* Throws "file" exeption with #path if file can not be opened
* sets $errno (arg)
utils.spell(string word[, string basedir])
plays each character in the word (e.g. utils.spell(321,wav/digits/) plays
wav/digits/3.wav, wav/digits/2.wav, wav/digits/1.wav
(like SayDigits from *)
* Throws "file" exeption with #path if file can not be opened
utils.rand(string varname [, int modulo])
generates random number: $varname=rand()%modulo or $varname = rand()
utils.srand()
seed the RNG with time().
mod_conference - Conferencing functions
Actions: conference.join(string roomname [, string mode]) mode = "" | speakonly | listenonly Throws "conference" Exception if conference can not be joined (currently never). conference.leave() destroy conference channel. Close playlist first!!!!! * Sets $errno (script). conference.rejoin(string roomname [, string mode]) mode = "" | speakonly | listenonly Throws "conference" Exception if conference can not be joined (currently never). conference.postEvent(string roomname, int event_id) * Sets $errno (arg). conference.setPlayoutType(string type) where type is one of ["adaptive", "jb", "simple"]
mod_mysql - MySQL DB access
mod_py - Embedded Python functions
mod_py - DSM actions and conditions in Python
=============================================
mod_py adds one action and one condition to DSM: py(...)
py() action may execute arbitrary python code; py() condition
python code needs to evaluate to an expression which returns
True or False (or and int value).
The 'type' and 'params' are accessible to determine the current
event's type and parameters.
py() actions and conditions can access session's variables using
session.var(name) and session.setvar(name, value). Locals stay
across different py(...) actions/conditions.
They may even directly use some media functionality implemented
in DSM sessions - see session module's help below. But,
conceptionally, mod_py is above DSM, so while it would be
possible it is not recommended to extend it with functions that
manipulate the AmSession directly.
Indentation is a little ugly with multi-line py() actions.
But this is Python's fault (in 21st century, who creates a
programming language with fixed indentation?).
mod_py MUST be preloaded, to initialize the python interpreter.
add preload=mod_py to dsm.conf.
locals
======
type - event type (dsm.Timer, dsm.Key, ...)
params - dictionary with event parameters
(e.g. params['id'] for event==dsm.Timer)
dsm - module to access dsm functions (see below)
session - module to access session functions (see below)
exceptions
==========
dsm.playPrompt, dsm.playFile, dsm.recordFile and dsm.setPromptSet
may throw an exception. This is a Python RuntimeError exception,
not the DSM exception handling (exceptions in interpreted code
can not be catched through outside of the interpreter).
The application needs to catch the exceptions in the python code
itself, and can then handle them or signal the DSM code to throw
an exception. example:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
initial state begin
enter {
py(dsm.INFO("hello dsm-world\n"))
py(#
try:
session.playFile('doesnotexist.wav')
except RuntimeError, e:
dsm.ERROR('thats a runtime error: ' + e.message + '!\n')
session.setvar('exception', e.message)
)
log(2, huhu, still there);
repost();
};
transition "catch py exception" begin - test($exception!="") / throw($exception) -> begin;
transition "normal exception handling" begin - exception; test(#type=="file") -> exception_state;
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
example
=======
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
import(mod_py);
initial state begin
enter {
py(dsm.INFO("hello dsm-world"))
py(#
session.setvar('some_variable','some val')
print session.var('some_variable')
print "dsm.Timer = ", dsm.Timer
)
setTimer(1, 5);
log(2, $some_variable)
repost();
};
transition "timer" begin - py(type == dsm.Timer and params['id'] == '1') /
py(session.playFile('wav/default_en.wav')); -> wait;
transition "key 1" begin - py(type == dsm.Key and params['key'] == '1') /
py(session.playFile('wav/default_en.wav')); -> wait;
state wait;
transition "bye recvd" (begin, wait) - hangup / stop -> end;
state end;
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
modules
=======
------------------------
help on module session:
NAME
session
FILE
(built-in)
FUNCTIONS
B2BconnectCallee(string remote_party, string remote_uri [, bool relayed_invite])
connect callee of B2B leg
B2BterminateOtherLeg()
terminate other leg of B2B call
addSeparator(string name[, bool front])
add a named separator to playlist
closePlaylist(bool notify)
close the playlist
connectMedia()
connect media (RTP processing)
disconnectMedia()
disconnect media (RTP processing)
getRecordDataSize()
get the data size of the current recording
getRecordLength()
get the length of the current recording
mute()
mute RTP)
playFile(string name [, bool loop[, bool front]])
play a file
playPrompt(string name [, bool loop])
play a prompt
recordFile(string name)
start recording to a file
select(string name)
get a session's select
setError(int errno)
set error (errno)
setPromptSet(string name)
set prompt set
setvar(string name, string value)
set a session's variable
stopRecord()
stop the running recording
unmute()
unmute RTP
var(string name)
get a session's variable
------------------------------------
Help on module dsm:
NAME
dsm
FILE
(built-in)
FUNCTIONS
DBG(string msg)
Log a message using SEMS' logging system, level debug
ERROR(string msg)
Log a message using SEMS' logging system, level error
INFO(string msg)
Log a message using SEMS' logging system, level info
WARN(string msg)
Log a message using SEMS' logging system, level warning
log(int level, string msg)
Log a message using SEMS' logging system
DATA
Any = 0
B2BOtherBye = 13
B2BOtherReply = 12
DSMEvent = 10
Hangup = 6
Hold = 7
Invite = 1
Key = 3
NoAudio = 5
PlaylistSeparator = 11
SessionStart = 2
Timer = 4
UnHold = 8
XmlrpcResponse = 9
how to debug memory leak:
-------------------------
Unfortunately, it seems to be not simple to get embedded
python interpreter leak free. Here is how to run with python's
mem debug:
compile python with --with-pydebug, e.g.
./configure --with-pydebug --prefix=/path/to/mod_dsm/python
make && make install
set debug python in Makefile, e.g. replace PY_VER/PY_EXE:
PY_VER = 2.5
PY_EXE = ./python/bin/python
make mod_py with -D PYDSM_WITH_MEM_DEBUG
run sems with -E, make calls, end sems with ctrl-c
from python do scripts/combinerefs.py refs.txt.
generate calls with e.g. sipp:
sipp -sn uac -i 192.168.5.106 -s 35 -d 500 -r 400 192.168.5.106:5070
(sudo su; ulimit -n 100000 before starting sems)
mod_aws - Amazon Web Services functions
mod_aws - Copyright (C) 2009 TelTech Systems Inc.
This module brings Amazon Web Services to DSM.
It uses and depends on libaws++, see http://aws.28msec.com/.
mod_aws must be preloaded. See aws.conf for configuration
(aws credentials, connection pool size etc).
Actions
-------
S3: bucket name from $aws.s3.bucket is used if not specified as
action parameter.
aws.s3.put(string filename [, string keyname])
put file into bucket, if specified under key keyname,
else keyname=filename
* throws Exception type="aws", with #cause
* sets $errno (aws_put)
aws.s3.putArray(string filename_array [, string keyname_array])
put files into bucket, if specified under key keyname,
else keyname=filename
array is $filename_0, $filename_1, .. $filename_size
* throws Exception type="aws", with #cause
* sets $errno (aws_put,arg)
SQS: queue name from $aws.sqs.queue is used if not specified as
action parameter.
aws.s3.createBucket([string bucket_name])
create bucket. created bucket name in $aws.s3.bucket (should be same
as bucket_name)
* throws Exception type="aws", with #cause
* sets $errno (aws_create,aws_conn,arg)
aws.sqs.createQueue(int visibilityTimeout [, string queue_name])
create queue
* sets $errno (aws_create,aws_conn,arg)
aws.sqs.deleteQueue([string queue_name])
delete queue
* sets $errno (aws_create,aws_conn,arg)
aws.sqs.sendMessage(string message [, string queue_name])
send a message in queue. sets
$aws.sqs.url
$aws.sqs.md5
* sets $errno (aws_create,aws_conn,arg)
Error handling
--------------
On error, errno is set, and aws.ereason has more detailed
description.
Errno codes:
#define DSM_ERRNO_AWS_CONN "50"
#define DSM_ERRNO_AWS_PUT "51"
#define DSM_ERRNO_AWS_CREATE "52"
#define DSM_ERRNO_AWS_DELETE "53"
#define DSM_ERRNO_AWS_SEND "54"
Links
Back to Application Modules Documentation, to Example Applications.
Generated on Wed Mar 17 14:34:30 2010 for SEMS by
1.6.1