George Joseph [Tue, 14 Dec 2021 17:39:16 +0000 (10:39 -0700)]
bundled_pjproject: Add more support for multipart bodies
Adding upstream patch for pull request...
https://github.com/pjsip/pjproject/pull/2920
---------------------------------------------------------------
sip_inv: Additional multipart support (#2919)
sip_inv.c:inv_check_sdp_in_incoming_msg() deals with multipart
message bodies in rdata correctly. In the case where early media is
involved though, the existing sdp has to be retrieved from the last
tdata sent in this transaction. This, however, always assumes that
the sdp sent is in a non-multipart body. While there's a function
to retrieve the sdp from multipart and non-multpart rdata bodies,
no similar function for tdata exists. So...
* The existing pjsip_rdata_get_sdp_info2 was refactored to
find the sdp in any body, multipart or non-multipart, and
from either an rdata or tdata. The new function is
pjsip_get_sdp_info. This new function detects whether the
pjsip_msg->body->data is the text representation of the sdp
from an rdata or an existing pjmedia_sdp_session object
from a tdata, or whether pjsip_msg->body is a multipart
body containing either of the two sdp formats.
* The exsting pjsip_rdata_get_sdp_info and pjsip_rdata_get_sdp_info2
functions are now wrappers that get the body and Content-Type
header from the rdata and call pjsip_get_sdp_info.
* Two new wrappers named pjsip_tdata_get_sdp_info and
pjsip_tdata_get_sdp_info2 have been created that get the body
from the tdata and call pjsip_get_sdp_info.
* inv_offer_answer_test.c was updated to test multipart scenarios.
Naveen Albert [Mon, 13 Dec 2021 22:49:52 +0000 (22:49 +0000)]
pbx_variables: initialize uninitialized variable
The variable cp4 in a variable substitution function
can potentially be used without being initialized
currently. This causes Asterisk to no longer compile.
This initializes cp4 to NULL to make the compiler
happy.
Kevin Harwell [Mon, 15 Nov 2021 21:35:05 +0000 (15:35 -0600)]
http.c: Add ability to create multiple HTTP servers
Previously, it was only possible to have one HTTP server in Asterisk.
With this patch it is now possible to have multiple HTTP servers
listening on different addresses.
Note, this behavior has only been made available through an API call
from within the TEST_FRAMEWORK. Specifically, this feature has been
added in order to allow unit test to create/start and stop servers,
if one has not been enabled through configuration.
Naveen Albert [Mon, 13 Dec 2021 00:08:26 +0000 (00:08 +0000)]
app.c: Throw warnings for nonexistent options
Currently, Asterisk doesn't throw warnings if options
are passed into applications that don't accept them.
This can confuse users if they're unaware that they
are doing something wrong.
This adds an additional check to parse_options so that
a warning is thrown anytime an option is parsed that
doesn't exist in the parsing application, so that users
are notified of the invalid usage.
Naveen Albert [Sun, 12 Dec 2021 02:11:21 +0000 (02:11 +0000)]
strings: Fix enum names in comment examples
The enum values for ast_strsep_flags includes
AST_STRSEP_STRIP. However, some comments reference
AST_SEP_STRIP, which doesn't exist. This fixes
these comments to use the correct value.
Naveen Albert [Wed, 17 Nov 2021 21:16:56 +0000 (21:16 +0000)]
configs: Updates to sample configs
Includes some minor updates to extensions.conf
and iax.conf. In particular, the demonstration
of macros in extensions.conf is removed, as
Macro is deprecated and will be removed soon.
These examples have been replaced with examples
demonstrating the usage of Gosub instead.
The older exten => ...,n syntax is also mostly
replaced with the same keyword to demonstrate the
newer, more concise way of defining extensions.
IAXTEL no longer exists, so this example is replaced
with something more generic.
Some documentation is also added to extensions.conf
and iax.conf to clarify some of the new expanded
encryption capabilities with IAX2.
Naveen Albert [Wed, 17 Nov 2021 21:39:34 +0000 (21:39 +0000)]
app_sendtext: Add ReceiveText application
Adds a ReceiveText application that can be used in
conjunction with SendText. Currently, there is no
way in Asterisk to receive text in the dialplan
(or anywhere else, really). This allows for Asterisk
to be the recipient of text instead of just the sender.
Naveen Albert [Sat, 20 Nov 2021 20:37:16 +0000 (20:37 +0000)]
pbx_variables: Increase parsing capabilities of MSet
Currently MSet can only parse a maximum of 24 variables.
If more variables are provided to MSet, the 24th variable
will simply contain the remainder of the string and the
remaining variables thereafter will never get set.
This increases the number of variables that can be parsed
in one go from 24 to 99. Additionally, documentation is added
since this limitation is currently undocumented and is
confusing to users who encounter this limitation.
Naveen Albert [Wed, 24 Nov 2021 02:21:23 +0000 (02:21 +0000)]
chan_sip: Fix crash when accessing RURI before initiating outgoing call
Attempting to access ${CHANNEL(ruri)} in a pre-dial handler before
initiating an outgoing call will cause Asterisk to crash. This is
because a null field is accessed, resulting in an offset from null and
subsequent memory access violation.
Since RURI is not guaranteed to exist, we now check if the base
pointer is non-null before calculating an offset.
Naveen Albert [Mon, 25 Oct 2021 21:19:49 +0000 (21:19 +0000)]
func_json: Adds JSON_DECODE function
Adds the JSON_DECODE function for parsing JSON in the
dialplan. JSON parsing already exists in the Asterisk
core and is used for many different things. This
function exposes the basic parsing capability to
the user in the dialplan, for instance, in conjunction
with CURL for using API responses.
Naveen Albert [Mon, 15 Nov 2021 21:08:11 +0000 (21:08 +0000)]
pbx: Add variable substitution API for extensions
Currently, variable substitution involving dialplan
extensions is quite clunky since it entails obtaining
the current dialplan location, backing it up, storing
the desired variables for substitution on the channel,
performing substitution, then restoring the original
location.
In addition to being clunky, things could also go wrong
if an async goto were to occur and change the dialplan
location during a substitution.
Fundamentally, there's no reason it needs to be done this
way, so new API is added to allow for directly passing in
the dialplan location for the purposes of variable
substitution so we don't need to mess with the channel
information anymore. Existing API is not changed.
Alexander Traud [Mon, 6 Dec 2021 10:25:20 +0000 (11:25 +0100)]
xmldoc: Avoid whitespace around value for parameter/required.
Otherwise, the value 'false' was not found in the enumerated set of
the XML DTD for the XML attribute 'required' in the XML element
'parameter'. Therefore, DTD validation of the runtime XML failed.
Alexander Traud [Tue, 23 Nov 2021 14:05:55 +0000 (15:05 +0100)]
progdocs: Update Makefile.
In developer mode, use internal documentation as well.
This should produce no warnings. Fix yours!
In noisy mode, output all possible warnings of Doxygen.
This creates zillion of warnings. Double-check your current module!
Any warnings are in the file './doxygen.log'. Beside that, this change
avoids deprecated parameters because the configuration file for Doxygen
contains only those parameters which differ from the default. This
avoids the need to update the file on each run. Furthermore, it adds
AST_VECTOR to be expanded. Finally, the default name for that file is
Doxyfile. Therefore, let us use that!
Sean Bright [Tue, 30 Nov 2021 20:16:45 +0000 (15:16 -0500)]
channel: Short-circuit ast_channel_get_by_name() on empty arg.
We know that passing a NULL or empty argument to
ast_channel_get_by_name() will never result in a matching channel and
will always result in an error being emitted, so just short-circuit
out in that case.
Mike Bradeen [Tue, 26 Oct 2021 21:12:18 +0000 (15:12 -0600)]
res_rtp_asterisk: Addressing possible rtp range issues
res/res_rtp_asterisk.c: Adding 1 to rtpstart if it is deteremined
that rtpstart was configured to be an odd value. Also adding a loop
counter to prevent a possible infinite loop when looking for a free
port.
Dustin Marquess [Tue, 9 Nov 2021 00:30:00 +0000 (18:30 -0600)]
res_fax_spandsp: Add spandsp 3.0.0+ compatibility
Newer versions of spandsp did refactoring of code to add new features
like color FAXing. This refactoring broke backwards compatibility.
Add support for the new version while retaining support for 0.0.6.
Alexander Traud [Fri, 12 Nov 2021 16:05:05 +0000 (17:05 +0100)]
progdocs: Fix grouping for latest Doxygen.
Since Doxygen 1.8.16, a special comment block is required. Otherwise
(pure C comment), the group command is ignored. Additionally, several
unbalanced group commands were fixed.
Naveen Albert [Thu, 25 Nov 2021 18:41:18 +0000 (18:41 +0000)]
documentation: Standardize examples
Most examples in the XML documentation use the
example tag to demonstrate examples, which gets
parsed specially in the Wiki to make it easier
to follow for users.
This fixes a few modules to use the example
tag instead of vanilla para tags to bring them
in line with the standard syntax.
Sean Bright [Sun, 28 Nov 2021 20:52:24 +0000 (15:52 -0500)]
config.c: Prevent UB in ast_realtime_require_field.
A backend's implementation of the realtime 'require' function may call
va_arg() and then fail, leaving the va_list in an undefined
state. Pass a copy of the va_list instead.
Refactors generic functions used for email generation
into utils.c so that they can be used by multiple
modules, including app_voicemail and app_minivm,
to avoid code duplication.
Jaco Kroon [Thu, 28 Oct 2021 07:28:57 +0000 (09:28 +0200)]
logger: use __FUNCTION__ instead of __PRETTY_FUNCTION__
This avoids a few long-name overflows, at the cost of less instructive
names in the case of C++ (specifically overloaded functions and class
methods). This in turn is offset against the fact that we're logging
the filename and line numbers in any case.
Mike Bradeen [Tue, 16 Nov 2021 22:34:28 +0000 (15:34 -0700)]
astobj2.c: Fix core when ref_log enabled
In the AO2_ALLOC_OPT_LOCK_NOLOCK case the referenced obj
structure is freed, but is then referenced later if ref_log is
enabled. The change is to store the obj->priv_data.options value
locally and reference it instead of the value from the freed obj
Joshua C. Colp [Tue, 16 Nov 2021 10:06:26 +0000 (06:06 -0400)]
bridge: Deny full Local channel pair in bridge.
Local channels are made up of two pairs - the 1 and 2
sides. When a frame goes in one side, it comes out the
other. Back and forth. When both halves are in a
bridge this creates an infinite loop of frames.
This change makes it so that bridging no longer
allows both of these sides to exist in the same
bridge.
Naveen Albert [Sat, 6 Nov 2021 23:35:16 +0000 (23:35 +0000)]
res_tonedetect: Add call progress tone detection
Makes basic call progress tone detection available
in a tech-agnostic manner with the addition of the
ToneScan application. This can determine if the channel
has encountered a busy signal, SIT tones, dial tone,
modem, fax machine, etc. A few basic async progress
tone detect options are also added to the TONE_DETECT
function.
Alexander Traud [Fri, 12 Nov 2021 18:41:09 +0000 (19:41 +0100)]
progdocs: Avoid 'name' with Doxygen \file.
Fixes four misuses of the parameter 'name'. Additionally, for
consistency and to avoid such an issue in future, those few other
places, which used '\file name', were changed just to '\file'. Then,
Doxygen uses the name of the current file.
Currently, when the t option is specified with no arguments,
the # character is still treated as a terminator, even though
no character should be treated as a terminator.
This is because a previous regression fix was modified to
remove the use of NULL as a default altogether. However,
NULL and an empty string actually refer to different
arrangements and should be treated differently. NULL is the
default terminator (#), while an empty string removes the
terminator altogether. This is the behavior being used by
the rest of the core.
Additionally, since S_OR catches empty strings as well as
NULL (not intended), this is changed to a ternary operator
instead, which fixes the behavior.
Naveen Albert [Sun, 24 Oct 2021 18:38:13 +0000 (18:38 +0000)]
res_pjsip_callerid: Fix OLI parsing
Fix parsing of ANI2/OLI information, since it was previously
parsing the user, when it should have been parsing other_param.
Also improves the parsing by using pjproject native functions
rather than trying to parse the parameters ourselves like
chan_sip did. A previous attempt at this caused a crash, but
this works correctly now.
Naveen Albert [Sun, 24 Oct 2021 11:55:50 +0000 (11:55 +0000)]
sig_analog: Fix truncated buffer copy
Fixes compiler warning caused by a truncated copy of the ANI2 into a
buffer of size 10. This could prevent the null terminator from being
copied if the copy value exceeds the size of the buffer. This increases
the buffer size to 101 to ensure there is no way for truncation to occur.
Sean Bright [Mon, 8 Nov 2021 15:01:09 +0000 (10:01 -0500)]
pbx.c: Don't remove dashes from hints on reload.
When reloading dialplan, hints created dynamically would lose any dash
characters. Now we ignore those dashes if we are dealing with a hint
during a reload.
Naveen Albert [Sun, 24 Oct 2021 12:31:37 +0000 (12:31 +0000)]
app_voicemail: Fix phantom voicemail bug on rerecord
If users are able to press # for options while leaving
a message and then press 3 to rerecord the message, if
the caller hangs up during the rerecord prompt but before
Asterisk starts recording a message, then an "empty"
voicemail gets processed whereby an email gets sent out
notifying the user of a 0:00 duration message. The file
doesn't actually exist, so playback will fail since there
was no message to begin with.
This adds a check after the streaming of the rerecord
announcement to see if the caller has hung up. If so,
we bail out early so that we can clean up properly.
Naveen Albert [Tue, 26 Oct 2021 00:47:02 +0000 (00:47 +0000)]
chan_iax2: Allow both secret and outkey at dial time
Historically, the dial syntax for IAX2 has held that
an outkey (used only for RSA authenticated calls)
and a secret (used only for plain text and MD5 authenticated
calls, historically) were mutually exclusive, and thus
the same position in the dial string was used for both
values.
Now that encryption is possible with RSA authentication,
this poses a limitation, since encryption requires a
secret and RSA authentication requires an outkey. Thus,
the dial syntax is extended so that both a secret and
an outkey can be specified.
The new extended syntax is backwards compatible with the
old syntax. However, a secret can now be specified after
the outkey, or the outkey can be specified after the secret.
This makes it possible to spawn an encrypted RSA authenticated
call without a corresponding peer being predefined in iax.conf.
George Joseph [Thu, 9 Sep 2021 14:39:27 +0000 (08:39 -0600)]
ast_coredumper: Refactor to better find things
The search for a running asterisk when --running is used
has been greatly simplified and in the event it doesn't
work, you can now specify a pid to use on the command
line with --pid.
The search for asterisk modules when --tarball-coredumps
is used has been enhanced to have a better chance of finding
them and in the event it doesn't work, you can now specify
--libdir on the command line to indicate the library directory
where they were installed.
The DATEFORMAT variable was renamed to DATEOPTS and is now
passed to the 'date' utility rather than running DATEFORMAT
as a command.
The coredump and output files are now renamed with DATEOPTS.
This can be disabled by specifying --no-rename.
Several confusing and conflicting options were removed:
--append-coredumps
--conffile
--no-default-search
--tarball-uniqueid
The script was re-structured to make it easier for follow.
projects / thirdparty / asterisk.git / log
