Mark Michelson [Tue, 16 Feb 2010 18:29:42 +0000 (18:29 +0000)]
Add unit test for dialplan pattern matching.
This test works by reading input from arrays to build a sample
dialplan. From there, patterns are attempted to be matched against
said dialplan, with the expected match given. We then search in our
example dialplan to see if we find a match and if what we find matches
what we expected it to match.
David Vossel [Tue, 16 Feb 2010 17:07:41 +0000 (17:07 +0000)]
fixes sample rate conversion issue with Monitor application
When using ast_seekstream with the read/write streams of a monitor,
the number of samples we are seeking must be of the same rate as the
stream or the jump calculation will be incorrect. This patch adds logic
to correctly convert the number of samples to jump to the sample rate
the read/write stream is using.
For example, if the call is G722 (16khz) and the read/write stream is
recording a 8khz wav, seeking 320 samples of 16khz audio is not the
same as seeking 320 samples of 8khz audio when performing the ast_seekstream
on the stream.
Tilghman Lesher [Tue, 16 Feb 2010 00:52:45 +0000 (00:52 +0000)]
Change the blanket rules to delete .lastclean on all CFLAGS menuselect targets to be more particular.
This change builds upon the recent change to menuselect to add 'touch_on_change'
as an attribute of both categories and members. This should allow only the most
invasive defines to cause a complete rebuild, while defines which only affect a
subset of modules will only cause a rebuild of that smaller set.
David Vossel [Mon, 15 Feb 2010 15:45:02 +0000 (15:45 +0000)]
chan_sip parse code refactoring plus two new unit tests
Code Refactoring Changes
- read_to_parts() moved to reqresp_parser.c and has been renamed as
get_name_and_number()
- get_in_brackets() moved to reqresp_parser.c
- find_closing_quotes() added to sip_utils.h
Logic Changes
- get_name_and_number() now uses parse_uri() and get_calleridname()
for parsing. Before this change only names within quotes were
found, when names not within quotes are possible.
New Unit Tests
-sip_get_name_and_number_test
-sip_get_in_brackets_test
On channel destruction the channel's datastores are removed and
destroyed. Since there are public API calls to find and remove
datastores on a channel, a lock should be held whenever datastores are
removed and destroyed. This resolves a crash caused by a race
condition in app_chanspy.c.
(closes issue #16678)
Reported by: tim_ringenbach
Patches:
datastore_destroy_race.diff uploaded by tim ringenbach (license 540)
Tested by: dvossel
........
David Vossel [Fri, 12 Feb 2010 17:59:39 +0000 (17:59 +0000)]
fixes areas where port should be removed from domain during parsing
A patch was committed recently that converted duplicate uri parsing
code to use the parse_uri function. There were two instances where
this conversion did not mimic previous behavior exactly because the
port was not being parsed off the end of the domain. In order to do
this, a dummy pointer argument needs to be passed into parse_uri so
it will know it must parse out the port from the domain. If a port
output paramenter is not present, the domain is returned with the
port still attached.
David Vossel [Thu, 11 Feb 2010 18:42:25 +0000 (18:42 +0000)]
astobj2 unit test and bug fix
A bug was discovered during the creation of the astobj2 unit test.
When OBJ_MULTIPLE | OBJ_UNLINK is used, the objects being returned
had a ref count issue. This patch resolves that.
Russell Bryant [Wed, 10 Feb 2010 23:19:16 +0000 (23:19 +0000)]
Add a test module for the event API, test_event.c.
This module includes a single test so far that creates events using two
different methods and does some verification on the result to make sure
the correct data can be retrieved from the event that was created.
One bug was found in the event API while developing this test, which makes
me happy. :-)
Jeff Peeler [Wed, 10 Feb 2010 16:47:37 +0000 (16:47 +0000)]
Change channel state on local channels for busy,answer,ring.
Previously local channels channel state never changed. This became problematic
when the state of the other side of the local channel was lost, for example
during a masquerade. Changing the state of the local channel allows for the
scenario to be detected when the channel state is set to ringing, but the peer
isn't ringing. The specific problem scenario is described in 164201. Although
this was noted on one of the issues, here is the tested dialplan verified to
work:
Tilghman Lesher [Wed, 10 Feb 2010 16:01:28 +0000 (16:01 +0000)]
Solaris doesn't like outputting a NULL to a %s in format strings.
Detect all platforms that don't like that, either, and ensure that when documentation is
missing, we pass a non-NULL pointer when outputting the corresponding documentation.
Russell Bryant [Tue, 9 Feb 2010 23:32:14 +0000 (23:32 +0000)]
Various updates to the unit test API.
1) It occurred to me that the difference in usage between the error ast_str and
the ast_test_update_status() usage has turned out to be a bit ambiguous in
practice. In a lot of cases, the same message was being sent to both.
In other cases, it was only sent to one or the other. My opinion now is that
in every case, I think it makes sense to do both; we should output it to the
CLI as well as save it off for logging purposes.
This change results in most of the changes in this diff, since it required
changes to all existing unit tests. It also allowed for some simplifications
of unit test API implementation code.
2) Update ast_test_status_update() to include the file, function, and line
number for the code providing the update.
3) There are some formatting tweaks here and there. Hopefully they aren't too
distracting for code review purposes. Reviewboard's diff viewer seems to do a
pretty good job of pointing out when something is a whitespace change.
4) I moved the md5_test and sha1_test into the test_utils module. It seemed
like a better approach since these tests are so tiny.
5) I changed the number of nodes used in heap_test_2 from 1 million to
100 thousand. The only reason for this was to reduce the time it took
for this test to run.
6) Remove an unused function prototype that was at the bottom of utils.h.
7) Simplify test_insert() using the LIST_INSERT_SORTALPHA() macro. The one
minor difference in behavior is that it no longer checks for a test registered
with the same name.
8) Expand the code in test_alloc() to provide specific error messages for each
failure case, to clearly inform developers if they forget to set the name,
summary, description, etc.
9) Tweak the output of the "test show registered" CLI command. I swapped the
name and category to have the category first. It seemed more natural since
that is the sort key.
10) Don't output the status ast_str in the "test show results" CLI command.
This is going to tend to be pretty verbose, so just leave that for the
detailed test logs (test generate results).
2^15 = 32768 which is the maximum allowed iax2 callnumber.
Creating the iaxs and iaxsl array of size 32768 means the maximum
callnumber is actually out of bounds. This causes a nasty crash.
Don't offer MMR or JBIG transcoding during T.38 negotiation.
After further discussion with Steve Underwood, we should not (yet) be offering
to receive MMR or JBIG transcoded streams from T.38 endpoints. A future spandsp
release will support those features, and then they can be enabled during
negotiation
Tilghman Lesher [Mon, 8 Feb 2010 22:31:40 +0000 (22:31 +0000)]
Actually use _ASTLDFLAGS in the main/ and channels/ Makefiles.
They were previously passed correctly, but they simply weren't used. This
caused issues with various platforms whose builds needed to pass special
linker flags via the configure script.
Russell Bryant [Mon, 8 Feb 2010 04:43:55 +0000 (04:43 +0000)]
Add a todo for pbx_gtkconsole for updating to gtk2.
This module needs to be converted to gtk2, or we will eventually have to just
remove it from the tree. gtk1 isn't even packaged anymore in the distro I'm
using. I suspect nobody uses this and that nobody would notice if we removed
it.
Mark Michelson [Sat, 6 Feb 2010 14:43:03 +0000 (14:43 +0000)]
Remove useless sip options related to hash table size.
First off, these options weren't actually doing anything.
By the time the options were parsed, the peer and dialog
containers had already been allocated with their default
values.
Second, hash table size is something that doesn't really
make sense to change in a config file. If a user is that
interested in changing the hashtable size, he can modify
the source itself.
I have removed the parsing of the hash_peer, hash_user,
and hash_dialog options. I have removed the hash_user_size
variable altogether since it is not used at all. I also
changed hash_peer_size and hash_dialog_size to be constant,
and have changed the symbols to be in all caps as constants
typically are. I have also removed the entire section in
sip.conf.sample regarding configurable hashtable sizes.
Change channel state on local channels for busy,answer,ring.
Previously local channels channel state never changed. This became problematic
when the state of the other side of the local channel was lost, for example
during a masquerade. Changing the state of the local channel allows for the
scenario to be detected when the channel state is set to ringing, but the peer
isn't ringing. The specific problem scenario is described in 164201. Although
this was noted on one of the issues, here is the tested dialplan verified to
work:
David Vossel [Thu, 4 Feb 2010 15:36:33 +0000 (15:36 +0000)]
fix truncated format string in 'test show registered'
When using the 'test show registered' cli command the 'Test Results'
category was truncating the last few characters making it look like
'Test Resul'. I also expanded other parts of the format to better
represent how long function names and categories will likely be.
Jeff Peeler [Wed, 3 Feb 2010 20:48:36 +0000 (20:48 +0000)]
Add some additional option support for non-default parking lots.
The options are: parkedcallparking, parkedcallhangup, parkedcallrecording, and
parkedcalltransfers. Previously these options were only available for the
default parking lot.
David Vossel [Wed, 3 Feb 2010 20:33:32 +0000 (20:33 +0000)]
-----Changes -----
New files
- channels/sip/sip.h – A new header for shared #define, enum, and struct
definitions.
- channels/sip/include/sip_utils.h – sip util functions shared among
the all the sip APIs
- channels/sip/include/config_parser.h – sip config-parser API
- channels/sip/config_parser.c – Contains sip.conf parsing helper functions
with unit tests.
- channels/sip/include/reqresp_parser.h – sip request response parser API
- channels/sip/reqresp_parser.c – Contains sip request and response parsing
helper functions with unit tests.
New Unit Tests
- sip_parse_uri_test
- sip_parse_host_test
- sip_parse_register_line_test
Code Refactoring
- All reusable #define, enum, and struct definitions were moved out of chan_sip.c
into sip.h. During this process formatting changes were made to comments
in both sip.h and chan_sip.c in order to better adhere to the coding guidelines.
- The beginnings of three new sip APIs, sip-utils.h, config-parser.h,
reqresp-parser.h using existing chan_sip.c functions.
- parse_uri() and get_calleridname() were moved from chan_sip.c to request-parser.c
along with unit tests for both functions.
- sip_parse_host() and sip_parse_register_line() were moved from chan_sip.c to
config-parser.c along with unit tests for both functions.
Changes to parse_uri()
-removal of the options parameter. It was never used and did not behave correctly.
-additional check for [?header] field. When this field was present, the transport
type was not being set correctly.
----- Overview -----
This patch is introduced with the hope that unit tests for all our sip parsing
functions will be written soon. chan_sip is a huge file, and with the addition of
each unit test chan_sip is going to grow larger and harder to maintain. I'm proposing
we begin refactoring chan_sip, starting with the parsing functions. With each parsing
function we move into a separate helper file, a unit test should accompany it. I've
attempted to lay down the ground work for this change by creating two new parser
helper files (config-parser.c and reqresp-parser.c) and moving all shared structs,
enums, and defines from chan_sip.c into a shared sip.h file. We can't verify everything
in Asterisk using unit tests, but string parsing is one area where unit tests make
the most sense. By beginning to restructure the code in this way, chan_sip not only
becomes less bloated, but Asterisk as a whole will become more stable.
Tilghman Lesher [Tue, 2 Feb 2010 20:32:29 +0000 (20:32 +0000)]
Properly respect GOSUB_RESULT as to what to do with the master channel.
Previously, we would parse GOSUB_RESULT, but not actually do anything with it.
Also, allow GOSUB_RETVAL to be inherited back across a peer/master channel.
Backup and restore original textfile, for prosthesis (gerund of prepend).
Also, fix menuselect such that changing voicemail build options correctly
causes rebuild.
(closes issue #16415)
Reported by: tomo1657
Patches:
prepention.patch uploaded by tomo1657 (license 484)
(with modifications by me to backport to 1.4)
........
Russell Bryant [Wed, 27 Jan 2010 17:32:07 +0000 (17:32 +0000)]
Fix the ability to specify an OSP token for an outbound IAX2 call.
When this patch was originally submitted, the code allowed for the token to be
set via a channel variable. I decided that a cleaner approach would be to
integrate it into the CHANNEL() function. Unfortunately, that is not a suitable
approach. It's not possible to get the value set on the channel soon enough
using that method. So, go back to the simple channel variable method.
Jeff Peeler [Tue, 26 Jan 2010 18:07:57 +0000 (18:07 +0000)]
Fix crash resulting from frames with invalid data pointers.
In ast_frdup the frame data union does not get set to point to malloced memory
if the datalen is zero, so make sure to handle the same case in ast_frisolate
appropriately.
David Vossel [Tue, 26 Jan 2010 17:40:02 +0000 (17:40 +0000)]
modify 'test show registered' cli output format
In order to improve readability, the output from 'test show
registered' has been modified to truncate fields to fit within
the format output if they are over a certain length.
David Vossel [Tue, 26 Jan 2010 16:30:08 +0000 (16:30 +0000)]
RFC compliant uri and display-name encode/decode
1. URI Encoding
This patch changes ast_uri_encode()'s behavior when doreserved is enabled.
Previously when doreserved was enabled only a small set of reserved
characters were encoded. This set was comprised primarily of the reserved
characters defined in RFC3261 section 25.1, but contained other characters as
well. Rather than only escaping the reserved set, doreserved now escapes
all characters not within the unreserved set as defined by RFC 3261 and
RFC 2396. Also, the 'doreserved' variable has been renamed to 'do_special_char'
in attempts to avoid confusion.
When doreserve is not enabled, the previous logic of only encoding the
characters <= 0X1F and > 0X7f remains, except for the '%' character, which
must always be encoded as it signifies a HEX escaped character during the decode
process.
2. URI Decoding: Break up URI before decode.
In chan_sip.c ast_uri_decode is called on the entire URI instead of it's
individual parts after it is parsed. This is not good as ast_uri_decode
can introduce special characters back into the URI which can mess up parsing.
This patch resolves this by not decoding a URI until parsing is completely
done. There are many instances where we check to see if pedantic checking
is enabled before we decode a URI. In these cases a new macro,
SIP_PEDANTIC_DECODE, is used on the individual parsed segments of the URI
rather than constantly putting if (pedantic) { decode() } checks everywhere
in the code. In the areas where ast_uri_decode is not dependent upon
pedantic checking this macro is not used, but decoding is still moved to
each individual part of the URI. The only behavior that should change from
this patch is the time at which decoding occurs.
Since I had to look over every place URI parsing occurs to create this
patch, I found several places where we use duplicate code for parsing.
To consolidate the code, those areas have updated to use the parse_uri()
function where possible.
3. SIP display-name decoding according to RFC3261 section 25.
To properly decode the display-name portion of a FROM header, chan_sip's
get_calleridname() function required a complete re-write. More information
about this change can be found in the comments at the beginning of this function.
4. Unit Tests.
Unit tests for ast_uri_encode, ast_uri_decode, and get_calleridname() have been
written. This involved the addition of the test_utils.c file for testing the
utils api.
projects / thirdparty / asterisk.git / log
