GitLab

This adds OCEndpointParseException that will be thrown
if stringToEndpoint fails to parse the passed in string.

Initially I would have liked to use the java.text.ParseException
but it expects the errorOffset value which is the position where
the parse error occured in the source string. We don't have a
mechanism for discovering the errorOffset so a regular exception
with a error message has been used instead.

Additionally updated the NullPointerException code for setDi
to use the C/C++ abstraction macros for Java method calls.

Change-Id: I76442508c0b5b6c84d8c5ba303f0d982d550e5b6
Signed-off-by: George Nash <george.nash@intel.com>

This commit removes the default constructor generation for
the OCRole class. At this time the user must use
OCObt.addRoleId to create and add roles for the onboarding
tool.

This also prevents SWIG generating setter functions for role
and authority.  These should also be assigned using the
addRoleId method.

Change-Id: I396f5f45724d7f714c035866eece18f079dcd54c
Signed-off-by: George Nash <george.nash@intel.com>

Change-Id: I1e2a868803f68ece5b78655c10c3b5c4784916fa
Signed-off-by: Larry Sachs <larry.j.sachs@intel.com>

If null is passed in for the endpoint or for di when
calling the OCEndpointUtil.setDi method it will now
throw a NullPointerException that will avoid
a segfault from the C code.

Change-Id: I3f29486c21e08f3718752efbd108ce0401defa24
Signed-off-by: George Nash <george.nash@intel.com>

If initPut or initPost methods fail this will release
the jni_sync_lock before returning to prevent the code
from deadlocking.

Change-Id: Ie85c9eee3a4e4649f4af8b960264a8d25955adfe
Signed-off-by: George Nash <george.nash@intel.com>

This exposes most of the data types in the new oc_acl_common.h
header to Java.

Updated cloud certification app and onboarding tools to match
what is found in the C version of the tools.

Change-Id: I413abb229749d93a1bfe97673ab51af7b61fb997
Signed-off-by: George Nash <george.nash@intel.com>

Change-Id: If3f43cace602572a89256904c7dd7296414557fc
Signed-off-by: George Nash <george.nash@intel.com>

Add the code to map oc_resource_set_properties_cbs
function to the Java resourceSetPropertiesHandlers method

Also add the two callbacks handler classes for
OCGetPropertiesHandler and OCSetPropertiesHandler

Some cleanup code for
  - OCCollections
  - OCCloudContex
and others

The structs oc_link_params_t and oc_rt_t were ignored in the
swig output. No public functions that take or return these
structs could be found.

Change-Id: Iebb1ca8387ff119623af1765a5c0717a34ecd2bd
Signed-off-by: George Nash <george.nash@intel.com>

This extends the endpoint struct for swig so the generated
code will call oc_new_endpoint and oc_free_endpoint when
new and finalize are called.

Endpoint(s) are set to NULL after calling oc_free_endpoint or
oc_free_server_endpoints is called from Java. This will prevent
double freeing memory which would result in a segfault.

code that returns or accept multiple endpoints in a list now
indicate that fact because the variable name is plural
'endpoints'

Change-Id: Id11a1c98e724bf44b7ab82e92a529f3859f6e936
Signed-off-by: George Nash <george.nash@intel.com>

The Java stringToEndpoint method now returns an OCEndpoint

In addition a collection of unit tests were added to verify
the code.

Change-Id: I6040d9b48ca214d83daf49a59f6b2ba8ff2903dd
Signed-off-by: George Nash <george.nash@intel.com>

There was an issue in the generated code. It was supposed
to automatically find the length of the pin without the user
needing to pass it as a sepparate variable. There was a small
typo in the swig typemap which caused the code to ask for the
pin length.

Change-Id: If3f3df8dd158fdb3b0cee3270bb98cc98f3cec79
Signed-off-by: George Nash <george.nash@intel.com>

The enum was renamed but the name change was missed
in the interface file.

Change-Id: I1ae294be4813f8cc38130ab63c4978c51c8f7b15
Signed-off-by: George Nash <george.nash@intel.com>

This expands the documentation to generate documentation
for the OCRep class when swig is run.

This also cleans up the some of documentation for the OCMain
class as well. Mostly whitespace cleanup.

In The process of developing the documentation the clearCborErrno
was added since the documentation process exposed the fact that
there was no way to clear the CborErrno from Java.

The Java function getRepError was removed it was returning the
same information as getCborErrno.

Change-Id: Iee237168c122f3d851b19c1cf967de304ab26147
Signed-off-by: George Nash <george.nash@intel.com>

Change-Id: I529e32d4f104e8d6e088d29f8e5da4ce986aeaa5
Signed-off-by: George Nash <george.nash@intel.com>

This change allows us to have a consistant Java JAR
API regardless of the SECURE build options.

Change-Id: I70061fa2bf9d06dc59f16c764d652919095f6b91
Signed-off-by: George Nash <george.nash@intel.com>

Add code to release the global ref after a callback will no
longer be called.

Change-Id: I1da2c96f0a672d7897473ef7391c99732c6b1e66
Signed-off-by: George Nash <george.nash@intel.com>

When built with SECURE=0 the following functions
implementation is ifdef'ed out and will cause an
UnsatisfiedLinkError when the functions are called
from a Java progeam.
 - oc_reset
 - oc_core_regen_unique_ids
 - oc_auto_assert_roles
 - oc_assert_all_roles

We call a dummy function that will just call the C function
if SECURED=0 it will do nothing and return.

When building SECURE=0 all of the methods from oc_obt.h
are excluded from the build. To make this possible some
more selective logic was added to the swig Makefile
to not process oc_obt.i and not copy files from the
oc folder that reference generated code from oc_obt.i.

Change-Id: I82eddbc894df67a51127929eecf650c98b115b4f
Signed-off-by: George Nash <george.nash@intel.com>

SWIG currently does not have generic way to generate
documentation in the output so this use hacks to generate
the required documentation.

This copies over the documentation from oc_api.h to the
Java output.

Change-Id: Id80acf08a0035517b57a83ca322645ea98914046
Signed-off-by: George Nash <george.nash@intel.com>

Two of the wraper functions were not returning their
return values.  This fixes that failure.

Change-Id: I22e97c81f97ffa556a51db7bfa350dda66e472fa
Signed-off-by: George Nash <george.nash@intel.com>

Change-Id: I8edc2a928fbcfd4954b17e2128aaa1c1ed38e442
Signed-off-by: George Nash <george.nash@intel.com>

Change-Id: Ie4be7e6ae32feac13bf106e89039889f8c9ba28f
Signed-off-by: George Nash <george.nash@intel.com>

This includes a fix for the windows build.

Change-Id: If2878c888a93752b36627bc6162c55c8db77be57
Signed-off-by: George Nash <george.nash@intel.com>

This fixes issues when building without the security build option

The introspection caused build issues because variables were not used

the security build option was not being exported to the SWIG make file

SWIG make file not checks for SECURE build option. If SECURE is not
explicity set to zero it defaults to build with security.

Change-Id: I626fdaf890e8b0fd953ad452d10ae9ebf3073ac0
Signed-off-by: George Nash <george.nash@intel.com>

Break oc_iotivity_lite_jni.h into header and code file,
this helps solve many build issues that resulted from not
doing this.

Moved the JNI_onload function from the storage code to
this new location. This makes it possible to store pointer to
the JavaVM as soon as the library is loaded. This helps remove
many of the workarounds that were added to the code to grab the
JavaVM pointer.

All of the Java jclass varaible are now loaded in the JNI_onload
function. This insures that the classes are avalible to the jni
code instantly.  No need to make sure the OCMain.init function
is called before using a Java class in the jni code.

Added some functions to manage the jni_callbacks list. Insuring
all jni_callback_data is managed and stored in this central location.

Change-Id: I426cb39e9a233cbddd754011b3a3939a72df3257
Signed-off-by: George Nash <george.nash@intel.com>

Found a few locations that still were not outputing
camelCase which is what we expect in Java.

Change-Id: I8987c806afb9e29f76abdb0cf315d93d5d836155
Signed-off-by: George Nash <george.nash@intel.com>

The toString function now returns a String instead
of having to pass in an array of size 1 which is
unusual for Java.

Change-Id: I30ddd6796277f1bd0cdb29174072d83c7437b246
Signed-off-by: George Nash <george.nash@intel.com>

Trying to keep the names of the Java handlers as close to the
C source code names.

Change-Id: I4125e4a6cb19549139eeac7ef0bc4ff55c8ee500
Signed-off-by: George Nash <george.nash@intel.com>

This Exposes oc_cloud.h and oc_session_events.h to
to the Java language bindings using swig.

A lot of ifdefs to handle cases when OC_CLOUD and OC_TCP
are not defined when the code is compiled.

Change-Id: I42340f6d71159d6d12ce7119bbe01142fc3f2c10
Signed-off-by: George Nash <george.nash@intel.com>

This adds the following functions to the swig generated java:
from oc_api.h:
 - oc_do_site_local_ipv6_discovery
 - oc_do_realm_local_ipv6_discovery
 - oc_do_realm_local_ipv6_multicast
 - oc_do_site_local_ipv6_multicast
from oc_obt.h:
 - oc_obt_discover_unowned_devices_site_local_ipv6
 - oc_obt_discover_unowned_devices_realm_local_ipv6
 - oc_obt_discover_owned_devices_site_local_ipv6
 - oc_obt_discover_owned_devices_realm_local_ipv6
ignore functions from oc_client_state.h:
 - oc_ri_free_client_cbs_by_endpoint
 - oc_ri_fre_client_cbs_by_mid

Change-Id: If8ce5c91bca88891c65ed20f997dc05e36a6fdc0
Signed-off-by: George Nash <george.nash@intel.com>

- swig/Makefile was using jni.h from Linux not Android. Although the
  jni.h files are very similar they are not the same resulting in
  different build warnings/errors.
- oc_introspection_wrap was using logging but did not include oc_log.h
- Fixed call to AttachCurrentThread when jni.h switched to Android variant.
- Add some build flags to reduce the build warnings from swig generated code.

Change-Id: Ibb25db39be0056dbea45863bea4c472e157ae423
Signed-off-by: George Nash <george.nash@intel.com>

This reworks that oc_ri.i interface to pull in the contents of
oc_ri.h from an %include statement. This has the advantage that
we don't have copies of large structs that must match between.
It also means going through and updating the ignore list since
we are not exposing any of the functions from oc_ri.h to Java
just structs and enums.

Fixed an error in how oc_core_add_new_device was exposed to Java
in oc_core_res.i

Swig does not understand the OC_LIST_STRUCT macro. Ignore struct
members that use that macro. May need to revisit those members
in the future for now there are no known instances that the
end user will access the member variables.

testGetQueryValues disabled till we can find a way to set the
query value using the framework not setting directly which
causes memory leaks.

Change-Id: I873b30fbc4420221c965a35e499dbe1acf5a8716
Signed-off-by: George Nash <george.nash@intel.com>

This adds some ifdefs to the that handles situation the
IDD build option is not specified so the build will not
fail do to a linkter issue.

At the same time I moved the code in oc_storage.i so
it was inline. This reduces the need to repeate the
function name so is is picked up by swig.

Change-Id: I5ac7a6f8c476fba441b6b6128d329964c5b2da8c
Signed-off-by: George Nash <george.nash@intel.com>

Add new swig interfaces to use C API from Java.
Add Makefile to build the swig module.

Change-Id: I68fd5f23960c58e550aed934630806c6e5bc2d71
Signed-off-by: Javier Guerra Melgares <javier.guerra@dekra.com>
Reviewed-by: Larry Sachs <larry.j.sachs@intel.com>
Reviewed-by: Rick Bell <richard.s.bell@intel.com>
Reviewed-by: George Nash <george.nash@intel.com>
Reviewed-by: Kishen Maloor <kishen.maloor@intel.com>

This exposes most of oc_core_res.h to the Java language. With a few
exceptions.

Functions oc_store_uri and oc_core_populate_resource are not exposed
yet because due to handling of callbacks and oc_string_t pointers.

The oc_platform_info_t struct was named not just typedefed
The oc_device_infor_t struct was named not just typedefed

Structs were given names so the swig scripts could ignore exposing
parts of the structs that are can not be used in Java.

Change-Id: I29e3461ea925d001df2769b23db5e5ff49a53a5b
Signed-off-by: George Nash <george.nash@intel.com>

This fixes the callback versions of the addDevice
and initPlatform functions.  I think I inadvertantly
removed the version of the functions that exposed
the callback in a past cleanup  this restors the verson
of the functions that calls the handler.

Change-Id: Iccab8ef4ca22761df440bae708563fc5aaa1fcd2
Signed-off-by: George Nash <george.nash@intel.com>

This will update oc_endpoint.i so it no longer generates
SWIGTYPEs in its output.

Change-Id: I19afcec7e8829b2d128afe047c9ebcd54ec05e9e
Signed-off-by: George Nash <george.nash@intel.com>

Change-Id: I4d01be4bd54790017d0bb8f8d8a891cb794b6723
Signed-off-by: George Nash <george.nash@intel.com>

Also add it to the java implementation of the obt.

Change-Id: I45b92f1cd031981bc243914a96d12a564aafb870
Signed-off-by: George Nash <george.nash@intel.com>
Reviewed-on: https://gerrit.iotivity.org/gerrit/29467Tested-by: IoTivity Jenkins <jenkins-daemon@iotivity.org>
Reviewed-by: Larry Sachs <larry.j.sachs@intel.com>
Reviewed-by: Rick Bell <richard.s.bell@intel.com>

This reverts commit 4def6e82.

Some how this commit made it into the branch without
going through the full review process. As indicated by the
lack of Reviewed-by tag.

Change-Id: Ibcc970b8882b1874aa82ae7be964169a2f207485
Signed-off-by: George Nash <george.nash@intel.com>
Reviewed-on: https://gerrit.iotivity.org/gerrit/29466Tested-by: IoTivity Jenkins <jenkins-daemon@iotivity.org>
Reviewed-by: Kishen Maloor <kishen.maloor@intel.com>

Add new swig interfaces to use C API from Java.
Add Makefile to build the swig module.

Change-Id: I55d8af5ca5ad73e3bb89f21824f74cffa8599ed0
Signed-off-by: Javier Guerra Melgares <javier.guerra@dekra.com>