Python module for Enduro/X offers complete ATMI API access from the Python3 programming language. The module includes such features as:
- A multi-threaded server
- Synchronous, asynchronous, conversational, event-based, and notification IPC APIs
- Support for UBF (emulation of Tuxedo FML32, i.e. binary coded key/value buffer), NULL, STRING, CARRAY, VIEW (C structures), and JSON buffers
- Support for nested UBF buffers with PTR (pointer to other XATMI buffers) and VIEW support
- Dictionary-like interface for UBF, having zero marshaling when client/server sends/receives UBF buffers, meaning ultra-fast processing times, even for large data. When buffers are exchanged between processes, only a few memcpy() operations are performed between processes and the OS kernel, as for UBF there is no serialization and when using UbfDict() interface, only requested fields from memory are exchanged with the Python VM upon get/set access.
- Receive a response even when the service returns TPFAIL (instead of exception)
- Enduro/X ATMI server extensions, such as periodic callbacks and resource polling
- Logging API.
- Simple, yet powerful Python to Executable linker tool (named expyld), which pre-compiles and embeds pure Python packages/modules into executable files. During the runtime these packages/modules are resolved directly from the process memory, i.e. no temp files are used.
- See documentation for full API description.
Python module for Enduro/X is written in C++11 and pybind11.
The module is supported by all platforms where Enduro/X runs, Python3 and C++11 are available.
The following packages must be installed before installing the Enduro/X Python module from sources:
- Enduro/X Core (binary downloads here https://www.mavimax.com) starting from version 8.0.6
- pkg-config
- CMake version 3.1 or later
To build and install endurox, clone or download this repository and then, from within the repository, run:
$ python3 ./setup.py install --user
Or if pip3 is available, install it in the following way:
$ pip3 install . --user
The unit tests can be started with:
$ python3 ./setup.py test
To build API documentation:
$ sudo pip3 intall sphinx
$ cd doc
$ make html
$ cd _build/html
-- launch the browser at index (to view the docs):
$ firefox index.html
If files binary release is required, prepare by:
$ sudo pip3 install wheel
$ python3 ./setup.py bdist_wheel
To prepare source distribution:
$ python3 ./setup.py sdist
The distribution files are unloaded to ./dist folder.
endurox
module supports all Enduro/X ATMI buffer types:
STRING
CARRAY
UBF
VIEW
JSON
NULL
All buffers are encapsulated in Python dictionary. For example UBF
(equivalent to Tuxedo FML32) buffer is encoded as:
{
'buftype': 'UBF'
, 'data':
{
'T_SHORT_FLD': [3200]
, 'T_LONG_FLD': [99999111]
, 'T_CHAR_FLD': ['X', 'Y', b'\x00']
, 'T_FLOAT_FLD': [1000.989990234375]
, 'T_DOUBLE_FLD': [1000111.99]
, 'T_STRING_FLD': ['HELLO INPUT']
, 'T_PTR_FLD': [{'buftype': 'STRING', 'data': 'HELLO WORLD'}]
, 'T_UBF_FLD': [{'T_SHORT_FLD': [99], 'T_UBF_FLD': [{'T_LONG_2_FLD': [1000091]}]}]
, 'T_VIEW_FLD': [{}, {'vname': 'UBTESTVIEW2', 'data': {
'tshort1': [5]
, 'tlong1': [100000]
, 'tchar1': ['J']
, 'tfloat1': [9999.900390625]
, 'tdouble1': [11119999.9]
, 'tstring1': ['HELLO VIEW']
, 'tcarray1': [b'\x00\x00', b'\x01\x01']
}}]
}
}
Or with UBF Dictionary interface class:
{
'buftype': 'UBF'
, 'data': e.UbfDict(
{
'T_SHORT_FLD': [3200]
, 'T_LONG_FLD': [99999111]
, 'T_CHAR_FLD': ['X', 'Y', b'\x00']
, 'T_FLOAT_FLD': [1000.989990234375]
, 'T_DOUBLE_FLD': [1000111.99]
, 'T_STRING_FLD': ['HELLO INPUT']
, 'T_PTR_FLD': [{'buftype': 'STRING', 'data': 'HELLO WORLD'}]
, 'T_UBF_FLD': [{'T_SHORT_FLD': [99], 'T_UBF_FLD': [{'T_LONG_2_FLD': [1000091]}]}]
, 'T_VIEW_FLD': [{}, {'vname': 'UBTESTVIEW2', 'data': {
'tshort1': [5]
, 'tlong1': [100000]
, 'tchar1': ['J']
, 'tfloat1': [9999.900390625]
, 'tdouble1': [11119999.9]
, 'tstring1': ['HELLO VIEW']
, 'tcarray1': [b'\x00\x00', b'\x01\x01']
}}]
})
}
buftype
is optional for CARRAY
, STRING
, UBF
and NULL
buffers. It is mandatory for JSON
and VIEW
buffers. For VIEW
buffers subtype
specifies view name.
Buffer data is present in data
root dictionary key.
CARRAY
is mapped to/from Python bytes
type.
STRING
is mapped to/from Python str
type.
UBF
(a FML32
Tuxedo emulation) is mapped to/from Python dict
type with field names
(str
) as keys and lists (list
) of different types (int
, str
, float
or dict
(for embedded BFLD_UBF
, BLFD_PTR
or BFLD_VIEW
) as values. This is default type for the
dict
buffer if for root dictionary buftype
key is not specified. dict
to UBF
conversion also treats types int
, str
, float
or dict
as lists with a
single element (the same rule applies to VIEW
buffer keys):
{'data':{'T_STRING_FLD': 'Single value'}}
converted to UBF
and then back to dict
becomes
{'data':{'T_STRING_FLD': ['Single value']}}
All ATMI functions that take buffer and length arguments in C take only buffer argument in Python.
endurox.tpcall()
and endurox.tpgetrply()
functions return a tuple with 3
elements or throw an exception when no data is received. In case if service returned
TPFAIL
status, the error is not thrown, but instead error code
endurox.TPESVCFAIL
is returned in first return value.
For all other errors, AtmiException
is thrown.
endurox.tpcall()
and endurox.tpgetrply()
returns following values:
- 0 or
TPESVCFAIL
tpurcode
(the second argument totpreturn
)- data buffer
import endurox
tperrno, tpurcode, data = endurox.tpcall('TESTSV', {'data':{'T_STRING_FLD': 'HELLO', 'T_STRING_4_FLD': 'WORLD'}})
if tperrno == 0:
# Service returned TPSUCCESS
else:
# tperrno == endurox.TPESVCFAIL
# Service returned TPFAIL
Enduro/X servers are written as Python classes. tpsvrinit
method of object will be
called when Enduro/X calls tpsvrinit()
function and it must return 0 on success
or -1 on error. A common task for tpsvrinit
is to advertise services the server
provides by calling endurox.tpadvertise()
with a service name. Function accepts
service name (string), service function name (string) and callback to service function.
tpsvrdone
, tpsvrthrinit
and tpsvrthrdone
will be called when Enduro/X calls
corresponding functions. All of these 4 methods are optional.
Each service method receives a single argument with incoming buffer and service must end
with either call to endurox.tpreturn()
or endurox.tpforward()
, however
some non ATMI code may be executed after these function calls. Service function
return may be written in following ways:
def ECHO(self, args):
return t.tpreturn(t.TPSUCCESS, 0, args.data)
def ECHO(self, args):
t.tpreturn(t.TPSUCCESS, 0, args.data)
To start Enduro/X ATMI server process endurox.run()
must be called with an instance of the class and command-line arguments.
#!/usr/bin/env python3
import sys
import endurox as e
class Server:
def tpsvrinit(self, args):
e.tpadvertise('TESTSV', 'TESTSV', self.TESTSV)
return 0
def tpsvrthrinit(self, args):
return 0
def tpsvrthrdone(self):
pass
def tpsvrdone(self):
pass
def TESTSV(self, args):
e.tplogprintubf(e.log_info, 'Incoming request', args.data)
args.data['data']['T_STRING_2_FLD']='Hello World from XATMI server'
return e.tpreturn(e.TPSUCCESS, 0, args.data)
if __name__ == '__main__':
e.tprun(Server(), sys.argv)
To use Python code as Enduro/X server the file itself must be executable (chmod +x *.py
)
and it must contain shebang line with Python:
#!/usr/bin/env python3
After that you can use the *.py
file as server executable in UBBCONFIG
:
<server name="testsv.py">
<min>1</min>
<max>1</max>
<srvid>200</srvid>
<sysopt>-e ${NDRX_ULOG}/testsv.log -r</sysopt>
</server>
Nothing special is needed to implement Enduro/X clients, just import the module and start calling XATMI functions.
#!/usr/bin/env python3
import endurox as e
tperrno, tpurcode, data = e.tpcall('TESTSV', {'data':{'T_STRING_FLD': 'HELLO', 'T_STRING_4_FLD': 'WORLD'}})
e.tplog_info("tperrno=%d tpurcode=%d" % (tperrno, tpurcode))
e.tplogprintubf(e.log_info, 'Got response', data)
# would print to log file:
# t:USER:4:c9e5ad48:413519:7f35b9ad7740:001:20220619:233508180644:tplog :/tplog.c:0582:tperrno=0 tpurcode=0
# t:USER:4:c9e5ad48:413519:7f35b9ad7740:001:20220619:233518812671:plogprintubf:bf/ubf.c:1790:Got response
# T_STRING_FLD HELLO
# T_STRING_2_FLD Hello World from XATMI server
# T_STRING_4_FLD WORLD
You can access Oracle database with cx_Oracle
library and local transactions by just
following the documentation of cx_Oracle
.
If client or server needs to be written in Python to participate in the global transaction, standard Enduro/X Oracle XA driver configuration is be applied, i.e. libndrxxaoras (for static registration) or libndrxxaorad (for dynamic XA registration) configured.
Transactions can be started and committed or aborted by using endurox.tpbegin()
, endurox.tpcommit()
, endurox.tpabort()
.
Client process example:
#!/usr/bin/env python3
import cx_Oracle
import endurox as e
e.tpopen()
db = cx_Oracle.connect(handle=e.xaoSvcCtx())
e.tpbegin(60)
with db.cursor() as cursor:
cursor.execute("delete from pyaccounts")
# Call any service in global transaction
e.tpcommit()
e.tpclose()
e.tpterm()
When running the Enduro/X client which must participate in global transaction, CC tag shall be set in the environment before running the client script:
$ NDRX_CCTAG="ORA1" ./dbclient.py
When running ATMI server in the global transaction, the <cctag> XML tag can be used to assign the DB configuration to it:
<server name="dbserver.py">
<min>1</min>
<max>1</max>
<srvid>200</srvid>
<sysopt>-e ${NDRX_ULOG}/dbserver.log -r</sysopt>
<cctag>ORA1</cctag>
</server>
For a multi-threaded server, new connections for each thread must be created in
tpsvrthrinit()
(instead of tpsvrinit()
) and stored in thread-local storage of threading.local()
.
app.ini settings for the Oracle DB:
[@global/ORA1] NDRX_XA_RES_ID=1 NDRX_XA_OPEN_STR=ORACLE_XA+SqlNet=SID1+ACC=P/user1/pass1+SesTM=180+LogDir=/tmp+nolocal=f+Threads=true NDRX_XA_CLOSE_STR=${NDRX_XA_OPEN_STR} NDRX_XA_DRIVERLIB=libndrxxaoras.so NDRX_XA_RMLIB=libclntsh.so NDRX_XA_LAZY_INIT=1
Additionally, Enduro/X transaction manager must be configured to run global transactions, e.g.:
<server name="tmsrv"> <min>1</min> <max>1</max> <srvid>40</srvid> <cctag>ORA1</cctag> <sysopt>-e ${NDRX_ULOG}/tmsrv-rm1.log -r -- -t1 -l${NDRX_APPHOME}/tmlogs/rm1</sysopt> </server>
Bname
and Bfldid
are available to find map from field identifier to name or the other way.
Functions to determine field number and type from identifier:
import endurox as e
assert e.Bfldtype(e.Bmkfldid(e.BFLD_STRING, 10)) == e.BFLD_STRING
assert e.Bfldno(e.Bmkfldid(e.BFLD_STRING, 10)) == 10
On errors, either AtmiException
or UbfException
are raised by the module. Exceptions contain
additional attribute code
that contains the Enduro/X error code and it can be
compared it with defined errors like TPENOENT
or TPESYSTEM
.
import endurox as e
try:
e.tpcall("whatever", {})
except e.AtmiException as ee:
if ee.code == e.TPENOENT:
print("Service does not exist")
The Enduro/X Python module contains all logging features provided by Enduro/X Core.
functions such as tplog()
(including syntactic sugars for log levels), tplogdump()
for dumping bytes to hex dumps in logs, request/session log file contexting and
APIs manipulation with logfile file descriptors.
import endurox as e
e.tplog_debug("This is debug message")
tests/
contains test for all Enduro/X ATMIs provided by the module.
These test cases can be studied to get familiar with module APIs.
The module is licensed under GNU Affero General Public License Version 3, which allows the use of the product for open-source software. Mavimax SIA also provides a commercial license EULA which allows to use Enduro/X Python module in closed source projects.
This document gave a short overview of the Enduro/X Python module. For a full API overview please see API descriptions at https://www.endurox.org/dokuwiki
As all API descriptions are embedded as PyDoc, Python shell can be utilized to get help with the overview, functions, constants, etc.
As the root package "endurox" actually embeds C biding code in another sub-module "endurox", then full API doc can be viewed by:
import endurox as e
help (e.endurox)
The individual identifiers can be looked by directly by:
import endurox as e
help (e.tpcall)
- Version 8.0.2 released on 18/08/2022 - First stable release
- Version 8.0.4 marked on 25/09/2022 - Feature #790