rscada / python-mbus Goto Github PK

Just running the tests that come with python-mbus causes the Python interpreter to segfault.

libmbus is git head as of this morning, tried both with commit widesky/libmbus@716d712 and widesky/libmbus@ef6c4be. libmbus itself appears to work judging from the test/example programs it comes with, so it looks to be something funny with python-mbus.

RC=0 stuartl@rikishi /tmp/python-mbus $ py.test 
======================================================================== test session starts ========================================================================
platform linux2 -- Python 2.7.7 -- py-1.4.30 -- pytest-2.7.2
rootdir: /tmp/python-mbus, inifile: pytest.ini
plugins: cov
collected 16 items 

tests/test_MBus_connect.py Segmentation fault

Changes to python-mbus:

diff --git a/mbus/MBus.py b/mbus/MBus.py
index 7f26165..671a18f 100644
--- a/mbus/MBus.py
+++ b/mbus/MBus.py
@@ -60,7 +60,7 @@ class MBus:
                 raise TypeError("port number not given as integer")

         if None == libpath:
-            libpath = "/usr/local/lib/libmbus.so"
+            libpath = "/usr/local/lib64/libmbus.so"

         self._libmbus = cdll.LoadLibrary(libpath)

diff --git a/pytest.ini b/pytest.ini
index f84356a..0cd849c 100644
--- a/pytest.ini
+++ b/pytest.ini
@@ -1,8 +1,8 @@
 [pytest]
-addopts = --pep8 --cov-config .coveragerc --cov . tests/
+addopts = --cov-config .coveragerc --cov . tests/

-serialdevice = /dev/adjustme
+serialdevice = /dev/ttyS0
 ;serialdevice = /dev/ttyUSB0

 markers =
-       serial: mark a test to be run with serial device attached
\ No newline at end of file
+       serial: mark a test to be run with serial device attached

I think there is a typo in https://github.com/rscada/python-mbus/blob/master/mbus/MBusFrame.py#L21
the field stop is a duplicate and the second instance probably should be type, according to https://github.com/rscada/libmbus/blob/master/mbus/mbus-protocol.h#L91.

I tried redefining the field as:

diff --git a/mbus/MBusFrame.py b/mbus/MBusFrame.py
index 36358e8..1b9c677 100644
--- a/mbus/MBusFrame.py
+++ b/mbus/MBusFrame.py
@@ -1,4 +1,4 @@
-from ctypes import Structure, c_uint8, c_long, c_size_t, POINTER
+from ctypes import Structure, c_int, c_uint8, c_long, c_size_t, POINTER
 
 c_time_t = c_long
 
@@ -18,7 +18,7 @@ MBusFrame._fields_ = [
         ("stop",     c_uint8),
         ("data",     c_uint8 * 252),
         ("data_size", c_size_t),
-        ("stop",      c_uint8),
+        ("type",      c_int),
         ("timestamp", c_time_t),
         ("next",      POINTER(MBusFrame))
 ]

However, even with this change, I wasn't able to read the type of the MBusFrame (also tried c_uint8 for the type). I'm not that familiar with ctypes, so maybe there is another place where the C struct is mapped to the Python object MBusFrame?

For testing I called the function frame_type from the C library directly and it returned 4 while queryting the type attribute in Python always returned 0.

Hi,

Is this a typo or why is start1 type c_uint8 * 16 in MBusFrame._fields_? According to https://github.com/rscada/libmbus/blob/master/mbus/mbus-protocol.h#L77 it shoul also be just c_uint8.

typedef struct _mbus_frame {
    unsigned char start1;
    unsigned char length1;
    unsigned char length2;
    unsigned char start2;
...

How do you fill and use the frame structure with such array definition?

I locally changed it to c_uint8

diff --git a/mbus/MBusFrame.py b/mbus/MBusFrame.py
index 9d1b2b5..8f511b6 100644
--- a/mbus/MBusFrame.py
+++ b/mbus/MBusFrame.py
@@ -7,7 +7,7 @@ class MBusFrame(Structure):
         return "MBusFrame: XXX"
 
 MBusFrame._fields_ = [
-        ("start1",   c_uint8 * 16),  # MBusFrameFixed
+        ("start1",   c_uint8),
         ("length1",  c_uint8),
         ("length2",  c_uint8),
         ("start2",   c_uint8),

Now I can easily create MBus frames with code like this:

        mbus_frame = MBusFrame()
        mbus_frame.start1 = MBUS_FRAME_LONG_START
        mbus_frame.length1 = data_size + 3 
        mbus_frame.length2 = mbus_frame.length1
        mbus_frame.start2 = mbus_frame.start1
...

Is this a mistake and should I send a pull request for that or is it possible to use this structure in any other way?

Hello

I've been testing this code but I can't do much. Seems that this was not designed for the latest version of libmbus.

Can you give me some help on this?

https://github.com/HakuFlugel/python-mbus

Hi,
I managed to get the libmbus library compiled and working on my pi.
I would like to use the python wrapper
to develop my Mbus scanner application on Python.

In the example code, I see an Mbus instance being initialised in line 16
mbus = MBus(host="mbus-gw1", port=8888)

I would like to have Mbus on serial port /dev/TTYAMA0.
Please guide me on initialising Mbus on the serial port instead of TCP.

is it possible to change the baudrate? i tried to call the function 'serial_set_baudrate' of MBusLowLevel but i don't know what mbus_handle_p object to pass.

The original code states no specific FLOSS-license. Which one to choose? I would go for a 3-clause BSD license like libmbus, but AGPL is also fine.

Cheers,
tpltnt