Before the change, profanity has shown
`TypeError: prof_pre_chat_message_display() missing 3 required positional arguments: 'barejid',
'resource', and 'message'`
error for received messages with incorrect encoding
instead of proper error handling for string convertion.
That is a temporary fix, correct fix probably would be
passing potentially problematic char* as `bytes` (using `y` instead of `s` format),
but this would require API changes and hence correction of plugins to
handle new input parameters properly (likely with "decode" function).
A bit hard to reproduce. You need to add a plugin with the following code
(or any other plugin with `prof_pre_chat_message_display` present, such as emoticons.py):
```python
def prof_pre_chat_message_display(barejid, resource, message):
return message
```
and then receive/send message with incorrectly encoded character, it can be any
invalid UTF8 symbol. You'll see error in console, errors don't represent actual
errors in plugins, but rather Profanity's shortcoming, though it make appear so
that the problem is in plugin.
Actual proper handling would likely be using `y` instead of `s` format
(see [reference](https://docs.python.org/3/c-api/arg.html#c.Py_BuildValue))
> s ([str](https://docs.python.org/3/library/stdtypes.html#str) or None) [const char *]
> Convert a null-terminated C string to a Python [str](https://docs.python.org/3/library/stdtypes.html#str)
object using 'utf-8' encoding. If the C string pointer is NULL, None is used.
to
> y ([bytes](https://docs.python.org/3/library/stdtypes.html#bytes)) [const char *]
> This converts a C string to a Python [bytes](https://docs.python.org/3/library/stdtypes.html#bytes)
> object. If the C string pointer is NULL, None is returned.
since there is a [problem](https://docs.python.org/3/c-api/arg.html) with `s`:
> s ([str](https://docs.python.org/3/library/stdtypes.html#str)) [const char *]
> Convert a Unicode object to a C pointer to a character string.
> A pointer to an existing string is stored in the character pointer variable whose address you pass.
> ... Unicode objects are converted to C strings using 'utf-8' encoding.
> **If this conversion fails**, a [UnicodeError](https://docs.python.org/3/library/exceptions.html#UnicodeError) is raised.
In python such problem can be handled using `errors='ignore'` in [bytes.decode](https://docs.python.org/3/library/stdtypes.html#bytes.decode) or in a more sophisticated manner, depending on needs and realization.
Previously it relied on AX_PYTHON_DEVEL, which in turn executes
python-config to get the build flags. However this does not work while
cross compiling because we can't execute the python-config build for the
target platform. To circumvent this problem the python build flags are
now queried via pkgconfig, which has the drawback of not having some
extra build flags, but they do not seem to be needed.
I tested this patch with the termux build system and it build without
their existing hack of injecting python after the configure step. I also
tested non cross compile build on Arch Linux and it also still works.
Fixes#851
ncursesw defines _XOPEN_SOURCE macro via command-line. In particular, it
is defined in ncursesw.pc and extracted via pkg-config. From other side,
Python defines the same macro unconditionally in pyconfig.h. Python-3.x
defines the macro with value different than ncursesw does. In turn, this
causes a warning that the macro is redefined. And warnings are treated
as errors.
Since both entities define the mecro unconditionally, we can't simply
reorder headers as Python developers suggest. So, undefine the macro
just before the <Python.h> to fix this silly issue.