Update _axclrt.py (#36)

After investigating the issue, I found that the current code directly assigns the C pointer returned from the AxcrustEngineGetOutputNameByIndex function to the name variable. This causes Python to display the raw C pointer address (e.g., <cdata 'char *' 0x8820f70>) instead of the actual string value.

The proper solution involves a two-step process:

First, store the C pointer in a variable named cffi_name
Then, convert the C pointer to a string using axclrt_cffi.string() and decode it as a UTF-8 string with decode("utf-8")
This modification ensures that C string pointers are properly converted to Python strings, allowing the correct names to appear in the output results. This approach is consistent with the method already used in the _get_inputs() method.

Before change:

axengine/_axclrt.py LineNo.279

name = axclrt_lib.axclrtEngineGetOutputNameByIndex(self._info[0], index)
After change:

axengine/_axclrt.py LineNo.279

cffi_name = axclrt_lib.axclrtEngineGetOutputNameByIndex(self._info[0], index)
name = axclrt_cffi.string(cffi_name).decode("utf-8")

Author: nnn112358

axengine/_axclrt.py

@@ -276,7 +276,8 @@ def _get_outputs(self):
         for group in range(self._shape_count):
             one_group_io = []
             for index in range(axclrt_lib.axclrtEngineGetNumOutputs(self._info[0])):
-                name = axclrt_lib.axclrtEngineGetOutputNameByIndex(self._info[0], index)
+                cffi_name  = axclrt_lib.axclrtEngineGetOutputNameByIndex(self._info[0], index)
+                name = axclrt_cffi.string(cffi_name).decode("utf-8")
 
                 cffi_dtype = axclrt_cffi.new("axclrtEngineDataType *")
                 ret = axclrt_lib.axclrtEngineGetOutputDataType(self._info[0], index, cffi_dtype)