Improve error handling for unsupported object types

If you pass pypyodbc a value of some type it doesn’t know how to convert, the end result is an error message which doesn’t give you much information about what you’ve done wrong.

Take for example the following code:

import pypyodbc as pyodbc

class SomeClass(object): pass

obj = SomeClass()

cnxn = pyodbc.connect('....')
cursor = cnxn.cursor()
cursor.execute("INSERT INTO some_table (a) VALUES (?)", [obj])

When I run this code, it generates the following traceback:

Traceback (most recent call last):
  File "C:\Users\Luke\StackOverflow\pypyodbctest.py", line 9, in <module>
    cursor.execute("INSERT INTO some_table (a) VALUES (?)", [obj])
  File "C:\Python27\lib\site-packages\pypyodbc.py", line 1470, in execute
    self._BindParams(param_types)
  File "C:\Python27\lib\site-packages\pypyodbc.py", line 1275, in _BindParams
    if param_types[col_num][0] == 'u':
TypeError: 'type' object has no attribute '__getitem__'

From this message alone it can be difficult to figure out what has been done wrong.

In my sample code I can fully expect that there will be a problem, as I haven’t specified anything about how a SomeClass instance should be converted into a SQL datatype. However, the same error message appears in this StackOverflow question where it was less clear to the questioner what their mistake was.

[LukeWoodward]

@keshavnagpal: the only thing I can suggest is that you edit the pypyodbc.py file, find the line return type(v) within the get_type function and replace it with something like the following:

    raise TypeError("Value {0} of type {1} is not supported".format(v, type(v)))

That should improve the error handling and tell you the type of the value you are passing in that is causing the problem.

[keshavnagpal]

@LukeWoodward : Thanks a lot!

edit the pypyodbc.py file, find the line return type(v) within the get_type function and replace it with something like the following:

raise TypeError("Value {0} of type {1} is not supported".format(v, type(v))

This worked