Skip to content

Magic

GPT-powered rough equivalent of the %debug Jupyter magic. After an error occurs, just run %duck in the next cell to get an explanation. This is very similar to using the errors module, but is less intrusive - you only call it when you want an explanation, rather than having to type y/n after each error. We also provide paste mode, which attempts to paste a solution into a new code cell below, and interactive mode, which throws you into a conversational debugging session (technically closer to the original %debug magic functionality.

Quickstart

# cell 1
from roboduck import magic

nums = [1, 2, 3]
nums.add(4)

# cell 2
%duck

Classes

DebugMagic

Bases: Magics

Enter a conversational debugging session after an error is thrown by a jupyter notebook code cell.

Examples:

After a cell execution throws an error, enter this in the next cell to get an explanation of what caused the error and how to fix it:

%duck

To instead enter an Interactive conversational debugging session:

%duck -i

If the -p flag is present, we will try to Paste a solution into a new code cell upon exiting the debugger containing the fixed code snippet.

%duck -p

Flags can be combined:

%duck -ip

Source code in lib/roboduck/magic.py 
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
@magics_class
class DebugMagic(Magics):
    """Enter a conversational debugging session after an error is thrown by a
    jupyter notebook code cell.

    Examples
    --------
    After a cell execution throws an error, enter this in the next cell to
    get an explanation of what caused the error and how to fix it:

    %duck

    To instead enter an Interactive conversational debugging session:

    %duck -i

    If the -p flag is present, we will try to Paste a solution into a new
    code cell upon exiting the debugger containing the fixed code snippet.

    %duck -p

    Flags can be combined:

    %duck -ip
    """

    @magic_arguments()
    @argument('-p', action='store_true',
              help='Boolean flag: if provided, try to PASTE a solution into a '
                   'new code cell below. If you also choose to use '
                   'interactive mode and receive multiple replies containing '
                   'code snippets, the last snippet to be generated will be '
                   'the one that is pasted in (we make the assumption that '
                   'users will prompt for increasingly correct code rather '
                   'than the reverse).')
    @argument('-i', action='store_true',
              help='Boolean flag: if provided, use INTERACTIVE mode. Start a '
                   'conversational debugger session and allow the user to ask '
                   'custom questions, just as they would if using '
                   'roboduck.debug.duck(). The default mode, meanwhile, '
                   'simply asks gpt what caused the error that just '
                   'occurred and then exits, rather than lingering in a '
                   'debugger session.')
    @argument('--prompt', type=str, default=None)
    @line_magic
    def duck(self, line=''):
        """Silence warnings for a cell. The -p flag can be used to make the
        change persist, at least until the user changes it again.

        Developer note: later found a needs_local_scope decorator
        https://ipython.readthedocs.io/en/stable/config/custommagics.html#\
            accessing-user-namespace-and-local-scope
        that might let us load state from ipython in a a more reliable way.
        Not certain this would be an improvement, and major refactor might be
        needed to make this work, but it's something to keep in mind.
        """
        args = parse_argstring(self.duck, line)
        if args.prompt:
            warnings.warn('Support for custom prompts is somewhat limited '
                          'at the moment - your prompt must use the default '
                          'parse_func (roboduck.utils.parse_completion).')
        if args.p and is_colab(self.shell):
            warnings.warn('Paste mode is unavailable in google colab, which '
                          'you appear to be using. Ignoring -p flag.')
        try:
            # Confine this import to this method rather than keeping a top
            # level import because importing this module overwrites
            # sys.excepthook which we don't necessarily want when
            # importing this module.
            # Note that this uses the `debug_stack_trace` prompt by default.
            # This does NOT include a user question param in the contextful
            # prompt, but in this setting we only use that method once (for our
            # automatic question) because excepthook transports us to one
            # step before the error. The user can ask followup questions but
            # the code state will not progress because we can't step through
            # any further, and thus the contextless method (which does include
            # a user question) is used from that point forward.
            # Color should be specified because errors module uses red
            # by default (whereas debug module uses green).
            from roboduck import errors
            kwargs = {'auto': True, 'interactive': args.i, 'color': 'green'}
            if args.prompt:
                kwargs['prompt'] = args.prompt
            errors.excepthook(sys.last_type, sys.last_value,
                              sys.last_traceback, **kwargs)
        except Exception as e:
            pass
        finally:
            errors.disable()

        # Insert suggested code into next cell.
        if args.p and CodeCompletionCache.get('last_completion'):
            self.shell.set_next_input(CodeCompletionCache.get('last_new_code'),
                                      replace=False)
        CodeCompletionCache.reset_class_vars()

Functions

duck(line='')

Silence warnings for a cell. The -p flag can be used to make the change persist, at least until the user changes it again.

Developer note: later found a needs_local_scope decorator https://ipython.readthedocs.io/en/stable/config/custommagics.html# accessing-user-namespace-and-local-scope that might let us load state from ipython in a a more reliable way. Not certain this would be an improvement, and major refactor might be needed to make this work, but it's something to keep in mind.

Source code in lib/roboduck/magic.py 
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
@magic_arguments()
@argument('-p', action='store_true',
          help='Boolean flag: if provided, try to PASTE a solution into a '
               'new code cell below. If you also choose to use '
               'interactive mode and receive multiple replies containing '
               'code snippets, the last snippet to be generated will be '
               'the one that is pasted in (we make the assumption that '
               'users will prompt for increasingly correct code rather '
               'than the reverse).')
@argument('-i', action='store_true',
          help='Boolean flag: if provided, use INTERACTIVE mode. Start a '
               'conversational debugger session and allow the user to ask '
               'custom questions, just as they would if using '
               'roboduck.debug.duck(). The default mode, meanwhile, '
               'simply asks gpt what caused the error that just '
               'occurred and then exits, rather than lingering in a '
               'debugger session.')
@argument('--prompt', type=str, default=None)
@line_magic
def duck(self, line=''):
    """Silence warnings for a cell. The -p flag can be used to make the
    change persist, at least until the user changes it again.

    Developer note: later found a needs_local_scope decorator
    https://ipython.readthedocs.io/en/stable/config/custommagics.html#\
        accessing-user-namespace-and-local-scope
    that might let us load state from ipython in a a more reliable way.
    Not certain this would be an improvement, and major refactor might be
    needed to make this work, but it's something to keep in mind.
    """
    args = parse_argstring(self.duck, line)
    if args.prompt:
        warnings.warn('Support for custom prompts is somewhat limited '
                      'at the moment - your prompt must use the default '
                      'parse_func (roboduck.utils.parse_completion).')
    if args.p and is_colab(self.shell):
        warnings.warn('Paste mode is unavailable in google colab, which '
                      'you appear to be using. Ignoring -p flag.')
    try:
        # Confine this import to this method rather than keeping a top
        # level import because importing this module overwrites
        # sys.excepthook which we don't necessarily want when
        # importing this module.
        # Note that this uses the `debug_stack_trace` prompt by default.
        # This does NOT include a user question param in the contextful
        # prompt, but in this setting we only use that method once (for our
        # automatic question) because excepthook transports us to one
        # step before the error. The user can ask followup questions but
        # the code state will not progress because we can't step through
        # any further, and thus the contextless method (which does include
        # a user question) is used from that point forward.
        # Color should be specified because errors module uses red
        # by default (whereas debug module uses green).
        from roboduck import errors
        kwargs = {'auto': True, 'interactive': args.i, 'color': 'green'}
        if args.prompt:
            kwargs['prompt'] = args.prompt
        errors.excepthook(sys.last_type, sys.last_value,
                          sys.last_traceback, **kwargs)
    except Exception as e:
        pass
    finally:
        errors.disable()

    # Insert suggested code into next cell.
    if args.p and CodeCompletionCache.get('last_completion'):
        self.shell.set_next_input(CodeCompletionCache.get('last_new_code'),
                                  replace=False)
    CodeCompletionCache.reset_class_vars()