This module is an extension of the pdb module of the standard library. It is meant to be fully compatible with its predecessor, yet it introduces a number of new features to make your debugging experience as nice as possible.
pdb
features include:
- colorful TAB completion of Python expressions (through fancycompleter)
- optional syntax highlighting of code listings (through pygments)
- sticky mode
- several new commands to be used from the interactive
(Pdb )
prompt- smart command parsing (hint: have you ever typed
r
orc
at the prompt to print the value of some variable?)- additional convenience functions in the
pdb
module, to be used from your program
pdb
is meant to be a drop-in replacement for pdb
. If you find some
unexpected behavior, please report it as a bug.
Since pdb
is not a valid identifier for pip
and easy_install
,
the package is named pdbpp
:
$ pip install pdbpp -- OR -- $ easy_install pdbpp
pdb
is also available via conda:
$ conda install -c conda-forge pdbpp
Alternatively, you can just put pdb.py
somewhere inside your
PYTHONPATH
.
Note that the module is called pdb.py
so that pdb
will automatically
be used in all places that do import pdb
(e.g. pytest --pdb
will
give you a pdb
prompt).
The old pdb
module is still available by doing e.g. import pdb;
pdb.pdb.set_trace()
.
The following are new commands that you can use from the interative
(Pdb )
prompt.
sticky [start end]
- Toggle sticky mode. When in this mode, every time the current position
changes, the screen is repainted and the whole function shown. Thus, when
doing step-by-step execution you can easily follow the flow of the
execution. If
start
andend
are given, sticky mode is enabled and only lines within that range (extremes included) will be displayed. longlist
(ll
)- List source code for the current function. Different from the normal pdb
list
command,longlist
displays the whole function. The current line is marked with->
. In case of post-mortem debugging, the line which actually raised the exception is marked with>>
. If thehighlight
config option is set and pygments is installed, the source code is highlighted. interact
- Start an interative interpreter whose global namespace contains all the names found in the current scope.
track EXPRESSION
- Display a graph showing which objects are the value of the expression refers
to and are referred by. This command requires the
pypy
source code to be importable. display EXPRESSION
- Add an expression to the display list; expressions in this list are evaluated at each step, and printed every time its value changes. WARNING: since these expressions are evaluated multiple times, make sure not to put expressions with side-effects in the display list.
undisplay EXPRESSION
:- Remove
EXPRESSION
from the display list. source EXPRESSION
- Show the source code for the given function/method/class.
edit EXPRESSION
- Open the editor in the right position to edit the given function/method/class. The editor used is specified in a config option.
hf_unhide
,hf_hide
,hf_list
- Some frames might be marked as "hidden" by e.g. using the @pdb.hideframe
function decorator. By default, hidden frames are not shown in the stack
trace, and cannot be reached using
up
anddown
. You can usehf_unhide
to tell pdb to ignore the hidden status (i.e., to treat hidden frames as normal ones), andhf_hide
to hide them again.hf_list
prints a list of hidden frames. The config optionenable_hidden_frames
can be used to disable handling of hidden frames in general.
By default, pdb tries hard to interpret what you enter at the command prompt as one of its builtin commands. However, this is inconvenient if you want to just print the value of a local variable which happens to have the same name as one of the commands. E.g.:
(Pdb) list 1 2 def fn(): 3 c = 42 4 import pdb;pdb.set_trace() 5 -> return c (Pdb) c
In the example above, instead of printing 42 pdb interprets the input as the
command continue
, and then you loose your prompt. It's even worse than
that, because it happens even if you type e.g. c.__class__
.
pdb fixes this unfriendly (from the author's point of view, of course :-))
behavior by always prefering variable in scope, if it exists. If you really
want to execute the corresponding command, you can prefix it with !!
.
Thus, the example above becomes:
(Pdb ) list 1 2 def fn(): 3 c = 42 4 import pdb;pdb.set_trace() 5 -> return c (Pdb ) c 42 (Pdb ) !!c
Note that the "smart" behavior takes place only when there is ambiguity, i.e. if there exists a variable with the same name as a command: in all other cases, everything works as usual.
Regarding the list
command itself, using list(…
is a special case
that gets handled as the Python builtin:
(Pdb ) list([1, 2]) [1, 2]
The pdb
module that comes with pdb includes all the functions and
classes that are in the module from the standard library. If you find any
difference, please report it as a bug.
In addition, there are some new convenience functions that are unique to pdb .
pdb.xpm()
- eXtended Post Mortem: it is equivalent to
pdb.post_mortem(sys.exc_info()[2])
. If used inside anexcept
clause, it will start a post-mortem pdb prompt from the line that raised the exception being caught. pdb.disable()
- Disable
pdb.set_trace()
: any subsequent call to it will be ignored. pdb.enable()
- Re-enable
pdb.set_trace()
, in case it was disabled bypdb.disable()
.
@pdb.hideframe
- A function decorator that tells pdb to hide the frame corresponding to the
function. Hidden frames do not show up when using interactive commands such
as
up
,down
orwhere
, unlesshf_unhide
is invoked. @pdb.break_on_setattr(attrname, condition=always)
class decorator: break the execution of the program every time the attribute
attrname
is set on any instance of the class.condition
is a callable that takes the target object of thesetattr
and the actual value; by default, it breaks every time the attribute is set. E.g.:@break_on_setattr('bar') class Foo(object): pass f = Foo() f.bar = 42 # the program breaks here
If can be used even after the class has already been created, e.g. if we want to break when some attribute of a particular object is set:
class Foo(object): pass a = Foo() b = Foo() def break_if_a(obj, value): return obj is a break_on_setattr('bar', condition=break_if_a)(Foo) b.bar = 10 # no break a.bar = 42 # the program breaks here
This can be used after
pdb.set_trace()
also:(Pdb ) import pdb (Pdb ) pdb.break_on_setattr('tree_id')(obj.__class__) (Pdb ) continue
To customize pdb , you can put a file named .pdbrc.py
in your home
directory. The file must contain a class named Config
inheriting from
pdb.DefaultConfig
and override the desired values.
The following is a list of the options you can customize, together with their default value:
prompt = '(Pdb ) '
- The prompt to show when in interactive mode.
highlight = True
- Highlight line numbers and the current line when showing the
longlist
of a function or when in sticky mode. encoding = 'utf-8'
- File encoding. Useful when there are international characters in your string literals or comments.
sticky_by_default = False
- Determine whether pdb starts in sticky mode or not.
line_number_color = Color.turquoise
- The color to use for line numbers.
filename_color = Color.yellow
- The color to use for file names when printing the stack entries.
current_line_color = "39;49;7"
- The SGR parameters for the ANSI escape sequence to highlight the current
line.
This is set inside the SGR escape sequence
\e[%sm
where\e
is the ESC character and%s
the given value. See SGR parameters. The following means "reset all colors" (0
), set foreground color to 18 (48;5;18
), and background to21
. The default uses the default foreground (39
) and background (49
) colors, inversed (7
). use_pygments = True
- If pygments is installed and
highlight == True
, apply syntax highlight to the source code when showing thelonglist
of a function or when in sticky mode. bg = 'dark'
- Passed directly to the
pygments.formatters.TerminalFormatter
constructor. Selects the color scheme to use, depending on the background color of your terminal. If you have a light background color, try to set it to'light'
. colorscheme = None
- Passed directly to the
pygments.formatters.TerminalFormatter
constructor. It expects a dictionary that maps token types to (lightbg, darkbg) color names orNone
(default:None
= use builtin colorscheme). editor = '${EDITOR:-vi}'
- The command to invoke when using the
edit
command. By default, it uses$EDITOR
if set, elsevi
. The command must support the standard notationCOMMAND n filename
to open filename at linen
.emacs
andvi
are known to support this. truncate_long_lines = True
- Truncate lines which exceed the terminal width.
exec_if_unfocused = None
- Shell command to execute when starting the pdb prompt and the terminal window is not focused. Useful to e.g. play a sound to alert the user that the execution of the program stopped. It requires the wmctrl module.
disable_pytest_capturing = False
- Old versions of pytest crash when you execute
pdb.set_trace()
in a test, but the standard output is captured (i.e., without the-s
option, which is the default behavior). When this option is on, the stdout capturing is automatically disabled before showing the interactive prompt. enable_hidden_frames = True
- Certain frames can be hidden by default.
If enabled, the commands
hf_unhide
,hf_hide
, andhf_list
can be used to control display of them. show_hidden_frames_count = True
- If
enable_hidden_frames
isTrue
this controls if the number of hidden frames gets displayed. def setup(self, pdb): pass
- This method is called during the initialization of the
Pdb
class. Useful to do complex setup. show_traceback_on_error = True
Display tracebacks for errors via
Pdb.error
, that come fromPdb.default
(i.e. the execution of an unrecognized pdb command), and are not a direct cause of the expression itself (e.g.NameError
with a command likedoesnotexist
).With this option disabled only
*** exception string
gets printed, which often misses useful context.show_traceback_on_error_limit = None
- This option sets the limit to be used with
traceback.format_exception
, whenshow_traceback_on_error
is enabled.
pdb
is developed using Test Driven Development, and we try to keep test
coverage high.
As a general rule, every commit should come with its own test. If it's a new feature, it should come with one or many tests which excercise it. If it's a bug fix, the test should fail before the fix, and pass after.
The goal is to make refactoring easier in the future: if you wonder why a certain line of code does something, in principle it should be possible to comment it out and see which tests fail.
In exceptional cases, the test might be too hard or impossible to write: in that cases it is fine to do a commmit without a test, but you should explain very precisely in the commit message why it is hard to write a test and how to reproduce the buggy behaviour by hand.
It is fine NOT to write a test in the following cases:
- typos, documentation, and in general any non-coding commit
- code refactorings which do not add any feature
- commits which fix an already failing test
- commits to silence warnings
- purely cosmetic changes, such as change the color of the output