pdb — The Python Debugger

Source code: Lib/pdb.py

---

The module pdb defines an interactive source code debugger for Python programs. It supports setting (conditional) breakpoints and single stepping at the source line level, inspection of stack frames, source code listing, and evaluation of arbitrary Python code in the context of any stack frame. It also supports post-mortem debugging and can be called under program control.

The debugger is extensible – it is actually defined as the class Pdb. This is currently undocumented but easily understood by reading the source. The extension interface uses the modules bdb and cmd.

See also

Module faulthandler

Used to dump Python tracebacks explicitly, on a fault, after a timeout, or on a user signal.

Module traceback

Standard interface to extract, format and print stack traces of Python programs.

The typical usage to break into the debugger is to insert:

import pdb; pdb.set_trace()

Or:

breakpoint()

at the location you want to break into the debugger, and then run the program. You can then step through the code following this statement, and continue running without the debugger using the continue command.

Changed in version 3.7: The built-in breakpoint(), when called with defaults, can be used instead of import pdb; pdb.set_trace().

def double(x):
   breakpoint()
   return x * 2
val = 3
print(f"{val} * 2 is {double(val)}")

The debugger’s prompt is (Pdb), which is the indicator that you are in debug mode:

> ...(2)double()
-> breakpoint()
(Pdb) p x
3
(Pdb) continue
3 * 2 is 6

Changed in version 3.3: Tab-completion via the readline module is available for commands and command arguments, e.g. the current global and local names are offered as arguments of the p command.

You can also invoke pdb from the command line to debug other scripts. For example:

python -m pdb myscript.py

When invoked as a module, pdb will automatically enter post-mortem debugging if the program being debugged exits abnormally. After post-mortem debugging (or after normal exit of the program), pdb will restart the program. Automatic restarting preserves pdb’s state (such as breakpoints) and in most cases is more useful than quitting the debugger upon program’s exit.

Changed in version 3.2: Added the -c option to execute commands as if given in a .pdbrc file; see Debugger Commands.

Changed in version 3.7: Added the -m option to execute modules similar to the way python -m does. As with a script, the debugger will pause execution just before the first line of the module.

Typical usage to execute a statement under control of the debugger is:

>>> import pdb
>>> def f(x):
...     print(1 / x)
>>> pdb.run("f(2)")
> <string>(1)<module>()
(Pdb) continue
0.5
>>>

The typical usage to inspect a crashed program is:

>>> import pdb
>>> def f(x):
...     print(1 / x)
...
>>> f(0)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<stdin>", line 2, in f
ZeroDivisionError: division by zero
>>> pdb.pm()
> <stdin>(2)f()
(Pdb) p x
0
(Pdb)

Changed in version 3.13: The implementation of PEP 667 means that name assignments made via pdb will immediately affect the active scope, even when running inside an optimized scope.

The module defines the following functions; each enters the debugger in a slightly different way:

pdb.run(statement, globals=None, locals=None)

Execute the statement (given as a string or a code object) under debugger control. The debugger prompt appears before any code is executed; you can set breakpoints and type continue, or you can step through the statement using step or next (all these commands are explained below). The optional globals and locals arguments specify the environment in which the code is executed; by default the dictionary of the module __main__ is used. (See the explanation of the built-in exec() or eval() functions.)

pdb.runeval(expression, globals=None, locals=None)

Evaluate the expression (given as a string or a code object) under debugger control. When runeval() returns, it returns the value of the expression. Otherwise this function is similar to run().

pdb.runcall(function, *args, **kwds)

Call the function (a function or method object, not a string) with the given arguments. When runcall() returns, it returns whatever the function call returned. The debugger prompt appears as soon as the function is entered.

pdb.set_trace(*, header=None)

Enter the debugger at the calling stack frame. This is useful to hard-code a breakpoint at a given point in a program, even if the code is not otherwise being debugged (e.g. when an assertion fails). If given, header is printed to the console just before debugging begins.

Changed in version 3.7: The keyword-only argument header.

Changed in version 3.13: set_trace() will enter the debugger immediately, rather than on the next line of code to be executed.

pdb.post_mortem(traceback=None)

Enter post-mortem debugging of the given traceback object. If no traceback is given, it uses the one of the exception that is currently being handled (an exception must be being handled if the default is to be used).

pdb.pm()

Enter post-mortem debugging of the exception found in sys.last_exc.

The run* functions and set_trace() are aliases for instantiating the Pdb class and calling the method of the same name. If you want to access further features, you have to do this yourself:

class pdb.Pdb(completekey='tab', stdin=None, stdout=None, skip=None, nosigint=False, readrc=True)

Pdb is the debugger class.

The completekey, stdin and stdout arguments are passed to the underlying cmd.Cmd class; see the description there.

The skip argument, if given, must be an iterable of glob-style module name patterns. The debugger will not step into frames that originate in a module that matches one of these patterns. [1]

By default, Pdb sets a handler for the SIGINT signal (which is sent when the user presses Ctrl-C on the console) when you give a continue command. This allows you to break into the debugger again by pressing Ctrl-C. If you want Pdb not to touch the SIGINT handler, set nosigint to true.

The readrc argument defaults to true and controls whether Pdb will load .pdbrc files from the filesystem.

Example call to enable tracing with skip:

import pdb; pdb.Pdb(skip=['django.*']).set_trace()

Raises an auditing event pdb.Pdb with no arguments.

Changed in version 3.1: Added the skip parameter.

Changed in version 3.2: Added the nosigint parameter. Previously, a SIGINT handler was never set by Pdb.

Changed in version 3.6: The readrc argument.

run(statement, globals=None, locals=None)

runeval(expression, globals=None, locals=None)

runcall(function, *args, **kwds)

set_trace()

See the documentation for the functions explained above.

Debugger Commands

The commands recognized by the debugger are listed below. Most commands can be abbreviated to one or two letters as indicated; e.g. h(elp) means that either h or help can be used to enter the help command (but not he or hel, nor H or Help or HELP). Arguments to commands must be separated by whitespace (spaces or tabs). Optional arguments are enclosed in square brackets ([]) in the command syntax; the square brackets must not be typed. Alternatives in the command syntax are separated by a vertical bar (|).

Entering a blank line repeats the last command entered. Exception: if the last command was a list command, the next 11 lines are listed.

Commands that the debugger doesn’t recognize are assumed to be Python statements and are executed in the context of the program being debugged. Python statements can also be prefixed with an exclamation point (!). This is a powerful way to inspect the program being debugged; it is even possible to change a variable or call a function. When an exception occurs in such a statement, the exception name is printed but the debugger’s state is not changed.

Changed in version 3.13: Expressions/Statements whose prefix is a pdb command are now correctly identified and executed.

The debugger supports aliases. Aliases can have parameters which allows one a certain level of adaptability to the context under examination.

Multiple commands may be entered on a single line, separated by ;;. (A single ; is not used as it is the separator for multiple commands in a line that is passed to the Python parser.) No intelligence is applied to separating the commands; the input is split at the first ;; pair, even if it is in the middle of a quoted string. A workaround for strings with double semicolons is to use implicit string concatenation ';'';' or ";"";".

To set a temporary global variable, use a convenience variable. A convenience variable is a variable whose name starts with $. For example, $foo = 1 sets a global variable $foo which you can use in the debugger session. The convenience variables are cleared when the program resumes execution so it’s less likely to interfere with your program compared to using normal variables like foo = 1.

There are three preset convenience variables:

• $_frame: the current frame you are debugging

• $_retval: the return value if the frame is returning

• $_exception: the exception if the frame is raising an exception

Added in version 3.12: Added the convenience variable feature.

If a file .pdbrc exists in the user’s home directory or in the current directory, it is read with 'utf-8' encoding and executed as if it had been typed at the debugger prompt, with the exception that empty lines and lines starting with # are ignored. This is particularly useful for aliases. If both files exist, the one in the home directory is read first and aliases defined there can be overridden by the local file.

Changed in version 3.2: .pdbrc can now contain commands that continue debugging, such as continue or next. Previously, these commands had no effect.

Changed in version 3.11: .pdbrc is now read with 'utf-8' encoding. Previously, it was read with the system locale encoding.

h(elp) [command]

Without argument, print the list of available commands. With a command as argument, print help about that command. help pdb displays the full documentation (the docstring of the pdb module). Since the command argument must be an identifier, help exec must be entered to get help on the ! command.

w(here)

Print a stack trace, with the most recent frame at the bottom. An arrow (>) indicates the current frame, which determines the context of most commands.

d(own) [count]

Move the current frame count (default one) levels down in the stack trace (to a newer frame).

u(p) [count]

Move the current frame count (default one) levels up in the stack trace (to an older frame).

b(reak) [([filename:]lineno | function) [, condition]]

With a lineno argument, set a break at line lineno in the current file. The line number may be prefixed with a filename and a colon, to specify a breakpoint in another file (possibly one that hasn’t been loaded yet). The file is searched on sys.path. Accepatable forms of filename are /abspath/to/file.py, relpath/file.py, module and package.module.

With a function argument, set a break at the first executable statement within that function. function can be any expression that evaluates to a function in the current namespace.

If a second argument is present, it is an expression which must evaluate to true before the breakpoint is honored.

Without argument, list all breaks, including for each breakpoint, the number of times that breakpoint has been hit, the current ignore count, and the associated condition if any.

Each breakpoint is assigned a number to which all the other breakpoint commands refer.

tbreak [([filename:]lineno | function) [, condition]]

Temporary breakpoint, which is removed automatically when it is first hit. The arguments are the same as for break.

cl(ear) [filename:lineno | bpnumber ...]

With a filename:lineno argument, clear all the breakpoints at this line. With a space separated list of breakpoint numbers, clear those breakpoints. Without argument, clear all breaks (but first ask confirmation).

disable bpnumber [bpnumber ...]

Disable the breakpoints given as a space separated list of breakpoint numbers. Disabling a breakpoint means it cannot cause the program to stop execution, but unlike clearing a breakpoint, it remains in the list of breakpoints and can be (re-)enabled.

enable bpnumber [bpnumber ...]

Enable the breakpoints specified.

ignore bpnumber [count]

Set the ignore count for the given breakpoint number. If count is omitted, the ignore count is set to 0. A breakpoint becomes active when the ignore count is zero. When non-zero, the count is decremented each time the breakpoint is reached and the breakpoint is not disabled and any associated condition evaluates to true.

condition bpnumber [condition]

Set a new condition for the breakpoint, an expression which must evaluate to true before the breakpoint is honored. If condition is absent, any existing condition is removed; i.e., the breakpoint is made unconditional.

commands [bpnumber]

Specify a list of commands for breakpoint number bpnumber. The commands themselves appear on the following lines. Type a line containing just end to terminate the commands. An example:

(Pdb) commands 1
(com) p some_variable
(com) end
(Pdb)

To remove all commands from a breakpoint, type commands and follow it immediately with end; that is, give no commands.

With no bpnumber argument, commands refers to the last breakpoint set.

You can use breakpoint commands to start your program up again. Simply use the continue command, or step, or any other command that resumes execution.

Specifying any command resuming execution (currently continue, step, next, return, jump, quit and their abbreviations) terminates the command list (as if that command was immediately followed by end). This is because any time you resume execution (even with a simple next or step), you may encounter another breakpoint—which could have its own command list, leading to ambiguities about which list to execute.

If you use the silent command in the command list, the usual message about stopping at a breakpoint is not printed. This may be desirable for breakpoints that are to print a specific message and then continue. If none of the other commands print anything, you see no sign that the breakpoint was reached.

s(tep)

Execute the current line, stop at the first possible occasion (either in a function that is called or on the next line in the current function).

n(ext)

Continue execution until the next line in the current function is reached or it returns. (The difference between next and step is that step stops inside a called function, while next executes called functions at (nearly) full speed, only stopping at the next line in the current function.)

unt(il) [lineno]

Without argument, continue execution until the line with a number greater than the current one is reached.

With lineno, continue execution until a line with a number greater or equal to lineno is reached. In both cases, also stop when the current frame returns.

Changed in version 3.2: Allow giving an explicit line number.

r(eturn)

Continue execution until the current function returns.

c(ont(inue))

Continue execution, only stop when a breakpoint is encountered.

j(ump) lineno

Set the next line that will be executed. Only available in the bottom-most frame. This lets you jump back and execute code again, or jump forward to skip code that you don’t want to run.

It should be noted that not all jumps are allowed – for instance it is not possible to jump into the middle of a for loop or out of a finally clause.

l(ist) [first[, last]]

List source code for the current file. Without arguments, list 11 lines around the current line or continue the previous listing. With . as argument, list 11 lines around the current line. With one argument, list 11 lines around at that line. With two arguments, list the given range; if the second argument is less than the first, it is interpreted as a count.

The current line in the current frame is indicated by ->. If an exception is being debugged, the line where the exception was originally raised or propagated is indicated by >>, if it differs from the current line.

Changed in version 3.2: Added the >> marker.

ll | longlist

List all source code for the current function or frame. Interesting lines are marked as for list.

Added in version 3.2.

a(rgs)

Print the arguments of the current function and their current values.

p expression

Evaluate expression in the current context and print its value.

Note

print() can also be used, but is not a debugger command — this executes the Python print() function.

pp expression

Like the p command, except the value of expression is pretty-printed using the pprint module.

whatis expression

Print the type of expression.

source expression

Try to get source code of expression and display it.

Added in version 3.2.

display [expression]

Display the value of expression if it changed, each time execution stops in the current frame.

Without expression, list all display expressions for the current frame.

Note

Display evaluates expression and compares to the result of the previous evaluation of expression, so when the result is mutable, display may not be able to pick up the changes.

Example:

lst = []
breakpoint()
pass
lst.append(1)
print(lst)

Display won’t realize lst has been changed because the result of evaluation is modified in place by lst.append(1) before being compared:

> example.py(3)<module>()
-> pass
(Pdb) display lst
display lst: []
(Pdb) n
> example.py(4)<module>()
-> lst.append(1)
(Pdb) n
> example.py(5)<module>()
-> print(lst)
(Pdb)

You can do some tricks with copy mechanism to make it work:

> example.py(3)<module>()
-> pass
(Pdb) display lst[:]
display lst[:]: []
(Pdb) n
> example.py(4)<module>()
-> lst.append(1)
(Pdb) n
> example.py(5)<module>()
-> print(lst)
display lst[:]: [1]  [old: []]
(Pdb)

Added in version 3.2.

undisplay [expression]

Do not display expression anymore in the current frame. Without expression, clear all display expressions for the current frame.

Added in version 3.2.

interact

Start an interactive interpreter (using the code module) in a new global namespace initialised from the local and global namespaces for the current scope. Use exit() or quit() to exit the interpreter and return to the debugger.

Note

As interact creates a new dedicated namespace for code execution, assignments to variables will not affect the original namespaces. However, modifications to any referenced mutable objects will be reflected in the original namespaces as usual.

Added in version 3.2.

Changed in version 3.13: exit() and quit() can be used to exit the interact command.

Changed in version 3.13: interact directs its output to the debugger’s output channel rather than sys.stderr.

alias [name [command]]

Create an alias called name that executes command. The command must not be enclosed in quotes. Replaceable parameters can be indicated by %1, %2, … and %9, while %* is replaced by all the parameters. If command is omitted, the current alias for name is shown. If no arguments are given, all aliases are listed.

Aliases may be nested and can contain anything that can be legally typed at the pdb prompt. Note that internal pdb commands can be overridden by aliases. Such a command is then hidden until the alias is removed. Aliasing is recursively applied to the first word of the command line; all other words in the line are left alone.

As an example, here are two useful aliases (especially when placed in the .pdbrc file):

# Print instance variables (usage "pi classInst")
alias pi for k in %1.__dict__.keys(): print(f"%1.{k} = {%1.__dict__[k]}")
# Print instance variables in self
alias ps pi self

unalias name

Delete the specified alias name.

! statement

Execute the (one-line) statement in the context of the current stack frame. The exclamation point can be omitted unless the first word of the statement resembles a debugger command, e.g.:

(Pdb) ! n=42
(Pdb)

To set a global variable, you can prefix the assignment command with a global statement on the same line, e.g.:

(Pdb) global list_options; list_options = ['-l']
(Pdb)

run [args ...]

restart [args ...]

Restart the debugged Python program. If args is supplied, it is split with shlex and the result is used as the new sys.argv. History, breakpoints, actions and debugger options are preserved. restart is an alias for run.

q(uit)

Quit from the debugger. The program being executed is aborted.

debug code

Enter a recursive debugger that steps through code (which is an arbitrary expression or statement to be executed in the current environment).

retval

Print the return value for the last return of the current function.

exceptions [excnumber]

List or jump between chained exceptions.

When using pdb.pm() or Pdb.post_mortem(...) with a chained exception instead of a traceback, it allows the user to move between the chained exceptions using exceptions command to list exceptions, and exception <number> to switch to that exception.

Example:

def out():
    try:
        middle()
    except Exception as e:
        raise ValueError("reraise middle() error") from e

def middle():
    try:
        return inner(0)
    except Exception as e:
        raise ValueError("Middle fail")

def inner(x):
    1 / x

 out()

calling pdb.pm() will allow to move between exceptions:

> example.py(5)out()
-> raise ValueError("reraise middle() error") from e

(Pdb) exceptions
  0 ZeroDivisionError('division by zero')
  1 ValueError('Middle fail')
> 2 ValueError('reraise middle() error')

(Pdb) exceptions 0
> example.py(16)inner()
-> 1 / x

(Pdb) up
> example.py(10)middle()
-> return inner(0)

Added in version 3.13.

Footnotes

[1]

Whether a frame is considered to originate in a certain module is determined by the __name__ in the frame globals.