#! /usr/bin/env python3
"""
The Python Debugger Pdb
=======================
To use the debugger in its simplest form:
>>> import pdb
>>> pdb.run('')
The debugger's prompt is '(Pdb) '. This will stop in the first
function call in .
Alternatively, if a statement terminated with an unhandled exception,
you can use pdb's post-mortem facility to inspect the contents of the
traceback:
>>>
>>> import pdb
>>> pdb.pm()
The commands recognized by the debugger are listed in the next
section. Most can be abbreviated as indicated; e.g., h(elp) means
that 'help' can be typed as 'h' or 'help' (but not as 'he' or 'hel',
nor as 'H' or 'Help' or 'HELP'). Optional arguments are enclosed in
square brackets. Alternatives in the command syntax are separated
by a vertical bar (|).
A blank line repeats the previous command literally, except for
'list', where it lists the next 11 lines.
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 variables or call functions.
When an exception occurs in such a statement, the exception name is
printed but the debugger's state is not changed.
The debugger supports aliases, which can save typing. And aliases can
have parameters (see the alias help entry) which allows one a certain
level of adaptability to the context under examination.
Multiple commands may be entered on a single line, separated by the
pair ';;'. No intelligence is applied to separating the commands; the
input is split at the first ';;', even if it is in the middle of a
quoted string.
If a file ".pdbrc" exists in your home directory or in the current
directory, it is read in and executed as if it had been typed at the
debugger prompt. 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. This behavior can be
disabled by passing the "readrc=False" argument to the Pdb constructor.
Aside from aliases, the debugger is not directly programmable; but it
is implemented as a class from which you can derive your own debugger
class, which you can make as fancy as you like.
Debugger commands
=================
"""
# NOTE: the actual command documentation is collected from docstrings of the
# commands and is appended to __doc__ after the class has been defined.
import os
import io
import re
import sys
import cmd
import bdb
import dis
import code
import glob
import token
import types
import codeop
import pprint
import signal
import inspect
import textwrap
import tokenize
import itertools
import traceback
import linecache
import _colorize
from contextlib import contextmanager
from rlcompleter import Completer
from types import CodeType
class Restart(Exception):
"""Causes a debugger to be restarted for the debugged python program."""
pass
__all__ = ["run", "pm", "Pdb", "runeval", "runctx", "runcall", "set_trace",
"post_mortem", "help"]
def find_first_executable_line(code):
""" Try to find the first executable line of the code object.
Equivalently, find the line number of the instruction that's
after RESUME
Return code.co_firstlineno if no executable line is found.
"""
prev = None
for instr in dis.get_instructions(code):
if prev is not None and prev.opname == 'RESUME':
if instr.positions.lineno is not None:
return instr.positions.lineno
return code.co_firstlineno
prev = instr
return code.co_firstlineno
def find_function(funcname, filename):
cre = re.compile(r'def\s+%s(\s*\[.+\])?\s*[(]' % re.escape(funcname))
try:
fp = tokenize.open(filename)
except OSError:
lines = linecache.getlines(filename)
if not lines:
return None
fp = io.StringIO(''.join(lines))
funcdef = ""
funcstart = None
# consumer of this info expects the first line to be 1
with fp:
for lineno, line in enumerate(fp, start=1):
if cre.match(line):
funcstart, funcdef = lineno, line
elif funcdef:
funcdef += line
if funcdef:
try:
code = compile(funcdef, filename, 'exec')
except SyntaxError:
continue
# We should always be able to find the code object here
funccode = next(c for c in code.co_consts if
isinstance(c, CodeType) and c.co_name == funcname)
lineno_offset = find_first_executable_line(funccode)
return funcname, filename, funcstart + lineno_offset - 1
return None
def lasti2lineno(code, lasti):
linestarts = list(dis.findlinestarts(code))
linestarts.reverse()
for i, lineno in linestarts:
if lasti >= i:
return lineno
return 0
class _rstr(str):
"""String that doesn't quote its repr."""
def __repr__(self):
return self
class _ExecutableTarget:
filename: str
code: CodeType | str
namespace: dict
class _ScriptTarget(_ExecutableTarget):
def __init__(self, target):
self._target = os.path.realpath(target)
if not os.path.exists(self._target):
print(f'Error: {target} does not exist')
sys.exit(1)
if os.path.isdir(self._target):
print(f'Error: {target} is a directory')
sys.exit(1)
# If safe_path(-P) is not set, sys.path[0] is the directory
# of pdb, and we should replace it with the directory of the script
if not sys.flags.safe_path:
sys.path[0] = os.path.dirname(self._target)
def __repr__(self):
return self._target
@property
def filename(self):
return self._target
@property
def code(self):
# Open the file each time because the file may be modified
with io.open_code(self._target) as fp:
return f"exec(compile({fp.read()!r}, {self._target!r}, 'exec'))"
@property
def namespace(self):
return dict(
__name__='__main__',
__file__=self._target,
__builtins__=__builtins__,
__spec__=None,
)
class _ModuleTarget(_ExecutableTarget):
def __init__(self, target):
self._target = target
import runpy
try:
_, self._spec, self._code = runpy._get_module_details(self._target)
except ImportError as e:
print(f"ImportError: {e}")
sys.exit(1)
except Exception:
traceback.print_exc()
sys.exit(1)
def __repr__(self):
return self._target
@property
def filename(self):
return self._code.co_filename
@property
def code(self):
return self._code
@property
def namespace(self):
return dict(
__name__='__main__',
__file__=os.path.normcase(os.path.abspath(self.filename)),
__package__=self._spec.parent,
__loader__=self._spec.loader,
__spec__=self._spec,
__builtins__=__builtins__,
)
class _ZipTarget(_ExecutableTarget):
def __init__(self, target):
import runpy
self._target = os.path.realpath(target)
sys.path.insert(0, self._target)
try:
_, self._spec, self._code = runpy._get_main_module_details()
except ImportError as e:
print(f"ImportError: {e}")
sys.exit(1)
except Exception:
traceback.print_exc()
sys.exit(1)
def __repr__(self):
return self._target
@property
def filename(self):
return self._code.co_filename
@property
def code(self):
return self._code
@property
def namespace(self):
return dict(
__name__='__main__',
__file__=os.path.normcase(os.path.abspath(self.filename)),
__package__=self._spec.parent,
__loader__=self._spec.loader,
__spec__=self._spec,
__builtins__=__builtins__,
)
class _PdbInteractiveConsole(code.InteractiveConsole):
def __init__(self, ns, message):
self._message = message
super().__init__(locals=ns, local_exit=True)
def write(self, data):
self._message(data, end='')
# Interaction prompt line will separate file and call info from code
# text using value of line_prefix string. A newline and arrow may
# be to your liking. You can set it once pdb is imported using the
# command "pdb.line_prefix = '\n% '".
# line_prefix = ': ' # Use this to get the old situation back
line_prefix = '\n-> ' # Probably a better default
class Pdb(bdb.Bdb, cmd.Cmd):
_previous_sigint_handler = None
# Limit the maximum depth of chained exceptions, we should be handling cycles,
# but in case there are recursions, we stop at 999.
MAX_CHAINED_EXCEPTION_DEPTH = 999
_file_mtime_table = {}
def __init__(self, completekey='tab', stdin=None, stdout=None, skip=None,
nosigint=False, readrc=True):
bdb.Bdb.__init__(self, skip=skip)
cmd.Cmd.__init__(self, completekey, stdin, stdout)
sys.audit("pdb.Pdb")
if stdout:
self.use_rawinput = 0
self.prompt = '(Pdb) '
self.aliases = {}
self.displaying = {}
self.mainpyfile = ''
self._wait_for_mainpyfile = False
self.tb_lineno = {}
# Try to load readline if it exists
try:
import readline
# remove some common file name delimiters
readline.set_completer_delims(' \t\n`@#%^&*()=+[{]}\\|;:\'",<>?')
except ImportError:
pass
self.allow_kbdint = False
self.nosigint = nosigint
# Consider these characters as part of the command so when the users type
# c.a or c['a'], it won't be recognized as a c(ontinue) command
self.identchars = cmd.Cmd.identchars + '=.[](),"\'+-*/%@&|<>~^'
# Read ~/.pdbrc and ./.pdbrc
self.rcLines = []
if readrc:
try:
with open(os.path.expanduser('~/.pdbrc'), encoding='utf-8') as rcFile:
self.rcLines.extend(rcFile)
except OSError:
pass
try:
with open(".pdbrc", encoding='utf-8') as rcFile:
self.rcLines.extend(rcFile)
except OSError:
pass
self.commands = {} # associates a command list to breakpoint numbers
self.commands_doprompt = {} # for each bp num, tells if the prompt
# must be disp. after execing the cmd list
self.commands_silent = {} # for each bp num, tells if the stack trace
# must be disp. after execing the cmd list
self.commands_defining = False # True while in the process of defining
# a command list
self.commands_bnum = None # The breakpoint number for which we are
# defining a list
self._chained_exceptions = tuple()
self._chained_exception_index = 0
def sigint_handler(self, signum, frame):
if self.allow_kbdint:
raise KeyboardInterrupt
self.message("\nProgram interrupted. (Use 'cont' to resume).")
self.set_step()
self.set_trace(frame)
def reset(self):
bdb.Bdb.reset(self)
self.forget()
def forget(self):
self.lineno = None
self.stack = []
self.curindex = 0
if hasattr(self, 'curframe') and self.curframe:
self.curframe.f_globals.pop('__pdb_convenience_variables', None)
self.curframe = None
self.tb_lineno.clear()
def setup(self, f, tb):
self.forget()
self.stack, self.curindex = self.get_stack(f, tb)
while tb:
# when setting up post-mortem debugging with a traceback, save all
# the original line numbers to be displayed along the current line
# numbers (which can be different, e.g. due to finally clauses)
lineno = lasti2lineno(tb.tb_frame.f_code, tb.tb_lasti)
self.tb_lineno[tb.tb_frame] = lineno
tb = tb.tb_next
self.curframe = self.stack[self.curindex][0]
# The f_locals dictionary used to be updated from the actual frame
# locals whenever the .f_locals accessor was called, so it was
# cached here to ensure that modifications were not overwritten. While
# the caching is no longer required now that f_locals is a direct proxy
# on optimized frames, it's also harmless, so the code structure has
# been left unchanged.
self.curframe_locals = self.curframe.f_locals
self.set_convenience_variable(self.curframe, '_frame', self.curframe)
if self._chained_exceptions:
self.set_convenience_variable(
self.curframe,
'_exception',
self._chained_exceptions[self._chained_exception_index],
)
if self.rcLines:
self.cmdqueue = [
line for line in self.rcLines
if line.strip() and not line.strip().startswith("#")
]
self.rcLines = []
# Override Bdb methods
def user_call(self, frame, argument_list):
"""This method is called when there is the remote possibility
that we ever need to stop in this function."""
if self._wait_for_mainpyfile:
return
if self.stop_here(frame):
self.message('--Call--')
self.interaction(frame, None)
def user_line(self, frame):
"""This function is called when we stop or break at this line."""
if self._wait_for_mainpyfile:
if (self.mainpyfile != self.canonic(frame.f_code.co_filename)):
return
self._wait_for_mainpyfile = False
if self.bp_commands(frame):
self.interaction(frame, None)
user_opcode = user_line
def bp_commands(self, frame):
"""Call every command that was set for the current active breakpoint
(if there is one).
Returns True if the normal interaction function must be called,
False otherwise."""
# self.currentbp is set in bdb in Bdb.break_here if a breakpoint was hit
if getattr(self, "currentbp", False) and \
self.currentbp in self.commands:
currentbp = self.currentbp
self.currentbp = 0
lastcmd_back = self.lastcmd
self.setup(frame, None)
for line in self.commands[currentbp]:
self.onecmd(line)
self.lastcmd = lastcmd_back
if not self.commands_silent[currentbp]:
self.print_stack_entry(self.stack[self.curindex])
if self.commands_doprompt[currentbp]:
self._cmdloop()
self.forget()
return
return 1
def user_return(self, frame, return_value):
"""This function is called when a return trap is set here."""
if self._wait_for_mainpyfile:
return
frame.f_locals['__return__'] = return_value
self.set_convenience_variable(frame, '_retval', return_value)
self.message('--Return--')
self.interaction(frame, None)
def user_exception(self, frame, exc_info):
"""This function is called if an exception occurs,
but only if we are to stop at or just below this level."""
if self._wait_for_mainpyfile:
return
exc_type, exc_value, exc_traceback = exc_info
frame.f_locals['__exception__'] = exc_type, exc_value
self.set_convenience_variable(frame, '_exception', exc_value)
# An 'Internal StopIteration' exception is an exception debug event
# issued by the interpreter when handling a subgenerator run with
# 'yield from' or a generator controlled by a for loop. No exception has
# actually occurred in this case. The debugger uses this debug event to
# stop when the debuggee is returning from such generators.
prefix = 'Internal ' if (not exc_traceback
and exc_type is StopIteration) else ''
self.message('%s%s' % (prefix, self._format_exc(exc_value)))
self.interaction(frame, exc_traceback)
# General interaction function
def _cmdloop(self):
while True:
try:
# keyboard interrupts allow for an easy way to cancel
# the current command, so allow them during interactive input
self.allow_kbdint = True
self.cmdloop()
self.allow_kbdint = False
break
except KeyboardInterrupt:
self.message('--KeyboardInterrupt--')
def _validate_file_mtime(self):
"""Check if the source file of the current frame has been modified since
the last time we saw it. If so, give a warning."""
try:
filename = self.curframe.f_code.co_filename
mtime = os.path.getmtime(filename)
except Exception:
return
if (filename in self._file_mtime_table and
mtime != self._file_mtime_table[filename]):
self.message(f"*** WARNING: file '{filename}' was edited, "
"running stale code until the program is rerun")
self._file_mtime_table[filename] = mtime
# Called before loop, handles display expressions
# Set up convenience variable containers
def _show_display(self):
displaying = self.displaying.get(self.curframe)
if displaying:
for expr, oldvalue in displaying.items():
newvalue = self._getval_except(expr)
# check for identity first; this prevents custom __eq__ to
# be called at every loop, and also prevents instances whose
# fields are changed to be displayed
if newvalue is not oldvalue and newvalue != oldvalue:
displaying[expr] = newvalue
self.message('display %s: %s [old: %s]' %
(expr, self._safe_repr(newvalue, expr),
self._safe_repr(oldvalue, expr)))
def _get_tb_and_exceptions(self, tb_or_exc):
"""
Given a tracecack or an exception, return a tuple of chained exceptions
and current traceback to inspect.
This will deal with selecting the right ``__cause__`` or ``__context__``
as well as handling cycles, and return a flattened list of exceptions we
can jump to with do_exceptions.
"""
_exceptions = []
if isinstance(tb_or_exc, BaseException):
traceback, current = tb_or_exc.__traceback__, tb_or_exc
while current is not None:
if current in _exceptions:
break
_exceptions.append(current)
if current.__cause__ is not None:
current = current.__cause__
elif (
current.__context__ is not None and not current.__suppress_context__
):
current = current.__context__
if len(_exceptions) >= self.MAX_CHAINED_EXCEPTION_DEPTH:
self.message(
f"More than {self.MAX_CHAINED_EXCEPTION_DEPTH}"
" chained exceptions found, not all exceptions"
"will be browsable with `exceptions`."
)
break
else:
traceback = tb_or_exc
return tuple(reversed(_exceptions)), traceback
@contextmanager
def _hold_exceptions(self, exceptions):
"""
Context manager to ensure proper cleaning of exceptions references
When given a chained exception instead of a traceback,
pdb may hold references to many objects which may leak memory.
We use this context manager to make sure everything is properly cleaned
"""
try:
self._chained_exceptions = exceptions
self._chained_exception_index = len(exceptions) - 1
yield
finally:
# we can't put those in forget as otherwise they would
# be cleared on exception change
self._chained_exceptions = tuple()
self._chained_exception_index = 0
def interaction(self, frame, tb_or_exc):
# Restore the previous signal handler at the Pdb prompt.
if Pdb._previous_sigint_handler:
try:
signal.signal(signal.SIGINT, Pdb._previous_sigint_handler)
except ValueError: # ValueError: signal only works in main thread
pass
else:
Pdb._previous_sigint_handler = None
_chained_exceptions, tb = self._get_tb_and_exceptions(tb_or_exc)
if isinstance(tb_or_exc, BaseException):
assert tb is not None, "main exception must have a traceback"
with self._hold_exceptions(_chained_exceptions):
self.setup(frame, tb)
# We should print the stack entry if and only if the user input
# is expected, and we should print it right before the user input.
# We achieve this by appending _pdbcmd_print_frame_status to the
# command queue. If cmdqueue is not exausted, the user input is
# not expected and we will not print the stack entry.
self.cmdqueue.append('_pdbcmd_print_frame_status')
self._cmdloop()
# If _pdbcmd_print_frame_status is not used, pop it out
if self.cmdqueue and self.cmdqueue[-1] == '_pdbcmd_print_frame_status':
self.cmdqueue.pop()
self.forget()
def displayhook(self, obj):
"""Custom displayhook for the exec in default(), which prevents
assignment of the _ variable in the builtins.
"""
# reproduce the behavior of the standard displayhook, not printing None
if obj is not None:
self.message(repr(obj))
@contextmanager
def _disable_command_completion(self):
completenames = self.completenames
try:
self.completenames = self.completedefault
yield
finally:
self.completenames = completenames
return
def _exec_in_closure(self, source, globals, locals):
""" Run source code in closure so code object created within source
can find variables in locals correctly
returns True if the source is executed, False otherwise
"""
# Determine if the source should be executed in closure. Only when the
# source compiled to multiple code objects, we should use this feature.
# Otherwise, we can just raise an exception and normal exec will be used.
code = compile(source, "", "exec")
if not any(isinstance(const, CodeType) for const in code.co_consts):
return False
# locals could be a proxy which does not support pop
# copy it first to avoid modifying the original locals
locals_copy = dict(locals)
locals_copy["__pdb_eval__"] = {
"result": None,
"write_back": {}
}
# If the source is an expression, we need to print its value
try:
compile(source, "", "eval")
except SyntaxError:
pass
else:
source = "__pdb_eval__['result'] = " + source
# Add write-back to update the locals
source = ("try:\n" +
textwrap.indent(source, " ") + "\n" +
"finally:\n" +
" __pdb_eval__['write_back'] = locals()")
# Build a closure source code with freevars from locals like:
# def __pdb_outer():
# var = None
# def __pdb_scope(): # This is the code object we want to execute
# nonlocal var
#