# The MIT License

#

# Copyright (c) 2009 the bpython authors.

#

# Permission is hereby granted, free of charge, to any person obtaining a copy

# of this software and associated documentation files (the "Software"), to deal

# in the Software without restriction, including without limitation the rights

# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell

# copies of the Software, and to permit persons to whom the Software is

# furnished to do so, subject to the following conditions:

#

# The above copyright notice and this permission notice shall be included in

# all copies or substantial portions of the Software.

#

# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR

# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE

# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER

# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,

# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN

# THE SOFTWARE.

#

from __future__ import with_statement

import code

import codecs

import inspect

import os

import pydoc

import re

import rlcompleter

import sys

import textwrap

import traceback

from glob import glob

from itertools import takewhile

from locale import getpreferredencoding

from urlparse import urljoin

from xmlrpclib import ServerProxy, Error as XMLRPCError

from pygments.lexers import PythonLexer

from pygments.token import Token

from bpython import importcompletion, inspection

from bpython.formatter import Parenthesis

py3 = sys.version_info[0] == 3

class Interpreter(code.InteractiveInterpreter):

def __init__(self, locals=None, encoding=sys.getdefaultencoding()):

"""The syntaxerror callback can be set at any time and will be called

on a caught syntax error. The purpose for this in bpython is so that

the repl can be instantiated after the interpreter (which it

necessarily must be with the current factoring) and then an exception

callback can be added to the Interpeter instance afterwards - more

specifically, this is so that autoindentation does not occur after a

traceback."""

self.encoding = encoding

self.syntaxerror_callback = None

# Unfortunately code.InteractiveInterpreter is a classic class, so no super()

code.InteractiveInterpreter.__init__(self, locals)

if not py3:

def runsource(self, source, filename='<input>', symbol='single'):

source = '# coding: %s\n%s' % (self.encoding,

source.encode(self.encoding))

return code.InteractiveInterpreter.runsource(self, source,

filename, symbol)

def showsyntaxerror(self, filename=None):

"""Override the regular handler, the code's copied and pasted from

code.py, as per showtraceback, but with the syntaxerror callback called

and the text in a pretty colour."""

if self.syntaxerror_callback is not None:

self.syntaxerror_callback()

type, value, sys.last_traceback = sys.exc_info()

sys.last_type = type

sys.last_value = value

if filename and type is SyntaxError:

# Work hard to stuff the correct filename in the exception

try:

msg, (dummy_filename, lineno, offset, line) = value.args

except:

# Not the format we expect; leave it alone

pass

else:

# Stuff in the right filename and right lineno

if not py3:

lineno -= 1

value = SyntaxError(msg, (filename, lineno, offset, line))

sys.last_value = value

list = traceback.format_exception_only(type, value)

self.writetb(list)

def showtraceback(self):

"""This needs to override the default traceback thing

so it can put it into a pretty colour and maybe other

stuff, I don't know"""

try:

t, v, tb = sys.exc_info()

sys.last_type = t

sys.last_value = v

sys.last_traceback = tb

tblist = traceback.extract_tb(tb)

del tblist[:1]

# Set the right lineno (encoding header adds an extra line)

lineno = tblist[0][1]

if not py3:

lineno -= 1

tblist[0] = (tblist[0][0], lineno) + tblist[0][2:]

l = traceback.format_list(tblist)

if l:

l.insert(0, "Traceback (most recent call last):\n")

l[len(l):] = traceback.format_exception_only(t, v)

finally:

tblist = tb = None

self.writetb(l)

def writetb(self, lines):

"""This outputs the traceback and should be overridden for anything

fancy."""

for line in lines:

self.write(line)

class History(object):

def __init__(self, entries=None):

if entries is None:

self.entries = ['']

else:

self.entries = list(entries)

self.index = 0

self.saved_line = ''

def append(self, line):

line = line.rstrip('\n')

if line:

self.entries.append(line)

def back(self):

if not self.is_at_end():

self.index += 1

return self.entries[-self.index]

def is_at_end(self):

return self.index >= len(self.entries)

def is_at_start(self):

return self.index == 0

def enter(self, line):

if self.index == 0:

self.saved_line = line

def forward(self):

if self.index > 1:

self.index -= 1

return self.entries[-self.index]

else:

self.index = 0

return self.saved_line

@classmethod

def from_filename(cls, filename):

history = cls()

history.load(filename)

return history

def load(self, filename, encoding):

with codecs.open(filename, 'r', encoding, 'ignore') as hfile:

for line in hfile:

self.append(line)

def reset(self):

self.index = 0

self.saved_line = ''

def save(self, filename, encoding, lines=0):

with codecs.open(filename, 'w', encoding, 'ignore') as hfile:

for line in self.entries[-lines:]:

hfile.write(line)

hfile.write('\n')

class MatchesIterator(object):

def __init__(self, current_word='', matches=[]):

self.current_word = current_word

self.matches = list(matches)

self.index = -1

def __nonzero__(self):

return self.index != -1

def __iter__(self):

return self

def current(self):

if self.index == -1:

raise ValueError('No current match.')

return self.matches[self.index]

def next(self):

self.index = (self.index + 1) % len(self.matches)

return self.matches[self.index]

def update(self, current_word='', matches=[]):

if current_word != self.current_word:

self.current_word = current_word

self.matches = list(matches)

self.index = -1

class Repl(object):

"""Implements the necessary guff for a Python-repl-alike interface

The execution of the code entered and all that stuff was taken from the

Python code module, I had to copy it instead of inheriting it, I can't

remember why. The rest of the stuff is basically what makes it fancy.

It reads what you type, passes it to a lexer and highlighter which

returns a formatted string. This then gets passed to echo() which

parses that string and prints to the curses screen in appropriate

colours and/or bold attribute.

The Repl class also keeps two stacks of lines that the user has typed in:

One to be used for the undo feature. I am not happy with the way this

works. The only way I have been able to think of is to keep the code

that's been typed in in memory and re-evaluate it in its entirety for each

"undo" operation. Obviously this means some operations could be extremely

slow. I'm not even by any means certain that this truly represents a

genuine "undo" implementation, but it does seem to be generally pretty

effective.

If anyone has any suggestions for how this could be improved, I'd be happy

to hear them and implement it/accept a patch. I researched a bit into the

idea of keeping the entire Python state in memory, but this really seems

very difficult (I believe it may actually be impossible to work) and has

its own problems too.

The other stack is for keeping a history for pressing the up/down keys

to go back and forth between lines.

XXX Subclasses should implement echo, current_line, cw

"""

def __init__(self, interp, idle=None):

"""Initialise the repl with, unfortunately, a curses screen passed to

it. This needs to be split up so the curses crap isn't in here.

interp is a Python code.InteractiveInterpreter instance

The optional 'idle' parameter is a function that the repl call while

it's blocking (waiting for keypresses). This, again, should be in a

different class"""

self.cut_buffer = ''

self.buffer = []

self.interp = interp

self.interp.syntaxerror_callback = self.clear_current_line

self.match = False

self.rl_history = History()

self.stdout_hist = ''

self.s_hist = []

self.history = []

self.evaluating = False

# Use the interpreter's namespace only for the readline stuff:

self.completer = rlcompleter.Completer(self.interp.locals)

self.completer.attr_matches = self.attr_matches

# Gna, Py 2.6's rlcompleter searches for __call__ inside the

# instance instead of the type, so we monkeypatch to prevent

# side-effects (__getattr__/__getattribute__)

self.completer._callable_postfix = self._callable_postfix

self.matches = []

self.matches_iter = MatchesIterator()

self.argspec = None

self.current_func = None

self.highlighted_paren = None

self.list_win_visible = False

self._C = {}

self.prev_block_finished = 0

sys.path.insert(0, '.')

pythonhist = os.path.expanduser(OPTS.hist_file)

if os.path.exists(pythonhist):

self.rl_history.load(pythonhist, getpreferredencoding())

# This was a feature request to have the PYTHONSTARTUP

# file executed on startup - I personally don't use this

# feature so please notify me of any breakage.

filename = os.environ.get('PYTHONSTARTUP')

if filename and os.path.isfile(filename):

with open(filename, 'r') as f:

self.interp.runsource(f.read(), filename, 'exec')

def attr_matches(self, text):

"""Taken from rlcompleter.py and bent to my will."""

m = re.match(r"(\w+(\.\w+)*)\.(\w*)", text)

if not m:

return []

expr, attr = m.group(1, 3)

if expr.isdigit():

# Special case: float literal, using attrs here will result in

# a SyntaxError

return []

obj = eval(expr, self.interp.locals)

with inspection.AttrCleaner(obj):

matches = self.attr_lookup(obj, expr, attr)

return matches

def attr_lookup(self, obj, expr, attr):

"""Second half of original attr_matches method factored out so it can

be wrapped in a safe try/finally block in case anything bad happens to

restore the original __getattribute__ method."""

words = dir(obj)

if hasattr(obj, '__class__'):

words.append('__class__')

words = words + rlcompleter.get_class_members(obj.__class__)

matches = []

n = len(attr)

for word in words:

if word[:n] == attr and word != "__builtins__":

matches.append("%s.%s" % (expr, word))

return matches

def _callable_postfix(self, value, word):

"""rlcompleter's _callable_postfix done right."""

with inspection.AttrCleaner(value):

if hasattr(value, '__call__'):

word += '('

return word

def current_string(self, concatenate=False):

"""Return the current string."""

tokens = self.tokenize(self.current_line())

string_tokens = list(takewhile(token_is_any_of([Token.String,

Token.Text]),

reversed(tokens)))

if not string_tokens:

return ''

opening = string_tokens.pop()[1]

string = list()

for (token, value) in reversed(string_tokens):

if token is Token.Text:

continue

elif opening is None:

opening = value

elif token is Token.String.Doc:

string.append(value[3:-3])

opening = None

elif value == opening:

opening = None

if not concatenate:

string = list()

else:

string.append(value)

if opening is None:

return ''

return ''.join(string)

def get_object(self, name):

if name in self.interp.locals:

return self.interp.locals[name]

else:

return eval(name, self.interp.locals)

def get_args(self):

"""Check if an unclosed parenthesis exists, then attempt to get the

argspec() for it. On success, update self.argspec and return True,

otherwise set self.argspec to None and return False"""

self.current_func = None

if not OPTS.arg_spec:

return False

# Get the name of the current function and where we are in

# the arguments

stack = [['', 0, '']]

try:

for (token, value) in PythonLexer().get_tokens(self.current_line()):

if token is Token.Punctuation:

if value in '([{':

stack.append(['', 0, value])

elif value in ')]}':

stack.pop()

elif value == ',':

try:

stack[-1][1] += 1

except TypeError:

stack[-1][1] = ''

elif (token is Token.Name or token in Token.Name.subtypes or

token is Token.Operator and value == '.'):

stack[-1][0] += value

elif token is Token.Operator and value == '=':

stack[-1][1] = stack[-1][0]

else:

stack[-1][0] = ''

while stack[-1][2] in '[{':

stack.pop()

_, arg_number, _ = stack.pop()

func, _, _ = stack.pop()

except IndexError:

return False

if not func:

return False

try:

f = self.get_object(func)

except (AttributeError, NameError, SyntaxError):

return False

if inspect.isclass(f):

try:

if f.__init__ is not object.__init__:

f = f.__init__

except AttributeError:

return None

self.current_func = f

self.argspec = inspection.getargspec(func, f)

if self.argspec:

self.argspec.append(arg_number)

return True

return False

def complete(self, tab=False):

"""Construct a full list of possible completions and construct and

display them in a window. Also check if there's an available argspec

(via the inspect module) and bang that on top of the completions too.

The return value is whether the list_win is visible or not."""

self.docstring = None

if not self.get_args():

self.argspec = None

elif self.current_func is not None:

self.docstring = pydoc.getdoc(self.current_func)

if not self.docstring:

self.docstring = None

cw = self.cw()

cs = self.current_string()

if not cw:

self.matches = []

self.matches_iter.update()

if not (cw or cs):

return bool(self.argspec)

if cs and tab:

# Filename completion

self.matches = list()

username = cs.split(os.path.sep, 1)[0]

user_dir = os.path.expanduser(username)

for filename in glob(os.path.expanduser(cs + '*')):

if os.path.isdir(filename):

filename += os.path.sep

if cs.startswith('~'):

filename = username + filename[len(user_dir):]

self.matches.append(filename)

self.matches_iter.update(cs, self.matches)

return bool(self.matches)

elif cs:

# Do not provide suggestions inside strings, as one cannot tab

# them so they would be really confusing.

self.matches_iter.update()

return False

# Check for import completion

e = False

matches = importcompletion.complete(self.current_line(), cw)

if matches is not None and not matches:

self.matches = []

self.matches_iter.update()

return False

if matches is None:

# Nope, no import, continue with normal completion

try:

self.completer.complete(cw, 0)

except Exception:

# This sucks, but it's either that or list all the exceptions that could

# possibly be raised here, so if anyone wants to do that, feel free to send me

# a patch. XXX: Make sure you raise here if you're debugging the completion

# stuff !

e = True

else:

matches = self.completer.matches

if not e and self.argspec:

matches.extend(name + '=' for name in self.argspec[1][0]

if name.startswith(cw))

if py3:

matches.extend(name + '=' for name in self.argspec[1][4]

if name.startswith(cw))

if e or not matches:

self.matches = []

self.matches_iter.update()

if not self.argspec:

return False

else:

# remove duplicates and restore order

self.matches = sorted(set(matches))

if len(self.matches) == 1 and not OPTS.auto_display_list:

self.list_win_visible = True

self.tab()

return False

self.matches_iter.update(cw, self.matches)

return True

def format_docstring(self, docstring, width, height):

"""Take a string and try to format it into a sane list of strings to be

put into the suggestion box."""

lines = docstring.split('\n')

out = []

i = 0

for line in lines:

i += 1

if not line.strip():

out.append('\n')

for block in textwrap.wrap(line, width):

out.append(' ' + block + '\n')

if i >= height:

return out

i += 1

# Drop the last newline

out[-1] = out[-1].rstrip()

return out

def getstdout(self):

"""This method returns the 'spoofed' stdout buffer, for writing to a

file or sending to a pastebin or whatever."""

return self.stdout_hist + '\n'

def formatforfile(self, s):

"""Format the stdout buffer to something suitable for writing to disk,

i.e. without >>> and ... at input lines and with "# OUT: " prepended to

output lines."""

def process():

for line in s.split('\n'):

if line.startswith('>>>') or line.startswith('...'):

yield line[4:]

elif line.rstrip():

yield "# OUT: %s" % (line,)

return "\n".join(process())

def write2file(self):

"""Prompt for a filename and write the current contents of the stdout

buffer to disk."""

try:

fn = self.statusbar.prompt('Save to file (Esc to cancel): ')

except ValueError:

self.statusbar.message("Save cancelled.")

return

if fn.startswith('~'):

fn = os.path.expanduser(fn)

s = self.formatforfile(self.getstdout())

try:

f = open(fn, 'w')

f.write(s)

f.close()

except IOError:

self.statusbar.message("Disk write error for file '%s'." % (fn, ))

else:

self.statusbar.message('Saved to %s' % (fn, ))

def pastebin(self):

"""Upload to a pastebin and display the URL in the status bar."""

pasteservice_url = OPTS.pastebin_url

pasteservice = ServerProxy(urljoin(pasteservice_url, '/xmlrpc/'))

s = self.getstdout()

self.statusbar.message('Posting data to pastebin...')

try:

paste_id = pasteservice.pastes.newPaste('pycon', s)

except XMLRPCError, e:

self.statusbar.message( 'Upload failed: %s' % (str(e), ) )

return

paste_url = urljoin(pasteservice_url, '/show/%s/' % (paste_id, ))

self.statusbar.message('Pastebin URL: %s' % (paste_url, ), 10)

def make_list(self, items):

"""Compile a list of items. At the moment this simply returns

the list; it's here in case I decide to add any more functionality.

I originally had this method return a list of items where each item

was prepended with a number/letter so the user could choose an option

but it doesn't seem appropriate for readline-like behaviour."""

return items

def push(self, s, insert_into_history=True):

"""Push a line of code onto the buffer so it can process it all

at once when a code block ends"""

s = s.rstrip('\n')

self.buffer.append(s)

if insert_into_history:

self.rl_history.append(s)

try:

more = self.interp.runsource('\n'.join(self.buffer))

except SystemExit:

# Avoid a traceback on e.g. quit()

self.do_exit = True

return False

if not more:

self.buffer = []

return more

def undo(self, n=1):

"""Go back in the undo history n steps and call reeavluate()

Note that in the program this is called "Rewind" because I

want it to be clear that this is by no means a true undo

implementation, it is merely a convenience bonus."""

if not self.history:

return None

if len(self.history) < n:

n = len(self.history)

self.history = self.history[:-n]

self.reevaluate()

def reevaluate(self):

"""Clear the buffer, redraw the screen and re-evaluate the history"""

self.evaluating = True

self.stdout_hist = ''

self.f_string = ''

self.buffer = []

self.scr.erase()

self.s_hist = []

# Set cursor position to -1 to prevent paren matching

self.cpos = -1

self.prompt(False)

self.iy, self.ix = self.scr.getyx()

for line in self.history:

if py3:

self.stdout_hist += line + '\n'

else:

self.stdout_hist += line.encode(getpreferredencoding()) + '\n'

self.print_line(line)

self.s_hist[-1] += self.f_string

# I decided it was easier to just do this manually

# than to make the print_line and history stuff more flexible.

self.scr.addstr('\n')

more = self.push(line)

self.prompt(more)

self.iy, self.ix = self.scr.getyx()

self.cpos = 0

indent = next_indentantion(self.s)

self.s = ''

self.scr.refresh()

for _ in xrange(indent):

self.tab()

self.evaluating = False

#map(self.push, self.history)

#^-- That's how simple this method was at first :(

def write(self, s):

"""For overriding stdout defaults"""

if '\x04' in s:

for block in s.split('\x04'):

self.write(block)

return

if s.rstrip() and '\x03' in s:

t = s.split('\x03')[1]

else:

t = s

if not py3 and isinstance(t, unicode):

t = t.encode(getpreferredencoding())

if not self.stdout_hist:

self.stdout_hist = t

else:

self.stdout_hist += t

self.echo(s)

self.s_hist.append(s.rstrip())

def flush(self):

"""Olivier Grisel brought it to my attention that the logging

module tries to call this method, since it makes assumptions

about stdout that may not necessarily be true. The docs for

sys.stdout say:

"stdout and stderr needn't be built-in file objects: any

object is acceptable as long as it has a write() method

that takes a string argument."

So I consider this to be a bug in logging, and this is a hack

to fix it, unfortunately. I'm sure it's not the only module

to do it."""

def close(self):

"""See the flush() method docstring."""

def tokenize(self, s, newline=False):

"""Tokenize a line of code."""

source = '\n'.join(self.buffer + [s])

cursor = len(source) - self.cpos

if self.cpos:

cursor += 1

stack = list()

all_tokens = list(PythonLexer().get_tokens(source))

# Unfortunately, Pygments adds a trailing newline and strings with

# no size, so strip them

while not all_tokens[-1][1]:

all_tokens.pop()

all_tokens[-1] = (all_tokens[-1][0], all_tokens[-1][1].rstrip('\n'))

line = pos = 0

parens = dict(zip('{([', '})]'))

line_tokens = list()

saved_tokens = list()

search_for_paren = True

for (token, value) in split_lines(all_tokens):

pos += len(value)

if token is Token.Text and value == '\n':

line += 1

# Remove trailing newline

line_tokens = list()

saved_tokens = list()

continue

line_tokens.append((token, value))

saved_tokens.append((token, value))

if not search_for_paren:

continue

under_cursor = (pos == cursor)

if token is Token.Punctuation:

if value in parens:

if under_cursor:

line_tokens[-1] = (Parenthesis.UnderCursor, value)

# Push marker on the stack

stack.append((Parenthesis, value))

else:

stack.append((line, len(line_tokens) - 1,

line_tokens, value))

elif value in parens.itervalues():

saved_stack = list(stack)

try:

while True:

opening = stack.pop()

if parens[opening[-1]] == value:

break

except IndexError:

# SyntaxError.. more closed parentheses than

# opened or a wrong closing paren

opening = None

if not saved_stack:

search_for_paren = False

else:

stack = saved_stack

if opening and opening[0] is Parenthesis:

# Marker found

line_tokens[-1] = (Parenthesis, value)

search_for_paren = False

elif opening and under_cursor and not newline:

if self.cpos:

line_tokens[-1] = (Parenthesis.UnderCursor, value)

else:

# The cursor is at the end of line and next to

# the paren, so it doesn't reverse the paren.

# Therefore, we insert the Parenthesis token

# here instead of the Parenthesis.UnderCursor

# token.

line_tokens[-1] = (Parenthesis, value)

(lineno, i, tokens, opening) = opening

if lineno == len(self.buffer):

self.highlighted_paren = (lineno, saved_tokens)

line_tokens[i] = (Parenthesis, opening)

else:

self.highlighted_paren = (lineno, list(tokens))

# We need to redraw a line

tokens[i] = (Parenthesis, opening)

self.reprint_line(lineno, tokens)

search_for_paren = False

elif under_cursor:

search_for_paren = False

if line != len(self.buffer):

return list()

return line_tokens

def clear_current_line(self):

"""This is used as the exception callback for the Interpreter instance.

It prevents autoindentation from occuring after a traceback."""

def next_indentantion(line):

"""Given a code line, return the indentation of the next line."""

line = line.expandtabs(OPTS.tab_length)

indentation = (len(line) - len(line.lstrip(' '))) // OPTS.tab_length

if line.rstrip().endswith(':'):

indentation += 1

return indentation

def next_token_inside_string(s, inside_string):

"""Given a code string s and an initial state inside_string, return

whether the next token will be inside a string or not."""

for token, value in PythonLexer().get_tokens(s):

if token is Token.String:

value = value.lstrip('bBrRuU')

if value in ['"""', "'''", '"', "'"]:

if not inside_string:

inside_string = value

elif value == inside_string:

inside_string = False

return inside_string

def split_lines(tokens):

for (token, value) in tokens:

if not value:

continue

while value:

head, newline, value = value.partition('\n')

yield (token, head)

if newline:

yield (Token.Text, newline)

def token_is(token_type):

"""Return a callable object that returns whether a token is of the

given type `token_type`."""

def token_is_type(token):

"""Return whether a token is of a certain type or not."""

token = token[0]

while token is not token_type and token.parent:

token = token.parent

return token is token_type

return token_is_type

def token_is_any_of(token_types):

"""Return a callable object that returns whether a token is any of the

given types `token_types`."""

is_token_types = map(token_is, token_types)

def token_is_any_of(token):

return any(check(token) for check in is_token_types)

return token_is_any_of