help_end transformer shouldn't pick up ? in multiline string

Closes gh-4134

IPython/core/inputtransformer.py

@@ -1,502 +1,531 b''
 import abc
 import functools
 import re
 from StringIO import StringIO
 
 from IPython.core.splitinput import LineInfo
 from IPython.utils import tokenize2
 from IPython.utils.openpy import cookie_comment_re
 from IPython.utils.tokenize2 import generate_tokens, untokenize, TokenError
 
 #-----------------------------------------------------------------------------
 # Globals
 #-----------------------------------------------------------------------------
 
 # The escape sequences that define the syntax transformations IPython will
 # apply to user input.  These can NOT be just changed here: many regular
 # expressions and other parts of the code may use their hardcoded values, and
 # for all intents and purposes they constitute the 'IPython syntax', so they
 # should be considered fixed.
 
 ESC_SHELL  = '!'     # Send line to underlying system shell
 ESC_SH_CAP = '!!'    # Send line to system shell and capture output
 ESC_HELP   = '?'     # Find information about object
 ESC_HELP2  = '??'    # Find extra-detailed information about object
 ESC_MAGIC  = '%'     # Call magic function
 ESC_MAGIC2 = '%%'    # Call cell-magic function
 ESC_QUOTE  = ','     # Split args on whitespace, quote each as string and call
 ESC_QUOTE2 = ';'     # Quote all args as a single string, call
 ESC_PAREN  = '/'     # Call first argument with rest of line as arguments
 
 ESC_SEQUENCES = [ESC_SHELL, ESC_SH_CAP, ESC_HELP ,\
                  ESC_HELP2, ESC_MAGIC, ESC_MAGIC2,\
                  ESC_QUOTE, ESC_QUOTE2, ESC_PAREN ]
 
 
 class InputTransformer(object):
     """Abstract base class for line-based input transformers."""
     __metaclass__ = abc.ABCMeta
     
     @abc.abstractmethod
     def push(self, line):
         """Send a line of input to the transformer, returning the transformed
         input or None if the transformer is waiting for more input.
         
         Must be overridden by subclasses.
         """
         pass
     
     @abc.abstractmethod
     def reset(self):
         """Return, transformed any lines that the transformer has accumulated,
         and reset its internal state.
         
         Must be overridden by subclasses.
         """
         pass
     
     @classmethod
     def wrap(cls, func):
         """Can be used by subclasses as a decorator, to return a factory that
         will allow instantiation with the decorated object.
         """
         @functools.wraps(func)
         def transformer_factory(**kwargs):
             return cls(func, **kwargs)
         
         return transformer_factory
 
 class StatelessInputTransformer(InputTransformer):
     """Wrapper for a stateless input transformer implemented as a function."""
     def __init__(self, func):
         self.func = func
     
     def __repr__(self):
         return "StatelessInputTransformer(func={0!r})".format(self.func)
     
     def push(self, line):
         """Send a line of input to the transformer, returning the
         transformed input."""
         return self.func(line)
     
     def reset(self):
         """No-op - exists for compatibility."""
         pass
 
 class CoroutineInputTransformer(InputTransformer):
     """Wrapper for an input transformer implemented as a coroutine."""
     def __init__(self, coro, **kwargs):
         # Prime it
         self.coro = coro(**kwargs)
         next(self.coro)
     
     def __repr__(self):
         return "CoroutineInputTransformer(coro={0!r})".format(self.coro)
     
     def push(self, line):
         """Send a line of input to the transformer, returning the
         transformed input or None if the transformer is waiting for more
         input.
         """
         return self.coro.send(line)
     
     def reset(self):
         """Return, transformed any lines that the transformer has
         accumulated, and reset its internal state.
         """
         return self.coro.send(None)
 
 class TokenInputTransformer(InputTransformer):
     """Wrapper for a token-based input transformer.
     
     func should accept a list of tokens (5-tuples, see tokenize docs), and
     return an iterable which can be passed to tokenize.untokenize().
     """
     def __init__(self, func):
         self.func = func
         self.current_line = ""
         self.line_used = False
         self.reset_tokenizer()
     
     def reset_tokenizer(self):
         self.tokenizer = generate_tokens(self.get_line)
     
     def get_line(self):
         if self.line_used:
             raise TokenError
         self.line_used = True
         return self.current_line
     
     def push(self, line):
         self.current_line += line + "\n"
         if self.current_line.isspace():
             return self.reset()
         
         self.line_used = False
         tokens = []
         stop_at_NL = False
         try:
             for intok in self.tokenizer:
                 tokens.append(intok)
                 t = intok[0]
                 if t == tokenize2.NEWLINE or (stop_at_NL and t == tokenize2.NL):
                     # Stop before we try to pull a line we don't have yet
                     break
                 elif t == tokenize2.ERRORTOKEN:
                     stop_at_NL = True
         except TokenError:
             # Multi-line statement - stop and try again with the next line
             self.reset_tokenizer()
             return None
         
         return self.output(tokens)
     
     def output(self, tokens):
         self.current_line = ""
         self.reset_tokenizer()
         return untokenize(self.func(tokens)).rstrip('\n')
     
     def reset(self):
         l = self.current_line
         self.current_line = ""
         self.reset_tokenizer()
         if l:
             return l.rstrip('\n')
 
 class assemble_python_lines(TokenInputTransformer):
     def __init__(self):
         super(assemble_python_lines, self).__init__(None)
     
     def output(self, tokens):
         return self.reset()
 
 @CoroutineInputTransformer.wrap
 def assemble_logical_lines():
     """Join lines following explicit line continuations (\)"""
     line = ''
     while True:
         line = (yield line)
         if not line or line.isspace():
             continue
         
         parts = []
         while line is not None:
             if line.endswith('\\') and (not has_comment(line)):
                 parts.append(line[:-1])
                 line = (yield None) # Get another line
             else:
                 parts.append(line)
                 break
         
         # Output
         line = ''.join(parts)
 
 # Utilities
 def _make_help_call(target, esc, lspace, next_input=None):
     """Prepares a pinfo(2)/psearch call from a target name and the escape
     (i.e. ? or ??)"""
     method  = 'pinfo2' if esc == '??' \
                 else 'psearch' if '*' in target \
                 else 'pinfo'
     arg = " ".join([method, target])
     if next_input is None:
         return '%sget_ipython().magic(%r)' % (lspace, arg)
     else:
         return '%sget_ipython().set_next_input(%r);get_ipython().magic(%r)' % \
            (lspace, next_input, arg)
     
 # These define the transformations for the different escape characters.
 def _tr_system(line_info):
     "Translate lines escaped with: !"
     cmd = line_info.line.lstrip().lstrip(ESC_SHELL)
     return '%sget_ipython().system(%r)' % (line_info.pre, cmd)
 
 def _tr_system2(line_info):
     "Translate lines escaped with: !!"
     cmd = line_info.line.lstrip()[2:]
     return '%sget_ipython().getoutput(%r)' % (line_info.pre, cmd)
 
 def _tr_help(line_info):
     "Translate lines escaped with: ?/??"
     # A naked help line should just fire the intro help screen
     if not line_info.line[1:]:
         return 'get_ipython().show_usage()'
 
     return _make_help_call(line_info.ifun, line_info.esc, line_info.pre)
 
 def _tr_magic(line_info):
     "Translate lines escaped with: %"
     tpl = '%sget_ipython().magic(%r)'
     if line_info.line.startswith(ESC_MAGIC2):
         return line_info.line
     cmd = ' '.join([line_info.ifun, line_info.the_rest]).strip()
     return tpl % (line_info.pre, cmd)
 
 def _tr_quote(line_info):
     "Translate lines escaped with: ,"
     return '%s%s("%s")' % (line_info.pre, line_info.ifun,
                          '", "'.join(line_info.the_rest.split()) )
 
 def _tr_quote2(line_info):
     "Translate lines escaped with: ;"
     return '%s%s("%s")' % (line_info.pre, line_info.ifun,
                            line_info.the_rest)
 
 def _tr_paren(line_info):
     "Translate lines escaped with: /"
     return '%s%s(%s)' % (line_info.pre, line_info.ifun,
                          ", ".join(line_info.the_rest.split()))
 
 tr = { ESC_SHELL  : _tr_system,
        ESC_SH_CAP : _tr_system2,
        ESC_HELP   : _tr_help,
        ESC_HELP2  : _tr_help,
        ESC_MAGIC  : _tr_magic,
        ESC_QUOTE  : _tr_quote,
        ESC_QUOTE2 : _tr_quote2,
        ESC_PAREN  : _tr_paren }
 
 @StatelessInputTransformer.wrap
 def escaped_commands(line):
     """Transform escaped commands - %magic, !system, ?help + various autocalls.
     """
     if not line or line.isspace():
         return line
     lineinf = LineInfo(line)
     if lineinf.esc not in tr:
         return line
     
     return tr[lineinf.esc](lineinf)
 
 _initial_space_re = re.compile(r'\s*')
 
 _help_end_re = re.compile(r"""(%{0,2}
                               [a-zA-Z_*][\w*]*        # Variable name
                               (\.[a-zA-Z_*][\w*]*)*   # .etc.etc
                               )
                               (\?\??)$                # ? or ??
                               """,
                               re.VERBOSE)
 
+# Extra pseudotokens for multiline strings and data structures
+_MULTILINE_STRING = object()
+_MULTILINE_STRUCTURE = object()
+
+def _line_tokens(line):
+    """Helper for has_comment and ends_in_comment_or_string."""
+    readline = StringIO(line).readline
+    toktypes = set()
+    try:
+        for t in generate_tokens(readline):
+            toktypes.add(t[0])
+    except TokenError as e:
+        # There are only two cases where a TokenError is raised.
+        if 'multi-line string' in e.args[0]:
+            toktypes.add(_MULTILINE_STRING)
+        else:
+            toktypes.add(_MULTILINE_STRUCTURE)
+    return toktypes
+
 def has_comment(src):
     """Indicate whether an input line has (i.e. ends in, or is) a comment.
 
     This uses tokenize, so it can distinguish comments from # inside strings.
 
     Parameters
     ----------
     src : string
       A single line input string.
 
     Returns
     -------
     comment : bool
         True if source has a comment.
     """
-    readline = StringIO(src).readline
-    toktypes = set()
-    try:
-        for t in generate_tokens(readline):
-            toktypes.add(t[0])
-    except TokenError:
-        pass
-    return(tokenize2.COMMENT in toktypes)
+    return (tokenize2.COMMENT in _line_tokens(src))
 
+def ends_in_comment_or_string(src):
+    """Indicates whether or not an input line ends in a comment or within
+    a multiline string.
+    
+    Parameters
+    ----------
+    src : string
+      A single line input string.
+
+    Returns
+    -------
+    comment : bool
+        True if source ends in a comment or multiline string.
+    """
+    toktypes = _line_tokens(src)
+    return (tokenize2.COMMENT in toktypes) or (_MULTILINE_STRING in toktypes)
+        
 
 @StatelessInputTransformer.wrap
 def help_end(line):
     """Translate lines with ?/?? at the end"""
     m = _help_end_re.search(line)
-    if m is None or has_comment(line):
+    if m is None or ends_in_comment_or_string(line):
         return line
     target = m.group(1)
     esc = m.group(3)
     lspace = _initial_space_re.match(line).group(0)
 
     # If we're mid-command, put it back on the next prompt for the user.
     next_input = line.rstrip('?') if line.strip() != m.group(0) else None
 
     return _make_help_call(target, esc, lspace, next_input)
 
 
 @CoroutineInputTransformer.wrap
 def cellmagic(end_on_blank_line=False):
     """Captures & transforms cell magics.
     
     After a cell magic is started, this stores up any lines it gets until it is
     reset (sent None).
     """
     tpl = 'get_ipython().run_cell_magic(%r, %r, %r)'
     cellmagic_help_re = re.compile('%%\w+\?')
     line = ''
     while True:
         line = (yield line)
         # consume leading empty lines
         while not line:
             line = (yield line)
         
         if not line.startswith(ESC_MAGIC2):
             # This isn't a cell magic, idle waiting for reset then start over
             while line is not None:
                 line = (yield line)
             continue
         
         if cellmagic_help_re.match(line):
             # This case will be handled by help_end
             continue
         
         first = line
         body = []
         line = (yield None)
         while (line is not None) and \
                                 ((line.strip() != '') or not end_on_blank_line):
             body.append(line)
             line = (yield None)
         
         # Output
         magic_name, _, first = first.partition(' ')
         magic_name = magic_name.lstrip(ESC_MAGIC2)
         line = tpl % (magic_name, first, u'\n'.join(body))
 
 
 def _strip_prompts(prompt_re, initial_re=None):
     """Remove matching input prompts from a block of input.
     
     Parameters
     ----------
     prompt_re : regular expression
         A regular expression matching any input prompt (including continuation)
     initial_re : regular expression, optional
         A regular expression matching only the initial prompt, but not continuation.
         If no initial expression is given, prompt_re will be used everywhere.
         Used mainly for plain Python prompts, where the continuation prompt
         ``...`` is a valid Python expression in Python 3, so shouldn't be stripped.
     
     If initial_re and prompt_re differ,
     only initial_re will be tested against the first line.
     If any prompt is found on the first two lines,
     prompts will be stripped from the rest of the block.
     """
     if initial_re is None:
         initial_re = prompt_re
     line = ''
     while True:
         line = (yield line)
         
         # First line of cell
         if line is None:
             continue
         out, n1 = initial_re.subn('', line, count=1)
         line = (yield out)
         
         if line is None:
             continue
         # check for any prompt on the second line of the cell,
         # because people often copy from just after the first prompt,
         # so we might not see it in the first line.
         out, n2 = prompt_re.subn('', line, count=1)
         line = (yield out)
         
         if n1 or n2:
             # Found a prompt in the first two lines - check for it in
             # the rest of the cell as well.
             while line is not None:
                 line = (yield prompt_re.sub('', line, count=1))
         
         else:
             # Prompts not in input - wait for reset
             while line is not None:
                 line = (yield line)
 
 @CoroutineInputTransformer.wrap
 def classic_prompt():
     """Strip the >>>/... prompts of the Python interactive shell."""
     # FIXME: non-capturing version (?:...) usable?
     prompt_re = re.compile(r'^(>>> ?|\.\.\. ?)')
     initial_re = re.compile(r'^(>>> ?)')
     return _strip_prompts(prompt_re, initial_re)
 
 @CoroutineInputTransformer.wrap
 def ipy_prompt():
     """Strip IPython's In [1]:/...: prompts."""
     # FIXME: non-capturing version (?:...) usable?
     # FIXME: r'^(In \[\d+\]: | {3}\.{3,}: )' clearer?
     prompt_re = re.compile(r'^(In \[\d+\]: |\ \ \ \.\.\.+: )')
     return _strip_prompts(prompt_re)
 
 
 @CoroutineInputTransformer.wrap
 def leading_indent():
     """Remove leading indentation.
     
     If the first line starts with a spaces or tabs, the same whitespace will be
     removed from each following line until it is reset.
     """
     space_re = re.compile(r'^[ \t]+')
     line = ''
     while True:
         line = (yield line)
         
         if line is None:
             continue
         
         m = space_re.match(line)
         if m:
             space = m.group(0)
             while line is not None:
                 if line.startswith(space):
                     line = line[len(space):]
                 line = (yield line)
         else:
             # No leading spaces - wait for reset
             while line is not None:
                 line = (yield line)
 
 
 @CoroutineInputTransformer.wrap
 def strip_encoding_cookie():
     """Remove encoding comment if found in first two lines
     
     If the first or second line has the `# coding: utf-8` comment,
     it will be removed.
     """
     line = ''
     while True:
         line = (yield line)
         # check comment on first two lines
         for i in range(2):
             if line is None:
                 break
             if cookie_comment_re.match(line):
                 line = (yield "")
             else:
                 line = (yield line)
         
         # no-op on the rest of the cell
         while line is not None:
             line = (yield line)
 
 
 assign_system_re = re.compile(r'(?P<lhs>(\s*)([\w\.]+)((\s*,\s*[\w\.]+)*))'
                               r'\s*=\s*!\s*(?P<cmd>.*)')
 assign_system_template = '%s = get_ipython().getoutput(%r)'
 @StatelessInputTransformer.wrap
 def assign_from_system(line):
     """Transform assignment from system commands (e.g. files = !ls)"""
     m = assign_system_re.match(line)
     if m is None:
         return line
     
     return assign_system_template % m.group('lhs', 'cmd')
 
 assign_magic_re = re.compile(r'(?P<lhs>(\s*)([\w\.]+)((\s*,\s*[\w\.]+)*))'
                              r'\s*=\s*%\s*(?P<cmd>.*)')
 assign_magic_template = '%s = get_ipython().magic(%r)'
 @StatelessInputTransformer.wrap
 def assign_from_magic(line):
     """Transform assignment from magic commands (e.g. a = %who_ls)"""
     m = assign_magic_re.match(line)
     if m is None:
         return line
     
     return assign_magic_template % m.group('lhs', 'cmd')

IPython/core/tests/test_inputtransformer.py

@@ -1,464 +1,470 b''
 import tokenize
 import nose.tools as nt
 
 from IPython.testing import tools as tt
 from IPython.utils import py3compat
 u_fmt = py3compat.u_format
 
 from IPython.core import inputtransformer as ipt
 
 def transform_and_reset(transformer):
     transformer = transformer()
     def transform(inp):
         try:
             return transformer.push(inp)
         finally:
             transformer.reset()
     
     return transform
 
 # Transformer tests
 def transform_checker(tests, transformer, **kwargs):
     """Utility to loop over test inputs"""
     transformer = transformer(**kwargs)
     try:
         for inp, tr in tests:
             if inp is None:
                 out = transformer.reset()
             else:
                 out = transformer.push(inp)
             nt.assert_equal(out, tr)
     finally:
         transformer.reset()
 
 # Data for all the syntax tests in the form of lists of pairs of
 # raw/transformed input.  We store it here as a global dict so that we can use
 # it both within single-function tests and also to validate the behavior of the
 # larger objects
 
 syntax = \
   dict(assign_system =
        [(i,py3compat.u_format(o)) for i,o in \
        [(u'a =! ls', "a = get_ipython().getoutput({u}'ls')"),
         (u'b = !ls', "b = get_ipython().getoutput({u}'ls')"),
         ('x=1', 'x=1'), # normal input is unmodified
         ('    ','    '),  # blank lines are kept intact
         ]],
 
        assign_magic =
        [(i,py3compat.u_format(o)) for i,o in \
        [(u'a =% who', "a = get_ipython().magic({u}'who')"),
         (u'b = %who', "b = get_ipython().magic({u}'who')"),
         ('x=1', 'x=1'), # normal input is unmodified
         ('    ','    '),  # blank lines are kept intact
         ]],
 
        classic_prompt =
        [('>>> x=1', 'x=1'),
         ('x=1', 'x=1'), # normal input is unmodified
         ('    ', '    '),  # blank lines are kept intact
         ],
 
        ipy_prompt =
        [('In [1]: x=1', 'x=1'),
         ('x=1', 'x=1'), # normal input is unmodified
         ('    ','    '),  # blank lines are kept intact
         ],
 
        strip_encoding_cookie =
        [
         ('# -*- encoding: utf-8 -*-', ''),
         ('# coding: latin-1', ''),
        ],
 
 
        # Tests for the escape transformer to leave normal code alone
        escaped_noesc =
        [ ('    ', '    '),
          ('x=1', 'x=1'),
          ],
 
        # System calls
        escaped_shell =
        [(i,py3compat.u_format(o)) for i,o in \
        [ (u'!ls', "get_ipython().system({u}'ls')"),
          # Double-escape shell, this means to capture the output of the
          # subprocess and return it
          (u'!!ls', "get_ipython().getoutput({u}'ls')"),
          ]],
 
        # Help/object info
        escaped_help =
        [(i,py3compat.u_format(o)) for i,o in \
        [ (u'?', 'get_ipython().show_usage()'),
          (u'?x1', "get_ipython().magic({u}'pinfo x1')"),
          (u'??x2', "get_ipython().magic({u}'pinfo2 x2')"),
          (u'?a.*s', "get_ipython().magic({u}'psearch a.*s')"),
          (u'?%hist1', "get_ipython().magic({u}'pinfo %hist1')"),
          (u'?%%hist2', "get_ipython().magic({u}'pinfo %%hist2')"),
          (u'?abc = qwe', "get_ipython().magic({u}'pinfo abc')"),
          ]],
 
       end_help =
       [(i,py3compat.u_format(o)) for i,o in \
       [ (u'x3?', "get_ipython().magic({u}'pinfo x3')"),
         (u'x4??', "get_ipython().magic({u}'pinfo2 x4')"),
         (u'%hist1?', "get_ipython().magic({u}'pinfo %hist1')"),
         (u'%hist2??', "get_ipython().magic({u}'pinfo2 %hist2')"),
         (u'%%hist3?', "get_ipython().magic({u}'pinfo %%hist3')"),
         (u'%%hist4??', "get_ipython().magic({u}'pinfo2 %%hist4')"),
         (u'f*?', "get_ipython().magic({u}'psearch f*')"),
         (u'ax.*aspe*?', "get_ipython().magic({u}'psearch ax.*aspe*')"),
         (u'a = abc?', "get_ipython().set_next_input({u}'a = abc');"
                       "get_ipython().magic({u}'pinfo abc')"),
         (u'a = abc.qe??', "get_ipython().set_next_input({u}'a = abc.qe');"
                           "get_ipython().magic({u}'pinfo2 abc.qe')"),
         (u'a = *.items?', "get_ipython().set_next_input({u}'a = *.items');"
                           "get_ipython().magic({u}'psearch *.items')"),
         (u'plot(a?', "get_ipython().set_next_input({u}'plot(a');"
                      "get_ipython().magic({u}'pinfo a')"),
         (u'a*2 #comment?', 'a*2 #comment?'),
         ]],
 
        # Explicit magic calls
        escaped_magic =
        [(i,py3compat.u_format(o)) for i,o in \
        [ (u'%cd', "get_ipython().magic({u}'cd')"),
          (u'%cd /home', "get_ipython().magic({u}'cd /home')"),
          # Backslashes need to be escaped.
          (u'%cd C:\\User', "get_ipython().magic({u}'cd C:\\\\User')"),
          (u'    %magic', "    get_ipython().magic({u}'magic')"),
          ]],
 
        # Quoting with separate arguments
        escaped_quote =
        [ (',f', 'f("")'),
          (',f x', 'f("x")'),
          ('  ,f y', '  f("y")'),
          (',f a b', 'f("a", "b")'),
          ],
 
        # Quoting with single argument
        escaped_quote2 =
        [ (';f', 'f("")'),
          (';f x', 'f("x")'),
          ('  ;f y', '  f("y")'),
          (';f a b', 'f("a b")'),
          ],
 
        # Simply apply parens
        escaped_paren =
        [ ('/f', 'f()'),
          ('/f x', 'f(x)'),
          ('  /f y', '  f(y)'),
          ('/f a b', 'f(a, b)'),
          ],
 
        # Check that we transform prompts before other transforms
        mixed =
        [(i,py3compat.u_format(o)) for i,o in \
        [ (u'In [1]: %lsmagic', "get_ipython().magic({u}'lsmagic')"),
          (u'>>> %lsmagic', "get_ipython().magic({u}'lsmagic')"),
          (u'In [2]: !ls', "get_ipython().system({u}'ls')"),
          (u'In [3]: abs?', "get_ipython().magic({u}'pinfo abs')"),
          (u'In [4]: b = %who', "b = get_ipython().magic({u}'who')"),
          ]],
        )
 
 # multiline syntax examples.  Each of these should be a list of lists, with
 # each entry itself having pairs of raw/transformed input.  The union (with
 # '\n'.join() of the transformed inputs is what the splitter should produce
 # when fed the raw lines one at a time via push.
 syntax_ml = \
   dict(classic_prompt =
        [ [('>>> for i in range(10):','for i in range(10):'),
           ('...     print i','    print i'),
           ('... ', ''),
           ],
          [('>>> a="""','a="""'),
           ('... 123"""','123"""'),
           ],
          [('a="""','a="""'),
           ('... 123','123'),
           ('... 456"""','456"""'),
           ],
          [('a="""','a="""'),
           ('>>> 123','123'),
           ('... 456"""','456"""'),
           ],
          [('a="""','a="""'),
           ('123','123'),
           ('... 456"""','... 456"""'),
           ],
          [('....__class__','....__class__'),
          ],
          [('a=5', 'a=5'),
           ('...', ''),
          ],
          [('>>> def f(x):', 'def f(x):'),
           ('...', ''),
           ('...     return x', '    return x'),
           ],
         ],
 
        ipy_prompt =
        [ [('In [24]: for i in range(10):','for i in range(10):'),
           ('   ....:     print i','    print i'),
           ('   ....: ', ''),
           ],
          [('In [2]: a="""','a="""'),
           ('   ...: 123"""','123"""'),
           ],
          [('a="""','a="""'),
           ('   ...: 123','123'),
           ('   ...: 456"""','456"""'),
           ],
          [('a="""','a="""'),
           ('In [1]: 123','123'),
           ('   ...: 456"""','456"""'),
           ],
          [('a="""','a="""'),
           ('123','123'),
           ('   ...: 456"""','   ...: 456"""'),
           ],
          ],
 
        strip_encoding_cookie =
        [
         [
             ('# -*- coding: utf-8 -*-', ''),
             ('foo', 'foo'),
         ],
         [
             ('#!/usr/bin/env python', '#!/usr/bin/env python'),
             ('# -*- coding: latin-1 -*-', ''),
             # only the first-two lines
             ('# -*- coding: latin-1 -*-', '# -*- coding: latin-1 -*-'),
         ],
        ],
 
        multiline_datastructure_prompt =
        [ [('>>> a = [1,','a = [1,'),
           ('... 2]','2]'),
          ],
        ],
         
        multiline_datastructure =
        [ [('b = ("%s"', None),
           ('# comment', None),
           ('%foo )', 'b = ("%s"\n# comment\n%foo )'),
          ],
        ],
        
+       multiline_string =
+       [ [("'''foo?", None),
+          ("bar'''", "'''foo?\nbar'''"),
+         ],
+       ],
+       
        leading_indent =
        [ [('    print "hi"','print "hi"'),
           ],
          [('  for a in range(5):','for a in range(5):'),
           ('    a*2','  a*2'),
           ],
          [('    a="""','a="""'),
           ('    123"""','123"""'),
            ],
          [('a="""','a="""'),
           ('    123"""','    123"""'),
           ],
        ],
        
        cellmagic =
        [ [(u'%%foo a', None),
           (None, u_fmt("get_ipython().run_cell_magic({u}'foo', {u}'a', {u}'')")),
           ],
          [(u'%%bar 123', None),
           (u'hello', None),
           (None , u_fmt("get_ipython().run_cell_magic({u}'bar', {u}'123', {u}'hello')")),
           ],
          [(u'a=5', 'a=5'),
           (u'%%cellmagic', '%%cellmagic'),
           ],
        ],
        
        escaped =
        [ [('%abc def \\', None),
           ('ghi', u_fmt("get_ipython().magic({u}'abc def ghi')")),
           ],
          [('%abc def \\', None),
           ('ghi\\', None),
           (None, u_fmt("get_ipython().magic({u}'abc def ghi')")),
           ],
        ],
        
        assign_magic =
        [ [(u'a = %bc de \\', None),
           (u'fg', u_fmt("a = get_ipython().magic({u}'bc de fg')")),
           ],
          [(u'a = %bc de \\', None),
           (u'fg\\', None),
           (None, u_fmt("a = get_ipython().magic({u}'bc de fg')")),
           ],
        ],
        
        assign_system =
        [ [(u'a = !bc de \\', None),
           (u'fg', u_fmt("a = get_ipython().getoutput({u}'bc de fg')")),
           ],
          [(u'a = !bc de \\', None),
           (u'fg\\', None),
           (None, u_fmt("a = get_ipython().getoutput({u}'bc de fg')")),
           ],
        ],
        )
 
 
 def test_assign_system():
     tt.check_pairs(transform_and_reset(ipt.assign_from_system), syntax['assign_system'])
 
 def test_assign_magic():
     tt.check_pairs(transform_and_reset(ipt.assign_from_magic), syntax['assign_magic'])
 
 def test_classic_prompt():
     tt.check_pairs(transform_and_reset(ipt.classic_prompt), syntax['classic_prompt'])
     for example in syntax_ml['classic_prompt']:
         transform_checker(example, ipt.classic_prompt)
     for example in syntax_ml['multiline_datastructure_prompt']:
         transform_checker(example, ipt.classic_prompt)
 
 
 def test_ipy_prompt():
     tt.check_pairs(transform_and_reset(ipt.ipy_prompt), syntax['ipy_prompt'])
     for example in syntax_ml['ipy_prompt']:
         transform_checker(example, ipt.ipy_prompt)
 
 def test_coding_cookie():
     tt.check_pairs(transform_and_reset(ipt.strip_encoding_cookie), syntax['strip_encoding_cookie'])
     for example in syntax_ml['strip_encoding_cookie']:
         transform_checker(example, ipt.strip_encoding_cookie)
 
 def test_assemble_logical_lines():
     tests = \
     [ [(u"a = \\", None),
        (u"123", u"a = 123"),
       ],
       [(u"a = \\", None),  # Test resetting when within a multi-line string
        (u"12 *\\", None),
        (None, u"a = 12 *"),
       ],
       [(u"# foo\\", u"# foo\\"), # Comments can't be continued like this
       ],
     ]
     for example in tests:
         transform_checker(example, ipt.assemble_logical_lines)
 
 def test_assemble_python_lines():
     tests = \
     [ [(u"a = '''", None),
        (u"abc'''", u"a = '''\nabc'''"),
       ],
       [(u"a = '''", None),  # Test resetting when within a multi-line string
        (u"def", None),
        (None, u"a = '''\ndef"),
       ],
       [(u"a = [1,", None),
        (u"2]", u"a = [1,\n2]"),
       ],
       [(u"a = [1,", None),  # Test resetting when within a multi-line string
        (u"2,", None),
        (None, u"a = [1,\n2,"),
       ],
     ] + syntax_ml['multiline_datastructure']
     for example in tests:
         transform_checker(example, ipt.assemble_python_lines)
 
 
 def test_help_end():
     tt.check_pairs(transform_and_reset(ipt.help_end), syntax['end_help'])
 
 def test_escaped_noesc():
     tt.check_pairs(transform_and_reset(ipt.escaped_commands), syntax['escaped_noesc'])
 
 
 def test_escaped_shell():
     tt.check_pairs(transform_and_reset(ipt.escaped_commands), syntax['escaped_shell'])
 
 
 def test_escaped_help():
     tt.check_pairs(transform_and_reset(ipt.escaped_commands), syntax['escaped_help'])
 
 
 def test_escaped_magic():
     tt.check_pairs(transform_and_reset(ipt.escaped_commands), syntax['escaped_magic'])
 
 
 def test_escaped_quote():
     tt.check_pairs(transform_and_reset(ipt.escaped_commands), syntax['escaped_quote'])
 
 
 def test_escaped_quote2():
     tt.check_pairs(transform_and_reset(ipt.escaped_commands), syntax['escaped_quote2'])
 
 
 def test_escaped_paren():
     tt.check_pairs(transform_and_reset(ipt.escaped_commands), syntax['escaped_paren'])
 
 
 def test_cellmagic():
     for example in syntax_ml['cellmagic']:
         transform_checker(example, ipt.cellmagic)
     
     line_example = [(u'%%bar 123', None),
                     (u'hello', None),
                     (u'' , u_fmt("get_ipython().run_cell_magic({u}'bar', {u}'123', {u}'hello')")),
                    ]
     transform_checker(line_example, ipt.cellmagic, end_on_blank_line=True)
 
 def test_has_comment():
     tests = [('text', False),
              ('text #comment', True),
              ('text #comment\n', True),
              ('#comment', True),
              ('#comment\n', True),
              ('a = "#string"', False),
              ('a = "#string" # comment', True),
              ('a #comment not "string"', True),
              ]
     tt.check_pairs(ipt.has_comment, tests)
 
 @ipt.TokenInputTransformer.wrap
 def decistmt(tokens):
     """Substitute Decimals for floats in a string of statements.
 
     Based on an example from the tokenize module docs.
     """
     result = []
     for toknum, tokval, _, _, _  in tokens:
         if toknum == tokenize.NUMBER and '.' in tokval:  # replace NUMBER tokens
             for newtok in [
                 (tokenize.NAME, 'Decimal'),
                 (tokenize.OP, '('),
                 (tokenize.STRING, repr(tokval)),
                 (tokenize.OP, ')')
             ]:
                 yield newtok
         else:
             yield (toknum, tokval)
 
 
 
 def test_token_input_transformer():
     tests = [(u'1.2', u_fmt(u"Decimal ({u}'1.2')")),
              (u'"1.2"', u'"1.2"'),
              ]
     tt.check_pairs(transform_and_reset(decistmt), tests)
     ml_tests = \
     [ [(u"a = 1.2; b = '''x", None),
        (u"y'''", u_fmt(u"a =Decimal ({u}'1.2');b ='''x\ny'''")),
       ],
       [(u"a = [1.2,", None),
        (u"3]", u_fmt(u"a =[Decimal ({u}'1.2'),\n3 ]")),
       ],
       [(u"a = '''foo", None),  # Test resetting when within a multi-line string
        (u"bar", None),
        (None, u"a = '''foo\nbar"),
       ],
     ]
     for example in ml_tests:
         transform_checker(example, decistmt)