Issue with multiline TokenInputTransformer in IPython extension - ipython

Question

I am trying to use a TokenInputTransformer within an IPython extension module, but it seems that there is something wrong with the standard implementation of token transformers with multiline input. Consider the following minimal extension:
from IPython.core.inputtransformer import TokenInputTransformer
#TokenInputTransformer.wrap
def test_transformer(tokens):
return tokens
def load_ipython_extension(ip):
for s in (ip.input_splitter, ip.input_transformer_manager):
s.python_line_transforms.extend([test_transformer()])
print "Test activated"
When I load the extension in IPython 1.1.0 I get a non-handled exception with multiline input:
In [1]: %load_ext test
Test activated
In [2]: abs(
...: 2
...: )
Traceback (most recent call last):
File "/Applications/anaconda/bin/ipython", line 6, in <module>
sys.exit(start_ipython())
File "/Applications/anaconda/lib/python2.7/site-packages/IPython/__init__.py", line 118, in start_ipython
return launch_new_instance(argv=argv, **kwargs)
File "/Applications/anaconda/lib/python2.7/site-packages/IPython/config/application.py", line 545, in launch_instance
app.start()
File "/Applications/anaconda/lib/python2.7/site-packages/IPython/terminal/ipapp.py", line 362, in start
self.shell.mainloop()
File "/Applications/anaconda/lib/python2.7/site-packages/IPython/terminal/interactiveshell.py", line 436, in mainloop
self.interact(display_banner=display_banner)
File "/Applications/anaconda/lib/python2.7/site-packages/IPython/terminal/interactiveshell.py", line 548, in interact
self.input_splitter.push(line)
File "/Applications/anaconda/lib/python2.7/site-packages/IPython/core/inputsplitter.py", line 620, in push
out = self.push_line(line)
File "/Applications/anaconda/lib/python2.7/site-packages/IPython/core/inputsplitter.py", line 655, in push_line
line = transformer.push(line)
File "/Applications/anaconda/lib/python2.7/site-packages/IPython/core/inputtransformer.py", line 152, in push
return self.output(tokens)
File "/Applications/anaconda/lib/python2.7/site-packages/IPython/core/inputtransformer.py", line 157, in output
return untokenize(self.func(tokens)).rstrip('\n')
File "/Applications/anaconda/lib/python2.7/site-packages/IPython/utils/_tokenize_py2.py", line 276, in untokenize
return ut.untokenize(iterable)
File "/Applications/anaconda/lib/python2.7/site-packages/IPython/utils/_tokenize_py2.py", line 214, in untokenize
self.add_whitespace(start)
File "/Applications/anaconda/lib/python2.7/site-packages/IPython/utils/_tokenize_py2.py", line 199, in add_whitespace
assert row >= self.prev_row
AssertionError
If you suspect this is an IPython bug, please report it at:
https://github.com/ipython/ipython/issues
or send an email to the mailing list at ipython-dev#scipy.org
You can print a more detailed traceback right now with "%tb", or use "%debug"
to interactively debug it.
Extra-detailed tracebacks for bug-reporting purposes can be enabled via:
%config Application.verbose_crash=True
Am I doing something wrong or is it really an IPython bug?

Answer 1

It is really an IPython bug, I think. Specifically, the way we handle tokenize fails when an expression involving brackets (()[]{}) is spread over more than one line. I'm trying to work out what we can do about this.

Answer 2

Kinda late answer but, I was trying to use it in my own extension and just had the same problem. I've solved it by simply removing NL from the list (it's not the same as NEWLINE token which end statement), NL token only appears inside [], (), {} so it should be safely removable.
from tokenize import NL
#TokenInputTransformer.wrap
def mat_transformer(tokens):
tokens = list(filter(lambda t: t.type != NL, tokens))
return tokens
If you are looking for full example, I've posted my goofy code there: https://github.com/Quinzel/pymat