MetaKernel do_complete

do_complete.py

"""
Completions should be case-sensitive

There are three potential objects:
- the token we are in the middle of
- partial version of that token
- the prior token representing a function call

For completions, use the partial object
For do_help, use the full token or the function call token

"""
import re
import os


def _listdir(root):
    "List directory 'root' appending the path separator to subdirs."
    res = []
    root = os.path.expanduser(root)
    try:
        for name in os.listdir(root):
            path = os.path.join(root, name)
            if os.path.isdir(path):
                name += os.sep
            res.append(name)
    except:
        pass  # no need to report invalid paths
    return res


def _complete_path(path=None):
    """Perform completion of filesystem path.
    http://stackoverflow.com/questions/5637124/tab-completion-in-pythons-raw-input
    """
    if not path:
        return _listdir('.')
    dirname, rest = os.path.split(path)
    tmp = dirname if dirname else '.'
    res = [p for p in _listdir(tmp) if p.startswith(rest)]
    # more than one match, or single match which does not exist (typo)
    if len(res) > 1 or not os.path.exists(path):
        return res
    # resolved to a single directory, so return list of files below it
    if os.path.isdir(path):
        return [p for p in _listdir(path)]
    # exact file match terminates this completion
    return [path + ' ']


class Parser(object):

    def __init__(self, identifier_regex, function_call_regex, magic_prefixes,
                 magic_suffixes):
        self.identifier_regex = identifier_regex
        self.func_call_regex = function_call_regex
        self.magic_prefixes = magic_prefixes
        self.magic_suffixes = magic_suffixes
        self._default_regex = r'[^\d\W]\w*'

    def parse_code(self, code, start=0, end=-1):

        if not code:
            return

        if end == -1:
            end = len(code)
        end = min(len(code), end)

        start = min(start, end)
        start = max(0, start)

        info = dict(code=code, start=start, end=end, pre=code[:start],
                    mid=code[start:end], post=code[end:], magic=dict(),
                    complete_obj='', help_obj='')

        info['magic'] = self.parse_magic(code[:end])

        id_regex = re.compile('(\{0}+{1}|{2})'.format(
            self.magic_prefixes['magic'], self._default_regex,
            self.identifier_regex), re.UNICODE)

        tokens = re.split(id_regex, code[:end], re.UNICODE)

        if not tokens:
            return info

        info['lines'] = lines = code[start:end].splitlines()
        info['line_num'] = line_num = len(lines)

        info['line'] = line = lines[-1]
        info['column'] = col = len(lines[-1])

        tokens = re.findall(id_regex, line)
        if tokens and line.endswith(tokens[-1]):
            obj = tokens[-1]
        else:
            obj = ''

        full_obj = obj

        if obj:
            full_line = code.splitlines()[line_num - 1]
            rest = full_line[col:]
            match = re.match(id_regex, rest)
            if match:
                full_obj = obj + match.group()

        func_call = re.findall(self.func_call_regex, line)
        if func_call and not obj:
            info['help_obj'] = func_call[-1]
            info['help_col'] = line.index(obj) + len(obj)
            info['help_pos'] = end - len(line) + col
        else:
            info['help_obj'] = full_obj
            info['help_col'] = col
            info['help_pos'] = end

        info['complete_obj'] = obj
        info['full_obj'] = full_obj

        return info

    def parse_magic(self, code):
        # find magic characters - help overrides any others
        info = {}
        prefixes = self.magic_prefixes
        suffixes = self.magic_suffixes

        id_regex = '|'.join([self._default_regex, self.identifier_regex])

        pre_magics = {}
        for (name, prefix) in prefixes.items():
            if name == 'shell':
                regex = r'(\%s+)( *)(%s)' % (prefix, id_regex)
            else:
                regex = r'(\%s+)(%s)' % (prefix, id_regex)
            match = re.search(regex, code, re.UNICODE)
            if match:
                pre_magics[name] = match.groups()

        post_magics = {}
        for (name, suffix) in suffixes.items():
            regex = r'(%s)(\%s+)' % (id_regex, suffix)
            match = re.search(regex, code, re.UNICODE)
            if match:
                post_magics[name] = match.groups()

        types = ['none', 'line', 'cell', 'sticky']

        if 'help' in pre_magics:
            info['name'] = 'help'
            pre, obj = pre_magics['help']
            info['type'] = types[len(pre)]
            info['index'] = code.index(pre + obj)

        elif 'help' in post_magics:
            info['name'] = 'help'
            obj, suf = post_magics['help']
            info['type'] = types[len(suf)]
            info['index'] = code.index(obj + suf)

        elif 'magic' in pre_magics:
            pre, obj = pre_magics['magic']
            info['type'] = types[len(pre)]
            info['name'] = obj
            info['index'] = code.index(pre + obj) + len(pre + obj)

        elif 'shell' in pre_magics:
            info['name'] = 'shell'
            pre, ws, obj = pre_magics['shell']
            info['type'] = types[len(pre)]
            info['index'] = code.index(pre + ws + obj) + len(pre + ws)

        else:
            return info

        info['rest'] = code[info['index']:].strip()

        lines = info['rest'].splitlines()
        info['args'] = lines[0].strip()
        info['code'] = '\n'.join(lines[1:])

        return info


def test_parser():
    identifier_regex = r'[^\d\W]\w*'
    function_call_regex = r'([^\d\W][\w\.]*)\([^\)\()]*\Z'
    magic_prefixes = dict(magic='%', shell='!', help='?')
    magic_suffixes = dict(help='?')
    p = Parser(identifier_regex, function_call_regex, magic_prefixes,
               magic_suffixes)

    info = p.parse_code('import nump')
    assert info['help_obj'] == info['complete_obj'] == 'nump'

    assert p.parse_code('%python impor', 0, 10)['magic']['name'] == 'python'
    assert p.parse_code('oct(a,')['help_obj'] == 'oct'
    assert p.parse_code('! ls')['magic']['name'] == 'shell'

    info = p.parse_code('%help %lsmagic', 0, 10)
    assert info['help_obj'] == '%lsmagic'
    assert info['complete_obj'] == '%lsm'

    info = p.parse_code('%%python\nprint("hello, world!",')
    assert info['help_obj'] == 'print'


def test_scheme_parser():
    identifier_regex = r'[^\d\W]\w*'
    function_call_regex = r'\(([^\d\W][\w\.]*)[^\)\()]*\Z'
    magic_prefixes = dict(magic='%', shell='!', help='?')
    magic_suffixes = dict(help='?')
    p = Parser(identifier_regex, function_call_regex, magic_prefixes,
               magic_suffixes)

    info = p.parse_code('(oct a b ')
    assert info['help_obj'] == 'oct'


def test_path_completions():
    if not os.name == 'nt':
        code = '/usr/bi'
        assert 'bin/' in _complete_path(code)
    code = '~/.bashr'
    assert '.bashrc' in _complete_path(code)
    for f in os.listdir('.'):
        if os.path.isdir(f):
            assert f + os.sep in _complete_path('.')
        else:
            assert f in _complete_path('.')

if __name__ == '__main__':
    test_parser()
    test_scheme_parser()
    test_path_completions()

    # TODO: adjust how these are used by the kernel and affected magics

Also, don't forget that languages like Scheme write their functions this way:

(oct a b c)

So, open paren in front, and no commas. Should also be able to do something useful perhaps for languages like Logo and Basic.

Hmm, that one is tricky. We could instead use a regex for a function call. So, the default would be:
r'[^\d\W]\(\w*' and for Scheme it would be r'\([^\d\W]\w*')

But how to tell if we are still inside the function or not...

Ok, for python: re.search(r'[^\d\W]\(\w*[^\)]*\Z', and for scheme: re.search(r'\([^\d\W]\w*[^\)]*\Z'

Updated.

I think the idea of using kernel-defined regexes is the way to go (and I hope that that is the way that IPython will work for when to call the auto-help when calling a function to give help on args.)

Updated with tests, including a Scheme test.

Looking good!

If you agree with the current implementation I'll go ahead and merge it.