Python - Multiple choice markup parsing

Question

Consider this text:

Would you like to have responses to your questions sent to you via email ?

I'm going to propose multiple choices for several words by marking up them like that:

Would you like [to get]|[having]|g[to have] responses to your questions sent [up to]|g[to]|[on] you via email ?

The choices are bracketted and separated by pipes
The good choice is preceded by a g

I would like to parse this sentence to get the text formatted like that:

Would you like __ responses to your questions sent __ you via email ?

With a list like:

[
  [
    {"to get":0},
    {"having":0},
    {"to have":1},
  ],
  [
    {"up to":0},
    {"to":1},
    {"on":0},
  ],
]

Is my markup design ok ?
How to regex the sentence to get the needed result and generate the list ?

edit: User oriented markup language needed

Answer 1

I would add some grouping parentheses {}, and output not list of list of dicts, but list of dicts.

Code:

import re

s = 'Would you like {[to get]|[having]|g[to have]} responses to your questions sent {[up to]|g[to]|[on]} you via email ?'

def variants_to_dict(variants):
    dct = {}
    for is_good, s in variants:
        dct[s] = 1 if is_good == 'g' else 0
    return dct

def question_to_choices(s):
    choices_re = re.compile(r'{[^}]+}')
    variants_re = re.compile(r'''\|?(g?)
                                 \[
                                    ([^\]]+)
                                 \]
                                ''', re.VERBOSE)
    choices_list = []
    for choices in choices_re.findall(s):
        choices_list.append(variants_to_dict(variants_re.findall(choices)))

    return choices_re.sub('___', s), choices_list

question, choices = question_to_choices(s)
print question
print choices

Output:

Would you like ___ responses to your questions sent ___ you via email ?
[{'to have': 1, 'to get': 0, 'having': 0}, {'to': 1, 'up to': 0, 'on': 0}]

Answer 2

Rough parsing implementation using regular expressions:

import re
s = "Would you like [to get]|[having]|g[to have] responses to your questions sent [up to]|g[to]|[on] you via email ?"   # pattern string

choice_groups = re.compile(r"((?:g?\[[^\]]+\]\|?)+)")  # regex to get choice groups
choices = re.compile(r"(g?)\[([^\]]+)\]")  # regex to extract choices within each group

# now, use the regexes to parse the string:
groups = choice_groups.findall(s)
# returns: ['[to get]|[having]|g[to have]', '[up to]|g[to]|[on]']

# parse each group to extract possible choices, along with if they are good
group_choices = [choices.findall(group) for group in groups]
# will contain [[('', 'to get'), ('', 'having'), ('g', 'to have')], [('', 'up to'), ('g', 'to'), ('', 'on')]]

# finally, substitute each choice group to form a template
template = choice_groups.sub('___', s)
# template is "Would you like ___ responses to your questions sent ___ you via email ?"

Parsing this to suit your format should be pretty easy now. Good luck :)

Answer 3

I will suggest my solution too:

Would you like {to get|having|+to have} responses to your questions sent {up to|+to|on} you via email ?

def extract_choices(text):
    choices = []

    def callback(match):
        variants = match.group().strip('{}')
        choices.append(dict(
            (v.lstrip('+'), v.startswith('+'))
            for v in variants.split('|')
        ))
        return '___'

    text = re.sub('{.*?}', callback, text)

    return text, choices

Lets try it:

>>> t = 'Would you like {to get|having|+to have} responses to your questions    sent {up to|+to|on} you via email?'
>>> pprint.pprint(extract_choices(t))
... ('Would you like ___ responses to your questions sent ___ you via email?',
... [{'having': False, 'to get': False, 'to have': True},
...  {'on': False, 'to': True, 'up to': False}])

Answer 4

I also think that for this task xml is much more appropriate because there are already a lot of tools available that will make parsing much easier and less error-prone.

Anyway, if you decide to use your design, I'd do something like this:

import re

question_str = ("Would you like [to get]|[having]|g[to have] "
                "responses to your questions sent "
                "[up to]|g[to]|[on] you via email ?")

def option_to_dict(option_str):
     if option_str.startswith('g'):
          name = option_str.lstrip('g')
          value = 1
     else:
          name = option_str
          value = 0
     name = name.strip('[]')
     return {name: value}

regex = re.compile('g?\[[^]]+\](\|g?\[[^]]+\])*')

options = [[option_to_dict(option_str)
            for option_str in match.group(0).split('|')]
           for match in regex.finditer(question_str)]
print options

question = regex.sub('___', question_str)
print question

Example output:

[[{'to get': 0}, {'having': 0}, {'to have': 1}], [{'up to': 0}, {'to': 1}, {'on': 0}]]
Would you like ___ responses to your questions sent ___ you via email ?

Note: Regarding the design, I think it would be better to have a mark to set start/end of the whole set of options (not just one for single options).