formfiller.py

'''
Helps you submit an HTML form like a browser would.

    form = Form(soup.find('form', name='my_form'))
    requests.request(
        form.get_method('login'),
        form.get_action('login', relative_to='http://example.com/'),
        data=form.fill('login', {
            'username': 'bob',
            'password': 'hunter2',
        })
    )

The values passed to Form.fill() override default values.

Above, 'login' is the submit button to use. If omitted, the first submit button
(<input type="submit"> or <button type="submit">) is used.

Known issues:
    - Cannot uncheck a checkbox whose default state is checked
    - Does not find the default selected <option> for a <select>
'''
import urllib.parse


# see https://www.w3.org/TR/html52/sec-forms.html
class Form:
    def __init__(self, form):
        assert form.name == 'form'  # feed me BeautifulSoup <form> tags

        self.form = form
        self._action = form.get('action', '')
        self._method = form.get('method', 'GET')

        self.fields = {}
        self.buttons = {}

        for field in form.find_all(('input', 'button', 'select', 'textarea')):
            name = field.get('name')
            if not name:
                continue

            self.fields[name] = (field, field.get('value'))
            if field.name in ('input', 'button'):
                if field.get('type') == 'submit':
                    self.buttons[name] = (field.get('formaction'),
                                          field.get('formmethod'))

    def _get_default_button(self):
        if self.buttons:
            return next(iter(self.buttons.keys()))

    def get_action(self, button=None, relative_to=''):
        # Get default submit button if none specified
        if button is None:
            button = self._get_default_button()

        # Use the submit button's formaction if available
        action = None if button is None else self.buttons[button][0]
        action = action or self._action
        return urllib.parse.urljoin(relative_to, action)

    def get_method(self, button=None):
        if button is None:
            button = self._get_default_button()
        
        method = None if button is None else self.buttons[button][1]
        return method or self._method

    def _fill_impl(self, button, values):
        filled = {}

        for form_name, (field, default_value) in self.fields.items():
            # Skip disabled fields
            if field.has_attr('disabled'):
                continue

            # Skip buttons that are not the submit button
            is_button = (field.name == 'button') or (field.name == 'input' and \
                field.get('type') in ('submit', 'image', 'reset', 'button'))
            if is_button and form_name != button:
                continue

            # Skip radio buttons and checkboxes that are not checked
            is_radio_or_checkbox = (field.name == 'input' and \
                field.get('type') in ('radio', 'checkbox'))
            if is_radio_or_checkbox and not field.has_attr('checked'):
                continue

            # Add the default value
            if default_value is None:
                if is_button:
                    default_value = 'Submit'
                elif is_radio_or_checkbox:
                    default_value = 'on'
                else:
                    default_value = ''
            filled[form_name] = default_value

        # Override any form values with our input
        filled.update(values)
        return filled

    def fill(self, *args):
        if len(args) == 0:
            return self._fill_impl(self._get_default_button(), {})
        elif len(args) == 1:
            return self._fill_impl(self._get_default_button(), args[0])
        elif len(args) == 2:
            return self._fill_impl(args[0], args[1])
        raise ValueError('Expected fill(values) or fill(button, values)')