davtelepot/davtelepot/utilities.py

"""Useful functions used by Davte when programming in python."""

# Standard library modules
import asyncio
import collections
import csv
import datetime
from difflib import SequenceMatcher
import json
import logging
import os
import random
import string
import time

# Third party modules
import aiohttp
from aiohttp import web
from bs4 import BeautifulSoup


def sumif(iterable, condition):
    """Sum all `iterable` items matching `condition`."""
    return sum(
        filter(
            condition,
            iterable
        )
    )


def markdown_check(text, symbols):
    """Check that  all `symbols` occur an even number of times in `text`."""
    for s in symbols:
        if (len(text.replace(s, "")) - len(text)) % 2 != 0:
            return False
    return True


def shorten_text(text, limit, symbol="[...]"):
    """Return a given text truncated at limit if longer than limit.

    On truncation, add symbol.
    """
    assert type(text) is str and type(symbol) is str and type(limit) is int
    if len(text) <= limit:
        return text
    return text[:limit-len(symbol)] + symbol


def extract(text, starter=None, ender=None):
    """Return string in text between starter and ender.

    If starter is None, truncate at ender.
    """
    if starter and starter in text:
        text = text.partition(starter)[2]
    if ender:
        return text.partition(ender)[0]
    return text


def make_button(text=None, callback_data='',
                prefix='', delimiter='|', data=[]):
    """Return a Telegram bot API-compliant button.

    callback_data can be either a ready-to-use string or a
        prefix + delimiter-joined data.
    If callback_data exceeds Telegram limits (currently 60 characters),
        it gets truncated at the last delimiter before that limit.
    If absent, text is the same as callback_data.
    """
    if len(data):
        callback_data += delimiter.join(map(str, data))
    callback_data = "{p}{c}".format(
        p=prefix,
        c=callback_data
    )
    if len(callback_data) > 60:
        callback_data = callback_data[:61]
        callback_data = callback_data[:-1-callback_data[::-1].find(delimiter)]
    if text is None:
        text = callback_data
    return dict(
        text=text,
        callback_data=callback_data
    )


def mkbtn(x, y):
    """Backward compatibility.

    Warning: this function will be deprecated sooner or later,
        stop using it and migrate to make_button.
    """
    return make_button(text=x, callback_data=y)


def make_lines_of_buttons(buttons, row_len=1):
    """Split `buttons` list in a list of lists having length = `row_len`."""
    return [
        buttons[i:i + row_len]
        for i in range(
            0,
            len(buttons),
            row_len
        )
    ]


def make_inline_keyboard(buttons, row_len=1):
    """Make a Telegram API compliant inline keyboard."""
    return dict(
        inline_keyboard=make_lines_of_buttons(
            buttons,
            row_len
        )
    )


async def async_get(url, mode='json', **kwargs):
    """Make an async get request.

    `mode`s allowed:
        * html
        * json
        * string

    Additional **kwargs may be passed.
    """
    if 'mode' in kwargs:
        mode = kwargs['mode']
        del kwargs['mode']
    return await async_request(
        url,
        type='get',
        mode=mode,
        **kwargs
    )


async def async_post(url, mode='html', **kwargs):
    """Make an async post request.

    `mode`s allowed:
        * html
        * json
        * string

    Additional **kwargs may be passed (will be converted to `data`).
    """
    return await async_request(
        url,
        type='post',
        mode=mode,
        **kwargs
    )


async def async_request(url, type='get', mode='json', encoding='utf-8',
                        **kwargs):
    """Make an async html request.

    `types` allowed
        * get
        * post

    `mode`s allowed:
        * html
        * json
        * string

    Additional **kwargs may be passed.
    """
    try:
        async with aiohttp.ClientSession() as s:
                async with (
                    s.get(url, timeout=30)
                    if type == 'get'
                    else s.post(url, timeout=30, data=kwargs)
                ) as r:
                    result = await r.read()
                    result = result.decode(encoding)
    except Exception as e:
        logging.error(
            'Error making async request to {}:\n{}'.format(
                url,
                e
            ),
            exc_info=False
        )  # Set exc_info=True to debug
        return e
    if mode == 'json':
        try:
            result = json.loads(
                result
            )
        except Exception as e:
            result = {}
    elif mode == 'html':
        result = BeautifulSoup(result, "html.parser")
    elif mode == 'string':
        result = result
    return result


def json_read(file_, default={}, encoding='utf-8', **kwargs):
    """Return json parsing of `file_`, or `default` if file does not exist.

    `encoding` refers to how the file should be read.
    `kwargs` will be passed to json.load()
    """
    if not os.path.isfile(file_):
        return default
    with open(file_, "r", encoding=encoding) as f:
        return json.load(f, **kwargs)


def json_write(what, file_, encoding='utf-8', **kwargs):
    """Store `what` in json `file_`.

    `encoding` refers to how the file should be written.
    `kwargs` will be passed to json.dump()
    """
    with open(file_, "w", encoding=encoding) as f:
        return json.dump(what, f, indent=4, **kwargs)


def csv_read(file_, default=[], encoding='utf-8',
             delimiter=',', quotechar='"', **kwargs):
    """Return csv parsing of `file_`, or `default` if file does not exist.

    `encoding` refers to how the file should be read.
    `delimiter` is the separator of fields.
    `quotechar` is the string delimiter.
    `kwargs` will be passed to csv.reader()
    """
    if not os.path.isfile(file_):
        return default
    result = []
    keys = []
    with open(file_, newline='', encoding=encoding) as csv_file:
        csv_reader = csv.reader(
            csv_file,
            delimiter=delimiter,
            quotechar=quotechar,
            **kwargs
        )
        for row in csv_reader:
            if not keys:
                keys = row
                continue
            item = collections.OrderedDict()
            for key, val in zip(keys, row):
                item[key] = val
            result.append(item)
    return result


def csv_write(info=[], file_='output.csv', encoding='utf-8',
              delimiter=',', quotechar='"', **kwargs):
    """Store `info` in CSV `file_`.

    `encoding` refers to how the file should be read.
    `delimiter` is the separator of fields.
    `quotechar` is the string delimiter.
    `encoding` refers to how the file should be written.
    `kwargs` will be passed to csv.writer()
    """
    assert (
        type(info) is list
        and len(info) > 0
    ), "info must be a non-empty list"
    assert all(
        isinstance(row, dict)
        for row in info
    ), "Rows must be dictionaries!"
    with open(file_, 'w', newline='', encoding=encoding) as csv_file:
        csv_writer = csv.writer(
            csv_file,
            delimiter=delimiter,
            quotechar=quotechar,
            **kwargs
        )
        csv_writer.writerow(info[0].keys())
        for row in info:
            csv_writer.writerow(row.values())
    return


class MyOD(collections.OrderedDict):
    """Subclass of OrderedDict.

    It features `get_by_val` and `get_by_key_val` methods.
    """

    def __init__(self, *args, **kwargs):
        """Return a MyOD instance."""
        super().__init__(*args, **kwargs)
        self._anti_list_casesensitive = None
        self._anti_list_caseinsensitive = None

    @property
    def anti_list_casesensitive(self):
        """Case-sensitive reverse dictionary.

        Keys and values are swapped.
        """
        if not self._anti_list_casesensitive:
            self._anti_list_casesensitive = {
                val: key
                for key, val in self.items()
            }
        return self._anti_list_casesensitive

    @property
    def anti_list_caseinsensitive(self):
        """Case-sensitive reverse dictionary.

        Keys and values are swapped and lowered.
        """
        if not self._anti_list_caseinsensitive:
            self._anti_list_caseinsensitive = {
                (val.lower() if type(val) is str else val): key
                for key, val in self.items()
            }
        return self._anti_list_caseinsensitive

    def get_by_val(self, val, case_sensitive=True):
        """Get key pointing to given val.

        Can be case-sensitive or insensitive.
        MyOD[key] = val <-- MyOD.get(key) = val <--> MyOD.get_by_val(val) = key
        """
        return (
            self.anti_list_casesensitive
            if case_sensitive
            else self.anti_list_caseinsensitive
        )[val]

    def get_by_key_val(self, key, val,
                       case_sensitive=True, return_value=False):
        """Get key (or val) of a dict-like object having key == val.

        Perform case-sensitive or insensitive search.
        """
        for x, y in self.items():
            if (
                (
                    y[key] == val
                    and case_sensitive
                ) or (
                    y[key].lower() == val.lower()
                    and not case_sensitive
                )
            ):
                return (
                    y if return_value
                    else x
                )
        return None


def line_drawing_unordered_list(l):
    """Draw an old-fashioned unordered list.

    Unorderd list example
    ├ An element
    ├ Another element
    └Last element
    """
    result = ""
    if l:
        for x in l[:-1]:
            result += "├ {}\n".format(x)
        result += "└ {}".format(l[-1])
    return result


def str_to_datetime(d):
    """Convert string to datetime.

    Dataset library often casts datetimes to str, this is a workaround.
    """
    if isinstance(d, datetime.datetime):
        return d
    return datetime.datetime.strptime(
        d,
        '%Y-%m-%d %H:%M:%S.%f'
    )


def datetime_to_str(d):
    """Cast datetime to string."""
    if isinstance(d, str):
        d = str_to_datetime(d)
    if not isinstance(d, datetime.datetime):
        raise TypeError(
            'Input of datetime_to_str function must be a datetime.datetime '
            'object. Output is a str'
        )
    return '{:%Y-%m-%d %H:%M:%S.%f}'.format(d)


class MyCounter():
    """Counter object, with a `lvl` method incrementing `n` property."""

    def __init__(self):
        """Initialize and get MyCounter instance."""
        self._n = 0
        return

    def lvl(self):
        """Increments and return self.n."""
        self._n += 1
        return self.n

    def reset(self):
        """Set self.n = 0."""
        self._n = 0
        return self.n

    def n(self):
        """Counter's value."""
        return self._n


def wrapper(func, *args, **kwargs):
    """Wrap a function so that it can be later called with one argument."""
    def wrapped(update):
        return func(update, *args, **kwargs)
    return wrapped


async def async_wrapper(func, *args, **kwargs):
    """Wrap a coroutine so that it can be later awaited with one argument."""
    async def wrapped(update):
        return await func(update, *args, **kwargs)
    return wrapped


def forwarded(by=None):
    """Check that update is forwarded, optionally `by` someone in particular.

    Decorator: such decorated functions have effect only if update
        is forwarded from someone (you can specify `by` whom).
    """
    def is_forwarded_by(update, by):
        if 'forward_from' not in update:
            return False
        if by and update['forward_from']['id'] != by:
            return False
        return True

    def decorator(view_func):
        if asyncio.iscoroutinefunction(view_func):
            async def decorated(update):
                if is_forwarded_by(update, by):
                    return await view_func(update)
        else:
            def decorated(update):
                if is_forwarded_by(update, by):
                    return view_func(update)
        return decorated
    return decorator


def chat_selective(chat_id=None):
    """Check that update comes from a chat, optionally having `chat_id`.

    Such decorated functions have effect only if update comes from
        a specific (if `chat_id` is given) or generic chat.
    """
    def check_function(update, chat_id):
        if 'chat' not in update:
            return False
        if chat_id:
            if update['chat']['id'] != chat_id:
                return False
        return True

    def decorator(view_func):
        if asyncio.iscoroutinefunction(view_func):
            async def decorated(update):
                if check_function(update, chat_id):
                    return await view_func(update)
        else:
            def decorated(update):
                if check_function(update, chat_id):
                    return view_func(update)
        return decorated
    return decorator


async def sleep_until(when):
    """Sleep until now > `when`.

    `when` could be a datetime.datetime or a datetime.timedelta instance.
    """
    if not (
        isinstance(when, datetime.datetime)
        or isinstance(when, datetime.timedelta)
    ):
        raise TypeError(
            "sleep_until takes a datetime.datetime or datetime.timedelta "
            "object as argument!"
        )
    if isinstance(when, datetime.datetime):
        delta = when - datetime.datetime.now()
    elif isinstance(when, datetime.timedelta):
        delta = when
    if delta.days >= 0:
        await asyncio.sleep(
            delta.seconds
        )
    return


async def wait_and_do(when, what, *args, **kwargs):
    """Sleep until `when`, then call `what` passing `args` and `kwargs`."""
    await sleep_until(when)
    return await what(*args, **kwargs)


def get_csv_string(list_, delimiter=',', quotechar='"'):
    """Return a `delimiter`-delimited string of `list_` items.

    Wrap strings in `quotechar`s.
    """
    return delimiter.join(
        str(item) if type(item) is not str
        else '{q}{i}{q}'.format(
            i=item,
            q=quotechar
        )
        for item in list_
    )


def case_accent_insensitive_sql(field):
    """Get a SQL string to perform a case- and accent-insensitive query.

    Given a `field`, return a part of SQL string necessary to perform
        a case- and accent-insensitive query.
    """
    replacements = [
        (' ', ''),
        ('à', 'a'),
        ('è', 'e'),
        ('é', 'e'),
        ('ì', 'i'),
        ('ò', 'o'),
        ('ù', 'u'),
    ]
    return "{r}LOWER({f}){w}".format(
        r="replace(".upper()*len(replacements),
        f=field,
        w=''.join(
            ", '{w[0]}', '{w[1]}')".format(w=w)
            for w in replacements
        )
    )


# Italian definite articles.
ARTICOLI = MyOD()
ARTICOLI[1] = {
    'ind': 'un',
    'dets': 'il',
    'detp': 'i',
    'dess': 'l',
    'desp': 'i'
}
ARTICOLI[2] = {
    'ind': 'una',
    'dets': 'la',
    'detp': 'le',
    'dess': 'lla',
    'desp': 'lle'
}
ARTICOLI[3] = {
    'ind': 'uno',
    'dets': 'lo',
    'detp': 'gli',
    'dess': 'llo',
    'desp': 'gli'
}
ARTICOLI[4] = {
    'ind': 'un',
    'dets': 'l\'',
    'detp': 'gli',
    'dess': 'll\'',
    'desp': 'gli'
}


class Gettable():
    """Gettable objects can be retrieved from memory without being duplicated.

    Key is the primary key.
    Use classmethod get to instantiate (or retrieve) Gettable objects.
    Assign SubClass.instances = {}, otherwise Gettable.instances will
        contain SubClass objects.
    """

    instances = {}

    @classmethod
    def get(cls, key, *args, **kwargs):
        """Instantiate and/or retrieve Gettable object.

        SubClass.instances is searched if exists.
        Gettable.instances is searched otherwise.
        """
        if key not in cls.instances:
            cls.instances[key] = cls(key, *args, **kwargs)
        return cls.instances[key]


class Confirmable():
    """Confirmable objects are provided with a confirm instance method.

    It evaluates True if it was called within self._confirm_timedelta,
        False otherwise.
    When it returns True, timer is reset.
    """

    CONFIRM_TIMEDELTA = datetime.timedelta(seconds=10)

    def __init__(self, confirm_timedelta=None):
        """Instantiate Confirmable instance.

        If `confirm_timedelta` is not passed,
            `self.__class__.CONFIRM_TIMEDELTA` is used as default.
        """
        if confirm_timedelta is None:
            confirm_timedelta = self.__class__.CONFIRM_TIMEDELTA
        self.set_confirm_timedelta(confirm_timedelta)
        self._confirm_datetimes = {}

    @property
    def confirm_timedelta(self):
        """Maximum timedelta between two calls of `confirm`."""
        return self._confirm_timedelta

    def confirm_datetime(self, who='unique'):
        """Get datetime of `who`'s last confirm.

        If `who` never called `confirm`, fake an expired call.
        """
        if who not in self._confirm_datetimes:
            self._confirm_datetimes[who] = (
                datetime.datetime.now()
                - 2*self.confirm_timedelta
            )
        return self._confirm_datetimes[who]

    def set_confirm_timedelta(self, confirm_timedelta):
        """Change self._confirm_timedelta."""
        if type(confirm_timedelta) is int:
            confirm_timedelta = datetime.timedelta(
                seconds=confirm_timedelta
            )
        assert isinstance(
            confirm_timedelta, datetime.timedelta
        ), "confirm_timedelta must be a datetime.timedelta instance!"
        self._confirm_timedelta = confirm_timedelta

    def confirm(self, who='unique'):
        """Return True if `confirm` was called by `who` recetly enough."""
        now = datetime.datetime.now()
        if now >= self.confirm_datetime(who) + self.confirm_timedelta:
            self._confirm_datetimes[who] = now
            return False
        self._confirm_datetimes[who] = now - 2*self.confirm_timedelta
        return True


class HasBot():
    """Objects having a telepot.Bot subclass object as `.bot` attribute.

    HasBot objects have a .bot and .db properties for faster access.
    """

    bot = None

    @property
    def bot(self):
        """Class bot."""
        return self.__class__.bot

    @property
    def db(self):
        """Class bot db."""
        return self.bot.db

    @classmethod
    def set_bot(cls, bot):
        """Change class bot."""
        cls.bot = bot


class CachedPage(Gettable):
    """Cache a webpage and return it during CACHE_TIME, otherwise refresh.

    Usage:
    cached_page = CachedPage.get(
        'https://www.google.com',
        datetime.timedelta(seconds=30),
        **kwargs
    )
    page = await cached_page.get_page()
    """

    CACHE_TIME = datetime.timedelta(minutes=5)
    instances = {}

    def __init__(self, url, cache_time=None, **async_get_kwargs):
        """Instantiate CachedPage object.

        `url`: the URL to be cached
        `cache_time`: timedelta from last_update during which
            page will be cached
        `**kwargs` will be passed to async_get function
        """
        self._url = url
        if type(cache_time) is int:
            cache_time = datetime.timedelta(seconds=cache_time)
        if cache_time is None:
            cache_time = self.__class__.CACHE_TIME
        assert type(cache_time) is datetime.timedelta, (
            "Cache time must be a datetime.timedelta object!"
        )
        self._cache_time = cache_time
        self._page = None
        self._last_update = datetime.datetime.now() - self.cache_time
        self._async_get_kwargs = async_get_kwargs

    @property
    def url(self):
        """Get cached page url."""
        return self._url

    @property
    def cache_time(self):
        """Get cache time."""
        return self._cache_time

    @property
    def page(self):
        """Get webpage."""
        return self._page

    @property
    def last_update(self):
        """Get datetime of last update."""
        return self._last_update

    @property
    def async_get_kwargs(self):
        """Get async get request keyword arguments."""
        return self._async_get_kwargs

    @property
    def is_old(self):
        """Evaluate True if `chache_time` has passed since last update."""
        return datetime.datetime.now() > self.last_update + self.cache_time

    async def refresh(self):
        """Update cached webpage."""
        try:
            self._page = await async_get(self.url, **self.async_get_kwargs)
            self._last_update = datetime.datetime.now()
            return 0
        except Exception as e:
            self._page = None
            logging.error(
                '{e}'.format(
                    e=e
                ),
                exc_info=False
            )  # Set exc_info=True to debug
            return 1
        return 1

    async def get_page(self):
        """Refresh if necessary and return webpage."""
        if self.is_old:
            await self.refresh()
        return self.page


class Confirmator(Gettable, Confirmable):
    """Gettable Confirmable object."""

    instances = {}

    def __init__(self, key, *args, confirm_timedelta=None):
        """Call Confirmable.__init__ passing `confirm_timedelta`."""
        Confirmable.__init__(self, confirm_timedelta)


def get_cleaned_text(update, bot=None, replace=[], strip='/ @'):
    """Clean `update`['text'] and return it.

    Strip `bot`.name and items to be `replace`d from the beginning of text.
    Strip `strip` characters from both ends.
    """
    if bot is not None:
        replace.append(
            '@{.name}'.format(
                bot
            )
        )
    text = update['text'].strip(strip)
    for s in replace:
        while s and text.lower().startswith(s.lower()):
            text = text[len(s):]
    return text.strip(strip)


def get_user(record):
    """Get an HTML Telegram tag for user `record`."""
    if not record:
        return
    from_ = {key: val for key, val in record.items()}
    first_name, last_name, username, id_ = None, None, None, None
    result = '{name}'
    if 'telegram_id' in from_:
        from_['id'] = from_['telegram_id']
    if (
        'id' in from_
        and from_['id'] is not None
    ):
        result = '<a href="tg://user?id={}">{{name}}</a>'.format(
            from_['id']
        )
    if 'username' in from_ and from_['username']:
        result = result.format(
            name=from_['username']
        )
    elif (
        'first_name' in from_
        and from_['first_name']
        and 'last_name' in from_
        and from_['last_name']
    ):
        result = result.format(
            name='{} {}'.format(
                from_['first_name'],
                from_['last_name']
            )
        )
    elif 'first_name' in from_ and from_['first_name']:
        result = result.format(
            name=from_['first_name']
        )
    elif 'last_name' in from_ and from_['last_name']:
        result = result.format(
            name=from_['last_name']
        )
    else:
        result = result.format(
            name="Utente anonimo"
        )
    return result


def datetime_from_utc_to_local(utc_datetime):
    """Convert `utc_datetime` to local datetime."""
    now_timestamp = time.time()
    offset = (
        datetime.datetime.fromtimestamp(now_timestamp)
        - datetime.datetime.utcfromtimestamp(now_timestamp)
    )
    return utc_datetime + offset


# TIME_SYMBOLS from more specific to less specific (avoid false positives!)
TIME_SYMBOLS = MyOD()
TIME_SYMBOLS["'"] = 'minutes'
TIME_SYMBOLS["settimana"] = 'weeks'
TIME_SYMBOLS["settimane"] = 'weeks'
TIME_SYMBOLS["weeks"] = 'weeks'
TIME_SYMBOLS["week"] = 'weeks'
TIME_SYMBOLS["giorno"] = 'days'
TIME_SYMBOLS["giorni"] = 'days'
TIME_SYMBOLS["secondi"] = 'seconds'
TIME_SYMBOLS["seconds"] = 'seconds'
TIME_SYMBOLS["secondo"] = 'seconds'
TIME_SYMBOLS["minuti"] = 'minutes'
TIME_SYMBOLS["minuto"] = 'minutes'
TIME_SYMBOLS["minute"] = 'minutes'
TIME_SYMBOLS["minutes"] = 'minutes'
TIME_SYMBOLS["day"] = 'days'
TIME_SYMBOLS["days"] = 'days'
TIME_SYMBOLS["ore"] = 'hours'
TIME_SYMBOLS["ora"] = 'hours'
TIME_SYMBOLS["sec"] = 'seconds'
TIME_SYMBOLS["min"] = 'minutes'
TIME_SYMBOLS["m"] = 'minutes'
TIME_SYMBOLS["h"] = 'hours'
TIME_SYMBOLS["d"] = 'days'
TIME_SYMBOLS["s"] = 'seconds'


def _interval_parser(text, result):
    text = text.lower()
    succeeded = False
    if result is None:
        result = []
    if len(result) == 0 or result[-1]['ok']:
        text_part = ''
        _text = text  # I need to iterate through _text modifying text
        for char in _text:
            if not char.isnumeric():
                break
            else:
                text_part += char
                text = text[1:]
        if text_part.isnumeric():
            result.append(
                dict(
                    unit=None,
                    value=int(text_part),
                    ok=False
                )
            )
            succeeded = True, True
            if text:
                dummy, result = _interval_parser(text, result)
    elif len(result) > 0 and not result[-1]['ok']:
        text_part = ''
        _text = text  # I need to iterate through _text modifying text
        for char in _text:
            if char.isnumeric():
                break
            else:
                text_part += char
                text = text[1:]
        for time_symbol, unit in TIME_SYMBOLS.items():
            if time_symbol in text_part:
                result[-1]['unit'] = unit
                result[-1]['ok'] = True
                succeeded = True, True
                break
        else:
            result.pop()
        if text:
            dummy, result = _interval_parser(text, result)
    return succeeded, result


def _date_parser(text, result):
    succeeded = False
    if 3 <= len(text) <= 10 and text.count('/') >= 1:
        if 3 <= len(text) <= 5 and text.count('/') == 1:
            text += '/{:%y}'.format(datetime.datetime.now())
        if 6 <= len(text) <= 10 and text.count('/') == 2:
            day, month, year = [
                int(n) for n in [
                    ''.join(char)
                    for char in text.split('/')
                    if char.isnumeric()
                ]
            ]
            if year < 100:
                year += 2000
            if result is None:
                result = []
            result += [
                dict(
                    unit='day',
                    value=day,
                    ok=True
                ),
                dict(
                    unit='month',
                    value=month,
                    ok=True
                ),
                dict(
                    unit='year',
                    value=year,
                    ok=True
                )
            ]
            succeeded = True, True
    return succeeded, result


def _time_parser(text, result):
    succeeded = False
    if (1 <= len(text) <= 8) and any(char.isnumeric() for char in text):
        text = ''.join(
            ':' if char == '.' else char
            for char in text
            if char.isnumeric() or char in (':', '.')
        )
        if len(text) <= 2:
            text = '{:02d}:00:00'.format(int(text))
        elif len(text) == 4 and ':' not in text:
            text = '{:02d}:{:02d}:00'.format(
                *[int(x) for x in (text[:2], text[2:])]
            )
        elif text.count(':') == 1:
            text = '{:02d}:{:02d}:00'.format(
                *[int(x) for x in text.split(':')]
            )
        if text.count(':') == 2:
            hour, minute, second = (int(x) for x in text.split(':'))
            if (
                0 <= hour <= 24
                and 0 <= minute <= 60
                and 0 <= second <= 60
            ):
                if result is None:
                    result = []
                result += [
                    dict(
                        unit='hour',
                        value=hour,
                        ok=True
                    ),
                    dict(
                        unit='minute',
                        value=minute,
                        ok=True
                    ),
                    dict(
                        unit='second',
                        value=second,
                        ok=True
                    )
                ]
                succeeded = True
    return succeeded, result


WEEKDAY_NAMES_ITA = ["Lunedì", "Martedì", "Mercoledì", "Giovedì",
                     "Venerdì", "Sabato", "Domenica"]
WEEKDAY_NAMES_ENG = ["Monday", "Tuesday", "Wednesday", "Thursday",
                     "Friday", "Saturday", "Sunday"]


def _period_parser(text, result):
    succeeded = False
    if text in ('every', 'ogni',):
        succeeded = True
    if text.title() in WEEKDAY_NAMES_ITA + WEEKDAY_NAMES_ENG:
        day_code = (WEEKDAY_NAMES_ITA + WEEKDAY_NAMES_ENG).index(text.title())
        if day_code > 6:
            day_code -= 7
        today = datetime.date.today()
        days = 1
        while (today + datetime.timedelta(days=days)).weekday() != day_code:
            days += 1
        if result is None:
            result = []
        result.append(
            dict(
                unit='days',
                value=days,
                ok=True,
                weekly=True
            )
        )
        succeeded = True
    else:
        succeeded, result = _interval_parser(text, result)
    return succeeded, result


TIME_WORDS = {
    'tra': dict(
        parser=_interval_parser,
        recurring=False,
        type_='delta'
    ),
    'in': dict(
        parser=_interval_parser,
        recurring=False,
        type_='delta'
    ),
    'at': dict(
        parser=_time_parser,
        recurring=False,
        type_='set'
    ),
    'on': dict(
        parser=_date_parser,
        recurring=False,
        type_='set'
    ),
    'alle': dict(
        parser=_time_parser,
        recurring=False,
        type_='set'
    ),
    'il': dict(
        parser=_date_parser,
        recurring=False,
        type_='set'
    ),
    'every': dict(
        parser=_period_parser,
        recurring=True,
        type_='delta'
    ),
    'ogni': dict(
        parser=_period_parser,
        recurring=True,
        type_='delta'
    ),
}


def parse_datetime_interval_string(text):
    """Parse `text` and return text, datetime and timedelta."""
    parsers = []
    result_text, result_datetime, result_timedelta = [], None, None
    is_quoted_text = False
    for word in text.split(' '):
        if word.count('"') % 2:
            is_quoted_text = not is_quoted_text
        if is_quoted_text or '"' in word:
            result_text.append(
                word.replace('"', '') if 'href=' not in word else word
            )
            continue
        result_text.append(word)
        word = word.lower()
        succeeded = False
        if len(parsers) > 0:
            succeeded, result = parsers[-1]['parser'](
                word,
                parsers[-1]['result']
            )
            if succeeded:
                parsers[-1]['result'] = result
        if not succeeded and word in TIME_WORDS:
            parsers.append(
                dict(
                    result=None,
                    parser=TIME_WORDS[word]['parser'],
                    recurring=TIME_WORDS[word]['recurring'],
                    type_=TIME_WORDS[word]['type_']
                )
            )
        if succeeded:
            result_text.pop()
            if len(result_text) > 0 and result_text[-1].lower() in TIME_WORDS:
                result_text.pop()
    result_text = escape_html_chars(
        ' '.join(result_text)
    )
    parsers = list(
        filter(
            lambda x: 'result' in x and x['result'],
            parsers
        )
    )
    recurring_event = False
    weekly = False
    _timedelta = datetime.timedelta()
    _datetime = None
    _now = datetime.datetime.now()
    for parser in parsers:
        if parser['recurring']:
            recurring_event = True
        type_ = parser['type_']
        for result in parser['result']:
            if not result['ok']:
                continue
            if recurring_event and 'weekly' in result and result['weekly']:
                weekly = True
            if type_ == 'set':
                if _datetime is None:
                    _datetime = _now
                _datetime = _datetime.replace(
                    **{
                        result['unit']: result['value']
                    }
                )
            elif type_ == 'delta':
                _timedelta += datetime.timedelta(
                    **{
                        result['unit']: result['value']
                    }
                )
    if _datetime:
        result_datetime = _datetime
    if _timedelta:
        if result_datetime is None:
            result_datetime = _now
        if recurring_event:
            result_timedelta = _timedelta
            if weekly:
                result_timedelta = datetime.timedelta(days=7)
        else:
            result_datetime += _timedelta
    while result_datetime and result_datetime < datetime.datetime.now():
        result_datetime += (
            result_timedelta
            if result_timedelta
            else datetime.timedelta(days=1)
        )
    return result_text, result_datetime, result_timedelta


DAY_GAPS = {
    -1: 'ieri',
    -2: 'avantieri',
    0: 'oggi',
    1: 'domani',
    2: 'dopodomani'
}

MONTH_NAMES_ITA = MyOD()
MONTH_NAMES_ITA[1] = "gennaio"
MONTH_NAMES_ITA[2] = "febbraio"
MONTH_NAMES_ITA[3] = "marzo"
MONTH_NAMES_ITA[4] = "aprile"
MONTH_NAMES_ITA[5] = "maggio"
MONTH_NAMES_ITA[6] = "giugno"
MONTH_NAMES_ITA[7] = "luglio"
MONTH_NAMES_ITA[8] = "agosto"
MONTH_NAMES_ITA[9] = "settembre"
MONTH_NAMES_ITA[10] = "ottobre"
MONTH_NAMES_ITA[11] = "novembre"
MONTH_NAMES_ITA[12] = "dicembre"


def beautytd(td):
    """Format properly timedeltas."""
    result = ''
    if type(td) is int:
        td = datetime.timedelta(seconds=td)
    assert isinstance(
        td,
        datetime.timedelta
    ), "td must be a datetime.timedelta object!"
    mtd = datetime.timedelta
    if td < mtd(minutes=1):
        result = "{:.0f} secondi".format(
            td.total_seconds()
        )
    elif td < mtd(minutes=60):
        result = "{:.0f} min{}".format(
            td.total_seconds()//60,
            (
                " {:.0f} s".format(
                    td.total_seconds() % 60
                )
            ) if td.total_seconds() % 60 else ''
        )
    elif td < mtd(days=1):
        result = "{:.0f} h{}".format(
            td.total_seconds()//3600,
            (
                " {:.0f} min".format(
                    (td.total_seconds() % 3600) // 60
                )
            ) if td.total_seconds() % 3600 else ''
        )
    elif td < mtd(days=30):
        result = "{} giorni{}".format(
            td.days,
            (
                " {:.0f} h".format(
                    td.total_seconds() % (3600*24) // 3600
                )
            ) if td.total_seconds() % (3600*24) else ''
        )
    return result


def beautydt(dt):
    """Format a datetime in a smart way."""
    if type(dt) is str:
        dt = str_to_datetime(dt)
    assert isinstance(
        dt,
        datetime.datetime
    ), "dt must be a datetime.datetime object!"
    now = datetime.datetime.now()
    gap = dt - now
    gap_days = (dt.date() - now.date()).days
    result = "{dt:alle %H:%M}".format(
        dt=dt
    )
    if abs(gap) < datetime.timedelta(minutes=30):
        result += "{dt::%S}".format(dt=dt)
    if -2 <= gap_days <= 2:
        result += " di {dg}".format(
            dg=DAY_GAPS[gap_days]
        )
    elif gap.days not in (-1, 0):
        result += " del {d}{m}".format(
            d=dt.day,
            m=(
                "" if now.year == dt.year and now.month == dt.month
                else " {m}{y}".format(
                    m=MONTH_NAMES_ITA[dt.month].title(),
                    y="" if now.year == dt.year
                    else " {}".format(dt.year)
                )
            )
        )
    return result


HTML_SYMBOLS = MyOD()
HTML_SYMBOLS["&"] = "&amp;"
HTML_SYMBOLS["<"] = "&lt;"
HTML_SYMBOLS[">"] = "&gt;"
HTML_SYMBOLS["\""] = "&quot;"
HTML_SYMBOLS["&lt;b&gt;"] = "<b>"
HTML_SYMBOLS["&lt;/b&gt;"] = "</b>"
HTML_SYMBOLS["&lt;i&gt;"] = "<i>"
HTML_SYMBOLS["&lt;/i&gt;"] = "</i>"
HTML_SYMBOLS["&lt;code&gt;"] = "<code>"
HTML_SYMBOLS["&lt;/code&gt;"] = "</code>"
HTML_SYMBOLS["&lt;pre&gt;"] = "<pre>"
HTML_SYMBOLS["&lt;/pre&gt;"] = "</pre>"
HTML_SYMBOLS["&lt;a href=&quot;"] = "<a href=\""
HTML_SYMBOLS["&quot;&gt;"] = "\">"
HTML_SYMBOLS["&lt;/a&gt;"] = "</a>"

HTML_TAGS = [
    None, "<b>", "</b>",
    None, "<i>", "</i>",
    None, "<code>", "</code>",
    None, "<pre>", "</pre>",
    None, "<a href=\"", "\">", "</a>",
    None
]


def remove_html_tags(text):
    """Remove HTML tags from `text`."""
    for tag in HTML_TAGS:
        if tag is None:
            continue
        text = text.replace(tag, '')
    return text


def escape_html_chars(text):
    """Escape HTML chars if not part of a tag."""
    for s, r in HTML_SYMBOLS.items():
        text = text.replace(s, r)
    copy = text
    expected_tag = None
    while copy:
        min_ = min(
            (
                dict(
                    position=copy.find(tag) if tag in copy else len(copy),
                    tag=tag
                )
                for tag in HTML_TAGS
                if tag
            ),
            key=lambda x: x['position'],
            default=0
        )
        if min_['position'] == len(copy):
            break
        if expected_tag and min_['tag'] != expected_tag:
            return text.replace('<', '_').replace('>', '_')
        expected_tag = HTML_TAGS[HTML_TAGS.index(min_['tag'])+1]
        copy = extract(copy, min_['tag'])
    return text


def accents_to_jolly(text, lower=True):
    """Replace letters with Italian accents with SQL jolly character."""
    to_be_replaced = ('à', 'è', 'é', 'ì', 'ò', 'ù')
    if lower:
        text = text.lower()
    else:
        to_be_replaced += tuple(s.upper() for s in to_be_replaced)
    for s in to_be_replaced:
        text = text.replace(s, '_')
    return text.replace("'", "''")


def get_secure_key(allowed_chars=None, length=6):
    """Get a randomly-generate secure key.

    You can specify a set of `allowed_chars` and a `length`.
    """
    if allowed_chars is None:
        allowed_chars = string.ascii_uppercase + string.digits
    return ''.join(
        random.SystemRandom().choice(
            allowed_chars
        )
        for _ in range(length)
    )


def round_to_minute(datetime_):
    """Round `datetime_` to closest minute."""
    return (
        datetime_ + datetime.timedelta(seconds=30)
    ).replace(second=0, microsecond=0)


def get_line_by_content(text, key):
    """Get line of `text` containing `key`."""
    for line in text.split('\n'):
        if key in line:
            return line
    return


def str_to_int(string):
    """Cast str to int, ignoring non-numeric characters."""
    string = ''.join(
        char
        for char in string
        if char.isnumeric()
    )
    if len(string) == 0:
        string = '0'
    return int(string)


def starting_with_or_similar_to(a, b):
    """Return similarity between two strings.

    Least similar equals 0, most similar evaluates 1.
    If similarity is less than 0.75, return 1 if one string starts with
        the other and return 0.5 if one string is contained in the other.
    """
    a = a.lower()
    b = b.lower()
    similarity = SequenceMatcher(None, a, b).ratio()
    if similarity < 0.75:
        if b.startswith(a) or a.startswith(b):
            return 1
        if b in a or a in b:
            return 0.5
    return similarity


def pick_most_similar_from_list(list_, item):
    """Return element from `list_` which is most similar to `item`.

    Similarity is evaluated using `starting_with_or_similar_to`.
    """
    return max(
        list_,
        key=lambda element: starting_with_or_similar_to(
            item,
            element
        )
    )


def run_aiohttp_server(app, *args, **kwargs):
    """Run an aiohttp web app, with its positional and keyword arguments.

    Useful to run apps in dedicated threads.
    """
    loop = asyncio.new_event_loop()
    asyncio.set_event_loop(loop)
    web.run_app(app, *args, **kwargs)


def custom_join(_list, joiner, final=None):
    """Join elements of `_list` using `joiner` (`final` as last joiner)."""
    _list = list(map(str, _list))
    if final is None:
        final = joiner
    if len(_list) == 0:
        return ''
    if len(_list) == 1:
        return _list[0]
    if len(_list) == 2:
        return final.join(_list)
    return joiner.join(_list[:-1]) + final + _list[-1]