def extract_datetime(text, anchorDate="DEFAULT", lang=None, default_time=None):
"""Extracts date and time information from a sentence.
Parses many of the common ways that humans express dates and times,
including relative dates like "5 days from today", "tomorrow', and
"Tuesday".
Vague terminology are given arbitrary values, like:
* morning = 8 AM
* afternoon = 3 PM
* evening = 7 PM
If a time isn't supplied or implied, the function defaults to 12 AM
Args:
text (str): the text to be interpreted
anchorDate (:obj:`datetime`, optional): the date to be used for
relative dating (for example, what does "tomorrow" mean?).
Defaults to the current local date/time.
lang (str): the BCP-47 code for the language to use, None uses default
default_time (datetime.time): time to use if none was found in
the input string.
Returns:
[:obj:`datetime`, :obj:`str`]: 'datetime' is the extracted date
as a datetime object in the user's local timezone.
'leftover_string' is the original phrase with all date and time
related keywords stripped out. See examples for further
clarification
Returns 'None' if no date or time related text is found.
Examples:
>>> extract_datetime(
... "What is the weather like the day after tomorrow?",
... datetime(2017, 06, 30, 00, 00)
... )
[datetime.datetime(2017, 7, 2, 0, 0), 'what is weather like']
>>> extract_datetime(
... "Set up an appointment 2 weeks from Sunday at 5 pm",
... datetime(2016, 02, 19, 00, 00)
... )
[datetime.datetime(2016, 3, 6, 17, 0), 'set up appointment']
>>> extract_datetime(
... "Set up an appointment",
... datetime(2016, 02, 19, 00, 00)
... )
None
"""
if anchorDate is None:
warn(
DeprecationWarning("extract_datetime(anchorDate=None) is "
"deprecated. This parameter can be omitted."))
if anchorDate is None or anchorDate == "DEFAULT":
anchorDate = now_local()
return _extract_datetime(text, anchorDate, lang or get_default_loc(),
default_time)
def test_load_language(self):
lingua_franca.load_language('en')
# Verify that English is loaded and, since it's the only language
# we've loaded, also the default.
self.assertEqual(lingua_franca.get_default_lang(), 'en')
# Verify that English's default full code is 'en-us'
self.assertEqual(lingua_franca.get_full_lang_code('en'), 'en-us')
# Verify that this is also our current full code
self.assertEqual(lingua_franca.get_default_loc(), 'en-us')
self.assertFalse('es' in lingua_franca.get_active_langs())
# Verify that unloaded languages can't be invoked explicitly
self.assertRaises(ModuleNotFoundError,
lingua_franca.parse.extract_number,
'uno', lang='es')
unload_all_languages()
def _duration_handler(time1,
lang=None,
speech=True,
*,
time2=None,
use_years=True,
clock=False,
resolution=TimeResolution.SECONDS):
"""Convert duration in seconds to a nice spoken timespan.
Used as a handler by nice_duration and nice_duration_dt.
Accepts:
datetime.timedelta, or
seconds (int/float), or
2 x datetime.datetime
Examples:
time1 = 60 -> "1:00" or "one minute"
time1 = 163 -> "2:43" or "two minutes forty three seconds"
time1 = timedelta(seconds=120) -> "2:00" or "two minutes"
time1 = datetime(2019, 3, 12),
time2 = datetime(2019, 1, 1) -> "seventy days"
Args:
time1: int/float seconds, OR datetime.timedelta, OR datetime.datetime
time2 (datetime, optional): subtracted from time1 if time1 is datetime
lang (str, optional): a BCP-47 language code, None for default
speech (bool, opt): format output for speech (True) or display (False)
use_years (bool, opt): rtn years and days if True, total days if False
clock (bool, opt): always format output like digital clock (see below)
resolution (mycroft.util.format.TimeResolution, optional): lower bound
mycroft.util.format.TimeResolution values:
TimeResolution.YEARS
TimeResolution.DAYS
TimeResolution.HOURS
TimeResolution.MINUTES
TimeResolution.SECONDS
TimeResolution.MILLISECONDS
NOTE: nice_duration will not produce milliseconds
unless that resolution is passed.
NOTE: clock will produce digital clock-like output appropriate to
resolution. Has no effect on resolutions DAYS or YEARS. Only
applies to displayed output.
Returns:
str: timespan as a string
"""
lang = lang or get_default_loc()
_leapdays = 0
_input_resolution = resolution
milliseconds = 0
type1 = type(time1)
if time2:
type2 = type(time2)
if type1 is not type2:
raise Exception("nice_duration() can't combine data types: "
"{} and {}".format(type1, type2))
elif type1 is datetime.datetime:
duration = time1 - time2
_leapdays = (abs(leapdays(time1.year, time2.year)))
# when operating on datetimes, refuse resolutions that
# would result in bunches of trailing zeroes
if all([
time1.second == 0, time2.second == 0,
resolution.value >= TimeResolution.SECONDS.value
]):
resolution = TimeResolution.MINUTES
if all([
time1.minute == 0, time2.minute == 0,
resolution.value == TimeResolution.MINUTES.value
]):
resolution = TimeResolution.HOURS
if all([
time1.hour == 0, time2.hour == 0,
resolution.value == TimeResolution.HOURS.value
]):
resolution = TimeResolution.DAYS
else:
_tmp = warnings.formatwarning
warnings.formatwarning = lambda msg, * \
args, **kwargs: "{}\n".format(msg)
warning = ("WARN: mycroft.util.format.nice_duration_dt() can't "
"subtract " + str(type1) + ". Ignoring 2nd "
"argument '" + str(time2) + "'.")
warnings.warn(warning)
warnings.formatwarning = _tmp
duration = time1
else:
duration = time1
# Pull decimal portion of seconds, if present, to use for milliseconds
if isinstance(duration, float):
milliseconds = str(duration).split('.')[1]
if speech:
milliseconds = milliseconds[:2]
else:
milliseconds = milliseconds[:3]
milliseconds = float("0." + milliseconds)
# Cast duration to datetime.timedelta for human-friendliness
if not isinstance(duration, datetime.timedelta):
duration = datetime.timedelta(seconds=duration)
days = duration.days
if use_years:
days -= _leapdays if days > 365 else 0
years = days // 365
else:
years = 0
days = days % 365 if years > 0 else days
# We already stored milliseconds. Now we want the integer part.
seconds = duration.seconds
minutes = seconds // 60
seconds %= 60
hours = minutes // 60
minutes %= 60
if speech:
out = ""
if years > 0:
out += pronounce_number(years, lang) + " "
out += _translate_word("year" if years == 1 else "years", lang)
if days > 0 and resolution.value > TimeResolution.YEARS.value:
if out:
out += " "
out += pronounce_number(days, lang) + " "
out += _translate_word("day" if days == 1 else "days", lang)
if hours > 0 and resolution.value > TimeResolution.DAYS.value:
if out:
out += " "
out += pronounce_number(hours, lang) + " "
out += _translate_word("hour" if hours == 1 else "hours", lang)
if minutes > 0 and resolution.value > TimeResolution.HOURS.value:
if out:
out += " "
out += pronounce_number(minutes, lang) + " "
out += _translate_word("minute" if minutes == 1 else "minutes",
lang)
if ((seconds > 0 and resolution.value >= TimeResolution.SECONDS.value)
or
(milliseconds > 0
and resolution.value == TimeResolution.MILLISECONDS.value)):
if resolution.value == TimeResolution.MILLISECONDS.value:
seconds += milliseconds
if out:
out += " "
# Throw "and" between minutes and seconds if duration < 1 hour
if len(out.split()) > 3 or seconds < 1:
out += _translate_word("and", lang) + " "
# speaking "zero point five seconds" is better than "point five"
out += pronounce_number(seconds, lang) + " "
out += _translate_word("second" if seconds == 1 else "seconds",
lang)
else:
# M:SS, MM:SS, H:MM:SS, Dd H:MM:SS format
_seconds_str = ("0" + str(seconds)) if seconds < 10 else str(seconds)
out = ""
if years > 0:
out = str(years) + "y "
if days > 0 and resolution.value > TimeResolution.YEARS.value:
out += str(days) + "d "
if (hours > 0 and resolution.value > TimeResolution.DAYS.value) or \
(clock and resolution is TimeResolution.HOURS):
out += str(hours)
if resolution.value == TimeResolution.MINUTES.value and not clock:
out += (("h " + str(minutes) +
"m") if hours > 0 else str(minutes) + "m")
elif (minutes > 0 and resolution.value > TimeResolution.HOURS.value) \
or (clock and resolution.value >= TimeResolution.HOURS.value):
if hours != 0 or (clock and resolution is TimeResolution.HOURS):
out += ":"
if minutes < 10:
out += "0"
out += str(minutes) + ":"
if (seconds > 0 and
resolution.value > TimeResolution.MINUTES.value) or clock:
out += _seconds_str
else:
out += "00"
# if we have seconds but no minutes...
elif (seconds > 0 or clock) and resolution.value > \
TimeResolution.MINUTES.value:
# check if output ends in hours
try:
if str(hours) == out.split()[-1]:
out += ":"
except IndexError:
pass
out += ("00:" if hours > 0 else "0:") + _seconds_str
if (milliseconds > 0 or clock) and resolution.value \
== TimeResolution.MILLISECONDS.value:
_mill = str(milliseconds).split(".")[1]
# right-pad milliseconds to three decimal places
while len(_mill) < 3:
_mill += "0"
# make sure output < 1s still formats correctly
if out == "":
out = "0:00"
else:
if (str(hours) == out.split()[-1]) and ":" not in out:
out += ":00:00"
# only append milliseconds to output that contains
# minutes and/or seconds
if ":" in out:
out += "." + _mill
# If this evaluates True, out currently ends in hours: "1d 12"
if out and all([
resolution.value >= TimeResolution.HOURS.value, ":" not in out,
out[-1] != "m", hours > 0
]):
# to "1d 12h"
out += "h"
out = out.strip()
if not out:
out = "zero " if speech else "0"
if _input_resolution == TimeResolution.YEARS:
out += "years" if speech else "y"
elif _input_resolution == TimeResolution.DAYS:
out += "days" if speech else "d"
elif _input_resolution == TimeResolution.HOURS:
out += "hours" if speech else "h"
elif _input_resolution == TimeResolution.MINUTES:
if speech:
out = "under a minute" if seconds > 0 else "zero minutes"
else:
out = "0m"
else:
out = "zero seconds" if speech else "0:00"
return out