def at_time(self, start, end=None, sort_diff=True, **window_kwargs):
if is_mp(start):
start = start.start_date
end = start.end_date
elif isinstance(start, str) and end is None:
mp = parse_date_tz(start)
start = mp.start_date
end = mp.end_date
elif isinstance(start, str) and isinstance(end, str):
mp = parse_date_tz(start)
start = mp.start_date
mp = parse_date_tz(end)
end = mp.end_date
elif end is None and window_kwargs:
end = start
elif end is None:
raise ValueError("Either a metaperiod, a date string, 2 times, or time + window_kwargs.")
self.infer_time()
if window_kwargs:
start = start - pd.Timedelta(**window_kwargs)
end = end + pd.Timedelta(**window_kwargs)
if self._start_col is None:
res = self[ab_overlap_c(start, end, self[self._time_col])]
else:
res = self[ab_overlap_cd(self[self._start_col], self[self._end_col], start, end)]
if not res.empty and sort_diff:
# avg_time = start + (end - start) / 2
# res["sort_score"] = -abs(res[self._time_col] - avg_time)
# res = res.sort_values('sort_score').drop('sort_score', axis=1)
res["sort_score"] = res[self._time_col]
res = res.sort_values("sort_score").drop("sort_score", axis=1)
return self.__class__(res)
def load(cls, nrows=None):
files = just.glob("~/nostalgia_data/input/mijn_chipkaart/*.csv")
data = pd.concat([pd.read_csv(x, sep=";", nrows=nrows) for x in files])
data["Bedrag"] = [float(x.replace(",", ".")) for x in data["Bedrag"]]
data["Datum"] = [
parse_date_tz(x + " " + y).start_date for x, y in zip(data.Datum, data["Check-uit"])
]
return cls(data)
def _select_at_day(self, day_or_class):
if isinstance(day_or_class, pd.DataFrame):
days = day_or_class.time.dt.date.unique()
return self.time.dt.date.isin(days)
elif isinstance(day_or_class, (list, tuple, set, pd.Series)):
return self.time.dt.date.isin(set(day_or_class))
else:
mp = parse_date_tz(day_or_class)
return (self.time.dt.date >= mp.start_date.date()) & (self.time.dt.date < mp.end_date.date())
def at(self, time_or_place):
if isinstance(time_or_place, NDF) and time_or_place.df_name.endswith("places"):
return self.when_at(time_or_place)
if isinstance(time_or_place, str):
mp = parse_date_tz(time_or_place)
if mp:
start = mp.start_date
end = mp.end_date
return self.at_time(start, end)
else:
return self.when_at(get_type_from_registry("places").containing(time_or_place))
raise ValueError("neither time nor place was passed")
def find_entities(sentence):
global ts
if ts is None:
ts = get_ts()
mp = parse_date_tz(sentence)
try:
ents = ts.findall(sentence)
except AttributeError:
raise AttributeError("No entities have been registered")
# remove metaperiod tokens from otherwise matching
if mp is not None:
wrongs = set()
for l, e in mp.spans:
wrongs.update(range(l, e))
ents = [x for x in ents if x.start not in wrongs and x.end not in wrongs]
ents.append(ResultInfo("MP", "filter", at_time_wrapper(mp), orig_word=" ".join(mp.matches)))
return ents
def at(self, time):
mp = parse_date_tz(time)
return self[self.index.overlaps(pd.Interval(mp.start_date, mp.end_date))]