#!/usr/bin/env python3
+# type: ignore
+# pylint: disable=W0201
+# pylint: disable=R0904
-"""
-Parse dates in a variety of formats.
+# © Copyright 2021-2022, Scott Gasch
-"""
+"""Parse dates in a variety of formats."""
import datetime
import functools
-import holidays # type: ignore
import logging
import re
import sys
import antlr4 # type: ignore
import dateutil.easter
import dateutil.tz
+import holidays # type: ignore
import pytz
import acl
import bootstrap
+import decorator_utils
+from dateparse.dateparse_utilsLexer import dateparse_utilsLexer # type: ignore
+from dateparse.dateparse_utilsListener import dateparse_utilsListener # type: ignore
+from dateparse.dateparse_utilsParser import dateparse_utilsParser # type: ignore
from datetime_utils import (
TimeUnit,
- n_timeunits_from_base,
- datetime_to_date,
date_to_datetime,
+ datetime_to_date,
+ n_timeunits_from_base,
)
-from dateparse.dateparse_utilsLexer import dateparse_utilsLexer # type: ignore
-from dateparse.dateparse_utilsListener import dateparse_utilsListener # type: ignore
-from dateparse.dateparse_utilsParser import dateparse_utilsParser # type: ignore
-import decorator_utils
-
logger = logging.getLogger(__name__)
"""An exception thrown during parsing because of unrecognized input."""
def __init__(self, message: str) -> None:
+ super().__init__()
self.message = message
def syntaxError(self, recognizer, offendingSymbol, line, column, msg, e):
raise ParseException(msg)
- def reportAmbiguity(
- self, recognizer, dfa, startIndex, stopIndex, exact, ambigAlts, configs
- ):
+ def reportAmbiguity(self, recognizer, dfa, startIndex, stopIndex, exact, ambigAlts, configs):
pass
def reportAttemptingFullContext(
):
pass
- def reportContextSensitivity(
- self, recognizer, dfa, startIndex, stopIndex, prediction, configs
- ):
+ def reportContextSensitivity(self, recognizer, dfa, startIndex, stopIndex, prediction, configs):
pass
),
)
class DateParser(dateparse_utilsListener):
+ """A class to parse dates expressed in human language."""
+
PARSE_TYPE_SINGLE_DATE_EXPR = 1
PARSE_TYPE_BASE_AND_OFFSET_EXPR = 2
PARSE_TYPE_SINGLE_TIME_EXPR = 3
'yea': TimeUnit.YEARS,
}
self.override_now_for_test_purposes = override_now_for_test_purposes
+
+ # Note: _reset defines several class fields. It is used both here
+ # in the c'tor but also in between parse operations to restore the
+ # class' state and allow it to be reused.
+ #
self._reset()
def parse(self, date_string: str) -> Optional[datetime.datetime]:
to timezone naive (i.e. tzinfo = None).
"""
dt = self.datetime
- if tz is not None:
- dt = dt.replace(tzinfo=None).astimezone(tz=tz)
+ if dt is not None:
+ if tz is not None:
+ dt = dt.replace(tzinfo=None).astimezone(tz=tz)
return dt
# -- helpers --
name = DateParser._normalize_special_day_name(self.context['special'])
# Yesterday, today, tomorrow -- ignore any next/last
- if name == 'today' or name == 'now':
+ if name in ('today', 'now'):
return today
if name == 'yeste':
return today + datetime.timedelta(days=-1)
day=self.context['day'],
)
- def _parse_tz(self, txt: str) -> Any:
+ @staticmethod
+ def _parse_tz(txt: str) -> Any:
if txt == 'Z':
txt = 'UTC'
# Try pytz
try:
- tz = pytz.timezone(txt)
- if tz is not None:
- return tz
+ tz1 = pytz.timezone(txt)
+ if tz1 is not None:
+ return tz1
except Exception:
pass
# Try dateutil
try:
- tz = dateutil.tz.gettz(txt)
- if tz is not None:
- return tz
+ tz2 = dateutil.tz.gettz(txt)
+ if tz2 is not None:
+ return tz2
except Exception:
pass
# Try constructing an offset in seconds
try:
- sign = txt[0]
- if sign == '-' or sign == '+':
- sign = +1 if sign == '+' else -1
+ txt_sign = txt[0]
+ if txt_sign in ('-', '+'):
+ sign = +1 if txt_sign == '+' else -1
hour = int(txt[1:3])
minute = int(txt[-2:])
offset = sign * (hour * 60 * 60) + sign * (minute * 60)
pass
return None
- def _get_int(self, txt: str) -> int:
+ @staticmethod
+ def _get_int(txt: str) -> int:
while not txt[0].isdigit() and txt[0] != '-' and txt[0] != '+':
txt = txt[1:]
while not txt[-1].isdigit():
txt = txt[:-1]
return int(txt)
- # -- overridden methods invoked by parse walk --
+ # -- overridden methods invoked by parse walk. Note: not part of the class'
+ # public API(!!) --
def visitErrorNode(self, node: antlr4.ErrorNode) -> None:
pass
micros = self.time.microsecond
self.datetime = datetime.datetime(
- year, month, day, hour, minute, second, micros, tzinfo=self.time.tzinfo
+ year,
+ month,
+ day,
+ hour,
+ minute,
+ second,
+ micros,
+ tzinfo=self.time.tzinfo,
)
# Apply resudual adjustments to times here when we have a
# datetime.
self.datetime = self.datetime + self.timedelta
+ assert self.datetime is not None
self.time = datetime.time(
self.datetime.hour,
self.datetime.minute,
# Adjust count's sign based on the presence of 'before' or 'after'.
if 'delta_before_after' in self.context:
before_after = self.context['delta_before_after'].lower()
- if (
- before_after == 'before'
- or before_after == 'until'
- or before_after == 'til'
- or before_after == 'to'
- ):
+ if before_after in ('before', 'until', 'til', 'to'):
count = -count
# What are we counting units of?
# Adjust count's sign based on the presence of 'before' or 'after'.
if 'time_delta_before_after' in self.context:
before_after = self.context['time_delta_before_after'].lower()
- if (
- before_after == 'before'
- or before_after == 'until'
- or before_after == 'til'
- or before_after == 'to'
- ):
+ if before_after in ('before', 'until', 'til', 'to'):
count = -count
# What are we counting units of... assume minutes.
elif unit == TimeUnit.HOURS:
self.timedelta = datetime.timedelta(hours=count)
else:
- raise ParseException()
+ raise ParseException(f'Invalid Unit: "{unit}"')
- def exitDeltaPlusMinusExpr(
- self, ctx: dateparse_utilsParser.DeltaPlusMinusExprContext
- ) -> None:
+ def exitDeltaPlusMinusExpr(self, ctx: dateparse_utilsParser.DeltaPlusMinusExprContext) -> None:
try:
n = ctx.nth()
if n is None:
raise ParseException(f'Bad N in Delta +/- Expr: {ctx.getText()}')
n = n.getText()
- n = self._get_int(n)
+ n = DateParser._get_int(n)
unit = self._figure_out_date_unit(ctx.deltaUnit().getText().lower())
- except Exception:
- raise ParseException(f'Invalid Delta +/-: {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Invalid Delta +/-: {ctx.getText()}') from e
else:
self.context['delta_int'] = n
self.context['delta_unit'] = unit
def exitNextLastUnit(self, ctx: dateparse_utilsParser.DeltaUnitContext) -> None:
try:
unit = self._figure_out_date_unit(ctx.getText().lower())
- except Exception:
- raise ParseException(f'Bad delta unit: {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Bad delta unit: {ctx.getText()}') from e
else:
self.context['delta_unit'] = unit
- def exitDeltaNextLast(
- self, ctx: dateparse_utilsParser.DeltaNextLastContext
- ) -> None:
+ def exitDeltaNextLast(self, ctx: dateparse_utilsParser.DeltaNextLastContext) -> None:
try:
txt = ctx.getText().lower()
- except Exception:
- raise ParseException(f'Bad next/last: {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Bad next/last: {ctx.getText()}') from e
if 'month' in self.context or 'day' in self.context or 'year' in self.context:
- raise ParseException(
- 'Next/last expression expected to be relative to today.'
- )
+ raise ParseException('Next/last expression expected to be relative to today.')
if txt[:4] == 'next':
self.context['delta_int'] = +1
self.context['day'] = self.now_datetime.day
try:
unit = self._figure_out_time_unit(ctx.deltaTimeUnit().getText().lower())
self.context['time_delta_unit'] = unit
- except Exception:
- raise ParseException(f'Bad delta unit: {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Bad delta unit: {ctx.getText()}') from e
if 'time_delta_before_after' not in self.context:
raise ParseException(f'Bad Before/After: {ctx.getText()}')
- def exitDeltaTimeFraction(
- self, ctx: dateparse_utilsParser.DeltaTimeFractionContext
- ) -> None:
+ def exitDeltaTimeFraction(self, ctx: dateparse_utilsParser.DeltaTimeFractionContext) -> None:
try:
txt = ctx.getText().lower()[:4]
if txt == 'quar':
self.context['time_delta_unit'] = TimeUnit.MINUTES
else:
raise ParseException(f'Bad time fraction {ctx.getText()}')
- except Exception:
- raise ParseException(f'Bad time fraction {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Bad time fraction {ctx.getText()}') from e
- def exitDeltaBeforeAfter(
- self, ctx: dateparse_utilsParser.DeltaBeforeAfterContext
- ) -> None:
+ def exitDeltaBeforeAfter(self, ctx: dateparse_utilsParser.DeltaBeforeAfterContext) -> None:
try:
txt = ctx.getText().lower()
- except Exception:
- raise ParseException(f'Bad delta before|after: {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Bad delta before|after: {ctx.getText()}') from e
else:
self.context['delta_before_after'] = txt
- def exitDeltaTimeBeforeAfter(
- self, ctx: dateparse_utilsParser.DeltaBeforeAfterContext
- ) -> None:
+ def exitDeltaTimeBeforeAfter(self, ctx: dateparse_utilsParser.DeltaBeforeAfterContext) -> None:
try:
txt = ctx.getText().lower()
- except Exception:
- raise ParseException(f'Bad delta before|after: {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Bad delta before|after: {ctx.getText()}') from e
else:
self.context['time_delta_before_after'] = txt
self.context['month'] = month
self.context['day'] = 1
self.main_type = DateParser.PARSE_TYPE_BASE_AND_OFFSET_EXPR
- except Exception:
- raise ParseException(f'Invalid nthWeekday expression: {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Invalid nthWeekday expression: {ctx.getText()}') from e
def exitFirstLastWeekdayInMonthMaybeYearExpr(
self,
def exitNth(self, ctx: dateparse_utilsParser.NthContext) -> None:
try:
- i = self._get_int(ctx.getText())
- except Exception:
- raise ParseException(f'Bad nth expression: {ctx.getText()}')
+ i = DateParser._get_int(ctx.getText())
+ except Exception as e:
+ raise ParseException(f'Bad nth expression: {ctx.getText()}') from e
else:
self.context['nth'] = i
txt = -1
else:
raise ParseException(f'Bad first|last expression: {ctx.getText()}')
- except Exception:
- raise ParseException(f'Bad first|last expression: {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Bad first|last expression: {ctx.getText()}') from e
else:
self.context['nth'] = txt
try:
dow = ctx.getText().lower()[:3]
dow = self.day_name_to_number.get(dow, None)
- except Exception:
- raise ParseException('Bad day of week')
+ except Exception as e:
+ raise ParseException('Bad day of week') from e
else:
self.context['dow'] = dow
if day[:3] == 'kal':
self.context['day'] = 1
return
- day = self._get_int(day)
+ day = DateParser._get_int(day)
if day < 1 or day > 31:
raise ParseException(f'Bad dayOfMonth expression: {ctx.getText()}')
- except Exception:
- raise ParseException(f'Bad dayOfMonth expression: {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Bad dayOfMonth expression: {ctx.getText()}') from e
self.context['day'] = day
def exitMonthName(self, ctx: dateparse_utilsParser.MonthNameContext) -> None:
month = self.month_name_to_number.get(month, None)
if month is None:
raise ParseException(f'Bad monthName expression: {ctx.getText()}')
- except Exception:
- raise ParseException(f'Bad monthName expression: {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Bad monthName expression: {ctx.getText()}') from e
else:
self.context['month'] = month
def exitMonthNumber(self, ctx: dateparse_utilsParser.MonthNumberContext) -> None:
try:
- month = self._get_int(ctx.getText())
+ month = DateParser._get_int(ctx.getText())
if month < 1 or month > 12:
raise ParseException(f'Bad monthNumber expression: {ctx.getText()}')
- except Exception:
- raise ParseException(f'Bad monthNumber expression: {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Bad monthNumber expression: {ctx.getText()}') from e
else:
self.context['month'] = month
def exitYear(self, ctx: dateparse_utilsParser.YearContext) -> None:
try:
- year = self._get_int(ctx.getText())
+ year = DateParser._get_int(ctx.getText())
if year < 1:
raise ParseException(f'Bad year expression: {ctx.getText()}')
- except Exception:
- raise ParseException(f'Bad year expression: {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Bad year expression: {ctx.getText()}') from e
else:
self.saw_overt_year = True
self.context['year'] = year
try:
special = ctx.specialDate().getText().lower()
self.context['special'] = special
- except Exception:
+ except Exception as e:
raise ParseException(
f'Bad specialDate expression: {ctx.specialDate().getText()}'
- )
+ ) from e
try:
mod = ctx.thisNextLast()
if mod is not None:
self.context['special_next_last'] = 'next'
elif mod.LAST() is not None:
self.context['special_next_last'] = 'last'
- except Exception:
- raise ParseException(f'Bad specialDateNextLast expression: {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Bad specialDateNextLast expression: {ctx.getText()}') from e
def exitNFoosFromTodayAgoExpr(
self, ctx: dateparse_utilsParser.NFoosFromTodayAgoExprContext
) -> None:
d = self.now_datetime
try:
- count = self._get_int(ctx.unsignedInt().getText())
+ count = DateParser._get_int(ctx.unsignedInt().getText())
unit = ctx.deltaUnit().getText().lower()
ago_from_now = ctx.AGO_FROM_NOW().getText()
- except Exception:
- raise ParseException(f'Bad NFoosFromTodayAgoExpr: {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Bad NFoosFromTodayAgoExpr: {ctx.getText()}') from e
if "ago" in ago_from_now or "back" in ago_from_now:
count = -count
count = +1
else:
raise ParseException(f'Bad This/Next/Last modifier: {mod}')
- except Exception:
- raise ParseException(f'Bad DeltaRelativeToTodayExpr: {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Bad DeltaRelativeToTodayExpr: {ctx.getText()}') from e
d = n_timeunits_from_base(count, TimeUnit(unit), d)
self.context['year'] = d.year
self.context['month'] = d.month
self.context['day'] = d.day
- def exitSpecialTimeExpr(
- self, ctx: dateparse_utilsParser.SpecialTimeExprContext
- ) -> None:
+ def exitSpecialTimeExpr(self, ctx: dateparse_utilsParser.SpecialTimeExprContext) -> None:
try:
txt = ctx.specialTime().getText().lower()
- except Exception:
- raise ParseException(f'Bad special time expression: {ctx.getText()}')
+ except Exception as e:
+ raise ParseException(f'Bad special time expression: {ctx.getText()}') from e
else:
- if txt == 'noon' or txt == 'midday':
+ if txt in ('noon', 'midday'):
self.context['hour'] = 12
self.context['minute'] = 0
self.context['seconds'] = 0
try:
tz = ctx.tzExpr().getText()
- self.context['tz'] = self._parse_tz(tz)
+ self.context['tz'] = DateParser._parse_tz(tz)
except Exception:
pass
- def exitTwelveHourTimeExpr(
- self, ctx: dateparse_utilsParser.TwelveHourTimeExprContext
- ) -> None:
+ def exitTwelveHourTimeExpr(self, ctx: dateparse_utilsParser.TwelveHourTimeExprContext) -> None:
try:
hour = ctx.hour().getText()
while not hour[-1].isdigit():
hour = hour[:-1]
- hour = self._get_int(hour)
- except Exception:
- raise ParseException(f'Bad hour: {ctx.hour().getText()}')
+ hour = DateParser._get_int(hour)
+ except Exception as e:
+ raise ParseException(f'Bad hour: {ctx.hour().getText()}') from e
if hour <= 0 or hour > 12:
raise ParseException(f'Bad hour (out of range): {hour}')
try:
- minute = self._get_int(ctx.minute().getText())
+ minute = DateParser._get_int(ctx.minute().getText())
except Exception:
minute = 0
if minute < 0 or minute > 59:
self.context['minute'] = minute
try:
- seconds = self._get_int(ctx.second().getText())
+ seconds = DateParser._get_int(ctx.second().getText())
except Exception:
seconds = 0
if seconds < 0 or seconds > 59:
self.context['seconds'] = seconds
try:
- micros = self._get_int(ctx.micros().getText())
+ micros = DateParser._get_int(ctx.micros().getText())
except Exception:
micros = 0
if micros < 0 or micros > 1000000:
try:
ampm = ctx.ampm().getText()
- except Exception:
- raise ParseException(f'Bad ampm: {ctx.ampm().getText()}')
+ except Exception as e:
+ raise ParseException(f'Bad ampm: {ctx.ampm().getText()}') from e
if hour == 12:
hour = 0
if ampm[0] == 'p':
try:
tz = ctx.tzExpr().getText()
- self.context['tz'] = self._parse_tz(tz)
+ self.context['tz'] = DateParser._parse_tz(tz)
except Exception:
pass
hour = ctx.hour().getText()
while not hour[-1].isdigit():
hour = hour[:-1]
- hour = self._get_int(hour)
- except Exception:
- raise ParseException(f'Bad hour: {ctx.hour().getText()}')
+ hour = DateParser._get_int(hour)
+ except Exception as e:
+ raise ParseException(f'Bad hour: {ctx.hour().getText()}') from e
if hour < 0 or hour > 23:
raise ParseException(f'Bad hour (out of range): {hour}')
self.context['hour'] = hour
try:
- minute = self._get_int(ctx.minute().getText())
+ minute = DateParser._get_int(ctx.minute().getText())
except Exception:
minute = 0
if minute < 0 or minute > 59:
self.context['minute'] = minute
try:
- seconds = self._get_int(ctx.second().getText())
+ seconds = DateParser._get_int(ctx.second().getText())
except Exception:
seconds = 0
if seconds < 0 or seconds > 59:
self.context['seconds'] = seconds
try:
- micros = self._get_int(ctx.micros().getText())
+ micros = DateParser._get_int(ctx.micros().getText())
except Exception:
micros = 0
if micros < 0 or micros >= 1000000:
try:
tz = ctx.tzExpr().getText()
- self.context['tz'] = self._parse_tz(tz)
+ self.context['tz'] = DateParser._parse_tz(tz)
except Exception:
pass
logger.exception(e)
print("Unrecognized.")
else:
+ assert dt is not None
print(dt.strftime('%A %Y/%m/%d %H:%M:%S.%f %Z(%z)'))
sys.exit(0)
projects / python_utils.git / blobdiff