summaryrefslogtreecommitdiff
path: root/dateparse/dateparse_utils.py
diff options
context:
space:
mode:
authorScott <[email protected]>2022-01-31 12:00:11 -0800
committerScott <[email protected]>2022-01-31 12:00:11 -0800
commite70297b7e29a07457d6c7195425e734a1290abeb (patch)
tree676cdb84991f936e45fc88f36c90e8dc0b0088ae /dateparse/dateparse_utils.py
parent679b41526d9327f5bcecc702a928a72b68a1e0c8 (diff)
Fix a couple of bugs in date parsing.
Diffstat (limited to 'dateparse/dateparse_utils.py')
-rwxr-xr-xdateparse/dateparse_utils.py253
1 files changed, 101 insertions, 152 deletions
diff --git a/dateparse/dateparse_utils.py b/dateparse/dateparse_utils.py
index 21fdb83..be5e1b5 100755
--- a/dateparse/dateparse_utils.py
+++ b/dateparse/dateparse_utils.py
@@ -21,7 +21,10 @@ import pytz
import acl
import bootstrap
from datetime_utils import (
- TimeUnit, n_timeunits_from_base, datetime_to_date, date_to_datetime
+ TimeUnit,
+ n_timeunits_from_base,
+ datetime_to_date,
+ date_to_datetime,
)
from dateparse.dateparse_utilsLexer import dateparse_utilsLexer # type: ignore
from dateparse.dateparse_utilsListener import dateparse_utilsListener # type: ignore
@@ -35,50 +38,46 @@ logger = logging.getLogger(__name__)
def debug_parse(enter_or_exit_f: Callable[[Any, Any], None]):
@functools.wraps(enter_or_exit_f)
def debug_parse_wrapper(*args, **kwargs):
- slf = args[0]
+ # slf = args[0]
ctx = args[1]
depth = ctx.depth()
logger.debug(
- ' ' * (depth-1) +
- f'Entering {enter_or_exit_f.__name__} ({ctx.invokingState} / {ctx.exception})'
+ ' ' * (depth - 1)
+ + f'Entering {enter_or_exit_f.__name__} ({ctx.invokingState} / {ctx.exception})'
)
for c in ctx.getChildren():
- logger.debug(
- ' ' * (depth-1) +
- f'{c} {type(c)}'
- )
+ logger.debug(' ' * (depth - 1) + f'{c} {type(c)}')
retval = enter_or_exit_f(*args, **kwargs)
return retval
+
return debug_parse_wrapper
class ParseException(Exception):
"""An exception thrown during parsing because of unrecognized input."""
+
def __init__(self, message: str) -> None:
self.message = message
class RaisingErrorListener(antlr4.DiagnosticErrorListener):
"""An error listener that raises ParseExceptions."""
- def syntaxError(
- self, recognizer, offendingSymbol, line, column, msg, e
- ):
+
+ def syntaxError(self, recognizer, offendingSymbol, line, column, msg, e):
raise ParseException(msg)
def reportAmbiguity(
- self, recognizer, dfa, startIndex, stopIndex, exact,
- ambigAlts, configs
+ self, recognizer, dfa, startIndex, stopIndex, exact, ambigAlts, configs
):
pass
def reportAttemptingFullContext(
- self, recognizer, dfa, startIndex, stopIndex, conflictingAlts,
- configs
+ self, recognizer, dfa, startIndex, stopIndex, conflictingAlts, configs
):
pass
def reportContextSensitivity(
- self, recognizer, dfa, startIndex, stopIndex, prediction, configs
+ self, recognizer, dfa, startIndex, stopIndex, prediction, configs
):
pass
@@ -90,13 +89,10 @@ class RaisingErrorListener(antlr4.DiagnosticErrorListener):
'enter*',
'exit*',
],
- denied_patterns=[
- 'enterEveryRule',
- 'exitEveryRule'
- ],
+ denied_patterns=['enterEveryRule', 'exitEveryRule'],
order_to_check_allow_deny=acl.Order.DENY_ALLOW,
- default_answer=False
- )
+ default_answer=False,
+ ),
)
class DateParser(dateparse_utilsListener):
PARSE_TYPE_SINGLE_DATE_EXPR = 1
@@ -104,11 +100,7 @@ class DateParser(dateparse_utilsListener):
PARSE_TYPE_SINGLE_TIME_EXPR = 3
PARSE_TYPE_BASE_AND_OFFSET_TIME_EXPR = 4
- def __init__(
- self,
- *,
- override_now_for_test_purposes = None
- ) -> None:
+ def __init__(self, *, override_now_for_test_purposes=None) -> None:
"""C'tor. Passing a value to override_now_for_test_purposes can be
used to force this instance to use a custom date/time for its
idea of "now" so that the code can be more easily unittested.
@@ -142,7 +134,7 @@ class DateParser(dateparse_utilsListener):
9: 30,
10: 31,
11: 30,
- 12: 31
+ 12: 31,
}
# N.B. day number is also synched with datetime_utils.TimeUnit values
@@ -198,7 +190,7 @@ class DateParser(dateparse_utilsListener):
This is the main entrypoint to this class for caller code.
"""
date_string = date_string.strip()
- date_string = re.sub('\s+', ' ', date_string)
+ date_string = re.sub(r'\s+', ' ', date_string)
self._reset()
listener = RaisingErrorListener()
input_stream = antlr4.InputStream(date_string)
@@ -246,9 +238,7 @@ class DateParser(dateparse_utilsListener):
self.today = datetime.date.today()
else:
self.now_datetime = self.override_now_for_test_purposes
- self.today = datetime_to_date(
- self.override_now_for_test_purposes
- )
+ self.today = datetime_to_date(self.override_now_for_test_purposes)
self.date: Optional[datetime.date] = None
self.time: Optional[datetime.time] = None
self.datetime: Optional[datetime.datetime] = None
@@ -275,16 +265,16 @@ class DateParser(dateparse_utilsListener):
return TimeUnit.MONTHS
txt = orig.lower()[:3]
if txt in self.day_name_to_number:
- return(TimeUnit(self.day_name_to_number[txt]))
+ return TimeUnit(self.day_name_to_number[txt])
elif txt in self.delta_unit_to_constant:
- return(TimeUnit(self.delta_unit_to_constant[txt]))
+ return TimeUnit(self.delta_unit_to_constant[txt])
raise ParseException(f'Invalid date unit: {orig}')
def _figure_out_time_unit(self, orig: str) -> int:
"""Figure out what unit a time expression piece is talking about."""
txt = orig.lower()[:3]
if txt in self.time_delta_unit_to_constant:
- return(self.time_delta_unit_to_constant[txt])
+ return self.time_delta_unit_to_constant[txt]
raise ParseException(f'Invalid time unit: {orig}')
def _parse_special_date(self, name: str) -> Optional[datetime.date]:
@@ -317,13 +307,9 @@ class DateParser(dateparse_utilsListener):
elif name == 'hallo':
return datetime.date(year=year, month=10, day=31)
- for holiday_date, holiday_name in sorted(
- holidays.US(years=year).items()
- ):
+ for holiday_date, holiday_name in sorted(holidays.US(years=year).items()):
if 'Observed' not in holiday_name:
- holiday_name = DateParser._normalize_special_day_name(
- holiday_name
- )
+ holiday_name = DateParser._normalize_special_day_name(holiday_name)
if name == holiday_name:
return holiday_date
if name == 'chriseve':
@@ -373,10 +359,7 @@ class DateParser(dateparse_utilsListener):
self.saw_overt_year = True
# Handling "ides" and "nones" requires both the day and month.
- if (
- self.context['day'] == 'ide' or
- self.context['day'] == 'non'
- ):
+ if self.context['day'] == 'ide' or self.context['day'] == 'non':
self.context['day'] = self._resolve_ides_nones(
self.context['day'], self.context['month']
)
@@ -452,8 +435,7 @@ class DateParser(dateparse_utilsListener):
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
@@ -464,7 +446,7 @@ class DateParser(dateparse_utilsListener):
self.datetime.minute,
self.datetime.second,
self.datetime.microsecond,
- self.datetime.tzinfo
+ self.datetime.tzinfo,
)
def enterDateExpr(self, ctx: dateparse_utilsParser.DateExprContext):
@@ -505,10 +487,10 @@ class DateParser(dateparse_utilsListener):
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'
+ before_after == 'before'
+ or before_after == 'until'
+ or before_after == 'til'
+ or before_after == 'to'
):
count = -count
@@ -516,11 +498,7 @@ class DateParser(dateparse_utilsListener):
if 'delta_unit' not in self.context:
raise ParseException('Missing delta_unit?!')
unit = self.context['delta_unit']
- dt = n_timeunits_from_base(
- count,
- TimeUnit(unit),
- date_to_datetime(self.date)
- )
+ dt = n_timeunits_from_base(count, TimeUnit(unit), date_to_datetime(self.date))
self.date = datetime_to_date(dt)
def exitTimeExpr(self, ctx: dateparse_utilsParser.TimeExprContext) -> None:
@@ -550,10 +528,10 @@ class DateParser(dateparse_utilsListener):
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'
+ before_after == 'before'
+ or before_after == 'until'
+ or before_after == 'til'
+ or before_after == 'to'
):
count = -count
@@ -577,23 +555,17 @@ class DateParser(dateparse_utilsListener):
try:
n = ctx.nth()
if n is None:
- raise ParseException(
- f'Bad N in Delta +/- Expr: {ctx.getText()}'
- )
+ raise ParseException(f'Bad N in Delta +/- Expr: {ctx.getText()}')
n = n.getText()
n = self._get_int(n)
- unit = self._figure_out_date_unit(
- ctx.deltaUnit().getText().lower()
- )
+ unit = self._figure_out_date_unit(ctx.deltaUnit().getText().lower())
except Exception:
raise ParseException(f'Invalid Delta +/-: {ctx.getText()}')
else:
self.context['delta_int'] = n
self.context['delta_unit'] = unit
- def exitNextLastUnit(
- self, ctx: dateparse_utilsParser.DeltaUnitContext
- ) -> None:
+ def exitNextLastUnit(self, ctx: dateparse_utilsParser.DeltaUnitContext) -> None:
try:
unit = self._figure_out_date_unit(ctx.getText().lower())
except Exception:
@@ -602,17 +574,13 @@ class DateParser(dateparse_utilsListener):
self.context['delta_unit'] = unit
def exitDeltaNextLast(
- self, ctx: dateparse_utilsParser.DeltaNextLastContext
+ self, ctx: dateparse_utilsParser.DeltaNextLastContext
) -> None:
try:
txt = ctx.getText().lower()
except Exception:
raise ParseException(f'Bad next/last: {ctx.getText()}')
- if (
- 'month' in self.context or
- 'day' in self.context or
- 'year' in self.context
- ):
+ 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.'
)
@@ -635,36 +603,26 @@ class DateParser(dateparse_utilsListener):
self, ctx: dateparse_utilsParser.CountUnitsBeforeAfterTimeExprContext
) -> None:
if 'nth' not in self.context:
- raise ParseException(
- f'Bad count expression: {ctx.getText()}'
- )
+ raise ParseException(f'Bad count expression: {ctx.getText()}')
try:
- unit = self._figure_out_time_unit(
- ctx.deltaTimeUnit().getText().lower()
- )
+ 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()}')
if 'time_delta_before_after' not in self.context:
- raise ParseException(
- f'Bad Before/After: {ctx.getText()}'
- )
+ raise ParseException(f'Bad Before/After: {ctx.getText()}')
def exitDeltaTimeFraction(
- self, ctx: dateparse_utilsParser.DeltaTimeFractionContext
+ self, ctx: dateparse_utilsParser.DeltaTimeFractionContext
) -> None:
try:
txt = ctx.getText().lower()[:4]
if txt == 'quar':
self.context['time_delta_int'] = 15
- self.context[
- 'time_delta_unit'
- ] = TimeUnit.MINUTES
+ self.context['time_delta_unit'] = TimeUnit.MINUTES
elif txt == 'half':
self.context['time_delta_int'] = 30
- self.context[
- 'time_delta_unit'
- ] = TimeUnit.MINUTES
+ self.context['time_delta_unit'] = TimeUnit.MINUTES
else:
raise ParseException(f'Bad time fraction {ctx.getText()}')
except Exception:
@@ -712,9 +670,7 @@ class DateParser(dateparse_utilsListener):
year = self.context.get('year', self.today.year)
if 'month' not in self.context:
- raise ParseException(
- f'Missing month expression: {ctx.getText()}'
- )
+ raise ParseException(f'Missing month expression: {ctx.getText()}')
month = self.context['month']
dow = self.context['dow']
@@ -746,9 +702,7 @@ class DateParser(dateparse_utilsListener):
self.context['day'] = 1
self.main_type = DateParser.PARSE_TYPE_BASE_AND_OFFSET_EXPR
except Exception:
- raise ParseException(
- f'Invalid nthWeekday expression: {ctx.getText()}'
- )
+ raise ParseException(f'Invalid nthWeekday expression: {ctx.getText()}')
def exitFirstLastWeekdayInMonthMaybeYearExpr(
self,
@@ -764,9 +718,7 @@ class DateParser(dateparse_utilsListener):
else:
self.context['nth'] = i
- def exitFirstOrLast(
- self, ctx: dateparse_utilsParser.FirstOrLastContext
- ) -> None:
+ def exitFirstOrLast(self, ctx: dateparse_utilsParser.FirstOrLastContext) -> None:
try:
txt = ctx.getText()
if txt == 'first':
@@ -774,9 +726,7 @@ class DateParser(dateparse_utilsListener):
elif txt == 'last':
txt = -1
else:
- raise ParseException(
- f'Bad first|last expression: {ctx.getText()}'
- )
+ raise ParseException(f'Bad first|last expression: {ctx.getText()}')
except Exception:
raise ParseException(f'Bad first|last expression: {ctx.getText()}')
else:
@@ -791,9 +741,7 @@ class DateParser(dateparse_utilsListener):
else:
self.context['dow'] = dow
- def exitDayOfMonth(
- self, ctx: dateparse_utilsParser.DayOfMonthContext
- ) -> None:
+ def exitDayOfMonth(self, ctx: dateparse_utilsParser.DayOfMonthContext) -> None:
try:
day = ctx.getText().lower()
if day[:3] == 'ide':
@@ -807,16 +755,12 @@ class DateParser(dateparse_utilsListener):
return
day = self._get_int(day)
if day < 1 or day > 31:
- raise ParseException(
- f'Bad dayOfMonth expression: {ctx.getText()}'
- )
+ raise ParseException(f'Bad dayOfMonth expression: {ctx.getText()}')
except Exception:
raise ParseException(f'Bad dayOfMonth expression: {ctx.getText()}')
self.context['day'] = day
- def exitMonthName(
- self, ctx: dateparse_utilsParser.MonthNameContext
- ) -> None:
+ def exitMonthName(self, ctx: dateparse_utilsParser.MonthNameContext) -> None:
try:
month = ctx.getText()
while month[0] == '/' or month[0] == '-':
@@ -824,27 +768,19 @@ class DateParser(dateparse_utilsListener):
month = month[:3].lower()
month = self.month_name_to_number.get(month, None)
if month is None:
- raise ParseException(
- f'Bad monthName expression: {ctx.getText()}'
- )
+ raise ParseException(f'Bad monthName expression: {ctx.getText()}')
except Exception:
raise ParseException(f'Bad monthName expression: {ctx.getText()}')
else:
self.context['month'] = month
- def exitMonthNumber(
- self, ctx: dateparse_utilsParser.MonthNumberContext
- ) -> None:
+ def exitMonthNumber(self, ctx: dateparse_utilsParser.MonthNumberContext) -> None:
try:
month = self._get_int(ctx.getText())
if month < 1 or month > 12:
- raise ParseException(
- f'Bad monthNumber expression: {ctx.getText()}'
- )
+ raise ParseException(f'Bad monthNumber expression: {ctx.getText()}')
except Exception:
- raise ParseException(
- f'Bad monthNumber expression: {ctx.getText()}'
- )
+ raise ParseException(f'Bad monthNumber expression: {ctx.getText()}')
else:
self.context['month'] = month
@@ -879,9 +815,7 @@ class DateParser(dateparse_utilsListener):
elif mod.LAST() is not None:
self.context['special_next_last'] = 'last'
except Exception:
- raise ParseException(
- f'Bad specialDateNextLast expression: {ctx.getText()}'
- )
+ raise ParseException(f'Bad specialDateNextLast expression: {ctx.getText()}')
def exitNFoosFromTodayAgoExpr(
self, ctx: dateparse_utilsParser.NFoosFromTodayAgoExprContext
@@ -892,18 +826,13 @@ class DateParser(dateparse_utilsListener):
unit = ctx.deltaUnit().getText().lower()
ago_from_now = ctx.AGO_FROM_NOW().getText()
except Exception:
- raise ParseException(
- f'Bad NFoosFromTodayAgoExpr: {ctx.getText()}'
- )
+ raise ParseException(f'Bad NFoosFromTodayAgoExpr: {ctx.getText()}')
if "ago" in ago_from_now or "back" in ago_from_now:
count = -count
unit = self._figure_out_date_unit(unit)
- d = n_timeunits_from_base(
- count,
- TimeUnit(unit),
- d)
+ d = n_timeunits_from_base(count, TimeUnit(unit), d)
self.context['year'] = d.year
self.context['month'] = d.month
self.context['day'] = d.day
@@ -911,29 +840,50 @@ class DateParser(dateparse_utilsListener):
def exitDeltaRelativeToTodayExpr(
self, ctx: dateparse_utilsParser.DeltaRelativeToTodayExprContext
) -> None:
+ # When someone says "next week" they mean a week from now.
+ # Likewise next month or last year. These expressions are now
+ # +/- delta.
+ #
+ # But when someone says "this Friday" they mean "this coming
+ # Friday". It would be weird to say "this Friday" if today
+ # was already Friday but I'm parsing it to mean: the next day
+ # that is a Friday. So when you say "next Friday" you mean
+ # the Friday after this coming Friday, or 2 Fridays from now.
+ #
+ # This set handles this weirdness.
+ weekdays = set(
+ [
+ TimeUnit.MONDAYS,
+ TimeUnit.TUESDAYS,
+ TimeUnit.WEDNESDAYS,
+ TimeUnit.THURSDAYS,
+ TimeUnit.FRIDAYS,
+ TimeUnit.SATURDAYS,
+ TimeUnit.SUNDAYS,
+ ]
+ )
d = self.now_datetime
try:
mod = ctx.thisNextLast()
+ unit = ctx.deltaUnit().getText().lower()
+ unit = self._figure_out_date_unit(unit)
if mod.LAST():
count = -1
elif mod.THIS():
- count = +1
+ if unit in weekdays:
+ count = +1
+ else:
+ count = 0
elif mod.NEXT():
- count = +2
+ if unit in weekdays:
+ count = +2
+ else:
+ count = +1
else:
- raise ParseException(
- f'Bad This/Next/Last modifier: {mod}'
- )
- unit = ctx.deltaUnit().getText().lower()
+ raise ParseException(f'Bad This/Next/Last modifier: {mod}')
except Exception:
- raise ParseException(
- f'Bad DeltaRelativeToTodayExpr: {ctx.getText()}'
- )
- unit = self._figure_out_date_unit(unit)
- d = n_timeunits_from_base(
- count,
- TimeUnit(unit),
- d)
+ raise ParseException(f'Bad DeltaRelativeToTodayExpr: {ctx.getText()}')
+ d = n_timeunits_from_base(count, TimeUnit(unit), d)
self.context['year'] = d.year
self.context['month'] = d.month
self.context['day'] = d.day
@@ -944,9 +894,7 @@ class DateParser(dateparse_utilsListener):
try:
txt = ctx.specialTime().getText().lower()
except Exception:
- raise ParseException(
- f'Bad special time expression: {ctx.getText()}'
- )
+ raise ParseException(f'Bad special time expression: {ctx.getText()}')
else:
if txt == 'noon' or txt == 'midday':
self.context['hour'] = 12
@@ -1076,6 +1024,7 @@ def main() -> None:
try:
dt = parser.parse(line)
except Exception as e:
+ logger.exception(e)
print("Unrecognized.")
else:
print(dt.strftime('%A %Y/%m/%d %H:%M:%S.%f %Z(%z)'))