From a1aba2dd768f21d9780534ce7ca8e6e624235c08 Mon Sep 17 00:00:00 2001
From: Tulir Asokan <tulir@maunium.net>
Date: Wed, 25 Mar 2020 23:43:30 +0200
Subject: [PATCH] Return unconsumed part instead of end index from matcher
Also removed unused regexes
---
reminder/locale_util.py | 62 ++++++++++++++++++++---------------------
reminder/locales.py | 44 ++++++++++++++++++-----------
reminder/util.py | 27 +-----------------
3 files changed, 58 insertions(+), 75 deletions(-)
diff --git a/reminder/locale_util.py b/reminder/locale_util.py
index c1e6551..604878b 100644
--- a/reminder/locale_util.py
+++ b/reminder/locale_util.py
@@ -53,12 +53,12 @@ if TYPE_CHECKING:
class MatcherReturn(NamedTuple):
params: 'RelativeDeltaParams'
- end: int
+ unconsumed: str
class Matcher(ABC):
@abstractmethod
- def match(self, val: str, start: int = 0) -> Optional[MatcherReturn]:
+ def match(self, val: str) -> Optional[MatcherReturn]:
pass
@@ -70,45 +70,44 @@ class RegexMatcher(Matcher):
self.regex = re.compile(pattern, re.IGNORECASE)
self.value_type = value_type
- def match(self, val: str, start: int = 0) -> Optional[MatcherReturn]:
- match = self.regex.match(val, pos=start)
- if match and match.end() > 0:
- return self._convert_match(match)
+ def match(self, val: str) -> Optional[MatcherReturn]:
+ match = self.regex.match(val)
+ if match and match.end() > 0 and len(match.groups()) > 0:
+ return MatcherReturn(params=self._convert_match(match), unconsumed=val[match.end():])
return None
- def _convert_match(self, match: Match) -> MatcherReturn:
- return MatcherReturn(params=self._convert_groups(match.groupdict()),
- end=match.end())
+ def _convert_match(self, match: Match) -> 'RelativeDeltaParams':
+ return self._convert_groups(match.groupdict())
def _convert_groups(self, groups: Dict[str, str]) -> 'RelativeDeltaParams':
return {key: self.value_type(value) for key, value in groups.items() if value}
class TimeMatcher(RegexMatcher):
- def _convert_match(self, match: Match) -> MatcherReturn:
+ def _convert_match(self, match: Match) -> 'RelativeDeltaParams':
groups = match.groupdict()
try:
meridiem = groups.pop("meridiem").lower()
- except KeyError:
+ except (KeyError, AttributeError):
meridiem = None
params = self._convert_groups(groups)
if meridiem == "pm":
params["hour"] += 12
elif meridiem == "am" and params["hour"] == 12:
params["hour"] = 0
- return MatcherReturn(params=params, end=match.end())
+ return params
class ShortYearMatcher(RegexMatcher):
- def _convert_match(self, match: Match) -> MatcherReturn:
- rtrn = super()._convert_match(match)
- if rtrn.params["year"] < 100:
+ def _convert_match(self, match: Match) -> 'RelativeDeltaParams':
+ params = super()._convert_match(match)
+ if params["year"] < 100:
year = datetime.now().year
current_century = year // 100
- if rtrn.params["year"] < year % 100:
+ if params["year"] < year % 100:
current_century += 1
- rtrn.params["year"] = (current_century * 100) + rtrn.params["year"]
- return rtrn
+ params["year"] = (current_century * 100) + params["year"]
+ return params
class WeekdayMatcher(Matcher):
@@ -121,13 +120,13 @@ class WeekdayMatcher(Matcher):
self.map = map
self.substr = substr
- def match(self, val: str, start: int = 0) -> Optional[MatcherReturn]:
- match = self.regex.match(val, pos=start)
+ def match(self, val: str) -> Optional[MatcherReturn]:
+ match = self.regex.match(val)
if match and match.end() > 0:
weekday = self.map[match.string[:self.substr].lower()]
if isinstance(weekday, int):
weekday = (datetime.now().weekday() + weekday) % 7
- return MatcherReturn(params={"weekday": weekday}, end=match.end())
+ return MatcherReturn(params={"weekday": weekday}, unconsumed=val[match.end():])
return None
@@ -151,26 +150,25 @@ class Locale(Matcher):
return Locale(name=name, timedelta=timedelta or self.timedelta, date=date or self.date,
weekday=weekday or self.weekday, time=time or self.time)
- def match(self, val: str, start: int = 0) -> Optional[MatcherReturn]:
- end = start
- found_delta = self.timedelta.match(val, start=end)
+ def match(self, val: str) -> Optional[MatcherReturn]:
+ found_delta = self.timedelta.match(val)
if found_delta:
- params, end = found_delta
+ params, val = found_delta
else:
params = {}
- found_day = self.weekday.match(val, start=end)
+ found_day = self.weekday.match(val)
if found_day:
- params, end = found_day
+ params, val = found_day
else:
- found_date = self.date.match(val, start=end)
+ found_date = self.date.match(val)
if found_date:
- params, end = found_date
+ params, val = found_date
- found_time = self.time.match(val, start=end)
+ found_time = self.time.match(val)
if found_time:
params = {**params, **found_time.params}
- end = found_time.end
- return MatcherReturn(params, end) if len(params) > 0 else None
+ val = found_time.unconsumed
+ return MatcherReturn(params=params, unconsumed=val) if len(params) > 0 else None
Locales = Dict[str, Locale]
diff --git a/reminder/locales.py b/reminder/locales.py
index 4b5471f..b29a12d 100644
--- a/reminder/locales.py
+++ b/reminder/locales.py
@@ -30,9 +30,10 @@ locales["en_iso"] = Locale(
rf"(?:(?P<days>[-+]?\d+)\s?d(?:ays?)?{td_sep_en})?"
rf"(?:(?P<hours>[-+]?\d+)\s?h(?:(?:r|our)?s?){td_sep_en})?"
rf"(?:(?P<minutes>[-+]?\d+)\s?m(?:in(?:ute)?s?)?{td_sep_en})?"
- r"(?:(?P<seconds>[-+]?\d+)\s?s(?:ec(?:ond)?s?)?)?"),
- date=RegexMatcher(r"(?P<year>\d{4})-(?P<month>\d{2})-(?P<day>\d{2})"),
- weekday=WeekdayMatcher(pattern=r"today"
+ r"(?:(?P<seconds>[-+]?\d+)\s?s(?:ec(?:ond)?s?)?)?"
+ r"(?:\s|$)"),
+ date=RegexMatcher(r"(?P<year>\d{4})-(?P<month>\d{2})-(?P<day>\d{2})\s"),
+ weekday=WeekdayMatcher(pattern=r"(?:today"
r"|tomorrow"
r"|mon(?:day)?"
r"|tues?(?:day)?"
@@ -40,7 +41,8 @@ locales["en_iso"] = Locale(
r"|thu(?:rs(?:day)?)?"
r"|fri(?:day)?"
r"|sat(?:urday)?"
- r"|sun(?:day)?",
+ r"|sun(?:day)?)"
+ r"(?:\s|$)",
map={
"tod": +0, "tom": +1, "mon": MO, "tue": TU, "wed": WE, "thu": TH,
"fri": FR, "sat": SA, "sun": SU,
@@ -48,22 +50,24 @@ locales["en_iso"] = Locale(
time=RegexMatcher(r"\s?(?:at\s)?"
r"(?P<hour>\d{2})"
r"[:.](?P<minute>\d{2})"
- r"(?:[:.](?P<second>\d{2}))?"),
+ r"(?:[:.](?P<second>\d{2}))?"
+ r"(?:\s|$)"),
)
time_12_en = TimeMatcher(r"\s?(?:at\s)?"
r"(?P<hour>\d{2})"
r"(?:[:.](?P<minute>\d{2}))?"
r"(?:[:.](?P<second>\d{2}))?"
- r"(?:\s(?P<meridiem>a\.?m|p\.?m)\.?)?")
+ r"(?:\s(?P<meridiem>a\.?m|p\.?m)\.?)?"
+ r"(?:\s|$)")
locales["en_us"] = locales["en_iso"].replace(
- name="English (US)", time=time_12_en,
- date=ShortYearMatcher(r"(?P<month>\d{1,2})/(?P<day>\d{1,2})(?:/(?P<year>\d{2}(?:\d{2})?))?"))
+ name="English (US)", time=time_12_en, date=ShortYearMatcher(
+ r"(?P<month>\d{1,2})/(?P<day>\d{1,2})(?:/(?P<year>\d{2}(?:\d{2})?))?(?:\s|$)"))
locales["en_uk"] = locales["en_iso"].replace(
- name="English (UK)", time=time_12_en,
- date=ShortYearMatcher(r"(?P<day>\d{1,2})/(?P<month>\d{1,2})(?:/(?P<year>\d{2}(?:\d{2})?))?"))
+ name="English (UK)", time=time_12_en, date=ShortYearMatcher(
+ r"(?P<day>\d{1,2})/(?P<month>\d{1,2})(?:/(?P<year>\d{2}(?:\d{2})?))?(?:\s|$)"))
td_sep_fi = r"(?:[\s,]{1,3}(?:ja\s)?)"
locales["fi_fi"] = Locale(
@@ -75,8 +79,9 @@ locales["fi_fi"] = Locale(
rf"(?:(?P<hours>[-+]?\d+)\s?t(?:un(?:nin?|tia))?{td_sep_fi})?"
rf"(?:(?P<minutes>[-+]?\d+)\s?m(?:in(?:uut(?:in?|tia))?)?{td_sep_fi})?"
r"(?:(?P<seconds>[-+]?\d+)\s?s(?:ek(?:un(?:nin?|tia))?)?)?"
- r"(?:\s(?:kuluttua|päästä?))?"),
- date=ShortYearMatcher(r"(?P<day>\d{1,2})\.(?P<month>\d{1,2})\.(?P<year>\d{2}(?:\d{2})?)"),
+ r"(?:\s(?:kuluttua|päästä?))?"
+ r"(?:\s|$)"),
+ date=ShortYearMatcher(r"(?P<day>\d{1,2})\.(?P<month>\d{1,2})\.(?P<year>\d{2}(?:\d{2})?)\s"),
weekday=WeekdayMatcher(pattern=r"(?:tänään"
r"|(?:yli)?huomen"
r"|ma(?:aanantai)?"
@@ -86,7 +91,8 @@ locales["fi_fi"] = Locale(
r"|pe(?:rjantai)?"
r"|la(?:uantai)?"
r"|su(?:nnuntai)?)"
- r"(?:na)?",
+ r"(?:na)?"
+ r"(?:\s|$)",
map={
"tä": +0, "hu": +1, "yl": +2,
"ma": MO, "ti": TU, "ke": WE, "to": TH, "pe": FR, "la": SA, "su": SU,
@@ -94,7 +100,8 @@ locales["fi_fi"] = Locale(
time=RegexMatcher(r"\s?(?:ke?ll?o\.?\s)?"
r"(?P<hour>\d{2})"
r"[:.](?P<minute>\d{2})"
- r"(?:[:.](?P<second>\d{2}))?"),
+ r"(?:[:.](?P<second>\d{2}))?"
+ r"(?:\s|$)"),
)
td_sep_de = r"(?:[\s,]{1,3}(?:und\s)?)"
@@ -108,7 +115,8 @@ locales["de_de"] = Locale(
rf"(?:(?P<hours>[-+]?\d+)\s?stunden?{td_sep_de})?"
rf"(?:(?P<minutes>[-+]?\d+)\s?minuten?{td_sep_de})?"
r"(?:(?P<seconds>[-+]?\d+)\s?sekunden?)?"),
- date=ShortYearMatcher(r"(?P<day>\d{1,2})\.(?P<month>\d{1,2})\.(?P<year>\d{2}(?:\d{2})?)"),
+ date=ShortYearMatcher(
+ r"(?P<day>\d{1,2})\.(?P<month>\d{1,2})\.(?P<year>\d{2}(?:\d{2})?)(?:\s|$)"),
weekday=WeekdayMatcher(pattern=r"(?:heute"
r"|(?:über)?morgen"
r"|mo(?:ntag)?"
@@ -117,7 +125,8 @@ locales["de_de"] = Locale(
r"|do(?:nnerstag)?"
r"|fr(?:eitag)?"
r"|sa(?:mstag)?"
- r"|so(?:nntag)?)",
+ r"|so(?:nntag)?)"
+ r"(?:\s|$)",
map={
"heu": +0, "mor": +1, "übe": +2, "mon": MO, "die": TU, "mit": WE,
"don": TH, "fre": FR, "sam": SA, "son": SU,
@@ -125,5 +134,6 @@ locales["de_de"] = Locale(
time=RegexMatcher(r"\s?(?:um\s)?"
r"(?P<hour>\d{2})"
r"[:.](?P<minute>\d{2})"
- r"(?:[:.](?P<second>\d{2}))?"),
+ r"(?:[:.](?P<second>\d{2}))?"
+ r"(?:\s|$)"),
)
diff --git a/reminder/util.py b/reminder/util.py
index 3482f82..a29992a 100644
--- a/reminder/util.py
+++ b/reminder/util.py
@@ -37,31 +37,6 @@ class Config(BaseProxyConfig):
helper.copy("base_command")
-timedelta_regex = re.compile(r"(?:(?P<years>[-+]?\d+)\s?y(?:ears?)?\s?)?"
- r"(?:(?P<months>[-+]?\d+)\s?months?\s?)?"
- r"(?:(?P<weeks>[-+]?\d+)\s?w(?:eeks?)?\s?)?"
- r"(?:(?P<days>[-+]?\d+)\s?d(?:ays?)?\s?)?"
- r"(?:(?P<hours>[-+]?\d+)\s?h(?:ours?)?\s?)?"
- r"(?:(?P<minutes>[-+]?\d+)\s?m(?:inutes?)?\s?)?"
- r"(?:(?P<seconds>[-+]?\d+)\s?s(?:econds?)?\s?)?",
- flags=re.IGNORECASE)
-date_regex = re.compile(r"(?P<year>\d{4})-(?P<month>\d{1,2})-(?P<day>\d{1,2})")
-day_regex = re.compile(r"today"
- r"|tomorrow"
- r"|mon(?:day)?"
- r"|tues?(?:day)?"
- r"|wed(?:nesday)?"
- r"|thu(?:rs(?:day)?)?"
- r"|fri(?:day)?"
- r"|sat(?:urday)?"
- r"|sun(?:day)?",
- flags=re.IGNORECASE)
-time_regex = re.compile(r"(?:\sat\s)?(?P<hour>\d{2})"
- r"[:.](?P<minute>\d{2})"
- r"(?:[:.](?P<second>\d{2}))?",
- flags=re.IGNORECASE)
-
-
class DateArgument(Argument):
def __init__(self, name: str, label: str = None, *, required: bool = False):
super().__init__(name, label=label, required=required, pass_raw=True)
@@ -79,7 +54,7 @@ class DateArgument(Argument):
match = locale.match(val)
if match:
date = (datetime.now(tz=tz) + relativedelta(**match.params)).astimezone(pytz.UTC)
- return val[match.end:], date
+ return match.unconsumed, date
return val, None
--
GitLab