Merge pull request #205 from aridevelopment-de/#198/dev

Added parsing and evaluating of `first monday of august`

Co-authored-by: @Inf-inity

datetimeparser/evaluator/evaluatorutils.py

@@ -112,6 +112,10 @@ def sanitize_input(
                             if current_time > test_out and not given_year:
                                 parsed_list = EvaluatorUtils.x_week_of_month(relative_dt, idx, pars2, year + 1)
 
+                else:
+                    if isinstance(parsed_list[idx + 1], AbsoluteDateTime):
+                        given_year = parsed_list[idx + 1].year
+
         return list(filter(lambda e: e not in Keywords.ALL and not isinstance(e, str), parsed_list)), given_year
 
     @staticmethod
@@ -124,8 +128,7 @@ def cut_time(time: datetime) -> datetime:
 
         return datetime(time.year, time.month, time.day, 0, 0, 0)
 
-    @staticmethod
-    def get_base(sanitized_input: list, year: int, current_time: datetime, forced: bool = False) -> datetime:
+    def get_base(self, sanitized_input: list, year: int, current_time: datetime, forced: bool = False) -> datetime:
         """
         Takes the last elements from the list and tries to generate a basis for further processing from them
         The base consists of at least one constant, to which values are then assigned
@@ -145,7 +148,37 @@ def get_base(sanitized_input: list, year: int, current_time: datetime, forced: b
                     dt: datetime = sanitized_input[-2].time_value(sanitized_input[-1].year)
                     day: int = sanitized_input[-3]
                     return datetime(dt.year, dt.month, day, dt.hour, dt.minute, dt.second)
-                return sanitized_input[-2].time_value(sanitized_input[-1].year)
+
+                if sanitized_input[-2].time_value:
+                    dt = sanitized_input[-2].time_value(sanitized_input[-1].year)
+
+                    if isinstance(sanitized_input[-3], Constant) and sanitized_input[-3].value:
+                        dt += relativedelta(days=sanitized_input[-3].value - 1)
+                    elif isinstance(sanitized_input[-3], Constant) and not sanitized_input[-3].value:
+                        val = sanitized_input[-4].value
+
+                        if sanitized_input[-3].name == "days":
+                            return datetime(dt.year, dt.month, val, dt.hour, dt.minute, dt.second)
+                        if sanitized_input[-3].name == "weeks":
+                            dt = self.get_week_of(dt)
+                            return dt + relativedelta(weeks=val-1)
+                        if sanitized_input[-3].name == "months":
+                            return datetime(dt.year, val, dt.day, dt.hour, dt.minute, dt.second)
+
+                        days_dict = {x.name: x.time_value(dt) for x in WeekdayConstants.ALL}
+                        if sanitized_input[-3].name in days_dict:
+                            dt = datetime.strptime(days_dict.get(sanitized_input[-3].name), "%Y-%m-%d %H:%M:%S")
+                            dt += relativedelta(weeks=val - 1)
+
+                    return dt
+
+                elif sanitized_input[-3].value:
+                    val: int = sanitized_input[-3].value
+
+                    if sanitized_input[-2].name == "days":
+                        return datetime(sanitized_input[-1].year, 1, val, 0, 0, 0)
+                    if sanitized_input[-2].name == "months":
+                        return datetime(sanitized_input[-1].year, val, 1, 0, 0, 0)
 
             # If a year is given but no months/days, they will be set to '1' because datetime can't handle month/day-values with '0'
             if sanitized_input[-1].year != 0:
@@ -169,15 +202,28 @@ def get_base(sanitized_input: list, year: int, current_time: datetime, forced: b
 
         # If no AbsoluteDatetime is given, the default year will be used instead
         elif isinstance(sanitized_input[-1], Constant):
+            dt: datetime = sanitized_input[-1].time_value(year)
             if isinstance(sanitized_input[-2], int):
-                dt: datetime = sanitized_input[-1].time_value(year)
                 day: int = sanitized_input[-2]
                 out = datetime(dt.year, dt.month, day, dt.hour, dt.minute, dt.second)
 
                 if out > current_time or forced:
                     return out
                 out += relativedelta(years=1)
                 return out
+            else:
+                if len(sanitized_input) == 3:
+                    val: int = sanitized_input[-3].value
+
+                    if sanitized_input[-2].name == "days":
+                        return datetime(dt.year, dt.month, dt.day + val, dt.hour, dt.minute, dt.second)
+                    if sanitized_input[-2].name == "weeks":
+                        dt = self.get_week_of(dt)
+                        return dt + relativedelta(weeks=val)
+                    if sanitized_input[-2].name == "months":
+                        return datetime(dt.year, dt.month + val, dt.day, dt.hour, dt.minute, dt.second)
+                if not isinstance(sanitized_input[-2], RelativeDateTime):
+                    return datetime(dt.year, dt.month, sanitized_input[-2].value, dt.hour, dt.minute, dt.second)
 
             # Checks if an event already happened this year (f.e. eastern). If so, the next year will be used
             if sanitized_input[-1].time_value(year) > current_time or forced:

datetimeparser/parser/parsermethods.py

@@ -676,6 +676,7 @@ class AbsolutePrepositionParser:
     )
 
     RELATIVE_DATETIME_CONSTANTS = (*NumberCountConstants.ALL, *NumberConstants.ALL, *DatetimeConstants.ALL)
+    ABSOLUTE_RELATIVE_DATETIME_CONSTANTS = (*WeekdayConstants.ALL, *MonthConstants.ALL)
     RELATIVE_TIME_CONSTANTS = DatetimeConstants.ALL
 
     RELATIVE_DATA_SKIPPABLE_WORDS = (
@@ -755,26 +756,40 @@ def _parse_relative_statement(self, relative_statement: str, preposition: str) -
                 returned_data.append(int(argument))
                 continue
 
+            found_keyword = False
+
             # '1st', '1.', 'first', ...
             # 'one', 'two', 'three', ...
             # 'seconds', 'minutes', 'hours', ...
             for keyword in self.RELATIVE_DATETIME_CONSTANTS:
                 if argument in keyword.get_all():
                     if keyword in DatetimeConstants.ALL:
-                        if not returned_data or not isinstance(returned_data[-1], int):
+                        if not returned_data or (not isinstance(returned_data[-1], int) and not returned_data[-1] in NumberCountConstants.ALL):
                             # For cases like 'day and month (before christmas)'
                             returned_data.append(1)
 
                         returned_data.append(keyword)
+                    elif keyword in NumberCountConstants.ALL:
+                        if not returned_data:
+                            # The keyword comes before the actual datetime constant 'second (monday of August)'
+                            returned_data.append(keyword)
                     else:
                         returned_data.append(keyword.value)
 
+                    found_keyword = True
                     break
-                else:
-                    continue
 
-                break
-            else:
+            # 'monday', 'tuesday', 'wednesday', ...
+            # 'january', 'february', 'march', ...
+            # GitHub issue #198
+            for keyword in self.ABSOLUTE_RELATIVE_DATETIME_CONSTANTS:
+                if argument in keyword.get_all():
+                    returned_data.append(keyword)
+
+                    found_keyword = True
+                    break
+
+            if not found_keyword:
                 return None
 
         return returned_data
@@ -824,6 +839,9 @@ def _concatenate_relative_data(
                     else:
                         # Otherwise, we cannot concatenate both parts of the data, so we just append the current one
                         returned_data.append(current_data)
+            elif value in NumberCountConstants.ALL and unit in (*DatetimeConstants.ALL, *WeekdayConstants.ALL, *MonthConstants.ALL):
+                returned_data.append(value)
+                returned_data.append(unit)
 
         return returned_data
 

tests/testcases.py

@@ -106,6 +106,8 @@ def is_leap_year(year: int) -> bool:
         # GitHub issue #176
         "10 days after pi-day": Expected(time_sensitive=True, month=3, day=14, delta=relativedelta(days=10)),
         "10 days before tau day": Expected(time_sensitive=True, month=6, day=28, delta=relativedelta(days=-10)),
+        # GitHub issue #198
+        "second monday of august 2023": Expected(year=2023, month=8, day=14),
     },
     # Relative Datetimes
     "relative_datetimes": {