Skip to content

Commit 6eaa0be

Browse files
rlskoeserrlskoeser
authored
Merge pull request #115 from dh-tech/feature/reorg-interval
Move interval object and tests into separate files
2 parents ea45598 + 1607b98 commit 6eaa0be

17 files changed

+300
-233
lines changed

‎docs/undate/core.rst

+5-2
Original file line numberDiff line numberDiff line change
@@ -1,13 +1,16 @@
11
Undate objects
22
==============
33

4-
undates and undate intervals
4+
dates, intervals, and calendar
55
------------------------------
66

77
.. autoclass:: undate.undate.Undate
88
:members:
99

10-
.. autoclass:: undate.undate.UndateInterval
10+
.. autoclass:: undate.undate.Calendar
11+
:members:
12+
13+
.. autoclass:: undate.interval.UndateInterval
1114
:members:
1215

1316
date, timedelta, and date precision

‎examples/notebooks/shxco_partial_date_durations.ipynb

+2-2
Original file line numberDiff line numberDiff line change
@@ -316,14 +316,14 @@
316316
},
317317
{
318318
"cell_type": "code",
319-
"execution_count": 4,
319+
"execution_count": 1,
320320
"metadata": {
321321
"id": "y_MqgrQW64uI"
322322
},
323323
"outputs": [],
324324
"source": [
325+
"from undate import UndateInterval\n",
325326
"from undate.date import ONE_DAY\n",
326-
"from undate.undate import UndateInterval\n",
327327
"from undate.converters.iso8601 import ISO8601DateFormat\n",
328328
"\n",
329329
"def undate_duration(start_date, end_date):\n",

‎src/undate/__init__.py

+3-2
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,7 @@
11
__version__ = "0.4.0.dev0"
22

33
from undate.date import DatePrecision
4-
from undate.undate import Undate, UndateInterval
4+
from undate.undate import Undate, Calendar
5+
from undate.interval import UndateInterval
56

6-
__all__ = ["Undate", "UndateInterval", "DatePrecision", "__version__"]
7+
__all__ = ["Undate", "UndateInterval", "Calendar", "DatePrecision", "__version__"]

‎src/undate/converters/calendars/hebrew/converter.py

+1-1
Original file line numberDiff line numberDiff line change
@@ -3,10 +3,10 @@
33
from convertdate import hebrew # type: ignore
44
from lark.exceptions import UnexpectedCharacters
55

6+
from undate import Undate, UndateInterval
67
from undate.converters.base import BaseCalendarConverter
78
from undate.converters.calendars.hebrew.parser import hebrew_parser
89
from undate.converters.calendars.hebrew.transformer import HebrewDateTransformer
9-
from undate.undate import Undate, UndateInterval
1010

1111

1212
class HebrewDateConverter(BaseCalendarConverter):

‎src/undate/converters/calendars/hebrew/transformer.py

+1-1
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
from lark import Transformer, Tree
22

3-
from undate.undate import Undate, Calendar
3+
from undate import Undate, Calendar
44

55

66
class HebrewUndate(Undate):

‎src/undate/converters/calendars/hijri/converter.py

+1-1
Original file line numberDiff line numberDiff line change
@@ -3,10 +3,10 @@
33
from convertdate import islamic # type: ignore
44
from lark.exceptions import UnexpectedCharacters
55

6+
from undate import Undate, UndateInterval
67
from undate.converters.base import BaseCalendarConverter
78
from undate.converters.calendars.hijri.parser import hijri_parser
89
from undate.converters.calendars.hijri.transformer import HijriDateTransformer
9-
from undate.undate import Undate, UndateInterval
1010

1111

1212
class HijriDateConverter(BaseCalendarConverter):

‎src/undate/converters/calendars/hijri/transformer.py

+1-1
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
from lark import Transformer, Tree
22

3-
from undate.undate import Undate, Calendar
3+
from undate import Undate, Calendar
44

55

66
class HijriUndate(Undate):

‎src/undate/converters/edtf/converter.py

+2-1
Original file line numberDiff line numberDiff line change
@@ -2,11 +2,12 @@
22

33
from lark.exceptions import UnexpectedCharacters
44

5+
from undate import Undate, UndateInterval
56
from undate.converters.base import BaseDateConverter
67
from undate.converters.edtf.parser import edtf_parser
78
from undate.converters.edtf.transformer import EDTFTransformer
89
from undate.date import DatePrecision
9-
from undate.undate import Undate, UndateInterval
10+
1011

1112
#: character for unspecified digits
1213
EDTF_UNSPECIFIED_DIGIT: str = "X"

‎src/undate/converters/edtf/transformer.py

+1-1
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
from lark import Token, Transformer, Tree
22

3-
from undate.undate import Undate, UndateInterval
3+
from undate import Undate, UndateInterval
44

55

66
class EDTFTransformer(Transformer):

‎src/undate/converters/iso8601.py

+1-1
Original file line numberDiff line numberDiff line change
@@ -1,7 +1,7 @@
11
from typing import Dict, List, Union
22

3+
from undate import Undate, UndateInterval
34
from undate.converters.base import BaseDateConverter
4-
from undate.undate import Undate, UndateInterval
55

66

77
class ISO8601DateFormat(BaseDateConverter):

‎src/undate/interval.py

+124
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,124 @@
1+
import datetime
2+
3+
# Pre 3.10 requires Union for multiple types, e.g. Union[int, None] instead of int | None
4+
from typing import Optional, Union
5+
6+
7+
from undate import Undate
8+
from undate.date import ONE_DAY, ONE_YEAR, Timedelta
9+
from undate.converters.base import BaseDateConverter
10+
11+
12+
class UndateInterval:
13+
"""A date range between two uncertain dates.
14+
15+
:param earliest: Earliest undate
16+
:type earliest: `undate.Undate`
17+
:param latest: Latest undate
18+
:type latest: `undate.Undate`
19+
:param label: A string to label a specific undate interval, similar to labels of `undate.Undate`.
20+
:type label: `str`
21+
"""
22+
23+
# date range between two undates
24+
earliest: Union[Undate, None]
25+
latest: Union[Undate, None]
26+
label: Union[str, None]
27+
28+
# TODO: let's think about adding an optional precision / length /size field
29+
# using DatePrecision
30+
31+
def __init__(
32+
self,
33+
earliest: Optional[Undate] = None,
34+
latest: Optional[Undate] = None,
35+
label: Optional[str] = None,
36+
):
37+
# for now, assume takes two undate objects;
38+
# support conversion from datetime
39+
if earliest and not isinstance(earliest, Undate):
40+
# NOTE: some overlap with Undate._comparison_type method
41+
# maybe support conversion from other formats later
42+
if isinstance(earliest, datetime.date):
43+
earliest = Undate.from_datetime_date(earliest)
44+
else:
45+
raise ValueError(
46+
f"earliest date {earliest} cannot be converted to Undate"
47+
)
48+
if latest and not isinstance(latest, Undate):
49+
if isinstance(latest, datetime.date):
50+
latest = Undate.from_datetime_date(latest)
51+
else:
52+
raise ValueError(f"latest date {latest} cannot be converted to Undate")
53+
54+
# check that the interval is valid
55+
if latest and earliest and latest <= earliest:
56+
raise ValueError(f"invalid interval {earliest}-{latest}")
57+
58+
self.earliest = earliest
59+
self.latest = latest
60+
self.label = label
61+
62+
def __str__(self) -> str:
63+
# using EDTF syntax for open ranges
64+
return "%s/%s" % (self.earliest or "..", self.latest or "")
65+
66+
def format(self, format) -> str:
67+
"""format this undate interval as a string using the specified format;
68+
for now, only supports named converters"""
69+
converter_cls = BaseDateConverter.available_converters().get(format, None)
70+
if converter_cls:
71+
return converter_cls().to_string(self)
72+
73+
raise ValueError(f"Unsupported format '{format}'")
74+
75+
def __repr__(self) -> str:
76+
if self.label:
77+
return "<UndateInterval '%s' (%s)>" % (self.label, self)
78+
return "<UndateInterval %s>" % self
79+
80+
def __eq__(self, other) -> bool:
81+
# consider interval equal if both dates are equal
82+
return self.earliest == other.earliest and self.latest == other.latest
83+
84+
def duration(self) -> Timedelta:
85+
"""Calculate the duration between two undates.
86+
Note that durations are inclusive (i.e., a closed interval), and
87+
include both the earliest and latest date rather than the difference
88+
between them.
89+
90+
:returns: A duration
91+
:rtype: Timedelta
92+
"""
93+
# what is the duration of this date range?
94+
95+
# if range is open-ended, can't calculate
96+
if self.earliest is None or self.latest is None:
97+
return NotImplemented
98+
99+
# if both years are known, subtract end of range from beginning of start
100+
if self.latest.known_year and self.earliest.known_year:
101+
return self.latest.latest - self.earliest.earliest + ONE_DAY
102+
103+
# if neither year is known...
104+
elif not self.latest.known_year and not self.earliest.known_year:
105+
# under what circumstances can we assume that if both years
106+
# are unknown the dates are in the same year or sequential?
107+
duration = self.latest.earliest - self.earliest.earliest
108+
# if we get a negative, we've wrapped from end of one year
109+
# to the beginning of the next;
110+
# recalculate assuming second date is in the subsequent year
111+
if duration.days < 0:
112+
end = self.latest.earliest + ONE_YEAR
113+
duration = end - self.earliest.earliest
114+
115+
# add the additional day *after* checking for a negative
116+
# or after recalculating with adjusted year
117+
duration += ONE_DAY
118+
119+
return duration
120+
121+
else:
122+
# is there any meaningful way to calculate duration
123+
# if one year is known and the other is not?
124+
raise NotImplementedError

‎src/undate/undate.py

+8-99
Original file line numberDiff line numberDiff line change
@@ -1,8 +1,12 @@
1-
import datetime
2-
import re
1+
from __future__ import annotations
32

3+
import datetime
44
from enum import auto
5+
import re
6+
from typing import TYPE_CHECKING
57

8+
if TYPE_CHECKING:
9+
from undate.interval import UndateInterval
610
try:
711
# StrEnum was only added in python 3.11
812
from enum import StrEnum
@@ -14,7 +18,7 @@
1418
from typing import Dict, Optional, Union
1519

1620
from undate.converters.base import BaseDateConverter
17-
from undate.date import ONE_DAY, ONE_MONTH_MAX, ONE_YEAR, Date, DatePrecision, Timedelta
21+
from undate.date import ONE_DAY, ONE_MONTH_MAX, Date, DatePrecision, Timedelta
1822

1923

2024
class Calendar(StrEnum):
@@ -218,7 +222,7 @@ def __repr__(self) -> str:
218222
return f"<Undate{label_str} {self} ({self.calendar.name.title()})>"
219223

220224
@classmethod
221-
def parse(cls, date_string, format) -> Union["Undate", "UndateInterval"]:
225+
def parse(cls, date_string, format) -> Union["Undate", UndateInterval]:
222226
"""parse a string to an undate or undate interval using the specified format;
223227
for now, only supports named converters"""
224228
converter_cls = BaseDateConverter.available_converters().get(format, None)
@@ -487,98 +491,3 @@ def _missing_digit_minmax(
487491
min_val = int("".join(new_min_val))
488492
max_val = int("".join(new_max_val))
489493
return (min_val, max_val)
490-
491-
492-
class UndateInterval:
493-
"""A date range between two uncertain dates.
494-
495-
:param earliest: Earliest undate
496-
:type earliest: `undate.Undate`
497-
:param latest: Latest undate
498-
:type latest: `undate.Undate`
499-
:param label: A string to label a specific undate interval, similar to labels of `undate.Undate`.
500-
:type label: `str`
501-
"""
502-
503-
# date range between two undates
504-
earliest: Union[Undate, None]
505-
latest: Union[Undate, None]
506-
label: Union[str, None]
507-
508-
# TODO: let's think about adding an optional precision / length /size field
509-
# using DatePrecision
510-
511-
def __init__(
512-
self,
513-
earliest: Optional[Undate] = None,
514-
latest: Optional[Undate] = None,
515-
label: Optional[str] = None,
516-
):
517-
# for now, assume takes two undate objects
518-
self.earliest = earliest
519-
self.latest = latest
520-
self.label = label
521-
522-
def __str__(self) -> str:
523-
# using EDTF syntax for open ranges
524-
return "%s/%s" % (self.earliest or "..", self.latest or "")
525-
526-
def format(self, format) -> str:
527-
"""format this undate interval as a string using the specified format;
528-
for now, only supports named converters"""
529-
converter_cls = BaseDateConverter.available_converters().get(format, None)
530-
if converter_cls:
531-
return converter_cls().to_string(self)
532-
533-
raise ValueError(f"Unsupported format '{format}'")
534-
535-
def __repr__(self) -> str:
536-
if self.label:
537-
return "<UndateInterval '%s' (%s)>" % (self.label, self)
538-
return "<UndateInterval %s>" % self
539-
540-
def __eq__(self, other) -> bool:
541-
# consider interval equal if both dates are equal
542-
return self.earliest == other.earliest and self.latest == other.latest
543-
544-
def duration(self) -> Timedelta:
545-
"""Calculate the duration between two undates.
546-
Note that durations are inclusive (i.e., a closed interval), and
547-
include both the earliest and latest date rather than the difference
548-
between them.
549-
550-
:returns: A duration
551-
:rtype: Timedelta
552-
"""
553-
# what is the duration of this date range?
554-
555-
# if range is open-ended, can't calculate
556-
if self.earliest is None or self.latest is None:
557-
return NotImplemented
558-
559-
# if both years are known, subtract end of range from beginning of start
560-
if self.latest.known_year and self.earliest.known_year:
561-
return self.latest.latest - self.earliest.earliest + ONE_DAY
562-
563-
# if neither year is known...
564-
elif not self.latest.known_year and not self.earliest.known_year:
565-
# under what circumstances can we assume that if both years
566-
# are unknown the dates are in the same year or sequential?
567-
duration = self.latest.earliest - self.earliest.earliest
568-
# if we get a negative, we've wrapped from end of one year
569-
# to the beginning of the next;
570-
# recalculate assuming second date is in the subsequent year
571-
if duration.days < 0:
572-
end = self.latest.earliest + ONE_YEAR
573-
duration = end - self.earliest.earliest
574-
575-
# add the additional day *after* checking for a negative
576-
# or after recalculating with adjusted year
577-
duration += ONE_DAY
578-
579-
return duration
580-
581-
else:
582-
# is there any meaningful way to calculate duration
583-
# if one year is known and the other is not?
584-
raise NotImplementedError

‎tests/test_converters/edtf/test_edtf_transformer.py

+2-1
Original file line numberDiff line numberDiff line change
@@ -1,7 +1,8 @@
11
import pytest
2+
3+
from undate import Undate, UndateInterval
24
from undate.converters.edtf.parser import edtf_parser
35
from undate.converters.edtf.transformer import EDTFTransformer
4-
from undate.undate import Undate, UndateInterval
56

67
# for now, just test that valid dates can be parsed
78

‎tests/test_converters/test_edtf.py

+1-1
Original file line numberDiff line numberDiff line change
@@ -1,7 +1,7 @@
11
import pytest
22
from undate.converters.edtf import EDTFDateConverter
33
from undate.date import DatePrecision
4-
from undate.undate import Undate, UndateInterval
4+
from undate import Undate, UndateInterval
55

66

77
class TestEDTFDateConverter:

‎tests/test_converters/test_iso8601.py

+1-1
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
1+
from undate import Undate, UndateInterval
12
from undate.converters.iso8601 import ISO8601DateFormat
2-
from undate.undate import Undate, UndateInterval
33

44

55
class TestISO8601DateFormat:

0 commit comments

Comments
 (0)