|
1
|
1 |
|
from typing import Any, Dict, List, Optional, Tuple |
|
2
|
|
|
|
|
3
|
|
|
from etlt.helper.Type2Helper import Type2Helper |
|
4
|
1 |
|
|
|
5
|
|
|
|
|
6
|
|
|
class Type2JoinHelper(Type2Helper): |
|
7
|
|
|
""" |
|
8
|
|
|
A helper class for joining data sets with date intervals. |
|
9
|
|
|
""" |
|
10
|
1 |
|
|
|
11
|
1 |
|
# ------------------------------------------------------------------------------------------------------------------ |
|
12
|
|
|
@staticmethod |
|
13
|
|
|
def _intersect(start1: int, end1: int, start2: int, end2: int) -> Tuple[Optional[int], Optional[int]]: |
|
14
|
|
|
""" |
|
15
|
|
|
Returns the intersection of two intervals. Returns (None,None) if the intersection is empty. |
|
16
|
|
|
|
|
17
|
|
|
:param start1: The start date of the first interval. |
|
18
|
|
|
:param end1: The end date of the first interval. |
|
19
|
|
|
:param start2: The start date of the second interval. |
|
20
|
|
|
:param end2: The end date of the second interval. |
|
21
|
|
|
""" |
|
22
|
1 |
|
start = max(start1, start2) |
|
23
|
1 |
|
end = min(end1, end2) |
|
24
|
|
|
|
|
25
|
1 |
|
if start > end: |
|
26
|
1 |
|
return None, None |
|
27
|
|
|
|
|
28
|
1 |
|
return start, end |
|
29
|
|
|
|
|
30
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
31
|
1 |
|
def _additional_rows_date2int(self, keys: List[Tuple[str, str]], rows: List[Dict[str, Any]]) -> None: |
|
32
|
|
|
""" |
|
33
|
|
|
Replaces start and end dates of the additional date intervals in the row set with their integer representation |
|
34
|
|
|
|
|
35
|
|
|
:param keys: The other keys with start and end date. |
|
36
|
|
|
:param rows: The list of rows. |
|
37
|
|
|
""" |
|
38
|
|
|
for row in rows: |
|
39
|
|
|
for key_start_date, key_end_date in keys: |
|
40
|
1 |
|
if key_start_date not in [self._key_start_date, self._key_end_date]: |
|
41
|
1 |
|
row[key_start_date] = self._date2int(row[key_start_date]) |
|
42
|
1 |
|
if key_end_date not in [self._key_start_date, self._key_end_date]: |
|
43
|
1 |
|
row[key_end_date] = self._date2int(row[key_end_date]) |
|
44
|
1 |
|
|
|
45
|
1 |
|
# ------------------------------------------------------------------------------------------------------------------ |
|
46
|
|
|
def _intersection(self, keys: List[Tuple[str, str]], rows: List[Dict[str, Any]]) -> List[Dict[str, Any]]: |
|
47
|
|
|
""" |
|
48
|
1 |
|
Computes the intersection of the date intervals of two or more reference data sets. If the intersection is empty |
|
49
|
|
|
the row is removed from the group. |
|
50
|
|
|
|
|
51
|
|
|
:param keys: The other keys with start and end date. |
|
52
|
|
|
:param rows: The list of rows. |
|
53
|
|
|
""" |
|
54
|
|
|
# If there are no other keys with start and end date (i.e. nothing to merge) return immediately. |
|
55
|
|
|
if not keys: |
|
56
|
|
|
return rows |
|
57
|
|
|
|
|
58
|
|
|
ret = list() |
|
59
|
1 |
|
for row in rows: |
|
60
|
1 |
|
start_date = row[self._key_start_date] |
|
61
|
|
|
end_date = row[self._key_end_date] |
|
62
|
1 |
|
for key_start_date, key_end_date in keys: |
|
63
|
1 |
|
start_date, end_date = Type2JoinHelper._intersect(start_date, |
|
64
|
1 |
|
end_date, |
|
65
|
1 |
|
row[key_start_date], |
|
66
|
1 |
|
row[key_end_date]) |
|
67
|
1 |
|
if not start_date: |
|
68
|
|
|
break |
|
69
|
|
|
if key_start_date not in [self._key_start_date, self._key_end_date]: |
|
70
|
|
|
del row[key_start_date] |
|
71
|
1 |
|
if key_end_date not in [self._key_start_date, self._key_end_date]: |
|
72
|
1 |
|
del row[key_end_date] |
|
73
|
1 |
|
|
|
74
|
1 |
|
if start_date: |
|
75
|
1 |
|
row[self._key_start_date] = start_date |
|
76
|
1 |
|
row[self._key_end_date] = end_date |
|
77
|
|
|
ret.append(row) |
|
78
|
1 |
|
|
|
79
|
1 |
|
return ret |
|
80
|
1 |
|
|
|
81
|
1 |
|
# ------------------------------------------------------------------------------------------------------------------ |
|
82
|
|
|
def merge(self, keys: List[Tuple[str, str]]) -> None: |
|
83
|
1 |
|
""" |
|
84
|
|
|
Merges the join on pseudo keys of two or more reference data sets. |
|
85
|
|
|
|
|
86
|
1 |
|
:param keys: For each data set the keys of the start and end date. |
|
87
|
|
|
""" |
|
88
|
|
|
deletes = [] |
|
89
|
|
|
for pseudo_key, rows in self._rows.items(): |
|
90
|
|
|
self._additional_rows_date2int(keys, rows) |
|
91
|
|
|
rows = self._intersection(keys, rows) |
|
92
|
1 |
|
if rows: |
|
93
|
1 |
|
rows = self._rows_sort(rows) |
|
94
|
1 |
|
self._rows[pseudo_key] = self._merge_adjacent_rows(rows) |
|
95
|
1 |
|
else: |
|
96
|
1 |
|
deletes.append(pseudo_key) |
|
97
|
1 |
|
|
|
98
|
1 |
|
for pseudo_key in deletes: |
|
99
|
|
|
del self._rows[pseudo_key] |
|
100
|
1 |
|
|
|
101
|
|
|
# ---------------------------------------------------------------------------------------------------------------------- |
|
102
|
|
|
|
PyETLT /
etlt
