|
1
|
|
|
""" |
|
2
|
|
|
ETLT |
|
3
|
|
|
|
|
4
|
|
|
Copyright 2016 Set Based IT Consultancy |
|
5
|
|
|
|
|
6
|
|
|
Licence MIT |
|
7
|
|
|
""" |
|
8
|
1 |
|
import copy |
|
9
|
|
|
|
|
10
|
1 |
|
from etlt.helper.Allen import Allen |
|
11
|
1 |
|
from etlt.helper.Type2Helper import Type2Helper |
|
12
|
|
|
|
|
13
|
|
|
|
|
14
|
1 |
|
class Type2CondenseHelper(Type2Helper): |
|
15
|
|
|
""" |
|
16
|
|
|
A helper class for deriving the distinct intervals in reference data with date intervals. |
|
17
|
|
|
""" |
|
18
|
|
|
|
|
19
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
20
|
1 |
|
@staticmethod |
|
21
|
|
|
def _distinct(row1, row2): |
|
22
|
|
|
""" |
|
23
|
|
|
Returns a list of distinct (or none overlapping) intervals if two intervals are not distinct. Returns None if |
|
24
|
|
|
the two intervals are distinct. The list can have 2 or 3 intervals. |
|
25
|
|
|
|
|
26
|
|
|
:param tuple[int,int] row1: The first interval. |
|
27
|
|
|
:param tuple[int,int] row2: The second interval. |
|
28
|
|
|
|
|
29
|
|
|
:rtype: None|tuple[int,int] |
|
30
|
|
|
""" |
|
31
|
1 |
|
relation = Allen.relation(row1[0], row1[1], row2[0], row2[1]) |
|
32
|
1 |
|
if relation == Allen.X_BEFORE_Y: |
|
33
|
|
|
# row1: |----| |
|
34
|
|
|
# row2: |-----| |
|
35
|
|
|
return None # [(row1[0], row1[1]), (row2[0], row2[1])] |
|
36
|
|
|
|
|
37
|
1 |
|
if relation == Allen.X_BEFORE_Y_INVERSE: |
|
38
|
|
|
# row1: |-----| |
|
39
|
|
|
# row2: |----| |
|
40
|
1 |
|
return None # [(row2[0], row2[1]), (row1[0], row1[1])] |
|
41
|
|
|
|
|
42
|
1 |
|
if relation == Allen.X_MEETS_Y: |
|
43
|
|
|
# row1: |-------| |
|
44
|
|
|
# row2: |-------| |
|
45
|
1 |
|
return None # [(row1[0], row1[1]), (row2[0], row2[1])] |
|
46
|
|
|
|
|
47
|
1 |
|
if relation == Allen.X_MEETS_Y_INVERSE: |
|
48
|
|
|
# row1: |-------| |
|
49
|
|
|
# row2: |-------| |
|
50
|
1 |
|
return None # [(row2[0], row2[1]), (row1[0], row1[1])] |
|
51
|
|
|
|
|
52
|
1 |
|
if relation == Allen.X_OVERLAPS_WITH_Y: |
|
53
|
|
|
# row1: |-----------| |
|
54
|
|
|
# row2: |----------| |
|
55
|
1 |
|
return [(row1[0], row2[0] - 1), (row2[0], row1[1]), (row1[1] + 1, row2[1])] |
|
56
|
|
|
|
|
57
|
1 |
|
if relation == Allen.X_OVERLAPS_WITH_Y_INVERSE: |
|
58
|
|
|
# row1: |----------| |
|
59
|
|
|
# row2: |-----------| |
|
60
|
1 |
|
return [(row2[0], row1[0] - 1), (row1[0], row2[1]), (row2[1] + 1, row1[1])] |
|
61
|
|
|
|
|
62
|
1 |
|
if relation == Allen.X_STARTS_Y: |
|
63
|
|
|
# row1: |------| |
|
64
|
|
|
# row2: |----------------| |
|
65
|
|
|
return [(row1[0], row1[1]), (row1[1] + 1, row2[1])] |
|
66
|
|
|
|
|
67
|
1 |
|
if relation == Allen.X_STARTS_Y_INVERSE: |
|
68
|
|
|
# row1: |----------------| |
|
69
|
|
|
# row2: |------| |
|
70
|
1 |
|
return [(row2[0], row2[1]), (row2[1] + 1, row1[1])] |
|
71
|
|
|
|
|
72
|
1 |
|
if relation == Allen.X_DURING_Y: |
|
73
|
|
|
# row1: |------| |
|
74
|
|
|
# row2: |----------------| |
|
75
|
1 |
|
return [(row2[0], row1[0] - 1), (row1[0], row1[1]), (row1[1] + 1, row2[1])] |
|
76
|
|
|
|
|
77
|
1 |
|
if relation == Allen.X_DURING_Y_INVERSE: |
|
78
|
|
|
# row1: |----------------| |
|
79
|
|
|
# row2: |------| |
|
80
|
1 |
|
return [(row1[0], row2[0] - 1), (row2[0], row2[1]), (row2[1] + 1, row1[1])] |
|
81
|
|
|
|
|
82
|
1 |
|
if relation == Allen.X_FINISHES_Y: |
|
83
|
|
|
# row1: |------| |
|
84
|
|
|
# row2: |----------------| |
|
85
|
1 |
|
return [(row2[0], row1[0] - 1), (row1[0], row1[1])] |
|
86
|
|
|
|
|
87
|
1 |
|
if relation == Allen.X_FINISHES_Y_INVERSE: |
|
88
|
|
|
# row1: |----------------| |
|
89
|
|
|
# row2: |------| |
|
90
|
1 |
|
return [(row1[0], row2[0] - 1), (row2[0], row2[1])] |
|
91
|
|
|
|
|
92
|
1 |
|
if relation == Allen.X_EQUAL_Y: |
|
93
|
|
|
# row1: |----------------| |
|
94
|
|
|
# row2: |----------------| |
|
95
|
1 |
|
return None # [(row1[0], row1[1])] |
|
96
|
|
|
|
|
97
|
|
|
# We got all 13 relation in Allen's interval algebra covered. |
|
98
|
|
|
raise ValueError('Unexpected relation {0:d}'.format(relation)) |
|
99
|
|
|
|
|
100
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
101
|
1 |
|
@staticmethod |
|
102
|
|
|
def _add_interval(all_intervals, new_interval): |
|
103
|
|
|
""" |
|
104
|
|
|
Adds a new interval to a set of distinct intervals. |
|
105
|
|
|
|
|
106
|
|
|
:param set[(int,int)] all_intervals: The set of distinct intervals. |
|
107
|
|
|
:param (int,int) new_interval: The new interval. |
|
108
|
|
|
""" |
|
109
|
1 |
|
intervals = None |
|
110
|
1 |
|
old_interval = None |
|
111
|
1 |
|
for old_interval in all_intervals: |
|
112
|
1 |
|
intervals = Type2CondenseHelper._distinct(new_interval, old_interval) |
|
113
|
1 |
|
if intervals: |
|
114
|
1 |
|
break |
|
115
|
|
|
|
|
116
|
1 |
|
if intervals: |
|
117
|
1 |
|
if old_interval: |
|
118
|
1 |
|
all_intervals.remove(old_interval) |
|
119
|
1 |
|
for distinct_interval in intervals: |
|
120
|
1 |
|
Type2CondenseHelper._add_interval(all_intervals, distinct_interval) |
|
121
|
|
|
else: |
|
122
|
1 |
|
all_intervals.add(new_interval) |
|
123
|
|
|
|
|
124
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
125
|
1 |
|
def _derive_distinct_intervals(self, rows): |
|
126
|
|
|
""" |
|
127
|
|
|
Returns the set of distinct intervals in a row set. |
|
128
|
|
|
|
|
129
|
|
|
:param list[dict[str,T]] rows: The rows set. |
|
130
|
|
|
|
|
131
|
|
|
:rtype: set[(int,int)] |
|
132
|
|
|
""" |
|
133
|
1 |
|
ret = set() |
|
134
|
1 |
|
for row in rows: |
|
135
|
1 |
|
self._add_interval(ret, (row[self._key_start_date], row[self._key_end_date])) |
|
136
|
|
|
|
|
137
|
1 |
|
return ret |
|
138
|
|
|
|
|
139
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
140
|
1 |
|
def condense(self): |
|
141
|
|
|
""" |
|
142
|
|
|
Condense the data set to the distinct intervals based on the natural key. |
|
143
|
|
|
""" |
|
144
|
1 |
|
for natural_key, rows in self.rows.items(): |
|
145
|
1 |
|
tmp1 = [] |
|
146
|
1 |
|
intervals = sorted(self._derive_distinct_intervals(rows)) |
|
147
|
1 |
|
for interval in intervals: |
|
148
|
1 |
|
tmp2 = copy.copy(rows[0]) |
|
149
|
1 |
|
tmp2[self._key_start_date] = interval[0] |
|
150
|
1 |
|
tmp2[self._key_end_date] = interval[1] |
|
151
|
1 |
|
tmp1.append(tmp2) |
|
152
|
|
|
|
|
153
|
1 |
|
self.rows[natural_key] = tmp1 |
|
154
|
|
|
|
|
155
|
|
|
# ---------------------------------------------------------------------------------------------------------------------- |
|
156
|
|
|
|
PyETLT /
etlt
