1
|
|
|
""" |
2
|
|
|
ETLT |
3
|
|
|
|
4
|
|
|
Copyright 2016 Set Based IT Consultancy |
5
|
|
|
|
6
|
|
|
Licence MIT |
7
|
|
|
""" |
8
|
1 |
|
import copy |
9
|
1 |
|
import datetime |
10
|
|
|
|
11
|
|
|
|
12
|
1 |
|
class Type2Helper: |
13
|
|
|
""" |
14
|
|
|
A helper class for reference data with date intervals. |
15
|
|
|
""" |
16
|
|
|
|
17
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
18
|
1 |
|
def __init__(self, key_start_date, key_end_date, natural_key): |
19
|
|
|
""" |
20
|
|
|
Object constructor. |
21
|
|
|
|
22
|
|
|
:param str key_start_date: The key of the start date in the rows. |
23
|
|
|
:param str key_end_date: The key of the end date in the rows. |
24
|
|
|
:param list[str] natural_key: The keys of the columns that form the natural key. |
25
|
|
|
""" |
26
|
1 |
|
self._natural_key = list(natural_key) |
27
|
|
|
""" |
28
|
|
|
The keys of the columns that form the natural key. |
29
|
|
|
|
30
|
|
|
:type: list[str] |
31
|
|
|
""" |
32
|
|
|
|
33
|
1 |
|
self._key_end_date = key_end_date |
34
|
|
|
""" |
35
|
|
|
The key of the end date in the rows. |
36
|
|
|
|
37
|
|
|
:type: str |
38
|
|
|
""" |
39
|
1 |
|
self._key_start_date = key_start_date |
40
|
|
|
""" |
41
|
|
|
The key of the start date in the rows. |
42
|
|
|
|
43
|
|
|
:type: str |
44
|
|
|
""" |
45
|
|
|
|
46
|
1 |
|
self.rows = dict() |
47
|
|
|
""" |
48
|
|
|
The data set. |
49
|
|
|
|
50
|
|
|
:type: dict |
51
|
|
|
""" |
52
|
|
|
|
53
|
1 |
|
self._copy = True |
54
|
|
|
""" |
55
|
|
|
If set to true a copy will be made from the original rows such that the original rows is not modified. |
56
|
|
|
|
57
|
|
|
:type: bool |
58
|
|
|
""" |
59
|
|
|
|
60
|
1 |
|
self._date_type = '' |
61
|
1 |
|
""" |
62
|
|
|
The type of the date fields. |
63
|
|
|
- date for datetime.date objects |
64
|
|
|
- str for strings in ISO 8601 (YYYY-MM-DD) format |
65
|
|
|
- int for integers |
66
|
|
|
|
67
|
|
|
:type: str |
68
|
|
|
""" |
69
|
|
|
|
70
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
71
|
1 |
|
def _get_natural_key(self, row): |
72
|
|
|
""" |
73
|
|
|
Returns the natural key in a row. |
74
|
|
|
|
75
|
|
|
:param dict row: The row. |
76
|
|
|
|
77
|
|
|
:rtype: tuple |
78
|
|
|
""" |
79
|
1 |
|
ret = list() |
80
|
1 |
|
for key in self._natural_key: |
81
|
1 |
|
ret.append(row[key]) |
82
|
|
|
|
83
|
1 |
|
return tuple(ret) |
84
|
|
|
|
85
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
86
|
1 |
|
@staticmethod |
87
|
|
|
def _date2int(date): |
88
|
|
|
""" |
89
|
|
|
Returns an integer representation of a date. |
90
|
|
|
|
91
|
|
|
:param str|datetime.date date: The date. |
92
|
|
|
|
93
|
|
|
:rtype: int |
94
|
|
|
""" |
95
|
1 |
|
if isinstance(date, str): |
96
|
1 |
|
if date.endswith(' 00:00:00') or date.endswith('T00:00:00'): |
97
|
|
|
# Ignore time suffix. |
98
|
1 |
|
date = date[0:-9] |
99
|
1 |
|
tmp = datetime.datetime.strptime(date, '%Y-%m-%d') |
100
|
1 |
|
return tmp.toordinal() |
101
|
|
|
|
102
|
|
|
if isinstance(date, datetime.date): |
103
|
|
|
return date.toordinal() |
104
|
|
|
|
105
|
|
|
if isinstance(date, int): |
106
|
|
|
return date |
107
|
|
|
|
108
|
|
|
raise ValueError('Unexpected type {0!s}'.format(date.__class__)) |
109
|
|
|
|
110
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
111
|
1 |
|
def _rows_date2int(self, rows): |
112
|
|
|
""" |
113
|
|
|
Replaces start and end dates in a row set with their integer representation |
114
|
|
|
|
115
|
|
|
:param list[dict[str,T]] rows: The list of rows. |
116
|
|
|
|
117
|
|
|
:rtype: list[dict[str,T]] |
118
|
|
|
""" |
119
|
1 |
|
ret = list() |
120
|
1 |
|
for row in rows: |
121
|
|
|
# Determine the type of dates based on the first start date. |
122
|
1 |
|
if not self._date_type: |
123
|
1 |
|
self._date_type = self._get_date_type(row[self._key_start_date]) |
124
|
|
|
|
125
|
|
|
# Convert dates to integers. |
126
|
1 |
|
row[self._key_start_date] = self._date2int(row[self._key_start_date]) |
127
|
1 |
|
row[self._key_end_date] = self._date2int(row[self._key_end_date]) |
128
|
1 |
|
ret.append(row) |
129
|
|
|
|
130
|
1 |
|
return ret |
131
|
|
|
|
132
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
133
|
1 |
|
def _rows_int2date(self, rows): |
134
|
|
|
""" |
135
|
|
|
Replaces start and end dates in the row set with their integer representation |
136
|
|
|
|
137
|
|
|
:param list[dict[str,T]] rows: The list of rows. |
138
|
|
|
""" |
139
|
1 |
|
for row in rows: |
140
|
1 |
|
if self._date_type == 'str': |
141
|
1 |
|
row[self._key_start_date] = datetime.date.fromordinal(row[self._key_start_date]).isoformat() |
142
|
1 |
|
row[self._key_end_date] = datetime.date.fromordinal(row[self._key_end_date]).isoformat() |
143
|
|
|
elif self._date_type == 'date': |
144
|
|
|
row[self._key_start_date] = datetime.date.fromordinal(row[self._key_start_date]) |
145
|
|
|
row[self._key_end_date] = datetime.date.fromordinal(row[self._key_end_date]) |
146
|
|
|
elif self._date_type == 'int': |
147
|
|
|
# Nothing to do. |
148
|
|
|
pass |
149
|
|
|
else: |
150
|
|
|
raise ValueError('Unexpected date type {0!s}'.format(self._date_type)) |
151
|
|
|
|
152
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
153
|
1 |
|
def _rows_sort(self, rows): |
154
|
|
|
""" |
155
|
|
|
Returns a list of rows sorted by start and end date. |
156
|
|
|
|
157
|
|
|
:param list[dict[str,T]] rows: The list of rows. |
158
|
|
|
|
159
|
|
|
:rtype: list[dict[str,T]] |
160
|
|
|
""" |
161
|
1 |
|
return sorted(rows, key=lambda row: (row[self._key_start_date], row[self._key_end_date])) |
162
|
|
|
|
163
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
164
|
1 |
|
@staticmethod |
165
|
|
|
def _get_date_type(date): |
166
|
|
|
""" |
167
|
|
|
Returns the type of a date. |
168
|
|
|
|
169
|
|
|
:param str|datetime.date date: The date. |
170
|
|
|
|
171
|
|
|
:rtype: str |
172
|
|
|
""" |
173
|
1 |
|
if isinstance(date, str): |
174
|
1 |
|
return 'str' |
175
|
|
|
|
176
|
|
|
if isinstance(date, datetime.date): |
177
|
|
|
return 'date' |
178
|
|
|
|
179
|
|
|
if isinstance(date, int): |
180
|
|
|
return 'int' |
181
|
|
|
|
182
|
|
|
raise ValueError('Unexpected type {0!s}'.format(date.__class__)) |
183
|
|
|
|
184
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
185
|
1 |
|
def enumerate(self, name, start=1): |
186
|
|
|
""" |
187
|
|
|
Enumerates all rows such that the natural key and the ordinal number are a unique key. |
188
|
|
|
|
189
|
|
|
:param str name: The key holding the ordinal number. |
190
|
|
|
:param int start: The start of the ordinal numbers. Foreach natural key the first row has this ordinal number. |
191
|
|
|
""" |
192
|
1 |
|
for natural_key, rows in self.rows.items(): |
193
|
1 |
|
rows = self._rows_sort(rows) |
194
|
1 |
|
ordinal = start |
195
|
1 |
|
for row in rows: |
196
|
1 |
|
row[name] = ordinal |
197
|
1 |
|
ordinal += 1 |
198
|
1 |
|
self.rows[natural_key] = rows |
199
|
|
|
|
200
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
201
|
1 |
|
def get_rows(self, sort=False): |
202
|
|
|
""" |
203
|
|
|
Returns the rows of this Type2Helper. |
204
|
|
|
|
205
|
|
|
:param bool sort: If True the rows are sorted by the natural key. |
206
|
|
|
""" |
207
|
1 |
|
ret = [] |
208
|
1 |
|
for _, rows in sorted(self.rows.items()) if sort else self.rows.items(): |
209
|
1 |
|
self._rows_int2date(rows) |
210
|
1 |
|
ret.extend(rows) |
211
|
|
|
|
212
|
1 |
|
return ret |
213
|
|
|
|
214
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
215
|
1 |
|
def prepare_data(self, rows): |
216
|
|
|
""" |
217
|
|
|
Sets and prepares the rows. The rows are stored in groups in a dictionary. A group is a list of rows with the |
218
|
|
|
same natural key. The key in the dictionary is a tuple with the values of the natural key. |
219
|
|
|
|
220
|
|
|
:param list[dict] rows: The rows |
221
|
|
|
""" |
222
|
1 |
|
self.rows = dict() |
223
|
1 |
|
for row in copy.copy(rows) if self._copy else rows: |
224
|
1 |
|
natural_key = self._get_natural_key(row) |
225
|
1 |
|
if natural_key not in self.rows: |
226
|
1 |
|
self.rows[natural_key] = list() |
227
|
1 |
|
self.rows[natural_key].append(row) |
228
|
|
|
|
229
|
|
|
# Convert begin and end dates to integers. |
230
|
1 |
|
self._date_type = None |
231
|
1 |
|
for natural_key, rows in self.rows.items(): |
232
|
1 |
|
self.rows[natural_key] = self._rows_date2int(rows) |
233
|
|
|
|
234
|
|
|
# ---------------------------------------------------------------------------------------------------------------------- |
235
|
|
|
|