|
1
|
|
|
"""The API for configuring datasets.""" |
|
2
|
|
|
|
|
3
|
|
|
from dataclasses import dataclass |
|
4
|
|
|
from typing import List, Set, Tuple, Union |
|
5
|
|
|
import collections.abc as cabc |
|
6
|
|
|
|
|
7
|
|
|
from airflow import DAG |
|
8
|
|
|
from airflow.operators import BaseOperator as Operator |
|
9
|
|
|
from airflow.utils.dates import days_ago |
|
10
|
|
|
from sqlalchemy import Column, ForeignKey, Integer, String, Table, orm |
|
11
|
|
|
from sqlalchemy.ext.declarative import declarative_base |
|
12
|
|
|
|
|
13
|
|
|
from egon.data import db |
|
14
|
|
|
|
|
15
|
|
|
Base = declarative_base() |
|
16
|
|
|
SCHEMA = "metadata" |
|
17
|
|
|
DEFAULTS = {"start_date": days_ago(1)} |
|
18
|
|
|
|
|
19
|
|
|
|
|
20
|
|
|
def setup(): |
|
21
|
|
|
"""Create the database structure for storing dataset information.""" |
|
22
|
|
|
# TODO: Move this into a task generating the initial database structure. |
|
23
|
|
|
db.execute_sql(f"CREATE SCHEMA IF NOT EXISTS {SCHEMA};") |
|
24
|
|
|
Model.__table__.create(bind=db.engine(), checkfirst=True) |
|
25
|
|
|
DependencyGraph.create(bind=db.engine(), checkfirst=True) |
|
26
|
|
|
|
|
27
|
|
|
|
|
28
|
|
|
DependencyGraph = Table( |
|
29
|
|
|
"dependency_graph", |
|
30
|
|
|
Base.metadata, |
|
31
|
|
|
Column( |
|
32
|
|
|
"dependency_id", |
|
33
|
|
|
Integer, |
|
34
|
|
|
ForeignKey(f"{SCHEMA}.datasets.id"), |
|
35
|
|
|
primary_key=True, |
|
36
|
|
|
), |
|
37
|
|
|
Column( |
|
38
|
|
|
"dependent_id", |
|
39
|
|
|
Integer, |
|
40
|
|
|
ForeignKey(f"{SCHEMA}.datasets.id"), |
|
41
|
|
|
primary_key=True, |
|
42
|
|
|
), |
|
43
|
|
|
schema="metadata", |
|
44
|
|
|
) |
|
45
|
|
|
|
|
46
|
|
|
|
|
47
|
|
|
class Model(Base): |
|
48
|
|
|
__tablename__ = "datasets" |
|
49
|
|
|
__table_args__ = {"schema": SCHEMA} |
|
50
|
|
|
id = Column(Integer, primary_key=True) |
|
51
|
|
|
name = Column(String, unique=True, nullable=False) |
|
52
|
|
|
version = Column(String, nullable=False) |
|
53
|
|
|
epoch = Column(Integer, default=0) |
|
54
|
|
|
dependencies = orm.relationship( |
|
55
|
|
|
"Model", |
|
56
|
|
|
secondary=DependencyGraph, |
|
57
|
|
|
primaryjoin=id == DependencyGraph.c.dependent_id, |
|
58
|
|
|
secondaryjoin=id == DependencyGraph.c.dependency_id, |
|
59
|
|
|
backref="dependents", |
|
60
|
|
|
) |
|
61
|
|
|
|
|
62
|
|
|
|
|
63
|
|
|
TaskGraph = Union[Operator, "ParallelTasks", "SequentialTasks"] |
|
64
|
|
|
ParallelTasks = Set[TaskGraph] |
|
65
|
|
|
SequentialTasks = Tuple[TaskGraph, ...] |
|
66
|
|
|
|
|
67
|
|
|
|
|
68
|
|
|
@dataclass |
|
69
|
|
|
class Tasks: |
|
70
|
|
|
first: Set[Operator] |
|
71
|
|
|
last: Set[Operator] |
|
72
|
|
|
all: Set[Operator] |
|
73
|
|
|
|
|
74
|
|
|
|
|
75
|
|
|
def connect(tasks: TaskGraph): |
|
76
|
|
|
"""Connect multiple tasks into a potentially complex graph. |
|
77
|
|
|
|
|
78
|
|
|
As per the type, a task graph can be given as a single operator, a tuple |
|
79
|
|
|
of task graphs or a set of task graphs. A tuple will be executed in the |
|
80
|
|
|
specified order, whereas a set means that the tasks in the graph will be |
|
81
|
|
|
executed in parallel. |
|
82
|
|
|
""" |
|
83
|
|
|
if isinstance(tasks, Operator): |
|
84
|
|
|
return Tasks(first={tasks}, last={tasks}, all={tasks}) |
|
85
|
|
|
elif isinstance(tasks, cabc.Sized) and len(tasks) == 0: |
|
86
|
|
|
return Tasks(first={}, last={}, all={}) |
|
87
|
|
|
elif isinstance(tasks, cabc.Set): |
|
88
|
|
|
results = [connect(subtasks) for subtasks in tasks] |
|
89
|
|
|
first = {task for result in results for task in result.first} |
|
90
|
|
|
last = {task for result in results for task in result.last} |
|
91
|
|
|
tasks = {task for result in results for task in result.all} |
|
92
|
|
|
return Tasks(first, last, tasks) |
|
93
|
|
|
elif isinstance(tasks, tuple): |
|
94
|
|
|
results = [connect(subtasks) for subtasks in tasks] |
|
95
|
|
|
for (left, right) in zip(results[:-1], results[1:]): |
|
96
|
|
|
for last in left.last: |
|
97
|
|
|
for first in right.first: |
|
98
|
|
|
last.set_downstream(first) |
|
99
|
|
|
first = results[0].first |
|
100
|
|
|
last = results[-1].last |
|
101
|
|
|
tasks = {task for result in results for task in result.all} |
|
102
|
|
|
return Tasks(first, last, tasks) |
|
103
|
|
|
else: |
|
104
|
|
|
raise ( |
|
105
|
|
|
TypeError( |
|
106
|
|
|
"`egon.data.datasets.connect` got an argument of type:\n\n" |
|
107
|
|
|
f" {type(tasks)}\n\n" |
|
108
|
|
|
"where only `Operator`s, `Set`s and `Tuple`s are allowed." |
|
109
|
|
|
) |
|
110
|
|
|
) |
|
111
|
|
|
|
|
112
|
|
|
|
|
113
|
|
|
@dataclass |
|
114
|
|
|
class Dataset: |
|
115
|
|
|
name: str |
|
116
|
|
|
version: str |
|
117
|
|
|
dependencies: List["Dataset"] |
|
118
|
|
|
graph: TaskGraph |
|
119
|
|
|
|
|
120
|
|
|
def __post_init__(self): |
|
121
|
|
|
self.tasks = connect(self.graph) |
|
122
|
|
|
predecessors = [p for d in self.dependencies for p in d.tasks.last] |
|
123
|
|
|
for p in predecessors: |
|
124
|
|
|
for first in self.tasks.first: |
|
125
|
|
|
p.set_downstream(first) |
|
126
|
|
|
|
|
127
|
|
|
def insert_into(self, dag: DAG): |
|
128
|
|
|
dag.add_tasks(self.tasks.all) |
|
129
|
|
|
|
openego /
eGon-data
