|
| 1 | +import logging |
| 2 | +import os |
| 3 | +from datetime import datetime |
| 4 | + |
| 5 | +import requests |
| 6 | +import sqlalchemy |
| 7 | +from sqlalchemy import create_engine, update |
| 8 | +from sqlalchemy.engine import Engine |
| 9 | +from sqlalchemy.orm import DeclarativeBase, sessionmaker |
| 10 | + |
| 11 | +logger = logging.getLogger(__name__) |
| 12 | + |
| 13 | +SQLALCHEMY_DATABASE_URI = os.environ.get( |
| 14 | + "SQLALCHEMY_DATABASE_URI", |
| 15 | + "mysql+mysqlconnector://root:ROOT_PASSWORD@localhost:3306/mysql_database", |
| 16 | +) |
| 17 | +engine: Engine = create_engine( |
| 18 | + SQLALCHEMY_DATABASE_URI, execution_options={"engine_id": "default"} |
| 19 | +) |
| 20 | +Session = sessionmaker(bind=engine) |
| 21 | + |
| 22 | + |
| 23 | +COVID_CAST_META_URL = os.environ.get( |
| 24 | + "COVID_CAST_META_URL", "https://api.delphi.cmu.edu/epidata/covidcast/meta" |
| 25 | +) |
| 26 | + |
| 27 | + |
| 28 | +class Base(DeclarativeBase): |
| 29 | + pass |
| 30 | + |
| 31 | + |
| 32 | +class Signal(Base): |
| 33 | + |
| 34 | + __tablename__ = "signals_signal" |
| 35 | + |
| 36 | + id = sqlalchemy.Column(sqlalchemy.Integer, primary_key=True) |
| 37 | + name = sqlalchemy.Column(sqlalchemy.String) |
| 38 | + display_name = sqlalchemy.Column(sqlalchemy.String) |
| 39 | + active = sqlalchemy.Column(sqlalchemy.Boolean) |
| 40 | + short_description = sqlalchemy.Column(sqlalchemy.String) |
| 41 | + description = sqlalchemy.Column(sqlalchemy.String) |
| 42 | + format_type = sqlalchemy.Column(sqlalchemy.String) |
| 43 | + time_type = sqlalchemy.Column(sqlalchemy.String) |
| 44 | + time_label = sqlalchemy.Column(sqlalchemy.String) |
| 45 | + is_smoothed = sqlalchemy.Column(sqlalchemy.Boolean) |
| 46 | + is_weighted = sqlalchemy.Column(sqlalchemy.Boolean) |
| 47 | + is_cumulative = sqlalchemy.Column(sqlalchemy.Boolean) |
| 48 | + has_stderr = sqlalchemy.Column(sqlalchemy.Boolean) |
| 49 | + has_sample_size = sqlalchemy.Column(sqlalchemy.Boolean) |
| 50 | + high_values_are = sqlalchemy.Column(sqlalchemy.String) |
| 51 | + base_id = sqlalchemy.Column(sqlalchemy.Integer) |
| 52 | + category_id = sqlalchemy.Column(sqlalchemy.Integer) |
| 53 | + source_id = sqlalchemy.Column(sqlalchemy.Integer) |
| 54 | + created = sqlalchemy.Column(sqlalchemy.DateTime) |
| 55 | + modified = sqlalchemy.Column(sqlalchemy.DateTime) |
| 56 | + last_updated = sqlalchemy.Column(sqlalchemy.Date) |
| 57 | + age_breakdown = sqlalchemy.Column(sqlalchemy.String) |
| 58 | + data_censoring = sqlalchemy.Column(sqlalchemy.Text) |
| 59 | + gender_breakdown = sqlalchemy.Column(sqlalchemy.Integer) |
| 60 | + missingness = sqlalchemy.Column(sqlalchemy.Text) |
| 61 | + race_breakdown = sqlalchemy.Column(sqlalchemy.Integer) |
| 62 | + reporting_cadence = sqlalchemy.Column(sqlalchemy.String) |
| 63 | + restrictions = sqlalchemy.Column(sqlalchemy.Text) |
| 64 | + severenity_pyramid_rungs = sqlalchemy.Column(sqlalchemy.String) |
| 65 | + temporal_scope_end = sqlalchemy.Column(sqlalchemy.String) |
| 66 | + temporal_scope_end_note = sqlalchemy.Column(sqlalchemy.Text) |
| 67 | + temporal_scope_start = sqlalchemy.Column(sqlalchemy.String) |
| 68 | + temporal_scope_start_note = sqlalchemy.Column(sqlalchemy.Text) |
| 69 | + typical_reporting_lag = sqlalchemy.Column(sqlalchemy.String) |
| 70 | + typical_revision_cadence = sqlalchemy.Column(sqlalchemy.String) |
| 71 | + license_id = sqlalchemy.Column(sqlalchemy.Integer) |
| 72 | + signal_type_id = sqlalchemy.Column(sqlalchemy.Integer) |
| 73 | + from_date = sqlalchemy.Column(sqlalchemy.Date) |
| 74 | + to_date = sqlalchemy.Column(sqlalchemy.Date) |
| 75 | + geographic_scope_id = sqlalchemy.Column(sqlalchemy.Integer) |
| 76 | + signal_availability_days = sqlalchemy.Column(sqlalchemy.Integer) |
| 77 | + |
| 78 | + |
| 79 | +class Source(Base): |
| 80 | + __tablename__ = "datasources_sourcesubdivision" |
| 81 | + |
| 82 | + id = sqlalchemy.Column(sqlalchemy.Integer, primary_key=True) |
| 83 | + name = sqlalchemy.Column(sqlalchemy.String) |
| 84 | + display_name = sqlalchemy.Column(sqlalchemy.String) |
| 85 | + description = sqlalchemy.Column(sqlalchemy.String) |
| 86 | + db_source = sqlalchemy.Column(sqlalchemy.String) |
| 87 | + data_source_id = sqlalchemy.Column(sqlalchemy.Integer) |
| 88 | + created = sqlalchemy.Column(sqlalchemy.DateTime) |
| 89 | + modified = sqlalchemy.Column(sqlalchemy.DateTime) |
| 90 | + external_name = sqlalchemy.Column(sqlalchemy.String) |
| 91 | + |
| 92 | + |
| 93 | +class SignalLastUpdatedParser: |
| 94 | + |
| 95 | + def __init__(self, covidcast_meta_data: list) -> None: |
| 96 | + self.covidcast_meta_data = covidcast_meta_data |
| 97 | + self.year_week_date_format = "%Y-%W-%w" |
| 98 | + self.year_month_day_date_format = "%Y%m%d" |
| 99 | + |
| 100 | + def format_date( |
| 101 | + self, |
| 102 | + date: str, |
| 103 | + ) -> datetime: |
| 104 | + """ |
| 105 | + Format the date string to a specific format. |
| 106 | +
|
| 107 | + :param date: The date string to format. |
| 108 | + :return: The formatted date string. |
| 109 | + :rtype: str |
| 110 | + """ |
| 111 | + formated_date: datetime |
| 112 | + if len(date) == 6: |
| 113 | + year, week = date[:4], date[4:] |
| 114 | + logger.info(f"Date: {date}, year: {year}, week: {int(week)-1}") |
| 115 | + formated_date = datetime.strptime( |
| 116 | + f"{int(year)}-{int(week)-1}-1", self.year_week_date_format |
| 117 | + ) |
| 118 | + elif len(date) == 8: |
| 119 | + formated_date = datetime.strptime(date, self.year_month_day_date_format) |
| 120 | + return formated_date |
| 121 | + |
| 122 | + def set_data(self) -> None: |
| 123 | + """ |
| 124 | + Set the last updated date for signals in the database. |
| 125 | + """ |
| 126 | + with Session() as session: |
| 127 | + for db_source in self.covidcast_meta_data: |
| 128 | + for signal_data in db_source["signals"]: |
| 129 | + source = ( |
| 130 | + session.query(Source) |
| 131 | + .filter(Source.name == signal_data["source"]) |
| 132 | + .first() |
| 133 | + ) |
| 134 | + last_updated = self.format_date(str(signal_data["max_issue"])) |
| 135 | + from_date = self.format_date(str(signal_data["min_time"])) |
| 136 | + to_date = self.format_date(str(signal_data["max_time"])) |
| 137 | + signal_availability_days = abs((to_date - from_date).days) |
| 138 | + try: |
| 139 | + session.execute( |
| 140 | + update(Signal) |
| 141 | + .where(Signal.name == signal_data["signal"]) |
| 142 | + .where(Signal.source_id == source.id) |
| 143 | + .values( |
| 144 | + last_updated=last_updated, |
| 145 | + from_date=from_date, |
| 146 | + to_date=to_date, |
| 147 | + signal_availability_days=signal_availability_days, |
| 148 | + ) |
| 149 | + ) |
| 150 | + session.commit() |
| 151 | + logger.info(f"Signal {signal_data['signal']} successfully updated.") |
| 152 | + except AttributeError: |
| 153 | + logger.error(f"Failed to update signal {signal_data['signal']}. Probably the issue is with the source or source with name {signal_data['source']} does not exist.") |
| 154 | + |
| 155 | + |
| 156 | +def main(): |
| 157 | + response = requests.get(COVID_CAST_META_URL) |
| 158 | + if response.status_code == 200: |
| 159 | + covidcast_meta_data = response.json() |
| 160 | + parser = SignalLastUpdatedParser(covidcast_meta_data) |
| 161 | + parser.set_data() |
| 162 | + else: |
| 163 | + logger.error(f"Failed to get data from {COVID_CAST_META_URL}") |
| 164 | + |
| 165 | + |
| 166 | +if __name__ == "__main__": |
| 167 | + main() |
0 commit comments