Rename code folder and add filters to parser

code/database/columns.py

@@ -0,0 +1,52 @@
+from db_setup import db
+from datetime import datetime
+
+
+class Glacier(db.Model):
+    uid = db.Column(db.Integer, primary_key=True)
+    country = db.Column(db.String(30))
+    name = db.Column(db.String(30))
+
+    def __init__(self, uid, country, name):
+        self.uid = uid
+        self.country = country
+        self.name = name
+
+
+class Annual_Data(db.Model):
+    glacier = db.relationship("Glacier", backref=db.backref("posts", lazy=True))
+    uid = db.Column(db.Integer, db.ForeignKey("glacier.uid"), nullable=False)
+    surface = db.Column(db.Integer)
+    volume = db.Column(db.Integer)
+    thickness = db.Column(db.Integer)
+    year = db.Column(db.Integer)
+
+    def __init__(self, surface, volume, thickness, year):
+        self.surface = surface
+        self.volume = volume
+        self.thickness = thickness
+        self.year = year
+
+
+class Annual_Change(db.Model):
+    glacier = db.relationship("Glacier", backref=db.backref("posts", lazy=True))
+    uid = db.Column(db.Integer, db.ForeignKey("glacier.uid"), nullable=False)
+    surface_change = db.Column(db.Integer)
+    volume_change = db.Column(db.Integer)
+    thickness_change = db.Column(db.Integer)
+    year = db.Column(db.Integer)
+
+    def __init__(self, surface_change, volume_change, thickness_change, year):
+        self.surface_change = surface_change
+        self.volume_change = volume_change
+        self.thickness_change = thickness_change
+        self.year = year
+
+
+class Users(db.Model):
+    uid = db.Column(db.Integer, primary_key=True)
+    registration_date = db.Column(db.DateTime, nullable=False)
+
+    def __init__(self, uid):
+        self.uid = uid
+        self.registration_date = datetime.today()

code/database/db_setup.py

@@ -0,0 +1,49 @@
+from subprocess import call
+from sys import argv
+from flask_sqlalchemy import SQLAlchemy
+from flask import Flask
+from fields import Glacier, Annual_Data, Annual_Change, User
+
+global db
+
+
+def create_database(db_name, db_user, db_pw):
+    script = "mariadb_setup.sh"
+    output = call([script + db_name + db_user + db_pw])
+    if output != 0:
+        print("Error: couldn't create database")
+        exit()
+
+
+def create_connection(db_name, db_user, db_pw):
+    host = "localhost:3306"
+    connection_uri = "mysql+pymysql://{user}:{pw}@{url}/{db}".format(
+        user=db_user, pw=db_pw, url=host, db=db_name
+    )
+    app = Flask(__name__)
+    app.config["SQLALCHEMY_DATABASE_URI"] = connection_uri
+    app.config["SQLALCHEMY_TRACK_MODIFICATIONS"] = False
+    global db
+    db = SQLAlchemy(app)
+
+
+def create_tables(db):
+    db.create_all()
+
+
+def main():
+    if len(argv) != 3:
+        print("Usage: " + argv[0] + " <db name> <db user> <db password>")
+        exit()
+
+    db_name = str(argv[1])
+    db_user = str(argv[2])
+    db_password = str(argv[3])
+
+    create_database(db_name, db_user, db_password)
+    create_connection(db_name, db_user, db_password)
+    create_tables(db)
+
+
+if __name__ == "__main__":
+    main()

code/database/mariadb_setup.sh

@@ -0,0 +1,19 @@
+#!/bin/sh
+
+EXPECTED_ARGS=3
+ERROR=1
+MYSQL=$(command -v which)
+
+if [ $# -ne $EXPECTED_ARGS ]
+then
+  echo "Usage: $0 <database name> <database user> <database password>"
+  exit $ERROR
+fi
+  
+Q1="CREATE DATABASE IF NOT EXISTS $1;"
+Q2="GRANT USAGE ON *.* TO $2@localhost IDENTIFIED BY '$3';"
+Q3="GRANT ALL PRIVILEGES ON $1.* TO $2@localhost;"
+Q4="FLUSH PRIVILEGES;"
+SQL="${Q1}${Q2}${Q3}${Q4}"
+  
+$MYSQL -uroot -p -e "$SQL"

code/database/parse_csv.py

@@ -0,0 +1,16 @@
+from pandas import read_csv, concat, DataFrame
+from iso3166 import countries
+
+
+def select_columns() -> [DataFrame]:
+    min_year = 2010
+    fields = ["POLITICAL_UNIT", "WGMS_ID", "YEAR"]
+    iter_csv = read_csv(
+        "../../Assets/WGMS-FoG-2019-12-D-CHANGE.csv",
+        skipinitialspace=True,
+        usecols=fields,
+        iterator=True,
+        chunksize=100,
+    )
+    data = concat([chunk[chunk["YEAR"] > min_year] for chunk in iter_csv])
+    return data