Ajouter la table avec les sportifs

data/make-sql.py

+"""
+Script pour générer le fichier SQL avec les données des JO de Paris 2024
+"""
+
 import json
-import random
 
-labels = {
+datasets = [
+	{
+		'filename': 'paris-2024-results-medals-oly-eng.json',
+		'url': 'https://data.education.gouv.fr/explore/dataset/paris-2024-results-medals-oly-eng/',
+		'table': 'medals',
+		'labels': {
 			'name': {'comment': "Nom de l'athlète", 'type': 'varchar(100)'},
 			'gender': {'comment': "Genre de l'athlète", 'type': 'varchar(1)'},
 			'discipline': {'comment': "Discipline", 'type': 'varchar(100)'},
 			'code_discipline': {'comment': "Code de la discipline", 'type': 'varchar(4)'},
-	'event': {'comment': "Sport", 'type': 'varchar(100)'},
+			'event': {'comment': "Compétition", 'type': 'varchar(100)'},
 			'country': {'comment': "Pays", 'type': 'varchar(100)'},
 			'country_code': {'comment': "Code du pays", 'type': 'varchar(4)'},
 			'medal_type': {'comment': "Type de la médaille", 'type': 'varchar(20)'},
 			'medal_code': {'comment': "Code de la médaille", 'type': 'int(1)'},
 			'medal_date': {'comment': "Date de la médaille", 'type': 'varchar(10)'},
-	# 'url_event': {'comment': "URL des résultats", 'type': 'varchar(100)'},
+			'url_event': {'comment': "URL des résultats", 'type': 'varchar(100)'},
 		}
+	},
+	{
+		'filename': 'paris-2024-liste-athletes-engages-olypara.json',
+		'url': 'https://data.education.gouv.fr/explore/dataset/paris-2024-liste-athletes-engages-olypara/',
+		'table': 'athletes',
+		'labels': {
+			'preferred_given_name': {'comment': "Prénom", 'type': 'varchar(100)'},
+			'preferred_family_name': {'comment': "Nom", 'type': 'varchar(100)'},
+			'nationality': {'comment': "Nationalité", 'type': 'varchar(4)'},
+			'national_olympic_committee': {'comment': "Comité Olympique National", 'type': 'varchar(4)'},
+			'date_of_birth': {'comment': "Date de naissance", 'type': 'varchar(10)'},
+			'gender': {'comment': "Genre", 'type': 'varchar(1)'},
+			'discipline': {'comment': "Discipline", 'type': 'varchar(4)'},
+			'event': {'comment': "Compétition", 'type': 'varchar(100)'},
+		}
+	},
+]
 
 
 beginning = """SET SQL_MODE = "NO_AUTO_VALUE_ON_ZERO";
@@ -31,50 +56,51 @@ ending = """
 /*!40101 SET CHARACTER_SET_RESULTS=@OLD_CHARACTER_SET_RESULTS */;
 /*!40101 SET COLLATION_CONNECTION=@OLD_COLLATION_CONNECTION */;"""
 
-insert_into = 'INSERT INTO `joparis2024` (`id`, ' + ', '.join(f'`{k}`' for k in labels) + ') VALUES\n'
 
+def make_table(dataset):
+	lines = []
+	lines.append(f"\nDROP TABLE IF EXISTS `{dataset['table']}`;\n")
+	lines.append(f"CREATE TABLE IF NOT EXISTS `{dataset['table']}` (\n")
+	lines.append("  `id` int(11) NOT NULL AUTO_INCREMENT,\n")
+	for k, v in dataset['labels'].items():
+		lines.append(f'  `{k}` {v["type"]} NOT NULL COMMENT "{v["comment"]}",\n')
+	lines.append("  PRIMARY KEY (`id`)")
+	lines.append("\n) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4;\n\n")
+	return lines
 
-def make_row(medal, i) -> str:
+def make_row(row, labels, i) -> str:
 	"""Faire une ligne de code SQL"""
 	values = [i]
 	for key in labels:
 		if 'varchar' in labels[key]['type']:
-			if key in medal and medal[key] is not None:
-				string = medal[key].replace('"', '')
+			if key in row and row[key] is not None:
+				string = row[key].replace('"', '')
 				value = f'"{string}"'
 			else:
 				value = '""'
 		else:
-			value = medal[key] if key in medal else 0
+			value = row[key] if key in row else 0
 		values.append(value)
 	return '(' + ', '.join(f'{v}' for v in values) + ')'
-	# return f'({i}, "{fields["acad_mies"]}", "{fields["acc_ab"]}", "{fields["acc_aca_orig"]}", "{fields["ville_etab"]}", "{fields["lib_for_voe_ins"]}"),\n'
 
 
 if __name__ == "__main__":
 
-	with open('data/paris-2024-results-medals-oly-eng.json', 'r') as f:
-		data = json.load(f)
-
 	lines = [beginning]
-	lines.append("DROP TABLE IF EXISTS `joparis2024`;\n")
-	lines.append("CREATE TABLE IF NOT EXISTS `joparis2024` (\n")
-	lines.append("  `id` int(11) NOT NULL AUTO_INCREMENT,\n")
-	for k, v in labels.items():
-		lines.append(f'  `{k}` {v["type"]} NOT NULL COMMENT "{v["comment"]}",\n')
-	lines.append("  PRIMARY KEY (`id`)")
-	lines.append("\n) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4;\n\n")
+	for dataset in datasets:
+		insert_into = f"INSERT INTO `{dataset['table']}` (`id`, " + ', '.join(f'`{k}`' for k in dataset['labels']) + ") VALUES\n"
+		lines += make_table(dataset)
+		with open(dataset['filename'], 'r') as f:
+			data = json.load(f)
 		lines.append(insert_into)
-
 		for i, datum in enumerate(data):
 			if i % 100 == 99:
-			lines.append(make_row(datum, i+1) + ";\n")
+				lines.append(make_row(datum, dataset['labels'], i+1) + ";\n")
 				lines.append(insert_into)
 			elif i == len(data) - 1:
-			lines.append(make_row(datum, i+1) + ";\n")
+				lines.append(make_row(datum, dataset['labels'], i+1) + ";\n")
 			else:
-			lines.append(make_row(datum, i+1) + ",\n")
-
+				lines.append(make_row(datum, dataset['labels'], i+1) + ",\n")
 	lines.append(ending)
 
 	with open('sae303.sql', 'w') as writer: