第一次合并clean代码

clean/stp2.2_remove_broken_pdf.py

@@ -0,0 +1,28 @@
+import sqlite3
+import mysql.connector
+import tqdm
+import os
+
+TABLE_NAME = 'phosphorus_synthesis_info'
+input('TABLE_NAME = {} ?'.format(TABLE_NAME))
+
+cur_dir = os.path.dirname(os.path.abspath(__file__))
+
+# MySQL connection setup
+mysql_connection = mysql.connector.connect(
+    host='100.84.94.73',
+    user='metadata_mat_papers',
+    password='siat-mic',
+    database='metadata_mat_papers'
+)
+mysql_cursor = mysql_connection.cursor()
+
+
+# 编写query语句
+query = f"SELECT pdf_url FROM {TABLE_NAME} WHERE scihub_downlowded = 'broken'"
+mysql_cursor.execute(query)
+records = mysql_cursor.fetchall()
+
+for record in tqdm.tqdm(records):
+    pdf_path = os.path.join(cur_dir, record[0])
+    os.remove(pdf_path)