第一次合并clean代码
This commit is contained in:
33
clean/step0_pdfs2sql.py
Normal file
33
clean/step0_pdfs2sql.py
Normal file
@@ -0,0 +1,33 @@
|
||||
import os
|
||||
import tqdm
|
||||
import sqlite3
|
||||
import mysql.connector
|
||||
|
||||
def main():
|
||||
cur_path = os.path.dirname(os.path.abspath(__file__))
|
||||
|
||||
TABLE_NAME = 'mp_cif_info'
|
||||
|
||||
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()
|
||||
|
||||
pdf_list = os.listdir(os.path.join(cur_path, 'mp_cif/pdfs'))
|
||||
|
||||
doi_list = [pdf.replace('.pdf', '') for pdf in pdf_list]
|
||||
|
||||
try:
|
||||
for doi in doi_list:
|
||||
sql = f"INSERT INTO {TABLE_NAME} (doi) VALUES (%s)"
|
||||
mysql_cursor.execute(sql, (doi,))
|
||||
mysql_connection.commit()
|
||||
finally:
|
||||
mysql_connection.close()
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
Reference in New Issue
Block a user