modify scripts
This commit is contained in:
@ -60,9 +60,9 @@ class SQLiteDBHandler:
|
||||
|
||||
return processed_data
|
||||
|
||||
def insert_or_update_common(self, data, tbl_name, uniq_key='url'):
|
||||
def insert_or_update_common(self, data, tbl_name, uniq_key='url', exists_do_nothing=False):
|
||||
if self.lower_sqlite_version:
|
||||
return self.insert_or_update_common_lower(data, tbl_name, uniq_key)
|
||||
return self.insert_or_update_common_lower(data, tbl_name, uniq_key, exists_do_nothing)
|
||||
|
||||
try:
|
||||
processed_data = self.check_and_process_data(data, tbl_name)
|
||||
@ -72,12 +72,17 @@ class SQLiteDBHandler:
|
||||
columns = ', '.join(processed_data.keys())
|
||||
values = list(processed_data.values())
|
||||
placeholders = ', '.join(['?' for _ in values])
|
||||
update_clause = ', '.join([f"{col}=EXCLUDED.{col}" for col in processed_data.keys() if col != uniq_key])
|
||||
|
||||
if exists_do_nothing:
|
||||
conflict_clause = f'ON CONFLICT ({uniq_key}) DO NOTHING'
|
||||
else:
|
||||
update_clause = ', '.join([f"{col}=EXCLUDED.{col}" for col in processed_data.keys() if col != uniq_key])
|
||||
conflict_clause = f"ON CONFLICT ({uniq_key}) DO UPDATE SET {update_clause}"
|
||||
|
||||
sql = f'''
|
||||
INSERT INTO {tbl_name} ({columns})
|
||||
VALUES ({placeholders})
|
||||
ON CONFLICT ({uniq_key}) DO UPDATE SET {update_clause}
|
||||
VALUES ({placeholders})
|
||||
{conflict_clause}
|
||||
'''
|
||||
self.cursor.execute(sql, values)
|
||||
self.conn.commit()
|
||||
@ -90,7 +95,7 @@ class SQLiteDBHandler:
|
||||
logging.error(f"Error inserting or updating data: {e}")
|
||||
return None
|
||||
|
||||
def insert_or_update_common_lower(self, data, tbl_name, uniq_key='url'):
|
||||
def insert_or_update_common_lower(self, data, tbl_name, uniq_key='url', exists_do_nothing=False):
|
||||
try:
|
||||
processed_data = self.check_and_process_data(data, tbl_name)
|
||||
if processed_data is None:
|
||||
@ -109,12 +114,13 @@ class SQLiteDBHandler:
|
||||
self.cursor.execute(sql, values)
|
||||
self.conn.commit()
|
||||
except sqlite3.IntegrityError: # 唯一键冲突,执行更新操作
|
||||
update_clause = ', '.join([f"{col}=?" for col in processed_data.keys() if col != uniq_key])
|
||||
update_values = [processed_data[col] for col in processed_data.keys() if col != uniq_key]
|
||||
update_values.append(data[uniq_key])
|
||||
sql = f"UPDATE {tbl_name} SET {update_clause} WHERE {uniq_key} = ?"
|
||||
self.cursor.execute(sql, update_values)
|
||||
self.conn.commit()
|
||||
if not exists_do_nothing:
|
||||
update_clause = ', '.join([f"{col}=?" for col in processed_data.keys() if col != uniq_key])
|
||||
update_values = [processed_data[col] for col in processed_data.keys() if col != uniq_key]
|
||||
update_values.append(data[uniq_key])
|
||||
sql = f"UPDATE {tbl_name} SET {update_clause} WHERE {uniq_key} = ?"
|
||||
self.cursor.execute(sql, update_values)
|
||||
self.conn.commit()
|
||||
|
||||
# 获取插入或更新后的记录 ID
|
||||
self.cursor.execute(f"SELECT id FROM {tbl_name} WHERE {uniq_key} = ?", (data[uniq_key],))
|
||||
|
||||
@ -76,7 +76,7 @@ class SQLitePipeline(SQLiteDBHandler):
|
||||
|
||||
def _process_u001_item(self, item, spider):
|
||||
logging.debug(f"insert one item. href:{spider.name}")
|
||||
return self.insert_or_update_common(item, tbl_name=self.tbl_name_u3c3, uniq_key='url')
|
||||
return self.insert_or_update_common(item, tbl_name=self.tbl_name_u3c3, uniq_key='url', exists_do_nothing=True)
|
||||
|
||||
def _process_sis001_item(self, item, spider):
|
||||
self.cursor.execute('''
|
||||
|
||||
Reference in New Issue
Block a user