PHP- Automated Housekeeping

PHP- Automated Housekeeping

Queries are run by users through the web interface and by administrators through either administrative web interfaces or from the MySQL command interpreter. However, sometimes automated querying is necessary to produce periodic reports, update data, or delete temporary data. We discuss how queries can be automated in this section.

To show how queries can be automated, consider an example from the online winestore. The shopping cart in the online winestore is implemented using the winestore database. As discussed in Chapter 12, when an anonymous user adds a wine to their shopping basket, an order row is added to the orders table. The row is for a dummy customer with a cust_id=-1, and the next available order_id for this dummy customer. A related items row is created for each item in the shopping cart. The order_id is maintained in the session variable order_no so that orders by different anonymous customers aren’t confused.

Our system requirements in Chapter 1 specify that if a customer doesn’t purchase the wines in their shopping cart within one day, then the shopping cart should be emptied. This is an example of a DELETE operation that should be automated. It is impractical to require the administrator to run this query each day to remove junk data.

The following query can be run from the Linux shell to remove all orders rows that are more than one day old and are for the dummy customer:

% /usr/local/mysql/bin/mysql -uusername -psecret

-e ‘USE winestore; DELETE FROM orders WHERE

unix_timestamp(date) < (unix_timestamp(date_add(now( ), interval -1 day))) AND cust_id = -1;’ The MySQL time and date function unix_timestamp( ) converts a timestamp attribute to an integer that is accurate to the nearest second. In this query, we compare the value of the entry in the orders table with the value of exactly one day earlier from the current date and time. If the row is older than one day, then it is deleted. The same query works for the items table, when orders is replaced with items in the FROM clause. 13.1.1 cron Jobs Having designed and tested the query, it can be inserted into a Unix cron table to automate the operation. The crond daemon is a process that runs by default in a Linux installation and continually checks the time. If any of the entries in user tables match the current time, then the commands in the entries are executed. Consider an example: 30 17 * * mon-fri echo ‘Go home!’ This prints the string at 5:30 p.m. each working day. The two asterisks mean every day of the month, and every month of the year respectively. The string mon-fri means the days Monday to Friday inclusive. More details about cron can be found by running man crontab in a Linux shell. We can add our housekeeping query to our cron table by running: % crontab -e This edits the user’s cron table. We have decided that the system should check for old shopping carts every 30 minutes. To do so, we add the following two lines to the file: 0 * * * * /usr/local/mysql/bin/mysql -uusername -psecret -e ‘USE winestore; DELETE FROM orders WHERE unix_timestamp(date) < (unix_timestamp(date_add(now( ), interval -1 day))) AND cust_id = -1;’ 30 * * * * /usr/local/mysql/bin/mysql -uusername -psecret -e ‘USE winestore; DELETE FROM items WHERE unix_timestamp(date) < (unix_timestamp(date_add(now( ), interval -1 day))) AND cust_id = -1;’ The first line contains the complete query command for the orders table from earlier in this section, and the second line the items query. The shopping cart orders DELETE query runs exactly on each hour, while the items DELETE query runs at 30 minutes past each hour. Different times are used to balance the DBMS load. Reports, updates, delete operations, and other tasks can be added to the cron table in a similar way. For example, we can output a simple report of the number of bottles purchased yesterday and send this to our email address each morning: 0 8 * * * mon-fri /usr/local/mysql/bin/mysql -uusername -psecret -e ‘USE winestore; SELECT sum(qty) FROM items WHERE unix_timestamp(date) >

(unix_timestamp(date_add(now( ), interval -1 day))) AND

cust_id != -1;’ | mail

We could also have automatically written the information to a log file or to a table in the database.

More PHP Tutorial