blog.mycado.fr

Queries on XML data with XQuery

We know how to store native XML data in our DB2 tables, and now we will see how we can access to these data. We can choose between standard SQL queries and XQuery.. or both ! The first solution, with a SQL query, only query at the column level of your table; this query will return the full XML data. The second solution, with XQuery, allow us to make a “query” inside our XML data.

For the SQL query, nothing more than a SELECT:

SELECT id, info from client

Easy, but not really powerful for XML data, let’s try with XQuery, who give us two functions for DB2. db2-fn:sqlquery and db2-fn:xmlcolumn. The first function retrieves a sequence that is the result of an SQL fullselect and the second retrieves a sequence from a column. One important thing you should keep in mind, SQL is not a case-sensitive language alors que XQuery is a case-sensitive language.

An example which return all the XML data from the “info” column:

XQUERY db2-fnxmlcolumn ('CLIENT.INFO')

Which is same as this SQL query:

SELECT info FROM client

Let’s see something more nice. This query will return all the elements in <name> , inside the “info” column, and with the <city> element which containt “Paris”:

XQUERY declare defaut element namespace "http://posample.org";
for $d in db2-fn:xmlcolumn('CLIENT.INFO')/clientinfo
  where $d/addr/city="Paris"
return <out>{$d/name}</out>

db2-fn:xmlcolumn retrieves the data from the “info” column in the “client” table. We add a $d variable, for each element of <clientinfo>, and we use a where to filter the <city> element which should be “Paris. To finish, we use <out> to output the data:

<out xmlns="http://posample.org">
  <name>Sophie Bool</name>
</out>

XQUERY declare default element namespace "http://posample.org";
for $d in db2-fn:sqlquery('SELECT info FROM client')/clientinfo
  where $d/addr/city="Paris
return <out>{$d/name}</out>

Store native XML in a DB2 table

IBM DB2 is the first RDBMS to provide a native XML facility, in a table. You will be able to insert in a column, some XML data. It will be easier to make direct requests to these data with XQuery.

So, it’s funny, but why it’s interesting to do that ? Because XML data are hierarchic (instead of relational data which are flat) and data self-describing through XML tags. XML also allows a better flexibility for data structures required to change very often. In the other hand, access time performances will be a little slower, we lose the integrity constraints, and OLAP queries will be more difficult. The important question is “Which flexibility-performance ratio do you need ?”.

Let’s start now with the creation of a database, which should be encoded with UTF-8, to store XML. For that, we use the CREATE DATABASE command:

CREATE DATABASE xmldb USING CODESET UTF-8 TERRITORY US

We have our table, now we should create the “client” table, with an “info” column who contain client information, in XML format:

CONNECT TO xmldb
CREATE TABLE client (id INT, info XML)

Let’s try to insert a new client with a SQL query:

INSERT INTO client (id, info) VALUES (1, '<clientinfo xmlns="http://posample.org" Cid="1"><name>Sophie Bool</name><addr country="France"><street>5 rue du chateau de stable</street><city>Paris</city></addr><phone type="work">01 72 92 02 88</phone></clientinfo>')

The first thing you will told me is this query is a normal SQL query, and this is right, insert XML isn’t more difficult. The second thing is about the XML, here we have a short XML data, but if we have more, it will be very difficult to use. This why, we will use XQuery to manipulate these data or use a XDS import. I’ll come back soon on these points.