- Create the file
/usr/local/sphinx/etc/sphinx-blog.confwith the following content:source blog { type = mysql sql_host = localhost sql_user = root sql_pass = sql_db = myblog sql_query = SELECT id, title, content FROM posts sql_query_info = SELECT id, title FROM posts WHERE ID=$id } index posts { source = blog path = /usr/local/sphinx/var/data/blog docinfo = extern charset_type = sbcs } indexer { mem_limit = 32M } - Run the following command to create the index:
$ /usr/local/sphinx/bin/indexer --config /usr/local/sphinx/etc/sphinx-blog.conf --all
- Now test the index by searching from the command line search utility:
$ /usr/local/sphinx/bin/search --config /usr/local/sphinx/etc/sphinx-blog.conf php
We created a configuration file /usr/local/sphinx/etc/sphinx-blog.conf, which is later used by Sphinx to create an index.
The first block in the configuration file defines the data source named blog which is of type mysql. We provided values for the options to connect to the database and a query (sql_query) which fetches all the records from the database for indexing. The last option in the configuration file is sql_query_info, which is used to display additional information related to the searched documents. Because of this we see the id and title in the search results.
The next block in the configuration file defines index. The index will be saved at /usr/local/sphinx/var/data/blog on the file system.To create the actual index we used the indexer program. indexer takes the path of the config file as the argument.
Another argument we passed was --all which says that all indexes defined in the configuration file should be indexed. In our case there was just one index and we had named it posts.
There are a number of arguments that can be passed to the indexer. To view a list of all the arguments issue the following command:
$ /usr/local/sphinx/bin/indexer
The last thing we did was perform a search for the term "php", which returned two documents. This concluded that our index is working fine.
Note
The search utility used to perform the search is one of the helper tools available to quickly test the index from the command line without writing the code to connect to the searchd server and process its response.
search is not intended to be used in a client application. You should use searchd and the bundle client APIs to perform a search from within your application. We will be taking a look at how to use searchd and a client API to perform search in Chapter 4,Searching.
Similar to the indexer program, search also takes a number of arguments. Since we were not using the default configuration file, we passed the path of configuration file as an argument to the search command. The search term, that is "php", should always be last in the list of arguments.
$ /usr/local/sphinx/bin/search
Indexing works in the same fashion with all the SQL drivers. When indexer is run, a database connection is established using the credentials provided in the configuration file. After that, the main query, the sql_query is fired to fetch the data to be indexed. Once this is done the connection to the database is closed and the indexer does the sorting phase.
The index we created for the blog posts is all good and fine, but it only works for full-text searching. What if we want to filter the results by author or date? That can't be done with the index that we created earlier. To solve this problem, Sphinx offers special fields in the index called attributes.
Let's add attributes to our index to hold the author_id and publish_date.