Basic sphinx api setup

I've seen this budding confusion a few times, so thought I'd give a slightly more detailed quick start. So, a basic introduction to get started looking for Sphinx:

Usually sphinx project files are stored in three directories: project/sphinx/etc

, project/sphinx/logs

and project/sphinx/data

. They can be changed as needed, but I am following this convention below.

Sphinx launch

After installing Sphinx on your system, you will have two executables: indexer

and searchd

. indexer

is a program that checks the configuration file and indexes the data you provide for quick access. searchd

- a background process (or daemon) that you usually want to keep running. Think of it like the Sphinx search engine. All requests from are sphinxapi

directed to searchd

, which will search the indexed data and return the record IDs in the order you specify. Examples of running both of these executables (on Linux, although in windows shouldn't be too different) are given below:

indexer -c /path/to/project/sphinx/etc/sphinx.conf --all

      

        
        
        
      

    

Specifies all sources specified in the config file. Additional options are available to index only specified sources. A detailed list of options is here .

searchd -c /path/to/project/sphinx/etc/sphinx.conf --rotate

      

        
        
        
      

    

This will initialize the daemon searchd

and deploy it in the background, where it will keep running until you kill it. This parameter rotate

allows you to keep the daemon running even if the indexes are updated in the future. You can use the switch --stop

to stop the demon. A detailed list of options is here .

Sphinx setup

This brings us to the configuration file. The config file consists of a set of configurations source

and index

together with a section searchd

used to customize the Sphinx itself. Many options can be configured here, but the basic one sphinx.conf

(usually the sphinx.conf.dist file found in the sphinx install folder) is a bit overwhelming at first. I'm just mentioning a few basic configuration options to get started. I assume you are using MySQL

, but this should be easily adaptable to any data provider.

searchd {
    /* Define your file paths. */
    log = /path/to/project/sphinx/logs/searchd.log
    query_log = /path/to/project/sphinx/logs/query.log
    pid_file = /path/to/project/sphinx/logs/searchd.pid

    /* Listen on port 9312 (This is the default port) */
    listen = localhost:9312
}

      

        
        
        
      

    

This is the basic configuration of searchd, defining features such as writing logs as well as a file .pid

to lock. The search daemon listens on the port 9312

to which it sphinxapi

forwards requests by default.

source text_search{
    /* Data provider details */
    type = mysql
    sql_host = localhost
    sql_user = sql_user_name
    sql_pass = sql_pass_word
    sql_db = my_db_name     

    /* The query used to index the data. A very basic example... */
    sql_query = SELECT id, text_field, status FROM text_search
    sql_attr_uint = status
}

      

        
        
        
      

    

This is the data source. This indexer

will run the provided query and index the results. id

is taken as a key since it is an integer field and does not explicitly mention what it is (as opposed to a field status

). Since text_field

is a text field (duh!), The indexes are sphinx, which is implicit for full text search. We also specify that status

is an integer field that we can use later to filter the results when doing a search.

index text_search_index{
    /* The data source that we have defined above. */
    source = text_search

    /* The path to store the index data/cache */
    path = /path/to/project/data/text_search

    /* Use stemming while searching */
    morphology = stem_en
}

      

        
        
        
      

    

This defines the details of the index that searchd

will be used for searches. A data source is provided. Many options are available to tailor the results to your needs. I have given just one example in which we indicate that searchd

the synchronization algorithm should be used to match requests. The details of all available options can be gleaned from reading:

• Source parameters
• Index parameters
• Search options

This is in no way close to detailing, but I hope it helps you get started ...