Creating an Index

You can create a GIN index to speed up text searches:

mogdb=# CREATE INDEX pgweb_idx_1 ON tsearch.pgweb USING gin(to_tsvector('english', body));

The to_tsvector function comes in to two versions: the 1-argument version and the 2-argument version. When the 1-argument version is used, the system uses the configuration specified by default_text_search_config by default.

Notice that the 2-argument version of to_tsvector is used for index creation. Only text search functions that specify a configuration name can be used in expression indexes. This is because the index contents must be unaffected by default_text_search_config, whose value can be changed at any time. If they were affected, the index contents might be inconsistent, because different entries could contain tsvectors that were created with different text search configurations, and there would be no way to guess which was which. It would be impossible to dump and restore such an index correctly.

Because the two-argument version of to_tsvector was used in the index above, only a query reference that uses the 2-argument version of to_tsvector with the same configuration name will use that index. That is, WHERE to_tsvector('english', body) @@ 'a & b' can use the index, but WHERE to_tsvector(body) @@ 'a & b' cannot. This ensures that an index will be used only with the same configuration used to create the index entries.

It is possible to set up more complex expression indexes wherein the configuration name is specified by another column. For example:

mogdb=# CREATE INDEX pgweb_idx_2 ON tsearch.pgweb USING gin(to_tsvector('ngram', body));

where body is a column in the pgweb table. This allows mixed configurations in the same index while recording which configuration was used for each index entry. This would be useful, for example, if the document collection contained documents in different languages. Again, queries that are meant to use the index must be phrased to match, for example, WHERE to_tsvector(config_name, body) @@ 'a & b' must match to_tsvector in the index.

Indexes can even concatenate columns:

mogdb=# CREATE INDEX pgweb_idx_3 ON tsearch.pgweb USING gin(to_tsvector('english', title || ' ' || body));

Another approach is to create a separate tsvector column to hold the output of to_tsvector. This example is a concatenation of title and body, using coalesce to ensure that one column will still be indexed when the other is NULL:

mogdb=# ALTER TABLE tsearch.pgweb ADD COLUMN textsearchable_index_col tsvector;
mogdb=# UPDATE tsearch.pgweb SET textsearchable_index_col = to_tsvector('english', coalesce(title,'') || ' ' || coalesce(body,''));

Then, create a GIN index to speed up the search:

mogdb=# CREATE INDEX textsearch_idx_4 ON tsearch.pgweb USING gin(textsearchable_index_col);

Now you are ready to perform a fast full text search:

mogdb=# SELECT title
FROM tsearch.pgweb
WHERE textsearchable_index_col @@ to_tsquery('north & america')
ORDER BY last_mod_date DESC
LIMIT 10;

 title
--------
 Canada
 Mexico
(2 rows)

One advantage of the separate-column approach over an expression index is that it is unnecessary to explicitly specify the text search configuration in queries in order to use the index. As shown in the preceding example, the query can depend on default_text_search_config. Another advantage is that searches will be faster, since it will not be necessary to redo the to_tsvector calls to verify index matches. The expression-index approach is simpler to set up, however, and it requires less disk space since the tsvector representation is not stored explicitly.