PostgreSQL Reverse LIKE

Question

I need to test if any part of a column value is in a given string, instead of whether the string is part of a column value. For instance:

This way, I can find if any

Answer 1

This query:

SELECT 
regexp_split_to_table(
  'The ships hung in the sky in much the same way that bricks don’t', 
  '\s' );

gives a following result:

| regexp_split_to_table |
|-----------------------|
|                   The |
|                 ships |
|                  hung |
|                    in |
|                   the |
|                   sky |
|                    in |
|                  much |
|                   the |
|                  same |
|                   way |
|                  that |
|                bricks |
|                 don’t |

Now just do a semijoin against a result of this query to get desired results

SELECT * FROM table t
WHERE EXISTS (
   SELECT * FROM (
      SELECT 
    regexp_split_to_table(
      'The ships hung in the sky in much the same way that bricks don’t', 
      '\s' ) x
   ) x
   WHERE t.column LIKE '%'|| x.x || '%'
)  

Answer 2

Your simple case can be solved with a simple query using the ANY construct and ~*:

SELECT *
FROM   tbl
WHERE  col ~* ANY (string_to_array('The ships hung in the sky ... bricks don’t', ' '));

~* is the case insensitive regular expression match operator. I use that instead of ILIKE so we can use original words in your string without the need to pad % for ILIKE. The result is the same - except for words containing special characters: %_\ for ILIKE and !$()*+.:<=>?[\]^{|}- for regular expression patterns. You may need to escape special characters either way to avoid surprises. Here is a function for regular expressions:

• Escape function for regular expression or LIKE patterns

But I have nagging doubts that will be all you need. See my comment. I suspect you need Full Text Search with a matching dictionary for your natural language to provide useful word stemming ...

Related:

• IN vs ANY operator in PostgreSQL
• PostgreSQL LIKE query performance variations
• Pattern matching with LIKE, SIMILAR TO or regular expressions in PostgreSQL