MySQL Comparative querying longtext columns

Question

I have a LONGTEXT column that I use to store paragraphs of text. Let's say that within these paragraphs there is something like:

Text text text text
COUNT: 75  text text
 text text text text
 text text text text

Would it be possible to do a comparative query the small string "COUNT: 75" out of all that text?

My first thought was something like

SELECT * FROM `books`
WHERE `paragraphs` LIKE '%COUNT: >0%'

Is this even possible?

Answer 1

Your SELECT will only find rows where the text contains exaclty the bit between the wildcard characters: you can't combine a LIKE with comparative logic like that.

What you can do, though, is to strip out the relevant sections of text using a regular expression and then analyse that.

Bear in mind, though, that combining

• large amounts of text
• textual content logic
• regex

all at once will not provide the best performance! I would suggest the following:

1. use a trigger to strip out a subsection of text so that you have something manageable (i.e. 50 characters or so) to work with, inserting this subtext into a separate table
2. use MySql regex or fulltext functions to analyse your COUNTs

So your trigger would have something like:

select 
ltrim(rtrim(substring(paragraphs, instr(paragraphs, 'count:') + 6, 10)))
from books 
where instr(paragraphs, 'count:') > 0

which would get you the next 10 characters after 'count:', trimmed of whitespace. you could then further refine that by e.g.

select substring(text_snippet, 1, instr(text_snippet, ' ')) as count_value
from
(
    select 
    ltrim(rtrim(substring(paragraphs, instr(paragraphs, 'count:') + 6, 10))) 
          as text_snippet
    from books 
    where instr(paragraphs, 'count:') > 0
) x
where isnumeric(substring(text_snippet, 1, instr(text_snippet, ' '))) = 1

to get rows where a numerical value follows the COUNT bit.

You can then extract numerical values next to COUNT, saving them as numbers in a separate table, and then use a JOIN like this:

select b.*
from books b inner join books_count_values v
on b.id = v.books_id
where v.count_value > 0

See here

http://dev.mysql.com/doc/refman/5.1/en/string-functions.html

for the functions at your disposal.