Upto now I have been using
WHERE col REGEXP 'IN (.*) WE TRUST'
But ever since adding the full index search to this column then this query is very slow.
I'm waning to know know how I could implement a wildcard search using full text index searches.
These are two queries that I have been playing with but still Im getting lots of results that is unexpected and not sure at all why my query is pulling those results.
WHERE MATCH (markIdentification) AGAINST ('IN (.*) WE TRUST')
WHERE MATCH (markIdentification) AGAINST ('+IN (.*) +WE +TRUST')
WHERE MATCH (markIdentification) AGAINST ('+IN * +WE +TRUST')
These are the only ones that seem to get even close.
Any suggestions?
Thank you
Update for question ref:
SELECT * from table
WHERE MATCH (col) AGAINST ('+IN * +WE +TRUST')
AND col LIKE '%IN (.*) WE TRUST%'
Is this correct?
If not then how would you do it?
The FULLTEXT search engine ignores words shorter than three characters. You can set the innodb_ft_min_token_size option to change that, then regenerate your FULLTEXT indexes.
The + (and -) syntax in AGAINST are boolean search mode things. So to use + you need
WHERE MATCH (markIdentification) AGAINST ('+IN +WE +TRUST' IN BOOLEAN MODE )
BOOLEAN mode has lots of special characters to control searches, but * standing alone is not one of them. You can say 'TRUST*' to match trust, trustee, and trusted.
Taking Gordon's suggestion, you might try this:
WHERE MATCH (markIdentification) AGAINST ('+IN +WE +TRUST' IN BOOLEAN MODE )
AND markIdentification REGEXP 'IN (.*) WE TRUST'
This will use your FULLTEXT index to look for possible matches, and REGEXP to get more exact results. The expensive REGEXP operation, then, can run on many fewer rows.
(Beware IN NATURAL LANGUAGE MODE when your tables don't have many rows. It can give strange results. The indexer decides which words are too common to bother with, and if you have a small number of words, that decision gets distorted.)
Related
I have a field called filepath that I'm trying to search. Here is an example path:
/mnt/qfs-X/Asset_Management/XG_Marketing_/Episodic-SG_1001_1233.jpg
I would like to be able to search the following and get a match:
search = "qf episodic sg_1001 JPG"
How would I do this with a fulltext search in mysql/myisam? What I have now is:
SELECT * FROM x_files2 WHERE MATCH(path)
AGAINST('qf episodic sg_1001 JPG' in boolean mode)
But it is returning way too many results (seems like it's returning if any terms are found instead of only those where all are found.
Put + in front of each 'word':
AGAINST('+qf* +episodic +sg_1001* +JPG' in boolean mode)
Do you have the min-word-length set to 2? If not, there could be other troubles.
The + avoids "too many".
Consider switching to InnoDB, now that it has FULLTEXT.
You may have to abandon use of FULLTEXT and switch to LIKE:
WHERE path LIKE '%qf%episodic%sg_1001%JPG%'
If performance is an issue, consider something like
WHERE MATCH(path) AGAINST('...' IN BOOLEAN MODE) -- using some of the words
AND path LIKE '...' -- as above
The MATCH will run first, whittling down the number of possible rows considerably, then the LIKE takes care of details.
Note that middles of words cannot be used in AGAINST. Those could be left out, relying on LIKE to take care of them.
I have a query like below:
SELECT prd_id FROM products WHERE MATCH (prd_search_field)
AGAINST ('+gul* +yetistiren* +adam*' in boolean mode);
This doesn't return the rows including 'gul'.
http://dev.mysql.com/doc/refman/5.0/en/fulltext-boolean.html
The document says this.
Then a search for '+word +the*' will likely return fewer rows than a
search for '+word +the':
The former query remains as is and requires both word and the* (a word starting with the) to be present in the document.
The latter query is transformed to +word (requiring only word to be present). the is both too short and a stopword, and either condition is enough to cause it to be ignored.
So as I understood the too short word condition must not be applied in my situation since I use * after each word. What's wrong with this?
As a solution I use the below query but since it's slow, I need to find another solution. Any idea would be appreciated? Thanks in advance..
SELECT prd_id FROM products WHERE 1 AND MATCH (prd_search_field)
AGAINST ('+yetistiren* +adam*' in boolean mode) AND prd_search_field
LIKE '%gul%';
As a note ft_min_word_length=4 as default in all shared hosting environments, and I cannot change it.
I'm pretty new to MySQL full-text searches and I ran into this problem today:
My company table has a record with "e-magazine AG" in the name column. I have a full-text index on the name column.
When I execute this query the record is not found:
SELECT id, name FROM company WHERE MATCH(name) AGAINST('+"e-magazi"*' IN BOOLEAN MODE);
I need to work with quotes because of the dash and to use the wildcard because I implement a "search as you type" functionality.
When I search for the whole term "e-magazine AG", the record is found.
Any ideas what I'm doing wrong here? I read about adding the dash to the list of word characters (config update needed) but I'm searching for a way to do this programmatically.
This clause
MATCH(name) AGAINST('+"e-magazi"*' IN BOOLEAN MODE);
Will search for a AND "e" AND NOT "magazi"; i.e. the - inside "e-magazi" will be interpreted as a not even though it is inside quotation marks.
For this reason it will not work as expected.
A solution is to apply an extra having clause with a LIKE.
I know this having is slow, but it will only be applied to the results of the match, so not too many rows should be involved.
I suggest something like:
SELECT id, name
FROM company
WHERE MATCH(name) AGAINST('magazine' IN BOOLEAN MODE)
HAVING name LIKE '%e-magazi%';
MySQL fulltext treats the word e-magazine in a text as a phrase and not as a word. Because of that it results the two words e and magazine. And while it builds the search index it does not add the e to the index because of the ft_min_word_len (default is 4 chars).
The same length limitation is used for the search query. That is the reason why a search for e-magazine returns exactly the same results as a-magazine because a and - is fully ignored.
But now you want to find the exact phrase e-magazine. By that you use the quotes and that is the complete correct way to find phrases, but MySQL does not support operators for phrases, only for words:
https://dev.mysql.com/doc/refman/5.7/en/fulltext-boolean.html
With this modifier, certain characters have special meaning at the beginning or end of words in the search string
Some people would suggest to use the following query:
SELECT id, name
FROM company
WHERE MATCH(name) AGAINST('e-magazi*' IN BOOLEAN MODE)
HAVING name LIKE 'e-magazi%';
As I said MySQL ignores the e- and searches for the wildcard word magazi*. After those results are optained it uses HAVING to aditionally filter the results for e-magazi* including the e-. By that you will find the phrase e-magazine AG. Of course HAVING is only needed if the search phrase contains the wildcard operator and you should never use quotes. This operator is used by your user and not you!
Note: As long you do not surround the search phrase with % it will find only fields that start with that word. And you do not want to surround it, because it would find bee-magazine as well. So maybe you need an additional OR HAVING name LIKE ' %e-magazi%' OR HAVING NAME LIKE '\\n%e-magazi%' to make it usable inside of texts.
Trick
But finally I prefer a trick so HAVING isn't needed at all:
If you add texts to your database table, add them additionally to a separate fulltext indexed column and replace words like up-to-date with up-to-date uptodate.
If a user searches for up-to-date replace it in the query with uptodate.
By that you can still find specific in user-specific but up-to-date as well (and not only date).
Bonus
If a user searches for -well-known huge ports MySQL treats that as not include *well*, could include *known* and *huge*. Of course you could solve that with an other extra query variant as well, but with the trick above you remove the hyphen so the search query looks simply like that:
SELECT id
FROM texts
WHERE MATCH(text) AGAINST('-wellknown huge ports' IN BOOLEAN MODE)
I am trying to follow: http://dev.mysql.com/doc/refman/4.1/en/fulltext-natural-language.html
in an attempt to improve search queries, both in speed and the ability to order by score.
However when using this SQL ("skitt" is used as a search term just so I can try match Skittles).
SELECT
id,name,description,price,image,
MATCH (name,description)
AGAINST ('skitt')
AS score
FROM
products
WHERE
MATCH (name,description)
AGAINST ('skitt')
it returns 0 results. I am trying to find out why, I think I might have set my index's up wrong I'm not sure, this is the first time I've strayed away from LIKE!
Here is my table structure and data:
Thank you!
By default certain words are excluded from the search. These are called stopwords. "a" is an example of a stopword. You could test your query by using a word that is not a stopword, or you can disable stopwords:
How can I write full search index query which will not consider any stopwords?
If you want to also match prefixes use the truncation operator in boolean mode:
*
The asterisk serves as the truncation (or wildcard) operator. Unlike the other operators, it should be appended to the word to be affected. Words match if they begin with the word preceding the * operator.
I'm trying to use wildcards to pass a stem of a word as part of a full text search in MySql. I would prefer to use match...against for the performance benefit instead of a like query.
I found this post which makes it sound as though this can be done:
MySQL fulltext with stems
...but I can't get it to work for me.
My data looks like this:
table name: "rxnorm_brands"
step_medname bn_name
Amoxicillin Wymox
Amoxicillin-Clavulanate Augmentin
This query works but uses "like":
select `step_medname`, `bn_name`
from `rxnorm_brands`
where (`bn_name` like 'Amox%' or `step_medname` like 'Amox%');
I want to use this query, but it returns nothing:
select `step_medname`, `bn_name`
from `rxnorm_brands`
where MATCH (`bn_name`, `step_medname`) AGAINST ('Amox*');
I do have a fulltext index on bn_name and step_medname. What am I doing wrong? Or can this not be done?
This can be done using IN BOOLEAN MODE, see: http://dev.mysql.com/doc/refman/5.1/en/fulltext-boolean.html.
So your query would become:
select `step_medname`, `bn_name`
from `rxnorm_brands`
where MATCH (`bn_name`, `step_medname`) AGAINST ('Amox*' IN BOOLEAN MODE);
but note that with the BOOLEAN MODE matching, rows either match or they don't - the results can no longer be ordered by relevance like they can with normal FULLTEXT searches.
The RxNorm API now has a method that will do matching of text that only approximately matches the RxNorm concept. See http://rxnav.nlm.nih.gov/RxNormAPI.html#label:23