问题
Mediawiki has a table in the database 'text' which contains the page content. It is saved as a [BLOB] file. I would like to run a query to search through all the text on the site to see which pages contain a certain 'string'. How do I run a query to search [blob] files?
回答1:
The Mediawiki markup text is stored in the old_text field, which is a mediumblob type. You can query it like any other text-based field. MySQL will cast your string into binary for the query. Note that this is a case-sensitive search!
select old_id from text where old_text like "%string%";
If you need case-insensitivity then you need to apply an appropriate character set with a case-insensitive collation to the column:
SELECT old_id from text where CONVERT(old_text USING latin1) like '%STRing%';
Be aware that if your table isn't small these queries will take a long time.
回答2:
As per the mediawiki documentation text table stores only the text for the revision. Hence to access the complete text, all revisions corresponding to a page need to be processed. It is better to use an API call to mediawiki search engine and process the results than search using SQL query.
来源：https://stackoverflow.com/questions/19563838/how-to-run-a-query-to-find-a-string-in-blob-files