WPDB code improvements

[hakre]

While taking care about various places over the past weeks I somehow returned back more and more often to WPDB starting to beautify and refactor the code in there.

I create this ticket now to collect changes on wpdb to improve the code.

Related:

• #6836 - Prepare queries using INSERT/UPDATE should use wpdb::insert() or wpdb::update()
• #7171 - wpdb::insert() & wpdb::update() need to allow typecasting of params.
• #11318 - $wpdb->prepare() Chokes on single literal percentage chars?
• #11372 - wpdb::query returns FALSE on TRUNCATE
• #11605 - wpdb::_weak_escape() is an alias to addslashes only
• #11608 - wpdb->prepare() is broken
• #11649 - Upgrades Fail Because of [11883]

The two most important things while I rewrite some parts of WPDB is code duplication and massive memory leaks in that class.

[hakre]

Bonus refactoring - Orphaned code removed, Docblock update, propper returns in mysql2date()

[hakre]

Todays Dev-Meeting feedback: wpmuguru merges in wpdb from WPMU the next day(s). I'll add my patch here when I at least get a first version on.

[hakre]

Finally I'm able to upload a first patch. I did not implement all of the changes I suggested in other patches around so that the changes are not so immersive. The focus is more on the general memory problems WPDB is having and the code quality. I was able to remove duplicate code blocks and to reduce complexity of many functions. The readablity did get a benefit as well. Docblocks have been extended / created and I tried to adopt the code to the coding standards.

Under the hood the class internally now uses arrays to store the resultset. That will lower the required memory. From the outside there is no difference, every public member can be used as it could before.

When was it introduced (compiled this for some @since tags )?

• num_rows - In [1180].
• suppress_errors - In [7269].
• last_error - In [7431].

While refactoring, I noticed the following things:

• #11804 - Object Access Violation in repair.php
• #11819 - mysql_real_escape_string available now / PHP 4.3 are minimum system requirements since 2.9

[hakre]

First one; Fix: Query $type was not case insensitive. Fixed.

[Denis-de-Bernardy]

this is a bit odd:

• initialise return

+ initi return values and objects state

[Denis-de-Bernardy]

isn't this missing a delimiter?

if ( preg_match( '(^\s*([a-z]{2,})\s+)i', $query, $matches ) ) 

also, I don't think it'll properly deal with:

query("
# some comments for debugging using SAVEQUERIES
INSERT INTO ...
");

but nor does the current version of the class, so that's probably another ticket...

also while we're at it, for reference, the code in wpmu does not deal with a tabbed update statements properly. (but yours does, since it follows the UPDATE with a \s instead of a space in the regexp.)

[hakre]

Replying to Denis-de-Bernardy:

this is a bit odd: + initi return values and objects state ...

Fixed.

Replying to Denis-de-Bernardy:

isn't this missing a delimiter?

if ( preg_match( '(^\s*([a-z]{2,})\s+)i', $query, $matches ) ) 

No, it's not, delimitter is () and it's valid and nice as well. matches for the group 0 to give a hint :).

also, I don't think it'll properly deal with:

query("
# some comments for debugging using SAVEQUERIES
INSERT INTO ...
");

but nor does the current version of the class, so that's probably another ticket...

jup, would say other dimension, this requires a new algo to copy with that. maybe mysql itself can tell the name of the last command. that could solve it.

also while we're at it, for reference, the code in wpmu does not deal with a tabbed update statements properly. (but yours does, since it follows the UPDATE with a \s instead of a space in the regexp.)

I think that's a valuable hint for wpmuguru.

[hakre]

I added another iteration of the patch. Now all functions and vars should have a properly @since tag. corrected some whitespace issues compared to the previous patch and fixed that comment.

[hakre]

First one - rev 4; Fix: get_var / hakre; Fix: Comment / Denis; Fix: Query $type was not case insensitive. Fixed.

[hakre]

I updated the patch again because it had a bug in the get_var subroutine.

Related: #11921

[ryan]

Patch breaks multisite and needs refresh.

[hakre]

Referenced: #10607

[hakre]

Patch is stale and does not support multisite. I now merged my changes against the current trunk (which is multisite) so there still is some benefit of the work. I've attached it to #11644.

I'll put this ticket on future release because right now it's more important to get the multisite feature done.

[hakre]

Related: #12257

[nacin]

​http://wpdevel.wordpress.com/2011/03/23/code-refactoring/

This ticket is horribly stale. Closing.