Parsing strange v1.1

Preview:

DESCRIPTION

WordCamp Chicago developer track talk on "URL to SQL to HTML", explaining how WordPress URL parsing and content selection works.

Citation preview

© 2010 Hal SternSome Rights Reserved

Parsing Strange:URL to SQL to HTML

Hal Sternhttp://snowmanonfire.com headshot by Richard Stevens

http://dieselsweeties.com

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 2

Why Do You Care?

• Database performance = user experience• A little database expertise goes a long way• Use taxonomies for more than sidebar lists• WordPress is a powerful CMS

>Change default behaviors>Defy the common wisdom>Integrate other content sources/filters

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 3

Disclaimers

• I’m somewhat social, for Jersey• I’m (old) old school

>If using PHP echo gives you hives……take a Benadryl now

• If “INNER JOIN” makes you giggle, you’re in the wrong session/conference/fantasy

• I suck at art and design

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 4

Flow of Control

• Web server URL manipulation>Real file or permalink URL?

• URL to query variables>What to display? Tag? Post? Category?

• Query variables to SQL generation>How exactly to get that content?

• Template file selection>How will content be displayed?

• Content manipulation

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 5

Whose File Is This?

• User URL request passed to web server• Web server checks.htaccess file>WP install root >Other .htaccess

files may interfere• Basic rewriting rules:

If file or directory URL doesn’t exist, start WordPress via index.php

<IfModule mod_rewrite.c>RewriteEngine OnRewriteBase /whereyouputWordPress/RewriteCond %{REQUEST_FILENAME} !-fRewriteCond %{REQUEST_FILENAME} !-dRewriteRule . /index.php [L]</IfModule>

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 6

What Happens Before The Loop

• Parse URL into a query• Set conditionals & select templates• Execute the query & cache results• Run the Loop:

<?phpif (have_posts()) : while (have_posts()) :

the_post(); //loop content endwhile;endif;?>

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 7

Examining the Query String

• SQL passed to MySQL in WP_Query object’s request element

• Brute force: edit theme footer.php to see main loop’s query for displayed page

<?php global $wp_query; echo ”SQL for this page "; echo $wp_query->request; echo "<br>";?>

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 8

“Home Page” Query Deconstruction

SELECT SQL_CALC_FOUND_ROWS wp_posts.* FROM wp_posts WHERE 1=1 AND wp_posts.post_type = 'post’ AND(wp_posts.post_status = 'publish' OR wp_posts.post_status = 'private’)ORDER BY wp_posts.post_date DESC LIMIT 0, 10

Get all fields from posts table, but limit number of returned rows

Only get posts, and those that are published or private to the user

Sort the results by date in descending order

Start results starting with record 0 and up to 10 more results

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 9

Query Parsing

• parse_request() method of WP_Query extracts query variables from URL

• Execute rewrite rules>Pick off ?p=67 style http GET variables>Match permalink structure>Match keywords like “author” and “tag”

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 10

Query Variables to SQL

• Query type: post by title, posts by category or tag, posts by date

• Variables for the query>Slug values for category/tags>Month/day numbers>Explicit variable values?p=67 for post_id

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 11

Simple Title Slug Parsing

• Rewrite matches root of permalink, extracts tail of URL as a title slug

SELECT wp_posts.* FROM wp_posts WHERE 1=1 AND YEAR(wp_posts.post_date)='2010' AND wp_posts.post_name = 'premio-sausage' AND wp_posts.post_type = 'post' ORDER BY wp_posts.post_date DESC

/2010/premio-sausage

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 12

Graphs and JOIN Operations

• WordPress treats tags and categories as “terms”, mapped 1:N to posts

• Relational databases aren’t ideal for this>INNER JOIN builds intermediate tables on

common key values• Following link in a social graph is

equivalent to an INNER JOIN on tables of linked items

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 13

WordPress Taxonomy Tables

• Term relationships table maps N:1 terms to each post

• Term taxonomy maps slugs 1:N to taxonomies

• Term table has slugs for URL mapping

wp_term_relationshipsobject_idterm_taxonomy_id

wp_postspost_id….post_date… post_content

wp_term_taxonomyterm_taxonomy_idterm_idtaxonomydescription

wp_termsterm_idnameslug

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 14

SELECT SQL_CALC_FOUND_ROWS wp_posts.* FROM wp_postsINNER JOIN wp_term_relationships ON(wp_posts.ID = wp_term_relationships.object_id)INNER JOIN wp_term_taxonomy ON (wp_term_relationships.term_taxonomy_id = wp_term_taxonomy.term_taxonomy_id)INNER JOIN wp_terms ON (wp_term_taxonomy.term_id = wp_terms.term_id)WHERE 1=1 AND wp_term_taxonomy.taxonomy = 'post_tag' AND wp_terms.slug IN ('premio') AND wp_posts.post_type = 'post' AND (wp_posts.post_status = 'publish' OR wp_posts.post_status = 'private') GROUP BY wp_posts.ID ORDER BY wp_posts.post_date DESC LIMIT 0, 10

Taxonomy Lookup/tag/premio

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 15

More on Canonical URLs

• Canonical URLs improve SEO• WordPress is really good about generating

301 Redirects for non-standard URLs• Example: URL doesn’t appear to match a

permalink, WordPress does prediction>Use “LIKE title%” in WHERE clause>Matches “title” as initial substring with %

wildcard

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 16

Modifying the Query

• Brute force isn’t necessarily good>Using query_posts() ignores all previous

parsing, runs a new SQL query• Filter query_vars

>Change default parsing (convert any day to a week’s worth of posts, for example)

• Actions parse_query & parse_request>Access WP_Query object before execution>is_xx() conditionals are already set

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 17

SQL Generation Filters

• posts_where>More explicit control over query variable to

SQL grammar mapping• posts_join

>Add or modify JOINS for other graph like relationships

• Many other filters>Change grouping of results>Change ordering of results

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 18

Applications

• Stylized listings>Category sorted alphabetically>Use posts as listings of resources (jobs,

clients, events)• Custom URL slugs

>Add rewrite rules to match slug and set query variables

• Joining other social graphs>Suggested/related content

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 19

Template File Selection

• is_x() conditionals set in query parsing• Used to drive template selection

>is_tag() looks for tag-slug, tag-id, then tag>Full search hierarchy in Codex

• template_redirect action>Called in the template loader>Add actions to override defaults

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 20

HTML Generation

• Done in the_post() method• Raw content retrieved from MySQL

>Short codes interpreted>CSS applied

• Some caching plugins generate and store HTML, so YMMV

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 21

Why Do You Care?

• User experience improvement>JOINS are expensive>Large table, repetitive SELECTs = slow>Running query once keeps cache warm>Category, permalink, title slug choices matter

• More CMS, less “blog”>Alphabetical sort>Adding taxonomy/social graph elements

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 22

Blunt Self-Promotion

• Brad and David are here, too

• They’re significantly more WP literate than me

• This slide was more helpful yesterday

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 23

Resources

• Core files where SQL stuff happens>query.php>post.php>canonical.php

• Template loader search path>http://codex.wordpress.org/Template_Hierarchy

© 2010 Hal SternSome Rights Reserved WordCamp Chicago 24

Contact

Hal Sternfreeholdhal@gmail.com@freeholdhalhttp://snowmanonfire.comhttp://facebook.com/hal.stern

http://slideshare.net/freeholdhal

Recommended