Skip to content

Instantly share code, notes, and snippets.

What would you like to do?
tables.sql for using sqlite with WikiDB Mediawiki extension
-- ---------------------------------------------------------------------------------
-- $Rev: 776 $
-- ---------------------------------------------------------------------------------
-- You are strongly recommended to use the supplied install/update scripts rather
-- than running this code directly.
-- /*$wgDBprefix*/ should be replaced with the DB prefix you are using for
-- your MediaWiki installation (if relevant).
-- This will be done automatically if you use the install/update scripts.
-- /*$wgDBTableOptions*/ is used in later versions of MediaWiki to set the
-- default table options, and will vary depending on the type of DB being
-- used (mysql/postgres/etc).
-- Currently these just set the type of storage engine for the table, and
-- this doesn't really matter from the point of view of the extension code (though
-- it may make a difference to the speed and/or storage space requirements).
-- In earlier versions of MediaWiki this setting is not present, meaning that
-- the default table type will be used, as per the way the DB engine has been
-- configured.
-- In earlier versions of WikiDB (before the SQL code and maintenance scripts were
-- included as part of the extension, and were just available on-wiki) I recommended
-- MyISAM table type, and now it will default to InnoDB. However this change is not
-- considered relevant, and so is not checked for by the install script. If this
-- distinction is important (and I am no DB expert!), somebody should let me know
-- and I'll update those scripts to test and fix the table type if necessary.
-- ---------------------------------------------------------------------------------
-- Table structure for table `wikidb_tables`.
-- This table holds the field definitions for all pages located in any
-- of the defined table namespaces. There is one row per page.
CREATE TABLE IF NOT EXISTS /*$wgDBprefix*/wikidb_tables (
-- ----- COLUMNS -----
-- These two fields point to the page on which the table is defined.
-- This will be a page in one of the table namespaces, as described in
-- $wgWikiDBNamespaces.
-- These (as a pair) are a foreign key into the standard MediaWiki `page` table,
-- though we don't actually create any JOINs between the two.
table_namespace int NOT NULL,
table_title varchar(255) binary NOT NULL,
-- This is the table definition itself. It is a serialized array containing the
-- result of parsing the page for field definitions.
table_def mediumblob NOT NULL default '',
-- If this page is a redirect to a page within a table namespace then we store
-- the details about the redirect here. It would be far too expensive to have to
-- process the page text when trying to find all pages that redirect to a specific
-- table (particularly as we need to follow chains of redirects) therefore we
-- store these two fields which point to the destination of the redirect so we can
-- use it as a quick look-up for this information.
-- We only log details about redirects which point to another page in a table
-- namespace. Pages pointing to non-table pages do not have their redirect
-- stored here as they are not relevant to the extension.
-- Note that a `redirect` table was added to MediaWiki in version 1.9, which could
-- be used instead of these fields, but for compatibility with earlier versions
-- we implement our own redirect cache instead.
-- These (as a pair) are a foreign key into the standard MediaWiki `page` table,
-- though we don't actually create any JOINs between the two.
redirect_namespace int DEFAULT NULL,
redirect_title varchar(255) binary DEFAULT NULL,
-- ----- INDEXES -----
-- Each page will have at most one row in this table, so we set this as
-- the primary key. We nearly always pull the data out of this table using an
-- exact match on these two fields, so it is important that they are indexed.
PRIMARY KEY (table_namespace,table_title)
-- We will also need to perform lookups on the destination page (to get pages that
-- point to it), so these need to be indexed too.
-- INDEX redirect_ns_title (redirect_namespace,redirect_title)
) /*$wgDBTableOptions*/;
CREATE INDEX IF NOT EXISTS redirect_ns_title ON /*$wgDBprefix*/wikidb_tables (table_namespace,table_title);
-- ---------------------------------------------------------------------------------
-- Table structure for table `wikidb_rowdata`.
-- This table holds one record per row of data. Data may defined on any page
-- on the wiki, and multiple data rows may be defined on a single page.
CREATE TABLE IF NOT EXISTS /*$wgDBprefix*/wikidb_rowdata (
-- ----- COLUMNS -----
-- An auto-generated unique RowID, used as the primary key.
row_id int(8) unsigned NOT NULL PRIMARY KEY auto_increment,
-- These two fields point to the page on which the data was defined. This page
-- should always exist on the wiki.
page_namespace int NOT NULL,
page_title varchar(255) binary NOT NULL,
-- These two fields point to the page in one of the table namespaces which
-- contains the definition. This page may or may not actually exist.
table_namespace int NOT NULL,
table_title varchar(255) binary NOT NULL,
-- This is the parsed data for the row, stored as a serialized PHP array.
-- It consists of a set of key => value pairs, where key is the field name
-- (normalised) and value is the data for the field (as entered).
parsed_data mediumblob NOT NULL default '',
-- This flag is set to true whenever the table definition for this row is changed.
-- The table definition is correctly resolved across table aliases, and if any
-- of the tables in the alias chain are modified, then this flag is set.
-- The system then works through the stale records in batches (a certain number per
-- page request) to avoid excessive load when a definition changes.
-- This is required because we format wikidb_fielddata.field_value according to the
-- current definition so that we can do efficient searches/sorting within MySQL, so
-- these fields will need updating whenever the field definition changes (as this
-- affects the sort/search order.
is_stale tinyint(1) unsigned NOT NULL default '0'
-- ----- INDEXES -----
-- Set the primary key to the unique row_id.
-- We most commonly pull data out of this table by performing an exact match on the
-- table that the data is from, so it is very important that this pair of fields
-- is indexed properly.
-- INDEX table_ns_title (table_namespace,table_title),
-- We only really use the source article fields as criteria when running the
-- DELETE queries after a page is updated (we delete all existing rows for that
-- page, re-parse the page and re-populate the table with any data records found).
-- This pair of fields is therefore still worth indexing, but less critical than
-- the above.
-- INDEX ns_title (page_namespace,page_title),
-- I don't know how useful it is to index on a boolean field, but we will be
-- performing lookups on this, so an index will hopefully speed things up a little
-- bit, at least.
-- INDEX is_stale (is_stale)
) /*$wgDBTableOptions*/;
CREATE INDEX IF NOT EXISTS is_stale ON /*$wgDBprefix*/wikidb_rowdata (is_stale);
CREATE INDEX IF NOT EXISTS table_ns_title ON /*$wgDBprefix*/wikidb_rowdata (table_namespace,table_title);
CREATE INDEX IF NOT EXISTS ns_title ON /*$wgDBprefix*/wikidb_rowdata (page_namespace,page_title);
-- ---------------------------------------------------------------------------------
-- Table structure for table `wikidb_fielddata`
-- This table holds one row per data field for each row of data. It is used
-- to create the temporary tables used for running queries on the WikiDB data.
-- The data here is also held in the parsed_data field of the appropriate row
-- in wikidb_rowdata, in the form of a serialized PHP array.
CREATE TABLE IF NOT EXISTS /*$wgDBprefix*/wikidb_fielddata (
-- ----- COLUMNS -----
-- Two fields to hold the table that the row belongs in. This is de-normalisation
-- of the data held in wikidb_rowdata.
-- Originally they were used for optimising the search, but actually it turns out
-- this is not required. I am not going to remove them for the time being, as it
-- may turn out that there is a use for them after all, and it will make the update
-- process tricky if we're removing and re-adding them.
table_namespace int NOT NULL,
table_title varchar(255) binary NOT NULL,
-- Foreign key to wikidb_rowdata.row_id, indicating the row of data to which
-- this field belongs.
row_id int(8) unsigned NOT NULL,
-- Field name - normalised from the original data entered on the wiki.
field_name varchar(255) NOT NULL default '',
-- Field value - as entered.
field_value varchar(255) NOT NULL default ''
-- ----- INDEXES -----
-- Create some indexes to speed up queries.
-- Index on row_id and the namespace/title pair, as these are used frequently when
-- querying/modifying the table.
-- INDEX row_id (row_id),
-- INDEX ns_title (table_namespace,table_title),
-- Create an index on the name/value pairs as these will be used to when creating
-- the temporary tables.
-- TODO: Actually, I'm not sure about this - this comes from my local DB but I can't
-- remember setting it up, so can't comment on how correct/useful it actually
-- is - to be investigated.
-- INDEX field_name (field_name(75),field_value(75))
) /*$wgDBTableOptions*/;
CREATE INDEX IF NOT EXISTS row_id ON /*$wgDBprefix*/wikidb_fielddata (row_id);
CREATE INDEX IF NOT EXISTS ns_title ON /*$wgDBprefix*/wikidb_fielddata (table_namespace,table_title);
CREATE INDEX IF NOT EXISTS field_name ON /*$wgDBprefix*/wikidb_fielddata (field_name(75),field_value(75));
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment