So I've done a bunch of Doctrine2 migrations (https://github.com/doctrine/migrations) but I have a question for a new migration I'm trying to do.
I've been digging into the library a little and I see that $this->addSql()
is used to build a list of SQL to execute and then it gets executed later.
I wanted to do something where I select some data, iterate over the rows, insert new data based on that, and then delete the data I selected. This lends itself to the DBAL library pretty easily, but I'm wondering, can I use the protected $connection
in a migration safely? Or is that bad because it would execute statements before any of my $this->addSql()
SQL gets executed? Also it seems like this would break the dry-run
setting from what I've seen in the code. Has anyone had any experience with this type of migration? Are there any best practices?
The following is the migration I want to do, but I'm not confident that this is supported by Doctrine Migrations:
public function up(Schema $schema)
{
// this up() migration is autogenerated, please modify it to your needs
$this->abortIf($this->connection->getDatabasePlatform()->getName() != "mysql");
$this->addSql("ALTER TABLE article_enclosures ADD is_scrape TINYINT(1) NOT NULL");
$this->addSql("ALTER TABLE images DROP FOREIGN KEY FK_E01FBE6AA536AAC7");
// now lets take all images with a scrape and convert the scrape to an enclosure
//
// Select all images where not scrape_id is null (join on article_image_scrape)
// for each image:
// insert into article_enclosures
// update image set enclosure_id = new ID
// delete from article_image_scrape where id...
//
// insert into article_enclosures select article_image_scrapes...
$sql = "SELECT i.id img_id, e.* FROM images i JOIN article_image_scrapes e ON i.scrape_id = e.id";
$stmt = $this->connection->prepare($sql);
$stmt->execute();
$scrapesToDelete = array();
while ($row = $stmt->fetch()) {
$scrapeArticle = $row['article_id'];
$scrapeOldId = $row['id'];
$scrapeUrl = $row['url'];
$scrapeExtension = $row['extension'];
$scrapeUrlHash = $row['url_hash'];
$imageId = $row['image_id'];
$this->connection->insert('article_enclosures', array(
'url' => $scrapeUrl,
'extension' => $scrapeExtension,
'url_hash' => $scrapeUrlHash
));
$scrapeNewId = $this->connection->lastInsertId();
$this->connection->update('images', array(
'enclosure_id' => $scrapeNewId,
'scrape_id' => null
), array(
'id' => $imageId
));
$scrapesToDelete[] = $scrapeOldId;
}
foreach ($scrapesToDelete as $id) {
$this->connection->delete('article_image_scrapes', array('id' => $id));
}
$this->addSql("INSERT INTO article_scrapes (article_id, url, extension, url_hash) "
."SELECT s.id, s.url, s.extension, s.url_hash"
."FROM article_image_scrapes s");
$this->addSql("DROP INDEX IDX_E01FBE6AA536AAC7 ON images");
$this->addSql("ALTER TABLE images DROP scrape_id, CHANGE enclosure_id enclosure_id INT NOT NULL");
}
You can use the $connection
like this
$result = $this->connection->fetchAssoc('SELECT id, name FROM table1 WHERE id = 1');
$this->abortIf(!$result, 'row with id not found');
$this->abortIf($result['name'] != 'jo', 'id 1 is not jo');
// etc..
You should only read the database and not use the connection to make update/delete so it won't break the dry-run option.
In your example, you should do two migrations. The first will do the two alter table. The second will do the "images with a scrape and convert the scrape to an enclosure" routine. Using multiple migration is easier to revert them if something goes wrong.