Joomla: instantiate ContentModelArticle really slow - php

Question

I have a json file of about 50M in size, about 10k records. I am using following code to insert into Joomla programmatically, it works pretty well as all relevant tables get updated simultaneously, e.g., #_assets, however the code runs really slow: about 40 hours passed now but only about 4k articles are processed. Can I speed up the process somehow?
I do know that if I do insert into #_content, #_tags, etc, it'll be much quicker, but that approach has some trivia complications that I try to avoid.
<?php
define( '_JEXEC', 1 );
define('JPATH_BASE', dirname(dirname(__FILE__)));
define( 'DS', DIRECTORY_SEPARATOR );
require_once (JPATH_BASE . DS . 'includes' . DS . 'defines.php');
require_once (JPATH_BASE . DS . 'includes' . DS . 'framework.php');
require_once (JPATH_BASE . DS . 'libraries' . DS . 'joomla' . DS . 'factory.php' );
define('JPATH_COMPONENT_ADMINISTRATOR', JPATH_BASE . DS . 'administrator' . DS . 'components' . DS . 'com_content');
$mainframe = JFactory::getApplication('site');
require_once (JPATH_ADMINISTRATOR.'/components/com_content/models/article.php');
$string = file_get_contents("items.json");
$json_str = json_decode($string, true);
foreach($json_str as $row){
$new_article = new ContentModelArticle();
$data = array(
'title' => $row['title'][0],
'alias' => $row['alias'],
'introtext' => $row['content'],
'state' => 1,
'catid' => 8, /* don't hard code here! */
'created' => $row['pdate'][0],
'created_by' => 798,
'created_by_alias' => $row['poster'][0],
'publish_up' => $row['pdate'][0],
'urls' => $row['urls'],
'access' => 1,
'metadata' => array(
'tags' => $row['tags'],
'robots' => "",
'author' => implode(" ", $row['poster']),
'rights' => "",
'xreference' => "",
),
);
$new_article->save($data);
}

Answer 1

If you were running this in Joomla as an extension I would suggest turning on debugging and having a look at the Profile Information in the Joomla! Debug Console to see where allthe time is being spent.
You're not, so the rest is deduction (therefore possibly faulty).
You're loading the ContentModelArticle for each insert this means that it creates and destroys not only the model class for each row but also the table class for each row.
If these are all inserts, then you don't need all the overhead of the model (e.g. checking if the article exists first).
If the data is all valid you also don't need the data preparation (prepareTable()) in the content model nor the check() in the content table.
If you don't need that you could load the JTable once, then call the table objects save() passing in your array for each article. That will run the bind(), check(), store() — where asset tracking is updated, and complete the insert.
So, you could try something like this, I think it will be faster, not sure by how much.
$contentTable = JTable::getInstance('Content', 'JTable', array());
foreach($json_str as $row){
$data = array(
'title' => $row['title'][0],
'alias' => $row['alias'],
'introtext' => $row['content'],
'state' => 1,
'catid' => 8, /* don't hard code here! */
'created' => $row['pdate'][0],
'created_by' => 798,
'created_by_alias' => $row['poster'][0],
'publish_up' => $row['pdate'][0],
'urls' => $row['urls'],
'access' => 1,
'metadata' => array(
'tags' => $row['tags'],
'robots' => "",
'author' => implode(" ", $row['poster']),
'rights' => "",
'xreference' => "",
),
$contentTable->save($data);
}
*Disclaimer: this isn't tested code it's just out of my head and into the browser.