SImplexml slow or is it WordPress update_meta?

Question

I have a script that uses simplexml_load_string to parse an 658kB XML file. The file is a property (real estate) feed with 118 different properties totalling 21000 lines. The script uses a lot of the following calls to extract the data from nodes:

(string)$properties->address->county

I'm also using Advanced Custom Fields in the script to update metadata custom fields in WordPress, lots more calls of:

update_field( 'field_59606d60525d3', (string)$properties->floorplans, $post_id );

On a Vagrant VVV box the script takes over 5 minutes to run, timing out after that. It manages to load into a custom post type about 46 out of the 118 properties What I don't know is the bottleneck. Is it:

• simplexml parsing the file?
• using update_field in ACF?

Webgrind (xdebug) appears to point to a lot of update_meta calls, but i'm not really sure what to look for and understand in a cachegrind file.

I suppose what I am asking is there a faster alternative to PHP native simpleXML and how does one interpret XDEBUG/webgrind output

This script will eventually be running on commodity hosting (no VPS/dedicated)

Skill level: procedural (functions, NOT classes)

xdebug output:

Call Stack
#   Time    Memory  Function    Location
1   0.2021  361704  {main}( )   .../test.php:0
2   0.6501  5888288 get_xml( )  .../test.php:163
3   544.3322    115472480   update_field( string(19), array(457), long )    .../test.php:115
4   544.3325    115472480   acf_update_value( array(457), long, array(24) ) .../api-template.php:1018
5   544.3325    115472536   apply_filters( string(30), array(457), long, array(24) )    .../api-value.php:350
6   544.3325    115472936   WP_Hook->apply_filters( array(457), array(3) )  .../plugin.php:203
7   544.3326    115473688   acf_field_repeater->update_value( array(457), long, array(24) ) .../class-wp-hook.php:298
8   556.4756    117433368   acf_field_repeater->update_row( array(2), long, array(24), long )   .../repeater.php:900
9   556.4756    117434744   acf_update_value( string(42), long, array(20) ) .../repeater.php:804
10  556.5003    117437600   acf_update_metadata( long, string(15), string(19), true )   .../api-value.php:368
11  556.5004    117438016   update_metadata( string(4), long, string(15), string(19), ??? ) .../api-value.php:101
12  556.5005    117438136   get_metadata( string(4), long, string(15), ??? )    .../meta.php:193
13  556.5005    117438512   update_meta_cache( string(4), array(1) )    .../meta.php:497
14  556.5124    118050992   intval ( string(3) )    .../meta.php:830

UPDATE #1 01/08/2017

I'm at a stage with this where I decided that file_get_contents might be the issue as each property in the feed has around 10 to 15 image URLs associated. 118 properties = just shy of 1800 images URL calls to make. I tried cUrl then stumbled on curl_multi.

I've now got working code below that will curl_multi on an array of image URLs, add them into WP as attachments and attach them to a specific post_id whilst updating an ACF gallery field. However I am still none the wiser on whether this is actually faster or not? How do I time something like this or work out if curl_multi is actually doing things asynchronously or if my code is correct?

require_once( '/srv/www/broadbean/wp-blog-header.php' );
require_once( '/srv/www/broadbean/wp-admin/includes/media.php' );
require_once( '/srv/www/broadbean/wp-admin/includes/file.php' );
require_once( '/srv/www/broadbean/wp-admin/includes/image.php' );

// https://stackoverflow.com/questions/15436388/download-multiple-images-from-remote-server-with-php-a-lot-of-images
// http://php.net/manual/en/function.curl-multi-init.php

$post_id = '2773';
$image_urls = array( 'http://target.domain.net/photos/1334268.jpg', 'http://target.domain.net/photos/1278564.jpg', 'http://target.domain.net/photos/1278565.jpg' );
$chs = array();
$upload_dir = wp_upload_dir();
$tc = count($image_urls);
$cmh = curl_multi_init();

for ($t = 0; $t < $tc; $t++)
{
    $chs[$t] = curl_init();
    curl_setopt($chs[$t], CURLOPT_URL, $image_urls[$t]);
    //curl_setopt($chs[$t], CURLOPT_FILE, $fp);
    curl_setopt($chs[$t], CURLOPT_FOLLOWLOCATION, 1);
    curl_setopt($chs[$t], CURLOPT_RETURNTRANSFER, 1);
    curl_setopt($chs[$t], CURLOPT_TIMEOUT, 120);
    curl_setopt($chs[$t], CURLOPT_USERAGENT, 'Mozilla/5.0');
    curl_multi_add_handle($cmh, $chs[$t]);
}

$running = null;

do {
    curl_multi_exec($cmh, $running);
} while ($running);


for ($t = 0; $t < $tc; $t++)
{
    $filename = basename( $image_urls[$t] );
    $image_file = $upload_dir['path'] . '/' . $filename;
    $fp = fopen($image_file, 'w+');

    fwrite($fp, curl_multi_getcontent( $chs[$t] ) );
    fclose($fp);

    $wp_filetype = wp_check_filetype($image_file, null );

    $attachment = array(
        'post_mime_type' => $wp_filetype['type'],
        'post_title' => sanitize_file_name( $filename),
        'post_content' => '',
        'post_status' => 'inherit'
    );


    $attach_id = wp_insert_attachment( $attachment, $image_file, $post_id );
    $attach_data = wp_generate_attachment_metadata( $attach_id, $image_file );
    $update_attach_metadata = wp_update_attachment_metadata( $attach_id, $attach_data );

    $add_gallery_images[] = $attach_id;

    curl_multi_remove_handle($cmh, $chs[$t]);
    curl_close($chs[$t]);

}
var_dump($add_gallery_images);
update_field( 'field_5973027c18fdc', $add_gallery_images , $post_id );

curl_multi_close($cmh);

Answer 1

Not really a specific answer to your problem, but since you're using Wordpress, have you tried using WP All Import? It has a good implementation for ACF too (I think u need the pro version in that case)