Custom functions used for WP All Import of articles

functions.php

<?php
function article_get_header_regex( $level ) {
	$level = (int) $level;
	return '/^\<h' . $level . '[^\>]*\>(.*)\<\/?h' . $level . '\>/i';
}

function article_get_end_of_url( $url ) {

	// Remove trailing "/".
	$last_chr = substr( $url, -1 );
	if ( '/' == $last_chr ) {
		$url = substr( $url, 0, -1 );
	}
	
	// Get last "part" of URL.
	preg_match( '/([^\/]+)$/i', $url, $matches );

	if ( ! empty( $matches[1] ) ) {
		return trim( $matches[1] );
	}

	return null;
}

function article_get_subheading( $content ) {
	
	// Remove title header.
	$content = preg_replace( article_get_header_regex( 3 ), '', $content );
	
	// Get subheading matches
	preg_match( article_get_header_regex( 4 ), $content, $matches );
	
	if ( ! empty( $matches[1] ) ) {
		return trim( strip_tags( $matches[1] ) );
	}

	return null;
}

function update_image_src( $content ) {
	
	preg_match_all( '/\<img[^\>]+src\=[\"\']+(\/portals[^\'\"]+)[\"\']+[^\/\>]*\/?\>/i', $content, $all_matches );
	
	if ( empty( $all_matches[1] ) ) {
		return $content;
	}

	$matches = $all_matches[1];

	// Trying to keep the work down on the import.
	if ( count( $matches ) > 2 ) {
		return $content;
	}

	$blog_url = '';

	// Build replacements.
	$replacements = [];
	foreach( $matches as $match ) {
		$replacements[] = $blog_url . $match;
	}

	$content = str_replace( $matches, $replacements, $content );

	return $content;
}

function article_optimize_content( $content ) {

	// Remove title header.
	$content = preg_replace( article_get_header_regex( 3 ), '', $content );
	
	// Remove subheading.
	$content = preg_replace( article_get_header_regex( 4 ), '', $content );
	
	// Update paths to images.
	if ( false !== strpos( $content, '/portals/' ) ) {
		$content = update_image_src( $content );
	}

	return trim( $content );
}