Test <title> and <meta type="description"> tags for SEO purposes against a CSV or regexp

test-seo.php

<?php

if (count($argv) != 3) {
    print("Usage: $argv[0] [domain] [csv file]\n\n");
    exit(0);
}

$domain       = $argv[1];
$csv_filename = $argv[2];

if (substr($domain, 0, 4) != 'http') {
    print("[domain] must be http url, for example: 'https://google.com'\n");
    exit(0);
}

if (!file_exists($csv_filename)) {
    print("Can't find file $csv_filename!\n");
    exit(0);
}

#
# CSV Format:
#
#  Url,Title,Description
#  [relataive path],[title_regex],[desc_regex]
#  ....
#

$csv = file_get_contents($csv_filename);
$lines = explode("\n", $csv);
$test_cases = [];
foreach($lines as $line) {
    $fields = str_getcsv($line);
    if (count($fields) == 3) {
        $test_cases[] = $fields;
    }
}

foreach($test_cases as $test_case) {
    $url = trim($test_case[0]);
    $title_regex = isset($test_case[1]) ? trim($test_case[1]) : '';
    $desc_regex  = isset($test_case[2]) ? trim($test_case[2]) : '';

    // header!
    if ($url == 'Url') {
        continue;
    }

    test_page($domain . $url, $title_regex, $desc_regex);
}

function test_page($url, $title_regex, $desc_regex) {

    libxml_use_internal_errors(true);
    $doc = new DOMDocument();
    $doc->loadHTMLFile($url);

    $title_value = get_title_tag_value($doc);

    $metadata = get_meta_tags($url);
    $desc_value  = $metadata['description'] ? $metadata['description'] : '';

    // Log 
    log_info();
    log_info("Testing $url ...");

    $failed = FALSE;
    // Test emmpty title
    if (strlen(trim($title_value)) == 0) {
        log_failure("<title> tag empty");
        $failed = TRUE;
    }

    // Compare with regex:
    if (preg_match("/$title_regex/", $title_value) != 1) {
        log_failure("<title> doesn't match. Expecting [$title_regex] found [$title_value]");
        $failed = TRUE;
    }
    else {
        log_success("<title> matches value: [$title_value]");
    }

    // Test emmpty description
    if (strlen(trim($desc_value)) == 0) {
        log_failure("description meta-tag empty.");
        $failed = TRUE;
    }

    // Compare with regex:
    if (preg_match("/$desc_regex/", $desc_value) != 1) {
        log_failure("description meta-tag data doesn't match. Expecting [$desc_regex] found [$desc_value]");
        $failed = TRUE;
    }
    else {
        log_success("description meta-tag matches value: [$desc_value]");
    }

    return $failed;
}

function get_title_tag_value($doc) {
    $count = 0;
    foreach($doc->getElementsByTagName('title') as $title){
        $title_value = $title->nodeValue;
        $count++;
    }

    if ($count != 1) {
        return FALSE;
    }

    return $title_value;
}

function log_info($message = NULL) {
    if ($message === NULL) {
	print("\n");
	return;
    }
    print(">> \033[34m" . $message . "\033[0m\n" );
}

function log_success($message) {
    print(" [✔] " . $message . "\n" );
}

function log_failure($message) {
    print(" [✖] " . $message . "\n" );
}