Question

I’m using a PHP script (using cURL) to check whether:

• The links in my database are correct (ie return HTTP status 200)
• The links are in fact redirected and redirect to an appropriate/similar page (using the contents of the page )

The results of this are saved to a log file and emailed to me as an attachment.

This is all fine and working, however it is slow as all hell and half the time it times out and aborts itself early. Of note, I have about 16,000 links to check.

Was wondering how best to make this run quicker, and what I’m doing wrong?

Code below:

function echoappend ($file,$tobewritten) {

        fwrite($file,$tobewritten);
        echo $tobewritten;
}

error_reporting(E_ALL);
ini_set('display_errors', '1');


$filename=date('YmdHis') . "linkcheck.htm";
echo $filename;
$file = fopen($filename,"w+");

try {
        $conn = new PDO('mysql:host=localhost;dbname=databasename',$un,$pw);
        $conn->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_EXCEPTION);
        echo '<b>connected to db</b><br /><br />';

        $sitearray = array("medical.posterous","ebm.posterous","behavenet","guidance.nice","www.rch","emedicine","www.chw","www.rxlist","www.cks.nhs.uk");

        foreach ($sitearray as $key => $value) {    
            $site=$value;

            echoappend ($file, "<h1>" . $site . "</h1>");

            $q="SELECT * FROM link WHERE url LIKE :site";
            $stmt = $conn->prepare($q);
            $stmt->execute(array(':site' => 'http://' . $site . '%'));
            $result = $stmt->fetchAll();

            $totallinks = 0;
            $workinglinks = 0;

            foreach($result as $row)
            {

                $ch = curl_init();
                $originalurl = $row['url'];

                curl_setopt($ch, CURLOPT_URL, $originalurl);
                curl_setopt($ch, CURLOPT_HEADER, 1);
                curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
                curl_setopt($ch, CURLOPT_NOBODY, true);
                curl_setopt($ch, CURLOPT_FOLLOWLOCATION, false);


                $output = curl_exec($ch);
                if ($output === FALSE) {
                    echo "cURL Error: " . curl_error($ch);
                }

                $urlinfo = curl_getinfo($ch);

                if ($urlinfo['http_code'] == 200)
                {
                    echoappend($file, $row['name'] . ": <b>working!</b><br />");
                    $workinglinks++;
                }
                else if ($urlinfo['http_code'] == 301 || 302)
                {
                    $redirectch = curl_init();                  
                    curl_setopt($redirectch, CURLOPT_URL, $originalurl);
                    curl_setopt($redirectch, CURLOPT_HEADER, 1);
                    curl_setopt($redirectch, CURLOPT_RETURNTRANSFER, 1);
                    curl_setopt($redirectch, CURLOPT_NOBODY, false);
                    curl_setopt($redirectch, CURLOPT_FOLLOWLOCATION, true);

                    $redirectoutput = curl_exec($redirectch);

                    $doc = new DOMDocument();
                    @$doc->loadHTML($redirectoutput);
                    $nodes = $doc->getElementsByTagName('title');

                    $title = $nodes->item(0)->nodeValue;

                    echoappend ($file, $row['name'] . ": <b>redirect ... </b>" . $title . " ... ");

                    if (strpos(strtolower($title),strtolower($row['name']))===false) {
                        echoappend ($file, "FAIL<br />");
                    }
                    else {
                        $header = curl_getinfo($redirectch);
                        echoappend ($file, $header['url']);
                        echoappend ($file, "SUCCESS<br />");
                    }

                    curl_close($redirectch);
                }
                else
                {
                    echoappend ($file, $row['name'] . ": <b>FAIL code</b>" . $urlinfo['http_code'] . "<br />");
                }

                curl_close($ch);

                $totallinks++;
            }
            echoappend ($file, '<br />');

            echoappend ($file, $site . ": " . $workinglinks . "/" . $totallinks . " links working. <br /><br />");


        }

        $conn = null;
        echo '<br /><b>connection closed</b><br /><br />';

    } catch(PDOException $e) {
            echo 'ERROR: ' . $e->getMessage();
    }

Answer 1

Short answer is use the curl_multi_* methods to parallelize your requests.

The reason for the slowness is that web requests are comparatively slow. Sometimes VERY slow. Using the curl_multi_* functions lets you run multiple requests simultaneously.

One thing to be careful about is to limit the number of requests you run at once. In other words, don’t run 16,000 requests at once. Maybe start at 16 and see how that goes.

The following example should help you get started:

<?php

//
// Fetch a bunch of URLs in parallel. Returns an array of results indexed
// by URL.
//
function fetch_urls($urls, $curl_options = array()) {
  $curl_multi = curl_multi_init();
  $handles = array();

  $options = $curl_options + array(
    CURLOPT_HEADER         => true,
    CURLOPT_RETURNTRANSFER => true,
    CURLOPT_NOBODY         => true,
    CURLOPT_FOLLOWLOCATION => true);

  foreach($urls as $url) {
    $handles[$url] = curl_init($url);
    curl_setopt_array($handles[$url], $options);
    curl_multi_add_handle($curl_multi, $handles[$url]);
  }

  $active = null;
  do {
    $status = curl_multi_exec($curl_multi, $active);
  } while ($status == CURLM_CALL_MULTI_PERFORM);

  while ($active && ($status == CURLM_OK)) {
    if (curl_multi_select($curl_multi) != -1) {
      do {
        $status = curl_multi_exec($curl_multi, $active);
      } while ($status == CURLM_CALL_MULTI_PERFORM);
    }
  }

  if ($status != CURLM_OK) {
    trigger_error("Curl multi read error $status\n", E_USER_WARNING);
  }

  $results = array();
  foreach($handles as $url => $handle) {
    $results[$url] = curl_getinfo($handle);
    curl_multi_remove_handle($curl_multi, $handle);
    curl_close($handle);    
  }
  curl_multi_close($curl_multi);

  return $results;
}

//
// The urls to test
//
$urls = array("http://google.com", "http://yahoo.com", "http://google.com/probably-bogus", "http://www.google.com.au");

//
// The number of URLs to test simultaneously
//
$request_limit = 2;

//
// Test URLs in batches
//
$redirected_urls = array();
for ($i = 0 ; $i < count($urls) ; $i += $request_limit) {
  $results = fetch_urls(array_slice($urls, $i, $request_limit));
  foreach($results as $url => $result) {
    if ($result['http_code'] == 200) {
      $status = "Worked!";
    } else {
      $status = "FAILED with {$result['http_code']}";
    }
    if ($result["redirect_count"] > 0) {
      array_push($redirected_urls, $url);
      echo "{$url}: ${status}\n";
    } else {
      echo "{$url}: redirected to {$result['url']} and {$status}\n";
    }
  }
}

//
// Handle redirected URLs
//
echo "Processing redirected URLs...\n";
for ($i = 0 ; $i < count($redirected_urls) ; $i += $request_limit) {
  $results = fetch_urls(array_slice($redirected_urls, $i, $request_limit), array(CURLOPT_FOLLOWLOCATION => false));
  foreach($results as $url => $result) {
    if ($result['http_code'] == 301) {
      echo "{$url} permanently redirected to {$result['url']}\n";
    } else if ($result['http_code'] == 302) {
      echo "{$url} termporarily redirected to {$result['url']}\n";
    } else {
      echo "{$url}: FAILED with {$result['http_code']}\n";
    }
  }
}

The above code processes a list of URLs in batches. It works in two passes. In the first pass, each request is configured to follow redirects and simply reports whether each URL ultimately lead to a successful request, or a failure.

The second pass processes any redirected URLs detected in the first pass and reports whether the redirect was a permanent redirection (meaning you can update your database with the new URL), or temporary (meaning you should NOT update your database).

NOTE:

In your original code, you have the following line, which will not work the way you expect it to:

else if ($urlinfo['http_code'] == 301 || 302)

The expression will ALWAYS return TRUE. The correct expression is:

else if ($urlinfo['http_code'] == 301 || $urlinfo['http_code'] == 302)