Question

I’m running the PHP code below from command line. The issue is, its memory consumption far more than what it should be. I can’t, for the life of me, figure out where the memory is getting consumed.

for ($i=0;$i<100;$i++)  
        {
            $classObject = $classObjects[$i];                       

            echo $i . "   :   " . memory_get_usage(true) . "\n";
            $classDOM = $scraper->scrapeClassInfo($classObject,$termMap,$subjectMap);           
            unset($classDOM);           
        }

According to me, the memory consumed by my script should remain more or less constant after every iteration of the loop. Any memory consumed by $scraper->scrapeClassInfo() should be freed when its members go out of scope.

This is the output file I get. For the sake of brevity, I’m showing every 10th line of the output:

0   :   5767168
10   :   12058624
20   :   18350080
30   :   24903680
40   :   30932992
50   :   37748736
60   :   43778048
70   :   49807360
80   :   55836672
90   :   62914560
97   :   66846720

Fatal error: Allowed memory size of 67108864 bytes exhausted (tried to allocate 44 bytes) in /home/content/60/8349160/html/drexel/simple_html_dom.php on line 1255

Finally, as far as I can see, what $scraper->scrapeClassInfo() is doing should not really be the culprit, but just in case, here is the code:

function scrapeClassInfo($class,$termMap,$subjectMap)
        {
            $ckfile = tempnam ("/tmp", "CURLCOOKIE");
            $ckfile2 = tempnam ("/tmp", "CURLCOOKIE2");
            $ckfile3 = tempnam ("/tmp", "CURLCOOKIE3");         

            $termpage = $termMap[$class['termcode']];
            $subjectpage = $subjectMap[$class['subjectcode']];
            $classpage = $class['classlink'];

            //hit the main page and get cookie
            $ch = curl_init();
            curl_setopt($ch, CURLOPT_COOKIEJAR, $ckfile);
            curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
            curl_setopt($ch, CURLOPT_URL, $this->mainURL);
            curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
            curl_exec($ch);
            curl_close($ch);

            //hit the term page and get cookie
            $ch = curl_init();
            curl_setopt($ch, CURLOPT_COOKIEFILE, $ckfile);
            curl_setopt($ch, CURLOPT_COOKIEJAR, $ckfile2);
            curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
            curl_setopt($ch, CURLOPT_URL, $termpage);
            curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
            curl_exec($ch);
            curl_close($ch);

            //hit the subject page and get cookie
            $ch = curl_init();
            curl_setopt($ch, CURLOPT_COOKIEJAR, $ckfile3);
            curl_setopt($ch, CURLOPT_COOKIEFILE, $ckfile2);
            curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
            curl_setopt($ch, CURLOPT_URL, $subjectpage);
            curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
            curl_exec($ch);
            curl_close($ch);

            //hit the class page and scrape
            $ch = curl_init();              
            curl_setopt($ch, CURLOPT_COOKIEFILE, $ckfile3);
            curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
            curl_setopt($ch, CURLOPT_URL, $classpage);
            curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
            $result = curl_exec($ch);
            curl_close($ch);

            return str_get_html($result);
        }

The method called in the last line, str_get_html() is a member of Simple HTML DOM Parser

Should it matter, this is how I am calling my script:

/usr/local/php5/bin/php index.php 2>&1 1>output

Answer 1

Alright, figured it out. Apparently, its a bug that all version of PHP prior to 5.3 suffer from. Setting CURLOPT_RETURNTRANSFER to true causes massive memory leaks.

I ran the script again, this time invoking the php 5.3 binary:

/web/cgi-bin/php5_3 index.php 2>&1 1>output

And the output file reads:

0   :   6291456
10   :   9437184
20   :   10747904
30   :   11534336
40   :   11534336
50   :   11534336
60   :   11534336
70   :   11534336
80   :   11534336
90   :   11534336
99   :   11534336
152.74998211861 sec

Now that’s what I’m talking about! Perfectly constant memory footprint.