phpcurlphp-curlcurl-multi

curl_multi_exec: some images downloaded are missing some data / stream incomplete

I have implemented a PHP function which checks & downloads a lot of images (> 1'000) - as passed to it using an array - using the PHP curl_multi_init() method.

After reworking this a few times already, because I was getting things like 0 bytes files, etc. I have a solution now which downloads all images - BUT every other image file downloaded is not complete.

It looks to me as if I use file_put_contents() "too early", meaning, before some of the images' data has been received completely using curl_multi_exec().

Unfortunately I didn't find any similar question nor any Google Results for my case, in which I need to use curl_multi_exec, but do NOT want to retrieve & save images using the curl-opt-header "CURLOPT_FILE".

Hope someone is able to help me out regarding what I'm missing and why I get some broken images saved locally.

Here are some examples of the broken images retrieved:

example image 1 example image 2

Here's an example Array which I pass to my Multi-CURL-Function:

$curl_httpresources = [
        [ 'http://www.gravatar.com/avatar/example?d=mm&r=x&s=427'
        ,'/srv/www/data/images/1_unsplash.jpg' ],
        [ 'http://www.gravatar.com/avatar/example?d=identicon&r=x&s=427'
        ,'/srv/www/data/images/2_unsplash.jpg' ],
        [ 'http://www.gravatar.com/avatar/example?d=monsterid&r=x&s=427'
        ,'/srv/www/data/images/3_unsplash.jpg' ],
        [ 'http://www.gravatar.com/avatar/example?d=wavatar&r=x&s=427'
        ,'/srv/www/data/images/4_unsplash.jpg' ],
        [ 'http://www.gravatar.com/avatar/example?d=retro&r=x&s=427'
        ,'/srv/www/data/images/5_unsplash.jpg' ],
        [ 'http://www.gravatar.com/avatar/example?d=mm&r=x&s=427'
        ,'/srv/www/data/images/6_unsplash.jpg' ],
        [ 'http://www.gravatar.com/avatar/example?d=identicon&r=x&s=427'
        ,'/srv/www/data/images/7_unsplash.jpg' ],
        [ 'http://www.gravatar.com/avatar/example?d=monsterid&r=x&s=427'
        ,'/srv/www/data/images/8_unsplash.jpg' ],
        [ 'http://www.gravatar.com/avatar/example?d=wavatar&r=x&s=427'
        ,'/srv/www/data/images/9_unsplash.jpg' ],
        [ 'http://www.gravatar.com/avatar/example?d=retro&r=x&s=427'
        ,'/srv/www/data/images/10_unsplash.jpg' ],
];

My Multi-cURL PHP Function

Now for the function I'm currently using - and kind of "works", except some partially downloaded files - this is the code:

function cURLfetch(array $resources)
{
    /** Disable PHP timelimit, because this could take a while... */
    set_time_limit(0);

    /** Validate the $resources Array (not empty, null, or alike) */
    $resources_num = count($resources);
    if ( empty($resources) || $resources_num <= 0 ) return false;

    /** Callback-Function for writing data to file */
    $callback = function($resource, $filepath)
    {
        file_put_contents($filepath, $resource);
        /** For Debug only: output <img>-Tag with saved $resource */
        printf('<img src="%s"><br>', str_replace('/srv/www', '', $filepath));
    };

    /**
     * Initialize CURL process for handling multiple parallel requests 
     */
    $curl_instance = curl_multi_init();
    $curl_multi_exec_active = null;
    $curl_request_options = [
                                CURLOPT_USERAGENT => 'PHP-Script/1.0 (+https://website.com/)',
                                CURLOPT_TIMEOUT => 10,
                                CURLOPT_FOLLOWLOCATION => true,
                                CURLOPT_VERBOSE => false,
                                CURLOPT_RETURNTRANSFER => true,
                            ];

    /**
     * Looping through all $resources
     *   $resources[$i][0] = HTTP resource
     *   $resources[$i][1] = Target Filepath
     */
    for ($i = 0; $i < $resources_num; $i++)
    {
        $curl_requests[$i] = curl_init($resources[$i][0]);
        curl_setopt_array($curl_requests[$i], $curl_request_options);
        curl_multi_add_handle($curl_instance, $curl_requests[$i]);
    }

    do {
        try {
            $curl_execute = curl_multi_exec($curl_instance, $curl_multi_exec_active);
        } catch (Exception $e) {
            error_log($e->getMessage());
        }
    } while ($curl_execute == CURLM_CALL_MULTI_PERFORM);


    /** Wait until data arrives on all sockets */
    $h = 0; // initialise a counter
    while ($curl_multi_exec_active && $curl_execute == CURLM_OK)
    {
        if (curl_multi_select($curl_instance) != -1)
        {
            do {
              $curl_data = curl_multi_exec($curl_instance, $curl_multi_exec_active);
              $curl_done = curl_multi_info_read($curl_instance);
              /** Check if there is data... */
              if ($curl_done['handle'] !== NULL)
              {
                  /** Continue ONLY if HTTP statuscode was OK (200) */
                  $info = curl_getinfo($curl_done['handle']);
                  if ($info['http_code'] == 200)
                  {
                      if (!empty(curl_multi_getcontent($curl_requests[$h]))) {
                          /** Curl request successful. Process data using the callback function. */
                          $callback(curl_multi_getcontent($curl_requests[$h]), $resources[$h][1]);
                      }
                      $h++; // count up
                   }
               }
            } while ($curl_data == CURLM_CALL_MULTI_PERFORM);
        }
    }

    /** Close all $curl_requests */
    foreach($curl_requests as $request) {
        curl_multi_remove_handle($curl_instance, $request);
    }
    curl_multi_close($curl_instance);

    return true;
}

/** Start fetching images from an Array */
cURLfetch($curl_httpresources);

Thanks a lot for any help, much appreciated!

Solution

  • I ended up using just regular cURL requests in a classical loop, to query all >1'000 images and download the ones with a "HTTP 200 OK" response. My initial concerns, that the server might cut the connection due to a potential falsely identified DDoS has had no effect, why this approach works well for my case.

    Here's the final function with regular cURL requests I'm using:

    function cURLfetchUrl($url, $save_as_file)
{
    /** Validate $url & $save_as_file (not empty, null, or alike) */
    if ( empty($url) || is_numeric($url) ) return false;
    if ( empty($save_as_file) || is_numeric($save_as_file) ) return false;

    /** Disable PHP timelimit, because this could take a while... */
    set_time_limit(0);

    try {
        /**
         * Set cURL options to be passed to a single request
         */
        $curl_request_options = [
                                    CURLOPT_USERAGENT => 'PHP-Script/1.0 (+https://website.com/)',
                                    CURLOPT_TIMEOUT => 5,
                                    CURLOPT_FOLLOWLOCATION => true,
                                    CURLOPT_RETURNTRANSFER => true,
                                ];

        /** Initialize & execute cURL-Request */
        $curl_instance = curl_init($url);
        curl_setopt_array($curl_instance, $curl_request_options);
        $curl_data = curl_exec($curl_instance);
        $curl_done = curl_getinfo($curl_instance);

        /** cURL request successful */
        if ($curl_done['http_code'] == 200)
        {
            /** Open a new file handle, write the file & close the file handle */
            if (file_put_contents($save_as_file, $curl_data) !== false) {
                // logging if file_put_contents was OK
            } else {
                // logging if file_put_contents FAILED
            }
        }

        /** Close the $curl_instance */
        curl_close($curl_instance);

        return true;

    } catch (Exception $e) {
        error_log($e->getMessage());
        return false;
    }
}

    And to execute it:

    $curl_httpresources = [
    [ 'http://www.gravatar.com/avatar/example?d=mm&r=x&s=427'
    ,'/srv/www/data/images/1_unsplash.jpg' ],
    [ 'http://www.gravatar.com/avatar/example?d=identicon&r=x&s=427'
    ,'/srv/www/data/images/2_unsplash.jpg' ],
    [ 'http://www.gravatar.com/avatar/example?d=monsterid&r=x&s=427'
    ,'/srv/www/data/images/3_unsplash.jpg' ],
    [ 'http://www.gravatar.com/avatar/example?d=wavatar&r=x&s=427'
    ,'/srv/www/data/images/4_unsplash.jpg' ],
    [ 'http://www.gravatar.com/avatar/example?d=retro&r=x&s=427'
    ,'/srv/www/data/images/5_unsplash.jpg' ],
    [ 'http://www.gravatar.com/avatar/example?d=mm&r=x&s=427'
    ,'/srv/www/data/images/6_unsplash.jpg' ],
    [ 'http://www.gravatar.com/avatar/example?d=identicon&r=x&s=427'
    ,'/srv/www/data/images/7_unsplash.jpg' ],
    [ 'http://www.gravatar.com/avatar/example?d=monsterid&r=x&s=427'
    ,'/srv/www/data/images/8_unsplash.jpg' ],
    [ 'http://www.gravatar.com/avatar/example?d=wavatar&r=x&s=427'
    ,'/srv/www/data/images/9_unsplash.jpg' ],
    [ 'http://www.gravatar.com/avatar/example?d=retro&r=x&s=427'
    ,'/srv/www/data/images/10_unsplash.jpg' ],
];

/** cURL all request from the $curl_httpresources Array */
if (count($curl_httpresources) > 0)
{
    foreach ($curl_httpresources as $resource)
    {
        cURLfetchUrl($resource[0], $resource[1]);
    }
}

    Still, if someone has an idea, how to properly retrieve the file data streams using curl_multi, that would be great, as my answer to the initial question just shows a different approach - rather than solving the initial approach.