Question

我编写了一个MVC框架，我使用ob_start（'error_handler'）来捕获致命错误。这很棒！

在boot.php中

// Start the error-logging
ob_start( 'error_logging' );

// Run the request
Request::run( URL::getInstance()->init()->formatURL() );

// Flush and disable output-buffering
ob_end_flush();

error_logging函数：

function error_logging( $output )
{

    // Get the last error
    $trace = error_get_last();
    // Initialize our final array of fields explaining the error
    $final = array();

    // Check if the error does not exist, return the original output unedited
    if( !isset( $trace ) ) return $output;

    // Loop through or error and build our array
    foreach( $trace as $info => $value ) $final[$info] = $value;

    // Initialize our Error-class
    $error = Error::factory();
    // Handle the error with the fields and build our error-message
    $newOutput = $error->handle( $final['type'] , $final['message'], $final['file'], $final['line'] );

    // Check if the new output is null, if yes set output to the original output, otherwise set it to the new output
    $output = ( $newOutput === null ) ? $output : $newOutput;

    // Return the output
    return $output;

}

同样，这可以很好地处理所有错误！我不在Error-class中使用任何flush或任何东西。

现在，我有一个使用ob_start的Controller，因为它是某个方法的相当长的运行时间，我想要向用户反馈正在发生的事情以及脚本的作用。我正在使用ob_flush（）。

然而，在实现这个ob_start（'error_logging'）之后; ob_flush似乎不起作用。他们等到整个脚本完成（大约需要8分钟）。

为什么？

我如何冲洗的一个例子是这样的：在Controller-method中：

$this->login->execute();

在Login :: execute：

中

public function execute( $site = 'mobile' )
{

    ob_start();

    $this->debug->log('--------------------------------------------<br />');
    $this->debug->log( 'Logging into site: ' . ucfirst( $site ) . '<br />' );

    // Initiate cookies
    if( $site == 'full' )
        $this->_initCookies();

    // Set up the URL
    $url = Config::get( $site ) . Config::get( 'url_login' );
    curl_setopt( $this->curl->get(), CURLOPT_URL, $url );

    // CURL-values
    curl_setopt( $this->curl->get(), CURLOPT_FOLLOWLOCATION, 1  );
    curl_setopt( $this->curl->get(), CURLOPT_RETURNTRANSFER, 1  );
    curl_setopt( $this->curl->get(), CURLOPT_POST,           1  );
    curl_setopt( $this->curl->get(), CURLOPT_POSTFIELDS,     Config::get( 'postfields' ) );
    curl_setopt( $this->curl->get(), CURLOPT_COOKIEFILE,     'resources/tmp/cookies.txt');
    curl_setopt( $this->curl->get(), CURLOPT_CONNECTTIMEOUT, 10 );
    curl_setopt( $this->curl->get(), CURLOPT_TIMEOUT,        40 );
    curl_setopt( $this->curl->get(), CURLOPT_USERAGENT,      'Mozilla/5.0 (Windows NT 6.1; WOW64; rv:6.0.1) Gecko/20100101 Firefox/6.0.1');

    // Get the page returned after we've logged in
    $this->login_html = str_get_html( curl_exec( $this->curl->get() ) );

    // Validate the page returned to see if we successfully logged in
    $this->_validateLogin( $site );

    $this->debug->log('--------------------------------------------<br />');

    ob_end_flush();

    // Return $this
    return $this;

}

每个$ this-＆gt; debug-＆gt; log（）调用都会像这样刷新输出：

public function log( $string )
{

    if( $this->active() || $this->logging() )
    {

        echo str_repeat("<!-- AGENT SMITH -->", 500);
        echo $string;

        ob_flush();
        flush();

    }

}

知道它为什么没有正确冲洗？告诉我你是否需要查看更多代码，我会提供它！

由于

:::::::: SOLUTION ::::::::::

我必须保存第一个缓冲区（error_logging），结束它，启动新的缓冲区，执行我的操作然后结束那个缓冲区，再次启动第一个缓冲区并回显保存的输出。

为它做了一个课程：

class Tanaxia_Buffer
{

    public static $instance = null;

    private $main = null;
    private $previousOutput = "";

    public static function getInstance()
    {

        if( Buffer::$instance === null )
            Buffer::$instance = new Buffer();

        return Buffer::$instance;

    }

    public function set_main( $ob )
    {

        $this->main = $ob;

    }

    public function start( $main = null )
    {

        if( $main !== null )
            $this->set_main( $main );

        ob_start( $this->main );

    }

    public function end( $type = 'clean' )
    {

        switch( $type )
        {

            case 'flush': ob_end_flush(); break;
            case 'clean': ob_end_clean(); break;

        }

    }

    public function start_separate()
    {

        $this->previousOutput = ob_get_contents();
        ob_end_clean();

        ob_start();

    }

    public function end_separate( $type = 'flush' )
    {

        switch( $type )
        {

            case 'flush': ob_end_flush(); break;
            case 'clean': ob_end_clean(); break;

        }

        ob_start( $this->main );
        echo $this->previousOutput;

        empty( $this->previousOutput );

    }

}

用法：

boot.php

// Start the error-logging
Buffer::getInstance()->start( 'error_logging' );

// Run the request
Request::run( URL::getInstance()->init()->formatURL() );

// Flush and disable output-buffering
Buffer::getInstance()->end( 'flush' );

登录::执行（）：

public function execute( $site = 'mobile' )
{

    Buffer::getInstance()->start_separate();

    // Everything in between

    Buffer::getInstance()->end_separate();

    // Return $this
    return $this;

}

完美的作品！不会支持几个嵌套的ob_start，因为它只保存了1个先前的输出。虽然现在可以修复但不是必需的！

Answer 1

我猜它与嵌套的ob_start()调用有关。我假设您希望将它们嵌套，但如果在execute()函数中，您可以在开始下一个之前调用ob_end_flush()或ob_end_clean()，那么它是否会按照您希望的方式进行刷新输出缓冲区？像这样：

public function execute( $site = 'mobile' )
{
    // save existing contents of output buffer
    $previousOutput = ob_get_contents();
    // wipe the output buffer itself
    ob_end_clean();
    // start a new output buffer
    ob_start();

    /* SNIP all the code that wouldn't change */

    // flush the function's output buffer
    ob_end_flush();
    // restart the old output buffer
    ob_start("error_logging");
    // take whatever was in the old buffer and make sure it shows up in the new one
    echo $previousOutput;

    return $this;
}

PHP ob_start调用两次，现在不会正确刷新

1 个答案: