MultiHttpClient.php

Go to the documentation of this file.

<?php
class MultiHttpClient {
    protected $multiHandle = null; // curl_multi handle
    protected $caBundlePath;
    protected $connTimeout = 10;
    protected $reqTimeout = 300;
    protected $usePipelining = false;
    protected $maxConnsPerHost = 50;
    protected $proxy;
    protected $userAgent = 'wikimedia/multi-http-client v1.0';

    public function __construct( array $options ) {
        if ( isset( $options['caBundlePath'] ) ) {
            $this->caBundlePath = $options['caBundlePath'];
            if ( !file_exists( $this->caBundlePath ) ) {
                throw new Exception( "Cannot find CA bundle: " . $this->caBundlePath );
            }
        }
        static $opts = [
            'connTimeout', 'reqTimeout', 'usePipelining', 'maxConnsPerHost', 'proxy', 'userAgent'
        ];
        foreach ( $opts as $key ) {
            if ( isset( $options[$key] ) ) {
                $this->$key = $options[$key];
            }
        }
    }

    final public function run( array $req, array $opts = [] ) {
        return $this->runMulti( [ $req ], $opts )[0]['response'];
    }

    public function runMulti( array $reqs, array $opts = [] ) {
        $chm = $this->getCurlMulti();

        // Normalize $reqs and add all of the required cURL handles...
        $handles = [];
        foreach ( $reqs as $index => &$req ) {
            $req['response'] = [
                'code'     => 0,
                'reason'   => '',
                'headers'  => [],
                'body'     => '',
                'error'    => ''
            ];
            if ( isset( $req[0] ) ) {
                $req['method'] = $req[0]; // short-form
                unset( $req[0] );
            }
            if ( isset( $req[1] ) ) {
                $req['url'] = $req[1]; // short-form
                unset( $req[1] );
            }
            if ( !isset( $req['method'] ) ) {
                throw new Exception( "Request has no 'method' field set." );
            } elseif ( !isset( $req['url'] ) ) {
                throw new Exception( "Request has no 'url' field set." );
            }
            $req['query'] = isset( $req['query'] ) ? $req['query'] : [];
            $headers = []; // normalized headers
            if ( isset( $req['headers'] ) ) {
                foreach ( $req['headers'] as $name => $value ) {
                    $headers[strtolower( $name )] = $value;
                }
            }
            $req['headers'] = $headers;
            if ( !isset( $req['body'] ) ) {
                $req['body'] = '';
                $req['headers']['content-length'] = 0;
            }
            $handles[$index] = $this->getCurlHandle( $req, $opts );
            if ( count( $reqs ) > 1 ) {
                // https://github.com/guzzle/guzzle/issues/349
                curl_setopt( $handles[$index], CURLOPT_FORBID_REUSE, true );
            }
        }
        unset( $req ); // don't assign over this by accident

        $indexes = array_keys( $reqs );
        if ( function_exists( 'curl_multi_setopt' ) ) { // PHP 5.5
            if ( isset( $opts['usePipelining'] ) ) {
                curl_multi_setopt( $chm, CURLMOPT_PIPELINING, (int)$opts['usePipelining'] );
            }
            if ( isset( $opts['maxConnsPerHost'] ) ) {
                // Keep these sockets around as they may be needed later in the request
                curl_multi_setopt( $chm, CURLMOPT_MAXCONNECTS, (int)$opts['maxConnsPerHost'] );
            }
        }

        // @TODO: use a per-host rolling handle window (e.g. CURLMOPT_MAX_HOST_CONNECTIONS)
        $batches = array_chunk( $indexes, $this->maxConnsPerHost );
        $infos = [];

        foreach ( $batches as $batch ) {
            // Attach all cURL handles for this batch
            foreach ( $batch as $index ) {
                curl_multi_add_handle( $chm, $handles[$index] );
            }
            // Execute the cURL handles concurrently...
            $active = null; // handles still being processed
            do {
                // Do any available work...
                do {
                    $mrc = curl_multi_exec( $chm, $active );
                    $info = curl_multi_info_read( $chm );
                    if ( $info !== false ) {
                        $infos[(int)$info['handle']] = $info;
                    }
                } while ( $mrc == CURLM_CALL_MULTI_PERFORM );
                // Wait (if possible) for available work...
                if ( $active > 0 && $mrc == CURLM_OK ) {
                    if ( curl_multi_select( $chm, 10 ) == -1 ) {
                        // PHP bug 63411; http://curl.haxx.se/libcurl/c/curl_multi_fdset.html
                        usleep( 5000 ); // 5ms
                    }
                }
            } while ( $active > 0 && $mrc == CURLM_OK );
        }

        // Remove all of the added cURL handles and check for errors...
        foreach ( $reqs as $index => &$req ) {
            $ch = $handles[$index];
            curl_multi_remove_handle( $chm, $ch );

            if ( isset( $infos[(int)$ch] ) ) {
                $info = $infos[(int)$ch];
                $errno = $info['result'];
                if ( $errno !== 0 ) {
                    $req['response']['error'] = "(curl error: $errno)";
                    if ( function_exists( 'curl_strerror' ) ) {
                        $req['response']['error'] .= " " . curl_strerror( $errno );
                    }
                }
            } else {
                $req['response']['error'] = "(curl error: no status set)";
            }

            // For convenience with the list() operator
            $req['response'][0] = $req['response']['code'];
            $req['response'][1] = $req['response']['reason'];
            $req['response'][2] = $req['response']['headers'];
            $req['response'][3] = $req['response']['body'];
            $req['response'][4] = $req['response']['error'];
            curl_close( $ch );
            // Close any string wrapper file handles
            if ( isset( $req['_closeHandle'] ) ) {
                fclose( $req['_closeHandle'] );
                unset( $req['_closeHandle'] );
            }
        }
        unset( $req ); // don't assign over this by accident

        // Restore the default settings
        if ( function_exists( 'curl_multi_setopt' ) ) { // PHP 5.5
            curl_multi_setopt( $chm, CURLMOPT_PIPELINING, (int)$this->usePipelining );
            curl_multi_setopt( $chm, CURLMOPT_MAXCONNECTS, (int)$this->maxConnsPerHost );
        }

        return $reqs;
    }

    protected function getCurlHandle( array &$req, array $opts = [] ) {
        $ch = curl_init();

        curl_setopt( $ch, CURLOPT_CONNECTTIMEOUT,
            isset( $opts['connTimeout'] ) ? $opts['connTimeout'] : $this->connTimeout );
        curl_setopt( $ch, CURLOPT_PROXY, isset( $req['proxy'] ) ? $req['proxy'] : $this->proxy );
        curl_setopt( $ch, CURLOPT_TIMEOUT,
            isset( $opts['reqTimeout'] ) ? $opts['reqTimeout'] : $this->reqTimeout );
        curl_setopt( $ch, CURLOPT_FOLLOWLOCATION, 1 );
        curl_setopt( $ch, CURLOPT_MAXREDIRS, 4 );
        curl_setopt( $ch, CURLOPT_HEADER, 0 );
        if ( !is_null( $this->caBundlePath ) ) {
            curl_setopt( $ch, CURLOPT_SSL_VERIFYPEER, true );
            curl_setopt( $ch, CURLOPT_CAINFO, $this->caBundlePath );
        }
        curl_setopt( $ch, CURLOPT_RETURNTRANSFER, 1 );

        $url = $req['url'];
        // PHP_QUERY_RFC3986 is PHP 5.4+ only
        $query = str_replace(
            [ '+', '%7E' ],
            [ '%20', '~' ],
            http_build_query( $req['query'], '', '&' )
        );
        if ( $query != '' ) {
            $url .= strpos( $req['url'], '?' ) === false ? "?$query" : "&$query";
        }
        curl_setopt( $ch, CURLOPT_URL, $url );

        curl_setopt( $ch, CURLOPT_CUSTOMREQUEST, $req['method'] );
        if ( $req['method'] === 'HEAD' ) {
            curl_setopt( $ch, CURLOPT_NOBODY, 1 );
        }

        if ( $req['method'] === 'PUT' ) {
            curl_setopt( $ch, CURLOPT_PUT, 1 );
            if ( is_resource( $req['body'] ) ) {
                curl_setopt( $ch, CURLOPT_INFILE, $req['body'] );
                if ( isset( $req['headers']['content-length'] ) ) {
                    curl_setopt( $ch, CURLOPT_INFILESIZE, $req['headers']['content-length'] );
                } elseif ( isset( $req['headers']['transfer-encoding'] ) &&
                    $req['headers']['transfer-encoding'] === 'chunks'
                ) {
                    curl_setopt( $ch, CURLOPT_UPLOAD, true );
                } else {
                    throw new Exception( "Missing 'Content-Length' or 'Transfer-Encoding' header." );
                }
            } elseif ( $req['body'] !== '' ) {
                $fp = fopen( "php://temp", "wb+" );
                fwrite( $fp, $req['body'], strlen( $req['body'] ) );
                rewind( $fp );
                curl_setopt( $ch, CURLOPT_INFILE, $fp );
                curl_setopt( $ch, CURLOPT_INFILESIZE, strlen( $req['body'] ) );
                $req['_closeHandle'] = $fp; // remember to close this later
            } else {
                curl_setopt( $ch, CURLOPT_INFILESIZE, 0 );
            }
            curl_setopt( $ch, CURLOPT_READFUNCTION,
                function ( $ch, $fd, $length ) {
                    $data = fread( $fd, $length );
                    $len = strlen( $data );
                    return $data;
                }
            );
        } elseif ( $req['method'] === 'POST' ) {
            curl_setopt( $ch, CURLOPT_POST, 1 );
            // Don't interpret POST parameters starting with '@' as file uploads, because this
            // makes it impossible to POST plain values starting with '@' (and causes security
            // issues potentially exposing the contents of local files).
            // The PHP manual says this option was introduced in PHP 5.5 defaults to true in PHP 5.6,
            // but we support lower versions, and the option doesn't exist in HHVM 5.6.99.
            if ( defined( 'CURLOPT_SAFE_UPLOAD' ) ) {
                curl_setopt( $ch, CURLOPT_SAFE_UPLOAD, true );
            } elseif ( is_array( $req['body'] ) ) {
                // In PHP 5.2 and later, '@' is interpreted as a file upload if POSTFIELDS
                // is an array, but not if it's a string. So convert $req['body'] to a string
                // for safety.
                $req['body'] = wfArrayToCgi( $req['body'] );
            }
            curl_setopt( $ch, CURLOPT_POSTFIELDS, $req['body'] );
        } else {
            if ( is_resource( $req['body'] ) || $req['body'] !== '' ) {
                throw new Exception( "HTTP body specified for a non PUT/POST request." );
            }
            $req['headers']['content-length'] = 0;
        }

        if ( !isset( $req['headers']['user-agent'] ) ) {
            $req['headers']['user-agent'] = $this->userAgent;
        }

        $headers = [];
        foreach ( $req['headers'] as $name => $value ) {
            if ( strpos( $name, ': ' ) ) {
                throw new Exception( "Headers cannot have ':' in the name." );
            }
            $headers[] = $name . ': ' . trim( $value );
        }
        curl_setopt( $ch, CURLOPT_HTTPHEADER, $headers );

        curl_setopt( $ch, CURLOPT_HEADERFUNCTION,
            function ( $ch, $header ) use ( &$req ) {
                $length = strlen( $header );
                $matches = [];
                if ( preg_match( "/^(HTTP\/1\.[01]) (\d{3}) (.*)/", $header, $matches ) ) {
                    $req['response']['code'] = (int)$matches[2];
                    $req['response']['reason'] = trim( $matches[3] );
                    return $length;
                }
                if ( strpos( $header, ":" ) === false ) {
                    return $length;
                }
                list( $name, $value ) = explode( ":", $header, 2 );
                $req['response']['headers'][strtolower( $name )] = trim( $value );
                return $length;
            }
        );

        if ( isset( $req['stream'] ) ) {
            // Don't just use CURLOPT_FILE as that might give:
            // curl_setopt(): cannot represent a stream of type Output as a STDIO FILE*
            // The callback here handles both normal files and php://temp handles.
            curl_setopt( $ch, CURLOPT_WRITEFUNCTION,
                function ( $ch, $data ) use ( &$req ) {
                    return fwrite( $req['stream'], $data );
                }
            );
        } else {
            curl_setopt( $ch, CURLOPT_WRITEFUNCTION,
                function ( $ch, $data ) use ( &$req ) {
                    $req['response']['body'] .= $data;
                    return strlen( $data );
                }
            );
        }

        return $ch;
    }

    protected function getCurlMulti() {
        if ( !$this->multiHandle ) {
            $cmh = curl_multi_init();
            if ( function_exists( 'curl_multi_setopt' ) ) { // PHP 5.5
                curl_multi_setopt( $cmh, CURLMOPT_PIPELINING, (int)$this->usePipelining );
                curl_multi_setopt( $cmh, CURLMOPT_MAXCONNECTS, (int)$this->maxConnsPerHost );
            }
            $this->multiHandle = $cmh;
        }
        return $this->multiHandle;
    }

    function __destruct() {
        if ( $this->multiHandle ) {
            curl_multi_close( $this->multiHandle );
        }
    }
}