MultiHttpClient.php

Go to the documentation of this file.

<?php
use MediaWiki\MediaWikiServices;
use Psr\Log\LoggerAwareInterface;
use Psr\Log\LoggerInterface;
use Psr\Log\NullLogger;
 
class MultiHttpClient implements LoggerAwareInterface {
    private const SENSITIVE_HEADERS = '/(^|-|_)(authorization|auth|password|cookie)($|-|_)/';
    protected $cmh;
    protected $caBundlePath;
    protected $connTimeout = 10;
    protected $maxConnTimeout = INF;
    protected $reqTimeout = 30;
    protected $maxReqTimeout = INF;
    protected $usePipelining = false;
    protected $maxConnsPerHost = 50;
    protected $proxy;
    protected $localProxy = false;
    protected $localVirtualHosts = [];
    protected $userAgent = 'wikimedia/multi-http-client v1.0';
    protected $logger;
 
    // In PHP 7 due to https://bugs.php.net/bug.php?id=76480 the request/connect
    // timeouts are periodically polled instead of being accurately respected.
    // The select timeout is set to the minimum timeout multiplied by this factor.
    private const TIMEOUT_ACCURACY_FACTOR = 0.1;
 
    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', 'maxConnTimeout', 'reqTimeout', 'maxReqTimeout',
            'usePipelining', 'maxConnsPerHost', 'proxy', 'userAgent', 'logger',
            'localProxy', 'localVirtualHosts',
        ];
        foreach ( $opts as $key ) {
            if ( isset( $options[$key] ) ) {
                $this->$key = $options[$key];
            }
        }
        if ( $this->logger === null ) {
            $this->logger = new NullLogger;
        }
    }
 
    public function run( array $req, array $opts = [] ) {
        return $this->runMulti( [ $req ], $opts )[0]['response'];
    }
 
    public function runMulti( array $reqs, array $opts = [] ) {
        $this->normalizeRequests( $reqs );
        $opts += [ 'connTimeout' => $this->connTimeout, 'reqTimeout' => $this->reqTimeout ];
 
        if ( $this->maxConnTimeout && $opts['connTimeout'] > $this->maxConnTimeout ) {
            $opts['connTimeout'] = $this->maxConnTimeout;
        }
        if ( $this->maxReqTimeout && $opts['reqTimeout'] > $this->maxReqTimeout ) {
            $opts['reqTimeout'] = $this->maxReqTimeout;
        }
 
        if ( $this->isCurlEnabled() ) {
            switch ( $opts['httpVersion'] ?? null ) {
                case 'v1.0':
                    $opts['httpVersion'] = CURL_HTTP_VERSION_1_0;
                    break;
                case 'v1.1':
                    $opts['httpVersion'] = CURL_HTTP_VERSION_1_1;
                    break;
                case 'v2':
                case 'v2.0':
                    $opts['httpVersion'] = CURL_HTTP_VERSION_2_0;
                    break;
                default:
                    $opts['httpVersion'] = CURL_HTTP_VERSION_NONE;
            }
            return $this->runMultiCurl( $reqs, $opts );
        } else {
            # TODO: Add handling for httpVersion option
            return $this->runMultiHttp( $reqs, $opts );
        }
    }
 
    protected function isCurlEnabled() {
        // Explicitly test if curl_multi* is blocked, as some users' hosts provide
        // them with a modified curl with the multi-threaded parts removed(!)
        return extension_loaded( 'curl' ) && function_exists( 'curl_multi_init' );
    }
 
    private function runMultiCurl( array $reqs, array $opts ) {
        $chm = $this->getCurlMulti( $opts );
 
        $selectTimeout = $this->getSelectTimeout( $opts );
 
        // Add all of the required cURL handles...
        $handles = [];
        foreach ( $reqs as $index => &$req ) {
            $handles[$index] = $this->getCurlHandle( $req, $opts );
            curl_multi_add_handle( $chm, $handles[$index] );
        }
        unset( $req ); // don't assign over this by accident
 
        $infos = [];
        // 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 ) {
                    // Note: cast to integer even works on PHP 8.0+ despite the
                    // handle being an object not a resource, because CurlHandle
                    // has a backwards-compatible cast_object handler.
                    $infos[(int)$info['handle']] = $info;
                }
            } while ( $mrc == CURLM_CALL_MULTI_PERFORM );
            // Wait (if possible) for available work...
            if ( $active > 0 && $mrc == CURLM_OK && curl_multi_select( $chm, $selectTimeout ) == -1 ) {
                // PHP bug 63411; https://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 );
                    }
                    $this->logger->warning( "Error fetching URL \"{$req['url']}\": " .
                        $req['response']['error'] );
                } else {
                    $this->logger->debug(
                        "HTTP complete: {method} {url} code={response_code} size={size} " .
                        "total={total_time} connect={connect_time}",
                        [
                            'method' => $req['method'],
                            'url' => $req['url'],
                            'response_code' => $req['response']['code'],
                            'size' => curl_getinfo( $ch, CURLINFO_SIZE_DOWNLOAD ),
                            'total_time' => $this->getCurlTime(
                                $ch, CURLINFO_TOTAL_TIME, 'CURLINFO_TOTAL_TIME_T'
                            ),
                            'connect_time' => $this->getCurlTime(
                                $ch, CURLINFO_CONNECT_TIME, 'CURLINFO_CONNECT_TIME_T'
                            ),
                        ]
                    );
                }
            } 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
 
        return $reqs;
    }
 
    protected function getCurlHandle( array &$req, array $opts ) {
        $ch = curl_init();
 
        curl_setopt( $ch, CURLOPT_PROXY, $req['proxy'] ?? $this->proxy );
        curl_setopt( $ch, CURLOPT_CONNECTTIMEOUT_MS, intval( $opts['connTimeout'] * 1e3 ) );
        curl_setopt( $ch, CURLOPT_TIMEOUT_MS, intval( $opts['reqTimeout'] * 1e3 ) );
        curl_setopt( $ch, CURLOPT_FOLLOWLOCATION, 1 );
        curl_setopt( $ch, CURLOPT_MAXREDIRS, 4 );
        curl_setopt( $ch, CURLOPT_HEADER, 0 );
        if ( $this->caBundlePath !== null ) {
            curl_setopt( $ch, CURLOPT_SSL_VERIFYPEER, true );
            curl_setopt( $ch, CURLOPT_CAINFO, $this->caBundlePath );
        }
        curl_setopt( $ch, CURLOPT_RETURNTRANSFER, 1 );
 
        $url = $req['url'];
        $query = http_build_query( $req['query'], '', '&', PHP_QUERY_RFC3986 );
        if ( $query != '' ) {
            $url .= strpos( $req['url'], '?' ) === false ? "?$query" : "&$query";
        }
        curl_setopt( $ch, CURLOPT_URL, $url );
        curl_setopt( $ch, CURLOPT_CUSTOMREQUEST, $req['method'] );
        curl_setopt( $ch, CURLOPT_NOBODY, ( $req['method'] === 'HEAD' ) );
        curl_setopt( $ch, CURLOPT_HTTP_VERSION, $opts['httpVersion'] ?? CURL_HTTP_VERSION_NONE );
 
        if ( $req['method'] === 'PUT' ) {
            curl_setopt( $ch, CURLOPT_PUT, 1 );
            // phpcs:ignore MediaWiki.Usage.ForbiddenFunctions.is_resource
            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,
                static function ( $ch, $fd, $length ) {
                    return (string)fread( $fd, $length );
                }
            );
        } elseif ( $req['method'] === 'POST' ) {
            curl_setopt( $ch, CURLOPT_POST, 1 );
            curl_setopt( $ch, CURLOPT_POSTFIELDS, $req['body'] );
        } else {
            // phpcs:ignore MediaWiki.Usage.ForbiddenFunctions.is_resource
            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,
            static function ( $ch, $header ) use ( &$req ) {
                if ( !empty( $req['flags']['relayResponseHeaders'] ) && trim( $header ) !== '' ) {
                    header( $header );
                }
                $length = strlen( $header );
                $matches = [];
                if ( preg_match( "/^(HTTP\/(?:1\.[01]|2)) (\d{3}) (.*)/", $header, $matches ) ) {
                    $req['response']['code'] = (int)$matches[2];
                    $req['response']['reason'] = trim( $matches[3] );
                    // After a redirect we will receive this again, but we already stored headers
                    // that belonged to a redirect response. Start over.
                    $req['response']['headers'] = [];
                    return $length;
                }
                if ( strpos( $header, ":" ) === false ) {
                    return $length;
                }
                list( $name, $value ) = explode( ":", $header, 2 );
                $name = strtolower( $name );
                $value = trim( $value );
                if ( isset( $req['response']['headers'][$name] ) ) {
                    $req['response']['headers'][$name] .= ', ' . $value;
                } else {
                    $req['response']['headers'][$name] = $value;
                }
                return $length;
            }
        );
 
        // This works with both file and php://temp handles (unlike CURLOPT_FILE)
        $hasOutputStream = isset( $req['stream'] );
        curl_setopt( $ch, CURLOPT_WRITEFUNCTION,
            static function ( $ch, $data ) use ( &$req, $hasOutputStream ) {
                if ( $hasOutputStream ) {
                    // @phan-suppress-next-line PhanTypePossiblyInvalidDimOffset False positive
                    return fwrite( $req['stream'], $data );
                } else {
                    // @phan-suppress-next-line PhanTypeArraySuspiciousNullable
                    $req['response']['body'] .= $data;
 
                    return strlen( $data );
                }
            }
        );
 
        return $ch;
    }
 
    protected function getCurlMulti( array $opts ) {
        if ( !$this->cmh ) {
            $cmh = curl_multi_init();
            // Limit the size of the idle connection cache such that consecutive parallel
            // request batches to the same host can avoid having to keep making connections
            curl_multi_setopt( $cmh, CURLMOPT_MAXCONNECTS, (int)$this->maxConnsPerHost );
            $this->cmh = $cmh;
        }
 
        $curlVersion = curl_version()['version'];
 
        // CURLMOPT_MAX_HOST_CONNECTIONS is available since PHP 7.0.7 and cURL 7.30.0
        if ( version_compare( $curlVersion, '7.30.0', '>=' ) ) {
            // Limit the number of in-flight requests for any given host
            $maxHostConns = $opts['maxConnsPerHost'] ?? $this->maxConnsPerHost;
            curl_multi_setopt( $this->cmh, CURLMOPT_MAX_HOST_CONNECTIONS, (int)$maxHostConns );
        }
 
        if ( $opts['usePipelining'] ?? $this->usePipelining ) {
            if ( version_compare( $curlVersion, '7.43', '<' ) ) {
                // The option is a boolean
                $pipelining = 1;
            } elseif ( version_compare( $curlVersion, '7.62', '<' ) ) {
                // The option is a bitfield and HTTP/1.x pipelining is supported
                $pipelining = CURLPIPE_HTTP1 | CURLPIPE_MULTIPLEX;
            } else {
                // The option is a bitfield but HTTP/1.x pipelining has been removed
                $pipelining = CURLPIPE_MULTIPLEX;
            }
            // Suppress deprecation, we know already (T264735)
            // phpcs:ignore Generic.PHP.NoSilencedErrors
            @curl_multi_setopt( $this->cmh, CURLMOPT_PIPELINING, $pipelining );
        }
 
        return $this->cmh;
    }
 
    private function getCurlTime( $ch, $oldOption, $newConstName ): string {
        if ( defined( $newConstName ) ) {
            return sprintf( "%.6F", curl_getinfo( $ch, constant( $newConstName ) ) / 1e6 );
        } else {
            return (string)curl_getinfo( $ch, $oldOption );
        }
    }
 
    private function runMultiHttp( array $reqs, array $opts = [] ) {
        $httpOptions = [
            'timeout' => $opts['reqTimeout'] ?? $this->reqTimeout,
            'connectTimeout' => $opts['connTimeout'] ?? $this->connTimeout,
            'logger' => $this->logger,
            'caInfo' => $this->caBundlePath,
        ];
        foreach ( $reqs as &$req ) {
            $reqOptions = $httpOptions + [
                'method' => $req['method'],
                'proxy' => $req['proxy'] ?? $this->proxy,
                'userAgent' => $req['headers']['user-agent'] ?? $this->userAgent,
                'postData' => $req['body'],
            ];
 
            $url = $req['url'];
            $query = http_build_query( $req['query'], '', '&', PHP_QUERY_RFC3986 );
            if ( $query != '' ) {
                $url .= strpos( $req['url'], '?' ) === false ? "?$query" : "&$query";
            }
 
            $httpRequest = MediaWikiServices::getInstance()->getHttpRequestFactory()->create(
                $url, $reqOptions, __METHOD__ );
            $httpRequest->setLogger( $this->logger );
            $sv = $httpRequest->execute()->getStatusValue();
 
            $respHeaders = array_map(
                static function ( $v ) {
                    return implode( ', ', $v );
                },
                $httpRequest->getResponseHeaders() );
 
            $req['response'] = [
                'code' => $httpRequest->getStatus(),
                'reason' => '',
                'headers' => $respHeaders,
                'body' => $httpRequest->getContent(),
                'error' => '',
            ];
 
            if ( !$sv->isOK() ) {
                $svErrors = $sv->getErrors();
                if ( isset( $svErrors[0] ) ) {
                    $req['response']['error'] = $svErrors[0]['message'];
 
                    // param values vary per failure type (ex. unknown host vs unknown page)
                    if ( isset( $svErrors[0]['params'][0] ) ) {
                        if ( is_numeric( $svErrors[0]['params'][0] ) ) {
                            if ( isset( $svErrors[0]['params'][1] ) ) {
                                // @phan-suppress-next-line PhanTypeInvalidDimOffset
                                $req['response']['reason'] = $svErrors[0]['params'][1];
                            }
                        } else {
                            $req['response']['reason'] = $svErrors[0]['params'][0];
                        }
                    }
                }
            }
 
            $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'];
        }
 
        return $reqs;
    }
 
    private function normalizeRequests( array &$reqs ) {
        foreach ( $reqs as &$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." );
            }
            if ( $this->localProxy !== false && $this->isLocalURL( $req['url'] ) ) {
                $this->useReverseProxy( $req, $this->localProxy );
            }
            $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;
            }
            // Redact some headers we know to have tokens before logging them
            $logHeaders = $req['headers'];
            foreach ( $logHeaders as $header => $value ) {
                if ( preg_match( self::SENSITIVE_HEADERS, $header ) === 1 ) {
                    $logHeaders[$header] = '[redacted]';
                }
            }
            $this->logger->debug( "HTTP start: {method} {url}",
                [
                    'method' => $req['method'],
                    'url' => $req['url'],
                    'headers' => $logHeaders,
                ]
            );
            $req['flags'] = $req['flags'] ?? [];
        }
    }
 
    private function useReverseProxy( array &$req, $proxy ) {
        $parsedProxy = wfParseUrl( $proxy );
        if ( $parsedProxy === false ) {
            throw new Exception( "Invalid reverseProxy configured: $proxy" );
        }
        $parsedUrl = wfParseUrl( $req['url'] );
        if ( $parsedUrl === false ) {
            throw new Exception( "Invalid url specified: ${req['url']}" );
        }
        // Set the current host in the Host header
        $req['headers']['Host'] = $parsedUrl['host'];
        // Replace scheme, host and port in the request
        $parsedUrl['scheme'] = $parsedProxy['scheme'];
        $parsedUrl['host'] = $parsedProxy['host'];
        if ( isset( $parsedProxy['port'] ) ) {
            $parsedUrl['port'] = $parsedProxy['port'];
        } else {
            unset( $parsedUrl['port'] );
        }
        $req['url'] = wfAssembleUrl( $parsedUrl );
        // Explicitly disable use of another proxy by setting to false,
        // since null will fallback to $this->proxy
        $req['proxy'] = false;
    }
 
    private function isLocalURL( $url ) {
        if ( !$this->localVirtualHosts ) {
            // Shortcut
            return false;
        }
 
        // Extract host part
        $matches = [];
        if ( preg_match( '!^https?://([\w.-]+)[/:].*$!', $url, $matches ) ) {
            $host = $matches[1];
            // Split up dotwise
            $domainParts = explode( '.', $host );
            // Check if this domain or any superdomain is listed as a local virtual host
            $domainParts = array_reverse( $domainParts );
 
            $domain = '';
            $countParts = count( $domainParts );
            for ( $i = 0; $i < $countParts; $i++ ) {
                $domainPart = $domainParts[$i];
                if ( $i == 0 ) {
                    $domain = $domainPart;
                } else {
                    $domain = $domainPart . '.' . $domain;
                }
 
                if ( in_array( $domain, $this->localVirtualHosts ) ) {
                    return true;
                }
            }
        }
 
        return false;
    }
 
    private function getSelectTimeout( $opts ) {
        $connTimeout = $opts['connTimeout'] ?? $this->connTimeout;
        $reqTimeout = $opts['reqTimeout'] ?? $this->reqTimeout;
        $timeouts = array_filter( [ $connTimeout, $reqTimeout ] );
        if ( count( $timeouts ) === 0 ) {
            return 1;
        }
 
        $selectTimeout = min( $timeouts ) * self::TIMEOUT_ACCURACY_FACTOR;
        // Minimum 10us
        if ( $selectTimeout < 10e-6 ) {
            $selectTimeout = 10e-6;
        }
        return $selectTimeout;
    }
 
    public function setLogger( LoggerInterface $logger ) {
        $this->logger = $logger;
    }
 
    public function __destruct() {
        if ( $this->cmh ) {
            curl_multi_close( $this->cmh );
        }
    }
}