ForeignResourceManager.php

Go to the documentation of this file.

<?php
namespace MediaWiki\ResourceLoader;
 
use Composer\Spdx\SpdxLicenses;
use Exception;
use MediaWiki\MainConfigNames;
use MediaWiki\MediaWikiServices;
use PharData;
use RecursiveDirectoryIterator;
use RecursiveIteratorIterator;
use SplFileInfo;
use Symfony\Component\Yaml\Yaml;
 
class ForeignResourceManager {
    private $defaultAlgo = 'sha384';
 
    private $hasErrors = false;
 
    private $registryFile;
 
    private $libDir;
 
    private $tmpParentDir;
 
    private $cacheDir;
 
    private $infoPrinter;
 
    private $errorPrinter;
    private $verbosePrinter;
 
    private $action;
 
    private $registry;
 
    public function __construct(
        $registryFile,
        $libDir,
        callable $infoPrinter = null,
        callable $errorPrinter = null,
        callable $verbosePrinter = null
    ) {
        $this->registryFile = $registryFile;
        $this->libDir = $libDir;
        $this->infoPrinter = $infoPrinter ?? static function ( $_ ) {
        };
        $this->errorPrinter = $errorPrinter ?? $this->infoPrinter;
        $this->verbosePrinter = $verbosePrinter ?? static function ( $_ ) {
        };
 
        // Support XDG_CACHE_HOME to speed up CI by avoiding repeated downloads.
        $conf = MediaWikiServices::getInstance()->getMainConfig();
        if ( ( $cacheHome = getenv( 'XDG_CACHE_HOME' ) ) !== false ) {
            $this->cacheDir = realpath( $cacheHome ) . '/mw-foreign';
        } elseif ( ( $cacheConf = $conf->get( MainConfigNames::CacheDirectory ) ) !== false ) {
            $this->cacheDir = "$cacheConf/ForeignResourceManager";
        } else {
            $this->cacheDir = "{$this->libDir}/.foreign/cache";
        }
    }
 
    public function run( $action, $module ) {
        $actions = [ 'update', 'verify', 'make-sri' ];
        if ( !in_array( $action, $actions ) ) {
            $this->error( "Invalid action.\n\nMust be one of " . implode( ', ', $actions ) . '.' );
            return false;
        }
        $this->action = $action;
        $this->setupTempDir( $action );
 
        $this->registry = Yaml::parseFile( $this->registryFile );
        if ( $module === 'all' ) {
            $modules = $this->registry;
        } elseif ( isset( $this->registry[$module] ) ) {
            $modules = [ $module => $this->registry[$module] ];
        } else {
            $this->error( "Unknown module name.\n\nMust be one of:\n" .
                wordwrap( implode( ', ', array_keys( $this->registry ) ), 80 ) .
                '.'
            );
            return false;
        }
 
        foreach ( $modules as $moduleName => $info ) {
            $this->verbose( "\n### {$moduleName}\n\n" );
 
            if ( $this->action === 'update' ) {
                $this->output( "... updating '{$moduleName}'\n" );
            } elseif ( $this->action === 'verify' ) {
                $this->output( "... verifying '{$moduleName}'\n" );
            } else {
                $this->output( "... checking '{$moduleName}'\n" );
            }
 
            // Do checks on yaml content (such as license existence, validity and type keys)
            // before doing any potentially destructive actions (potentially deleting directories,
            // depending on action.
 
            if ( !isset( $info['type'] ) ) {
                throw new Exception( "Module '$moduleName' must have a 'type' key." );
            }
 
            $this->validateLicense( $moduleName, $info );
 
            if ( $info['type'] === 'doc-only' ) {
                $this->output( "... {$moduleName} is documentation-only, skipping\n" );
                continue;
            }
 
            $destDir = "{$this->libDir}/$moduleName";
 
            if ( $this->action === 'update' ) {
                $this->verbose( "... emptying directory for $moduleName\n" );
                wfRecursiveRemoveDir( $destDir );
            }
 
            $this->verbose( "... preparing {$this->tmpParentDir}\n" );
            wfRecursiveRemoveDir( $this->tmpParentDir );
            if ( !wfMkdirParents( $this->tmpParentDir ) ) {
                throw new Exception( "Unable to create {$this->tmpParentDir}" );
            }
 
            switch ( $info['type'] ) {
                case 'tar':
                case 'zip':
                    $this->handleTypeTar( $moduleName, $destDir, $info, $info['type'] );
                    break;
                case 'file':
                    $this->handleTypeFile( $moduleName, $destDir, $info );
                    break;
                case 'multi-file':
                    $this->handleTypeMultiFile( $moduleName, $destDir, $info );
                    break;
                default:
                    throw new Exception( "Unknown type '{$info['type']}' for '$moduleName'" );
            }
 
            if ( $this->action === 'update' ) {
                foreach ( $info['transforms'] ?? [] as $file => $transforms ) {
                    $fullFilePath = "$destDir/$file";
                    if ( !file_exists( $fullFilePath ) ) {
                        throw new Exception( "$moduleName: invalid transform target $file" );
                    }
                    if ( !is_array( $transforms ) || !array_is_list( $transforms ) ) {
                        $transforms = [ $transforms ];
                    }
                    foreach ( $transforms as $transform ) {
                        if ( $transform === 'nomin' ) {
                            // not super efficient but these files aren't expected to be large
                            file_put_contents( $fullFilePath, "/*@nomin*/\n" . file_get_contents( $fullFilePath ) );
                        } else {
                            throw new Exception( "$moduleName: invalid transform $transform" );
                        }
                    }
                }
            }
        }
 
        $this->cleanUp();
        if ( $this->hasErrors ) {
            // The "verify" action should check all modules and files and fail after, not during.
            // We don't throw on the first issue so that developers enjoy access to all actionable
            // information at once (given we can't have cascading errors).
            // The "verify" action prints errors along the way and simply exits here.
            return false;
        }
 
        return true;
    }
 
    private function setupTempDir( $action ) {
        if ( $action === 'verify' ) {
            $this->tmpParentDir = wfTempDir() . '/ForeignResourceManager';
        } else {
            // Use a temporary directory under the destination directory instead
            // of wfTempDir() because PHP's rename() does not work across file
            // systems, and the user's /tmp and $IP may be on different filesystems.
            $this->tmpParentDir = "{$this->libDir}/.foreign/tmp";
        }
    }
 
    private function cacheKey( $src, $integrity, $moduleName ) {
        $key = $moduleName
            . '_' . hash( 'fnv132', $integrity )
            . '_' . hash( 'fnv132', $src )
            // Append readable filename to aid cache inspection and debugging
            . '_' . basename( $src );
        $key = preg_replace( '/[.\/+?=_-]+/', '_', $key );
        return rtrim( $key, '_' );
    }
 
    private function cacheGet( $key ) {
        // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged
        return @file_get_contents( "{$this->cacheDir}/$key.data" );
    }
 
    private function cacheSet( $key, $data ) {
        // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged
        @mkdir( $this->cacheDir, 0777, true );
        file_put_contents( "{$this->cacheDir}/$key.data", $data, LOCK_EX );
    }
 
    private function fetch( string $src, $integrity, string $moduleName ) {
        if ( $integrity !== null ) {
            $key = $this->cacheKey( $src, $integrity, $moduleName );
            $data = $this->cacheGet( $key );
            if ( $data ) {
                return $data;
            }
        }
 
        $req = MediaWikiServices::getInstance()->getHttpRequestFactory()
            ->create( $src, [ 'method' => 'GET', 'followRedirects' => false ], __METHOD__ );
        if ( !$req->execute()->isOK() ) {
            throw new Exception( "Failed to download resource at {$src}" );
        }
        if ( $req->getStatus() !== 200 ) {
            throw new Exception( "Unexpected HTTP {$req->getStatus()} response from {$src}" );
        }
        $data = $req->getContent();
        $algo = $integrity === null ? $this->defaultAlgo : explode( '-', $integrity )[0];
        $actualIntegrity = $algo . '-' . base64_encode( hash( $algo, $data, true ) );
        if ( $integrity === $actualIntegrity ) {
            $this->verbose( "... passed integrity check for {$src}\n" );
            $key = $this->cacheKey( $src, $actualIntegrity, $moduleName );
            $this->cacheSet( $key, $data );
        } elseif ( $this->action === 'make-sri' ) {
            $this->output( "Integrity for {$src}\n\tintegrity: {$actualIntegrity}\n" );
        } else {
            $expectedIntegrity = $integrity ?? 'null';
            throw new Exception( "Integrity check failed for {$src}\n" .
                "\tExpected: {$expectedIntegrity}\n" .
                "\tActual: {$actualIntegrity}"
            );
        }
        return $data;
    }
 
    private function handleTypeFile( $moduleName, $destDir, array $info ) {
        if ( !isset( $info['src'] ) ) {
            throw new Exception( "Module '$moduleName' must have a 'src' key." );
        }
        $data = $this->fetch( $info['src'], $info['integrity'] ?? null, $moduleName );
        $dest = $info['dest'] ?? basename( $info['src'] );
        $path = "$destDir/$dest";
        if ( $this->action === 'verify' && sha1_file( $path ) !== sha1( $data ) ) {
            $this->error( "File for '$moduleName' is different.\n" );
        }
        if ( $this->action === 'update' ) {
            wfMkdirParents( $destDir );
            file_put_contents( "$destDir/$dest", $data );
        }
    }
 
    private function handleTypeMultiFile( $moduleName, $destDir, array $info ) {
        if ( !isset( $info['files'] ) ) {
            throw new Exception( "Module '$moduleName' must have a 'files' key." );
        }
        foreach ( $info['files'] as $dest => $file ) {
            if ( !isset( $file['src'] ) ) {
                throw new Exception( "Module '$moduleName' file '$dest' must have a 'src' key." );
            }
            $data = $this->fetch( $file['src'], $file['integrity'] ?? null, $moduleName );
            $path = "$destDir/$dest";
            if ( $this->action === 'verify' && sha1_file( $path ) !== sha1( $data ) ) {
                $this->error( "File '$dest' for '$moduleName' is different.\n" );
            } elseif ( $this->action === 'update' ) {
                wfMkdirParents( $destDir );
                file_put_contents( "$destDir/$dest", $data );
            }
        }
    }
 
    private function handleTypeTar( $moduleName, $destDir, array $info, string $fileType ) {
        $info += [ 'src' => null, 'integrity' => null, 'dest' => null ];
        if ( $info['src'] === null ) {
            throw new Exception( "Module '$moduleName' must have a 'src' key." );
        }
        // Download the resource to a temporary file and open it
        $data = $this->fetch( $info['src'], $info['integrity'], $moduleName );
        $tmpFile = "{$this->tmpParentDir}/$moduleName." . $fileType;
        $this->verbose( "... writing '$moduleName' src to $tmpFile\n" );
        file_put_contents( $tmpFile, $data );
        $p = new PharData( $tmpFile );
        $tmpDir = "{$this->tmpParentDir}/$moduleName";
        $p->extractTo( $tmpDir );
        unset( $data, $p );
 
        if ( $info['dest'] === null ) {
            // Default: Replace the entire directory
            $toCopy = [ $tmpDir => $destDir ];
        } else {
            // Expand and normalise the 'dest' entries
            $toCopy = [];
            foreach ( $info['dest'] as $fromSubPath => $toSubPath ) {
                // Use glob() to expand wildcards and check existence
                $fromPaths = glob( "{$tmpDir}/{$fromSubPath}", GLOB_BRACE );
                if ( !$fromPaths ) {
                    throw new Exception( "Path '$fromSubPath' of '$moduleName' not found." );
                }
                foreach ( $fromPaths as $fromPath ) {
                    $toCopy[$fromPath] = $toSubPath === null
                        ? "$destDir/" . basename( $fromPath )
                        : "$destDir/$toSubPath/" . basename( $fromPath );
                }
            }
        }
        foreach ( $toCopy as $from => $to ) {
            if ( $this->action === 'verify' ) {
                $this->verbose( "... verifying $to\n" );
                if ( is_dir( $from ) ) {
                    $rii = new RecursiveIteratorIterator( new RecursiveDirectoryIterator(
                        $from,
                        RecursiveDirectoryIterator::SKIP_DOTS
                    ) );
                    foreach ( $rii as $file ) {
                        $remote = $file->getPathname();
                        $local = strtr( $remote, [ $from => $to ] );
                        if ( sha1_file( $remote ) !== sha1_file( $local ) ) {
                            $this->error( "File '$local' is different.\n" );
                        }
                    }
                } elseif ( sha1_file( $from ) !== sha1_file( $to ) ) {
                    $this->error( "File '$to' is different.\n" );
                }
            } elseif ( $this->action === 'update' ) {
                $this->verbose( "... moving $from to $to\n" );
                wfMkdirParents( dirname( $to ) );
                if ( !rename( $from, $to ) ) {
                    throw new Exception( "Could not move $from to $to." );
                }
            }
        }
    }
 
    private function verbose( $text ) {
        ( $this->verbosePrinter )( $text );
    }
 
    private function output( $text ) {
        ( $this->infoPrinter )( $text );
    }
 
    private function error( $text ) {
        $this->hasErrors = true;
        ( $this->errorPrinter )( $text );
    }
 
    private function cleanUp() {
        wfRecursiveRemoveDir( $this->tmpParentDir );
 
        // Prune the cache of files we don't recognise.
        $knownKeys = [];
        foreach ( $this->registry as $module => $info ) {
            if ( $info['type'] === 'file' || $info['type'] === 'tar' ) {
                $knownKeys[] = $this->cacheKey( $info['src'], $info['integrity'], $module );
            } elseif ( $info['type'] === 'multi-file' ) {
                foreach ( $info['files'] as $file ) {
                    $knownKeys[] = $this->cacheKey( $file['src'], $file['integrity'], $module );
                }
            }
        }
        foreach ( glob( "{$this->cacheDir}/*" ) as $cacheFile ) {
            if ( !in_array( basename( $cacheFile, '.data' ), $knownKeys ) ) {
                unlink( $cacheFile );
            }
        }
    }
 
    private function validateLicense( $moduleName, $info ) {
        if ( !isset( $info['license'] ) || !is_string( $info['license'] ) ) {
            throw new Exception( "Module '$moduleName' needs a valid SPDX license; no license is currently present" );
        }
        $licenses = new SpdxLicenses();
        if ( !$licenses->validate( $info['license'] ) ) {
            $this->error(
                "Module '$moduleName' has an invalid SPDX license identifier '{$info['license']}', "
                . 'see <https://spdx.org/licenses/>'
            );
        }
    }
}
 
class_alias( ForeignResourceManager::class, 'ForeignResourceManager' );