Fix external media download

_test/ConfigTest.php

@@ -50,6 +50,7 @@ public function testDefaults(): void
         $this->assertNull($config->getSavedSelection(), 'default savedselection');
         $this->assertSame('', $config->getExportId(), 'default exportid');
         $this->assertTrue($config->useCache(), 'default usecache');
+        $this->assertSame(2 * 1024 * 1024, $config->getFetchSize(), 'default fetchsize');
 
         $this->assertSame('A4', $mpdfConfig['format'], 'default pagesize/orientation');
         $this->assertSame(11, $mpdfConfig['default_font_size'], 'default font-size');
@@ -86,6 +87,7 @@ public function testloadPluginConfig(): void
             'template' => 'modern',
             'output' => 'inline',
             'usestyles' => 'wrap,foo ',
+            'fetchsize' => 1024,
             'watermark' => 'CONFIDENTIAL',
             'qrcodescale' => '2.5',
             'debug' => 1,
@@ -111,6 +113,7 @@ public function testloadPluginConfig(): void
         $this->assertSame('modern', $config->getTemplateName(), 'from template');
         $this->assertSame('inline', $config->getOutputTarget(), 'from output');
         $this->assertSame(['wrap', 'foo'], $config->getStyledExtensions(), 'from usestyles');
+        $this->assertSame(1024, $config->getFetchSize(), 'from fetchsize');
         $this->assertSame('CONFIDENTIAL', $config->getWatermarkText(), 'from watermark');
         $this->assertTrue($mpdfConfig['showWatermarkText'], 'from watermark');
         $this->assertSame(2.5, $config->getQRScale(), 'from qrcodescale');

_test/DokuAssetFetcherTest.php

@@ -0,0 +1,123 @@
+<?php
+
+namespace dokuwiki\plugin\dw2pdf\test;
+
+use dokuwiki\plugin\dw2pdf\src\Config;
+use dokuwiki\plugin\dw2pdf\src\DokuMpdf;
+use DokuWikiTest;
+use Mpdf\AssetFetcherInterface;
+use Mpdf\Mpdf;
+use ReflectionProperty;
+
+/**
+ * Tests for handing resolved media files over to mpdf.
+ *
+ * @group plugin_dw2pdf
+ * @group plugins
+ */
+class DokuAssetFetcherTest extends DokuWikiTest
+{
+    /**
+     * Get the asset fetcher of a fresh DokuMpdf instance with the given basepath setting.
+     *
+     * Depending on basepathIsLocal mpdf uses either the first or the second argument of
+     * fetchDataFromPath(), so both cases need to be tested. The flag is false on any install
+     * where the baseurl host differs from the request host, a non-default port being enough.
+     *
+     * @param bool $basepathIsLocal Value to force for the mpdf flag
+     * @return AssetFetcherInterface
+     */
+    protected function getFetcher(bool $basepathIsLocal): AssetFetcherInterface
+    {
+        $mpdf = new DokuMpdf(new Config(), 'en');
+        $mpdf->basepathIsLocal = $basepathIsLocal;
+
+        return (new ReflectionProperty(Mpdf::class, 'assetFetcher'))->getValue($mpdf);
+    }
+
+    /**
+     * @return array<string, array{0:string,1:string,2:bool}>
+     */
+    public static function fetchProvider(): array
+    {
+        global $conf;
+
+        $media = DOKU_URL . 'lib/exe/fetch.php?media=wiki:dokuwiki-128.png';
+        $mediaFile = $conf['mediadir'] . '/wiki/dokuwiki-128.png';
+        $static = DOKU_URL . 'lib/images/throbber.gif';
+        $staticFile = DOKU_INC . 'lib/images/throbber.gif';
+
+        return [
+            'fetch url, local basepath' => [$media, $mediaFile, true],
+            'fetch url, remote basepath' => [$media, $mediaFile, false],
+            'static file, local basepath' => [$static, $staticFile, true],
+            'static file, remote basepath' => [$static, $staticFile, false],
+        ];
+    }
+
+    /**
+     * The local file has to be read regardless of the basepath setting.
+     *
+     * The second argument is the unmodified source as mpdf passes it for every image.
+     *
+     * @dataProvider fetchProvider
+     */
+    public function testFetchDataFromPathReadsLocalFile(string $source, string $expected, bool $basepathIsLocal): void
+    {
+        $data = $this->getFetcher($basepathIsLocal)->fetchDataFromPath($source, $source);
+
+        $this->assertSame(file_get_contents($expected), $data);
+    }
+
+    /**
+     * The resolved file has to win over the source mpdf was originally given.
+     *
+     * With a local basepath mpdf prefers the second argument whenever it can open it. For a
+     * media URL that means an anonymous HTTP request back into the wiki instead of our local
+     * copy. A readable file stands in for such a loadable source here.
+     */
+    public function testFetchDataFromPathIgnoresOriginalSource(): void
+    {
+        global $conf;
+        $file = $conf['mediadir'] . '/wiki/dokuwiki-128.png';
+        $source = DOKU_URL . 'lib/exe/fetch.php?media=wiki:dokuwiki-128.png';
+        $loadable = DOKU_INC . 'lib/images/throbber.gif';
+
+        $data = $this->getFetcher(true)->fetchDataFromPath($source, $loadable);
+
+        $this->assertSame(file_get_contents($file), $data);
+    }
+
+    /**
+     * Our pseudo scheme has no stream wrapper, so it can only ever be read locally.
+     *
+     * @testWith [true]
+     *           [false]
+     */
+    public function testFetchDataFromPathReadsDw2pdfScheme(bool $basepathIsLocal): void
+    {
+        global $conf;
+        $file = $conf['mediadir'] . '/wiki/dokuwiki-128.png';
+        $source = 'dw2pdf://' . $file;
+
+        $data = $this->getFetcher($basepathIsLocal)->fetchDataFromPath($source, $source);
+
+        $this->assertSame(file_get_contents($file), $data);
+    }
+
+    /**
+     * Media this wiki serves is never requested back over HTTP when it cannot be resolved.
+     *
+     * @testWith [true]
+     *           [false]
+     */
+    public function testFetchDataFromPathRefusesUnresolvableMedia(bool $basepathIsLocal): void
+    {
+        $source = DOKU_URL . 'lib/exe/fetch.php?media=wiki:no-such-image.png';
+        $this->expectLogMessage('Media not available for PDF export');
+
+        $data = $this->getFetcher($basepathIsLocal)->fetchDataFromPath($source, $source);
+
+        $this->assertSame('', $data);
+    }
+}

_test/HttpClientTest.php

@@ -0,0 +1,30 @@
+<?php
+
+namespace dokuwiki\plugin\dw2pdf\test;
+
+use dokuwiki\plugin\dw2pdf\src\Config;
+use dokuwiki\plugin\dw2pdf\src\HttpClient;
+use DokuWikiTest;
+use Mpdf\PsrHttpMessageShim\Request;
+
+/**
+ * Tests for the HTTP client mpdf loads remote assets through.
+ *
+ * @group plugin_dw2pdf
+ * @group plugins
+ */
+class HttpClientTest extends DokuWikiTest
+{
+    /**
+     * No remote host may be contacted when downloading is switched off.
+     */
+    public function testSendRequestRefusesWhenDownloadingIsDisabled(): void
+    {
+        $this->expectLogMessage('Remote asset not downloaded for PDF export');
+
+        $request = new Request('GET', 'https://example.org/does-not-matter.png');
+        $response = (new HttpClient(new Config(['fetchsize' => 0])))->sendRequest($request);
+
+        $this->assertSame(403, $response->getStatusCode());
+    }
+}

_test/MediaLinkResolverTest.php

@@ -2,6 +2,7 @@
 
 namespace dokuwiki\plugin\dw2pdf\test;
 
+use dokuwiki\plugin\dw2pdf\src\Config;
 use dokuwiki\plugin\dw2pdf\src\MediaLinkResolver;
 use DokuWikiTest;
 
@@ -18,7 +19,7 @@ class MediaLinkResolverTest extends DokuWikiTest
     public function setUp(): void
     {
         parent::setUp();
-        $this->resolver = new MediaLinkResolver();
+        $this->resolver = new MediaLinkResolver(new Config());
     }
 
     /**
@@ -80,16 +81,19 @@ public function testResolveDw2pdfScheme(): void
     }
 
     /**
+     * External media is downloaded even when the wiki does not allow fetch.php to do so.
+     *
      * @group internet
      */
     public function testResolveFetchesExternalMedia(): void
     {
         global $conf;
-        $conf['fetchsize'] = 512 * 1024; // 512 KB
+        $conf['fetchsize'] = 0;
+        $resolver = new MediaLinkResolver(new Config(['fetchsize' => 512 * 1024]));
 
         $external = 'https://php.net/images/php.gif';
         $input = DOKU_URL . 'lib/exe/fetch.php?media=' . rawurlencode($external);
-        $resolved = $this->resolver->resolve($input);
+        $resolved = $resolver->resolve($input);
 
         if ($resolved === null) {
             $this->markTestSkipped('External media fetching is not available in this environment.');
@@ -101,6 +105,85 @@ public function testResolveFetchesExternalMedia(): void
         $this->assertSame(2523, filesize($resolved['path']));
     }
 
+    /**
+     * Downloading external media can be switched off in the plugin configuration.
+     */
+    public function testResolveSkipsExternalMediaWhenDownloadingIsDisabled(): void
+    {
+        $resolver = new MediaLinkResolver(new Config(['fetchsize' => 0]));
+
+        $external = 'https://php.net/images/php.gif';
+        $input = DOKU_URL . 'lib/exe/fetch.php?media=' . rawurlencode($external);
+
+        $this->assertNull($resolver->resolve($input));
+    }
+
+    /**
+     * Media marked nocache is still downloaded.
+     *
+     * @group internet
+     */
+    public function testResolveFetchesNocacheExternalMedia(): void
+    {
+        $resolver = new MediaLinkResolver(new Config(['fetchsize' => 512 * 1024]));
+
+        $external = 'https://php.net/images/php.gif';
+        $input = DOKU_URL . 'lib/exe/fetch.php?media=' . rawurlencode($external) . '&cache=nocache';
+        $resolved = $resolver->resolve($input);
+
+        if ($resolved === null) {
+            $this->markTestSkipped('External media fetching is not available in this environment.');
+        }
+
+        $this->assertSame(2523, filesize($resolved['path']));
+    }
+
+    /**
+     * Without a cache parameter an existing copy is used no matter how old it is.
+     */
+    public function testResolveKeepsCachedExternalMediaByDefault(): void
+    {
+        $external = 'http://127.0.0.1:1/unreachable.png';
+        $cacheFile = $this->seedOutdatedCache($external);
+
+        $resolver = new MediaLinkResolver(new Config(['fetchsize' => 1024 * 1024]));
+        $input = DOKU_URL . 'lib/exe/fetch.php?media=' . rawurlencode($external);
+
+        $resolved = $resolver->resolve($input);
+
+        $this->assertNotNull($resolved);
+        $this->assertSame($cacheFile, $resolved['path']);
+    }
+
+    /**
+     * Media marked recache expires, so an outdated copy is fetched again.
+     */
+    public function testResolveExpiresCachedExternalMediaOnRecache(): void
+    {
+        $external = 'http://127.0.0.1:1/unreachable.png';
+        $this->seedOutdatedCache($external);
+
+        $resolver = new MediaLinkResolver(new Config(['fetchsize' => 1024 * 1024]));
+        $input = DOKU_URL . 'lib/exe/fetch.php?media=' . rawurlencode($external) . '&cache=recache';
+
+        $this->assertNull($resolver->resolve($input));
+    }
+
+    /**
+     * Write a week old cache file for the given external media URL.
+     *
+     * @param string $url The external media URL.
+     * @return string Absolute path to the cache file.
+     */
+    protected function seedOutdatedCache(string $url): string
+    {
+        $cacheFile = getCacheName(strtolower($url), '.media.png');
+        io_saveFile($cacheFile, 'outdated bytes');
+        touch($cacheFile, time() - 7 * 86400);
+
+        return $cacheFile;
+    }
+
     /**
      * Non-image payloads should never be returned to the PDF generator.
      */

conf/default.php

@@ -12,6 +12,7 @@
 $conf['template']         = 'default';
 $conf['output']           = 'browser';
 $conf['usecache']         = 1;
+$conf['fetchsize']        = 2 * 1024 * 1024;
 $conf['usestyles']        = 'wrap,';
 $conf['qrcodescale']       = '1';
 $conf['showexportbutton'] = 1;

conf/metadata.php

@@ -12,6 +12,7 @@
 $meta['template']         = array('dirchoice', '_dir' => DOKU_PLUGIN . 'dw2pdf/tpl/');
 $meta['output']           = array('multichoice', '_choices' => array('browser', 'file'));
 $meta['usecache']         = array('onoff');
+$meta['fetchsize']        = array('numeric');
 $meta['usestyles']        = array('string');
 $meta['qrcodescale']       = array('string', '_pattern' => '/^(|\d+(\.\d+)?)$/');
 $meta['showexportbutton'] = array('onoff');

lang/en/settings.php

@@ -22,6 +22,7 @@
 $lang['output_o_browser']        = 'Show in browser';
 $lang['output_o_file']           = 'Download the PDF';
 $lang['usecache']                = 'Should PDFs be cached? Embedded images won\'t be ACL checked then, disable if that\'s a security concern for you.';
+$lang['fetchsize']               = 'Maximum size (bytes) an export may download per remote file, such as an external image or font. <code>0</code> disables downloading, those files are then missing from the PDF.';
 $lang['usestyles']               = 'You can give a comma separated list of plugins of which the <code>style.css</code> or <code>screen.css</code> should be used for PDF generation. By default only <code>print.css</code> and <code>pdf.css</code> are used.';
 $lang['qrcodescale']             = 'Size scaling of the embedded QR code. Empty or <code>0</code> to disable.';
 $lang['showexportbutton']        = 'Show PDF export button (only when supported by your template)';

src/Config.php

@@ -35,6 +35,8 @@ class Config
     #[FromConfig]
     protected bool $usecache = true;
     #[FromConfig]
+    protected int $fetchSize = 2 * 1024 * 1024;
+    #[FromConfig]
     protected array $useStyles = [];
     #[FromConfig]
     protected float $qrCodeScale = 0.0;
@@ -196,6 +198,16 @@ public function useCache(): bool
         return $this->usecache;
     }
 
+    /**
+     * Get the maximum number of bytes to download per remote file
+     *
+     * @return int
+     */
+    public function getFetchSize(): int
+    {
+        return $this->fetchSize;
+    }
+
     /**
      * Get a list of extensions whose screen styles should be applied
      *