Added "page_include_parse" theme event

[bookstack] / tests / Entity / PageContentTest.php

diff --git a/tests/Entity/PageContentTest.php b/tests/Entity/PageContentTest.php
index ca2a241314baa7082f2101c7717a72d1c36c32ee..f88e4d513f02d5ce9f188025834b8387551b97ef 100644 (file)
--- a/tests/Entity/PageContentTest.php
+++ b/tests/Entity/PageContentTest.php
@@ -62,7 +62,9 @@ class PageContentTest extends TestCase
 
     public function test_page_includes_do_not_break_tables()
     {
+        /** @var Page $page */
         $page = Page::query()->first();
+        /** @var Page $secondPage */
         $secondPage = Page::query()->where('id', '!=', $page->id)->first();
 
         $content = '<table id="table"><tbody><tr><td>test</td></tr></tbody></table>';
@@ -72,9 +74,26 @@ class PageContentTest extends TestCase
         $page->html = "{{@{$secondPage->id}#table}}";
         $page->save();
 
-        $this->asEditor();
-        $pageResp = $this->get($page->getUrl());
-        $pageResp->assertSee($content);
+        $pageResp = $this->asEditor()->get($page->getUrl());
+        $pageResp->assertSee($content, false);
+    }
+
+    public function test_page_includes_do_not_break_code()
+    {
+        /** @var Page $page */
+        $page = Page::query()->first();
+        /** @var Page $secondPage */
+        $secondPage = Page::query()->where('id', '!=', $page->id)->first();
+
+        $content = '<pre id="bkmrk-code"><code>var cat = null;</code></pre>';
+        $secondPage->html = $content;
+        $secondPage->save();
+
+        $page->html = "{{@{$secondPage->id}#bkmrk-code}}";
+        $page->save();
+
+        $pageResp = $this->asEditor()->get($page->getUrl());
+        $pageResp->assertSee($content, false);
     }
 
     public function test_page_includes_rendered_on_book_export()
@@ -106,7 +125,7 @@ class PageContentTest extends TestCase
 
         $pageView = $this->get($page->getUrl());
         $pageView->assertStatus(200);
-        $pageView->assertDontSee($script);
+        $pageView->assertDontSee($script, false);
         $pageView->assertSee('abc123abc123');
     }
 
@@ -130,8 +149,8 @@ class PageContentTest extends TestCase
 
             $pageView = $this->get($page->getUrl());
             $pageView->assertStatus(200);
-            $pageView->assertElementNotContains('.page-content', '<script>');
-            $pageView->assertElementNotContains('.page-content', '</script>');
+            $this->withHtml($pageView)->assertElementNotContains('.page-content', '<script>');
+            $this->withHtml($pageView)->assertElementNotContains('.page-content', '</script>');
         }
     }
 
@@ -166,13 +185,14 @@ class PageContentTest extends TestCase
 
             $pageView = $this->get($page->getUrl());
             $pageView->assertStatus(200);
-            $pageView->assertElementNotContains('.page-content', '<iframe>');
-            $pageView->assertElementNotContains('.page-content', '<img');
-            $pageView->assertElementNotContains('.page-content', '</iframe>');
-            $pageView->assertElementNotContains('.page-content', 'src=');
-            $pageView->assertElementNotContains('.page-content', 'javascript:');
-            $pageView->assertElementNotContains('.page-content', 'data:');
-            $pageView->assertElementNotContains('.page-content', 'base64');
+            $html = $this->withHtml($pageView);
+            $html->assertElementNotContains('.page-content', '<iframe>');
+            $html->assertElementNotContains('.page-content', '<img');
+            $html->assertElementNotContains('.page-content', '</iframe>');
+            $html->assertElementNotContains('.page-content', 'src=');
+            $html->assertElementNotContains('.page-content', 'javascript:');
+            $html->assertElementNotContains('.page-content', 'data:');
+            $html->assertElementNotContains('.page-content', 'base64');
         }
     }
 
@@ -194,8 +214,8 @@ class PageContentTest extends TestCase
 
             $pageView = $this->get($page->getUrl());
             $pageView->assertStatus(200);
-            $pageView->assertElementNotContains('.page-content', '<a id="xss"');
-            $pageView->assertElementNotContains('.page-content', 'href=javascript:');
+            $this->withHtml($pageView)->assertElementNotContains('.page-content', '<a id="xss"');
+            $this->withHtml($pageView)->assertElementNotContains('.page-content', 'href=javascript:');
         }
     }
 
@@ -218,11 +238,11 @@ class PageContentTest extends TestCase
 
             $pageView = $this->get($page->getUrl());
             $pageView->assertStatus(200);
-            $pageView->assertElementNotContains('.page-content', '<button id="xss"');
-            $pageView->assertElementNotContains('.page-content', '<input id="xss"');
-            $pageView->assertElementNotContains('.page-content', '<form id="xss"');
-            $pageView->assertElementNotContains('.page-content', 'action=javascript:');
-            $pageView->assertElementNotContains('.page-content', 'formaction=javascript:');
+            $this->withHtml($pageView)->assertElementNotContains('.page-content', '<button id="xss"');
+            $this->withHtml($pageView)->assertElementNotContains('.page-content', '<input id="xss"');
+            $this->withHtml($pageView)->assertElementNotContains('.page-content', '<form id="xss"');
+            $this->withHtml($pageView)->assertElementNotContains('.page-content', 'action=javascript:');
+            $this->withHtml($pageView)->assertElementNotContains('.page-content', 'formaction=javascript:');
         }
     }
 
@@ -243,10 +263,10 @@ class PageContentTest extends TestCase
 
             $pageView = $this->get($page->getUrl());
             $pageView->assertStatus(200);
-            $pageView->assertElementNotContains('.page-content', '<meta>');
-            $pageView->assertElementNotContains('.page-content', '</meta>');
-            $pageView->assertElementNotContains('.page-content', 'content=');
-            $pageView->assertElementNotContains('.page-content', 'external_url');
+            $this->withHtml($pageView)->assertElementNotContains('.page-content', '<meta>');
+            $this->withHtml($pageView)->assertElementNotContains('.page-content', '</meta>');
+            $this->withHtml($pageView)->assertElementNotContains('.page-content', 'content=');
+            $this->withHtml($pageView)->assertElementNotContains('.page-content', 'external_url');
         }
     }
 
@@ -260,8 +280,8 @@ class PageContentTest extends TestCase
 
         $pageView = $this->get($page->getUrl());
         $pageView->assertStatus(200);
-        $pageView->assertDontSee($script);
-        $pageView->assertSee('<p>Hello</p>');
+        $pageView->assertDontSee($script, false);
+        $pageView->assertSee('<p>Hello</p>', false);
     }
 
     public function test_more_complex_inline_on_attributes_escaping_scenarios()
@@ -286,7 +306,7 @@ class PageContentTest extends TestCase
 
             $pageView = $this->get($page->getUrl());
             $pageView->assertStatus(200);
-            $pageView->assertElementNotContains('.page-content', 'onclick');
+            $this->withHtml($pageView)->assertElementNotContains('.page-content', 'onclick');
         }
     }
 
@@ -301,15 +321,18 @@ class PageContentTest extends TestCase
         $page->save();
 
         $pageView = $this->get($page->getUrl());
-        $pageView->assertSee($script);
+        $pageView->assertSee($script, false);
         $pageView->assertDontSee('abc123abc123');
     }
 
-    public function test_svg_xlink_hrefs_are_removed()
+    public function test_svg_script_usage_is_removed()
     {
         $checks = [
             '<svg id="test" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" width="100" height="100"><a xlink:href="javascript:alert(document.domain)"><rect x="0" y="0" width="100" height="100" /></a></svg>',
             '<svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink"><use xlink:href="data:application/xml;base64 ,PHN2ZyB4bWxucz0iaHR0cDovL3d3dy53My5vcmcvMjAwMC9zdmciIHhtbG5zOnhsaW5rPSJodHRwOi8vd3d3LnczLm9yZy8xOTk5L3hsaW5rIj4KPGRlZnM+CjxjaXJjbGUgaWQ9InRlc3QiIHI9IjAiIGN4PSIwIiBjeT0iMCIgc3R5bGU9ImZpbGw6ICNGMDAiPgo8c2V0IGF0dHJpYnV0ZU5hbWU9ImZpbGwiIGF0dHJpYnV0ZVR5cGU9IkNTUyIgb25iZWdpbj0nYWxlcnQoZG9jdW1lbnQuZG9tYWluKScKb25lbmQ9J2FsZXJ0KCJvbmVuZCIpJyB0bz0iIzAwRiIgYmVnaW49IjBzIiBkdXI9Ijk5OXMiIC8+CjwvY2lyY2xlPgo8L2RlZnM+Cjx1c2UgeGxpbms6aHJlZj0iI3Rlc3QiLz4KPC9zdmc+#test"/></svg>',
+            '<svg><animate href=#xss attributeName=href values=javascript:alert(1) /></svg>',
+            '<svg><animate href="#xss" attributeName="href" values="a;javascript:alert(1)" /></svg>',
+            '<svg><animate href="#xss" attributeName="href" values="a;data:alert(1)" /></svg>',
         ];
 
         $this->asEditor();
@@ -321,9 +344,11 @@ class PageContentTest extends TestCase
 
             $pageView = $this->get($page->getUrl());
             $pageView->assertStatus(200);
-            $pageView->assertElementNotContains('.page-content', 'alert');
-            $pageView->assertElementNotContains('.page-content', 'xlink:href');
-            $pageView->assertElementNotContains('.page-content', 'application/xml');
+            $html = $this->withHtml($pageView);
+            $html->assertElementNotContains('.page-content', 'alert');
+            $html->assertElementNotContains('.page-content', 'xlink:href');
+            $html->assertElementNotContains('.page-content', 'application/xml');
+            $html->assertElementNotContains('.page-content', 'javascript');
         }
     }
 
@@ -338,8 +363,8 @@ class PageContentTest extends TestCase
         $page->save();
 
         $pageView = $this->get($page->getUrl());
-        $pageView->assertSee($script);
-        $pageView->assertDontSee('<p>Hello</p>');
+        $pageView->assertSee($script, false);
+        $pageView->assertDontSee('<p>Hello</p>', false);
     }
 
     public function test_duplicate_ids_does_not_break_page_render()
@@ -487,7 +512,7 @@ class PageContentTest extends TestCase
         $this->assertStringContainsString('</tbody>', $page->html);
 
         $pageView = $this->get($page->getUrl());
-        $pageView->assertElementExists('.page-content table tbody td');
+        $this->withHtml($pageView)->assertElementExists('.page-content table tbody td');
     }
 
     public function test_page_markdown_task_list_rendering()
@@ -507,8 +532,8 @@ class PageContentTest extends TestCase
         $this->assertStringContainsString('type="checkbox"', $page->html);
 
         $pageView = $this->get($page->getUrl());
-        $pageView->assertElementExists('.page-content li.task-list-item input[type=checkbox]');
-        $pageView->assertElementExists('.page-content li.task-list-item input[type=checkbox][checked=checked]');
+        $this->withHtml($pageView)->assertElementExists('.page-content li.task-list-item input[type=checkbox]');
+        $this->withHtml($pageView)->assertElementExists('.page-content li.task-list-item input[type=checkbox][checked]');
     }
 
     public function test_page_markdown_strikethrough_rendering()
@@ -526,7 +551,7 @@ class PageContentTest extends TestCase
         $this->assertStringMatchesFormat('%A<s%A>some crossed out text</s>%A', $page->html);
 
         $pageView = $this->get($page->getUrl());
-        $pageView->assertElementExists('.page-content p > s');
+        $this->withHtml($pageView)->assertElementExists('.page-content p > s');
     }
 
     public function test_page_markdown_single_html_comment_saving()
@@ -545,7 +570,7 @@ class PageContentTest extends TestCase
 
         $pageView = $this->get($page->getUrl());
         $pageView->assertStatus(200);
-        $pageView->assertSee($content);
+        $pageView->assertSee($content, false);
     }
 
     public function test_base64_images_get_extracted_from_page_content()
@@ -596,16 +621,24 @@ class PageContentTest extends TestCase
 
     public function test_base64_images_within_html_blanked_if_not_supported_extension_for_extract()
     {
-        $this->asEditor();
-        $page = Page::query()->first();
+        // Relevant to https://github.com/BookStackApp/BookStack/issues/3010 and other cases
+        $extensions = [
+            'jiff', 'pngr', 'png ', ' png', '.png', 'png.', 'p.ng', ',png',
+            'data:image/png', ',data:image/png',
+        ];
 
-        $this->put($page->getUrl(), [
-            'name' => $page->name, 'summary' => '',
-            'html' => '<p>test<img src="data:image/jiff;base64,' . $this->base64Jpeg . '"/></p>',
-        ]);
+        foreach ($extensions as $extension) {
+            $this->asEditor();
+            $page = Page::query()->first();
 
-        $page->refresh();
-        $this->assertStringContainsString('<img src=""', $page->html);
+            $this->put($page->getUrl(), [
+                'name' => $page->name, 'summary' => '',
+                'html' => '<p>test<img src="data:image/' . $extension . ';base64,' . $this->base64Jpeg . '"/></p>',
+            ]);
+
+            $page->refresh();
+            $this->assertStringContainsString('<img src=""', $page->html);
+        }
     }
 
     public function test_base64_images_get_extracted_from_markdown_page_content()
@@ -630,17 +663,78 @@ class PageContentTest extends TestCase
         $this->deleteImage($imagePath);
     }
 
-    public function test_base64_images_within_markdown_blanked_if_not_supported_extension_for_extract()
+    public function test_markdown_base64_extract_not_limited_by_pcre_limits()
     {
+        $pcreBacktrackLimit = ini_get('pcre.backtrack_limit');
+        $pcreRecursionLimit = ini_get('pcre.recursion_limit');
+
         $this->asEditor();
         $page = Page::query()->first();
 
+        ini_set('pcre.backtrack_limit', '500');
+        ini_set('pcre.recursion_limit', '500');
+
+        $content = str_repeat('a', 5000);
+        $base64Content = base64_encode($content);
+
         $this->put($page->getUrl(), [
             'name'     => $page->name, 'summary' => '',
-            'markdown' => 'test ![test](data:image/jiff;base64,' . $this->base64Jpeg . ')',
+            'markdown' => 'test ![test](data:image/jpeg;base64,' . $base64Content . ') ![test](data:image/jpeg;base64,' . $base64Content . ')',
         ]);
 
         $page->refresh();
-        $this->assertStringContainsString('<img src=""', $page->html);
+        $this->assertStringMatchesFormat('<p%A>test <img src="http://localhost/uploads/images/gallery/%A.jpeg" alt="test"> <img src="http://localhost/uploads/images/gallery/%A.jpeg" alt="test">%A</p>%A', $page->html);
+
+        $matches = [];
+        preg_match('/src="https:\/\/p.rizon.top:443\/http\/localhost(.*?)"/', $page->html, $matches);
+        $imagePath = $matches[1];
+        $imageFile = public_path($imagePath);
+        $this->assertEquals($content, file_get_contents($imageFile));
+
+        $this->deleteImage($imagePath);
+        ini_set('pcre.backtrack_limit', $pcreBacktrackLimit);
+        ini_set('pcre.recursion_limit', $pcreRecursionLimit);
+    }
+
+    public function test_base64_images_within_markdown_blanked_if_not_supported_extension_for_extract()
+    {
+        $page = Page::query()->first();
+
+        $this->asEditor()->put($page->getUrl(), [
+            'name'     => $page->name, 'summary' => '',
+            'markdown' => 'test ![test](data:image/jiff;base64,' . $this->base64Jpeg . ')',
+        ]);
+
+        $this->assertStringContainsString('<img src=""', $page->refresh()->html);
+    }
+
+    public function test_nested_headers_gets_assigned_an_id()
+    {
+        $page = Page::query()->first();
+
+        $content = '<table><tbody><tr><td><h5>Simple Test</h5></td></tr></tbody></table>';
+        $this->asEditor()->put($page->getUrl(), [
+            'name'    => $page->name,
+            'html'    => $content,
+        ]);
+
+        // The top level <table> node will get assign the bkmrk-simple-test id because the system will
+        // take the node value of h5
+        // So the h5 should get the bkmrk-simple-test-1 id
+        $this->assertStringContainsString('<h5 id="bkmrk-simple-test-1">Simple Test</h5>', $page->refresh()->html);
+    }
+
+    public function test_non_breaking_spaces_are_preserved()
+    {
+        /** @var Page $page */
+        $page = Page::query()->first();
+
+        $content = '<p>&nbsp;</p>';
+        $this->asEditor()->put($page->getUrl(), [
+            'name'    => $page->name,
+            'html'    => $content,
+        ]);
+
+        $this->assertStringContainsString('<p id="bkmrk-%C2%A0">&nbsp;</p>', $page->refresh()->html);
     }
 }