Merge pull request #112 from j0k3r/fix/1.x/tidy-php85

Fix bad encoding for `tidy_repair_string`

.github/workflows/coding-standards.yml

@@ -16,7 +16,7 @@ env:
 jobs:
   coding-standards:
     name: "CS Fixer & PHPStan"
-    runs-on: "ubuntu-22.04"
+    runs-on: "ubuntu-latest"
 
     strategy:
       matrix:

.github/workflows/continuous-integration.yml

@@ -3,11 +3,9 @@ name: "CI"
 on:
   pull_request:
     branches:
-      - "master"
       - "1.x"
   push:
     branches:
-      - "master"
       - "1.x"
 
 env:
@@ -16,14 +14,11 @@ env:
 jobs:
   phpunit:
     name: "PHPUnit (PHP ${{ matrix.php }})"
-    runs-on: "ubuntu-22.04"
+    runs-on: "ubuntu-latest"
 
     strategy:
       matrix:
         php:
-          - "5.6"
-          - "7.0"
-          - "7.1"
           - "7.2"
           - "7.3"
           - "7.4"
@@ -32,6 +27,7 @@ jobs:
           - "8.2"
           - "8.3"
           - "8.4"
+          - "8.5"
 
     steps:
       - name: "Checkout"
@@ -66,7 +62,7 @@ jobs:
 
   phpunit-coverage:
     name: "PHPUnit coverage (PHP ${{ matrix.php }})"
-    runs-on: "ubuntu-22.04"
+    runs-on: "ubuntu-latest"
 
     strategy:
       matrix:
@@ -117,12 +113,12 @@ jobs:
 
   phpunit-lowest:
     name: "PHPUnit lowest deps (PHP ${{ matrix.php }})"
-    runs-on: "ubuntu-22.04"
+    runs-on: "ubuntu-latest"
 
     strategy:
       matrix:
         php:
-          - "7.2"
+          - "7.4"
 
     steps:
       - name: "Checkout"
@@ -155,43 +151,3 @@ jobs:
 
       - name: "Run PHPUnit"
         run: "php vendor/bin/simple-phpunit -v"
-
-  phpunit-composerv2:
-    name: "PHPUnit with Composer v1 (PHP ${{ matrix.php }})"
-    runs-on: "ubuntu-20.04"
-
-    strategy:
-      matrix:
-        php:
-          - "7.4"
-
-    steps:
-      - name: "Checkout"
-        uses: "actions/checkout@v2"
-        with:
-          fetch-depth: 2
-
-      - name: "Install PHP"
-        uses: "shivammathur/setup-php@v2"
-        with:
-          php-version: "${{ matrix.php }}"
-          coverage: "none"
-          tools: composer:v1
-          extensions: tidy
-          ini-values: "date.timezone=Europe/Paris"
-        env:
-          COMPOSER_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-
-      - name: "Remove useless deps"
-        run: "composer remove friendsofphp/php-cs-fixer --dev --no-progress --no-update"
-
-      - name: "Install dependencies with Composer"
-        uses: "ramsey/composer-install@v1"
-        with:
-          composer-options: "--optimize-autoloader --prefer-dist"
-
-      - name: "Setup logs"
-        run: "mkdir -p build/logs"
-
-      - name: "Run PHPUnit"
-        run: "php vendor/bin/simple-phpunit -v"

composer.json

@@ -24,14 +24,14 @@
         "role": "Developer (original JS version)"
     }],
     "require": {
-        "php": ">=5.6.0",
+        "php": ">=7.2",
         "ext-mbstring": "*",
-        "psr/log": "^1.0",
+        "psr/log": "^1.0.1 || ^2.0 || ^3.0",
         "masterminds/html5": "^2.7"
     },
     "require-dev": {
         "friendsofphp/php-cs-fixer": "^2.14",
-        "monolog/monolog": "^1.24|^2.1",
+        "monolog/monolog": "^1.24|^2.1|^3.0",
         "symfony/phpunit-bridge": "^4.4|^5.3|^6.0|^7.0"
     },
     "suggest": {

src/Readability.php

@@ -118,7 +118,7 @@ class Readability implements LoggerAwareInterface
         'enclose-text' => true,
         'merge-divs' => true,
         // 'merge-spans' => true,
-        'input-encoding' => '????',
+        'input-encoding' => 'utf8',
         'output-encoding' => 'utf8',
         'hide-comments' => true,
     ];
@@ -183,10 +183,7 @@ class Readability implements LoggerAwareInterface
         $this->loadHtml();
     }
 
-    /**
+    public function setLogger(LoggerInterface $logger): void
-     * @return void
-     */
-    public function setLogger(LoggerInterface $logger)
     {
         $this->logger = $logger;
     }
@@ -1435,7 +1432,7 @@ class Readability implements LoggerAwareInterface
             unset($tidy);
         }
 
-        $this->html = self::ensureMetaCharset((string) $this->html);
+        $this->html = self::entitizeNonAscii((string) $this->html);
 
         if ('html5lib' === $this->parser || 'html5' === $this->parser) {
             $this->dom = (new HTML5())->loadHTML($this->html);
@@ -1455,43 +1452,19 @@ class Readability implements LoggerAwareInterface
     }
 
     /**
-     * Tries to insert `meta[charset]` tag into the proper place in the passed HTML document.
+     * Converts non-ASCII UTF-8 characters to numeric HTML entities.
      *
      * `DOMDocument::loadHTML` will parse HTML documents as ISO-8859-1 if there is no `meta[charset]` tag.
      * This means that UTF-8-encoded HTML fragments such as those coming from JSON-LD `articleBody` field would be parsed with incorrect encoding.
-     * Unfortunately, we cannot just put the tag at the start of the HTML fragment, since that would cause parser to auto-insert a `html` element, losing the attributes of the original `html` tag.
      *
      * @param string $html UTF-8 encoded document
      */
-    private static function ensureMetaCharset($html)
+    private static function entitizeNonAscii($html)
     {
-        $charsetTag = '<meta charset="utf-8">';
+        $convmap = [
+            0x80, 0x1FFFFF, 0, 0x10FFFF,
+        ];
 
-        // Only look at first 1024 bytes since, according to HTML5 specification,
+        return mb_encode_numericentity($html, $convmap, 'utf8', true);
-        // that’s where <meta> elements declaring a character encoding must be located.
-        // https://developer.mozilla.org/en-US/docs/Web/HTML/Element/meta#charset
-        $start = substr($html, 0, 1000);
-
-        if (1 === preg_match('/<meta[^>]+charset/i', $start)) {
-            // <meta> tag is already present, no need for modification.
-            return $html;
-        }
-
-        if (1 === preg_match('/<head[^>]*>/i', $start)) {
-            // <head> tag was located, <meta> tags go there.
-            $html = preg_replace('/<head[^>]*>/i', '$0' . $charsetTag, $html, 1);
-
-            return $html;
-        }
-
-        if (1 === preg_match('/<html[^>]*>/i', $start)) {
-            // <html> tag was located, let’s put it inside and have parser create <head>.
-            $html = preg_replace('/<html[^>]*>/i', '$0' . $charsetTag, $html, 1);
-
-            return $html;
-        }
-
-        // Fallback – just plop the <meta> at the start of the fragment.
-        return $charsetTag . $html;
     }
 }

tests/ReadabilityTest.php

@@ -17,6 +17,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
     public function testConstructDefault()
     {
         $readability = $this->getReadability('');
+        $this->assertSame('utf8', $readability->tidy_config['input-encoding']);
 
         $this->assertNull($readability->url);
         $this->assertInstanceOf(\DOMDocument::class, $readability->dom);
@@ -486,6 +487,20 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertStringContainsString('<a href="alice-I.html">Down the Rabbit-Hole</a>', $readability->getContent()->getInnerHtml());
     }
 
+    // https://github.com/wallabag/wallabag/issues/8158
+    public function testCharsetAfterTitle()
+    {
+        $readability = $this->getReadability('<!DOCTYPE html><html lang="et"><head><title>Tõde ja õigus I</title> <meta charset="utf-8"></head><body><p>See oli läinud aastasaja kolmanda veerandi lõpul. Päike lähenes silmapiirile, seistes sedavõrd madalas, et enam ei ulatunud valgustama ei mäkke ronivat hobust, kes puutelgedega vankrit vedas, ei vankril istuvat noort naist ega ka ligi kolmekümnelist meest, kes kõndis vankri kõrval.</p></body></html>', 'https://et.wikisource.org/wiki/T%C3%B5de_ja_%C3%B5igus_I/I');
+        $readability->convertLinksToFootnotes = true;
+        $res = $readability->init();
+
+        $this->assertTrue($res);
+        $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
+        $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
+        $this->assertSame('Tõde ja õigus I', $readability->getTitle()->getInnerHtml());
+        $this->assertStringContainsString('Päike lähenes', $readability->getContent()->getInnerHtml());
+    }
+
     /**
      * @return array<string, array{0: string, 1: string, 2?: bool}>
      */
@@ -493,21 +508,21 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
     {
         return [
             'meta' => [
-                '<html lang="fr"><head><meta charset="utf-8"></head><body><article>' . str_repeat('<p>This is the awesome content :)</p>', 7) . '</article></body></html>',
+                '<html lang="fr"><head><meta charset="utf-8"></head><body><article>' . str_repeat('<p>Tous les êtres humains naissent libres et égaux en dignité et en droits. Ils sont doués de raison et de conscience et doivent agir les uns envers les autres dans un esprit de fraternité.</p>', 7) . '</article></body></html>',
                 'fr',
             ],
             'head' => [
-                '<html lang="fr"><head><title>Foo</title></head><body><article>' . str_repeat('<p>This is the awesome content :)</p>', 7) . '</article></body></html>',
+                '<html lang="fr"><head><title>Foo</title></head><body><article>' . str_repeat('<p>Tous les êtres humains naissent libres et égaux en dignité et en droits. Ils sont doués de raison et de conscience et doivent agir les uns envers les autres dans un esprit de fraternité.</p>', 7) . '</article></body></html>',
                 'fr',
             ],
             'headless' => [
-                '<html lang="fr"><body><article>' . str_repeat('<p>This is the awesome content :)</p>', 7) . '</article></body></html>',
+                '<html lang="fr"><body><article>' . str_repeat('<p>Tous les êtres humains naissent libres et égaux en dignité et en droits. Ils sont doués de raison et de conscience et doivent agir les uns envers les autres dans un esprit de fraternité.</p>', 7) . '</article></body></html>',
                 'fr',
                 // tidy would add <head> tag.
                 false,
             ],
             'fragment' => [
-                '<article>' . str_repeat('<p>This is the awesome content :)</p>', 7) . '</article>',
+                '<article>' . str_repeat('<p>Tous les êtres humains naissent libres et égaux en dignité et en droits. Ils sont doués de raison et de conscience et doivent agir les uns envers les autres dans un esprit de fraternité.</p>', 7) . '</article>',
                 '',
                 // tidy would add <html>.
                 false,
@@ -526,6 +541,8 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf(\DOMDocument::class, $readability->dom);
         $this->assertSame($lang, $readability->dom->documentElement->getAttribute('lang'));
+        $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
+        $this->assertStringContainsString('êtres', $readability->getContent()->getInnerHtml());
     }
 
     private function getReadability($html, $url = null, $parser = 'libxml', $useTidy = true)