Merge pull request #57 from j0k3r/fix/html5-parser

Use a new deps for HTML5 parser

.editorconfig

@@ -4,8 +4,6 @@ root = true
 ; Unix-style newlines
 [*]
 end_of_line = LF
-
-[*.php]
 indent_style = space
 indent_size = 4
 trim_trailing_whitespace = true

.gitattributes

@@ -4,7 +4,6 @@
 /.scrutinizer.yml export-ignore
 /.travis.yml export-ignore
 /.php_cs export-ignore
-/README.md export-ignore
 /phpunit.xml.dist export-ignore
-/tests export-ignore
 /phpstan.neon export-ignore
+/tests export-ignore

.gitignore

@@ -2,3 +2,4 @@ vendor/
 coverage/
 composer.lock
 .php_cs.cache
+.phpunit.result.cache

.scrutinizer.yml

@@ -1,2 +1,12 @@
 tools:
     external_code_coverage: false
+
+build:
+    nodes:
+        analysis:
+            tests:
+                override:
+                    - php-scrutinizer-run
+            environment:
+                php:
+                    version: 7.2

.travis.yml

@@ -11,8 +11,10 @@ jobs:
     include:
         - php: 7.2
           env: SYMFONY_PHPUNIT_VERSION=7.5
+        - php: 7.2
+          env: SYMFONY_PHPUNIT_VERSION=7.5 COMPOSER_FLAGS="--prefer-lowest"
         - php: 7.3
-          env: SYMFONY_PHPUNIT_VERSION=7.5 CS_FIXER=run
+          env: SYMFONY_PHPUNIT_VERSION=7.5 CS_FIXER=run COVERAGE=run
         - php: 7.4
           env: SYMFONY_PHPUNIT_VERSION=7.5
         - php: nightly
@@ -21,30 +23,32 @@ jobs:
     allow_failures:
         - php: nightly
 
-# cache vendor dirs
 cache:
     directories:
-        - vendor
-        - $HOME/.composer/cache
+        - $HOME/.composer/cache/files
 
 before_install:
     - if [ -n "$GH_TOKEN" ]; then composer config github-oauth.github.com ${GH_TOKEN}; fi;
 
 install:
-    - composer self-update
+    # only install phpstan when we need it
+    - if [ "$CS_FIXER" = "run" ]; then composer require phpstan/phpstan phpstan/phpstan-phpunit --dev --no-progress --no-suggest ; fi;
+    - composer update --prefer-dist --no-progress --no-suggest -o $COMPOSER_FLAGS
+    - php vendor/bin/simple-phpunit install
 
 before_script:
-    - if [ "$CS_FIXER" = "run" ]; then composer require phpstan/phpstan phpstan/phpstan-phpunit --dev -n ; fi;
-    - composer install -o --prefer-dist --no-interaction
+    - mkdir -p build/logs
+    - if [ "$COVERAGE" = "run" ]; then PHPUNIT_FLAGS="--coverage-clover build/logs/clover.xml"; fi;
 
 script:
-    - mkdir -p build/logs
-    - php vendor/bin/simple-phpunit -v --coverage-clover build/logs/clover.xml
+    - php vendor/bin/simple-phpunit -v $PHPUNIT_FLAGS
     - if [ "$CS_FIXER" = "run" ]; then php vendor/bin/php-cs-fixer fix --verbose --dry-run ; fi;
-    # only install phpstan when we need it
-    - if [ "$CS_FIXER" = "run" ]; then composer require phpstan/phpstan --no-progress --no-suggest ; fi;
-    - if [ "$CS_FIXER" = "run" ]; then composer require phpstan/phpstan-phpunit --no-progress --no-suggest ; fi;
     - if [ "$CS_FIXER" = "run" ]; then php vendor/bin/phpstan analyse src tests --no-progress --level 1 ; fi;
 
-after_script:
-    - php vendor/bin/php-coveralls -v -x build/logs/clover.xml
+after_success:
+    - |
+        if [ "$COVERAGE" = "run" ]; then
+            wget https://github.com/php-coveralls/php-coveralls/releases/download/v2.2.0/php-coveralls.phar
+            chmod +x php-coveralls.phar
+            php php-coveralls.phar -v -x build/logs/clover.xml
+        fi

composer.json

@@ -27,12 +27,11 @@
         "php": ">=5.6.0",
         "ext-mbstring": "*",
         "psr/log": "^1.0",
-        "electrolinux/php-html5lib": "^0.1.0"
+        "masterminds/html5": "^2.7"
     },
     "require-dev": {
-        "php-coveralls/php-coveralls": "^2.1",
         "friendsofphp/php-cs-fixer": "^2.14",
-        "monolog/monolog": "^1.24",
+        "monolog/monolog": "^1.24|^2.1",
         "symfony/phpunit-bridge": "^4.2.3"
     },
     "suggest": {

phpunit.xml.dist

@@ -26,6 +26,6 @@
     </filter>
 
     <!-- <logging>
-        <log type="coverage-html" target="coverage" title="Readability" charset="UTF-8" yui="true" highlight="true" lowUpperBound="35" highLowerBound="70"/>
+        <log type="coverage-html" target="coverage" lowUpperBound="35" highLowerBound="70"/>
     </logging> -->
 </phpunit>

src/Readability.php

@@ -2,7 +2,7 @@
 
 namespace Readability;
 
-use HTML5Lib\Parser;
+use Masterminds\HTML5;
 use Psr\Log\LoggerAwareInterface;
 use Psr\Log\LoggerInterface;
 use Psr\Log\NullLogger;
@@ -616,7 +616,6 @@ class Readability implements LoggerAwareInterface
      */
     public function clean(\DOMElement $e, $tag)
     {
-        $currentItem = null;
         $targetList = $e->getElementsByTagName($tag);
         $isEmbed = ('audio' === $tag || 'video' === $tag || 'iframe' === $tag || 'object' === $tag || 'embed' === $tag);
 
@@ -657,7 +656,6 @@ class Readability implements LoggerAwareInterface
 
         $tagsList = $e->getElementsByTagName($tag);
         $curTagsLength = $tagsList->length;
-        $node = null;
 
         /*
          * Gather counts for other typical elements embedded within.
@@ -1108,7 +1106,6 @@ class Readability implements LoggerAwareInterface
          */
         if ($this->flagIsActive(self::FLAG_STRIP_UNLIKELYS) && $xpath) {
             $candidates = $xpath->query('.//*[(self::footer and count(//footer)<2) or (self::aside and count(//aside)<2)]', $page->documentElement);
-            $node = null;
 
             for ($c = $candidates->length - 1; $c >= 0; --$c) {
                 $node = $candidates->item($c);
@@ -1120,7 +1117,6 @@ class Readability implements LoggerAwareInterface
             }
 
             $candidates = $xpath->query('.//*[not(self::body) and (@class or @id or @style) and ((number(@readability) < 40) or not(@readability))]', $page->documentElement);
-            $node = null;
 
             for ($c = $candidates->length - 1; $c >= 0; --$c) {
                 $node = $candidates->item($c);
@@ -1424,8 +1420,8 @@ class Readability implements LoggerAwareInterface
 
         $this->html = mb_convert_encoding($this->html, 'HTML-ENTITIES', 'UTF-8');
 
-        if ('html5lib' === $this->parser) {
-            $this->dom = Parser::parse($this->html);
+        if ('html5lib' === $this->parser || 'html5' === $this->parser) {
+            $this->dom = (new HTML5())->loadHTML($this->html);
         }
 
         if ('libxml' === $this->parser) {