update the orphans filter

Author: mlebkowski

src/DependencyInjection/Configuration.php

@@ -61,6 +61,9 @@ public function getConfigTreeBuilder()
         $srcset->children()->integerNode('image_width_threshold')->defaultValue(100);
         $srcset->children()->arrayNode('breakpoints')->prototype('scalar');
 
+        $typography = $rootNode->children()->arrayNode('typography')->addDefaultsIfNotSet();
+        $typography->children()->scalarNode('orphans')->defaultNull();
+
         return $treeBuilder;
 
     }

src/DependencyInjection/KunstmaanExtraExtension.php

@@ -75,6 +75,8 @@ public function load(array $configs, ContainerBuilder $container)
             $loader->load('imagine_imgix_services.yml');
         }
 
+        $container->setParameter('kunstmaan_extra.typography.orphans', $configs['typography']['orphans']);
+
         if ($configs['generate_controller']) {
             $container->setDefinition('kunstmaan_extra.typehinting_controller_cache_warmer', (new Definition)
                 ->setClass(TypehintingControllerCacheWarmer::class)

src/Resources/config/twig_extensions.yml

@@ -74,6 +74,7 @@ services:
 
     ars_thanea_kunstmaan_extra.twig.typography_twig_extension:
         class: '%ars_thanea_kunstmaan_extra.twig.typography_twig_extension.class%'
+        arguments: ['%kunstmaan_extra.typography.orphans%']
         public: false
         tags:
             - { name: twig.extension }

src/Resources/doc/misc.md

@@ -128,6 +128,22 @@ Replaces phases like `a cat` with `a cat` to keep them in the same line.
 }}
 ```
 
+This filter tries very hard not to break HTML. You can override the used regex with `kunstmaan_extra.typography.orphans` setting; the default is:
+
+```regexp
+/
+    \b            # start at a word boundary
+    (?<!\<)       # exclude opening HTML tags
+    (             # first matching group is important for replace
+        \w{1,2}\.?      # orphan optionally followed by a period
+        (?:
+            <\/?\w+>    # don’t let a closing HTML tag brake the match
+        )*?
+    )
+    \s+           # match one or many whitespace characters
+/umx
+```
+
 ## AdminList Generator
 
 To make routes work, add this entry to your bundle routing config: 

src/Twig/TypographyTwigExtension.php

@@ -4,6 +4,24 @@
 
 class TypographyTwigExtension extends \Twig_Extension
 {
+    private $regex;
+
+    public function __construct($regex = null)
+    {
+        $this->regex = $regex ?: '/
+            \b            # start at a word boundary
+            (?<!\<)       # exclude opening HTML tags
+            (             # first matching group is important for replace
+                \w{1,2}\.?      # orphan optionally followed by a period
+                (?:
+                    <\/?\w+>    # don’t let a closing HTML tag brake the match
+                )*?
+            )
+            \s+           # match one or many whitespace characters
+        /umx';
+    }
+
+
     public function getFilters()
     {
         return [
@@ -20,7 +38,7 @@ public function orphans($text)
     {
         $nonBreakingSpace = ' ';
 
-        return preg_replace("/\\b(\\w{1,2}\\.?) /um", '$1' . $nonBreakingSpace, $text);
+        return preg_replace($this->regex, '$1' . $nonBreakingSpace, $text);
     }
 
 

tests/Typography/TypographyTwigExtensionTest.php

@@ -0,0 +1,38 @@
+<?php
+
+namespace ArsThanea\KunstmaanExtraBundle\Typography;
+
+use ArsThanea\KunstmaanExtraBundle\Twig\TypographyTwigExtension;
+
+class TypographyTwigExtensionTest extends \PHPUnit_Framework_TestCase
+{
+    /**
+     * @param string $input
+     * @param string $expected
+     * @dataProvider data
+     */
+    public function testOrphans($input, $expected)
+    {
+        $service = new TypographyTwigExtension();
+
+        $result = $service->orphans($input);
+
+        $this->assertEquals($expected, $result);
+    }
+
+    public function data()
+    {
+        return [
+            ['this is an orphan', 'this is an orphan'],
+            ['<a href="http://foo.bar">baz a bar</a>', '<a href="http://foo.bar">baz a bar</a>'],
+            ['<a title="this is an orphan">an orphan</a>', '<a title="this is an orphan">an orphan</a>', ],
+            ['<span>this is</span><a> an orphan', '<span>this is</span><a> an orphan'],
+            ['<span>this is</span></a> an orphan', '<span>this is</span></a> an orphan'],
+            ['is <a href="foo">bar</a>', 'is <a href="foo">bar</a>'],
+            ['<span>this is</span></a> an orphan <span>other a</span> orphan', '<span>this is</span></a> an orphan <span>other a</span> orphan'],
+            ['this is  an  orphan', 'this is an orphan'],
+            ['10 000', '10 000'],
+            ['foo an bar', 'foo an bar'],
+        ];
+    }
+}