From 84a073e6751c06758e2283931db188abe1df99a7 Mon Sep 17 00:00:00 2001
From: Colin Tucker <colin@praxis.net.au>
Date: Tue, 22 May 2018 14:57:04 +1000
Subject: [PATCH] Improved method of generating summary from content field.

---
 src/Extensions/Model/MetaDataExtension.php |   7 +-
 src/Tools/StringTools.php                  | 152 +++++++++++++++++++++
 2 files changed, 156 insertions(+), 3 deletions(-)
 create mode 100644 src/Tools/StringTools.php

diff --git a/src/Extensions/Model/MetaDataExtension.php b/src/Extensions/Model/MetaDataExtension.php
index 7ff89be..0e4ce08 100644
--- a/src/Extensions/Model/MetaDataExtension.php
+++ b/src/Extensions/Model/MetaDataExtension.php
@@ -31,6 +31,7 @@
 use SilverWare\Forms\DimensionsField;
 use SilverWare\Forms\FieldSection;
 use SilverWare\Tools\ImageTools;
+use SilverWare\Tools\StringTools;
 use SilverWare\Tools\ViewTools;
 use SilverWare\View\GridAware;
 
@@ -599,7 +600,7 @@ public function getMetaSummary()
         }
         
         if ($content = $this->owner->getMetaContent()) {
-            return DBField::create_field('HTMLFragment', sprintf('<p>%s</p>', $this->getContentSummary($content)));
+            return DBField::create_field('HTMLFragment', sprintf('<p>%s</p>', $this->owner->getContentSummary($content)));
         }
     }
     
@@ -1141,7 +1142,7 @@ public function getFieldFromParent($name)
      *
      * @return string
      */
-    protected function getContentSummary(DBHTMLText $content)
+    public function getContentSummary(DBHTMLText $content)
     {
         switch ($this->owner->SummaryMode) {
             case self::SUMMARY_FIRST_SENTENCE:
@@ -1149,7 +1150,7 @@ protected function getContentSummary(DBHTMLText $content)
             case self::SUMMARY_FIRST_PARAGRAPH:
                 return $content->FirstParagraph();
             default:
-                return $content->Summary();
+                return StringTools::singleton()->getContentSummary($content);
         }
     }
     
diff --git a/src/Tools/StringTools.php b/src/Tools/StringTools.php
new file mode 100644
index 0000000..3f8dfa1
--- /dev/null
+++ b/src/Tools/StringTools.php
@@ -0,0 +1,152 @@
+<?php
+
+/**
+ * This file is part of SilverWare.
+ *
+ * PHP version >=5.6.0
+ *
+ * For full copyright and license information, please view the
+ * LICENSE.md file that was distributed with this source code.
+ *
+ * @package SilverWare\Tools
+ * @author Colin Tucker <colin@praxis.net.au>
+ * @copyright 2018 Praxis Interactive
+ * @license https://opensource.org/licenses/BSD-3-Clause BSD-3-Clause
+ * @link https://github.com/praxisnetau/silverware
+ */
+
+namespace SilverWare\Tools;
+
+use SilverStripe\Core\Convert;
+use SilverStripe\Core\Injector\Injectable;
+use SilverStripe\ORM\FieldType\DBHTMLText;
+
+/**
+ * A singleton providing utility functions for use with strings.
+ *
+ * @package SilverWare\Tools
+ * @author Colin Tucker <colin@praxis.net.au>
+ * @copyright 2018 Praxis Interactive
+ * @license https://opensource.org/licenses/BSD-3-Clause BSD-3-Clause
+ * @link https://github.com/praxisnetau/silverware
+ */
+class StringTools
+{
+    use Injectable;
+    
+    /**
+     * Answers a summary of the given content field using an improved method of removing tags.
+     *
+     * @param DBHTMLText $content
+     * @param integer $maxWords
+     * @param string $add
+     *
+     * @return string
+     */
+    public function getContentSummary(DBHTMLText $content, $maxWords = 50, $add = '...')
+    {
+        // Get Plain Text Version:
+        
+        $value = $this->getContentAsPlainText($content);
+        
+        // Bail Early (if empty):
+        
+        if (!$value) {
+            return '';
+        }
+        
+        // Split on Sentences (do not remove period):
+        
+        $sentences = array_filter(array_map(function ($str) {
+            return trim($str);
+        }, preg_split('@(?<=\.)@', $value)));
+        
+        $wordCount = count(preg_split('#\s+#u', $sentences[0]));
+        
+        // If the First Sentence is Too Long, Show Only the First $maxWords Words:
+        
+        if ($wordCount > $maxWords) {
+            return implode(' ', array_slice(explode(' ', $sentences[0]), 0, $maxWords)) . $add;
+        }
+        
+        // Add Each Sentence (while there are enough words to do so):
+        
+        $result = '';
+        
+        do {
+            
+            // Add Next Sentence:
+            
+            $result .= ' ' . array_shift($sentences);
+            
+            // If More Sentences, Count Number of Words:
+            
+            if ($sentences) {
+                $wordCount += count(preg_split('#\s+#u', $sentences[0]));
+            }
+            
+        } while ($wordCount < $maxWords && $sentences && trim($sentences[0]));
+        
+        // Answer Result String:
+        
+        return trim($result);
+    }
+    
+    /**
+     * Converts the given content field to a plain text string using an improved method of removing tags.
+     *
+     * @param DBHTMLText $content
+     *
+     * @return string
+     */
+    public function getContentAsPlainText(DBHTMLText $content)
+    {
+        // Preserve Line Breaks:
+        
+        $text = preg_replace('/\<br(\s*)?\/?\>/i', "\n", $content->RAW());
+        
+        // Convert Paragraph Breaks to Multi-Lines:
+        
+        $text = preg_replace('/\<\/p\>/i', "\n\n", $text);
+        
+        // Remove HTML Tags:
+        
+        $text = $this->removeTags($text);
+        
+        // Implode >3 Consecutive Linebreaks into 2:
+        
+        $text = preg_replace('~(\R){2,}~', "\n\n", $text);
+        
+        // Decode HTML Entities Back to Plain Text:
+        
+        return trim(Convert::xml2raw($text));
+    }
+    
+    /**
+     * Removes HTML tags from the given string while maintaining whitespace.
+     *
+     * @param string $string
+     *
+     * @return string
+     */
+    public function removeTags($string)
+    {
+        // Remove HTML Tags:
+        
+        $string = preg_replace('/<[^>]*>/', ' ', $string);
+        
+        // Remove Control Characters:
+        
+        $string = str_replace("\r", '',  $string);  // replace with empty space
+        $string = str_replace("\n", ' ', $string); // replace with single space
+        $string = str_replace("\t", ' ', $string); // replace with single space
+        
+        // Remove Multiple Spaces:
+        
+        $string = trim(preg_replace('/ {2,}/', ' ', $string));
+        
+        // Answer String:
+        
+        return $string;
+    }
+}