From d14bb89583e79bb60ac23b9964e66a216d4a4c72 Mon Sep 17 00:00:00 2001
From: Dennis Eichhorn <spl1nes.com@googlemail.com>
Date: Sat, 2 Sep 2017 20:41:09 +0200
Subject: [PATCH] fixes #70

---
 Utils/StringCompare.php | 168 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 168 insertions(+)
 create mode 100644 Utils/StringCompare.php

diff --git a/Utils/StringCompare.php b/Utils/StringCompare.php
new file mode 100644
index 000000000..c78fa86db
--- /dev/null
+++ b/Utils/StringCompare.php
@@ -0,0 +1,168 @@
+<?php
+/**
+ * Orange Management
+ *
+ * PHP Version 7.1
+ *
+ * @category   TBD
+ * @package    TBD
+ * @copyright  Dennis Eichhorn
+ * @license    OMS License 1.0
+ * @version    1.0.0
+ * @link       http://orange-management.com
+ */
+declare(strict_types=1);
+
+namespace phpOMS\Utils;
+
+/**
+ * String comparison class.
+ *
+ * This class helps to compare two strings
+ *
+ * @category   Framework
+ * @package    phpOMS\Utils
+ * @license    OMS License 1.0
+ * @link       http://orange-management.com
+ * @since      1.0.0
+ */
+class StringCompare
+{
+    /**
+     * Dictionary.
+     *
+     * @var array
+     * @since 1.0.0
+     */
+    private $dictionary = [];
+
+    /**
+     * Constructor.
+     * 
+     * @param array $dictionary Dictionary
+     *
+     * @since  1.0.0
+     */
+    public function __construct(array $dictionary)
+    {
+        $this->dictionary = $dictionary;
+    }
+
+    /**
+     * Match word against dictionary.
+     *
+     * @param string $match Word to match against dictionary
+     *
+     * @return string Best match
+     *
+     * @since  1.0.0
+     */
+    public function matchDictionary(string $match) : string
+    {
+        $bestScore = PHP_INT_MAX;
+        $bestMatch = '';
+
+        foreach($dictionary as $word) {
+            $score = self::fuzzyMatch($word, $match);
+
+            if($score < $bestScore) {
+                $bestMatch = $word;
+            }
+        }
+
+        return $bestMatch;
+    }
+
+    /**
+     * Calculate word match score.
+     *
+     * @param string $s1 Word 1
+     * @param string $s2 Word 2
+     *
+     * @return int
+     *
+     * @since  1.0.0
+     */
+    public static function valueWords(string $s1, string $s2) : int
+    {
+        $words1 = preg_split('/[ _-]/', $s1);
+        $words2 = preg_split('/[ _-]/', $s2);
+        $total = 0;
+
+        foreach($words1 as $word1) {
+            $best = strlen($s2);
+
+            foreach($words2 as $word2) {
+                $wordDist = levenshtein($word1, $word2);
+
+                if($wordDist < $best) {
+                    $best = $wordDist;
+                }
+
+                if($wordDist === 0) {
+                    break;
+                }
+            }
+
+            $total += $total + $best;
+        }
+
+        return $total;
+    }
+
+    /**
+     * Calculate phrase match score.
+     *
+     * @param string $s1 Word 1
+     * @param string $s2 Word 2
+     *
+     * @return int
+     *
+     * @since  1.0.0
+     */
+    public static function valuePhrase(string $s1, string $s2) : int
+    {
+        return levenshtein($s1, $s2);
+    }
+
+    /**
+     * Calculate word length score.
+     *
+     * @param string $s1 Word 1
+     * @param string $s2 Word 2
+     *
+     * @return int
+     *
+     * @since  1.0.0
+     */
+    public static function valueLength(string $s1, string $s2) : int
+    {
+        return abs(strlen($s1) - strlen($s2));
+    }
+
+    /**
+     * Calculate fuzzy match score.
+     *
+     * @param string $s1 Word 1
+     * @param string $s2 Word 2
+     * @param float $prhaseWeight Weighting for phrase score
+     * @param float $wordWeight Weighting for word score
+     * @param float $minWeight Min weight
+     * @param float $maxWeight Max weight
+     * @param float $lengthWeight Weighting for word length
+     *
+     * @return float
+     *
+     * @since  1.0.0
+     */
+    public static function fuzzyMatch(string $s1, string $s2, float $phraseWeight = 0.5, float $wordWeight = 1, float $minWeight = 10, float $maxWeight = 1, float $lengthWeight = -0.3) : float
+    {
+        $phraseValue = valuePhrase($s1, $s2);
+        $wordValue = valueWords($s1, $s2);
+        $lengthValue = valueLength($s1, $s2);
+
+        return min($phraseValue * $phraseWeight, $wordValue * $wordWeight) * $minWeight
+            + max($phraseValue * $phraseWeight, $wordValue * $wordWeight) * $maxWeight
+            + $lengthValue * $lengthWeight;
+    } 
+}
\ No newline at end of file