From 0732dc7d2c239d792e2c9f02fa378a18a6d01070 Mon Sep 17 00:00:00 2001
From: Tom Lingham <tjlingham@gmail.com>
Date: Sat, 14 May 2016 18:14:03 +1000
Subject: [PATCH] Add Fuzzy Search driver with unicode support

---
 config/searchy.php                                 |  4 ++
 src/Matchers/AcronymUnicodeMatcher.php             | 31 +++++++++++++++
 .../ConsecutiveCharactersUnicodeMatcher.php        | 45 ++++++++++++++++++++++
 src/Matchers/StudlyCaseUnicodeMatcher.php          | 36 +++++++++++++++++
 src/SearchDrivers/FuzzySearchUnicodeDriver.php     | 20 ++++++++++
 5 files changed, 136 insertions(+)
 create mode 100644 src/Matchers/AcronymUnicodeMatcher.php
 create mode 100644 src/Matchers/ConsecutiveCharactersUnicodeMatcher.php
 create mode 100644 src/Matchers/StudlyCaseUnicodeMatcher.php
 create mode 100644 src/SearchDrivers/FuzzySearchUnicodeDriver.php

diff --git a/config/searchy.php b/config/searchy.php
index ccbf005..06fceac 100755
--- a/config/searchy.php
+++ b/config/searchy.php
@@ -12,6 +12,10 @@ return [
             'class' => 'TomLingham\Searchy\SearchDrivers\FuzzySearchDriver',
         ],
 
+        'ufuzzy' => [
+            'class' => 'TomLingham\Searchy\SearchDrivers\FuzzySearchUnicodeDriver',
+        ],
+
         'simple' => [
             'class' => 'TomLingham\Searchy\SearchDrivers\SimpleSearchDriver',
         ],
diff --git a/src/Matchers/AcronymUnicodeMatcher.php b/src/Matchers/AcronymUnicodeMatcher.php
new file mode 100644
index 0000000..94d7a14
--- /dev/null
+++ b/src/Matchers/AcronymUnicodeMatcher.php
@@ -0,0 +1,31 @@
+<?php
+
+namespace TomLingham\Searchy\Matchers;
+
+/**
+ * Matches strings for Acronym 'like' matches but does NOT return Studly Case Matches.
+ *
+ * for example, a search for 'fb' would match; 'foo bar' or 'Fred Brown' but not 'FreeBeer'.
+ *
+ * Class AcronymMatcher
+ */
+class AcronymUnicodeMatcher extends BaseMatcher
+{
+    /**
+     * @var string
+     */
+    protected $operator = 'LIKE';
+
+    /**
+     * @param $searchString
+     *
+     * @return mixed|string
+     */
+    public function formatSearchString($searchString)
+    {
+        $results = [];
+        preg_match_all('/./u', mb_strtoupper($searchString, 'UTF-8'), $results);
+
+        return implode('% ', $results[0]).'%';
+    }
+}
diff --git a/src/Matchers/ConsecutiveCharactersUnicodeMatcher.php b/src/Matchers/ConsecutiveCharactersUnicodeMatcher.php
new file mode 100644
index 0000000..9aeeeb0
--- /dev/null
+++ b/src/Matchers/ConsecutiveCharactersUnicodeMatcher.php
@@ -0,0 +1,45 @@
+<?php
+
+namespace TomLingham\Searchy\Matchers;
+
+/**
+ * Matches strings that include all the characters in the search relatively position within the string.
+ * It also calculates the percentage of characters in the string that are matched and applies the multiplier accordingly.
+ *
+ * For Example, a search for 'fba' would match; 'Foo Bar' or 'Afraid of bats'
+ *
+ * Class ConsecutiveCharactersMatcher
+ */
+class ConsecutiveCharactersUnicodeMatcher extends BaseMatcher
+{
+    /**
+     * @var string
+     */
+    protected $operator = 'LIKE';
+
+    /**
+     * @param $searchString
+     *
+     * @return string
+     */
+    public function formatSearchString($searchString)
+    {
+        $results = [];
+        preg_match_all('/./u', $searchString, $results);
+
+        return '%'.implode('%', $results[0]).'%';
+    }
+
+    /**
+     * @param $column
+     * @param $rawString
+     *
+     * @return mixed|string
+     */
+    public function buildQueryString($column, $rawString)
+    {
+        $searchString = $this->formatSearchString($rawString);
+
+        return "IF( REPLACE($column, '\.', '') {$this->operator} '$searchString', ROUND({$this->multiplier} * ( CHAR_LENGTH( '$rawString' ) / CHAR_LENGTH( REPLACE($column, ' ', '') ))), 0)";
+    }
+}
diff --git a/src/Matchers/StudlyCaseUnicodeMatcher.php b/src/Matchers/StudlyCaseUnicodeMatcher.php
new file mode 100644
index 0000000..e8444ec
--- /dev/null
+++ b/src/Matchers/StudlyCaseUnicodeMatcher.php
@@ -0,0 +1,36 @@
+<?php
+
+namespace TomLingham\Searchy\Matchers;
+
+/**
+ * Matches Studly Case strings using the first letters of the words only.
+ *
+ * For example a search for 'hp' would match; 'HtmlServiceProvider' or 'HashParser' but not 'hasProvider'
+ *
+ * Class StudlyCaseMatcher
+ */
+class StudlyCaseUnicodeMatcher extends BaseMatcher
+{
+    /**
+     * @var string
+     */
+    protected $operator = 'LIKE BINARY';
+
+    /**
+     * @param $searchString
+     *
+     * @return string
+     */
+    public function formatSearchString($searchString)
+    {
+        $results = [];
+        preg_match_all('/./u', mb_strtoupper($searchString, 'UTF-8'), $results);
+
+        return implode('%', $results[0]).'%';
+    }
+
+    public function buildQueryString($column, $searchString)
+    {
+        return "IF( CHAR_LENGTH( TRIM( $column )) = CHAR_LENGTH( REPLACE( TRIM( $column ), ' ', '')) AND $column {$this->operator} '{$this->formatSearchString($searchString)}', {$this->multiplier}, 0)";
+    }
+}
diff --git a/src/SearchDrivers/FuzzySearchUnicodeDriver.php b/src/SearchDrivers/FuzzySearchUnicodeDriver.php
new file mode 100644
index 0000000..d3a3b41
--- /dev/null
+++ b/src/SearchDrivers/FuzzySearchUnicodeDriver.php
@@ -0,0 +1,20 @@
+<?php
+
+namespace TomLingham\Searchy\SearchDrivers;
+
+class FuzzySearchUnicodeDriver extends BaseSearchDriver
+{
+    /**
+     * @var array
+     */
+    protected $matchers = [
+        \TomLingham\Searchy\Matchers\ExactMatcher::class                        => 100,
+        \TomLingham\Searchy\Matchers\StartOfStringMatcher::class                => 50,
+        \TomLingham\Searchy\Matchers\AcronymUnicodeMatcher::class               => 42,
+        \TomLingham\Searchy\Matchers\ConsecutiveCharactersUnicodeMatcher::class => 40,
+        \TomLingham\Searchy\Matchers\StartOfWordsMatcher::class                 => 35,
+        \TomLingham\Searchy\Matchers\StudlyCaseUnicodeMatcher::class            => 32,
+        \TomLingham\Searchy\Matchers\InStringMatcher::class                     => 30,
+        \TomLingham\Searchy\Matchers\TimesInStringMatcher::class                => 8,
+  ];
+}