From 82a891f6e0ed2e88aad4f52c7fa1726e063db1b2 Mon Sep 17 00:00:00 2001
From: Seb35 <seb35@seb35.fr>
Date: Sat, 29 Sep 2018 14:05:00 +0200
Subject: [PATCH] Adapt for multibyte strings as UTF-8

There is no native function mb_str_split, hence writing a loop.
---
 src/MatchesSolver.php                     |  2 +-
 src/Solver.php                            | 12 +++++++++---
 test/suite/MatchesArrayProvidersTrait.php | 23 +++++++++++++++++++++++
 3 files changed, 33 insertions(+), 4 deletions(-)

diff --git a/src/MatchesSolver.php b/src/MatchesSolver.php
index 5079be1..d76c259 100644
--- a/src/MatchesSolver.php
+++ b/src/MatchesSolver.php
@@ -46,7 +46,7 @@ protected function result(
         array $matrix
     ) {
         return array_map(function (Match $result) use ($stringA) {
-            $result->value = substr($stringA, $result->index(), $result->length);
+            $result->value = mb_substr($stringA, $result->index(), $result->length);
 
             return $result;
         }, $longestIndexes);
diff --git a/src/Solver.php b/src/Solver.php
index 74ee735..665ec15 100644
--- a/src/Solver.php
+++ b/src/Solver.php
@@ -37,7 +37,7 @@ protected function result(
         string $stringB,
         array $matrix
     ) {
-        return count($longestIndexes) === 0 ? '' : substr($stringA, $longestIndexes[0], $longestLength);
+        return count($longestIndexes) === 0 ? '' : mb_substr($stringA, $longestIndexes[0], $longestLength);
     }
 
     /**
@@ -55,8 +55,14 @@ public function solve(string $stringA, string $stringB)
             return call_user_func_array([$this, 'solve'], $arguments);
         }
 
-        $charsA = str_split($stringA);
-        $charsB = str_split($stringB);
+        $charsA = [];
+        $charsB = [];
+        for ($i=0; $i < max(mb_strlen($stringA), 1); $i++) {
+                $charsA[] = mb_substr($stringA, $i, 1);
+        }
+        for ($i=0; $i < max(mb_strlen($stringB), 1); $i++) {
+                $charsB[] = mb_substr($stringB, $i, 1);
+        }
 
         $matrix = array_fill_keys(array_keys($charsA), array_fill_keys(array_keys($charsB), 0));
         $longestLength = 0;
diff --git a/test/suite/MatchesArrayProvidersTrait.php b/test/suite/MatchesArrayProvidersTrait.php
index 0e715c3..33986bd 100644
--- a/test/suite/MatchesArrayProvidersTrait.php
+++ b/test/suite/MatchesArrayProvidersTrait.php
@@ -55,6 +55,29 @@ public function twoStringsOrderedMatchesArrayProvider()
                     ],
                 ],
             ],
+            'UTF-8'              => [
+                'L’été était chaud.',
+                'L’hiver était froid.',
+                [
+                    [
+                        'value'   => ' était ',
+                        'length'  => 7,
+                        'indexes' => [5, 7],
+                    ],
+                ],
+            ],
+            // In UTF-8: é = 0xC3A9 and © = 0xC2A9 (the last byte is the same but the Unicode characters are different)
+            'UTF-8 (nasty)'      => [
+                'L’été était chaud.',
+                'L’hiver ©tait froid.',
+                [
+                    [
+                        'value'   => 'tait ',
+                        'length'  => 5,
+                        'indexes' => [7, 9],
+                    ],
+                ],
+            ],
         ];
     }
 }