Merge pull request #4 from matecat/fix-positions

WholeTextFinder::find multi byte safe

README.md

@@ -48,6 +48,35 @@ $matches = WholeTextFinder::find($haystack, $needle);
 
 ```
 
+### Multi bytes strings
+
+Please note that `WholeTextFinder::find` function is multi byte safe and returns the correct word positions in the original phrase. Take a look here:
+
+```php
+//..
+use Finder\WholeTextFinder;
+
+$haystack  = "La casa è bella bella";
+$needle = "bella";
+
+$matches = WholeTextFinder::find($haystack, $needle, true, true, true);
+
+// $matches is equals to:
+// array (
+//    0 =>
+//        array (
+//            0 => 'bella',
+//            1 => 10,
+//        ),
+//    1 =>
+//        array (
+//            0 => 'bella',
+//            1 => 16,
+//        ),
+//)
+
+```
+
 ## Find and Replace
 
 There is also available a `findAndReplace` method:
@@ -74,14 +103,14 @@ $matches = WholeTextFinder::findAndReplace($haystack, $needle, $replacement);
 //       [0]=>
 //       string(6) "και"
 //       [1]=>
-//       int(122)
+//       int(66)
 //     }
 //     [1]=>
 //     array(2) {
 //       [0]=>
 //       string(6) "και"
 //       [1]=>
-//       int(213)
+//       int(123)
 //     }
 //   }
 // } 

src/Helper/Replacer.php

@@ -14,9 +14,9 @@ class Replacer
      *
      * $expected = "Beauty -> test Anti-Akne Gesichtsreiniger Schlankmacher <g id=\"2\">XXX</g>";
      *
-     * @param $pattern
-     * @param $replacement
-     * @param $haystack
+     * @param string $pattern
+     * @param string $replacement
+     * @param string $haystack
      *
      * @return string|string[]
      */
@@ -38,7 +38,7 @@ public static function replace($pattern, $replacement, $haystack)
      *
      * /(\|\|\|\||<.*?>|%{.*?})(*SKIP)(*FAIL)|ciao/iu
      *
-     * @param $pattern
+     * @param string $pattern
      *
      * @return string
      */

src/Helper/Strings.php

@@ -50,4 +50,14 @@ public static function token($length = 8)
 
         return $key;
     }
+
+    /**
+     * @param string $string
+     *
+     * @return bool
+     */
+    public static function isMultibyte($string)
+    {
+        return ((strlen($string) - mb_strlen($string)) > 0);
+    }
 }

src/WholeTextFinder.php

@@ -24,9 +24,54 @@ public static function find($haystack, $needle, $skipHtmlEntities = true, $exact
 
         preg_match_all($patternAndHaystack['pattern'], $patternAndHaystack['haystack'], $matches, PREG_OFFSET_CAPTURE);
 
+        self::mbCorrectMatchPositions($patternAndHaystack['haystack'], $matches);
+
         return $matches[0];
     }
 
+    /**
+     * Correct position for multi byte strings
+     *
+     * @param string $haystack
+     * @param array $matches
+     *
+     * @return mixed
+     */
+    private static function mbCorrectMatchPositions( $haystack, &$matches)
+    {
+        if(!Strings::isMultibyte($haystack) ){
+            return $matches[0];
+        }
+
+        foreach ($matches[0] as $index => $match){
+            $word = $match[0];
+            $position = $match[1];
+
+            $correctPosition = self::mbFindTheCorrectPosition($haystack, $word, $position);
+            $matches[0][$index][1] = $correctPosition;
+        }
+    }
+
+    /**
+     * @param string $haystack
+     * @param string $word
+     * @param int $position
+     *
+     * @return int
+     */
+    private static function mbFindTheCorrectPosition( $haystack, $word, &$position)
+    {
+        $wordCheck = mb_substr($haystack, $position, mb_strlen($word));
+
+        if($wordCheck !== $word){
+            $position = $position - 1;
+
+            self::mbFindTheCorrectPosition($haystack, $word, $position);
+        }
+
+        return $position;
+    }
+
     /**
      * @param string $haystack
      * @param string $needle

tests/StringsTest.php

@@ -17,4 +17,16 @@ public function html_entity_decode()
 
         $this->assertEquals(Strings::htmlEntityDecode($input), $output);
     }
+
+    /**
+     * @test
+     */
+    public function is_multibyte()
+    {
+        $string = "La casa e bella";
+        $string2 = "La casa è bella";
+
+        $this->assertFalse(Strings::isMultibyte($string));
+        $this->assertTrue(Strings::isMultibyte($string2));
+    }
 }

tests/WholeTextFinderReplacementTest.php

@@ -19,8 +19,8 @@ public function find_and_replace_test_on_greek_text()
         $expected = [
             'replacement' => 'Δύο παράγοντες καθόρισαν την αντίληψή μου για την Τενεσί Ουίλιαμς test τη σκηνική παρουσίαση των κειμένων: η Maria Britneva test η Annette Saddik, αφετέρου.',
             'occurrencies' => [
-                [$needle, 122],
-                [$needle, 213],
+                [$needle, 66],
+                [$needle, 123],
             ],
         ];
         $matches = WholeTextFinder::findAndReplace($haystack, $needle, $replacement);
@@ -64,7 +64,7 @@ public function find_and_replace_must_skip_matecat_ph_tags()
     public function find_and_replace_must_skip_matecat_html_tags()
     {
         $haystack = "Beauty -> 0 Anti-Akne Gesichtsreiniger Schlankmacher <g id=\"2\">XXX</g>";
-        $needle = 0;
+        $needle = "0";
         $replacement = "test";
 
         $expected = "Beauty -> test Anti-Akne Gesichtsreiniger Schlankmacher <g id=\"2\">XXX</g>";

tests/WholeTextFinderTest.php

@@ -7,6 +7,21 @@
 
 class WholeTextFinderTest extends TestCase
 {
+    /**
+     * @test
+     */
+    public function can_detect_positions()
+    {
+        $haystack  = "La casa è bella bella";
+        $needle = "bella";
+
+        $matches = WholeTextFinder::find($haystack, $needle, true, true, true);
+
+        $this->assertCount(2, $matches);
+        $this->assertEquals(10, $matches[0][1]);
+        $this->assertEquals(16, $matches[1][1]);
+    }
+
     /**
      * @test
      */
@@ -138,17 +153,17 @@ public function find_should_return_correct_matches()
         $expected = [
             [
                 0 => 'ggio',
-                1 => 23
+                1 => 22
             ]
         ];
         $expected2 = [
             [
                 0 => 'ggio',
-                1 => 18
+                1 => 17
             ],
             [
                 0 => 'ggio',
-                1 => 23
+                1 => 22
             ]
         ];