Fixed patterns of typo, corrected and context. Added abstract function clearText()

src/TyposClientInterface.php

@@ -2,8 +2,6 @@
 /**
  * Created by PhpStorm.
  * User: ambulance
- * Date: 25.05.18
- * Time: 19:39
  */
 
 namespace Etersoft\Typos;
@@ -210,7 +208,7 @@ abstract class TyposClientInterface
      * @param string $typo
      * @param string $corrected
      * @param string $context
-     * @param $text
+     * @param string $text
      *
      * @return string       Text with typo replaced by corrected
      * @throws \Exception If something goes wrong
@@ -219,10 +217,8 @@ abstract class TyposClientInterface
 		"Context not found", 405
      */
     private function replaceTypoInText(string $typo, string $corrected, string $context, string $text) {
-        // Strip all tags from text
-        // Also copy input string
+        // Copy input string
         $originalText = $text;
-        $text = strip_tags($text);
 /*
         // BUG# 13121 
         $typo = str_replace("\xc2\xa0", " ", $typo);
@@ -230,10 +226,18 @@ abstract class TyposClientInterface
         $context = str_replace("\xc2\xa0", " ", $context);
         $text = str_replace("\xc2\xa0", " ", $text);
 */
+		// Clear texts from tags, replaces special characters with standard ones 
+//		$typo		= $this->clearText($typo);
+//		$corrected	= $this->clearText($corrected);
+		$context	= $this->clearText($context);
+		$text		= $this->clearText($text);
+
         // Find all typos in text, capture an offset of each typo
-        if (!preg_match_all("#{$typo}#", $text, $typos, PREG_OFFSET_CAPTURE)) {
+		$pattern = preg_quote ($typo, '#');
+        if (!preg_match_all("#{$pattern}#u", $text, $typos, PREG_OFFSET_CAPTURE)) {
             // Check for already fixed typo
-            if (preg_match("#{$corrected}#", $text, $typos, PREG_OFFSET_CAPTURE)) {
+			$pattern = preg_quote ($corrected, '#');
+            if (preg_match("#{$pattern}#u", $text, $typos, PREG_OFFSET_CAPTURE)) {
                 throw new \Exception("Already fixed", 208);
             }
 
@@ -243,7 +247,8 @@ abstract class TyposClientInterface
         // Find a context in text, capture it offset
         
 
-        if (!preg_match("#{$context}#", $text, $contextMatch, PREG_OFFSET_CAPTURE)) {
+		$pattern = preg_quote ($context, '#');
+		if (!preg_match("#{$pattern}#u", $text, $contextMatch, PREG_OFFSET_CAPTURE)) {
 			// If a context was changed then report an error,
 			// cannot locate typo in a new context, must be
 			// fixed manually
@@ -264,7 +269,8 @@ abstract class TyposClientInterface
 
         // Fix a match with index = $indexOfTypo
         $index = 0;
-        return preg_replace_callback("#{$typo}#",
+		$pattern = preg_quote ($typo, '#');
+        return preg_replace_callback("#{$pattern}#u",
             function($match) use(&$index, $indexOfTypo, $corrected) {
                 $index++;
                 if (($index - 1) == $indexOfTypo) {
@@ -276,4 +282,14 @@ abstract class TyposClientInterface
             $originalText);
     }
 
+    /**
+     * Clears text from tags, replaces special characters with standard ones. 
+	 * Removes whitespace characters from the beginning and end of the text.
+	 * In a sense, the function is the inverse for wptexturize(); - WordPress function.
+     *
+     * @param string $text
+     *
+     * @return string       Cleared text
+     */
+    protected abstract function clearText(string $text);
 }