MDL-46256 count_words: fix handling of paragraphs

lib/moodlelib.php

@@ -8375,10 +8375,27 @@ function moodle_setlocale($locale='') {
  * Words are defined as things between whitespace.
  *
  * @category string
- * @param string $string The text to be searched for words.
+ * @param string $string The text to be searched for words. May be HTML.
  * @return int The count of words in the specified string
  */
 function count_words($string) {
+    // Before stripping tags, add a space after the close tag of anything that is not obviously inline.
+    // Also, br is a special case because it definitely delimits a word, but has no close tag.
+    $string = preg_replace('~
+            (                                   # Capture the tag we match.
+                </                              # Start of close tag.
+                (?!                             # Do not match any of these specific close tag names.
+                    a> | b> | del> | em> | i> |
+                    ins> | s> | small> |
+                    strong> | sub> | sup> | u>
+                )
+                \w+                             # But, apart from those execptions, match any tag name.
+                >                               # End of close tag.
+            |
+                <br> | <br\s*/>                 # Special cases that are not close tags.
+            )
+            ~x', '$1 ', $string); // Add a space after the close tag.
+    // Now remove HTML tags.
     $string = strip_tags($string);
     // Decode HTML entities.
     $string = html_entity_decode($string);