MDL-46256 count_words: fix handling of paragraphs

lib/moodlelib.php

@@ -8352,10 +8352,27 @@ 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);

lib/tests/moodlelib_test.php

@@ -3792,7 +3792,7 @@ public function test_username_load_fields_from_object() {
     }
 
     /**
-     * Test function count_words().
+     * Test function {@see count_words()}.
      *
      * @dataProvider count_words_testcases
      * @param int $expectedcount number of words in $string.
@@ -3809,16 +3809,30 @@ public function test_count_words(int $expectedcount, string $string): void {
      */
     public function count_words_testcases(): array {
         return [
+            [0, ''],
+            [4, 'one two three four'],
             [3, "one two three'four"],
             [3, 'one+two three’four'],
             [2, 'one"two three-four'],
             [4, 'one@two three_four'],
             [4, 'one\two three/four'],
+            [4, '<p>one two<br></br>three four</p>'],
+            [4, '<p>one two<br>three four</p>'],
+            [4, '<p>one two<br />three four</p>'], // XHTML style.
             [4, ' one ... two &nbsp; three...four '],
             [4, 'one.2 3,four'],
             [4, '1³ £2 €3.45 $6,789'],
             [4, 'one—two ブルース カンベッル'],
             [4, 'one…two ブルース … カンベッル'],
+            [4, '<p>one two</p><p>three four</p>'],
+            [4, '<p>one two</p><p><br/></p><p>three four</p>'],
+            [4, '<p>one</p><ul><li>two</li><li>three</li></ul><p>four.</p>'],
+            [1, '<p>em<b>phas</b>is.</p>'],
+            [1, '<p>em<i>phas</i>is.</p>'],
+            [1, '<p>em<strong>phas</strong>is.</p>'],
+            [1, '<p>em<em>phas</em>is.</p>'],
+            [2, "one\ntwo"],
+            [1, "SO<sub>4</sub><sup>2-</sup>"],
         ];
     }