From c110115cc09448b20eb1a1528e63fcb56ec8ceff Mon Sep 17 00:00:00 2001
From: Tim Hunt <T.J.Hunt@open.ac.uk>
Date: Tue, 19 Jan 2021 15:29:30 +0000
Subject: [PATCH 1/3] MDL-46256 word count: Convert existing tests to a data
 provider

---
 lib/tests/moodlelib_test.php | 57 +++++++++++++++++-------------------
 1 file changed, 27 insertions(+), 30 deletions(-)

diff --git a/lib/tests/moodlelib_test.php b/lib/tests/moodlelib_test.php
index 69621008bfd..00ff10e1827 100644
--- a/lib/tests/moodlelib_test.php
+++ b/lib/tests/moodlelib_test.php
@@ -3793,38 +3793,35 @@ class core_moodlelib_testcase extends advanced_testcase {
 
     /**
      * Test function count_words().
+     *
+     * @dataProvider count_words_testcases
+     * @param int $expectedcount number of words in $string.
+     * @param string $string the test string to count the words of.
      */
-    public function test_count_words() {
-        $count = count_words("one two three'four");
-        $this->assertEquals(3, $count);
-
-        $count = count_words('one+two three’four');
-        $this->assertEquals(3, $count);
-
-        $count = count_words('one"two three-four');
-        $this->assertEquals(2, $count);
-
-        $count = count_words('one@two three_four');
-        $this->assertEquals(4, $count);
-
-        $count = count_words('one\two three/four');
-        $this->assertEquals(4, $count);
-
-        $count = count_words(' one ... two &nbsp; three...four ');
-        $this->assertEquals(4, $count);
-
-        $count = count_words('one.2 3,four');
-        $this->assertEquals(4, $count);
-
-        $count = count_words('1³ £2 €3.45 $6,789');
-        $this->assertEquals(4, $count);
-
-        $count = count_words('one—two ブルース カンベッル');
-        $this->assertEquals(4, $count);
-
-        $count = count_words('one…two ブルース … カンベッル');
-        $this->assertEquals(4, $count);
+    public function test_count_words(int $expectedcount, string $string): void {
+        $this->assertEquals($expectedcount, count_words($string));
     }
+
+    /**
+     * Data provider for {@see test_count_words}.
+     *
+     * @return array of test cases.
+     */
+    public function count_words_testcases(): array {
+        return [
+            [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, ' 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'],
+        ];
+    }
+
     /**
      * Tests the getremoteaddr() function.
      */

From 82a050d67bc260066848009463076a9c0f704a5a Mon Sep 17 00:00:00 2001
From: Tim Hunt <T.J.Hunt@open.ac.uk>
Date: Tue, 19 Jan 2021 15:50:05 +0000
Subject: [PATCH 2/3] MDL-46256 count_letters: an unit tests and fix HTML
 editity handling

---
 lib/moodlelib.php            |  3 ++-
 lib/tests/moodlelib_test.php | 25 +++++++++++++++++++++++++
 2 files changed, 27 insertions(+), 1 deletion(-)

diff --git a/lib/moodlelib.php b/lib/moodlelib.php
index 59fcd1c132f..5a6ef0577e3 100644
--- a/lib/moodlelib.php
+++ b/lib/moodlelib.php
@@ -8375,11 +8375,12 @@ function count_words($string) {
  * Letters are defined as chars not in tags and different from whitespace.
  *
  * @category string
- * @param string $string The text to be searched for letters.
+ * @param string $string The text to be searched for letters. May be HTML.
  * @return int The count of letters in the specified text.
  */
 function count_letters($string) {
     $string = strip_tags($string); // Tags are out now.
+    $string = html_entity_decode($string);
     $string = preg_replace('/[[:space:]]*/', '', $string); // Whitespace are out now.
 
     return core_text::strlen($string);
diff --git a/lib/tests/moodlelib_test.php b/lib/tests/moodlelib_test.php
index 00ff10e1827..8c02d58b1b8 100644
--- a/lib/tests/moodlelib_test.php
+++ b/lib/tests/moodlelib_test.php
@@ -3822,6 +3822,31 @@ class core_moodlelib_testcase extends advanced_testcase {
         ];
     }
 
+    /**
+     * Test function {@see count_letters()}.
+     *
+     * @dataProvider count_letters_testcases
+     * @param int $expectedcount number of characters in $string.
+     * @param string $string the test string to count the letters of.
+     */
+    public function test_count_letters(int $expectedcount, string $string): void {
+        $this->assertEquals($expectedcount, count_letters($string));
+    }
+
+    /**
+     * Data provider for {@see count_letters_testcases}.
+     *
+     * @return array of test cases.
+     */
+    public function count_letters_testcases(): array {
+        return [
+            [0, ''],
+            [1, 'x'],
+            [1, '&amp;'],
+            [4, '<p>frog</p>'],
+        ];
+    }
+
     /**
      * Tests the getremoteaddr() function.
      */

From 6a62cbe59998c2939fa0ef8c390fb11a2830542c Mon Sep 17 00:00:00 2001
From: Tim Hunt <T.J.Hunt@open.ac.uk>
Date: Tue, 19 Jan 2021 15:39:24 +0000
Subject: [PATCH 3/3] MDL-46256 count_words: fix handling of paragraphs

---
 lib/moodlelib.php            | 19 ++++++++++++++++++-
 lib/tests/moodlelib_test.php | 16 +++++++++++++++-
 2 files changed, 33 insertions(+), 2 deletions(-)

diff --git a/lib/moodlelib.php b/lib/moodlelib.php
index 5a6ef0577e3..2a7b5a10ff3 100644
--- a/lib/moodlelib.php
+++ b/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);
diff --git a/lib/tests/moodlelib_test.php b/lib/tests/moodlelib_test.php
index 8c02d58b1b8..d3733c768a2 100644
--- a/lib/tests/moodlelib_test.php
+++ b/lib/tests/moodlelib_test.php
@@ -3792,7 +3792,7 @@ class core_moodlelib_testcase extends advanced_testcase {
     }
 
     /**
-     * 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 @@ class core_moodlelib_testcase extends advanced_testcase {
      */
     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>"],
         ];
     }