From 1e3449ca28e94f556cbb4caae7aa7b6f40080056 Mon Sep 17 00:00:00 2001
From: Paul Holden <paulh@moodle.com>
Date: Wed, 25 Nov 2020 21:35:14 +0000
Subject: [PATCH] MDL-70295 libraries: upgrade to version 4.3.1 of Html2Text.

---
 lib/html2text/Html2Text.php     | 61 ++++++++++++++++++++++++---------
 lib/html2text/readme_moodle.txt |  9 ++---
 lib/thirdpartylibs.xml          |  2 +-
 3 files changed, 50 insertions(+), 22 deletions(-)

diff --git a/lib/html2text/Html2Text.php b/lib/html2text/Html2Text.php
index f279879d6f8..9fd91235927 100644
--- a/lib/html2text/Html2Text.php
+++ b/lib/html2text/Html2Text.php
@@ -28,14 +28,14 @@ class Html2Text
     /**
      * Contains the HTML content to convert.
      *
-     * @type string
+     * @var string $html
      */
     protected $html;
 
     /**
      * Contains the converted, formatted text.
      *
-     * @type string
+     * @var string $text
      */
     protected $text;
 
@@ -43,7 +43,7 @@ class Html2Text
      * List of preg* regular expression patterns to search for,
      * used in conjunction with $replace.
      *
-     * @type array
+     * @var array $search
      * @see $replace
      */
     protected $search = array(
@@ -54,6 +54,7 @@ class Html2Text
         '/<style\b[^>]*>.*?<\/style>/i',                  // <style>s -- which strip_tags supposedly has problems with
         '/<i\b[^>]*>(.*?)<\/i>/i',                        // <i>
         '/<em\b[^>]*>(.*?)<\/em>/i',                      // <em>
+        '/<ins\b[^>]*>(.*?)<\/ins>/i',                    // <ins>
         '/(<ul\b[^>]*>|<\/ul>)/i',                        // <ul> and </ul>
         '/(<ol\b[^>]*>|<\/ol>)/i',                        // <ol> and </ol>
         '/(<dl\b[^>]*>|<\/dl>)/i',                        // <dl> and </dl>
@@ -73,7 +74,7 @@ class Html2Text
     /**
      * List of pattern replacements corresponding to patterns searched.
      *
-     * @type array
+     * @var array $replace
      * @see $search
      */
     protected $replace = array(
@@ -84,6 +85,7 @@ class Html2Text
         '',                              // <style>s -- which strip_tags supposedly has problems with
         '_\\1_',                         // <i>
         '_\\1_',                         // <em>
+        '_\\1_',                         // <ins>
         "\n\n",                          // <ul> and </ul>
         "\n\n",                          // <ol> and </ol>
         "\n\n",                          // <dl> and </dl>
@@ -104,7 +106,7 @@ class Html2Text
      * List of preg* regular expression patterns to search for,
      * used in conjunction with $entReplace.
      *
-     * @type array
+     * @var array $entSearch
      * @see $entReplace
      */
     protected $entSearch = array(
@@ -118,7 +120,7 @@ class Html2Text
     /**
      * List of pattern replacements corresponding to patterns searched.
      *
-     * @type array
+     * @var array $entReplace
      * @see $entSearch
      */
     protected $entReplace = array(
@@ -133,7 +135,7 @@ class Html2Text
      * List of preg* regular expression patterns to search for
      * and replace using callback function.
      *
-     * @type array
+     * @var array $callbackSearch
      */
     protected $callbackSearch = array(
         '/<(h)[123456]( [^>]*)?>(.*?)<\/h[123456]>/i',           // h1 - h6
@@ -141,6 +143,7 @@ class Html2Text
         '/<(br)[^>]*>[ ]*/i',                                    // <br> with leading whitespace after the newline.
         '/<(b)( [^>]*)?>(.*?)<\/b>/i',                           // <b>
         '/<(strong)( [^>]*)?>(.*?)<\/strong>/i',                 // <strong>
+        '/<(del)( [^>]*)?>(.*?)<\/del>/i',                       // <del>
         '/<(th)( [^>]*)?>(.*?)<\/th>/i',                         // <th> and </th>
         '/<(a) [^>]*href=("|\')([^"\']+)\2([^>]*)>(.*?)<\/a>/i'  // <a href="">
     );
@@ -149,7 +152,7 @@ class Html2Text
      * List of preg* regular expression patterns to search for in PRE body,
      * used in conjunction with $preReplace.
      *
-     * @type array
+     * @var array $preSearch
      * @see $preReplace
      */
     protected $preSearch = array(
@@ -163,7 +166,7 @@ class Html2Text
     /**
      * List of pattern replacements corresponding to patterns searched for PRE body.
      *
-     * @type array
+     * @var array $preReplace
      * @see $preSearch
      */
     protected $preReplace = array(
@@ -177,21 +180,21 @@ class Html2Text
     /**
      * Temporary workspace used during PRE processing.
      *
-     * @type string
+     * @var string $preContent
      */
     protected $preContent = '';
 
     /**
      * Contains the base URL that relative links should resolve to.
      *
-     * @type string
+     * @var string $baseurl
      */
     protected $baseurl = '';
 
     /**
      * Indicates whether content in the $html variable has been converted yet.
      *
-     * @type boolean
+     * @var boolean $converted
      * @see $html, $text
      */
     protected $converted = false;
@@ -199,7 +202,7 @@ class Html2Text
     /**
      * Contains URL addresses from links to be rendered in plain text.
      *
-     * @type array
+     * @var array $linkList
      * @see buildlinkList()
      */
     protected $linkList = array();
@@ -207,7 +210,7 @@ class Html2Text
     /**
      * Various configuration options (able to be set in the constructor)
      *
-     * @type array
+     * @var array $options
      */
     protected $options = array(
         'do_links' => 'inline', // 'none'
@@ -281,7 +284,7 @@ class Html2Text
     /**
      * Returns the text, converted from HTML.
      *
-     * @return string
+     * @return string Plain text
      */
     public function getText()
     {
@@ -414,7 +417,7 @@ class Html2Text
         }
 
         // Ignored link types
-        if (preg_match('!^(javascript:|mailto:|#)!i', $link)) {
+        if (preg_match('!^(javascript:|mailto:|#)!i', html_entity_decode($link))) {
             return $display;
         }
 
@@ -450,6 +453,11 @@ class Html2Text
         }
     }
 
+    /**
+     * Helper function for PRE body conversion.
+     *
+     * @param string &$text HTML content
+     */
     protected function convertPre(&$text)
     {
         // get the content of PRE element
@@ -486,7 +494,7 @@ class Html2Text
     /**
      * Helper function for BLOCKQUOTE body conversion.
      *
-     * @param string $text HTML content
+     * @param string &$text HTML content
      */
     protected function convertBlockquotes(&$text)
     {
@@ -563,6 +571,8 @@ class Html2Text
             case 'b':
             case 'strong':
                 return $this->toupper($matches[3]);
+            case 'del':
+                return $this->tostrike($matches[3]);
             case 'th':
                 return $this->toupper("\t\t" . $matches[3] . "\n");
             case 'h':
@@ -628,4 +638,21 @@ class Html2Text
 
         return $str;
     }
+
+    /**
+     * Helper function for DEL conversion.
+     *
+     * @param  string $text HTML content
+     * @return string Converted text
+     */
+    protected function tostrike($str)
+    {
+        $rtn = '';
+        for ($i = 0; $i < mb_strlen($str); $i++) {
+            $chr = mb_substr($str, $i, 1);
+            $combiningChr = chr(0xC0 | 0x336 >> 6). chr(0x80 | 0x336 & 0x3F);
+            $rtn .= $chr . $combiningChr;
+        }
+        return $rtn;
+    }
 }
diff --git a/lib/html2text/readme_moodle.txt b/lib/html2text/readme_moodle.txt
index 111f11adcd4..37cab254a25 100644
--- a/lib/html2text/readme_moodle.txt
+++ b/lib/html2text/readme_moodle.txt
@@ -1,4 +1,4 @@
-Description of Html2Text library import into Moodle
+Description of Html2Text v4.3.1 library import into Moodle
 
 Please note that we override some mb_* functions in Html2Text's namespace at
 run time. Until Html2Text adds some sort of fallback for the mb_* functions
@@ -7,7 +7,8 @@ running PHP without mbstring don't see nasty undefined function errors.
 
 Instructions
 ------------
-1. Clone https://github.com/mtibben/html2text.git into an unrelated directory
-2. Copy /path/to/html2text/src/Html2Text.php to lib/html2text/
+1. Download the latest release of Html2Text from https://github.com/mtibben/html2text/releases/
+2. Extract the contents of the release archive into a directory.
+3. Copy src/Html2Text.php to lib/html2text/
 
-Imported from: https://github.com/mtibben/html2text.git
+Imported from: https://github.com/mtibben/html2text/releases/
diff --git a/lib/thirdpartylibs.xml b/lib/thirdpartylibs.xml
index 356cc82341e..4d385afc898 100644
--- a/lib/thirdpartylibs.xml
+++ b/lib/thirdpartylibs.xml
@@ -158,7 +158,7 @@
     <location>html2text</location>
     <name>HTML2Text</name>
     <license>GPL</license>
-    <version>4.2.1</version>
+    <version>4.3.1</version>
     <licenseversion>2.0+</licenseversion>
   </library>
   <library>