From 61001dc4c935a43e8532266927a39ffa60de3cb1 Mon Sep 17 00:00:00 2001
From: Hunter Ashton <hbashton13@gmail.com>
Date: Tue, 9 May 2017 04:34:19 -0500
Subject: [PATCH] HTML parser (#109)

* pwr requests (#108)

* on/off switcher for pwrtelegram peer info requests

* Fix PWRTelgram requets if we have no info about peer

* Apply fixes from StyleCI

* Add a new function, html_fixtags

This parses any message with parse_mode of Markdown or HTML, and converts it into a format parse-able by Dom  (if it wasn't already readable)

* Apply fixes from Style CI
---
 .../MadelineProto/TL/Conversion/BotAPI.php    | 26 +++++++++++++++++++
 1 file changed, 26 insertions(+)

diff --git a/src/danog/MadelineProto/TL/Conversion/BotAPI.php b/src/danog/MadelineProto/TL/Conversion/BotAPI.php
index 553b4e185..8191cf98b 100644
--- a/src/danog/MadelineProto/TL/Conversion/BotAPI.php
+++ b/src/danog/MadelineProto/TL/Conversion/BotAPI.php
@@ -431,6 +431,7 @@ trait BotAPI
         if (preg_match('/html/i', $arguments['parse_mode'])) {
             $nmessage = '';
             try {
+                $arguments['message'] = $this->html_fixtags($arguments['message']);
                 $dom = new \DOMDocument();
                 $dom->loadHTML(mb_convert_encoding($arguments['message'], 'HTML-ENTITIES', 'UTF-8'));
                 if (!isset($arguments['entities'])) {
@@ -487,4 +488,29 @@ trait BotAPI
 
         return $finalArray;
     }
+
+    public function html_fixtags($text)
+    {
+        preg_match_all("#(.*?)(<(a|b|strong|em|i|code|pre)[^>]*>)(.*?)(<\/\\3>)(.*)?#is", $text, $matches, PREG_SET_ORDER);
+        if ($matches) {
+            $last = count($matches) - 1;
+            foreach ($matches as $val) {
+                if (trim($val[1]) != '') {
+                    $text = str_replace($val[1], htmlentities($val[1]), $text);
+                }
+                $text = str_replace($val[4], htmlentities(trim($val[4])), $text);
+                if ($val == $matches[$last]) {
+                    $text = str_replace($val[6], $this->html_fixtags($val[6]), $text);
+                }
+            }
+            preg_match_all("#<a href=\x22(.+?)\x22>#is", $text, $matches);
+            foreach ($matches[1] as $match) {
+                $text = str_replace($match, htmlentities($match), $text);
+            }
+
+            return $text;
+        } else {
+            return htmlentities($text);
+        }
+    }
 }