From f1382840bae481c8f264d24648feb40a5707d93d Mon Sep 17 00:00:00 2001
From: "vscode iu2frl.it" <vscode@iu2frl.it>
Date: Tue, 5 Nov 2024 20:19:38 +0100
Subject: [PATCH] Escaping characters before being sent

---
 frlbot.py | 40 ++++++++++++++++++++++++++++++++++------
 1 file changed, 34 insertions(+), 6 deletions(-)

diff --git a/frlbot.py b/frlbot.py
index 00c16a9..b464ee9 100644
--- a/frlbot.py
+++ b/frlbot.py
@@ -331,6 +331,29 @@ def remove_old_news(max_days: int = -1) -> int:
         logging.error("Cannot delete older news. " + str(returned_exception))
         return -1
 
+# Escape characters that cannot be parsed
+def escape_markdown_v2(text: str) -> str:
+    """
+    Escape special characters for Telegram MarkdownV2 format.
+    Characters that need escaping: '_', '*', '[', ']', '(', ')', '~', '>', '#', '+', '-', '=', '|', '{', '}', '.', '!'
+    
+    Args:
+        text (str): The text to escape
+        
+    Returns:
+        str: The escaped text safe for MarkdownV2
+    """
+    special_chars = ['_', '*', '[', ']', '(', ')', '~', '>', '#', '+', '-', '=', '|', '{', '}', '.', '!', '\'']
+    escaped_text = ''
+    
+    for char in text:
+        if char in special_chars:
+            escaped_text += f'\\{char}'
+        else:
+            escaped_text += char
+            
+    return escaped_text
+
 # Main code
 def main():
     """Main robot code"""
@@ -370,12 +393,12 @@ def main():
                 emoji_calendar = emoji.emojize(":spiral_calendar:", language="alias")
                 emoji_link = emoji.emojize(":link:", language="alias")
                 try:
-                    telegram_payload = f"{emoji_flag_it} {translate_text(single_news.title, 'it')}\n" + \
-                                        f"{emoji_flag_en} {translate_text(single_news.title, 'en')}\n" + \
-                                        f"\n{emoji_pencil} {single_news.author}\n" + \
+                    telegram_payload = f"{emoji_flag_it} {escape_markdown_v2(translate_text(single_news.title, 'it'))}\n" + \
+                                        f"{emoji_flag_en} {escape_markdown_v2(translate_text(single_news.title, 'en'))}\n" + \
+                                        f"\n{emoji_pencil} {escape_markdown_v2(single_news.author)}\n" + \
                                         f"{emoji_calendar} {single_news.date.strftime('%Y/%m/%d, %H:%M')}\n" + \
-                                        f"\n{emoji_flag_it} {translate_text(single_news.summary, 'it')}\n" + \
-                                        f"\n{emoji_flag_en} {translate_text(single_news.summary, 'en')}\n" + \
+                                        f"\n{emoji_flag_it} {escape_markdown_v2(translate_text(single_news.summary, 'it'))}\n" + \
+                                        f"\n{emoji_flag_en} {escape_markdown_v2(translate_text(single_news.summary, 'en'))}\n" + \
                                         f"\n{emoji_link} {single_news.link}"
                     if not dryRun:
                         telegramBot.send_message(get_target_chat_from_env(), telegram_payload, parse_mode="MARKDOWN")
@@ -389,8 +412,13 @@ def main():
                     news_cnt += 1
                 except Exception as returned_exception:
                     logging.error(str(returned_exception))
-                    exception_cnt += 1
                     exception_message = str(returned_exception)
+                    if "can\'t parse entities:" in returned_exception:
+                        logging.warning("Skipping [" + single_news.checksum + "] due to Telegram parsing error")
+                        sql_connector.cursor().execute("INSERT INTO news(date, checksum) VALUES(?, ?)", [single_news.date, single_news.checksum])
+                        sql_connector.commit()
+                    else:
+                        exception_cnt += 1
         # This message was already posted
         else:
             logging.debug("Post at [" + single_news.link + "] was already sent")