theme finder in separated definition

src/autobioebbing.py

@@ -10,6 +10,88 @@ import random
 import re
 
 
+def find_infect_theme(text):
+    # NOTE: theme hell... any ideas for improvment required
+    # but not use huge regular expression like|that|fuckin|way|a|aaaa|aaaaaaaa
+    # because it makes re.findall like mess...
+    default_bioexpr_theme = r"Прибыль: ([0-9\.\,k]+)"
+    default_infected_days_theme = r' на ([0-9\ ]+) д.*'
+    default_pathogen_remaining_theme = r'Осталось: ([0-9\ ]+)'
+    bio_attack_themes = (  # I guess if too many themes it will be slow, but acceptable, because python slow as is.
+        # current order in theme:
+        # ('infected', 'bio_expr', 'infected days', 'pathogen remaining')
+        # UA theme
+        (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> йобнув.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
+         r"([0-9\.\,k]+) біо-ресурса",
+         default_infected_days_theme,
+         default_pathogen_remaining_theme),
+        # RU theme
+        (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> подверг.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
+         default_bioexpr_theme,
+         default_infected_days_theme,
+         default_pathogen_remaining_theme),
+        # EN theme
+        (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> infected.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
+         r"([0-9\.\,k]+) pcs\.",
+         r' for ([0-9\ ]+) d.*',
+         r'Remaining: ([0-9\ ]+)'),
+        # AZ theme
+        (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> сикди.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
+         r"верир: ([0-9\.\,k]+)",
+         default_infected_days_theme,
+         default_pathogen_remaining_theme),
+        # "ПК гик" theme
+        (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> насрал.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
+         r"потеряет: ([0-9\.\,k]+)",
+         default_infected_days_theme,
+         default_pathogen_remaining_theme),
+        # "Новогодняя" theme
+        (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> подверг заморозке.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
+         default_bioexpr_theme,
+         default_infected_days_theme,
+         default_pathogen_remaining_theme),
+        # "Сексуальная индустрия" theme
+        (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a>.+выебал.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
+         r"кончила ([0-9\.\,k]+)",
+         r' ещё ([0-9\ ]+) д.*',
+         default_pathogen_remaining_theme),
+        # "Аферисты в сетях" theme
+        (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> атаковал.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
+         r"приносит: ([0-9\.\,k]+)",
+         default_infected_days_theme,
+         default_pathogen_remaining_theme),
+        # UA theme [via trust]
+        (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> за допомогою довіреності зазнала зараження.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
+         r"([0-9\.\,k]+) біо-ресурса",
+         default_infected_days_theme,
+         default_pathogen_remaining_theme),
+        # RU theme [via trust]
+        (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> при помощи доверенности подвергла заражению.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
+         default_bioexpr_theme,
+         default_infected_days_theme,
+         default_pathogen_remaining_theme),
+        # EN theme [via trust]
+        (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> by authorization infected.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
+         r"([0-9\.\,k]+) pcs\.",
+         r' for ([0-9\ ]+) d.*',
+         r'Remaining: ([0-9\ ]+)'),
+        # idk what is theme [via trust]
+        (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> при помощи анонимуса атаковала.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
+         r'приносит: ([0-9\.\,k]+)',
+         default_infected_days_theme,
+         default_pathogen_remaining_theme),
+    )
+    for theme in bio_attack_themes:
+        trying_theme_index = bio_attack_themes.index(theme)
+        logger.debug(f'trying theme {trying_theme_index}...')
+        r = re.findall(theme[0], text)
+        if r:
+            logger.debug(f'found theme {trying_theme_index}')
+            return r, bio_attack_themes[trying_theme_index]
+    return r, None
+
+
+
 async def eb(client, c, conn, con, d, get_id, my_id, message_q):
     @client.on(events.NewMessage(
                pattern='.*йобнув.*|.*подверг(ла)?.*|.*infected.*|.*сикди.*|.*насрал.*|.*выебал.*|.*за допомогою довіреності.*|.*by authorization infected.*|.*при помощи анонимуса атаковала.*',
@@ -24,87 +106,10 @@ async def eb(client, c, conn, con, d, get_id, my_id, message_q):
         logger.debug(f"in chat '{chat_name}'")
         states.stats_most_infect_spam_chats[chat_name] += 1
         t = m.raw_text
-        # NOTE: theme hell... any ideas for improvment required
-        # but not use huge regular expression like|that|fuckin|way|a|aaaa|aaaaaaaa
-        # because it makes re.findall like mess...
-        default_bioexpr_theme = r"Прибыль: ([0-9\.\,k]+)"
-        default_infected_days_theme = r' на ([0-9\ ]+) д.*'
-        default_pathogen_remaining_theme = r'Осталось: ([0-9\ ]+)'
-        bio_attack_themes = (  # I guess if too many themes it will be slow, but acceptable, because python slow as is.
-            # current order in theme:
-            # ('infected', 'bio_expr', 'infected days', 'pathogen remaining')
-            # UA theme
-            (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> йобнув.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
-             r"([0-9\.\,k]+) біо-ресурса",
-             default_infected_days_theme,
-             default_pathogen_remaining_theme),
-            # RU theme
-            (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> подверг.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
-             default_bioexpr_theme,
-             default_infected_days_theme,
-             default_pathogen_remaining_theme),
-            # EN theme
-            (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> infected.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
-             r"([0-9\.\,k]+) pcs\.",
-             r' for ([0-9\ ]+) d.*',
-             r'Remaining: ([0-9\ ]+)'),
-            # AZ theme
-            (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> сикди.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
-             r"верир: ([0-9\.\,k]+)",
-             default_infected_days_theme,
-             default_pathogen_remaining_theme),
-            # "ПК гик" theme
-            (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> насрал.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
-             r"потеряет: ([0-9\.\,k]+)",
-             default_infected_days_theme,
-             default_pathogen_remaining_theme),
-            # "Новогодняя" theme
-            (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> подверг заморозке.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
-             default_bioexpr_theme,
-             default_infected_days_theme,
-             default_pathogen_remaining_theme),
-            # "Сексуальная индустрия" theme
-            (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a>.+выебал.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
-             r"кончила ([0-9\.\,k]+)",
-             r' ещё ([0-9\ ]+) д.*',
-             default_pathogen_remaining_theme),
-            # "Аферисты в сетях" theme
-            (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> атаковал.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
-             r"приносит: ([0-9\.\,k]+)",
-             default_infected_days_theme,
-             default_pathogen_remaining_theme),
-            # UA theme [via trust]
-            (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> за допомогою довіреності зазнала зараження.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
-             r"([0-9\.\,k]+) біо-ресурса",
-             default_infected_days_theme,
-             default_pathogen_remaining_theme),
-            # RU theme [via trust]
-            (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> при помощи доверенности подвергла заражению.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
-             default_bioexpr_theme,
-             default_infected_days_theme,
-             default_pathogen_remaining_theme),
-            # EN theme [via trust]
-            (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> by authorization infected.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
-             r"([0-9\.\,k]+) pcs\.",
-             r' for ([0-9\ ]+) d.*',
-             r'Remaining: ([0-9\ ]+)'),
-            # idk what is theme [via trust]
-            (r'<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">.*</a> при помощи анонимуса атаковала.+<a href="(tg://openmessage\?user_id=\d+|https://t\.me/\w+)">',
-             r'приносит: ([0-9\.\,k]+)',
-             default_infected_days_theme,
-             default_pathogen_remaining_theme),
-        )
-
         if len(m.entities) > 1:
             h = utils.sanitize_parse_mode(
                 'html').unparse(t, m.entities)  # HTML
-            for theme in bio_attack_themes:
+            r, bio_attack_theme = find_infect_theme(h)
-                trying_theme_index = bio_attack_themes.index(theme)
-                logger.debug(f'trying theme {trying_theme_index}...')
-                r = re.findall(theme[0], h)
-                if r:
-                    logger.debug(f'found theme {trying_theme_index}')
-                    break
             if r == []:
                 logger.warning(
                     'theme not found or lost part of message, showing original message: ' + m.text)
@@ -116,12 +121,11 @@ async def eb(client, c, conn, con, d, get_id, my_id, message_q):
                 u2id = await get_id(u2url)
                 bio_excludes = [x[0] for x in c.execute(
                     'select user_id from avocado_exclude').fetchall()]
-                # print(f'{u1url} [@{u1id}] подверг(ла) {u2url} [@{u2id}]')#показать
                 when = int(datetime.timestamp(m.date))
-                days = int(re.findall(bio_attack_themes[trying_theme_index][2], t)[
+                days = int(re.findall(bio_attack_theme[2], t)[
                            0].replace(' ', ''))
                 experience = re.findall(
-                    bio_attack_themes[trying_theme_index][1], t)[0].strip()
+                    bio_attack_theme[1], t)[0].strip()
                 if ',' in experience:
                     experience = re.sub(r',', r'.', experience)
                 if 'k' in experience:
@@ -130,7 +134,7 @@ async def eb(client, c, conn, con, d, get_id, my_id, message_q):
                 else:
                     exp_int = int(experience)
                 pathogen_remaining = int(re.findall(
-                    bio_attack_themes[trying_theme_index][3], t)[0])
+                    bio_attack_theme[3], t)[0])
                 if pathogen_remaining <= states.auto_bioeb_pathogen_threshold and u1id == my_id:
                     states.auto_bioeb_sleep_interval = states.auto_bioeb_max_interval
                     logger.warning(