delojza/delojza.py

#!/usr/bin/env python3

import errno
import logging
import os
import pprint
import re
import shutil
import sqlite3
import subprocess
import sys
import tempfile
import unicodedata
from configparser import ConfigParser
from datetime import datetime, timedelta
from glob import glob
from operator import itemgetter
from random import random
from time import sleep

import acoustid
import filetype
import markovify
import mutagen.id3
import pytumblr
import requests
import telegram
import youtube_dl
from telegram.ext import Updater, CommandHandler, MessageHandler
from youtube_dl import DownloadError
from youtube_dl.version import __version__ as YTDL_VERSION


def mkdir_p(path):
    try:
        os.makedirs(path)
    except OSError as exc:
        if exc.errno == errno.EEXIST and os.path.isdir(path):
            pass
        else:
            raise


def datestr(date):
    return date.strftime("%Y-%m-%d@%H%M")


class DelojzaDB:
    def __init__(self, db_path):
        self.db_path = db_path
        self.db = None

    def initialize(self):
        if self.db is None:
            self.db = sqlite3.connect(self.db_path)

    def get_protected_tags(self):
        results = self.db.execute("SELECT tag FROM tags WHERE protected == 1")
        return [res[0] for res in results.fetchall()]

    def get_protected_chats(self):
        results = self.db.execute("SELECT id FROM chats WHERE protected == 1")
        return [res[0] for res in results.fetchall()]

    def get_chat(self, id):
        return self.db.execute("SELECT id, protected FROM chats WHERE id == ?", (id,)).fetchone()

    def set_chat_protected(self, id, protected):
        chat_in_db = self.get_chat(id)
        if chat_in_db:
            self.db.execute("UPDATE chats SET protected = ? WHERE id = ?", (protected, id))
        else:
            self.db.execute("INSERT INTO chats (id, protected) VALUES (?, ?)", (id, protected))
        self.db.commit()

    def get_tag(self, tag):
        return self.db.execute("SELECT id, tag, protected FROM tags WHERE tag == ?", (tag,)).fetchone()

    def set_tag_protected(self, tag, protected):
        tag_in_db = self.get_tag(tag)
        if tag_in_db:
            self.db.execute("UPDATE tags SET protected = ? WHERE tag = ?", (protected, tag))
        else:
            self.db.execute("INSERT INTO tags (tag, protected) VALUES (?, ?)", (tag, protected))
        self.db.commit()


class DelojzaBot:
    def __init__(self, tg_api_key, out_dir, tmp_dir=None, db_path=None, protected_password=None,
                 acoustid_key=None, tumblr_name=None, tumblr_keys=None, markov=None):
        self.logger = logging.getLogger("delojza")

        self.db = DelojzaDB(db_path or os.path.join(os.path.dirname(os.path.realpath(__file__)), "delojza.db"))

        self.out_dir = os.path.abspath(out_dir)
        self.logger.debug('OUT_DIR: ' + out_dir)
        self.tmp_dir = tmp_dir if tmp_dir else tempfile.gettempdir()
        self.logger.debug('TMP_DIR: ' + tmp_dir)
        self.markov = markov

        self.updater = Updater(tg_api_key)
        dp = self.updater.dispatcher

        dp.add_handler(CommandHandler("start", self.tg_start))
        dp.add_error_handler(self.tg_error)
        dp.add_handler(CommandHandler("stats", self.tg_stats))
        dp.add_handler(CommandHandler("orphans", self.tg_orphan))
        dp.add_handler(CommandHandler("orphans_full", self.tg_orphan_full))
        dp.add_handler(CommandHandler("retag", self.tg_retag))
        dp.add_handler(CommandHandler("delete", self.tg_delete))
        dp.add_handler(CommandHandler("protect", self.tg_protect))
        dp.add_handler(CommandHandler("version", self.tg_version))
        dp.add_handler(MessageHandler(None, self.tg_handle))

        self.acoustid_key = acoustid_key

        if tumblr_name and tumblr_keys:
            self.tumblr_name = tumblr_name
            self.tumblr_client = pytumblr.TumblrRestClient(*tumblr_keys)
        else:
            self.tumblr_client = None

        self.protected_password = protected_password
        self.last_downloaded = {}
        self.last_hashtags = {}

    @staticmethod
    def ytdl_can(url):
        ies = youtube_dl.extractor.gen_extractors()
        for ie in ies:
            if ie.suitable(url) and ie.IE_NAME != 'generic' \
                    and '/channel/' not in url:
                # Site has dedicated extractor
                return True
        return False

    # https://github.com/django/django/blob/master/django/utils/text.py#L393
    @staticmethod
    def sanitize(filepath):
        if filepath is None:
            return None
        filepath = unicodedata.normalize('NFKD', filepath).encode('ascii', 'ignore').decode('ascii')
        return re.sub(r'[^\w.()\[\]{}#-]', '_', filepath)

    @staticmethod
    def _tag_file(filepath, artist, title):
        try:
            id3 = mutagen.id3.ID3(filepath)
        except mutagen.id3.ID3NoHeaderError:
            mutafile = mutagen.File(filepath)
            mutafile.add_tags()
            mutafile.save()
            id3 = mutagen.id3.ID3(filepath)
        id3.add(mutagen.id3.TIT2(encoding=3, text=title))
        if artist:
            id3.add(mutagen.id3.TOPE(encoding=3, text=artist))
            id3.add(mutagen.id3.TPE1(encoding=3, text=artist))
        id3.save()

    def _autotag_file(self, filepath, message, info=None):
        if info is None:
            info = {}

        title = None
        artist = None
        source = None

        best_acoustid_score = 0

        if self.acoustid_key:
            try:
                self.logger.debug("Requesting AcoustID for {}".format(filepath))
                results = sorted(acoustid.match(self.acoustid_key, filepath), key=itemgetter(0), reverse=True)
                if len(results) > 0:
                    score, rid, aid_title, aid_artist = results[0]
                    if score > .4:
                        title = aid_title
                        artist = aid_artist
                        best_acoustid_score = score
                        source = "AcoustID ({}%)".format(round(score * 100))
            except acoustid.NoBackendError:
                self.logger.warning("chromaprint library/tool not found")
            except acoustid.FingerprintGenerationError:
                self.logger.warning("fingerprint could not be calculated")
            except acoustid.WebServiceError as exc:
                self.logger.warning("web service request failed: {}".format(exc.message))

        if best_acoustid_score < .8:
            if 'track' in info:
                title = info['track']
            if 'artist' in info:
                artist = info['artist']

            if 'track' in info or 'artist' in info:
                source = "supplied metadata"

            if title is None and artist is None and '-' in info.get("title", ""):
                split = info['title'].split("-")
                artist = split[0]
                title = split[1]
                source = "fallback (artist - title)"

            if title is None and 'title' in info:
                title = info['title']
                source = "full title fallback"

            if 'soundcloud' in info.get("extractor", "") and artist is None:
                artist = info['uploader']
                source = "soundcloud \"fallback\""

        artist = artist.strip() if artist else None
        title = title.strip() if title else None

        if title is None and artist is None:
            message.reply_text("Tried tagging, found nothing :(")
            return

        message.reply_text("Tagging as \"{}\" by \"{}\"\nvia {}".format(title, artist, source))
        self.logger.info("Tagging {} w/ {} - {} [{}]...".format(filepath, title, artist, source))
        self._tag_file(filepath, artist, title)

    # noinspection PyUnusedLocal
    def download_ytdl(self, urls, out_path, date, message, audio=False, filetitle=None):
        ytdl = {
            'noplaylist': True,
            'restrictfilenames': True,
            'outtmpl': os.path.join(self.tmp_dir, '{}__%(title)s__%(id)s.%(ext)s'.format(datestr(date)))
        }
        if audio:
            ytdl['format'] = 'bestaudio/best'
            ytdl['postprocessors'] = [{
                'key': 'FFmpegExtractAudio',
                'preferredcodec': 'mp3',
                'preferredquality': '256'
            }]
            ytdl['postprocessor_args'] = ['-ar', '44100']
        filenames = []
        with youtube_dl.YoutubeDL(ytdl) as ytdl:
            attempts = 0
            while True:
                try:
                    ytdl.download(urls)
                    break
                except DownloadError as exc:
                    attempts += 1
                    if '403' in str(exc) and attempts < 5:
                        self.logger.warning("Received a 403!")
                        sleep(1.357)
                        if self.markov:
                            message.reply_text(self.markov.make_sentence)
                    else:
                        raise exc
            for info in [ytdl.extract_info(url, download=False) for url in urls]:
                filename = ytdl.prepare_filename(info)
                globbeds = glob(os.path.splitext(filename)[0] + '.*')
                for globbed in globbeds:
                    if globbed.endswith("mp3"):
                        self._autotag_file(globbed, message, info=info)
                    self.logger.info("Moving %s to %s..." % (globbed, out_path))
                    dest = shutil.move(globbed, out_path)
                    filenames.append(dest)
        return filenames

    def download_raw(self, urls, out_path, date, message, audio=False, filetitle=None):
        filenames = []
        for url in urls:
            local_filename = os.path.join(out_path, "{}__{}".format(datestr(date),
                                                                    self.sanitize(filetitle or url.split('/')[-1])))
            final_filename = local_filename
            is_mp3 = local_filename.endswith("mp3")

            r = requests.get(url, stream=True)
            with open(local_filename, 'wb') as f:
                for chunk in r.iter_content(chunk_size=1024):
                    if chunk:
                        f.write(chunk)

            if not re.match(r'.*\..{3,5}$', os.path.split(local_filename)[-1]):
                kind = filetype.guess(local_filename)
                if kind is None:
                    self.logger.error("File has no extension and could not be determined!")
                else:
                    self.logger.info('Moving file without extension... %s?' % kind.extension)
                    final_filename = shutil.move(local_filename, local_filename + '.' + kind.extension)
                    is_mp3 = kind.extension == "mp3"

            filenames.append(final_filename)

            if audio and is_mp3:
                try:
                    id3 = mutagen.id3.ID3(final_filename)
                    untagged = 'TIT2' not in id3
                except mutagen.id3.ID3NoHeaderError:
                    untagged = True
                if untagged:
                    self._autotag_file(final_filename, message)

        return filenames

    @staticmethod
    def extract_hashtags(message):
        hashtags = list(map(message.parse_entity,
                            list(filter(lambda e: e.type == 'hashtag', message.entities))))
        hashtags += list(map(message.parse_caption_entity,
                             list(filter(lambda e: e.type == 'hashtag', message.caption_entities))))
        if len(hashtags) > 0:
            hashtags = [hashtag[1:].upper() for hashtag in hashtags]
            for i, hashtag in enumerate(hashtags):
                if "PRAS" in hashtag:
                    hashtags[i] = "PRAS"
        return hashtags

    def _get_hashtags(self, message):
        hashtags = self.extract_hashtags(message)
        if len(hashtags) == 0 and self.last_hashtags.get(message.chat.id) is not None:
            user, ts, last_hashtags = self.last_hashtags[message.chat.id]
            if user == message.from_user and ts > datetime.now() - timedelta(hours=1):
                hashtags = last_hashtags
        return hashtags

    def handle_text(self, message, hashtags):
        if len(hashtags) == 0 or hashtags[0] not in ('TEXT', 'TXT'):
            return

        info_line = self.sanitize("-".join(re.sub(r'#[\w]+', '', message.text).strip().split()[:7]))
        if len(info_line) > 64:
            info_line = info_line[:64]

        filename = '{}__{}.txt'.format(datestr(message.date), info_line)

        out_path = os.path.join(self.out_dir, *hashtags[1:] or ['TEXT'])
        file_path = os.path.join(out_path, filename)

        mkdir_p(out_path)

        with open(file_path, 'w') as out_file:
            out_file.write(message.text)

        message.reply_text("Saved text to \"{}\"...".format(os.path.join(*hashtags[1:] or ['TEXT'], filename)))

    # noinspection PyBroadException
    def handle(self, urls, message, hashtags, download_fn, filetitle=None):
        self.db.initialize()

        try:
            if len(hashtags) == 0:
                self.logger.info("Ignoring %s due to no hashtag present..." % urls)
                return False

            if any(hashtag in self.db.get_protected_tags() for hashtag in hashtags):
                if message.chat.id not in self.db.get_protected_chats():
                    self.logger.info("Redirecting {} in chat {} due to protected hashtags: {}..."
                                     .format(urls, message.chat.title, hashtags))
                    hashtags.insert(0, "PUBLIC")

            self.last_hashtags[message.chat.id] = None

            self.logger.info("Downloading %s under '%s'" % (urls, "/".join(hashtags)))

            out_path = os.path.join(self.out_dir, *hashtags)
            mkdir_p(out_path)

            reply = 'Downloading to "{}"...'.format("/".join(hashtags))

            audio = any([any([tag in hashtag for tag in ('AUDIO', 'RADIO')]) for hashtag in hashtags])
            if audio and download_fn != self.download_raw:
                reply += ' (And also guessing you want to extract the audio)'
            message.reply_text(reply)

            filenames = download_fn(urls, out_path, message.date, message, audio=audio, filetitle=filetitle)

            cmd_hashtag = hashtags[0]

            tumblr_ids = []
            if cmd_hashtag in ('TUMBLR', 'TUMBLR_NOW') and self.tumblr_client:
                now = cmd_hashtag == 'TUMBLR_NOW'
                reply = '(btw, {})'.format("***FIRING TO TUMBLR RIGHT AWAY***" if now else "queueing to tumblr")
                message.reply_text(reply, parse_mode=telegram.ParseMode.MARKDOWN)
                for filename in filenames:
                    if filename.endswith(".mp4"):
                        try:
                            output_filename = filename[:-len(".mp4")] + ".gif"
                            subprocess.check_output(['ffmpeg', '-i', filename, output_filename])
                            filename = output_filename
                        except subprocess.CalledProcessError:
                            message.reply_text("Conversion to gif failed, sorry! Check log...")
                            continue
                    response = self.tumblr_client.create_photo(self.tumblr_name, data=filename,
                                                               state="published" if now else "queue")
                    if 'id' in response:
                        tumblr_ids.append(response['id'])
                    else:
                        self.logger.warning("Did not receive 'id' in tumblr response: \n" + pprint.pformat(response))
                        message.reply_text('Something weird happened with the tumblrs, check it!')

            self.last_downloaded[message.chat.id] = filenames, hashtags, tumblr_ids
            return True
        except:
            exc_type, exc_value, __ = sys.exc_info()
            if "Timed out" not in str(exc_value):
                message.reply_text("Something is FUCKED: [{}] {}".format(exc_type, exc_value))
            return False

    def handle_tg_message(self, message, bot, hashtag):
        file, filetitle, tumblr = None, None, False
        if len(message.photo) > 0:
            photo = max(message.photo, key=lambda p: p.width)
            file = photo.file_id
        elif message.document is not None:
            filetitle = message.document.file_name
            file = message.document.file_id
        elif message.audio is not None:
            filetitle = message.audio.title
            file = message.audio.file_id
        elif message.video is not None:
            file = message.video.file_id
        elif message.video_note is not None:
            file = message.video_note.file_id
        elif message.voice is not None:
            file = message.voice.file_id

        if file is not None:
            url = bot.getFile(file).file_path
            return self.handle([url], message, hashtag, self.download_raw, filetitle=filetitle)
        else:
            return False

    def handle_urls(self, message, hashtags):
        urls = list(map(lambda e: message.parse_entity(e),
                        filter(lambda e: e.type == 'url', message.entities)))

        ytdl_res = False
        ytdl_urls = [url for url in urls if self.ytdl_can(url)]
        if len(ytdl_urls) > 0:
            ytdl_res = self.handle(ytdl_urls, message, hashtags, self.download_ytdl)

        raw_res = False
        normal_urls = [url for url in urls if not self.ytdl_can(url)]
        if len(normal_urls) > 0:
            file_urls = [url for url in normal_urls if
                         "text" not in requests.head(url).headers.get("Content-Type", "text")]
            if len(file_urls) > 0:
                raw_res = self.handle(file_urls, message, hashtags, self.download_raw)

        return ytdl_res or raw_res

    def tg_handle(self, bot, update):
        url_res = self.handle_urls(update.message, self._get_hashtags(update.message))
        if url_res:
            return

        msg_res = self.handle_tg_message(update.message, bot, self._get_hashtags(update.message))
        if msg_res:
            return

        hashtags = self.extract_hashtags(update.message)
        if len(hashtags) > 0:
            self.handle_text(update.message.reply_to_message or update.message, hashtags)

            if update.message.reply_to_message:
                self.handle_tg_message(update.message.reply_to_message, bot, hashtags)
                self.handle_urls(update.message.reply_to_message, hashtags)
            else:
                self.last_hashtags[update.message.chat.id] = update.message.from_user, datetime.now(), hashtags
        else:
            if self.markov:
                self.markov.add_to_corpus(update.message.text)

    def _get_tag_dirs(self):
        return list(filter(lambda x: x.upper() == x,
                           filter(lambda directory: os.path.isdir(os.path.join(self.out_dir, directory)),
                                  os.listdir(self.out_dir))))

    def tg_stats(self, _, update):
        tag_dirs = self._get_tag_dirs()
        reply = "Total number of tags: {}\n\n".format(len(tag_dirs))
        counts = [(directory, os.listdir(os.path.join(self.out_dir, directory))) for directory in tag_dirs]
        counts.sort(key=itemgetter(0))
        counts.sort(key=lambda x: len(x[1]), reverse=True)
        for directory, files in counts:
            if len(files) == 1:
                break
            abs_paths = [os.path.join(self.out_dir, directory, file) for file in files]
            abs_files = list(filter(os.path.isfile, abs_paths))
            # mimes = [magic.from_file(path, mime=True).split("/")[0] for path in abs_files]
            # mime_counts = [(mime, mimes.count(mime)) for mime in set(mimes)]
            exts = [ext[1:] for ext in [os.path.splitext(path)[1] for path in abs_files] if len(ext) > 0]
            ext_counts = [(ext, exts.count(ext)) for ext in set(exts)]
            dir_cnt = len(abs_paths) - len(abs_files)
            type_counts = ext_counts + ([("directorie", dir_cnt)] if dir_cnt > 0 else [])
            details = ", ".join(["{} {}s".format(cnt, mime) for mime, cnt in
                                 sorted(type_counts, key=itemgetter(1), reverse=True)])
            if len(type_counts) == 1:
                reply += "<b>{}:</b> {}\n".format(directory, details)
            else:
                reply += "<b>{}:</b> {} files ({})\n".format(directory, len(files), details)
        orphans = list(filter(lambda cnt: len(cnt[1]) <= 1, counts))
        if len(orphans) > 0:
            reply += "\nFollowing tags are orphans: " + ", ".join(map(itemgetter(0), orphans))
        update.message.reply_text(reply, parse_mode=telegram.ParseMode.HTML)

    def _get_orphan_tags(self):
        result = []
        for directory in self._get_tag_dirs():
            files = os.listdir(os.path.join(self.out_dir, directory))
            if len(files) == 1:
                result.append((directory, files[0]))
            if len(files) == 0:
                result.append((directory, "NO FILE AT ALL..."))
        return sorted(result, key=itemgetter(0))

    def tg_orphan(self, _, update):
        orphans = self._get_orphan_tags()
        if len(orphans) == 0:
            update.message.reply_text("Good job, no orphan tags!")
        else:
            update.message.reply_text("The following tags only contain a single file:\n" +
                                      ", ".join(map(itemgetter(0), orphans)))

    def tg_orphan_full(self, _, update):
        orphans = self._get_orphan_tags()
        if len(orphans) == 0:
            update.message.reply_text("Good job, no orphan tags!")
        else:
            tmp_reply = "The following tags only contain a single file:\n"
            for directory, file in orphans:
                line = "{}: {}\n".format(directory, file)
                if len(tmp_reply + line) > 4096:
                    update.message.reply_text(tmp_reply)
                    tmp_reply = ""
                tmp_reply += line
            if len(tmp_reply) > 0:
                update.message.reply_text(tmp_reply)

    def tg_retag(self, _, update):
        if self.last_downloaded.get(update.message.chat.id) is not None and " - " in update.message.text:
            files, hashtags, tumblr_ids = self.last_downloaded[update.message.chat.id]
            mp3s = [filename for filename in files if filename.endswith("mp3")]
            if len(mp3s) > 0:
                tagline = re.sub(r'^/[\w]+', '', update.message.text).split(" - ")
                artist = tagline[0].strip()
                title = tagline[1].strip()
                for mp3 in mp3s:
                    self._tag_file(mp3, artist, title)
                    update.message.reply_text("Tagging \"{}\" as \"{}\" by \"{}\"!"
                                              .format(mp3[len(self.out_dir) + 1:], title, artist))
                return
        update.message.reply_text((self.markov.make_sentence() if self.markov and random() > .7 else "") + "???")

    def tg_delete(self, _, update):
        if self.last_downloaded.get(update.message.chat.id) is not None:
            files, hashtags, tumblr_ids = self.last_downloaded[update.message.chat.id]
            for file in files:
                update.message.reply_text("Removing \"{}\"!".format(file[len(self.out_dir) + 1:]))
                os.remove(file)
                parent_dir = os.path.dirname(file)
                while True:
                    if len(os.listdir(parent_dir)) == 0:
                        update.message.reply_text("Removing directory \"{}\" as it's empty..."
                                                  .format(parent_dir[len(self.out_dir) + 1:]))
                        os.rmdir(parent_dir)
                    parent_dir = os.path.dirname(parent_dir)
                    if parent_dir == self.out_dir:
                        break
            if len(tumblr_ids) > 0:
                plural = "s (all {} of them)".format(len(tumblr_ids)) if len(tumblr_ids) > 1 else ""
                update.message.reply_text("Also deleting tumblr post{}!".format(plural))
                for tumblr_id in tumblr_ids:
                    if self.tumblr_client:
                        self.tumblr_client.delete_post(self.tumblr_name, tumblr_id)
            self.last_downloaded[update.message.chat.id] = None
            return
        update.message.reply_text("Nothing to remove!")

    def tg_protect(self, _, update):
        self.db.initialize()

        msg_split = update.message.text.split(" ")
        if len(msg_split) != 3:
            update.message.reply_text((self.markov.make_sentence() if self.markov and random() > .7 else "") + "???")
            return

        chat_in_db = self.db.get_chat(update.message.chat.id)

        cmd = msg_split[1]
        if cmd == 'tag':
            if chat_in_db and chat_in_db[1]:
                tag = msg_split[2].upper()
                tag_in_db = self.db.get_tag(tag)
                if tag_in_db:
                    _, _, protected = tag_in_db
                    end_protected = not protected
                else:
                    end_protected = True

                self.db.set_tag_protected(tag, end_protected)
                update.message.reply_text(f"got it, will {'NOT ' if not end_protected else ''}protect tag {tag}!")
            else:
                update.message.reply_text((self.markov.make_sentence() if self.markov and random() > .7 else "hublubl"))
        elif cmd == 'chat':
            password = msg_split[2]
            if password == self.protected_password:
                if chat_in_db:
                    _, protected = chat_in_db
                    end_protected = not protected
                else:
                    end_protected = True

                self.db.set_chat_protected(update.message.chat.id, end_protected)
                update.message.reply_text(f"got it, will {'NOT ' if not end_protected else ''}protect this chat!")
            else:
                update.message.reply_text((self.markov.make_sentence() if self.markov and random() > .7 else "hublubl"))
        else:
            update.message.reply_text((self.markov.make_sentence() if self.markov and random() > .7 else "") + "???")

    # noinspection PyMethodMayBeStatic
    def tg_version(self, _, update):
        delojza_date = datetime.fromtimestamp(os.path.getmtime(os.path.realpath(__file__))) \
            .strftime('%Y/%m/%d - %H:%M:%S')
        update.message.reply_text("delojza modified date: {}\nyoutube-dl version: {}"
                                  .format(delojza_date, YTDL_VERSION))

    def tg_start(self, _, update):
        update.message.reply_text(self.markov.make_sentence() if self.markov else "HELLO")

    def tg_error(self, bot, update, error):
        self.logger.error(error)
        if "Timed out" in str(error):
            if update is not None:
                default = "Mmmm, I like it..."
                update.message.reply_text((self.markov.make_sentence(tries=100) if self.markov else default) or default)
                self.tg_handle(bot, update)
        else:
            if update is not None:
                update.message.reply_text("Something is fucked: %s" % error)

    def run_idle(self):
        self.updater.start_polling()
        self.logger.info("Started Telegram bot...")
        self.updater.idle()


class MarkovBlabberer:
    def __init__(self, filepath):
        self.logger = logging.getLogger('markov')
        self.filepath = filepath

        with open(filepath) as f:
            text = f.read()
            self.markov = markovify.NewlineText(text.lower())
            self.logger.info("Sentence of the day: " + self.make_sentence())

    def make_sentence(self, tries=100):
        return self.markov.make_sentence(tries=tries) or "???"

    def add_to_corpus(self, text):
        text = text.lower()
        new_sentence = markovify.NewlineText(text)
        self.markov = markovify.combine([self.markov, new_sentence])
        with open(self.filepath, 'a') as f:
            f.write(text + '\n')


if __name__ == '__main__':
    logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(name)s - %(levelname)s - %(message)s')

    _DIR_ = os.path.dirname(os.path.realpath(__file__))
    CONFIG_PATHS = ['/etc/delojza/delojza.ini',
                    os.path.join(os.getenv("HOME") or "", ".config/delojza/delojza.ini"),
                    os.path.join(_DIR_, "delojza.ini")]

    config = ConfigParser()
    try:
        CONF_FILE = next(conf_path for conf_path in CONFIG_PATHS if os.path.isfile(conf_path))
        config.read(CONF_FILE)
    except StopIteration:
        logging.error("No config file found, stopping.")
        sys.exit(-1)

    try:
        markov = MarkovBlabberer("initial.txt")
    except FileNotFoundError:
        logging.warning("Didn't find `initial.txt`, continuing without markov blabbering!")
        markov = None

    delojza = DelojzaBot(config.get('delojza', 'tg_api_key'),
                         config.get('delojza', 'OUT_DIR', fallback=os.path.join(_DIR_, "out")),
                         tmp_dir=config.get('delojza', 'tmp_dir', fallback=tempfile.gettempdir()),
                         protected_password=config.get('delojza', 'protected_password', fallback=None),
                         acoustid_key=config.get('delojza', 'acoustid_api_key', fallback=None),
                         tumblr_name=config.get('tumblr', 'blog_name', fallback=None),
                         tumblr_keys=(config.get('tumblr', 'consumer_key', fallback=None),
                                      config.get('tumblr', 'consumer_secret', fallback=None),
                                      config.get('tumblr', 'oauth_key', fallback=None),
                                      config.get('tumblr', 'oauth_secret', fallback=None)),
                         markov=markov)
    delojza.run_idle()