delojza/delojza.py

#!/usr/bin/env python3

import errno
import logging
import os
import pprint
import re
import shutil
import sys
import tempfile
import unicodedata
from configparser import ConfigParser
from datetime import datetime, timedelta
from glob import glob
from operator import itemgetter

import acoustid
import filetype
import markovify
import mutagen.id3
import pytumblr
import requests
import telegram
import youtube_dl
from telegram import MessageEntity
from telegram.ext import Updater, CommandHandler, MessageHandler, Filters
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 DelojzaBot:
    def __init__(self, tg_api_key, out_dir, tmp_dir=None,
                 protected_chats=None, protected_tags=None,
                 acoustid_key=None, tumblr_name=None, tumblr_keys=None, markov=None):
        self.logger = logging.getLogger("delojza")

        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)
        self.tg_url_handler = MessageHandler(Filters.entity(MessageEntity.URL), self.tg_handle_url)
        dp.add_handler(self.tg_url_handler)
        self.tg_rest_handler = MessageHandler(Filters.photo | Filters.video | Filters.video_note |
                                              Filters.audio | Filters.voice | Filters.document, self.tg_handle_rest)
        dp.add_handler(self.tg_rest_handler)
        dp.add_handler(MessageHandler(Filters.entity(MessageEntity.HASHTAG), self.tg_handle_hashtag))
        dp.add_handler(MessageHandler(Filters.text, self.tg_handle_text))
        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("delete", self.tg_delete))
        dp.add_handler(CommandHandler("version", self.tg_version))

        self.acoustid_key = acoustid_key

        if tumblr_name and tumblr_keys:
            self.tumblr_name = tumblr_name
            self.tumblr_client = pytumblr.TumblrRestClient(*tumblr_keys)

        self.protected_chats = protected_chats or []
        self.protected_tags = protected_tags or []

        self.last_downloaded = None
        self.last_hashtags = None

    @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)

    def tag_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))
        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()

    # noinspection PyUnusedLocal
    def download_ytdl(self, urls, out_path, date, message, audio=False, filetitle=None):
        ydl_opts = {
            'noplaylist': True,
            'restrictfilenames': True,
            'outtmpl': os.path.join(self.tmp_dir, '{}__%(title)s__%(id)s.%(ext)s'.format(datestr(date)))
        }
        if audio:
            ydl_opts['format'] = 'bestaudio/best'
            ydl_opts['postprocessors'] = [{
                'key': 'FFmpegExtractAudio',
                'preferredcodec': 'mp3',
                'preferredquality': '256'
            }]
        filenames = []
        with youtube_dl.YoutubeDL(ydl_opts) as ydl:
            ydl.download(urls)
            for info in [ydl.extract_info(url, download=False) for url in urls]:
                filename = ydl.prepare_filename(info)
                globbeds = glob(os.path.splitext(filename)[0] + '.*')
                for globbed in globbeds:
                    if globbed.endswith("mp3"):
                        self.tag_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.tag_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 is not None:
            user, chat, ts, last_hashtags = self.last_hashtags
            if user == message.from_user and chat == message.chat and ts > datetime.now() - timedelta(hours=1):
                hashtags = last_hashtags
        return hashtags

    def tg_handle_hashtag(self, bot, update):
        hashtags = self.extract_hashtags(update.message)

        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.from_user, update.message.chat, datetime.now(), 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:], filename)))

    # noinspection PyBroadException
    def handle(self, urls, message, hashtags, download_fn, filetitle=None):
        try:
            if len(hashtags) == 0:
                self.logger.info("Ignoring %s due to no hashtag present..." % urls)
                return

            if any(hashtag in self.protected_tags for hashtag in hashtags):
                if message.chat.title not in self.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 = 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 filetitle in filenames:
                    response = self.tumblr_client.create_photo(self.tumblr_name, data=filetitle,
                                                               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, filenames, hashtags, tumblr_ids
            return filenames
        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))

    def tg_handle_url(self, _, update):
        self.handle_urls(update.message, self.get_hashtags(update.message))

    def handle_urls(self, message, hashtags):
        urls = list(map(lambda e: message.parse_entity(e),
                        filter(lambda e: e.type == 'url', message.entities)))
        ytdl_urls = [url for url in urls if self.ytdl_can(url)]
        normal_urls = [url for url in urls if not self.ytdl_can(url)]
        if len(ytdl_urls) > 0:
            self.handle(ytdl_urls, message, hashtags, self.download_ytdl)
        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:
                self.handle(file_urls, message, hashtags, self.download_raw)

    def tg_handle_rest(self, bot, update):
        self.handle_tg_message(update.message, bot, self.get_hashtags(update.message))

    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
            self.handle([url], message, hashtag, self.download_raw, filetitle=filetitle)

    def tg_handle_text(self, _, update):
        if self.markov:
            self.markov.add_to_corpus(update.message.text)

    def 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.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 orphans(self):
        result = []
        tag_dirs = self.tag_dirs()
        for directory in 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.orphans()
        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.orphans()
        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_delete(self, _, update):
        if self.last_downloaded is not None:
            chat, files, hashtags, tumblr_ids = self.last_downloaded
            if chat == update.message.chat:
                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 = None
                return

        update.message.reply_text("Nothing to remove!")

    # 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_rest(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)

    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:
        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_chats=config.get('delojza', 'protected_chats').split(";"),
                         protected_tags=config.get('delojza', 'protected_tags').split(";"),
                         acoustid_key=config.get('delojza', 'acoustid_api_key'),
                         tumblr_name=config.get('tumblr', 'blog_name'),
                         tumblr_keys=(config.get('tumblr', 'consumer_key'),
                                      config.get('tumblr', 'consumer_secret'),
                                      config.get('tumblr', 'oauth_key'),
                                      config.get('tumblr', 'oauth_secret')),
                         markov=markov)
    delojza.run_idle()