# functions that deal with parsing data from youtube's internal API ("innertube")

from urllib.parse import parse_qs, urlparse

def listfind(obj, key):
    """
    given a list of dicts, where one dict contains a given key, return said key.
    """
    return next(iter([ obj[key] for obj in obj if key in obj.keys() ]),{})
def listget(obj, index, fallback=None):
    return next(iter(obj[index:]), fallback)

def prepare_searchresults(yt_results):
    contents = listfind(yt_results, 'response') \
        .get('contents',{})\
        .get('twoColumnSearchResultsRenderer',{})\
        .get('primaryContents',{})\
        .get('sectionListRenderer',{})\
        .get('contents',[])
    contents = listfind(contents, 'itemSectionRenderer').get('contents',[])

    return parse_result_items(contents)
def prepare_infocards(metadata):
    cards = metadata.get('cards',{}).get('cardCollectionRenderer',{}).get('cards',[])
    return [parse_infocard(card) for card in cards if card is not None]

def prepare_endcards(metadata):
    endsc = metadata.get('endscreen',{}).get('endscreenRenderer',{}).get('elements',[])
    return [parse_endcard(card) for card in endsc if card is not None]

def mkthumbs(thumbs):
    output = {str(e['height']): e['url'] for e in thumbs}
    largest=next(iter(sorted(output.keys(),reverse=True,key=int)),None)
    return {**output, 'largest': largest}

def clean_url(url):
    # externals URLs are redirected through youtube.com/redirect, but we
    # may encounter internal URLs, too
    return parse_qs(urlparse(url).query).get('q',[url])[0]

def toInt(s, fallback=0):
    if s is None:
        return fallback
    try:
        return int(''.join(filter(str.isdigit, s)))
    except ValueError:
        return fallback

# Remove left-/rightmost word from string:
delL = lambda s: s.partition(' ')[2]

def age(s):
    if s is None: # missing from autogen'd music, some livestreams
        return None
    # Some livestreams have "Streamed 7 hours ago"
    s = s.replace("Streamed ","")
    # Now, everything should be in the form "1 year ago"
    value, unit, _ = s.split(" ")
    suffix = dict(
        month='mn',
        months='mn',
    ).get(unit, unit[0]) # first letter otherwise (e.g. year(s) => y)

    return f"{value}{suffix}"

def parse_result_items(items):
    # TODO: use .get() for most non-essential attributes
    """
    parses youtube search response into an easier to use format.
    """
    results = []
    for item in items:
        key = next(iter(item.keys()), None)
        content = item[key]
        if key == 'videoRenderer':
            is_live = listfind(content.get('badges',[]), 'metadataBadgeRenderer').get('style') == 'BADGE_STYLE_TYPE_LIVE_NOW'
            results.append({'type': 'VIDEO', 'content': {
                'video_id': content['videoId'],
                'title': content['title']['runs'][0]['text'],
                'author': content['longBylineText']['runs'][0]['text'] or \
                          content['shortBylineText']['runs'][0]['text'],
                'channel_id': content['ownerText']['runs'][0] \
                    ['navigationEndpoint']['browseEndpoint']['browseId'],
                'length': content.get('lengthText',{}).get('simpleText') \
                    if not is_live else 'LIVE', # "44:07", "1:41:50"
                'views': toInt(content.get('viewCountText',{}).get('simpleText') or # "123,456 views"
                    listget(content.get('viewCountText',{}).get('runs'),0,{}).get('text')), # "1,234 watching"
                'published': age(content.get('publishedTimeText',{}).get('simpleText')),
            }})
        elif key == 'playlistRenderer':
            results.append({'type': 'PLAYLIST', 'content': {
                'playlist_id': content['navigationEndpoint']['watchEndpoint']['playlistId'],
                'video_id': content['navigationEndpoint']['watchEndpoint']['videoId'],
                'title': content['title']['simpleText'],
                'author': content['longBylineText']['runs'][0]['text'] or
                     content['shortBylineText']['runs'][0]['text'],
                'channel_id': content['longBylineText']['runs'][0]['navigationEndpoint']['browseEndpoint']['browseId'], # OR .shortBylineText
                'n_videos': toInt(content['videoCount']),
            }})
        elif key == 'radioRenderer': # "Mix" playlists
            results.append({'type': 'PLAYLIST', 'content': {
                'playlist_id': content['playlistId'],
                'video_id': content['navigationEndpoint']['watchEndpoint']['videoId'],
                'title': content['title']['simpleText'],
                'author': content['longBylineText']['simpleText'] or \
                    content['shortBylineText']['simpleText'] , # always "YouTube"
                'channel_id': None,
                'n_videos': content['videoCountShortText']['runs'][0]['text'] or \
                    content['videoCountText']['runs'][0]['text'],
                    # videoCountShortText: "50+"; videoCountText: "50+ videos"
            }})
        elif key == 'channelRenderer':
            results.append({'type': 'CHANNEL', 'content': {
                'channel_id': content['channelId'],
                'title': content['title']['simpleText'],
                'icons': mkthumbs(content['thumbnail']['thumbnails']),
                'subscribers': content['subscriberCountText']['simpleText'], # "2.47K subscribers"
            }})
        elif key == 'shelfRenderer':
            results.extend([item for item in
                parse_result_items(content['content']['verticalListRenderer']['items'])
            ])
        elif key == 'movieRenderer': # movies to buy/rent
            pass
        elif key == 'horizontalCardListRenderer':
            # suggested searches: .cards[].searchRefinementCardRenderer.query.runs[].text
            pass
        else:
            import pprint
            content = {'error': f"{key} is not implemented; <pre>{pprint.pformat(item)}</pre>"}
            results.append({'type': key, 'content': content})
    return results

def parse_infocard(card):
    """
    parses a single infocard into a format that's easier to handle.
    """
    card = card['cardRenderer']
    ctype = list(card['content'].keys())[0]
    content = card['content'][ctype]
    if ctype == "pollRenderer":
        return {'type': "POLL", 'content': {
            'question': content['question']['simpleText'],
            'answers': [(a['text']['simpleText'],a['numVotes']) \
                for a in content['choices']],
        }}
    elif ctype == "videoInfoCardContentRenderer":
        is_live = content.get('badge',{}).get('liveBadgeRenderer') is not None
        return {'type': "VIDEO", 'content': {
            'video_id': content['action']['watchEndpoint']['videoId'],
            'title': content['videoTitle']['simpleText'],
            'author': delL(content['channelName']['simpleText']),
            'length': content.get('lengthString',{}).get('simpleText') \
                if not is_live else "LIVE", # "23:03"
            'views': toInt(content.get('viewCountText',{}).get('simpleText')),
                # XXX: views sometimes "Starts: July 31, 2020 at 1:30 PM"
        }}
    elif ctype == "playlistInfoCardContentRenderer":
        return {'type': "PLAYLIST", 'content': {
            'playlist_id': content['action']['watchEndpoint']['playlistId'],
            'video_id': content['action']['watchEndpoint']['videoId'],
            'title': content['playlistTitle']['simpleText'],
            'author': delL(content['channelName']['simpleText']),
            'n_videos': toInt(content['playlistVideoCount']['simpleText']),
        }}
    elif ctype == "simpleCardContentRenderer" and \
            'urlEndpoint' in content['command']:
        return {'type': "WEBSITE", 'content': {
            'url': clean_url(content['command']['urlEndpoint']['url']),
            'domain': content['displayDomain']['simpleText'],
            'title': content['title']['simpleText'],
            # XXX: no thumbnails for infocards
        }}
    elif ctype == "collaboratorInfoCardContentRenderer":
        return {'type': "CHANNEL", 'content': {
            'channel_id': content['endpoint']['browseEndpoint']['browseId'],
            'title': content['channelName']['simpleText'],
            'icons': mkthumbs(content['channelAvatar']['thumbnails']),
            'subscribers': content.get('subscriberCountText',{}).get('simpleText',''), # "545K subscribers"
        }}
    else:
        import pprint
        content = {'error': f"{ctype} is not implemented; <pre>{pprint.pformat(card)}</pre>"}
        # TODO!!!
        return None

def parse_endcard(card):
    """
    parses a single endcard into a format that's easier to handle.
    """
    card = card.get('endscreenElementRenderer', card) #only sometimes nested
    ctype = card['style']
    if ctype == "CHANNEL":
        return {'type': ctype, 'content': {
            'channel_id': card['endpoint']['browseEndpoint']['browseId'],
            'title': card['title']['simpleText'],
            'icons': mkthumbs(card['image']['thumbnails']),
        }}
    elif ctype == "VIDEO":
        return {'type': ctype, 'content': {
            'video_id': card['endpoint']['watchEndpoint']['videoId'],
            'title': card['title']['simpleText'],
            'length': card['videoDuration']['simpleText'],  # '12:21'
            'views': toInt(card['metadata']['simpleText']),
            # XXX: no channel name
        }}
    elif ctype == "PLAYLIST":
        return {'type': ctype, 'content': {
            'playlist_id': card['endpoint']['watchEndpoint']['playlistId'],
            'video_id': card['endpoint']['watchEndpoint']['videoId'],
            'title': card['title']['simpleText'],
            'author': delL(card['metadata']['simpleText']),
            'n_videos': toInt(card['playlistLength']['simpleText']),
        }}
    elif ctype == "WEBSITE" or ctype == "CREATOR_MERCHANDISE":
        url = clean_url(card['endpoint']['urlEndpoint']['url'])
        return {'type': "WEBSITE", 'content': {
            'url': url,
            'domain': urlparse(url).netloc,
            'title': card['title']['simpleText'],
            'icons': mkthumbs(card['image']['thumbnails']),
        }}
    else:
        import pprint
        content = {'error': f"{ctype} is not implemented; <pre>{pprint.pformat(card)}</pre>"}
        # TODO!!!
        return None