def unit_get_monthly_livelist(year, month):
url = url_prefix_dict["unit"] + year + "-" + month + "/"
r = requests.get(url)
soup = BeautifulSoup(r.text, "html.parser")
livelist = soup.select(".p-schedule__item")
liveinfo_list = []
for live in livelist:
try:
artists = live.select(".p-schedule__item-lineup")[0].text
except IndexError:
artists = "リンク先をみてください"
artists_list = LiveDataCleaner().get_artists_name(artists)
live_name = live.select(".p-schedule__item-title")[0].text
live_url = live.find("a").get("href")
img_url = live.find("img").get("src")
day = live.select(".p-schedule__item-date")[0].text
day = LiveDataCleaner().get_day(day)
live_info = {
"artists": artists,
#"artists_list": artists_list,
"live_name": live_name,
"live_url": live_url,
"img_url": img_url,
"day": day,
"month": month,
"year": year,
"livehouse": "unit"
}
liveinfo_list.append(live_info)
return {"live_list": liveinfo_list}
def www_get_monthly_livelist(year, month):
url = url_prefix_dict['www'] + year + month + ".php"
r = requests.get(url)
soup = BeautifulSoup(r.text, "html.parser")
livelist = soup.find_all("article")
liveinfo_list = []
for live in livelist:
artists = live.select(".title")[0].find("span").text
artists_list = LiveDataCleaner().get_artists_name(
live.select(".title")[0].find("span").text)
live_name = live.select('.exp')[0].text
live_url = live.select('a[href^="https://"]')[0].get('href')
img_url = live.select(".visual")[0].find("img").get("src")
day = live.select('.day')[0].text
live_info = {
"artists": artists,
"live_name": live_name,
"live_url": live_url,
"img_url": img_url,
"day": day,
"month": month,
"year": year,
"livehouse": "www"
}
liveinfo_list.append(live_info)
return {"live_list": liveinfo_list}
def create_liveartist_liveid(liveid, db: Session = Depends(get_db)):
live = db.query(Live).filter(Live.pk_live_id == liveid).first()
Cleaner = LiveDataCleaner()
artists_list = Cleaner.remove_ng_artistname(
Cleaner.get_artists_name(live.artists))
liveartist_in = []
for artist in artists_list:
artist_info = spotify_api.spotify_artist_search(artist)['artist_info']
liveartist_in.append(
LiveArtist(live_id=live.pk_live_id,
artist_name=artist_info['name'],
spotify_id=artist_info['id'],
apple_id="apple",
img_url=artist_info['img_url'],
correction=artist_info['correction']))
db.add_all(liveartist_in)
db.commit()
return artist_info
def create_liveartist(livehouse, year, month, db: Session = Depends(get_db)):
lives = db.query(Live).filter(Live.livehouse == livehouse).filter(
Live.year == int(year)).filter(Live.month == int(month)).all()
Cleaner = LiveDataCleaner()
for live in lives:
artists_list = Cleaner.remove_ng_artistname(
Cleaner.get_artists_name(live.artists))
liveartist_in = []
for artist in artists_list:
artist_info = spotify_api.spotify_artist_search(
artist)['artist_info']
liveartist_in.append(
LiveArtist(live_id=live.pk_live_id,
artist_name=artist_info['name'],
spotify_id=artist_info['id'],
apple_id="apple",
img_url=artist_info['img_url'],
correction=artist_info['correction']))
db.add_all(liveartist_in)
db.commit()
return artist_info
def liquidroom_get_monthly_livelist(year, month):
url = url_prefix_dict['liquidroom'] + year + "/" + month
r = requests.get(url)
soup = BeautifulSoup(r.text, "html.parser")
livelist = soup.find_all("article")
liveinfo_list = []
for live in livelist:
try:
artists = live.select(".header")[0].find("h2").text
except IndexError:
artists = "リンク先をみてください"
try:
artists_list = LiveDataCleaner().get_artists_name(l)
except NameError:
artists_list = []
try:
live_name = live.select(".subtitle")[0].text
except IndexError:
live_name = "リンク先をみてください"
live_url = live.select(".s_link")[0].get("href")
try:
img_url = live.select(".header")[0].find("img").get("src")
except AttributeError:
img_url = "none"
day = live.select(".f1B")[0].text[:2]
live_info = {
"artists": artists,
"live_name": live_name,
"live_url": live_url,
"img_url": img_url,
"day": day,
"month": month,
"year": year,
"livehouse": "liquidroom"
}
liveinfo_list.append(live_info)
return {"live_list": liveinfo_list}
from fastapi import APIRouter, HTTPException, Depends
import requests
from bs4 import BeautifulSoup
from sqlalchemy.orm import Session, sessionmaker
from starlette.requests import Request
from db import Live, LiveKey, LiveArtist, LiveArtistWrong, LiveHouse, SongRanking, engine, session
from external_api import spotify_api
from data_cleaner.live_data_cleaner import LiveDataCleaner
SessionLocal = sessionmaker(autocommmit=False, autoflushs=False, bind=engine)
Cleaner = LiveDataCleaner()
def get_db(request: Request):
return request.state.db
router = APIRouter()
@router.post("/db/create/ranking/jwave")
def cerate_ranking_jwave(db: Session = Depends(get_db)):
url = "https://www.j-wave.co.jp/original/tokiohot100/chart/main.htm"
r = requests.get(url)
soup = BeautifulSoup(r.content, "html.parser")
music_list = soup.select("table")[0].select("table")[4].select(
"tr")[0::2][1:]
ranking_in = []
for music in music_list:
def get_liveartist_live_id(live_id, db: Session = Depends(get_db)):
content_list = []
spotify_info_list = []
spotify_prefix = "https://open.spotify.com/embed/artist/"
apple_prefix = ""
liveartists = db.query(LiveArtist).filter(
LiveArtist.live_id == int(live_id)).all()
live = db.query(Live).filter(Live.pk_live_id == live_id).first()
if not len(liveartists) == 0:
for liveartist in liveartists:
if liveartist.correction == 1:
related_artists_list = spotify_api.related_get_id(
liveartist.spotify_id)
liveartist_info = {
'artist_name': liveartist.artist_name,
'playlist_url': spotify_prefix + liveartist.spotify_id,
'related_artists_list': related_artists_list,
'live_id': live_id,
'spotify_id': liveartist.spotify_id,
}
else:
liveartist_info = {
'artist_name': liveartist.artist_name,
'playlist_url': spotify_prefix + liveartist.spotify_id,
'related_artists_list': [],
'live_id': live_id,
'spotify_id': liveartist.spotify_id,
}
content_list.append(liveartist_info)
else:
"""
LiveArtist をまだ保存してないLiveの場合
応急処置的に表示する
"""
artists = live.artists
Cleaner = LiveDataCleaner()
artist_list = Cleaner.remove_ng_artistname(
Cleaner.get_artists_name(live.artists))
for artist in artist_list:
spotify_id = spotify_api.spotify_artist_search(
artist)['artist_info']['id']
if not spotify_id == 'none':
related_artists_list = spotify_api.related_get(artist)
else:
"""
この状態のものの処理を工夫していきたい
"""
related_artists_list = []
liveartist_info = {
'artist_name': artist,
'playlist_url': spotify_prefix + spotify_id,
'related_artists_list': related_artists_list,
'live_id': live_id,
'spotify_id': spotify_id,
}
content_list.append(liveartist_info)
response = {
"liveartist_list": content_list,
"live_id": live_id,
"year": live.year,
"month": live.month,
"day": live.day,
"live_name": live.live_name
}
return response