youtube_api_client.py

# -*- coding: latin-1 -*-
import json
import requests
import urllib
import re
import logging
from youtube_client import YoutubeClient
log = logging.getLogger('flask_web_server.service')
class YoutubeApiClient(YoutubeClient):
    DEFAULT_VIDEO_ID = "GfKs8oNP9m8"
    
    def __init__(self, apiKey):
        self.key = apiKey
        
    def get_youtube_link(self, song_title, artist):
        log.info(song_title)
        # Build the query as song_title + artists (from the Spotify API)
        temp_query = song_title + " " + artist

        log.info("Searching Youtube for: " + temp_query)

        params_for_query = {"part": "snippet",
                            "maxResults": 3,
                            "order": "relevance",
                            "pageToken": "",
                            "q": temp_query,
                            "key": self.key,
                            "type": "video",
                            }
        url1 = "https://www.googleapis.com/youtube/v3/search"

        # Get top 5 video results ordered by relevance
        # TODO: Still some matching problems - how to get the most relevant video with the most views from official sources?
        page_query = requests.request(method="get", url=url1, params=params_for_query)
        j_results_query = json.loads(page_query.text)

        if(page_query.status_code != 200):
            log.error("Youtube API Error")
            log.error(j_results_query)
            return self.DEFAULT_VIDEO_ID
        
        max_viewcount = 0
        max_viewcount_title = ""
        max_viewcount_id = ""
        better_pick = ""

        if 'items' in j_results_query:
            # Iterate through the first 5 results to find the relevant video with most views
            for item in j_results_query['items']:

                # Video title of a video to match the query string
                video_title = item['snippet']['title']

                # Get the video ID of a video to match the query string
                video_id = item['id']['videoId']

                # Next, using this video ID we need to get the view count of that video from the API
                params_for_stats = {"part": "statistics",
                                    "id": video_id,
                                    "key": self.key,
                                }
                url2 = "https://www.googleapis.com/youtube/v3/videos"
                page_stats = requests.request(method="get", url=url2, params=params_for_stats)
                j_results_stats = json.loads(page_stats.text)

                # Check if problem with retrieving statistics
                if 'items' in j_results_stats:
                    # Get the view count
                    view_count  = j_results_stats['items'][0]['statistics']['viewCount']

                    if int(view_count) > max_viewcount:
                        max_viewcount = int(view_count)
                        max_viewcount_title = video_title
                        max_viewcount_id = video_id
                        if song_title in video_title.lower():
                            better_pick = video_id

            log.info("VIDEO CHOICE IS: " + max_viewcount_title + " ---- VIEWCOUNT: " + str(max_viewcount) + " ---- ID: " + max_viewcount_id)
            if better_pick == "" or better_pick == max_viewcount_id:
                return max_viewcount_id
            else:
                log.info("Actually, " + better_pick + " seems to be a better pick because the title of the song is in the title of the video.")
                return better_pick
        else:
            log.warn("Playing default video - couldn't find match")
            return self.DEFAULT_VIDEO_ID