트위터 데이터 크롤링 및 분석에 Scrapy를 실제로 적용

TWITTER_CONSUMER_KEY = 'your_consumer_key'
TWITTER_CONSUMER_SECRET = 'your_consumer_secret'
TWITTER_ACCESS_TOKEN = 'your_access_token'
TWITTER_ACCESS_TOKEN_SECRET = 'your_access_token_secret'

class TweetItem(scrapy.Item):
    text = scrapy.Field()
    created_at = scrapy.Field()
    user_screen_name = scrapy.Field()

class TwitterSpider(scrapy.Spider):
    name = 'twitter'
    allowed_domains = ['twitter.com']
    start_urls = ['https://twitter.com/search?f=tweets&q=keyword%20since%3A2021-01-01%20until%3A2021-12-31&src=typd']

    def parse(self, response):
        tweets = response.css('.tweet')
        for tweet in tweets:
            item = TweetItem()
            item['text'] = tweet.css('.tweet-text::text').extract_first().strip()
            item['created_at'] = tweet.css('._timestamp::text').extract_first()
            item['user_screen_name'] = tweet.css('.username b::text').extract_first().strip()
            yield item

class TwitterAnalyzer():
    def __init__(self, data=[]):
        self.data = data
        self.texts = [d['text'] for d in data]
        self.dates = [dt.strptime(d['created_at'], '%a %b %d %H:%M:%S %z %Y').date() for d in data]

    def get_top_hashtags(self, n=5):
        hashtags = Counter([re.findall(r'(?i)#w+', t) for t in self.texts])
        return hashtags.most_common(n)

    def get_top_users(self, n=5):
        users = Counter([d['user_screen_name'] for d in self.data])
        return users.most_common(n)

    def get_dates_histogram(self, step='day'):
        if step == 'day':
            return Counter(self.dates)
        elif step == 'week':
            return Counter([date.fromisoformat(str(dt).split()[0]) for dt in pd.date_range(min(self.dates), max(self.dates), freq='W')])

analyzer = TwitterAnalyzer(data)
print(analyzer.get_top_hashtags())
print(analyzer.get_top_users())
print(analyzer.get_dates_histogram('day'))