summarizer.py

# Import necessary libraries
from transformers import pipeline

class ContentSummarizer:
    def __init__(self, data):
        self.data = data
        self.summarizer = pipeline('summarization')

    def summarize(self):
        summaries = []
        for index, row in self.data.iterrows():
            summary = self.summarizer(row['content'], max_length=150, min_length=30, do_sample=False)
            summaries.append({'url': row['url'], 'summary': summary[0]['summary_text']})
        return summaries