Clean up & add requirements.txt
This commit is contained in:
parent
7f682e6daf
commit
c284a84b13
119
main.py.old
119
main.py.old
|
@ -1,119 +0,0 @@
|
|||
import os
|
||||
import requests
|
||||
import requests_html
|
||||
from requests_html import HTMLSession
|
||||
from bs4 import BeautifulSoup
|
||||
|
||||
class NewsScraper:
|
||||
def __init__(self, url, title_selector, time_selector, item_selector):
|
||||
self.url = url
|
||||
self.title_selector = title_selector
|
||||
self.time_selector = time_selector
|
||||
self.item_selector = item_selector
|
||||
|
||||
def scrape(self):
|
||||
response = requests.get(self.url)
|
||||
html = response.content
|
||||
soup = BeautifulSoup(html, 'html.parser')
|
||||
|
||||
news_items = []
|
||||
|
||||
for item in soup.select(self.item_selector):
|
||||
title_element = item.select_one(self.title_selector)
|
||||
time_element = item.select_one(self.time_selector)
|
||||
|
||||
if title_element and time_element:
|
||||
title = title_element.text.strip()
|
||||
time = time_element.text.strip()
|
||||
news_item_html = """
|
||||
<div class="news-item">
|
||||
<h2 class="news-title">{title}</h2>
|
||||
<p class="news-time">{time}</p>
|
||||
</div>
|
||||
""".format(title=title, time=time)
|
||||
news_items.append(news_item_html)
|
||||
|
||||
return news_items
|
||||
|
||||
def write_to_html(self, template_path, output_path, news_items, heading):
|
||||
with open(template_path, 'r') as f:
|
||||
html_template = f.read()
|
||||
|
||||
news_html = '\n'.join(news_items)
|
||||
html_output = html_template.replace('</body>', f'<h1>{heading}</h1>\n{news_html}\n</body>')
|
||||
|
||||
with open(output_path, 'a') as f:
|
||||
f.write(html_output)
|
||||
|
||||
class RequestsHTMLNewsScraper(NewsScraper):
|
||||
def __init__(self, url, title_selector, time_selector, item_selector, content_selector):
|
||||
super().__init__(url, title_selector, time_selector, item_selector)
|
||||
self.content_selector = content_selector
|
||||
|
||||
def scrape(self):
|
||||
session = HTMLSession()
|
||||
response = session.get(self.url)
|
||||
response.html.render(timeout=20) # Render the JavaScript content (adjust the timeout as needed)
|
||||
|
||||
soup = BeautifulSoup(response.html.html, 'html.parser')
|
||||
|
||||
news_items = []
|
||||
|
||||
for item in soup.select(self.item_selector):
|
||||
title_element = item.select_one(self.title_selector)
|
||||
time_element = item.select_one(self.time_selector)
|
||||
content_element = item.select_one(self.content_selector)
|
||||
|
||||
if title_element and time_element and content_element:
|
||||
title = title_element.text.strip()
|
||||
time = time_element.text.strip()
|
||||
content = content_element.decode_contents(formatter="html")
|
||||
news_item_html = """
|
||||
<div class="news-item">
|
||||
<h2 class="news-title">{title}</h2>
|
||||
<p class="news-time">{time}</p>
|
||||
<div class="news-content">{content}</div>
|
||||
</div>
|
||||
""".format(title=title, time=time, content=content)
|
||||
news_items.append(news_item_html)
|
||||
|
||||
return news_items
|
||||
|
||||
# Define the scrapers for each news source
|
||||
|
||||
almayadeen_scraper = NewsScraper(
|
||||
url="https://english.almayadeen.net/shortnews",
|
||||
title_selector="h4",
|
||||
time_selector="div.post-tag.day-time",
|
||||
item_selector="div.item"
|
||||
)
|
||||
|
||||
middleeasteye_scraper = NewsScraper(
|
||||
url="https://www.middleeasteye.net/live/israels-war-gaza-live-israel-pounds-rafah-overnight-strikes",
|
||||
title_selector=".views-field.views-field-title-1 span.field-content a",
|
||||
time_selector=".views-field.views-field-changed span.field-content",
|
||||
item_selector=".views-row"
|
||||
)
|
||||
|
||||
aljazeera_scraper = RequestsHTMLNewsScraper(
|
||||
url="https://www.aljazeera.com/news/liveblog/2024/4/20/israels-war-on-gaza-live-israelis-continue-intense-raid-on-nur-shams-camp",
|
||||
title_selector="h2",
|
||||
time_selector=".date-relative__time",
|
||||
item_selector=".card-live",
|
||||
content_selector=".wysiwyg-content"
|
||||
)
|
||||
|
||||
|
||||
# Scrape and write to HTML for each news source
|
||||
|
||||
if os.path.exists('./index.html'):
|
||||
os.remove('./index.html')
|
||||
|
||||
news_items = almayadeen_scraper.scrape()
|
||||
almayadeen_scraper.write_to_html('./template.html', './index.html', news_items, 'Al Mayadeen')
|
||||
|
||||
news_items = aljazeera_scraper.scrape()
|
||||
aljazeera_scraper.write_to_html('./template.html', './index.html', news_items, 'Al Jazeera')
|
||||
|
||||
news_items = middleeasteye_scraper.scrape()
|
||||
aljazeera_scraper.write_to_html('./template.html', './index.html', news_items, 'Middle East Eye')
|
3
requirements.txt
Normal file
3
requirements.txt
Normal file
|
@ -0,0 +1,3 @@
|
|||
requests
|
||||
requests_html
|
||||
lxml_html_clean
|
Loading…
Reference in a new issue