def process_pages(self):
skipped = []
pbar = ProgressBar(widgets=['Processing pages: ', SimpleProgress()], maxval=len(self.urls)).start()
i = 0
for (num, url) in self.urls:
pbar.update(int(num))
if (num and url):
html = helpers.get_html(num, url)
if html is not None:
self.urls_with_nums[url] = num
soup = BeautifulSoup(html.encode('utf-8', 'ignore'), 'lxml')
page = Page(title=soup.title.string, num=num, html=soup.prettify(), url=url, text=soup.body.get_text())
page.index = i
self.indices_with_pages[i] = page
if page.ID not in self.pages_with_ids.keys():
self.pages_with_ids[page.ID] = page
else:
raise RuntimeError('COLLISION: %s collides with %s with hash %s.' % (page.num, self.pages_with_ids[page.ID].num, page.ID))
for link in soup.find_all('a'):
if link.get('href') and 'mailto:' != link.get('href').strip()[0:7]:
page.a.append(link)
self.pages.append(page)
i += 1
else:
skipped.append(num)
else:
skipped.append(num)
pbar.finish()
print "Skipped page(s) %s because of an error." % (', '.join(skipped))
def calc_page_ranks(self, d=0.85):
self.adj = numpy.zeros( (len(self.pages_with_ids),len(self.pages_with_ids)) )
pbar = ProgressBar(widgets=['Processing links: ', SimpleProgress()], maxval=len(self.pages_with_ids.keys())).start()
progress = 1
for (ID, page) in self.pages_with_ids.iteritems():
pbar.update(progress)
# magic PageRank
for a in page.a:
href = a.get('href')
# normalize URLS
url = page.normalize_url(href)
if url in self.S:
soup = BeautifulSoup(helpers.get_html(self.urls_with_nums[url]).encode('utf-8', 'ignore'), 'lxml')
ID = helpers.page_hash(soup.prettify())
if ID in self.pages_with_ids.keys():
#print "%s (#%d) cites %s (#%d)" % (page.num, page.index, self.pages_with_ids[ID].num, self.pages_with_ids[ID].index)
#print self.urls[int(self.pages_with_ids[ID].num)-1]
self.adj[page.index][self.pages_with_ids[ID].index] = 1.0
progress += 1
# Normalize adjacency matrix into PageRanks
pbar = ProgressBar(widgets=['Normalizing adjacencies: ', SimpleProgress()], maxval=len(self.pages_with_ids.keys())).start()
progress = 1
col_sums = numpy.sum(self.adj, axis=1)
for (ID, page) in self.pages_with_ids.iteritems():
pbar.update(progress)
for k in xrange(len(self.adj[page.index])):
if col_sums[page.index] != 0:
self.adj[page.index][k] = self.adj[page.index][k] / col_sums[page.index]
else:
self.adj[page.index][k] = 0.0
self.indices_with_pages[k]
progress += 1
pbar.finish()
numpy.savetxt("adj.txt", self.adj)
# Run PageRank and converge to principal eigenvector of adj matrix
self.ranks = numpy.ones(len(self.pages_with_ids.keys()))
z = numpy.ones(len(self.pages_with_ids.keys()))
b = 1.0 - d
pbar = ProgressBar(widgets=['Running PageRank: ', SimpleProgress()], maxval=1000).start()
for m in xrange(1000):
pbar.update(m)
u = numpy.dot(self.adj, self.ranks)
e = d*u
f = b*z
self.ranks = e+f
pbar.finish()
# Updating ranks of the pages
pbar = ProgressBar(widgets=['Updating pages with new ranks: ', SimpleProgress()], maxval=len(self.pages_with_ids.keys())).start()
progress = 1
for (ID, page) in self.pages_with_ids.iteritems():
pbar.update(progress)
page.rank = self.ranks[page.index]
progress += 1
pbar.finish()
numpy.savetxt("page_ranks.txt", self.ranks)
def extract_data(url):
"""Given news article URL, return a dict with its data organised"""
#author link?
parsed_url = urlparse(url)
html = helpers.get_html(url)
if parsed_url.netloc == 'www.canarias7.es':
try:
text = html.find(attrs={'itemprop':'articleBody'}).get_text().strip()\
if html.find(attrs={'itemprop':'articleBody'}) else ""
headline = html.find(attrs={'itemprop':'headline'}).get_text().strip()\
if html.find(attrs={'itemprop':'headline'}) else ""
subheadline = html.find(attrs={'class':'subheadline'}).get_text().strip()\
if html.find(attrs={'class':'subheadline'}) else ""
date = html.find(attrs={'class':'datefrom'}).get_text().strip()\
if html.find(attrs={'class':'datefrom'}) else ""
author = html.find(attrs={'itemprop':'author'}).get_text().strip().title()\
if html.find(attrs={'itemprop':'author'}) else 'Anónimo'
n_comments = html.find(attrs={'class':'numComments'}).get_text().strip()\
if html.find(attrs={'class':'numComments'}) else 0
categories = parsed_url.path.split('/')[1:3]
labels = [topic.find('a').get_text() if topic.find('a') else ""\
for topic in html.find_all(attrs={'class':'topic'})]
except Exception as e:
print(e, parsed_url.path[1:])
if parsed_url.netloc == 'www.laprovincia.es':
try:
text = html.find(attrs={'itemprop':'articleBody'}).get_text().strip()\
if html.find(attrs={'itemprop':'articleBody'}) else ""
headline = html.find(attrs={'itemprop':'headline'}).get_text().strip()\
if html.find(attrs={'itemprop':'headline'}) else ""
subheadline = html.find(attrs={'itemprop':'description'}).get_text().strip()\
if html.find(attrs={'itemprop':'description'}) else ""
date = html.find(attrs={'itemprop':'dateCreated'}).get_text().split('|')[0].strip()\
if html.find(attrs={'itemprop':'dateCreated'}) else ""
author = html.find(attrs={'itemprop':'author'}).get_text().strip().title()\
if html.find(attrs={'itemprop':'author'}) else 'Anónimo'
n_comments = html.find(attrs={'class':'textveces'}).get_text().strip()\
if html.find(attrs={'class':'textveces'}) else 0
categories = parsed_url.path.split('/')[1:3]
labels = [x.get_text() for x in html.find(attrs={'id':'listaTags'}).findChildren('a')[1:]]\
if html.find(attrs={'id':'listaTags'}) else []
except Exception as e:
print(e, parsed_url.path[1:])
if parsed_url.netloc == 'www.eldia.es':
try:
text = html.find(attrs={'itemprop':'articleBody'}).get_text().strip()\
if html.find(attrs={'itemprop':'articleBody'}) else ""
headline = html.find(attrs={'itemprop':'headline'}).get_text().strip()\
if html.find(attrs={'itemprop':'headline'}) else ""
subheadline = html.find(attrs={'itemprop':'description'}).get_text().strip()\
if html.find(attrs={'itemprop':'description'}) else ""
date = html.find(attrs={'itemprop':'dateCreated'}).get_text().split('|')[0].strip()\
if html.find(attrs={'itemprop':'dateCreated'}) else ""
author = html.find(attrs={'itemprop':'author'}).get_text().strip().title()\
if html.find(attrs={'itemprop':'author'}) else 'Anónimo'
n_comments = html.find(attrs={'class':'textveces'}).get_text().strip()\
if html.find(attrs={'class':'textveces'}) else 0
categories = parsed_url.path.split('/')[1:3]
labels = [x.get_text() for x in html.find(attrs={'id':'listaTags'}).findChildren('a')[1:]]\
if html.find(attrs={'id':'listaTags'}) else []
except Exception as e:
print(e, parsed_url.path[1:])
if parsed_url.netloc == 'www.noticanarias.com':
try:
text = ' '.join([x.get_text().strip() for x in html.find(attrs={'itemprop':'articleBody'}).findChildren('p')])\
if html.find(attrs={'itemprop':'articleBody'}) else ""
headline = html.find(attrs={'itemprop':'headline'}).get_text().strip()\
if html.find(attrs={'itemprop':'headline'}) else ""
subheadline = ""
date = html.find(attrs={'class':'vw-post-date updated'}).findChildren('time')[0]['datetime'].split('T')[0].strip()\
if html.find(attrs={'class':'vw-post-date updated'}) else ""
author = html.find(attrs={'itemprop':'name'}).get_text().strip().title()\
if html.find(attrs={'itemprop':'name'}) else 'Anónimo'
n_comments = ""
categories = ['','']
labels = [x.get_text() for x in html.find_all('a',attrs={'rel':'tag'})]\
if html.find_all('a',attrs={'rel':'tag'}) else []
except Exception as e:
print(e, parsed_url.path[1:])
if parsed_url.netloc == 'www.canarias24horas.com':
try:
text = html.find(attrs={'class':'itemFullText'}).get_text().strip()\
if html.find(attrs={'class':'itemFullText'}) else ""
headline = html.find(attrs={'class':'itemTitle'}).get_text().strip()\
if html.find(attrs={'class':'itemTitle'}) else ""
subheadline = html.find(attrs={'class':'itemIntroText'}).get_text().strip()\
if html.find(attrs={'class':'itemIntroText'}) else ""
date = html.find(attrs={'class':'gkDate'}).get_text().strip()\
if html.find(attrs={'class':'gkDate'}) else ""
author = html.find(attrs={'class':'itemAuthor'}).get_text().strip().title().split()[-1]\
if html.find(attrs={'class':'itemAuthor'}) else 'Anónimo'
n_comments = ""
categories = parsed_url.path.split('/')[1:3]
labels = [topic.get_text() for topic in html.find(attrs={'class':'itemTags'}).findChildren('a')]\
if html.find(attrs={'class':'itemTags'}) else []
except Exception as e:
print(e, parsed_url.path[1:])
if parsed_url.netloc == 'canariasnoticias.es':
try:
text = ' '.join([x.get_text().strip() for x in html.find(attrs={'class':'noticia-body'}).findChildren('p')])\
if html.find(attrs={'class':'noticia-body'}) else ""
headline = html.find('h1', attrs={'class':'title'}).get_text().strip()\
if html.find('h1', attrs={'class':'title'}) else ""
subheadline = html.find('h3', attrs={'class':'subtitle'}).get_text().strip()\
if html.find('h3', attrs={'class':'subtitle'}) else ""
date = html.find(attrs={'class':'date'}).get_text().strip()\
if html.find(attrs={'class':'date'}) else ""
author = html.find(attrs={'class':'author'}).get_text().strip().title()\
if html.find(attrs={'class':'author'}) else 'Anónimo'
n_comments = html.find(attrs={'class':'comment-count'}).get_text().strip()\
if html.find(attrs={'class':'comment-count'}) else 0
categories = parsed_url.path.split('/')[1:3]
labels = []
except Exception as e:
print(e, parsed_url.path[1:])
if parsed_url.netloc == 'tribunadecanarias.es':
try:
text = ' '.join([x.get_text().strip() for x in html.find(attrs={'itemprop':'articleBody'}).findChildren('p')])\
if html.find(attrs={'itemprop':'articleBody'}) else ""
headline = html.find(attrs={'itemprop':'headline'}).get_text().strip()\
if html.find(attrs={'itemprop':'headline'}) else ""
subheadline = html.find(attrs={'class':'subheadline'}).get_text().strip()\
if html.find(attrs={'class':'subheadline'}) else ""
date = html.find(attrs={'id':'t1'}).get_text().strip()\
if html.find(attrs={'id':'t1'}) else datetime.datetime.now().strftime('%Y-%m-%d')
author = html.find(attrs={'itemprop':'author'}).get_text().strip().title()\
if html.find(attrs={'itemprop':'author'}) else 'Anónimo'
n_comments = html.find(attrs={'class':'numComments'}).get_text().strip()\
if html.find(attrs={'class':'numComments'}) else 0
categories = parsed_url.path.split('/')[1:3]
labels = []
except Exception as e:
print(e, parsed_url.path[1:])
#save categories in one
# extracted data in dict form
data_dict = {
'newspaper':parsed_url.netloc,
'news_link':parsed_url.path[1:],
'headline':headline,
'subhead':subheadline,
'author':author,
'date':date,
'raw_text':text,
'n_comments':n_comments,
'main_cat':categories[1],
'sub_cat':categories[0],
'labels':labels
}
return data_dict
def test_get_html(self):
self.assertIsNotNone(helpers.get_html(self.url))
def get_links(url,n_links=5):
"""Given a news outlet main website, return its news links. Identifies where links are depending on outlet"""
# it doesnt get content that loads with JS - solution would query directly those urls
# https://gohighbrow.com/scraping-javascript-heavy-websites/ not implemented
parsed_url = urlparse(url)
html = helpers.get_html(url) #get bs4 object
links = [] #store news links
# give parameters to scrape website depending on link
if parsed_url.netloc == 'www.canarias7.es':
for link in html.find_all(['h2','h3','div'], attrs={'class':'headline'}): #include div (more news but more noise)
if link.parent.has_attr('href'):# skip None links - without the structure (normally voting polls etc)
links.append(link.parent['href'])
if parsed_url.netloc == 'www.laprovincia.es':
for link in html.find_all('a', attrs={'data-tipo':'noticia'}):
links.append(link['href'])
if parsed_url.netloc == 'www.eldia.es':
for link in html.find_all('a', attrs={'data-tipo':'noticia'}):
links.append(link['href'])
if parsed_url.netloc == 'www.noticanarias.com':
for link in html.find_all('a', attrs={'itemprop':'url'}):
links.append(link['href'])
if parsed_url.netloc == 'www.canarias24horas.com':
for link in html.find_all('h4', attrs={'class':'nspHeader'}):
links.append(link.findChildren('a')[0]['href'])
#add data scrapers#add data scrapers#add data scrapers#add data scrapers#add data scrapers
if parsed_url.netloc == 'canariasnoticias.es':
for link in html.find_all(attrs={'class':'title'}):
links.append(link.find('a')['href'])
if parsed_url.netloc == 'www.sanborondon.info':
for link in html.find_all(attrs={'class':'nspHeader'}):
links.append(link.find('a')['href'])
if parsed_url.netloc == 'tribunadecanarias.es':
for link in html.find_all(attrs={'class':'ns2-title'}):
links.append(link.find('a')['href'])
if parsed_url.netloc == 'www.canariasdiario.com':
for link in html.find_all(attrs={'itemprop':'mainEntityOfPage'}):
links.append(link['href'])
if parsed_url.netloc == 'www.europapress.es':
for link in html.find_all(attrs={'itemprop':'headline'}):
print(link)
links.append(link.find('a')['href'])
if parsed_url.netloc == 'www.efe.com':
for link in html.find_all('a', attrs={'itemprop':'url'}):
links.append(link['href'])
return links
