Mercurio de Valpo

mercuvalpo.py

import datetime, requests
import io, glob, os
import PyPDF2
from bs4 import BeautifulSoup

fecha = datetime.datetime.now().strftime('%Y%m%d')
print('FECHA:', fecha)
root='https://www.mercuriovalpo.cl/impresa/%s/%s/%s/' %(fecha[:4], fecha[4:6], fecha[6:])
base = root + 'papel'

bs = BeautifulSoup(requests.get(base).text, 'lxml')
links = set([link['href'] for link in bs.find_all('a') if root in link.get('href','')])

pdfWriter = PyPDF2.PdfFileWriter()
for link in sorted(links):
    print(link)
    bs2 = BeautifulSoup(requests.get(link).text, 'lxml')
    pdf = bs2.find('section', attrs={'ref':'page'})['data-page']
    pdfFileObj = io.BytesIO(requests.get(pdf).content)
    pdfReader = PyPDF2.PdfFileReader(pdfFileObj)
    pdfWriter.addPage(pdfReader.getPage(0))

pdfWriter.write(open(f'MercurioValpo_{fecha}.pdf', 'wb'))

data.quant.cl:8080/valpo

soon as a chalice lambda!