示例#1
0
def main(model, dummy):
    if len(model) == 1:
        model = model[0]
    dispatcher = Dispatcher()
    if model not in available_models:
        print('Model', model, 'is not in the folder "scrape_models".')
        print('These models are available:')
        pprint.pprint(available_models, compact=True)
        return
    imported = vars(importlib.import_module(f'scrape_models.{model}')).values()
    scrape_models = [model for model in imported if type(model) == ScrapeModel]
    dispatcher.add_scraper(scrape_models, dummy=dummy)
    dispatcher.run()
示例#2
0
from modelscraper.dispatcher import Dispatcher

import nu_nl
import metronieuws
import parool
import volkskrant

disp = Dispatcher()
disp.add_scraper([volkskrant.volkskrant])
disp.run()
示例#3
0
product = Phase(
    sources=sources,
    templates=[
        Template(
            name='product',
            db_type='mongo_db',
            db='makro',
            table='products',
            attrs=[
                Attr(name='name', selector='h1', func='sel_text'),
                Attr(name='price_gross',
                     selector='.price-gross',
                     func='sel_text'),  #  kws={'replacers': '€ '}),
                Attr(name='price_net', selector='.price-net',
                     func='sel_text'),  # kws={'replacers': '€ '}),
                Attr(name='sku', selector='.articlenumber', func='sel_text'),
                Attr(name='description', selector='.tab-1', func='sel_text'),
                Attr(name='category', selector='li.normal', func='sel_text')
            ])
    ])

makro = ScrapeModel(name='makro',
                    domain='https://www.makro.nl/',
                    num_getters=1,
                    phases=[product])

d = Dispatcher()
d.add_scraper(makro)
d.run()