def test_get_articles(self):
collector = CollectArticles(("brokenpromises.channels.guardian",), "2014", "1")
results = collector.run()
print
print "results:", len(results)
assert len(results) > 0
for result in results:
assert result.ref_dates, "%s : %s" % (result, result.url)
assert collector.get_report()
assert collector.get_report().collector == "brokenpromises.operations.CollectArticles", collector.get_report().collector
assert collector.get_report().meta['count'] == len(results)
assert collector.get_report().meta['related_articles'] <= len(results)
assert len(collector.get_report().meta['urls_found']) == len(results)
def test_get_articles_with_storage(self):
from brokenpromises import Article
searched_date = (2014, 1, None)
collector = CollectArticles(("brokenpromises.channels.nytimes", ),
*searched_date,
use_storage=True)
# replace storage with custom storage (testing db)
collector.storage = self.testing_storage
results = collector.run()
print
print "results:", len(results)
assert len(results) > 0
for result in results:
assert result.ref_dates, "%s : %s" % (result, result.url)
assert collector.get_report()
assert collector.get_report(
).collector == "brokenpromises.operations.CollectArticles"
assert collector.get_report().meta['count'] == len(results)
assert collector.get_report().meta['related_articles'] <= len(results)
assert len(collector.get_report().meta['urls_found']) == len(results)
assert len(
self.testing_storage.get_reports(
name="collector", searched_date=searched_date,
status="done")) == 1, self.testing_storage.get_reports(
searched_date)
results = collector.run()
assert len(results) > 0, results
assert type(results[0]) is Article, type(results[0])
assert len(
self.testing_storage.get_reports(searched_date=searched_date)) == 2
assert len(
self.testing_storage.get_reports(name="collector",
searched_date=searched_date)) == 2
assert len(
self.testing_storage.get_reports(name="collector",
searched_date=searched_date,
status="escaped")) == 1
def test_get_articles_with_storage(self):
from brokenpromises import Article
searched_date = (2014, 1, None)
collector = CollectArticles(("brokenpromises.channels.nytimes",), *searched_date, use_storage=True)
# replace storage with custom storage (testing db)
collector.storage = self.testing_storage
results = collector.run()
print
print "results:", len(results)
assert len(results) > 0
for result in results:
assert result.ref_dates, "%s : %s" % (result, result.url)
assert collector.get_report()
assert collector.get_report().collector == "brokenpromises.operations.CollectArticles"
assert collector.get_report().meta['count'] == len(results)
assert collector.get_report().meta['related_articles'] <= len(results)
assert len(collector.get_report().meta['urls_found']) == len(results)
assert len(self.testing_storage.get_reports(name="collector", searched_date=searched_date, status="done")) == 1, self.testing_storage.get_reports(searched_date)
results = collector.run()
assert len(results) > 0, results
assert type(results[0]) is Article, type(results[0])
assert len(self.testing_storage.get_reports(searched_date=searched_date)) == 2
assert len(self.testing_storage.get_reports(name="collector", searched_date=searched_date)) == 2
assert len(self.testing_storage.get_reports(name="collector", searched_date=searched_date, status="escaped")) == 1
def test_get_articles(self):
collector = CollectArticles(("brokenpromises.channels.guardian", ),
"2014", "1")
results = collector.run()
print
print "results:", len(results)
assert len(results) > 0
for result in results:
assert result.ref_dates, "%s : %s" % (result, result.url)
assert collector.get_report()
assert collector.get_report(
).collector == "brokenpromises.operations.CollectArticles", collector.get_report(
).collector
assert collector.get_report().meta['count'] == len(results)
assert collector.get_report().meta['related_articles'] <= len(results)
assert len(collector.get_report().meta['urls_found']) == len(results)
