def setUp(self):
original_mongo_uri = settings.MONGODB_URI
original_db = original_mongo_uri.split("/")[-1]
self.test_db = "test" + original_db
test_uri = "/".join(
original_mongo_uri.split("/")[0:-1]) + "/" + self.test_db
self.testing_storage = Storage(uri=test_uri)
class MrClean(Collector): """ delete articles older than 1 week """ def __init__(self): super(MrClean, self).__init__() self.storage = Storage() def run(self, **kwargs): articles = self.storage.get_articles() week_before = datetime.date.today() - datetime.timedelta(7) # 1 week delta removed_articles = [] for article in articles: for ref_date in article.ref_dates: year = ref_date['date'][0] month = ref_date['date'][1] or 12 day = ref_date['date'][2] or calendar.monthrange(year, month)[1] r_date = datetime.date(year, month, day) if r_date >= week_before: break else: removed_articles.append(article.__dict__) self.storage.remove_article(article._id) self.set_report( status = "done", count = len(removed_articles), removed_articles = removed_articles) self.storage.save_report(self.get_report())
class MrClean(Collector):
"""
delete articles older than 1 week
"""
def __init__(self):
super(MrClean, self).__init__()
self.storage = Storage()
def run(self, **kwargs):
articles = self.storage.get_articles()
week_before = datetime.date.today() - datetime.timedelta(
7) # 1 week delta
removed_articles = []
for article in articles:
for ref_date in article.ref_dates:
year = ref_date['date'][0]
month = ref_date['date'][1] or 12
day = ref_date['date'][2] or calendar.monthrange(year,
month)[1]
r_date = datetime.date(year, month, day)
if r_date >= week_before:
break
else:
removed_articles.append(article.__dict__)
self.storage.remove_article(article._id)
self.set_report(status="done",
count=len(removed_articles),
removed_articles=removed_articles)
self.storage.save_report(self.get_report())
def __init__(self,
channels,
year,
month=None,
day=None,
report_extra={},
use_storage=False,
force_collect=False):
"""
force_collect : if use_storage is enable, force the collect even if there is already a report for this searched date
"""
super(CollectArticles, self).__init__()
self.use_storage = use_storage
self.channels = [
Channel() for Channel in
brokenpromises.channels.perform_channels_import(channels)
]
self.date = (year and int(year) or None, month and int(month)
or None, day and int(day) or None)
self.force_collect = force_collect
self.storage = self.use_storage and Storage() or None
self.report_extra = report_extra
def tearDown(self):
Storage().get_connection().drop_database(self.test_db)
class TestOperations(unittest.TestCase):
'''Test Class'''
def setUp(self):
original_mongo_uri = settings.MONGODB_URI
original_db = original_mongo_uri.split("/")[-1]
self.test_db = "test" + original_db
test_uri = "/".join(
original_mongo_uri.split("/")[0:-1]) + "/" + self.test_db
self.testing_storage = Storage(uri=test_uri)
def tearDown(self):
Storage().get_connection().drop_database(self.test_db)
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_queue(self):
# need to explicitly import the runnable object
from brokenpromises.operations import CollectArticles
from brokenpromises.worker import worker
collector = CollectArticles(("brokenpromises.channels.guardian", ),
"2014",
1,
use_storage=False)
worker.run(collector)
def test_retrieve_referenced_dates(self):
dates = (
("10 October 2013", (2013, 10, 10)),
("10 october, 2013", (2013, 10, 10)),
("4 by October 2013", (2013, 10, 4)),
("10 by October 2013", (2013, 10, 10)),
("10 by October, 2013", (2013, 10, 10)),
("Jan 2014", (2014, 1, None)),
("10 in October 2013", (2013, 10, 10)),
("10 in October, 2013", (2013, 10, 10)),
("10 of October 2013", (2013, 10, 10)),
("10 of October, 2013", (2013, 10, 10)),
("10th October 2013", (2013, 10, 10)),
("10th by October 2013", (2013, 10, 10)),
("10th by October, 2013", (2013, 10, 10)),
("10th in october 2013", (2013, 10, 10)),
("10th in October, 2013", (2013, 10, 10)),
("10th of October 2013", (2013, 10, 10)),
("10th of October, 2013", (2013, 10, 10)),
("2013-10-10", (2013, 10, 10)),
("2013/10/10", (2013, 10, 10)),
("August, 2013", (2013, 8, None)),
("2013", (2013, None, None)),
("November 04, 2013", (2013, 11, 4)),
("November 4, 2013", (2013, 11, 4)),
)
text = " bla bli 123. Bu \n pouet12 \n 12412 ".join(
[_[0] for _ in dates])
refs = CollectArticles.retrieve_referenced_dates(text)
date_found = [_['extracted_date'] for _ in refs]
for searched_date in dates:
try:
ref = filter(lambda _: _["extracted_date"] == searched_date[0],
refs)[0]
except:
raise Exception(
"\"%s\" not found in document. Date found:\n%s" %
(searched_date[0], "\n".join(date_found)))
assert ref['extracted_date'] in searched_date[0]
assert ref['date'] == searched_date[1], "%s != %s" % (
ref['date'], searched_date[1])
date_found.remove(ref['extracted_date'])
assert len(refs) == len(dates), "%s != %s\nToo much : %s" % (
len(refs), len(dates), date_found)
def __init__(self):
super(MrClean, self).__init__()
self.storage = Storage()
def setUp(self):
original_mongo_uri = settings.MONGODB_URI
original_db = original_mongo_uri.split("/")[-1]
self.test_db = "test" + original_db
test_uri = "/".join(original_mongo_uri.split("/")[0:-1]) + "/" + self.test_db
self.testing_storage = Storage(uri=test_uri)
class TestOperations(unittest.TestCase):
'''Test Class'''
def setUp(self):
original_mongo_uri = settings.MONGODB_URI
original_db = original_mongo_uri.split("/")[-1]
self.test_db = "test" + original_db
test_uri = "/".join(original_mongo_uri.split("/")[0:-1]) + "/" + self.test_db
self.testing_storage = Storage(uri=test_uri)
def tearDown(self):
Storage().get_connection().drop_database(self.test_db)
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_queue(self):
# need to explicitly import the runnable object
from brokenpromises.operations import CollectArticles
from brokenpromises.worker import worker
collector = CollectArticles(("brokenpromises.channels.guardian",), "2014", 1, use_storage=False)
worker.run(collector)
def test_retrieve_referenced_dates(self):
dates = (
("10 October 2013" , (2013, 10, 10)),
("10 october, 2013" , (2013, 10, 10)),
("4 by October 2013" , (2013, 10, 4)),
("10 by October 2013" , (2013, 10, 10)),
("10 by October, 2013" , (2013, 10, 10)),
("Jan 2014" , (2014, 1, None)),
("10 in October 2013" , (2013, 10, 10)),
("10 in October, 2013" , (2013, 10, 10)),
("10 of October 2013" , (2013, 10, 10)),
("10 of October, 2013" , (2013, 10, 10)),
("10th October 2013" , (2013, 10, 10)),
("10th by October 2013" , (2013, 10, 10)),
("10th by October, 2013" , (2013, 10, 10)),
("10th in october 2013" , (2013, 10, 10)),
("10th in October, 2013" , (2013, 10, 10)),
("10th of October 2013" , (2013, 10, 10)),
("10th of October, 2013" , (2013, 10, 10)),
("2013-10-10" , (2013, 10, 10)),
("2013/10/10" , (2013, 10, 10)),
("August, 2013" , (2013, 8, None)),
("2013" , (2013, None, None)),
("November 04, 2013" , (2013, 11, 4)),
("November 4, 2013" , (2013, 11, 4)),
)
text = " bla bli 123. Bu \n pouet12 \n 12412 ".join([_[0] for _ in dates])
refs = CollectArticles.retrieve_referenced_dates(text)
date_found = [_['extracted_date'] for _ in refs]
for searched_date in dates:
try:
ref = filter(lambda _: _["extracted_date"] == searched_date[0], refs)[0]
except:
raise Exception("\"%s\" not found in document. Date found:\n%s" % (searched_date[0], "\n".join(date_found)))
assert ref['extracted_date'] in searched_date[0]
assert ref['date'] == searched_date[1], "%s != %s" % (ref['date'], searched_date[1])
date_found.remove(ref['extracted_date'])
assert len(refs) == len(dates), "%s != %s\nToo much : %s" % (len(refs), len(dates), date_found)
def __init__(self): super(MrClean, self).__init__() self.storage = Storage()
from flask import Flask, render_template, request, send_file, \
send_from_directory, Response, abort, session, redirect, url_for, make_response, json
from flask.ext.assets import Environment
from flask.ext.login import LoginManager, login_user, login_required, UserMixin, logout_user
from flask.ext.cache import Cache
from rq_dashboard import RQDashboard
from brokenpromises.storage import Storage
from brokenpromises.channels import get_available_channels
from brokenpromises.operations import CollectArticlesAndSendEmail
from brokenpromises.worker import worker
import os
import datetime
STORAGE = Storage()
class CustomFlask(Flask):
jinja_options = Flask.jinja_options.copy()
jinja_options.update(
dict(block_start_string='[%',
block_end_string='%]',
variable_start_string='[[',
variable_end_string=']]',
comment_start_string='[#',
comment_end_string='#]'))
app = CustomFlask(__name__)
app.config.from_envvar("WEBAPP_SETTINGS")