def test_list_newer_album_photos(self, mocked_get):
"""Should return a list of photos when given a recent photo album url"""
expected = [
{"caption": "intro || photo 1 caption", "url": u"pic1.jpg"},
{"caption": u"photo 2 caption", "url": u"pic2.jpg"},
{"caption": u"photo 3 caption", "url": u"pic3.jpg"},
]
self.assertEqual(expected, list_album_photos("bla"))
def download_album(name, path, url):
"""Downloads a photo album if necessary"""
if not os.path.exists(path):
photos = list_album_photos(url)
if not photos:
return
log.info(u'Downloading: "{0}"'.format(name))
os.makedirs(path)
i = 0
for photo in photos:
i += 1
orig_filename = photo['url'].split('/')[-1]
file_path = os.path.join(path, '{0} - {1}'.format(i, orig_filename))
try:
response = http.get(photo['url'], stream=True)
except http.DownloadError:
continue
with open(file_path, 'wb') as f:
shutil.copyfileobj(response.raw, f)
if not file_path.endswith('.gif'):
write_caption(file_path, photo['caption'])
else:
log.info(u'Photo album "{0}" already downloaded, skipping...'.format(name))
def test_list_album_photos(self, mocked_get):
"""Should return a list of photos when given a photo album url"""
expected = [
{
"caption": (
u"NASA's Cassini Spacecraft is now reaching the end of its four-year prime mission "
"(on June 30th), and about to enter into its extended mission. What a nice excuse "
"for a retrospective of some of the great images sent back home by Cassini over "
"the past four years. || The Sun is on the opposite side, so all of Saturn is "
"backlit. Courtesy NASA/JPL-Caltech"
),
"url": u"http://cache.boston.com/universal/site_graphics/blogs/bigpicture/saturn_05_30/cassini1.jpg",
},
{
"caption": u"Swirls in Saturn's cloud-tops. Courtesy NASA/JPL-Caltech",
"url": u"http://cache.boston.com/universal/site_graphics/blogs/bigpicture/saturn_05_30/cassini2.jpg",
},
{
"caption": u"The surface of Saturn's moon Dione, up close. Courtesy NASA/JPL-Caltech",
"url": u"http://cache.boston.com/universal/site_graphics/blogs/bigpicture/saturn_05_30/cassini3.jpg",
},
]
self.assertEqual(expected, list_album_photos("bla"))
def test_list_album_photos_removed_photos(self, mocked_get):
"""Should return a list of photos when the album contains removed photos"""
# e.g. http://www.boston.com/bigpicture/2010/01/earthquake_in_haiti.html
self.assertEqual(41, len(list_album_photos("bla")))
def test_list_album_photos_containing_youtube(self, mocked_get):
"""Should return a list of photos when the album contains an embedded youtube"""
# e.g. http://www.boston.com/bigpicture/2008/10/nachtweys_wish_awareness_of_xd.html
self.assertEqual(1, len(list_album_photos("bla")))
def test_list_album_photos_non_gallery(self, mocked_get):
"""Should return an empty list when scraping a non-gallery page"""
# e.g. http://www.boston.com/bigpicture/2008/10/a_quick_note_1.html
self.assertEqual([], list_album_photos("bla"))
