def test_core_class_vs_method_settings(self):
url = 'http://lassie.it/core/class_vs_method_settings.html'
l = Lassie()
data = l.fetch(url)
self.assertEqual(len(data['images']), 1)
l.open_graph = False
data = l.fetch(url)
# open_graph is set to False so there shouldn't be any images in the list this time around
self.assertEqual(len(data['images']), 0)
def test_core_class_vs_method_settings(self):
url = "http://lassie.it/core/class_vs_method_settings.html"
l = Lassie()
data = l.fetch(url)
self.assertEqual(len(data["images"]), 1)
l.open_graph = False
data = l.fetch(url)
# open_graph is set to False so there shouldn't be any images in the list this time around
self.assertEqual(len(data["images"]), 0)
def fetch(url, **kwargs):
"""Constructs and sends a :class:`Lassie <Lassie>`
Retrieves content from the specified url, parses it, and returns
a beautifully crafted dictionary of important information about that
web page.
Priority tree is as follows:
1. Open Graph
2. Twitter Card
3. Other meta content (i.e. description, keywords)
:param url: URL to send a GET request to
:param open_graph: (optional) If ``True``, filters web page content for Open Graph meta tags. The content of these properties have top priority on return values.
:type open_graph: bool
:param twitter_card: (optional) If ``True``, filters web page content for Twitter Card meta tags
:type twitter_card: bool
:param touch_icon: (optional) If ``True``, retrieves Apple touch icons and includes them in the response ``images`` array
:type touch_icon: bool
:param favicon: (optional) If ``True``, retrieves any favicon images and includes them in the response ``images`` array
:type favicon: bool
:param all_images: (optional) If ``True``, retrieves images inside web pages body and includes them in the response ``images`` array. Default: False
:type all_images: bool
:param parser: (optional) String reference for the parser that BeautifulSoup will use
:type parser: string
"""
l = Lassie()
return l.fetch(url, **kwargs)
def fetch(url, **kwargs):
"""Constructs and sends a :class:`Lassie <Lassie>`
Retrieves content from the specified url, parses it, and returns
a beautifully crafted dictionary of important information about that
web page.
Priority tree is as follows:
1. Open Graph
2. Twitter Card
3. Other meta content (i.e. description, keywords)
:param url: URL to send a GET request to
:param open_graph: (optional) If ``True``, filters web page content for Open Graph meta tags. The content of these properties have top priority on return values.
:type open_graph: bool
:param twitter_card: (optional) If ``True``, filters web page content for Twitter Card meta tags
:type twitter_card: bool
:param touch_icon: (optional) If ``True``, retrieves Apple touch icons and includes them in the response ``images`` array
:type touch_icon: bool
:param favicon: (optional) If ``True``, retrieves any favicon images and includes them in the response ``images`` array
:type favicon: bool
:param all_images: (optional) If ``True``, retrieves images inside web pages body and includes them in the response ``images`` array. Default: False
:type all_images: bool
:param parser: (optional) String reference for the parser that BeautifulSoup will use
:type parser: string
"""
l = Lassie()
return l.fetch(url, **kwargs)
def test_core_class_setting_is_none(self):
url = 'http://lassie.it/core/class_setting_is_none.html'
# This is a really odd use-case where they'd set the class attr to None, but it might happen so oh wellz.
l = Lassie()
l.open_graph = None
data = l.fetch(url, open_graph=False)
self.assertEqual(len(data['images']), 0)
def test_core_class_setting_is_none(self):
url = "http://lassie.it/core/class_setting_is_none.html"
# This is a really odd use-case where they'd set the class attr to None, but it might happen so oh wellz.
l = Lassie()
l.open_graph = None
data = l.fetch(url, open_graph=False)
self.assertEqual(len(data["images"]), 0)
def test_bad_image_dimensions(self):
url = 'http://lassie.it/core/bad_image_dimensions.html'
l = Lassie()
data = l.fetch(url, all_images=True)
# lassie.utils.convert_to_int will except a TypeError or ValueError and pass (not setting a width/height on the image)
image = data['images'][0]
self.assertTrue(not 'width' in image)
self.assertTrue(not 'height' in image)
def test_bad_image_dimensions(self):
url = "http://lassie.it/core/bad_image_dimensions.html"
l = Lassie()
data = l.fetch(url, all_images=True)
# lassie.utils.convert_to_int will except a TypeError or ValueError and pass (not setting a width/height on the image)
image = data["images"][0]
self.assertTrue(not "width" in image)
self.assertTrue(not "height" in image)
def test_core_retrieve_all_images(self):
url = 'http://lassie.it/core/retrieve_all_images.html'
l = Lassie()
l.all_images = True
data = l.fetch(url)
self.assertEqual(len(data['images']), 3)
last_image = data['images'][2]
self.assertEqual(last_image['width'], 550)
self.assertEqual(last_image['height'], 365)
def test_core_retrieve_all_images(self):
url = "http://lassie.it/core/retrieve_all_images.html"
l = Lassie()
l.all_images = True
data = l.fetch(url)
self.assertEqual(len(data["images"]), 3)
last_image = data["images"][2]
self.assertEqual(last_image["width"], 550)
self.assertEqual(last_image["height"], 365)
def test_image_dimensions(self):
url = 'http://lassie.it/core/image_dimensions.html'
l = Lassie()
data = l.fetch(url, all_images=True)
self.assertEqual(len(data['images']), 4)
image = data['images'][0]
self.assertEqual(image['width'], 100)
self.assertEqual(image['height'], 100)
image = data['images'][1]
self.assertEqual(image['width'], 100)
self.assertEqual(image['height'], 100)
image = data['images'][2]
self.assertEqual(image['width'], 100)
self.assertEqual(image['height'], 100)
image = data['images'][3]
self.assertEqual(image['width'], 100)
self.assertEqual(image['height'], 100)
def test_image_dimensions(self):
url = "http://lassie.it/core/image_dimensions.html"
l = Lassie()
data = l.fetch(url, all_images=True)
self.assertEqual(len(data["images"]), 4)
image = data["images"][0]
self.assertEqual(image["width"], 100)
self.assertEqual(image["height"], 100)
image = data["images"][1]
self.assertEqual(image["width"], 100)
self.assertEqual(image["height"], 100)
image = data["images"][2]
self.assertEqual(image["width"], 100)
self.assertEqual(image["height"], 100)
image = data["images"][3]
self.assertEqual(image["width"], 100)
self.assertEqual(image["height"], 100)