def test_request_format(mocked_post):
image_describer = ImageDescriptionRetriever(1)
formatted_request = image_describer._format_single_request("url")
assert isinstance(formatted_request, dict)
assert "image" in formatted_request
assert isinstance(formatted_request["image"], dict)
assert "source" in formatted_request["image"]
assert isinstance(formatted_request["image"]["source"], dict)
assert "imageUri" in formatted_request["image"]["source"]
assert formatted_request["image"]["source"]["imageUri"] == "url"
assert "features" in formatted_request
assert isinstance(formatted_request["features"], list)
assert isinstance(formatted_request["features"][0], dict)
assert "maxResults" in formatted_request["features"][0]
assert "type" in formatted_request["features"][0]
assert formatted_request["features"][0]["type"] == "WEB_DETECTION"
assert "imageContext" in formatted_request
assert isinstance(formatted_request["imageContext"], dict)
assert "webDetectionParams" in formatted_request["imageContext"]
assert isinstance(formatted_request["imageContext"]["webDetectionParams"],
dict)
assert "includeGeoResults" \
in formatted_request["imageContext"]["webDetectionParams"]
assert \
formatted_request["imageContext"][
"webDetectionParams"]["includeGeoResults"] \
== "true"
def test_request_ordering():
image_describer = ImageDescriptionRetriever(3)
image_describer.all_responses_list = list()
image_describer.all_responses_list.extend([(1, ["one"]), (0, ["zero"]),
(2, ["two"])])
assert image_describer._get_ordered_and_combined_request_responses() == \
["zero", "one", "two"]
def test_default_response_with_no_colors(mocked_post):
image_describer = ImageDescriptionRetriever(1)
image_response = image_describer.get_description_for_images(
["img_url_with_no_image_color_annotation"])[0]
assert image_response["image_colors"] == [(-1, -1, -1)]
def test_request_number(mocked_post):
image_describer = ImageDescriptionRetriever(1)
actual_request_number = 1
returned_request_number, _ = image_describer._make_post_request(
["https://tinyurl.com/y9t35t3z"], actual_request_number)
assert actual_request_number == returned_request_number
def test_url_batch_splitting_for_multiple_of_batch_size():
image_describer = ImageDescriptionRetriever(3)
num_batches = 5
image_describer.image_urls = [
"url" for i in range(num_batches * image_describer.BATCH_SIZE)
]
image_describer._split_into_batches()
batch_split_urls = image_describer.image_url_batches
assert len(batch_split_urls) == num_batches
for batch in batch_split_urls:
assert len(batch) == image_describer.BATCH_SIZE
def test_url_batch_splitting_for_multiple_of_batch_size_plus_one():
image_describer = ImageDescriptionRetriever(3)
num_batches = 5
image_describer.image_urls = [
"url" for i in range(num_batches * image_describer.BATCH_SIZE + 1)
]
image_describer._split_into_batches()
batch_split_urls = image_describer.image_url_batches
assert len(batch_split_urls) == num_batches + 1
for batch_index in range(num_batches + 1):
if batch_index == num_batches:
assert len(batch_split_urls[batch_index]) == 1
else:
assert len(
batch_split_urls[batch_index]) == image_describer.BATCH_SIZE
def test_web_entity_detection(mocked_post):
image_describer = ImageDescriptionRetriever(1)
reponse_for_image_url_1 = image_describer.get_description_for_images(
["https://tinyurl.com/y7how2rj"])[0]
assert "sundar pichai" in reponse_for_image_url_1["label"]
assert "sundar pichai Alphabet" in reponse_for_image_url_1["entities"]
reponse_for_image_url_2 = image_describer.get_description_for_images(
["https://tinyurl.com/y9bvoehm"])[0]
assert "larry page" in reponse_for_image_url_2["label"]
assert "larry page google" in reponse_for_image_url_2["entities"]
reponse_for_image_url_3 = image_describer.get_description_for_images(
["https://tinyurl.com/y9t35t3z"])[0]
assert "sergey brin" in reponse_for_image_url_3["label"]
assert "sergey brin google" in reponse_for_image_url_3["entities"]
def _fetch_media_embeddings(self):
image_describer = ImageDescriptionRetriever()
# fetch all image descriptions+embeddings together - reduces latency
self.media_description_embeddings \
= self.sentence_embedding_model.encode([
self.get_condensed_image_description(image_description)
for image_description in
image_describer.get_description_for_images(
[media.img_url for media in self.media_content_list]
)
])
self.media_attribute_embeddings \
= self.sentence_embedding_model.encode([
self.get_condensed_image_attributes(image) for image in
self.media_content_list
])
def test_bad_request(mocked_post):
image_describer = ImageDescriptionRetriever(1)
with pytest.raises(BadRequestError) as error:
image_describer.get_description_for_images(["bad_url"])
assert error.message \
== "The API call was unsuccessful with status code: 400"