def test_update_tag(name):
anon = Agent()
tag = Tag(name)
data = anon.update(tag)
Tag.clear_cache()
def setup_function():
Account.clear_cache()
Comment.clear_cache()
Location.clear_cache()
Media.clear_cache()
Story.clear_cache()
Tag.clear_cache()
def test_update_tag(login, password, name):
agent = AgentAccount(login, password)
tag = Tag(name)
data = agent.update(tag)
Tag.clear_cache()
def setup_function():
Account.clear_cache()
Media.clear_cache()
Location.clear_cache()
Tag.clear_cache()
if not anon["global_delay"] is None:
min_delay = anon["global_delay"].get("min", 0)
max_delay = anon["global_delay"].get("max", 120)
sleep(random() * (max_delay - min_delay) + min_delay)
def get_posts_by_tag_name(tagname, num=None, path=None):
agent = Agent()
agent.update(Tag(tagname))
tag = Tag(tagname)
media = set()
pointer = None
if num == None:
media_count = tag.media_count
else:
media_count = num
limit = 50
batch_num = math.ceil(media_count / limit)
for i in range(batch_num):
if i == batch_num - 1:
count = media_count - limit * (batch_num - 1)
batch_media, pointer = agent.get_media(tag,
pointer=pointer,
count=count)
else:
batch_media, pointer = agent.get_media(tag,
pointer=pointer,
count=limit)
for j, item in enumerate(batch_media):
print("Getting media: " + str(i * 50 + j + 1) + " / " +
str(media_count))
agent.update(Media(item.code))
media.add(Media(item.code))
media_posts = {}
for i, item in enumerate(media):
post_info = copy.copy(item)
post_info.likes = dict(post_info.likes)
post_info.comments = dict(post_info.comments)
post_info.location = str(post_info.location)
media_posts[i] = post_info.__dict__
media_dict = {"posts": media_posts}
media_json = json.dumps(media_dict, indent=2)
print(media_json)
if path == None:
path = './data/tag__' + tagname
pathlib.Path(path).mkdir(parents=True, exist_ok=True)
filename = path + '/tag__' + tagname + '__last_posts.json'
with open(filename, 'w', newline='', encoding='utf8') as f:
f.write(media_json)
return media
def test_get_media_tag_long(login, password, count, name):
agent = AgentAccount(login, password)
tag = Tag(name)
data, pointer = agent.get_media(tag, count=count)
assert (min(tag.media_count, count) == len(data))
assert ((pointer is None) == (tag.media_count <= count))
Tag.clear_cache()
Media.clear_cache()
def test_get_media_tag_long(count, name):
anon = Agent()
tag = Tag(name)
data, pointer = anon.get_media(tag, count=count)
assert (min(tag.media_count, count) == len(data))
assert ((pointer is None) == (tag.media_count <= count))
Tag.clear_cache()
Media.clear_cache()
def test_get_media_tag_pointer(count, name):
anon = Agent()
tag = Tag(name)
pointer = None
data = []
for i in range(count):
tmp, pointer = anon.get_media(tag, pointer=pointer)
data.extend(tmp)
assert ((pointer is None) == (tag.media_count <= count))
Tag.clear_cache()
Media.clear_cache()
def test_get_media_tag_pointer(login, password, count, name):
agent = AgentAccount(login, password)
tag = Tag(name)
pointer = None
data = []
for i in range(count):
tmp, pointer = agent.get_media(tag, pointer=pointer)
data.extend(tmp)
assert ((pointer is None) == (tag.media_count <= count))
Account.clear_cache()
Media.clear_cache()
Tag.clear_cache()
async def test_async_update_tag(async_agent, settings, name):
tag = Tag(name)
data = await async_agent.update(tag, settings=settings)
assert not data is None
assert not tag.name is None
assert not tag.media_count is None
assert tag.top_posts
async def test_async_get_media_tag(async_agent, delay, settings, count, name):
tag = Tag(name)
data, pointer = await async_agent.get_media(tag,
count=count,
delay=delay,
settings=settings)
assert min(tag.media_count, count) == len(data)
assert (pointer is None) == (tag.media_count <= count)
def test_get_media_tag_pointer(agent, delay, settings, count, name):
tag = Tag(name)
pointer = None
data = []
for _ in range(count):
tmp, pointer = agent.get_media(tag, pointer=pointer, settings=settings)
sleep(delay)
data.extend(tmp)
assert (pointer is None) == (tag.media_count == len(data))
def get_tag_info(tagname, path=None):
agent = Agent()
agent.update(Tag(tagname))
tag = Tag(tagname)
tag_info = copy.copy(tag)
tag_info.media = dict(tag_info.media)
tag_dict = {"tag": tag_info.__dict__}
tag_json = json.dumps(tag_dict, indent=2)
if path == None:
path = './data/tag__' + tagname
pathlib.Path(path).mkdir(parents=True, exist_ok=True)
filename = path + '/tag__' + tagname + '__tag_info.json'
with open(filename, 'w', newline='', encoding='utf8') as f:
f.write(tag_json)
return tag
def get_comments_by_tag(word, numberOfPost):
try:
os.mkdir("{}".format(word))
except FileExistsError:
pass
tag = Tag(word)
medias = agent.get_media(obj=tag, count = numberOfPost ) #get tuple with post identifiers and tag identifier
counter = 0
for media in medias[0]:
comments = ""
post = agent.update(obj=Media(media))#get post with comments
for t in post["edge_media_to_comment"]["edges"]:
comment = str(t["node"]["text"]) #create str from comments
comment = preprocess_text(comment)
if comment != "":
comments += (comment + "\n")
if comments != "":
counter += 1
with open("{}/{}.txt".format(word, media), "w", encoding = "utf-8", ) as f:
f.write(comments)
return counter
def test_clear_cache_tag():
tag = Tag("test")
Tag.clear_cache()
assert(Tag._cache == dict())
def test_clear_cache_tag(id):
tag = Tag(id)
assert Tag.cache == {id: tag}
Tag.clear_cache()
assert Tag.cache == dict()
