def setup_function():
Account.clear_cache()
Comment.clear_cache()
Location.clear_cache()
Media.clear_cache()
Story.clear_cache()
Tag.clear_cache()
def test_clear_cache_comment(id):
account = Account("test")
media = Media("test")
comment = Comment(id, media=media, owner=account, text="test", created_at=0)
assert Comment.cache == {id: comment}
Comment.clear_cache()
assert Comment.cache == dict()
assert Media.cache == {"test": media}
assert Account.cache == {"test": account}
def test_comment(login, password, shortcode):
agent = AgentAccount(login, password)
media = Media(shortcode)
comment = agent.add_comment(media, "test")
agent.delete_comment(comment)
Account.clear_cache()
Media.clear_cache()
Comment.clear_cache()
def test_clear_cache_comment():
account = Account("test")
media = Media("test")
comment = Comment(1488, media=media, owner=account, text="test",
created_at=0)
Media.clear_cache()
Comment.clear_cache()
assert(Comment._cache == dict())
assert(Media._cache == dict())
def add_comment(self, media, text, settings={}):
if not isinstance(media, Media):
raise TypeError("'media' must be Media type")
if not isinstance(text, str):
raise TypeError("'text' must be str type")
if not isinstance(settings, dict):
raise TypeError("'settings' must be dict type")
if media.id is None:
self.update(media)
response = self._action_request(
referer="https://www.instagram.com/p/%s/" % media.code,
url="https://www.instagram.com/web/comments/%s/add/" % media.id,
data={"comment_text": text},
)
try:
data = response.json()
if data["status"] == "ok":
comment = Comment(
data["id"],
media=media,
owner=self,
text=data["text"],
created_at=data["created_time"],
)
return comment
return None
except (ValueError, KeyError) as exception:
raise UnexpectedResponse(exception, response.url, response.text)
def get_post_comments(post, num=None, path=None):
agent = Agent()
comments = set()
pointer = None
if num == None:
comments_count = int(post.comments_count)
else:
comments_count = num
limit = 50
batch_num = math.ceil(comments_count / limit)
for i in range(batch_num):
if i == batch_num - 1:
count = comments_count - limit * (batch_num - 1)
batch_comments, pointer = agent.get_comments(post,
pointer=pointer,
count=count)
else:
batch_comments, pointer = agent.get_comments(post,
pointer=pointer,
count=limit)
for j, item in enumerate(batch_comments):
comments.add(Comment(item.id))
comments_info = {}
for i, item in enumerate(comments):
comment_info = copy.copy(item)
comment_info.media = str(comment_info.media)
comment_info.owner = str(comment_info.owner)
comments_info[i] = comment_info.__dict__
comments_dict = {"comments": comments_info}
comments_json = json.dumps(comments_dict, indent=2)
if path == None:
path = './data'
pathlib.Path(path + '/comments').mkdir(parents=True, exist_ok=True)
postcode = post.code
filename = path + '/comments/' + postcode + '__last_comments.json'
with open(filename, 'w', newline='', encoding='utf8') as f:
f.write(comments_json)
return comments
def get_comments(self,
media,
pointer=None,
count=35,
settings={},
limit=50):
if not isinstance(settings, dict):
raise TypeError("'settings' must be dict type")
if not isinstance(count, int):
raise TypeError("'count' must be int type")
if not isinstance(media, Media):
raise TypeError("'media' must be Media type")
if not isinstance(limit, int):
raise TypeError("'limit' must be int type")
data = self.update(media, settings)
comments = []
if pointer is None:
try:
data = data["edge_media_to_comment"]
edges = data["edges"]
page_info = data["page_info"]
for index in range(min(len(edges), count)):
node = edges[index]["node"]
c = Comment(node["id"],
media=media,
owner=Account(node["owner"]["username"]),
text=node["text"],
created_at=node["created_at"])
media.comments.add(c)
comments.append(c)
if page_info["has_next_page"]:
pointer = page_info["end_cursor"]
if len(edges) < count and not pointer is None:
count = count - len(edges)
else:
return comments, pointer
except (ValueError, KeyError) as exception:
raise UnexpectedResponse(exception, media)
variables_string = '{{"shortcode":"{code}","first":{first},"after":"{after}"}}'
while True:
data = {
"after": pointer,
"code": media.code,
"first": min(limit, count)
}
response = self._graphql_request(
query_hash="33ba35852cb50da46f5b5e889df7d159",
variables=variables_string.format(**data),
settings=settings,
)
try:
data = response.json(
)["data"]["shortcode_media"]["edge_media_to_comment"]
media.comments_count = data["count"]
edges = data["edges"]
page_info = data["page_info"]
for index in range(min(len(edges), count)):
node = edges[index]["node"]
c = Comment(node["id"],
media=media,
owner=Account(node["owner"]["username"]),
text=node["text"],
created_at=node["created_at"])
media.comments.add(c)
comments.append(c)
if page_info["has_next_page"]:
pointer = page_info["end_cursor"]
else:
pointer = None
if len(edges) < count and page_info["has_next_page"]:
count = count - len(edges)
else:
return comments, pointer
except (ValueError, KeyError) as exception:
raise UnexpectedResponse(exception, response.url,
response.text)
def get_comments(self, media, pointer=None, count=35, settings={}, limit=50):
if not isinstance(settings, dict):
raise TypeError("'settings' must be dict type")
if not isinstance(count, int):
raise TypeError("'count' must be int type")
if not isinstance(media, Media):
raise TypeError("'media' must be Media type")
if not isinstance(limit, int):
raise TypeError("'limit' must be int type")
data = self.update(media, settings)
query_hash = "33ba35852cb50da46f5b5e889df7d159"
variables_string = '{{"shortcode":"{code}","first":{first},"after":"{after}"}}'
comments = []
if pointer is None:
try:
data = data["edge_media_to_comment"]
edges = data["edges"]
page_info = data["page_info"]
for index in range(min(len(edges), count)):
node = edges[index]["node"]
c = Comment(node["id"], media=media,
owner=Account(node["owner"]["username"]),
text=node["text"],
created_at=node["created_at"])
media.comments.add(c)
comments.append(c)
if page_info["has_next_page"]:
pointer = page_info["end_cursor"]
if len(edges) < count and not pointer is None:
count = count-len(edges)
else:
return comments, pointer
except (ValueError, KeyError):
raise UnexpectedResponse(media)
if not "params" in settings:
settings["params"] = {"query_hash": query_hash}
else:
settings["params"]["query_hash"] = query_hash
while True:
data = {"after": pointer, "code": media.code, "first": min(limit, count)}
settings["params"]["variables"] = variables_string.format(**data)
if not "headers" in settings:
settings["headers"] = {
"X-Instagram-GIS": "%s:%s" % (self._rhx_gis, settings["params"]["variables"]),
}
else:
settings["headers"]["X-Instagram-GIS"] = \
"%s:%s" % (self._rhx_gis, settings["params"]["variables"])
settings["headers"]["X-Instagram-GIS"] = \
hashlib.md5(settings["headers"]["X-Instagram-GIS"].encode("utf-8")).hexdigest()
settings["headers"]["X-Requested-With"] = "XMLHttpRequest"
response = self._get_request("https://www.instagram.com/graphql/query/", **settings)
try:
data = response.json()["data"]["shortcode_media"]["edge_media_to_comment"]
media.comments_count = data["count"]
edges = data["edges"]
page_info = data["page_info"]
for index in range(min(len(edges), count)):
node = edges[index]["node"]
c = Comment(node["id"],
media=media,
owner=Account(node["owner"]["username"]),
text=node["text"],
created_at=node["created_at"])
media.comments.add(c)
comments.append(c)
if page_info["has_next_page"]:
pointer = page_info["end_cursor"]
else:
pointer = None
if len(edges) < count and page_info["has_next_page"]:
count = count-len(edges)
else:
return comments, pointer
except (ValueError, KeyError):
raise UnexpectedResponse(response.url, response.text)