Better cache handling

The root API instance will handle item caching to bring it closer to the external-facing side of things. This also means that caches are instance-based now, which may be seen as either a pro or con.
Additionally, User and Submission classes will no longer serialise their API instances when pickling.
This commit is contained in:
Thomas Wade 2020-04-22 15:46:13 +09:30
parent 042f732c69
commit d4fff8d61a

View File

@ -29,8 +29,6 @@ class Submission:
A FurAffinity submission. A FurAffinity submission.
""" """
_submissioncache = {}
def __init__(self, api, id: int): def __init__(self, api, id: int):
self._api = api self._api = api
self.id: int = id self.id: int = id
@ -173,13 +171,6 @@ class Submission:
return self._recommended return self._recommended
@staticmethod
def new_or_cached(api, id: int):
if id not in Submission._submissioncache:
Submission._submissioncache[id] = Submission(api, id)
return Submission._submissioncache[id]
def fave(self): def fave(self):
if self.faved is False: # Usage of the faved property is deliberate to ensure _get_submission() has run if self.faved is False: # Usage of the faved property is deliberate to ensure _get_submission() has run
self._toggle_fave() self._toggle_fave()
@ -193,7 +184,7 @@ class Submission:
# Parse all the easy stuff # Parse all the easy stuff
self._title = soup.select('.submission-title')[0].text.strip() self._title = soup.select('.submission-title')[0].text.strip()
self._author = User.new_or_cached(self._api, soup.select('.submission-id-sub-container a')[0].text.strip()) self._author = self._api.get_user(soup.select('.submission-id-sub-container a')[0].text.strip())
self._description = soup.select('.submission-description')[0].text.strip() self._description = soup.select('.submission-description')[0].text.strip()
self._tags = [tag.text.strip() for tag in soup.select('.submission-sidebar .tags')] self._tags = [tag.text.strip() for tag in soup.select('.submission-sidebar .tags')]
self._timestamp = datetime.strptime(soup.select('.popup_date')[0].get('title'), '%b %d, %Y %H:%M %p') # e.g. Nov 26, 2019 02:47 PM self._timestamp = datetime.strptime(soup.select('.popup_date')[0].get('title'), '%b %d, %Y %H:%M %p') # e.g. Nov 26, 2019 02:47 PM
@ -243,7 +234,7 @@ class Submission:
comment.hidden_by_page_owner = False comment.hidden_by_page_owner = False
# Parse content from non-hidden comments # Parse content from non-hidden comments
comment.author = User.new_or_cached(self._api, comment_element.select('.comment_username')[0].text.strip()) comment.author = self._api.get_user(comment_element.select('.comment_username')[0].text.strip())
comment.timestamp = datetime.fromtimestamp(int(comment_element.get('data-timestamp'))) # FIXME: Needs to account for timezone difference since these are server-local epochs. Seems to be hosted in New York's TZ comment.timestamp = datetime.fromtimestamp(int(comment_element.get('data-timestamp'))) # FIXME: Needs to account for timezone difference since these are server-local epochs. Seems to be hosted in New York's TZ
self._comments.append(comment) self._comments.append(comment)
@ -252,7 +243,7 @@ class Submission:
self._recommended = [] self._recommended = []
for recommended in soup.select('.preview-gallery-container a'): for recommended in soup.select('.preview-gallery-container a'):
id = int(recommended.get('href')[6:-1]) # e.g. /view/35992314/ id = int(recommended.get('href')[6:-1]) # e.g. /view/35992314/
self._recommended.append(Submission.new_or_cached(self._api, id)) self._recommended.append(self._api.get_submission(id))
def _toggle_fave(self): def _toggle_fave(self):
soup = self._api.get_soup(self._fave_url) soup = self._api.get_soup(self._fave_url)
@ -260,6 +251,18 @@ class Submission:
self._faved = '-' in soup.select('.fav')[0].text self._faved = '-' in soup.select('.fav')[0].text
self._fave_url = FA_BASE_URL + soup.select('.fav a')[0].get('href') self._fave_url = FA_BASE_URL + soup.select('.fav a')[0].get('href')
def __getstate__(self):
# Get a copy of the object state without the API instance
state = self.__dict__.copy()
del (state['_api'])
return state
def __setstate__(self, state):
# Set up the object with the given state
self.__dict__.update(state)
self._api = None
def __str__(self): def __str__(self):
return '{} - {} ({}/view/{})'.format(self.author, self.title, FA_BASE_URL, self.url) return '{} - {} ({}/view/{})'.format(self.author, self.title, FA_BASE_URL, self.url)
@ -269,8 +272,6 @@ class User:
A FurAffinity user. A FurAffinity user.
""" """
_usercache = {}
def __init__(self, api, username: str): def __init__(self, api, username: str):
self._api = api self._api = api
self.username: str = username self.username: str = username
@ -315,13 +316,6 @@ class User:
return self._watchers return self._watchers
@staticmethod
def new_or_cached(api, username: str): # -> User:
if username not in User._usercache:
User._usercache[username] = User(api, username)
return User._usercache[username]
def _get_gallery(self) -> List[Submission]: def _get_gallery(self) -> List[Submission]:
submissions = [] submissions = []
next_page_url = '{}/gallery/{}/'.format(FA_BASE_URL, self.username) next_page_url = '{}/gallery/{}/'.format(FA_BASE_URL, self.username)
@ -338,7 +332,7 @@ class User:
# Get all the submissions on this page # Get all the submissions on this page
for item in soup.select('.gallery figure'): for item in soup.select('.gallery figure'):
sub = Submission.new_or_cached(self._api, int(item.get('id')[4:])) # IDs look like 'sid-35908275', so we just skip the 'sid-' sub = self._api.get_submission(int(item.get('id')[4:])) # IDs look like 'sid-35908275', so we just skip the 'sid-'
# Try cache the preview URL early # Try cache the preview URL early
try: try:
@ -367,7 +361,7 @@ class User:
# Get all the submissions on this page # Get all the submissions on this page
for item in soup.select('.gallery figure'): for item in soup.select('.gallery figure'):
sub = Submission.new_or_cached(self._api, int(item.get('id')[4:])) # IDs look like 'sid-35908275', so we just skip the 'sid-' sub = self._api.get_submission(int(item.get('id')[4:])) # IDs look like 'sid-35908275', so we just skip the 'sid-'
# Try cache the preview URL early # Try cache the preview URL early
try: try:
@ -392,7 +386,7 @@ class User:
break break
for item in users: for item in users:
watching.append(User.new_or_cached(self._api, item.text.strip())) watching.append(self._api.get_user(item.text.strip()))
# Try get the next page # Try get the next page
try: try:
@ -416,7 +410,7 @@ class User:
break break
for item in users: for item in users:
watchers.append(User.new_or_cached(self._api, item.text.strip())) watchers.append(self._api.get_user(item.text.strip()))
# Try get the next page # Try get the next page
try: try:
@ -428,6 +422,18 @@ class User:
return watchers return watchers
def __getstate__(self):
# Get a copy of the object state without the API instance
state = self.__dict__.copy()
del(state['_api'])
return state
def __setstate__(self, state):
# Set up the object with the given state
self.__dict__.update(state)
self._api = None
def __str__(self): def __str__(self):
return self.username return self.username
@ -446,6 +452,8 @@ class FAAPI:
self._driver = None self._driver = None
self.logged_in = False self.logged_in = False
self.username = None self.username = None
self._usercache = {}
self._submissioncache = {}
def login(self, cookies=None) -> bool: def login(self, cookies=None) -> bool:
""" """
@ -520,16 +528,17 @@ class FAAPI:
:param username: Username :param username: Username
:return: User :return: User
""" """
soup = self.get_soup('{}/user/{}/'.format(FA_BASE_URL, username)) # Return a new cached user instance if one doesn't exist already
if username not in self._usercache:
user = User(self, username)
self._usercache[username] = user
return user
try: # Get the cached user, updating the API instance if necessary
# Try extract the username, stripping the ~/∞/! in front user = self._usercache[username]
username = soup.select('div.username h2')[0].text.strip()[1:] if not user._api:
except IndexError: user._api = self
# No username, give up return user
return None
return User.new_or_cached(self, username)
def get_self(self) -> User: def get_self(self) -> User:
""" """
@ -544,4 +553,14 @@ class FAAPI:
:param id: Submission ID :param id: Submission ID
:return: Submission :return: Submission
""" """
return Submission.new_or_cached(self, id) # Return a new cached submission instance if one doesn't exist already
if id not in self._submissioncache:
submission = Submission(self, id)
self._submissioncache[id] = submission
return submission
# Get the cached submission, updating the API instance if necessary
submission = self._submissioncache[id]
if not submission._api:
submission._api = self
return submission