def load_train_data(categories, size=SIZE, localization=True, verbose=False):
x = []
y = []
log('Status', 'Processing... ' + str(categories))
for t in categories:
folder = TRAIN_FOLDER + t
files = os.listdir(folder)
log(t, len(files), suffix='files')
for filename in files:
img = load_image(folder + '/' + filename,
size=size,
expand_dims=False)
x.append(img)
y.append(t)
log('Status', 'DONE')
X = normalize(np.array(x))
log('X shape', X.shape)
log('X size', bytesto(X.nbytes, 'm'), suffix='MB')
Y = preprocessing.LabelEncoder().fit_transform(np.array(y))
log('Y shape', Y.shape)
log('Y size', bytesto(Y.nbytes, 'k'), suffix='KB')
return X, Y
def load_regression_data(categories, size=SIZE, verbose=False):
x = []
y = []
log('Status', 'Processing... ' + str(categories))
for t in categories:
folder = TRAIN_FOLDER + t
files = os.listdir(folder)
log(t, len(files), suffix='files')
for filename in files:
img = load_image(folder + '/' + filename,
size=size,
expand_dims=False)
a = annotations.for_image(filename, t)
if a != None:
x.append(img)
y.append([a['x'], a['y']])
log('Status', 'DONE')
X = normalize(np.array(x))
log('X shape', X.shape)
log('X size', bytesto(X.nbytes, 'm'), suffix='MB')
Y = np.array(y)
log('Y shape', Y.shape)
log('Y size', bytesto(Y.nbytes, 'k'), suffix='KB')
return X, Y
def get_db_size(human=False):
if human:
size = db_common.search(Common.data == "db_size")
size = sorted(size, key=lambda k: k['db_size_timestamp'])
return bytesto(size[-1].get("value") if size else 0, "m")
else:
size = db_common.search(Common.data == "db_size")
return size[-1].get("value") if size else 0
def set_db_size():
if os.path.isfile(MAIN_DB):
size = os.path.getsize(MAIN_DB)
db_common.insert({"data": "db_size", "value": size, "value_mb": bytesto(size, "m"), "db_size_timestamp": int(time.time()), "timestamp": int(time.time())})
hours_ago_24 = int((int(time.time()) - DAY))
db_common.remove(Common.db_size_timestamp < hours_ago_24)
else:
log.info('no database found in: {}'.format(MAIN_DB))
def load_test_data(folder):
t = []
files = os.listdir(folder)
log('Status', 'Processing... ' + str(len(files)) + ' files')
for filename in files:
path = folder + '/' + filename
img = load_image(path, expand_dims=False)
t.append(img)
log('Status', 'DONE')
T = normalize(np.array(t))
log('Shape', T.shape)
log('Size', bytesto(T.nbytes, 'm'), suffix='MB')
return T, files
def learn(subreddit=None):
log.info("trying to learn")
if os.path.isfile(MAIN_DB):
size = os.path.getsize(MAIN_DB)
log.info("db size: " + str(bytesto(size, "m")))
else:
size = 0
if size > MAIN_DB_MAX_SIZE: # learn faster early on
log.info("DB size has reached limit: {}".format(
bytesto(MAIN_DB_MAX_SIZE, "m")))
return
try:
if subreddit: # learning from a supplied
if isinstance(subreddit, str):
log.info("learning from: " + subreddit)
sub = reddit.api.subreddit(subreddit)
else:
log.info(
'I have no idea what subreddit you gave me. not going to learn.'
)
return
elif SUBREDDIT_LIST: # learn from one of the filtered subreddits
sub_name = random.choice(SUBREDDIT_LIST)
log.info("SUBREDDIT_LIST is active")
log.info("learning from: {}".format(sub_name))
sub = reddit.api.subreddit(sub_name)
else: # no subreddit supplied, so we learn from a random one
subok = False
while subok == False:
sub = reddit.api.subreddit("random")
if sub.over18 == False: # we don't want nsfw sub
if (sub.subscribers > 100000
): # easier to get away with stuff on big subs
log.info("checking if subreddit is blacklisted")
with open('src/banned_subreddits.txt'
) as blacklisted_subreddits:
list_of_blacklisted_subreddits = blacklisted_subreddits.read(
).splitlines()
if (str(sub)
) not in list_of_blacklisted_subreddits:
log.info("not blacklisted")
log.info("found: " + str(sub.display_name))
subok = True
else:
log.info("r/" + str(sub) + " is blacklisted")
sub_db = "{}/{}.db".format(DB_DIR, str(sub.display_name))
log.info("active db : {}".format(sub_db))
sub_brain = Brain(sub_db)
sub_hot = sub.hot()
log.info("looping through submissions")
# Loop through each submission
for submission in sub_hot:
log.info("checking submission")
# Replace the "MoreReplies" with all of the submission replies
submission.comments.replace_more(limit=0)
# Get a list of all the comments flattened
comments = submission.comments.list()
log.debug("looping through comments")
# Loop through each comment from the submission
for comment in comments:
if len(comment.body
) < 240: # long text tends to turn into noise.
log.debug("comment score: {}".format(comment.score))
if (comment.score > 20
): # We only want to learn things people like to hear.
if (comment.author != submission.author
): # We only want to learn comments as an ovserver
comment_body = comment.body
for dis_word in DISALLOWED_WORDS:
if dis_word in comment_body:
comment_body.replace(dis_word, "")
if LOG_LEARNED_COMMENTS:
log.info(
"learning comment. score: {}; comment: {}".
format(comment.score,
comment_body.encode("utf8")))
base_brain.learn(comment_body.encode(
"utf8")) # Tell the bot to learn this comment
sub_brain.learn(comment_body.encode(
"utf8")) # Tell the bot to learn this comment
# else:
# log.info("Did not learn form this comment as it contained a word from the dissallowed word list.")
log.info("done learning")
except Exception as e:
# If any errors occur just print it to the console
log.info(e, exc_info=True)
def download_photos(self):
total_photos_from_neighborhood = 0
total_size_downloaded_from_neighborhood = 0
page = 1
while (True):
print "\n-> PAGINA: %s" % (str(page))
total_photos_in_page = 0
total_size_downloaded_in_page = 0
url_params = {'idciudad': Constants.CITY_ID_TO_DOWNLOAD, 'idzona': Constants.NEIGHBORHOOD_ID_TO_DOWNLOAD, 'pagina': page}
request_response = requests.get(Constants.IMOVELWEB_API_QUERY_URL, params=url_params, headers=Constants.DEFAULT_IMOVELWEB_API_HEADER)
json_response = request_response.json()
ads = json_response['data']['avisos']
for ad in ads:
number_of_photos_from_ad = 0
total_size_downloaded_from_ad = 0
for attribute, value in ad.iteritems():
if attribute=='fotos':
for photo in value:
photo_url = photo.get('url').replace("580x465", Constants.PHOTO_SIZE_TO_DOWNLOAD); # getting smaller images
local_dir_saved_photo = Constants.LOCAL_DIR_SAVE_PHOTO + "/" + str(ad.get('idpropiedad'))
local_path_to_save_photo = local_dir_saved_photo + "/" + photo_url.rpartition('/')[2]
if not os.path.exists(local_dir_saved_photo):
os.mkdir(local_dir_saved_photo) #creating aviso directory to save photo
urllib.urlretrieve(photo_url, local_path_to_save_photo) #downloading photo
size = format(utils.bytesto(os.path.getsize(local_path_to_save_photo), 'm'),'.4f') #size in megabytes of each photo
number_of_photos_from_ad += 1
total_size_downloaded_from_ad = total_size_downloaded_from_ad + float(size)
print "[#AD] Photos from: %s: %s | Tamanho:%s\n" % (ad.get('idpropiedad'),str(number_of_photos_from_ad),str(total_size_downloaded_from_ad))
total_photos_in_page = total_photos_in_page + number_of_photos_from_ad
total_size_downloaded_in_page = total_size_downloaded_in_page + total_size_downloaded_from_ad
print "[#PAGE] Photos from page %s: %s | Tamanho:%s\n" % (page,str(total_photos_in_page),str(total_size_downloaded_in_page))
total_photos_from_neighborhood = total_photos_from_neighborhood + total_photos_in_page
total_size_downloaded_from_neighborhood = total_size_downloaded_from_neighborhood + total_size_downloaded_in_page
#checking if it's the end of the listings
if len(ads)<Constants.NUMBER_OF_ADS_RETURNED_API:
break
page += 1
print "[#TOTAL] Photos from all pages: %s | Tamanho:%s\n" % (str(total_photos_from_neighborhood),str(total_size_downloaded_from_neighborhood))
from logger import log
from requests import get
try:
ip = get('https://api.ipify.org').text
print 'My public IP address is:', ip
except Exception as e:
print "counld not check external ip"
limit = 52428800
log.info('------------new bot run--------------')
log.info("user is " + str(reddit.api.user.me()))
if __name__ == '__main__':
log.info('db size size to start replying:' + str(bytesto(limit, 'm')))
while True:
if os.path.isfile(MAIN_DB):
size = os.path.getsize(MAIN_DB)
log.info('db size: ' + str(bytesto(size, 'm')))
else:
size = 0
if size < limit: # learn faster early on
log.info('fast learning')
learn()
try:
log.info('new db size: ' +
str(bytesto(os.path.getsize(MAIN_DB), 'm')))
except:
def learn(subreddit=None):
log.info("trying to learn")
if os.path.isfile(MAIN_DB):
size = os.path.getsize(MAIN_DB)
log.info("db size: " + str(bytesto(size, "m")))
else:
size = 0
if size > MAIN_DB_MAX_SIZE: # learn faster early on
log.info("DB size has reached limit: {}".format(
bytesto(MAIN_DB_MAX_SIZE, "m")))
return
try:
if subreddit:
log.info("learning from: " + subreddit)
sub = reddit.api.subreddit(subreddit)
else: # no subreddit supplied, so we learn from a random one
subok = False
while subok == False:
sub = reddit.api.subreddit("random")
if sub.over18 == False: # we don't want nsfw sub
if (sub.subscribers > 100000
): # easier to get away with stuff on big subs
log.info("found: " + str(sub.display_name))
subok = True
sub_db = "{}/{}.db".format(DB_DIR, str(sub.display_name))
log.info("active db : {}".format(sub_db))
sub_brain = Brain(sub_db)
sub_hot = sub.hot()
log.info("looping through submissions")
# Loop through each submission
for submission in sub_hot:
log.info("checking submission")
# Replace the "MoreReplies" with all of the submission replies
submission.comments.replace_more(limit=0)
# Get a list of all the comments flattened
comments = submission.comments.list()
log.debug("looping through comments")
# Loop through each comment from the submission
for comment in comments:
if len(comment.body
) < 240: # long text tends to turn into noise.
log.debug("comment score: {}".format(comment.score))
if (comment.score > 20
): # We only want to learn things people like to hear.
if (comment.author != submission.author
): # We only want to learn comments as an ovserver
log.info(
"learning comment. score: {}; comment: {}".
format(comment.score,
comment.body.encode("utf8")))
base_brain.learn(comment.body.encode(
"utf8")) # Tell the bot to learn this comment
sub_brain.learn(comment.body.encode(
"utf8")) # Tell the bot to learn this comment
log.info("done learning")
except Exception as e:
# If any errors occur just print it to the console
log.info(e, exc_info=True)
def init():
log.info("db size size to start replying:" +
str(bytesto(MAIN_DB_MIN_SIZE, "m")))
reddit.shadow_check()
# check if this is the first time running the bot
set_user_info()
check_first_run()
set_db_size()
while True:
if get_db_size(
) < MAIN_DB_MIN_SIZE and not COMMENTS_DISABLED: # learn faster early on
log.info("""
THE BOT IS WORKING. IT WILL TAKE ABOUT 8 HOURS FOR IT TO LEARN AND START COMMENTING.
""")
log.info("fast learning")
learn()
try:
log.info("new db size: " + str(bytesto(get_db_size(), "m")))
except:
pass
set_db_size()
countdown(2)
if (get_db_size() > MAIN_DB_MIN_SIZE or COMMENTS_DISABLED
): # once we learn enough start submissions and replies
log.info("database size is big enough")
if USE_SLEEP_SCHEDULE:
while should_we_sleep():
log.info("zzzzzzzz :snore:")
time.sleep(60)
for action in reddit_bot:
if action.rate_limit_unlock_epoch != 0:
if action.rate_limit_unlock_epoch > get_current_epoch():
log.info(
"{} hit RateLimit recently we need to wait {} seconds with this"
.format(
action.name,
action.rate_limit_unlock_epoch -
get_current_epoch(),
))
continue
else:
action._replace(rate_limit_unlock_epoch=0)
else:
if prob(action.probability):
log.info("making a random {}".format(action.name))
try:
action.action()
except praw.exceptions.APIException as e:
secs_to_wait = get_seconds_to_wait(str(e))
action._replace(
rate_limit_unlock_epoch=(get_current_epoch() +
secs_to_wait))
log.info(
"{} hit RateLimit, need to sleep for {} seconds"
.format(action.name, secs_to_wait))
except Exception as e:
log.error("something weird happened, {}".format(e),
exc_info=True)
if prob(PROBABILITIES["LEARN"]): # chance we'll learn more
log.info("going to learn")
learn()
# Wait 10 minutes to comment and post because of reddit rate limits
countdown(1)
log.info("end main loop")
log.error("counld not check external ip")
RATE_LIMIT = 0
NEED_TO_WAIT = 0
log.info("------------new bot run--------------")
log.info("user is " + str(reddit.api.user.me()))
reddit_bot = [
reddit_bot_action("reply", reddit.random_reply, PROBABILITIES["REPLY"], 0),
reddit_bot_action("delete", reddit.delete_comments,
PROBABILITIES["DELETE"], 0),
]
if __name__ == "__main__":
log.info("db size size to start replying:" +
str(bytesto(MAIN_DB_MIN_SIZE, "m")))
while True:
if os.path.isfile(MAIN_DB):
size = os.path.getsize(MAIN_DB)
log.info("db size: " + str(bytesto(size, "m")))
else:
size = 0
if size < MAIN_DB_MIN_SIZE: # learn faster early on
log.info("fast learning")
learn()
try:
log.info("new db size: " +
str(bytesto(os.path.getsize(MAIN_DB), "m")))
except:
