def memes_post(user, forum):
topics = Topic.query.filter(Topic.forum_id == forum.id).all()
topic = random.choice(topics)
text_model = utils.load_model(model_fname())
rand_val = random.random()
title = topic.title
if not title.endswith('meme'):
title += "meme"
print(title)
if rand_val < 0.2:
utils.save_post(forum, user, topic, text_model)
elif rand_val >= 0.12 and rand_val <= 0.92:
if random.random() > 0.5:
gan_path_prefix = "{}/{}/memes_gan/".format(PROJECT_DIR, DATA_DIR)
meme_files = [
gan_path_prefix + fname
for fname in os.listdir(gan_path_prefix)
]
meme_file = random.choice(meme_files)
url = utils.upload_image(meme_file).link
else:
url = google_scraper.run(100, topic.title)
utils.save_image_post_markov(forum, user, topic, text_model, url)
else:
url = google_scraper.run(100, topic.title)
utils.save_image_post_caption(forum, user, topic, text_model, url)
def generate_title(subreddit):
text_model = utils.load_model(model_fname(subreddit, 'title'))
title = text_model.make_sentence(tries=100)
if title.startswith("Rex Tillerson Brushes Off Drama"):
return None
return title
def save_thread(user, forum, subreddit):
thread_name = generate_title(subreddit)
text_model = utils.load_model(model_fname(subreddit, 'post'))
post_content = utils.generate_body(text_model)
post = Post(content=post_content)
thread = Topic(title=thread_name)
thread.save(user=user, forum=forum, post=post)
def load_thread_model(thread_id):
if os.path.exists(thread_fname(thread_id)):
return utils.load_model(thread_fname(thread_id))
else:
data_objects = RawData.query.filter(
RawData.thread_id == thread_id).all()
data_str = get_data_from_threads(data_objects)
print("new one created!", thread_fname(thread_id))
return utils.create_model_from_text(data_str, thread_fname(thread_id))
def get_thread_model(forum, thread_id):
thread_corpus_size = RawData.query.filter(
RawData.thread_id == thread_id).count()
base_model_url = get_base_model_url(forum)
if thread_corpus_size > THREAD_CORPUS_MIN:
thread_model = get_combined_model_for_thread(base_model_url, thread_id)
else:
thread_model = utils.load_model(base_model_url)
return thread_model
def reddit_post(user, forum):
rand_val = random.random()
subreddits = ["politics", "The_Donald"]
subreddit = random.choice(subreddits)
if rand_val > THREAD_TO_POST_RATIO:
save_thread(user, forum, subreddit)
else:
topics = Topic.query.filter(Topic.forum_id == forum.id).all()
topic = random.choice(topics)
text_model = utils.load_model(model_fname(subreddit, 'post'))
utils.post_or_image(forum, user, topic, text_model)
def seed_topics():
f = open("{}/{}/meme_threads.txt".format(PROJECT_DIR, DATA_DIR), 'r')
memes = f.read().split('\n')
forum = Forum.query.filter(Forum.id == MEME_FORUM_ID).all()[0]
text_model = utils.load_model(model_fname())
users = User.query.all()
for meme in memes[48:]:
user = random.choice(users)
if len(meme) > 1:
title = meme
url = google_scraper.run(100, title + " meme")
post_content = "".format(url)
post_content += "\n"
post_content += "\n"
post_content += text_model.make_sentence()
rand_val = random.random()
if rand_val > 0.6:
title += " meme"
post = Post(content=post_content)
topic = Topic(title=title)
topic.save(user=user, forum=forum, post=post)
def get_combined_model_for_thread(base_model_url, thread_id):
model_a = utils.load_model(base_model_url)
model_b = load_thread_model(thread_id)
print('combining models')
return markovify.combine([model_a, model_b], [1, 2])