1
0
Fork 0
mirror of https://github.com/Findus23/se-simulator.git synced 2024-09-19 15:53:45 +02:00
se-simulator/todb.py

83 lines
2.6 KiB
Python
Raw Normal View History

#!/usr/bin/env python
2018-03-25 23:01:32 +02:00
import random
2018-03-22 22:52:14 +01:00
from datetime import datetime
2018-03-30 15:07:18 +02:00
import sys
2018-03-19 22:03:32 +01:00
from slugify import slugify
2018-03-23 21:28:37 +01:00
import utils
2018-03-19 22:03:32 +01:00
from models import *
from text_generator import get_chain, generate_text
2018-03-25 23:01:32 +02:00
def get_unused_users(site, count):
return User.select().join(Question, JOIN.LEFT_OUTER).switch(User).join(Answer, JOIN.LEFT_OUTER) \
.where((User.site == site) & (Question.id.is_null()) & (Answer.id.is_null()))
2018-03-19 22:03:32 +01:00
2018-03-25 23:01:32 +02:00
def add_username(site, count=500):
2018-04-03 21:52:18 +02:00
print("usernames")
2018-03-19 22:03:32 +01:00
chain = get_chain(site.url, "Usernames")
for _ in range(count):
username = generate_text(chain, "Usernames")
User.create(username=username, site=site)
def add_title(site, count=100):
2018-04-03 21:52:18 +02:00
print("titles")
2018-03-24 17:33:10 +01:00
# TODO: Make sure that every slug is unique
2018-03-19 22:03:32 +01:00
chain = get_chain(site.url, "Titles")
for _ in range(count):
title = generate_text(chain, "Titles")
slug = slugify(title, max_length=70, word_boundary=True)
Title.create(text=title, slug=slug, site=site)
2018-03-25 23:01:32 +02:00
def add_answer(site, count=300):
2018-04-03 21:52:18 +02:00
print("answers")
2018-03-25 23:01:32 +02:00
users = get_unused_users(site, count)
chain = get_chain(site.url, "Answers")
2018-03-19 22:03:32 +01:00
for i in range(count):
2018-03-25 23:01:32 +02:00
text = generate_text(chain, "Answers")
2018-03-19 22:03:32 +01:00
user = users[i]
2018-03-22 22:52:14 +01:00
time = datetime.now()
2018-03-25 23:01:32 +02:00
Answer.create(text=text, user_id=user, site_id=site, datetime=time)
2018-03-19 22:03:32 +01:00
2018-03-25 23:01:32 +02:00
def add_question(site, count=100):
2018-04-03 21:52:18 +02:00
print("Questions")
2018-03-25 23:01:32 +02:00
users = get_unused_users(site, count)
titles = Title.select().join(Question, JOIN.LEFT_OUTER) \
.where((Title.site == site) & (Question.id.is_null())) \
.limit(count)
chain = get_chain(site.url, "Questions")
2018-03-25 20:35:03 +02:00
for i in range(count):
text = generate_text(chain, "Questions")
2018-03-25 23:01:32 +02:00
title = titles[i]
2018-03-25 20:35:03 +02:00
user = users[i]
time = datetime.now()
question = Question.create(text=text, title_id=title, user_id=user, site_id=site, datetime=time)
2018-03-25 23:01:32 +02:00
num_answers = random.randint(1, 4)
answers = Answer.select().where((Answer.site == site) & (Answer.question.is_null())).limit(num_answers)
for answer in answers:
answer.question = question
answer.save()
2018-03-25 20:35:03 +02:00
2018-03-19 22:03:32 +01:00
if __name__ == "__main__":
2018-03-30 15:07:18 +02:00
if len(sys.argv) > 1:
sites = sys.argv[1:]
query = Site.select().where((Site.last_download.is_null(False)) & (Site.url.in_(sites)))
else:
2018-04-03 21:52:18 +02:00
query = Site.select().join(Question, JOIN.LEFT_OUTER).where(
(Question.id.is_null()) & (Site.last_download.is_null(False))).limit(10)
print([i.name for i in query])
2018-03-23 21:28:37 +01:00
for s in query:
2018-04-03 21:52:18 +02:00
print(s.name)
2018-03-23 21:28:37 +01:00
add_username(s)
add_title(s)
2018-03-25 23:01:32 +02:00
add_answer(s)
2018-03-23 21:28:37 +01:00
add_question(s)