guest_worker/sorteerhoed/HITStore.py

218 lines
7.2 KiB
Python
Raw Normal View History

2019-10-23 10:56:28 +02:00
import logging
from sqlalchemy.ext.declarative import declarative_base
2019-10-31 14:35:24 +01:00
from sqlalchemy import Column, Integer, String, DateTime, Float
2019-10-23 10:56:28 +02:00
from sqlalchemy.orm import relationship
from sqlalchemy.sql.schema import ForeignKey, Sequence
from sqlalchemy.engine import create_engine
from sqlalchemy.orm.session import sessionmaker
import datetime
from contextlib import contextmanager
import uuid
import os
import coloredlogs
import argparse
from sqlalchemy.sql.functions import func
mainLogger = logging.getLogger("sorteerhoed")
logger = mainLogger.getChild("store")
Base = declarative_base()
"""
HIT lifetime:
created
accepted
(returned!)
working
awaiting amazon confirmation (submitted on page)
submitted
Actions:
creating Hit (creating hit with scanned image)
Scanning
2019-10-23 10:56:28 +02:00
"""
class HIT(Base):
__tablename__ = 'hits'
id = Column(Integer, Sequence('hit_id'), primary_key=True) # our sequential hit id
hit_id = Column(String(255)) # amazon's hit id
2019-10-30 12:44:25 +01:00
created_at = Column(DateTime, default=datetime.datetime.utcnow)
updated_at = Column(DateTime, default=datetime.datetime.utcnow)
scanned_at = Column(DateTime, default=None)
deleted_at = Column(DateTime, default=None)
assignments = relationship("Assignment", back_populates="hit", order_by="Assignment.created_at")
2019-10-31 14:35:24 +01:00
fee = Column(Float(precision=2), default=None)
# previous hit so we can load the corrent image
# previous_hit_id = Column(Integer, ForeignKey('hits.id'), default=None)
# previous_hit = relationship("HIT")
2019-10-23 22:33:37 +02:00
def getImagePath(self):
return os.path.join('scanimation/interfaces/frames', f"{self.id:06d}.jpg")
2019-11-02 18:09:21 +01:00
def getSvgImageUrl(self):
2019-11-02 18:14:05 +01:00
return f"scans/{self.id:06d}.svg"
def getSvgImagePath(self):
return os.path.join('www', self.getSvgImageUrl())
2019-10-23 10:56:28 +02:00
def getLastAssignment(self):
if not len(self.assignments):
return None
return self.assignments[-1]
def getAssignmentById(self, assignmentId):
for a in self.assignments:
if a.assignment_id == assignmentId:
return
2019-10-30 12:44:25 +01:00
def getStatus(self):
if self.scanned_at:
return "completed"
if self.submit_hit_at:
return "submission confirmed"
if self.submit_page_at:
return "submitted by worker"
if self.open_page_at:
2019-11-01 20:31:39 +01:00
return "working"
2019-10-30 12:44:25 +01:00
if self.accept_time:
return "accepted by worker"
2019-11-03 16:21:08 +01:00
# on abandon:
if self.worker_id:
2019-11-04 10:36:26 +01:00
return "abandoned by worker"
2019-11-03 16:21:08 +01:00
return "awaiting worker"
2019-10-30 12:44:25 +01:00
class Assignment(Base):
__tablename__ = 'assignments'
id = Column(Integer, Sequence('assignment_id'), primary_key=True) # our sequential hit id
assignment_id = Column(String(255)) # amazon's assignment id
hit_id = Column(Integer, ForeignKey('hits.id')) # our sequential hit id
hit = relationship("HIT", back_populates="assignments")
uuid = Column(String(32), default=lambda : uuid.uuid4().hex)
created_at = Column(DateTime, default=datetime.datetime.utcnow)
updated_at = Column(DateTime, default=datetime.datetime.utcnow)
assignment_id = Column(String(255), default = None)
worker_id = Column(String(255), default = None)
accept_at = Column(DateTime, default=None)
# open_page_at = Column(DateTime, default=None)
submit_page_at = Column(DateTime, default=None) # Submit the page
confirmed_at = Column(DateTime, default=None) # validate with UUID when getting Message from Amazon
abandoned_at = Column(DateTime, default=None)
rejected_at = Column(DateTime, default=None)
answer = Column(String(255), default=None)
turk_ip = Column(String(255), default=None)
turk_country = Column(String(255), default=None)
2019-10-23 10:56:28 +02:00
class Store:
def __init__(self, db_filename, logLevel=0):
path = os.path.abspath(db_filename)
if logLevel <= logging.DEBUG:
logging.getLogger('sqlalchemy.engine').setLevel(logging.INFO)
2019-10-23 22:33:37 +02:00
self.engine = create_engine('sqlite:///'+path, echo=False, connect_args={'check_same_thread': False})
2019-10-23 10:56:28 +02:00
Base.metadata.create_all(self.engine)
self.Session = sessionmaker(bind=self.engine)
2019-10-23 22:33:37 +02:00
self.session = self.Session()
2019-11-01 17:02:38 +01:00
self.currentHit = None # mirrors Centralmanagmenet, stored here so we can quickly access it from webserver classes
2019-10-23 10:56:28 +02:00
@contextmanager
def getSession(self):
"""Provide a transactional scope around a series of operations."""
try:
2019-10-23 22:33:37 +02:00
yield self.session
self.session.commit()
2019-10-23 10:56:28 +02:00
except:
2019-10-23 22:33:37 +02:00
self.session.rollback()
2019-10-23 10:56:28 +02:00
raise
2019-10-23 10:56:28 +02:00
def getHits(self, session):
2019-10-23 22:33:37 +02:00
return self.session.query(Source).order_by(HIT.created_at.desc())
2019-10-23 22:33:37 +02:00
def getHitById(self, hitId):
return self.session.query(HIT).\
filter(HIT.id==hitId).one()
2019-10-23 22:33:37 +02:00
def getHitByRemoteId(self, amazonHitId):
return self.session.query(HIT).\
filter(HIT.hit_id==amazonHitId).one()
2019-10-23 22:33:37 +02:00
def getLastSubmittedHit(self):
return self.session.query(HIT).\
filter(HIT.submit_page_at!=None).\
order_by(HIT.submit_page_at.desc()).first()
def createHIT(self) -> HIT:
2019-10-23 22:33:37 +02:00
with self.getSession() as s:
hit = HIT()
s.add(hit)
s.flush()
s.refresh(hit)
logger.info(f"Created HIT {hit.id}")
return hit
def newAssignment(self, hit: HIT) -> Assignment:
with self.getSession() as s:
assignment = Assignment()
hit.assignments.append(assignment)
s.add(assignment)
s.flush()
s.refresh(hit)
logger.info(f"Created Assignment {assignment.id}")
return assignment
2019-10-23 22:33:37 +02:00
def saveHIT(self, hit):
with self.getSession() as s:
logger.info(f"Updating hit! {hit.id}")
# s.flush()
2019-10-23 10:56:28 +02:00
def addHIT(self, hit: HIT):
with self.getSession() as s:
s.add(hit)
s.flush()
s.refresh(hit)
2019-10-23 22:33:37 +02:00
logger.info(f"Added {hit.id}")
2019-10-30 12:44:25 +01:00
def getAvgDurationOfPreviousNHits(self, n) -> int:
latest_hits = self.session.query(HIT).\
filter(HIT.submit_hit_at!=None).\
filter(HIT.accept_time!=None).\
order_by(HIT.submit_hit_at.desc()).limit(n)
durations = []
for hit in latest_hits:
durations.append((hit.submit_hit_at - hit.accept_time).total_seconds())
if not len(durations):
return int(2.5*60)
return int(sum(durations) / len(durations))
def getEstimatedHitDuration(self):
return self.getAvgDurationOfPreviousNHits(5)
def getHitTimeout(self):
2019-11-04 10:48:05 +01:00
return 160 # max(160, self.getAvgDurationOfPreviousNHits(5)*2)
2019-10-31 14:35:24 +01:00
def getHITs(self, n = 100):
return self.session.query(HIT).\
filter(HIT.submit_hit_at != None).\
order_by(HIT.submit_hit_at.desc()).limit(n)
2019-10-23 10:56:28 +02:00
# def rmSource(self, id: int):
# with self.getSession() as session:
# source = session.query(Source).get(id)
# if not source:
# logging.warning(f"Source nr {id} not found")
# else:
# logging.info(f"Deleting source {source.id}: {source.url}")
# session.delete(source)
#
2019-10-23 10:56:28 +02:00
# def getRandomNewsItem(self, session) -> NewsItem:
# return session.query(NewsItem).order_by(func.random()).limit(1).first()