moviewyrm/bookwyrm/models/import_job.py

191 lines
6 KiB
Python
Raw Normal View History

2021-03-08 16:49:10 +00:00
""" track progress of goodreads imports """
import re
import dateutil.parser
from django.apps import apps
from django.db import models
from django.utils import timezone
2021-01-02 16:14:28 +00:00
from bookwyrm.connectors import connector_manager
from bookwyrm.models import ReadThrough, User, Book
from .fields import PrivacyLevels
# Mapping goodreads -> bookwyrm shelf titles.
GOODREADS_SHELVES = {
2021-03-08 16:49:10 +00:00
"read": "read",
"currently-reading": "reading",
"to-read": "to-read",
}
2021-03-08 16:49:10 +00:00
def unquote_string(text):
2021-04-26 16:15:42 +00:00
"""resolve csv quote weirdness"""
match = re.match(r'="([^"]*)"', text)
if match:
return match.group(1)
return text
def construct_search_term(title, author):
2021-04-26 16:15:42 +00:00
"""formulate a query for the data connector"""
# Strip brackets (usually series title from search term)
2021-03-08 16:49:10 +00:00
title = re.sub(r"\s*\([^)]*\)\s*", "", title)
# Open library doesn't like including author initials in search term.
2021-03-08 16:49:10 +00:00
author = re.sub(r"(\w\.)+\s*", "", author)
2021-03-08 16:49:10 +00:00
return " ".join([title, author])
2020-05-09 21:26:27 +00:00
class ImportJob(models.Model):
2021-04-26 16:15:42 +00:00
"""entry for a specific request for book data import"""
2021-03-08 16:49:10 +00:00
user = models.ForeignKey(User, on_delete=models.CASCADE)
created_date = models.DateTimeField(default=timezone.now)
task_id = models.CharField(max_length=100, null=True)
include_reviews = models.BooleanField(default=True)
complete = models.BooleanField(default=False)
privacy = models.CharField(
2021-03-08 16:49:10 +00:00
max_length=255, default="public", choices=PrivacyLevels.choices
)
2020-11-13 17:02:41 +00:00
retry = models.BooleanField(default=False)
def save(self, *args, **kwargs):
2021-04-26 16:15:42 +00:00
"""save and notify"""
super().save(*args, **kwargs)
if self.complete:
notification_model = apps.get_model(
2021-03-08 16:49:10 +00:00
"bookwyrm.Notification", require_ready=True
)
notification_model.objects.create(
user=self.user,
2021-03-08 16:49:10 +00:00
notification_type="IMPORT",
related_import=self,
)
class ImportItem(models.Model):
2021-04-26 16:15:42 +00:00
"""a single line of a csv being imported"""
2021-03-08 16:49:10 +00:00
job = models.ForeignKey(ImportJob, on_delete=models.CASCADE, related_name="items")
index = models.IntegerField()
2021-03-19 19:43:36 +00:00
data = models.JSONField()
2021-03-08 16:49:10 +00:00
book = models.ForeignKey(Book, on_delete=models.SET_NULL, null=True, blank=True)
fail_reason = models.TextField(null=True)
def resolve(self):
2021-04-26 16:15:42 +00:00
"""try various ways to lookup a book"""
if self.isbn:
self.book = self.get_book_from_isbn()
else:
# don't fall back on title/author search is isbn is present.
# you're too likely to mismatch
self.book = self.get_book_from_title_author()
def get_book_from_isbn(self):
2021-04-26 16:15:42 +00:00
"""search by isbn"""
2021-01-02 16:14:28 +00:00
search_result = connector_manager.first_search_result(
self.isbn, min_confidence=0.999
2020-10-29 22:29:23 +00:00
)
if search_result:
# raises ConnectorException
return search_result.connector.get_or_create_book(search_result.key)
return None
def get_book_from_title_author(self):
2021-04-26 16:15:42 +00:00
"""search by title and author"""
2021-03-08 16:49:10 +00:00
search_term = construct_search_term(self.title, self.author)
2021-01-02 16:14:28 +00:00
search_result = connector_manager.first_search_result(
search_term, min_confidence=0.999
2020-10-29 22:29:23 +00:00
)
if search_result:
2020-11-13 17:47:35 +00:00
# raises ConnectorException
return search_result.connector.get_or_create_book(search_result.key)
return None
2020-11-13 17:02:41 +00:00
@property
def title(self):
2021-04-26 16:15:42 +00:00
"""get the book title"""
2021-03-08 16:49:10 +00:00
return self.data["Title"]
2020-11-13 17:02:41 +00:00
@property
def author(self):
2021-04-26 16:15:42 +00:00
"""get the book title"""
2021-03-08 16:49:10 +00:00
return self.data["Author"]
2020-11-13 17:02:41 +00:00
@property
def isbn(self):
2021-04-26 16:15:42 +00:00
"""pulls out the isbn13 field from the csv line data"""
2021-03-08 16:49:10 +00:00
return unquote_string(self.data["ISBN13"])
@property
def shelf(self):
2021-04-26 16:15:42 +00:00
"""the goodreads shelf field"""
2021-03-08 16:49:10 +00:00
if self.data["Exclusive Shelf"]:
return GOODREADS_SHELVES.get(self.data["Exclusive Shelf"])
return None
@property
def review(self):
2021-04-26 16:15:42 +00:00
"""a user-written review, to be imported with the book data"""
2021-03-08 16:49:10 +00:00
return self.data["My Review"]
@property
def rating(self):
2021-04-26 16:15:42 +00:00
"""x/5 star rating for a book"""
2021-05-10 21:11:28 +00:00
if self.data.get("My Rating", None):
return int(self.data["My Rating"])
return None
@property
def date_added(self):
2021-04-26 16:15:42 +00:00
"""when the book was added to this dataset"""
2021-03-08 16:49:10 +00:00
if self.data["Date Added"]:
return timezone.make_aware(dateutil.parser.parse(self.data["Date Added"]))
return None
2021-02-20 16:02:36 +00:00
@property
def date_started(self):
2021-04-26 16:15:42 +00:00
"""when the book was started"""
2021-03-08 16:49:10 +00:00
if "Date Started" in self.data and self.data["Date Started"]:
return timezone.make_aware(dateutil.parser.parse(self.data["Date Started"]))
2021-02-20 16:02:36 +00:00
return None
@property
def date_read(self):
2021-04-26 16:15:42 +00:00
"""the date a book was completed"""
2021-03-08 16:49:10 +00:00
if self.data["Date Read"]:
return timezone.make_aware(dateutil.parser.parse(self.data["Date Read"]))
return None
@property
def reads(self):
2021-04-26 16:15:42 +00:00
"""formats a read through dataset for the book in this line"""
2021-02-20 16:02:36 +00:00
start_date = self.date_started
# Goodreads special case (no 'date started' field)
2021-03-08 16:49:10 +00:00
if (
(self.shelf == "reading" or (self.shelf == "read" and self.date_read))
and self.date_added
and not start_date
):
2021-02-20 16:02:36 +00:00
start_date = self.date_added
2021-03-08 16:49:10 +00:00
if start_date and start_date is not None and not self.date_read:
2021-02-20 16:02:36 +00:00
return [ReadThrough(start_date=start_date)]
if self.date_read:
start_date = start_date if start_date < self.date_read else None
2021-03-08 16:49:10 +00:00
return [
ReadThrough(
start_date=start_date,
finish_date=self.date_read,
)
]
return []
def __repr__(self):
2021-03-08 16:49:10 +00:00
return "<{!r}Item {!r}>".format(self.data["import_source"], self.data["Title"])
def __str__(self):
2021-03-08 16:49:10 +00:00
return "{} by {}".format(self.data["Title"], self.data["Author"])