csv import and export fixes

Adds shelved and published dates for books and their imported reviews.
Provides option to create new (custom) shelves when importing books.

fixes #3004
fixes #2846
fixes #2666
fixes #2411

bookwyrm/importers/__init__.py

@@ -1,6 +1,7 @@
 """ import classes """
 
 from .importer import Importer
+from .bookwyrm_import import BookwyrmBooksImporter
 from .calibre_import import CalibreImporter
 from .goodreads_import import GoodreadsImporter
 from .librarything_import import LibrarythingImporter

bookwyrm/importers/bookwyrm_import.py

@@ -0,0 +1,14 @@
+""" handle reading a csv from BookWyrm """
+
+from typing import Any
+from . import Importer
+
+
+class BookwyrmBooksImporter(Importer):
+    """Goodreads is the default importer, we basically just use the same structure"""
+
+    service = "BookWyrm"
+
+    def __init__(self, *args: Any, **kwargs: Any):
+        self.row_mappings_guesses.append(("shelf_name", ["shelf_name"]))
+        super().__init__(*args, **kwargs)

bookwyrm/importers/importer.py

@@ -18,14 +18,14 @@ class Importer:
     row_mappings_guesses = [
         ("id", ["id", "book id"]),
         ("title", ["title"]),
-        ("authors", ["author", "authors", "primary author"]),
+        ("authors", ["author_text", "author", "authors", "primary author"]),
         ("isbn_10", ["isbn10", "isbn", "isbn/uid"]),
         ("isbn_13", ["isbn13", "isbn", "isbns", "isbn/uid"]),
         ("shelf", ["shelf", "exclusive shelf", "read status", "bookshelf"]),
-        ("review_name", ["review name"]),
-        ("review_body", ["my review", "review"]),
+        ("review_name", ["review_name", "review name"]),
+        ("review_body", ["review_content", "my review", "review"]),
         ("rating", ["my rating", "rating", "star rating"]),
-        ("date_added", ["date added", "entry date", "added"]),
+        ("date_added", ["date_added", "date added", "entry date", "added"]),
         ("date_started", ["date started", "started"]),
         ("date_finished", ["date finished", "last date read", "date read", "finished"]),
     ]
@@ -38,7 +38,12 @@ class Importer:
 
     # pylint: disable=too-many-locals
     def create_job(
-        self, user: User, csv_file: Iterable[str], include_reviews: bool, privacy: str
+        self,
+        user: User,
+        csv_file: Iterable[str],
+        include_reviews: bool,
+        create_shelves: bool,
+        privacy: str,
     ) -> ImportJob:
         """check over a csv and creates a database entry for the job"""
         csv_reader = csv.DictReader(csv_file, delimiter=self.delimiter)
@@ -55,6 +60,7 @@ class Importer:
         job = ImportJob.objects.create(
             user=user,
             include_reviews=include_reviews,
+            create_shelves=create_shelves,
             privacy=privacy,
             mappings=mappings,
             source=self.service,
@@ -114,7 +120,7 @@ class Importer:
         shelf = [
             s for (s, gs) in self.shelf_mapping_guesses.items() if shelf_name in gs
         ]
-        return shelf[0] if shelf else None
+        return shelf[0] if shelf else normalized_row.get("shelf") or None
 
     # pylint: disable=no-self-use
     def normalize_row(
@@ -149,6 +155,7 @@ class Importer:
         job = ImportJob.objects.create(
             user=user,
             include_reviews=original_job.include_reviews,
+            create_shelves=original_job.create_shelves,
             privacy=original_job.privacy,
             source=original_job.source,
             # TODO: allow users to adjust mappings

bookwyrm/migrations/0189_importjob_create_shelves.py

@@ -0,0 +1,18 @@
+# Generated by Django 3.2.23 on 2023-11-25 05:49
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("bookwyrm", "0188_theme_loads"),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name="importjob",
+            name="create_shelves",
+            field=models.BooleanField(default=True),
+        ),
+    ]

bookwyrm/models/import_job.py

@@ -4,6 +4,7 @@ import math
 import re
 import dateutil.parser
 
+from django.core.exceptions import ObjectDoesNotExist
 from django.db import models
 from django.utils import timezone
 from django.utils.translation import gettext_lazy as _
@@ -59,6 +60,7 @@ class ImportJob(models.Model):
     created_date = models.DateTimeField(default=timezone.now)
     updated_date = models.DateTimeField(default=timezone.now)
     include_reviews: bool = models.BooleanField(default=True)
+    create_shelves: bool = models.BooleanField(default=True)
     mappings = models.JSONField()
     source = models.CharField(max_length=100)
     privacy = models.CharField(max_length=255, default="public", choices=PrivacyLevels)
@@ -245,6 +247,11 @@ class ImportItem(models.Model):
         """the goodreads shelf field"""
         return self.normalized_data.get("shelf")
 
+    @property
+    def shelf_name(self):
+        """the goodreads shelf field"""
+        return self.normalized_data.get("shelf_name")
+
     @property
     def review(self):
         """a user-written review, to be imported with the book data"""
@@ -388,11 +395,36 @@ def handle_imported_book(item):
 
     # shelve the book if it hasn't been shelved already
     if item.shelf and not existing_shelf:
-        desired_shelf = Shelf.objects.get(identifier=item.shelf, user=user)
+
         shelved_date = item.date_added or timezone.now()
-        ShelfBook(
-            book=item.book, shelf=desired_shelf, user=user, shelved_date=shelved_date
-        ).save(priority=IMPORT_TRIGGERED)
+
+        try:
+
+            desired_shelf = Shelf.objects.get(identifier=item.shelf, user=user)
+            shelved_date = item.date_added or timezone.now()
+            ShelfBook(
+                book=item.book,
+                shelf=desired_shelf,
+                user=user,
+                shelved_date=shelved_date,
+            ).save(priority=IMPORT_TRIGGERED)
+
+        except ObjectDoesNotExist:
+            if job.create_shelves:
+                shelfname = getattr(item, "shelf_name", item.shelf)
+                new_shelf = Shelf.objects.create(
+                    user=user,
+                    identifier=item.shelf,
+                    name=shelfname,
+                    privacy=job.privacy,
+                )
+
+                ShelfBook(
+                    book=item.book,
+                    shelf=new_shelf,
+                    user=user,
+                    shelved_date=shelved_date,
+                ).save(priority=IMPORT_TRIGGERED)
 
     for read in item.reads:
         # check for an existing readthrough with the same dates
@@ -408,9 +440,9 @@ def handle_imported_book(item):
         read.save()
 
     if job.include_reviews and (item.rating or item.review) and not item.linked_review:
-        # we don't know the publication date of the review,
-        # but "now" is a bad guess
-        published_date_guess = item.date_read or item.date_added
+        # we don't necessarily know the publication date of the review,
+        # but "now" is a bad guess unless we have no choice
+        published_date_guess = item.date_read or item.date_added or timezone.now()
         if item.review:
             # pylint: disable=consider-using-f-string
             review_title = "Review of {!r} on {!r}".format(

bookwyrm/templates/import/import.html

@@ -70,6 +70,9 @@
                             <option value="Calibre" {% if current == 'Calibre' %}selected{% endif %}>
                                 {% trans "Calibre (CSV)" %}
                             </option>
+                            <option value="BookWyrm" {% if current == 'BookWyrm' %}selected{% endif %}>
+                                {% trans "BookWyrm (CSV)" %}
+                            </option>
                         </select>
                     </div>
 
@@ -94,9 +97,14 @@
                         <input type="checkbox" name="include_reviews" checked> {% trans "Include reviews" %}
                     </label>
                 </div>
+                <div class="field">
+                    <label class="label">
+                        <input type="checkbox" name="create_shelves" checked> {% trans "Create new shelves if they do not exist" %}
+                    </label>
+                </div>
                 <div class="field">
                     <label class="label" for="privacy_import">
-                        {% trans "Privacy setting for imported reviews:" %}
+                        {% trans "Privacy setting for imported reviews and shelves:" %}
                     </label>
                     {% include 'snippets/privacy_select.html' with no_label=True privacy_uuid="import" %}
                 </div>

bookwyrm/views/imports/import_data.py

@@ -15,6 +15,7 @@ from django.views import View
 
 from bookwyrm import forms, models
 from bookwyrm.importers import (
+    BookwyrmBooksImporter,
     CalibreImporter,
     LibrarythingImporter,
     GoodreadsImporter,
@@ -67,7 +68,7 @@ class Import(View):
         return TemplateResponse(request, "import/import.html", data)
 
     def post(self, request):
-        """ingest a goodreads csv"""
+        """ingest a book data csv"""
         site = models.SiteSettings.objects.get()
         if not site.imports_enabled:
             raise PermissionDenied()
@@ -77,11 +78,16 @@ class Import(View):
             return HttpResponseBadRequest()
 
         include_reviews = request.POST.get("include_reviews") == "on"
+        create_shelves = request.POST.get("create_shelves") == "on"
         privacy = request.POST.get("privacy")
         source = request.POST.get("source")
 
         importer = None
-        if source == "LibraryThing":
+
+        if source == "BookWyrm":
+            importer = BookwyrmBooksImporter()
+            print("BookwyrmBooksImporter")
+        elif source == "LibraryThing":
             importer = LibrarythingImporter()
         elif source == "Storygraph":
             importer = StorygraphImporter()
@@ -98,6 +104,7 @@ class Import(View):
                 request.user,
                 TextIOWrapper(request.FILES["csv_file"], encoding=importer.encoding),
                 include_reviews,
+                create_shelves,
                 privacy,
             )
         except (UnicodeDecodeError, ValueError, KeyError):

bookwyrm/views/preferences/export.py

@@ -48,7 +48,16 @@ class Export(View):
         fields = (
             ["title", "author_text"]
             + deduplication_fields
-            + ["rating", "review_name", "review_cw", "review_content"]
+            + [
+                "rating",
+                "review_published",
+                "review_name",
+                "review_cw",
+                "review_content",
+                "shelf",
+                "shelf_name",
+                "date_added",
+            ]
         )
         writer.writerow(fields)
 
@@ -72,9 +81,23 @@ class Export(View):
                 .first()
             )
             if review:
+                book.review_published = review.published_date
                 book.review_name = review.name
                 book.review_cw = review.content_warning
-                book.review_content = review.raw_content
+                book.review_content = (
+                    review.raw_content
+                )  # do imported reviews not have raw content?
+
+            shelfbook = (
+                models.ShelfBook.objects.filter(book=book, user=request.user)
+                .order_by("shelved_date")
+                .last()
+            )
+            if shelfbook:
+                book.shelf = shelfbook.shelf.identifier
+                book.shelf_name = shelfbook.shelf.name
+                book.date_added = shelfbook.shelved_date
+
             writer.writerow([getattr(book, field, "") or "" for field in fields])
 
         return HttpResponse(