Merge pull request #2834 from zachflanders/2678

Add support for title sort to ignore initial article

bookwyrm/forms/books.py

@@ -20,6 +20,7 @@ class EditionForm(CustomForm):
         model = models.Edition
         fields = [
             "title",
+            "sort_title",
             "subtitle",
             "description",
             "series",
@@ -45,6 +46,9 @@ class EditionForm(CustomForm):
         ]
         widgets = {
             "title": forms.TextInput(attrs={"aria-describedby": "desc_title"}),
+            "sort_title": forms.TextInput(
+                attrs={"aria-describedby": "desc_sort_title"}
+            ),
             "subtitle": forms.TextInput(attrs={"aria-describedby": "desc_subtitle"}),
             "description": forms.Textarea(
                 attrs={"aria-describedby": "desc_description"}

bookwyrm/forms/lists.py

@@ -24,7 +24,7 @@ class SortListForm(forms.Form):
     sort_by = ChoiceField(
         choices=(
             ("order", _("List Order")),
-            ("title", _("Book Title")),
+            ("sort_title", _("Book Title")),
             ("rating", _("Rating")),
         ),
         label=_("Sort By"),

bookwyrm/migrations/0179_populate_sort_title.py

@@ -0,0 +1,49 @@
+import re
+from itertools import chain
+
+from django.db import migrations, transaction
+from django.db.models import Q
+
+from bookwyrm.settings import LANGUAGE_ARTICLES
+
+
+def set_sort_title(edition):
+    articles = chain(
+        *(LANGUAGE_ARTICLES.get(language, ()) for language in tuple(edition.languages))
+    )
+    edition.sort_title = re.sub(
+        f'^{" |^".join(articles)} ', "", str(edition.title).lower()
+    )
+    return edition
+
+
+@transaction.atomic
+def populate_sort_title(apps, schema_editor):
+    Edition = apps.get_model("bookwyrm", "Edition")
+    db_alias = schema_editor.connection.alias
+    editions_wo_sort_title = Edition.objects.using(db_alias).filter(
+        Q(sort_title__isnull=True) | Q(sort_title__exact="")
+    )
+    batch_size = 1000
+    start = 0
+    end = batch_size
+    while True:
+        batch = editions_wo_sort_title[start:end]
+        if not batch.exists():
+            break
+        Edition.objects.bulk_update(
+            (set_sort_title(edition) for edition in batch), ["sort_title"]
+        )
+        start = end
+        end += batch_size
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("bookwyrm", "0178_auto_20230328_2132"),
+    ]
+
+    operations = [
+        migrations.RunPython(populate_sort_title),
+    ]

bookwyrm/models/book.py

@@ -1,4 +1,5 @@
 """ database schema for books and shelves """
+from itertools import chain
 import re
 
 from django.contrib.postgres.search import SearchVectorField
@@ -17,6 +18,7 @@ from bookwyrm.preview_images import generate_edition_preview_image_task
 from bookwyrm.settings import (
     DOMAIN,
     DEFAULT_LANGUAGE,
+    LANGUAGE_ARTICLES,
     ENABLE_PREVIEW_IMAGES,
     ENABLE_THUMBNAIL_GENERATION,
 )
@@ -363,6 +365,19 @@ class Edition(Book):
             for author_id in self.authors.values_list("id", flat=True):
                 cache.delete(f"author-books-{author_id}")
 
+        # Create sort title by removing articles from title
+        if self.sort_title in [None, ""]:
+            if self.sort_title in [None, ""]:
+                articles = chain(
+                    *(
+                        LANGUAGE_ARTICLES.get(language, ())
+                        for language in tuple(self.languages)
+                    )
+                )
+                self.sort_title = re.sub(
+                    f'^{" |^".join(articles)} ', "", str(self.title).lower()
+                )
+
         return super().save(*args, **kwargs)
 
     @classmethod

bookwyrm/settings.py

@@ -312,6 +312,9 @@ LANGUAGES = [
     ("zh-hant", _("繁體中文 (Traditional Chinese)")),
 ]
 
+LANGUAGE_ARTICLES = {
+    "English": {"the", "a", "an"},
+}
 
 TIME_ZONE = "UTC"
 

bookwyrm/templates/book/edit/edit_book_form.html

@@ -28,6 +28,15 @@
                     {% include 'snippets/form_errors.html' with errors_list=form.title.errors id="desc_title" %}
                 </div>
 
+                <div class="field">
+                    <label class="label" for="id_sort_title">
+                        {% trans "Sort Title:" %}
+                    </label>
+                    <input type="text" name="sort_title" value="{{ form.sort_title.value|default:'' }}" maxlength="255" class="input" required="" id="id_sort_title" aria-describedby="desc_sort_title">
+
+                    {% include 'snippets/form_errors.html' with errors_list=form.sort_title.errors id="desc_sort_title" %}
+                </div>
+
                 <div class="field">
                     <label class="label" for="id_subtitle">
                         {% trans "Subtitle:" %}

bookwyrm/templates/shelf/shelf.html

@@ -145,7 +145,7 @@
             <thead>
                 <tr>
                     <th>{% trans "Cover"%}</th>
-                    <th>{% trans "Title"  as text %}{% include 'snippets/table-sort-header.html' with field="title" sort=sort text=text %}</th>
+                    <th>{% trans "Title"  as text %}{% include 'snippets/table-sort-header.html' with field="sort_title" sort=sort text=text %}</th>
                     <th>{% trans "Author"  as text %}{% include 'snippets/table-sort-header.html' with field="author" sort=sort text=text %}</th>
                     {% if request.user.is_authenticated %}
                         {% if is_self %}

bookwyrm/tests/models/test_book_model.py

@@ -132,3 +132,14 @@ class Book(TestCase):
         self.assertIsNotNone(book.cover_bw_book_xlarge_jpg.url)
         self.assertIsNotNone(book.cover_bw_book_xxlarge_webp.url)
         self.assertIsNotNone(book.cover_bw_book_xxlarge_jpg.url)
+
+    def test_populate_sort_title(self):
+        """The sort title should remove the initial article on save"""
+        books = (
+            models.Edition.objects.create(
+                title=f"{article} Test Edition", languages=[langauge]
+            )
+            for langauge, articles in settings.LANGUAGE_ARTICLES.items()
+            for article in articles
+        )
+        self.assertTrue(all(book.sort_title == "test edition" for book in books))

bookwyrm/views/list/list.py

@@ -129,7 +129,7 @@ def sort_list(request, items):
     """helper to handle the surprisingly involved sorting"""
     # sort_by shall be "order" unless a valid alternative is given
     sort_by = request.GET.get("sort_by", "order")
-    if sort_by not in ("order", "title", "rating"):
+    if sort_by not in ("order", "sort_title", "rating"):
         sort_by = "order"
 
     # direction shall be "ascending" unless a valid alternative is given
@@ -139,7 +139,7 @@ def sort_list(request, items):
 
     directional_sort_by = {
         "order": "order",
-        "title": "book__title",
+        "sort_title": "book__sort_title",
         "rating": "average_rating",
     }[sort_by]
     if direction == "descending":

bookwyrm/views/shelf/shelf.py

@@ -128,7 +128,7 @@ class Shelf(View):
 def sort_books(books, sort):
     """Books in shelf sorting"""
     sort_fields = [
-        "title",
+        "sort_title",
         "author",
         "shelved_date",
         "start_date",