Display search results in api mode and regular

This commit is contained in:
Mouse Reeve 2021-09-16 10:44:33 -07:00
parent 1f06d1a1d8
commit 98325818b2
6 changed files with 155 additions and 41 deletions

120
bookwyrm/book_search.py Normal file
View file

@ -0,0 +1,120 @@
""" using a bookwyrm instance as a source of book data """
from functools import reduce
import operator
from django.contrib.postgres.search import SearchRank, SearchQuery
from django.db.models import OuterRef, Subquery, F, Q
from bookwyrm import models
from bookwyrm.connectors.abstract_connector import SearchResult
from bookwyrm.settings import MEDIA_FULL_URL
# pylint: disable=arguments-differ
def search(query, min_confidence=0, filters=None):
"""search your local database"""
filters = filters or []
if not query:
return []
# first, try searching unqiue identifiers
results = search_identifiers(query, *filters)
if not results:
# then try searching title/author
results = search_title_author(query, min_confidence, *filters)
return results
def isbn_search(query):
"""search your local database"""
if not query:
return []
filters = [{f: query} for f in ["isbn_10", "isbn_13"]]
results = models.Edition.objects.filter(
reduce(operator.or_, (Q(**f) for f in filters))
).distinct()
# when there are multiple editions of the same work, pick the default.
# it would be odd for this to happen.
default_editions = models.Edition.objects.filter(
parent_work=OuterRef("parent_work")
).order_by("-edition_rank")
results = (
results.annotate(default_id=Subquery(default_editions.values("id")[:1])).filter(
default_id=F("id")
)
or results
)
return results
def format_search_result(search_result):
"""convert a book object into a search result object"""
cover = None
if search_result.cover:
cover = f"{MEDIA_FULL_URL}{search_result.cover}"
return SearchResult(
title=search_result.title,
key=search_result.remote_id,
author=search_result.author_text,
year=search_result.published_date.year
if search_result.published_date
else None,
cover=cover,
confidence=search_result.rank if hasattr(search_result, "rank") else 1,
connector="",
).json()
def search_identifiers(query, *filters):
"""tries remote_id, isbn; defined as dedupe fields on the model"""
# pylint: disable=W0212
or_filters = [
{f.name: query}
for f in models.Edition._meta.get_fields()
if hasattr(f, "deduplication_field") and f.deduplication_field
]
results = models.Edition.objects.filter(
*filters, reduce(operator.or_, (Q(**f) for f in or_filters))
).distinct()
if results.count() <= 1:
return results
# when there are multiple editions of the same work, pick the default.
# it would be odd for this to happen.
default_editions = models.Edition.objects.filter(
parent_work=OuterRef("parent_work")
).order_by("-edition_rank")
return (
results.annotate(default_id=Subquery(default_editions.values("id")[:1])).filter(
default_id=F("id")
)
or results
)
def search_title_author(query, min_confidence, *filters):
"""searches for title and author"""
query = SearchQuery(query, config="simple") | SearchQuery(query, config="english")
results = (
models.Edition.objects.filter(*filters, search_vector=query)
.annotate(rank=SearchRank(F("search_vector"), query))
.filter(rank__gt=min_confidence)
.order_by("-rank")
)
# when there are multiple editions of the same work, pick the closest
editions_of_work = results.values("parent_work__id").values_list("parent_work__id")
# filter out multiple editions of the same work
for work_id in set(editions_of_work):
editions = results.filter(parent_work=work_id)
default = editions.order_by("-edition_rank").first()
default_rank = default.rank if default else 0
# if mutliple books have the top rank, pick the default edition
if default_rank == editions.first().rank:
yield default
else:
yield editions.first()

View file

@ -60,7 +60,33 @@
<ul class="is-flex-grow-1"> <ul class="is-flex-grow-1">
{% for result in result_set.results %} {% for result in result_set.results %}
<li class="mb-5"> <li class="mb-5">
{% include 'snippets/search_result_text.html' with result=result remote_result=True %} <div class="columns is-mobile is-gapless">
<div class="columns is-mobile is-gapless">
{% include 'snippets/book_cover.html' with book=result cover_class='is-w-xs is-h-xs' external_path=True %}
</div>
<div class="column is-10 ml-3">
<p>
<strong>
<a
href="{{ result.view_link|default:result.key }}"
rel="noopener"
target="_blank"
>{{ result.title }}</a>
</strong>
</p>
<p>
{{ result.author }}
{% if result.year %}({{ result.year }}){% endif %}
</p>
<form class="mt-1" action="/resolve-book" method="post">
{% csrf_token %}
<input type="hidden" name="remote_id" value="{{ result.key }}">
<button type="submit" class="button is-small is-link">
{% trans "Import book" %}
</button>
</form>
</div>
</div>
</li> </li>
{% endfor %} {% endfor %}
</ul> </ul>

View file

@ -1,37 +0,0 @@
{% load i18n %}
<div class="columns is-mobile is-gapless">
<div class="column is-cover">
{% include 'snippets/book_cover.html' with book=result cover_class='is-w-xs is-h-xs' external_path=True %}
</div>
<div class="column is-10 ml-3">
<p>
<strong>
<a
href="{{ result.view_link|default:result.key }}"
rel="noopener"
target="_blank"
>{{ result.title }}</a>
</strong>
</p>
<p>
{% if result.author %}
{{ result.author }}
{% endif %}
{% if result.year %}
({{ result.year }})
{% endif %}
</p>
<form class="mt-1" action="/resolve-book" method="post">
{% csrf_token %}
<input type="hidden" name="remote_id" value="{{ result.key }}">
<button type="submit" class="button is-small is-link">
{% trans "Import book" %}
</button>
</form>
</div>
</div>

View file

@ -221,6 +221,7 @@ urlpatterns = [
), ),
# search # search
re_path(r"^search/?$", views.Search.as_view(), name="search"), re_path(r"^search/?$", views.Search.as_view(), name="search"),
re_path(r"^search.json/?$", views.Search.as_view(), name="search"),
# imports # imports
re_path(r"^import/?$", views.Import.as_view(), name="import"), re_path(r"^import/?$", views.Import.as_view(), name="import"),
re_path(r"^import/(\d+)/?$", views.ImportStatus.as_view(), name="import-status"), re_path(r"^import/(\d+)/?$", views.ImportStatus.as_view(), name="import-status"),

View file

@ -32,7 +32,9 @@ def get_user_from_username(viewer, username):
def is_api_request(request): def is_api_request(request):
"""check whether a request is asking for html or data""" """check whether a request is asking for html or data"""
return "json" in request.headers.get("Accept", "") or request.path[-5:] == ".json" return "json" in request.headers.get("Accept", "") or re.match(
r".*\.json/?$", request.path
)
def is_bookwyrm_request(request): def is_bookwyrm_request(request):

View file

@ -10,7 +10,7 @@ from django.views import View
from bookwyrm import models from bookwyrm import models
from bookwyrm.connectors import connector_manager from bookwyrm.connectors import connector_manager
from bookwyrm.book_search import search from bookwyrm.book_search import search, format_search_result
from bookwyrm.settings import PAGE_LENGTH from bookwyrm.settings import PAGE_LENGTH
from bookwyrm.utils import regex from bookwyrm.utils import regex
from .helpers import is_api_request, privacy_filter from .helpers import is_api_request, privacy_filter
@ -33,7 +33,9 @@ class Search(View):
if is_api_request(request): if is_api_request(request):
# only return local book results via json so we don't cascade # only return local book results via json so we don't cascade
book_results = search(query, min_confidence=min_confidence) book_results = search(query, min_confidence=min_confidence)
return JsonResponse([r.json() for r in book_results], safe=False) return JsonResponse(
[format_search_result(r) for r in book_results], safe=False
)
if query and not search_type: if query and not search_type:
search_type = "user" if "@" in query else "book" search_type = "user" if "@" in query else "book"