From 45767d0927b38db09462d703400a0ac392d20d7a Mon Sep 17 00:00:00 2001 From: Ben Silverman Date: Wed, 27 Mar 2024 11:56:41 -0400 Subject: [PATCH] Allow bulk manifest metadata update via CSV --- apps/iiif/manifests/admin.py | 10 +++- apps/iiif/manifests/forms.py | 48 ++++++++++++++++++- .../templates/admin/change_list_override.html | 26 ++++++++++ .../templates/manifest_metadata_import.html | 30 ++++++++++++ apps/iiif/manifests/views.py | 42 +++++++++++++++- 5 files changed, 152 insertions(+), 4 deletions(-) create mode 100644 apps/iiif/manifests/templates/admin/change_list_override.html create mode 100644 apps/iiif/manifests/templates/manifest_metadata_import.html diff --git a/apps/iiif/manifests/admin.py b/apps/iiif/manifests/admin.py index 50288374c..d0ec3aef2 100644 --- a/apps/iiif/manifests/admin.py +++ b/apps/iiif/manifests/admin.py @@ -2,6 +2,7 @@ from django.contrib import admin from django.http import HttpResponseRedirect from django.http.request import HttpRequest +from django.urls import reverse from django.urls.conf import path from import_export import resources, fields from import_export.admin import ImportExportModelAdmin @@ -9,7 +10,7 @@ from django_summernote.admin import SummernoteModelAdmin from .models import Manifest, Note, ImageServer, RelatedLink from .forms import ManifestAdminForm -from .views import AddToCollectionsView +from .views import AddToCollectionsView, MetadataImportView from ..kollections.models import Collection class ManifestResource(resources.ModelResource): @@ -56,6 +57,7 @@ class ManifestAdmin(ImportExportModelAdmin, SummernoteModelAdmin, admin.ModelAdm form = ManifestAdminForm actions = ['add_to_collections_action'] inlines = [RelatedLinksInline] + change_list_template = 'admin/change_list_override.html' def add_to_collections_action(self, request, queryset): """Action choose manifests to add to collections""" @@ -72,6 +74,12 @@ def get_urls(self): self.admin_site.admin_view(AddToCollectionsView.as_view()), {'model_admin': self, }, name="AddManifestsToCollections", + ), + path( + 'manifest_metadata_import/', + self.admin_site.admin_view(MetadataImportView.as_view()), + {'model_admin': self, }, + name="MultiManifestMetadataImport", ) ] return my_urls + urls diff --git a/apps/iiif/manifests/forms.py b/apps/iiif/manifests/forms.py index 6df732c25..0755a12af 100644 --- a/apps/iiif/manifests/forms.py +++ b/apps/iiif/manifests/forms.py @@ -1,9 +1,14 @@ """Django Forms for export.""" +import csv +from io import StringIO import logging from django import forms from django.contrib.admin import site as admin_site, widgets -from .models import Language, Manifest -from ..canvases.models import Canvas +from django.core.validators import FileExtensionValidator + +from apps.iiif.manifests.models import Manifest +from apps.iiif.canvases.models import Canvas +from apps.ingest.services import normalize_header LOGGER = logging.getLogger(__name__) @@ -49,3 +54,42 @@ def __init__(self, *args, **kwargs): self.instance._meta.get_field('collections').remote_field, admin_site, ) + +class ManifestCSVImportForm(forms.Form): + """Form to import a CSV and update the metadata for multiple manifests""" + + csv_file = forms.FileField( + required=True, + validators=[FileExtensionValidator(allowed_extensions=['csv'])], + label="CSV File", + help_text=""" +

Provide a CSV with a pid column, whose value in each row must match + the PID of an existing volume. Additional columns will be used to update the volume's + metadata.

+

Columns matching Manifest model field names will update those + fields directly, and any additional columns will be used to populate the volume's + metadata JSON field.

+ """, + ) + + def clean(self): + # check csv has pid column + super().clean() + csv_file = self.cleaned_data.get('csv_file') + if csv_file: + reader = csv.DictReader( + normalize_header( + StringIO(csv_file.read().decode('utf-8')) + ), + ) + if 'pid' not in reader.fieldnames: + self.add_error( + 'metadata_spreadsheet', + forms.ValidationError( + """Spreadsheet must have pid column. Check to ensure there + are no stray characters in the header row.""" + ), + ) + # return back to start of file so we can read again + csv_file.seek(0, 0) + return self.cleaned_data diff --git a/apps/iiif/manifests/templates/admin/change_list_override.html b/apps/iiif/manifests/templates/admin/change_list_override.html new file mode 100644 index 000000000..1f76d59dc --- /dev/null +++ b/apps/iiif/manifests/templates/admin/change_list_override.html @@ -0,0 +1,26 @@ +{% extends "admin/change_list.html" %} + +{% load i18n %} + +{# adapted from django-import-export #} +{% block object-tools %} + +{% endblock %} diff --git a/apps/iiif/manifests/templates/manifest_metadata_import.html b/apps/iiif/manifests/templates/manifest_metadata_import.html new file mode 100644 index 000000000..7191c6630 --- /dev/null +++ b/apps/iiif/manifests/templates/manifest_metadata_import.html @@ -0,0 +1,30 @@ +{% extends "admin/base_site.html" %} +{% load i18n admin_urls %} +{% load static %} + +{% block extrastyle %} + +{% endblock %} + +{% block breadcrumbs %} +
+ {% trans 'Home' %} + › Manifests + › {{ title }} +
+{% endblock %} + +{% block content %} +
{% csrf_token %} +
+
+ {{ form }} +
+
+ {% block submit_buttons_bottom %} +
+ +
+ {% endblock %} +
+{% endblock %} diff --git a/apps/iiif/manifests/views.py b/apps/iiif/manifests/views.py index 395932a37..07ce81927 100644 --- a/apps/iiif/manifests/views.py +++ b/apps/iiif/manifests/views.py @@ -1,4 +1,6 @@ """Django views for manifests""" +import csv +from io import StringIO import json import logging from datetime import datetime @@ -10,8 +12,10 @@ from django.core.serializers import serialize from django.contrib.sitemaps import Sitemap from django.urls import reverse + +from apps.ingest.services import normalize_header, set_metadata from .models import Manifest -from .forms import ManifestsCollectionsForm +from .forms import ManifestCSVImportForm, ManifestsCollectionsForm LOGGER = logging.getLogger(__name__) @@ -154,3 +158,39 @@ def get_success_url(self): self.request, messages.SUCCESS, 'Successfully added manifests to collections' ) return reverse('admin:manifests_manifest_changelist') + +class MetadataImportView(FormView): + """Admin page to import a CSV and update multiple Manifests' metadata""" + + template_name = 'manifest_metadata_import.html' + form_class = ManifestCSVImportForm + + def get_context_data(self, **kwargs): + """Set page title on context data""" + context = super().get_context_data(**kwargs) + context['title'] = 'Manifest metadata bulk update (CSV import)' + return context + + def form_valid(self, form): + """Read the CSV file and, find associated manifests, and update metadata""" + csv_file = form.cleaned_data.get('csv_file') + csv_io = StringIO(csv_file.read().decode('utf-8')) + reader = csv.DictReader(normalize_header(csv_io)) + for row in reader: + try: + # try to find manifest + manifest = Manifest.objects.get(pid=row['pid']) + # use ingest set_metadata function + set_metadata(manifest, metadata=row) + except Manifest.DoesNotExist: + messages.add_message( + self.request, messages.ERROR, f'Manifest with pid {row["pid"]} not found' + ) + return super().form_valid(form) + + def get_success_url(self): + """Return to the manifest change list with a success message""" + messages.add_message( + self.request, messages.SUCCESS, 'Successfully updated manifests' + ) + return reverse('admin:manifests_manifest_changelist')