Add custom metadata facets to search form and view (#960)

apps/readux/forms.py

@@ -2,6 +2,7 @@
 
 from dateutil import parser
 from django import forms
+from django.conf import settings
 from django.template.defaultfilters import truncatechars
 
 
@@ -139,12 +140,41 @@ class ManifestSearchForm(forms.Form):
         )
     )
 
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+
+        # pull additional facets from Elasticsearch
+        if (
+            settings
+            and hasattr(settings, "CUSTOM_METADATA")
+            and isinstance(settings.CUSTOM_METADATA, dict)
+        ):
+            # should be a dict like {meta_key: {"multi": bool, "separator": str}}
+            for key in settings.CUSTOM_METADATA.keys():
+                self.fields[
+                    # use django-friendly form field names
+                    key.casefold().replace(" ", "_")
+                ] = FacetedMultipleChoiceField(
+                    label=key,
+                    required=False,
+                    widget=forms.SelectMultiple(
+                        attrs={
+                            "aria-label": f"Filter volumes by {key}",
+                            "class": "uk-input",
+                        },
+                    ),
+                )
+
     def set_facets(self, facets):
         """Use facets from Elasticsearch to populate form fields"""
         for name, buckets in facets.items():
             if name in self.fields:
                 # Assumes that name passed in the view's facets list matches form field name
                 self.fields[name].populate_from_buckets(buckets)
+            elif name.casefold().replace(" ", "_") in self.fields:
+                self.fields[name.casefold().replace(" ", "_")].populate_from_buckets(
+                    buckets
+                )
 
     def set_date(self, min_date, max_date):
         """Use min and max aggregations from Elasticsearch to populate date range fields"""

apps/readux/views.py

@@ -340,6 +340,20 @@ class VolumeSearchView(ListView, FormMixin):
         "sort": "label_alphabetical"
     }
 
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        # pull additional facets from Elasticsearch
+        if (
+            settings
+            and hasattr(settings, "CUSTOM_METADATA")
+            and isinstance(settings.CUSTOM_METADATA, dict)
+        ):
+            for key in settings.CUSTOM_METADATA.keys():
+                self.facets.append((key, NestedFacet(
+                    "metadata",
+                    TermsFacet(field=f"metadata.{key}", size=2000, min_doc_count=1),
+                )))
+
     # regex to match terms in doublequotes
     re_exact_match = re.compile(r'\B(".+?")\B')
 
@@ -368,6 +382,18 @@ def get_form_kwargs(self):
 
     def get_context_data(self, **kwargs):
         context_data = super().get_context_data(**kwargs)
+        # add configured custom metadata keys to context data
+        context_data["CUSTOM_METADATA_KEYS"] = (
+            [
+                # use django-friendly form field names
+                key.casefold().replace(" ", "_")
+                for key in settings.CUSTOM_METADATA.keys()
+            ]
+            if hasattr(settings, "CUSTOM_METADATA")
+            and isinstance(settings.CUSTOM_METADATA, dict)
+            else []
+        )
+
         volumes_response = self.get_queryset().execute()
         # populate a dict with "buckets" of extant categories for each facet
         facets = {}
@@ -519,6 +545,18 @@ def get_queryset(self):
         if max_date_filter:
             volumes = volumes.filter("range", date_latest={"lte": max_date_filter})
 
+        # filter on custom metadata fields
+        if hasattr(settings, "CUSTOM_METADATA") and isinstance(
+            settings.CUSTOM_METADATA, dict
+        ):
+            for key in settings.CUSTOM_METADATA.keys():
+                field_name = key.casefold().replace(" ", "_")
+                meta_filter = form_data.get(field_name) or ""
+                if meta_filter:
+                    volumes = volumes.filter("nested", path="metadata", query=Q(
+                        "terms", **{f"metadata.{key}": meta_filter}
+                    ))
+
         # create aggregation buckets for facet fields
         for (facet_name, facet) in self.facets:
             volumes.aggs.bucket(facet_name, facet.get_aggregation())

apps/templates/search_results.html

@@ -1,5 +1,5 @@
 {% extends "base.html" %}
-{% load wagtailcore_tags static %}
+{% load wagtailcore_tags static readux_extras %}
 
 {% block extra_head %}
     <meta property="og:title" content="Readux - Search" />
@@ -112,6 +112,16 @@ <h4 class="uk-flex uk-flex-center">Filters</h4>
                         {{ form.end_date }}
                     </noscript>
                 </fieldset>
+                {% for key in CUSTOM_METADATA_KEYS %}
+                    <fieldset class="uk-margin">
+                        <div class="uk-width-1-1">
+                            {% with form|dict_item:key as field %}
+                                <div class="uk-form-label">{{ field.label }}</div>
+                                {{ field }}
+                            {% endwith %}
+                        </div>
+                    </fieldset>
+                {% endfor %}
                 <fieldset class="uk-margin uk-width-1-1 uk-flex uk-flex-between">
                     <button id="reset-filters" class="uk-button uk-button-danger" type="button">
                         Reset filters