Add settings for custom metadata facets and index them (#960)

apps/iiif/manifests/documents.py

@@ -1,6 +1,7 @@
 """Elasticsearch indexing rules for IIIF manifests"""
 
 from html import unescape
+from django.conf import settings
 from django_elasticsearch_dsl import Document, fields
 from django_elasticsearch_dsl.registries import registry
 from elasticsearch_dsl import MetaField, Keyword, analyzer
@@ -106,6 +107,38 @@ def prepare_languages(self, instance):
             return [lang.name for lang in instance.languages.all()]
         return ["[no language]"]
 
+    def prepare_metadata(self, instance):
+        """use custom metadata settings to prepare metadata field"""
+        custom_metadata = {}
+
+        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, opts in settings.CUSTOM_METADATA.items():
+                val = None
+                # each key in CUSTOM_METADATA dict should be a metadata key.
+                # however, instance.metadata will generally be a list rather than a dict: it's a
+                # jsonfield that maps to the IIIF manifest metadata field, which is a list
+                # consisting of dicts like { label: str, value: str }
+                if isinstance(instance.metadata, list):
+                    # find matching value by "label" == key
+                    for obj in instance.metadata:
+                        if "label" in obj and obj["label"] == key and "value" in obj:
+                            val = obj["value"]
+                            break
+                elif isinstance(instance.metadata, dict):
+                    # in some cases it may be just a dict, so in that case, use get()
+                    val = instance.metadata.get(key, None)
+                # should have "multi" bool and if multi is True, "separator" string
+                if val and opts.get("multi", False) == True:
+                    val = val.split(opts.get("separator", ";"))
+                custom_metadata[key] = val
+
+        return custom_metadata
+
     def prepare_summary(self, instance):
         """Strip HTML tags from summary"""
         return unescape(strip_tags(instance.summary))

config/settings/local.dst

@@ -171,3 +171,24 @@ LOGGING = {
 
 # Must be configured to the host and port running Elasticsearch
 ELASTICSEARCH_DSL['default']['hosts'] = 'user:pass@localhost:9200'
+
+# Custom faceted metadata fields configuration. Must be a dict with key names
+# matching exactly the ingest metadata spradsheet column names to be indexed.
+# For each column, configure options for whether it should accept multiple
+# values in a single record ("multi": True), and if so, what separator will be
+# used ("separator": ";", semicolon by default).
+# Note that these key names will also correspond to "label" names in the Manifest
+# metadata list.
+# 
+# Example:
+# CUSTOM_METADATA = {
+#     "Column Name": {
+#         "multi": True,
+#         "separator": ";",
+#     },
+#     "Other Column": {
+#         "multi": False,
+#     },
+# }
+
+CUSTOM_METADATA = {}