Allow filter/sort to include inferred/document dates (#1347)

geniza/corpus/forms.py

@@ -216,6 +216,13 @@ class DocumentSearchForm(RangeForm):
         ),
     )
 
+    exclude_inferred = forms.BooleanField(
+        # Translators: label for "exclude inferred dates" search form filter
+        label=_("Exclude inferred dates"),
+        required=False,
+        widget=forms.CheckboxInput,
+    )
+
     doctype = FacetChoiceField(
         # Translators: label for document type search form filter
         label=_("Document type"),

geniza/corpus/models.py

@@ -1220,6 +1220,10 @@ def index_data(self):
                 ),
                 # combined original/standard document date for display
                 "document_date_t": strip_tags(self.document_date) or None,
+                # inferred document date for display
+                "document_dating_t": standard_date_display(
+                    "/".join([d.isoformat() for d in self.dating_range() if d])
+                ),
                 # date range for filtering
                 "document_date_dr": self.solr_date_range(),
                 # date range for filtering, but including inferred datings if any exist
@@ -1232,6 +1236,17 @@ def index_data(self):
                 "end_date_i": (
                     self.end_date.numeric_format(mode="max") if self.end_date else None
                 ),
+                # start/end of document date or date range, including inferred datings, for sort
+                "start_dating_i": (
+                    self.dating_range()[0].numeric_format()
+                    if self.dating_range()[0]
+                    else None
+                ),
+                "end_dating_i": (
+                    self.dating_range()[1].numeric_format(mode="max")
+                    if self.dating_range()[1]
+                    else None
+                ),
                 # library/collection possibly redundant?
                 "collection_ss": [str(f.collection) for f in fragments],
                 "tags_ss_lower": [t.name for t in self.tags.all()],

geniza/corpus/solr_queryset.py

@@ -48,6 +48,7 @@ class DocumentSolrQuerySet(AliasedSolrQuerySet):
         "shelfmark": "shelfmark_s",  # string version for display
         "shelfmarks": "fragment_shelfmark_ss",
         "document_date": "document_date_t",  # text version for search & display
+        "document_dating": "document_dating_t",  # inferred date for display
         "original_date_t": "original_date",
         "collection": "collection_ss",
         "tags": "tags_ss_lower",

geniza/corpus/templates/corpus/document_list.html

@@ -56,6 +56,11 @@ <h1>{{ page_title }}</h1>
                     {# NOTE: stimulus action is configured via django widget attrs #}
                     {{ form.docdate }}
                 </label>
+                <label for="{{ form.exclude_inferred.auto_id }}">
+                    {% render_field form.exclude_inferred data-action="search#update" %}
+                    <span>{{ form.exclude_inferred.label }}</span>
+                    <div class="thumb" aria-hidden="true"></div>
+                </label>
             </div>
             <fieldset class="includes-fields">
                 <legend><span class="fieldname">{% translate "Includes" %}</span></legend>

geniza/corpus/templates/corpus/snippets/document_result.html

@@ -21,6 +21,16 @@ <h2 class="title">
                             {{ document.document_date.0 }} {# indexed as _t which is a multival field #}
                         </time>
                     </dd>
+                {% elif document.document_dating %}
+                    <dt>
+                        {# Translators: label for inferred date on a document #}
+                        {% translate "Inferred date" %}
+                    </dt>
+                    <dd>
+                        <time>
+                            {{ document.document_dating.0 }} {# indexed as _t which is a multival field #}
+                        </time>
+                    </dd>
                 {% endif %}
                 {% if document.languages|length %}
                     <dt>

geniza/corpus/tests/test_corpus_views.py

@@ -547,8 +547,8 @@ def test_get_range_stats(self, mock_solr_queryset):
             # mock_queryset_cls.return_value.stats.return_value.get_stats.return_value = {
             mock_queryset_cls.return_value.get_stats.return_value = {
                 "stats_fields": {
-                    "start_date_i": {"min": None},
-                    "end_date_i": {"max": None},
+                    "start_dating_i": {"min": None},
+                    "end_dating_i": {"max": None},
                 }
             }
             docsearch_view = DocumentSearchView()
@@ -560,14 +560,14 @@ def test_get_range_stats(self, mock_solr_queryset):
             )
             assert stats == {"docdate": (None, None)}
             mock_queryset_cls.return_value.stats.assert_called_with(
-                "start_date_i", "end_date_i"
+                "start_dating_i", "end_dating_i"
             )
 
             # convert integer date to year
             mock_queryset_cls.return_value.get_stats.return_value = {
                 "stats_fields": {
-                    "start_date_i": {"min": 10380101.0},
-                    "end_date_i": {"max": 10421231.0},
+                    "start_dating_i": {"min": 10380101.0},
+                    "end_dating_i": {"max": 10421231.0},
                 }
             }
             stats = docsearch_view.get_range_stats(
@@ -577,7 +577,7 @@ def test_get_range_stats(self, mock_solr_queryset):
 
             # test three-digit year
             mock_queryset_cls.return_value.get_stats.return_value["stats_fields"][
-                "start_date_i"
+                "start_dating_i"
             ]["min"] = 8430101.0
             stats = docsearch_view.get_range_stats(
                 queryset_cls=mock_queryset_cls, field_name="docdate"
@@ -979,6 +979,14 @@ def test_get_solr_sort(self):
         assert random_sort.startswith("random_")
         assert int(random_sort.split("_")[1])
 
+        # doc dating without exclude_inferred: should include inferred
+        dating_sort = docsearch_view.get_solr_sort("docdate_asc")
+        assert dating_sort.startswith("start_dating_")
+
+        # with exclude_inferred: should use start_dating, which is dates without inferred
+        dating_sort = docsearch_view.get_solr_sort("docdate_asc", "true")
+        assert dating_sort.startswith("start_date_")
+
     def test_random_page_redirect(self, client):
         # any page of results other than one should redirect to the first page
         docsearch_url = reverse("corpus:document-search")

geniza/corpus/views.py

@@ -53,13 +53,13 @@ def get_range_stats(self, queryset_cls, field_name):
             the key is not added to a dictionary.
         :rtype: dict
         """
-        stats = queryset_cls().stats("start_date_i", "end_date_i").get_stats()
+        stats = queryset_cls().stats("start_dating_i", "end_dating_i").get_stats()
         if stats.get("stats_fields"):
             # use minimum from start date and max from end date
             # - we're storing YYYYMMDD as 8-digit number for this we only want year
             # convert to str, take first 4 digits, then convert back to int
-            min_val = stats["stats_fields"]["start_date_i"]["min"]
-            max_val = stats["stats_fields"]["end_date_i"]["max"]
+            min_val = stats["stats_fields"]["start_dating_i"]["min"]
+            max_val = stats["stats_fields"]["end_dating_i"]["max"]
 
             # trim from the end to handle 3-digit years; includes .0 at end
             min_year = int(str(min_val)[:-6]) if min_val else None
@@ -96,6 +96,8 @@ class DocumentSearchView(
         "shelfmark": "shelfmark_natsort",
         "docdate_asc": "start_date_i",
         "docdate_desc": "-end_date_i",
+        "docdating_asc": "start_dating_i",
+        "docdating_desc": "-end_dating_i",
     }
 
     def dispatch(self, request, *args, **kwargs):
@@ -116,13 +118,16 @@ def last_modified(self):
             return None
         return super().last_modified()
 
-    def get_solr_sort(self, sort_option):
+    def get_solr_sort(self, sort_option, exclude_inferred=False):
         """Return solr sort field for user-seleted sort option;
         generates random sort field using solr random dynamic field;
         otherwise uses solr sort field from :attr:`solr_sort`"""
         if sort_option == "random":
             # use solr's random dynamic field to sort randomly
             return "random_%s" % randint(1000, 9999)
+        elif "docdate" in sort_option and not exclude_inferred:
+            # use inferred datings if exclude_inferred is not true
+            sort_option = sort_option.replace("date", "dating")
         return self.solr_sort[sort_option]
 
     def get_form_kwargs(self):
@@ -249,7 +254,11 @@ def get_queryset(self):
                 )  # include relevance score in results
 
             # order by sort option
-            documents = documents.order_by(self.get_solr_sort(search_opts["sort"]))
+            documents = documents.order_by(
+                self.get_solr_sort(
+                    search_opts["sort"], search_opts.get("exclude_inferred", False)
+                )
+            )
 
             # filter by type if specified
             if search_opts["doctype"]:
@@ -286,9 +295,13 @@ def get_queryset(self):
             if search_opts["docdate"]:
                 # date range filter; returns tuple of value or None for open-ended range
                 start, end = search_opts["docdate"]
-                documents = documents.filter(
-                    document_date_dr="[%s TO %s]" % (start or "*", end or "*")
+                date_filter = "[%s TO %s]" % (start or "*", end or "*")
+                date_field = (
+                    "document_date_dr"
+                    if search_opts.get("exclude_inferred", False)
+                    else "document_dating_dr"
                 )
+                documents = documents.filter(**{date_field: date_filter})
                 label = "%s–%s" % (start, end)
                 if start and not end:
                     label = _("After %s") % start

sitemedia/scss/base/_colors.scss

@@ -24,6 +24,7 @@ $off-white: #f7f7f7;
     --primary-80: rgba(67, 96, 67, 0.8);
     --secondary: #567856;
     --secondary-80: rgba(86, 120, 86, 0.8);
+    --secondary-40: rgba(86, 120, 86, 0.4);
     --tertiary: #7bac7b;
     --link-primary: #567856;
     --link-secondary: #436043;
@@ -107,6 +108,7 @@ $off-white: #f7f7f7;
     --primary-80: rgba(183, 102, 128, 0.8);
     --secondary: #b76680;
     --secondary-80: rgba(195, 127, 151, 0.8);
+    --secondary-40: rgba(195, 127, 151, 0.4);
     --tertiary: #b05070;
     --link-primary: #c37f97;
     --link-secondary: #b76680;

sitemedia/scss/components/_searchform.scss

@@ -277,6 +277,61 @@ main.people {
     }
 }
 
+main.search form fieldset#filters div.fieldset-left-column {
+    @include breakpoints.for-tablet-landscape-up {
+        gap: 1rem;
+    }
+    label[for="id_exclude_inferred"] {
+        position: relative;
+        cursor: pointer;
+        input[type="checkbox"] + span::before {
+            display: none;
+        }
+        input[type="checkbox"] + span {
+            display: flex;
+            justify-content: space-between;
+            align-items: center;
+            width: 100%;
+        }
+        input[type="checkbox"] + span::after {
+            display: block;
+            content: "";
+            border-radius: 7px;
+            width: 34px;
+            height: 14px;
+            background-color: var(--background-gray);
+            transition: background-color 0.1s ease-in-out;
+        }
+        input[type="checkbox"]:active + span::after,
+        input[type="checkbox"]:checked + span::after {
+            background-color: var(--secondary-40);
+        }
+        input[type="checkbox"]:checked:active + span::after {
+            background-color: var(--background-gray);
+        }
+        .thumb {
+            display: block;
+            width: 20px;
+            height: 20px;
+            position: absolute;
+            top: calc(50% - 10px);
+            right: 14px;
+            transition: right 0.1s ease-in-out,
+                background-color 0.1s ease-in-out;
+            border-radius: 50%;
+            background-color: var(--disabled-on-background-light);
+            box-shadow: 0px 1px 3px 0px #00000033;
+            box-shadow: 0px 2px 1px 0px #0000001f;
+            box-shadow: 0px 1px 1px 0px #00000024;
+        }
+        input[type="checkbox"]:active + span + .thumb,
+        input[type="checkbox"]:checked + span + .thumb {
+            right: 0;
+            background-color: var(--secondary);
+        }
+    }
+}
+
 // tweaks for RTL search form for hebrew, arabic
 html[dir="rtl"] main.search form {
     // search query box and button