From 6c0d4671f7611afb78b008ae04f9a1080bde6627 Mon Sep 17 00:00:00 2001
From: Friedel Wolff <friedel.wolff@nwu.ac.za>
Date: Fri, 2 Aug 2024 21:51:54 +0200
Subject: [PATCH] Rework filtering and searching for projects

Two intertwined issues are fixed:
 - An empty search query would cause no projects to be returned. Now
   we only search if a query is given.
 - We weren't filtering projects at all. We're really going against
   the intention of django_filters with this one, but let's see...

The test was simply wrong (and duplicated).
---
 app/general/filters.py           | 46 +++++++++++++++++---------------
 app/general/tests/test_filter.py | 10 ++-----
 2 files changed, 26 insertions(+), 30 deletions(-)

diff --git a/app/general/filters.py b/app/general/filters.py
index e75f96bc..72fe7ef9 100644
--- a/app/general/filters.py
+++ b/app/general/filters.py
@@ -14,7 +14,7 @@
 
 
 class DocumentFileFilter(django_filters.FilterSet):
-    search = django_filters.CharFilter(method="filter_search", label="Search")
+    search = django_filters.CharFilter(method="ignore", label="Search")
 
     institution = ModelMultipleChoiceFilter(
         queryset=Institution.objects.all(), widget=forms.CheckboxSelectMultiple
@@ -37,7 +37,6 @@ class Meta:
     def filter_queryset(self, queryset):
         # More information about weighting and normalization in postgres:
         # https://www.postgresql.org/docs/current/textsearch-controls.html#TEXTSEARCH-RANKING
-        queryset = super().filter_queryset(queryset)
 
         search = self.form.cleaned_data.get("search", "").strip()
         query = SearchQuery(search)
@@ -60,20 +59,18 @@ def filter_queryset(self, queryset):
         project_search_vector = SearchVector("name", weight="A") + SearchVector(
             "description", weight="B"
         )
-        project_query = (
-            Project.objects.annotate(
-                heading=F("name"),
-                view=Value("project_detail"),
-                logo_url=F("logo"),
-                associated_url=F("url"),
-                search_headline=SearchHeadline("description", query, max_words=15, min_words=10),
-                rank=SearchRank(project_search_vector, query, normalization=16),
-                search=project_search_vector,
-            )
-            .filter(search=query)
-            .values(*fields)
+        project_query = Project.objects.annotate(
+            heading=F("name"),
+            view=Value("project_detail"),
+            logo_url=F("logo"),
+            associated_url=F("url"),
+            search_headline=SearchHeadline("description", query, max_words=15, min_words=10),
+            rank=SearchRank(project_search_vector, query, normalization=16),
         )
 
+        queryset = super().filter_queryset(queryset)
+        project_query = super().filter_queryset(project_query)
+
         # We limit the headline to limit the performance impact. On very large
         # documents, this slows things down if unconstrained.
         search_headline = SearchHeadline(
@@ -87,13 +84,18 @@ def filter_queryset(self, queryset):
             associated_url=F("url"),
             rank=search_rank,
             search_headline=search_headline,
-        ).values(*fields)
-        return queryset.union(project_query, all=True).order_by("-rank")
+        )
+        if search:
+            # An empty search on Project filters out everything.
+            queryset = queryset.filter(search_vector=query)
+            project_query = project_query.annotate(search=project_search_vector).filter(
+                search=query
+            )
 
-    def filter_search(self, queryset, name, value):
-        if value:
-            query = SearchQuery(value.strip())
-            return queryset.filter(search_vector=query)
+        queryset = queryset.values(*fields)
+        project_query = project_query.values(*fields)
+        return queryset.union(project_query, all=True).order_by("-rank")
 
-        else:
-            return queryset
+    def ignore(self, queryset, name, value):
+        # All fields are handled in `.filter_queryset()`
+        return queryset
diff --git a/app/general/tests/test_filter.py b/app/general/tests/test_filter.py
index 9eea3982..bc328769 100644
--- a/app/general/tests/test_filter.py
+++ b/app/general/tests/test_filter.py
@@ -50,14 +50,8 @@ def test_institution_filter(self):
         data = {"institution": [self.institution1.id]}
         filter = DocumentFileFilter(data=data)
         qs = filter.qs
-        self.assertEqual(len(qs), 1)
-        self.assertEqual(qs[0]["id"], self.doc1.id)
-
-    def test_institution_filter(self):
-        data = {"institution": [self.institution1.id]}
-        filter = DocumentFileFilter(data=data)
-        qs = filter.qs
-        self.assertEqual(len(qs), 1)
+        self.assertEqual(len(qs), 2)
+        # TODO: ordering between documents and projects are not yet defined
         self.assertEqual(qs[0]["id"], self.doc1.id)
 
     def test_subjects_filter(self):