Container autocomplete: improve search

2 files changed, 94 insertions, 51 deletions

diff --git a/archaeological_warehouse/tests.py b/archaeological_warehouse/tests.py
index f6d877ec4..21833a08f 100644
--- a/archaeological_warehouse/tests.py
+++ b/archaeological_warehouse/tests.py
@@ -514,8 +514,8 @@ class ContainerTest(FindInit, TestCase):
         )
 
     def test_container_search(self):
-        ct = models.ContainerType.objects.all()[0]
-        ct2 = models.ContainerType.objects.all()[0]
+        ct = models.ContainerType.objects.order_by("id").all()[0]
+        ct2 = models.ContainerType.objects.order_by("id").all()[1]
         container_1 = models.Container.objects.create(
             reference="Test", responsible=self.main_warehouse,
             location=self.main_warehouse,
@@ -547,9 +547,8 @@ class ContainerTest(FindInit, TestCase):
         self.assertEqual(response.status_code, 200)
         c = json.loads(response.content.decode())
         self.assertEqual(len(c), 1)
-        response = client.get(
-            url,
-            {"term": "{} Test {} Test 35000".format(ct.label, ct2.label)})
+        full_path = "{} Test {} 35000".format(ct.label, ct2.label)
+        response = client.get(url, {"term": full_path})
         self.assertEqual(response.status_code, 200)
         c = json.loads(response.content.decode())
         self.assertEqual(len(c), 1)
diff --git a/archaeological_warehouse/views.py b/archaeological_warehouse/views.py
index efa5bdf19..a9d1bf409 100644
--- a/archaeological_warehouse/views.py
+++ b/archaeological_warehouse/views.py
@@ -113,63 +113,106 @@ def autocomplete_container(request, warehouse_id=None):
     containers += list(models.Container.objects.filter(
         query).values('id', 'cached_label')[:limit])
     limit = 15 - len(containers)
-    splitted = [s for s in term.split(' ') if s]
-    if limit > 0:
-        if len(splitted) > 1 and not len(splitted) % 2:
-            # group by container type, ref tuple
-            groups = [(splitted[idx * 2], splitted[idx * 2 + 1])
-                      for idx in range(int(len(splitted) / 2))]
+    splitted = [s.lower() for s in term.split(' ') if s]
+
+    if limit > 0 and len(splitted) > 1:
+        type_positions = []  # container_type ID, pos inf, pos sup
+        container_types = [
+            (c[0], c[1])
+            for c in models.ContainerType.objects.values_list("id", "label")
+        ]
+        for container_type_id, value in container_types:
+            if value.lower() in term.lower():  # container_type is in search q
+                values = [v.lower() for v in value.split(" ") if v]
+
+                # verify that all term match in splitted
+                try:
+                    index = splitted.index(values[0].lower())
+                except ValueError:
+                    index = None
+                index_is_ok = False
+                while not index_is_ok and index is not None:
+                    for idx, v in enumerate(values):
+                        try:
+                            assert splitted[index + idx] == v
+                        except (ValueError, AssertionError):
+                            break
+                        index_is_ok = True
+                    if not index_is_ok:
+                        try:
+                            index = splitted.index(values[0].lower(), index + 1)
+                        except ValueError:
+                            index = None
+                if index_is_ok:
+                    type_positions.append(
+                        (container_type_id, index, index + len(values)))
+
+        query = base_query
+        if not warehouse_id and type_positions and type_positions[0][1] > 0:
+            for idx in range(type_positions[0][1]):
+                query &= Q(location__name__icontains=splitted[idx])
+        # group by container type, ref tuple
+        groups = []
+        for idx, (container_type_id, pos_inf, pos_sup) \
+                in enumerate(type_positions):
+            if len(type_positions) == idx + 1:  # last
+                value = " ".join(splitted[pos_sup:])
+            else:
+                value = " ".join(splitted[pos_sup:type_positions[idx + 1][1]])
+            groups.append((container_type_id, value))
+        if groups:
             query = base_query
             for idx, g in enumerate(reversed(groups)):
                 base_key = "parent__" * idx
-                key1 = base_key + "container_type__label__unaccent__iexact"
+                key1 = base_key + "container_type_id"
                 key2 = base_key + "reference__unaccent__iexact"
                 query &= Q(**{key1: g[0], key2: g[1]})
-            ids = set([c["id"] for c in containers])
+            ids = {c["id"] for c in containers}
             containers += list(models.Container.objects.filter(
                 query).exclude(pk__in=ids).values('id', 'cached_label')[
                                :limit])
 
-        if len(splitted) > 1 and (15 - len(containers)) > 0:
-            # group to do a "type" "reference" search
-            for idx in range(1, len(splitted)):
-                group_1 = splitted[:idx]
-                group_2 = splitted[idx:]
-                extra = Q(
-                    container_type__label__unaccent__iexact=" ".join(group_1),
-                    reference__unaccent__iexact=" ".join(group_2))
-                query = base_query & extra
-                ids = set([c["id"] for c in containers])
-                containers += list(models.Container.objects.filter(
-                    query).exclude(pk__in=ids).values('id', 'cached_label')[
-                                   :limit])
-                if (15 - len(containers)) <= 0:
-                    break
-        if (15 - len(containers)) > 0:
+    if len(splitted) > 1 and len(containers) < 15:
+        # group to do a "type" "reference" search
+        for idx in range(1, len(splitted)):
+            group_1 = splitted[:idx]
+            group_2 = splitted[idx:]
+            extra = Q(
+                container_type__label__unaccent__iexact=" ".join(group_1),
+                reference__unaccent__iexact=" ".join(group_2))
+            query = base_query & extra
+            ids = {c["id"] for c in containers}
+            containers += list(models.Container.objects.filter(
+                query).exclude(pk__in=ids).values('id', 'cached_label')[
+                               :limit])
+            if len(containers) >= 15:
+                break
+
+    if len(containers) < 15:
+        query = base_query
+        for q in splitted:
+            extra = Q(reference__unaccent__iexact=q)
+            query = query & extra
+        ids = {c["id"] for c in containers}
+        containers += list(models.Container.objects.filter(
+            query).exclude(pk__in=ids).values('id', 'cached_label')[:limit])
+        limit = 15 - len(containers)
+        if limit > 0:
             query = base_query
             for q in splitted:
-                extra = Q(reference__unaccent__iexact=q)
+                extra = Q(container_type__label__unaccent__icontains=q) | \
+                    Q(container_type__reference__unaccent__icontains=q) | \
+                    Q(reference__unaccent__icontains=q) | \
+                    Q(cached_label__unaccent__icontains=q)
+                if not warehouse_id:
+                    extra |= Q(location__name__unaccent=q) | Q(
+                        location__town__unaccent=q)
                 query = query & extra
-            ids = set([c["id"] for c in containers])
-            containers += list(models.Container.objects.filter(
-                query).exclude(pk__in=ids).values('id', 'cached_label')[:limit])
-            limit = 15 - len(containers)
-            if limit > 0:
-                query = base_query
-                for q in splitted:
-                    extra = Q(container_type__label__unaccent__icontains=q) | \
-                        Q(container_type__reference__unaccent__icontains=q) | \
-                        Q(reference__unaccent__icontains=q) | \
-                        Q(cached_label__unaccent__icontains=q)
-                    if not warehouse_id:
-                        extra |= Q(location__name__unaccent=q) | Q(
-                            location__town__unaccent=q)
-                    query = query & extra
-                ids = set([c["id"] for c in containers])
-                containers += list(
-                    models.Container.objects.filter(query).exclude(
-                        pk__in=ids
-                    ).values('id', 'cached_label')[:limit])
+            ids = {c["id"] for c in containers}
+            containers += list(
+                models.Container.objects.filter(query).exclude(
+                    pk__in=ids
+                ).values('id', 'cached_label')[:limit])
     data = json.dumps(
         [{'id': container['id'], 'value': container['cached_label']}
          for container in containers])
@@ -304,7 +347,8 @@ def container_treatment_add(request, pk, current_right=None):
     except models.Container.DoesNotExist:
         raise Http404()
     return treatment_add(
-        request, ",".join([str(f.pk) for f in container.finds.all()]))
+        request, ",".join(str(f.pk) for f in container.finds.all())
+    )
 
 """
 warehouse_packaging_wizard = ItemSourceWizard.as_view([