Find moj data 69/glossary

home/service/glossary.py

@@ -0,0 +1,43 @@
+from data_platform_catalogue.search_types import MultiSelectFilter, ResultType
+from django.core.exceptions import ObjectDoesNotExist
+from copy import deepcopy
+from itertools import groupby
+
+from .base import GenericService
+
+
+class GlossaryService(GenericService):
+    def __init__(self):
+        # Can we put the client instantiation in the base class?
+        self.client = self._get_catalogue_client()
+        self.context = self._get_context()
+
+    def _get_context(self):
+        """Returns a glossary context which is grouped by parent term"""
+        glossary_search_results = self.client.get_glossary_terms()
+        total_results = glossary_search_results.total_results
+
+        def sorter(result):
+            first_parent = result.metadata.get("parentNodes", [])
+            if first_parent:
+                return first_parent[0]["properties"]["name"]
+            if not first_parent:
+                return "Unsorted"
+
+        page_results_copy = sorted(glossary_search_results.page_results, key=sorter)
+        sorted_total_results = [
+            {"name": key, "members": list(group)}
+            for key, group
+            in groupby(page_results_copy, key=sorter)
+        ]
+        # Adding the description in the list comprehension doesn't seem to work
+        for parent_term in sorted_total_results:
+            if parent_term["members"][0].metadata.get("parentNodes"):
+                parent_term["description"] = parent_term["members"][0].metadata["parentNodes"][0]["properties"]["description"]
+            else:
+                parent_term["description"] = ""
+
+        context = {"results": sorted_total_results}
+
+        return context
+

home/urls.py

@@ -7,6 +7,7 @@
 urlpatterns = [
     path("", views.home_view, name="home"),
     path("search", views.search_view, name="search"),
+    path("glossary", views.glossary_view, name="glossary"),
     path(
         "details/<str:result_type>/<str:id>",
         views.details_view,

home/views.py

@@ -5,6 +5,7 @@
 from home.forms.search import SearchForm
 from home.service.details import DataProductDetailsService, DatasetDetailsService
 from home.service.search import SearchService
+from home.service.glossary import GlossaryService
 
 
 def home_view(request):
@@ -55,3 +56,7 @@ def search_view(request, page: str = "1"):
 
     search_service = SearchService(form=form, page=page)
     return render(request, "search.html", search_service.context)
+
+def glossary_view(request):
+    glossary_service = GlossaryService()
+    return render(request, "glossary.html", glossary_service.context)

pyproject.toml

@@ -11,7 +11,7 @@ django = "^5.0.1"
 pyyaml = "^6.0.1"
 gunicorn = "^21.2.0"
 whitenoise = "^6.6.0"
-ministryofjustice-data-platform-catalogue = "^0.15.0"
+ministryofjustice-data-platform-catalogue = "^0.16.1"
 markdown = "^3.5.2"
 python-dotenv = "^1.0.1"
 

templates/base/navigation.html

@@ -26,6 +26,7 @@
         <ul id="navigation" class="govuk-header__navigation-list">
           {% url 'home:home' as home_url %}
           {% url 'home:search' as search_url %}
+          {% url 'home:glossary' as glossary_url %}
           <li class="govuk-header__navigation-item {% if request.path == home_url %}govuk-header__navigation-item--active{%endif%}">
             <a class="govuk-header__link" href="{% url 'home:home' %}">
               Home
@@ -36,6 +37,11 @@
               Search
             </a>
           </li>
+          <li class="govuk-header__navigation-item {% if request.path == glossary_url %}govuk-header__navigation-item--active{%endif%}">
+            <a class="govuk-header__link" href="{% url 'home:glossary' %}?new=True">
+              Glossary
+            </a>
+          </li>
         </ul>
       </nav>
     </div>

templates/glossary.html

@@ -0,0 +1,46 @@
+{% extends "base/base.html" %}
+{% load markdown %}
+
+{% block content %}
+<div class="govuk-grid-row">
+  <div class="govuk-grid-column-one-quarter">
+    <h1 class="govuk-heading-l">Glossary</h1>
+  </div>
+</div>
+<div class="govuk-grid-row">
+    <div class="govuk-grid-column-one-quarter">
+        <ul class="govuk-list">
+            {% for parent_term in results %}
+            <li>
+                <div><a class="govuk-link govuk-link--no-visited-state govuk-link--no-underline" href="#{{ parent_term }}">
+                    <strong>{{ parent_term.name }}</strong>
+                </a></div>
+                {% for member in parent_term.members %}
+                    <div><a class="govuk-link govuk-link--no-visited-state govuk-link--no-underline" href="#{{ member.name }}">
+                        {{ member.name }}
+                    </a></div>
+                {%endfor%}
+                <br>
+            </li>
+            {%endfor%}
+        </ul>
+    </div>
+    <div class="govuk-grid-column-three-quarters">
+        {% for parent_term in results %}
+        <div>
+            <h1 class="govuk-heading-l" id="{{ parent_term }}">{{ parent_term.name }}</h1>
+            <p class="govuk-body-l">{{ parent_term.description }}</p>
+            <hr class="govuk-section-break govuk-section-break--m govuk-section-break--visible">
+        </div>
+            {% for member in parent_term.members %}
+            <div>
+                <h2 class="govuk-heading-m" id="{{ member.name }}">{{ member.name }}</h2>
+                <p class="govuk-body">{{ member.description|markdown }}</p>
+            </div>
+            {%endfor%}
+            <br>
+        {%endfor%}
+    </div>
+</div>
+
+{% endblock content %}

templates/home.html

@@ -6,4 +6,8 @@
   <a onclick="history.back()" class="govuk-back-link">Back</a>
   <h1 class="govuk-heading-xl">Customised page template</h1>
 
+  <p class="govuk-body">
+    <a href="{% url 'home:glossary' %}" class="govuk-link">Glossary</a>.
+  </p>
+
 {% endblock content %}

tests/conftest.py

@@ -20,6 +20,7 @@
 from home.forms.search import SearchForm
 from home.service.details import DataProductDetailsService
 from home.service.search import SearchService
+from home.service.glossary import GlossaryService
 
 from datahub.metadata.schema_classes import (
     DataProductPropertiesClass,
@@ -92,6 +93,7 @@ def mock_catalogue():
         mock_catalogue, page_results=generate_page(), total_results=100
     )
     mock_search_facets_response(mock_catalogue, domains=generate_options())
+    mock_get_glossary_terms_response(mock_catalogue)
     mock_list_data_product_response(
         mock_catalogue,
         page_results=generate_page(page_size=1, result_type=ResultType.TABLE),
@@ -132,6 +134,51 @@ def mock_get_dataproduct_aspect(mock_catalogue):
     )
     mock_catalogue.graph.get_aspect.return_value = response
 
+def mock_get_glossary_terms_response(mock_catalogue):
+    mock_catalogue.get_glossary_terms.return_value = SearchResponse(
+        total_results=3,
+        page_results=[
+            SearchResult(
+                id="urn:li:glossaryTerm:022b9b68-c211-47ae-aef0-2db13acfeca8",
+                name="IAO",
+                description="Information asset owner.\n",
+                metadata={
+                    "parentNodes": [
+                        {
+                            "properties": {
+                                "name": "Data protection terms",
+                                "description": "Data protection terms",
+                            }
+                        }
+                    ]
+                },
+                result_type="GLOSSARY_TERM",
+            ),
+            SearchResult(
+                id="urn:li:glossaryTerm:022b9b68-c211-47ae-aef0-2db13acfeca8",
+                name="Other term",
+                description="Term description to test groupings work",
+                metadata={
+                    "parentNodes": [
+                        {
+                            "properties": {
+                                "name": "Data protection terms",
+                                "description": "Data protection terms",
+                            }
+                        }
+                    ]
+                },
+                result_type="GLOSSARY_TERM",
+            ),
+            SearchResult(
+                id="urn:li:glossaryTerm:0eb7af28-62b4-4149-a6fa-72a8f1fea1e6",
+                name="Security classification",
+                description="Only data that is 'official'",
+                metadata={"parentNodes": []},
+                result_type="GLOSSARY_TERM",
+            ),
+        ],
+    )
 
 @pytest.fixture
 def valid_form():

tests/selenium/conftest.py

@@ -84,6 +84,9 @@ class HomePage(Page):
     def search_nav_link(self) -> WebElement:
         return self.selenium.find_element(By.LINK_TEXT, "Search")
 
+    def glossary_nav_link(self) -> WebElement:
+        return self.selenium.find_element(By.LINK_TEXT, "Glossary")
+
 
 class SearchResultWrapper:
     def __init__(self, element: WebElement):

tests/selenium/test_search_scenarios.py

@@ -30,8 +30,11 @@ def setup(
         self.search_page = search_page
         self.details_data_product_page = details_data_product_page
         self.chromedriver_path = chromedriver_path
-        self.details_data_product_page = details_data_product_page
-        self.chromedriver_path = chromedriver_path
+
+    def verify_glossary_link_from_homepage_works(self):
+        self.start_on_the_home_page()
+        self.click_on_the_glossary_link()
+        self.verify_i_am_on_the_glossary_page()
 
     def test_browse_to_first_item_data_product(self, mock_catalogue):
         """
@@ -206,13 +209,19 @@ def start_on_the_search_page(self):
     def click_on_the_search_link(self):
         self.home_page.search_nav_link().click()
 
+    def click_on_the_glossary_link(self):
+        self.home_page.glossary_nav_link().click()
+
     def click_on_the_search_button(self):
         self.search_page.search_button().click()
 
     def verify_i_am_on_the_search_page(self):
         assert "Search" in self.selenium.title
         assert "Find MOJ Data" in self.search_page.primary_heading().text
 
+    def verify_i_am_on_the_glossary_page(self):
+        assert "Glossary" in self.selenium.title
+
     def verify_i_have_results(self):
         result_count = self.search_page.result_count().text
         assert re.match(r"[1-9]\d* Results", result_count)

tests/test_services.py

@@ -1,10 +1,14 @@
 from types import GeneratorType
 
 import pytest
-from data_platform_catalogue.search_types import ResultType
-from home.service.search import SearchForm, SearchService
-
-from home.service.search import domains_with_their_subdomains
+from data_platform_catalogue.search_types import (
+    ResultType,
+    SearchResponse,
+    SearchResult,
+)
+from unittest.mock import patch
+from home.service.search import SearchForm, SearchService, domains_with_their_subdomains
+from home.service.glossary import GlossaryService
 
 
 class TestSearchService:
@@ -125,6 +129,77 @@ def test_get_context_data_product_tables(
         assert detail_dataproduct_context["tables"][0] == mock_table
 
 
+class TestGlossaryService:
+    def test_get_context(self):
+        glossary_context = GlossaryService()
+        expected_context = {
+            "results": [
+                {
+                    "name": "Data protection terms",
+                    "members": [
+                        SearchResult(
+                            id="urn:li:glossaryTerm:022b9b68-c211-47ae-aef0-2db13acfeca8",
+                            result_type="GLOSSARY_TERM",
+                            name="IAO",
+                            description="Information asset owner.\n",
+                            matches={},
+                            metadata={
+                                "parentNodes": [
+                                    {
+                                        "properties": {
+                                            "name": "Data protection terms",
+                                            "description": "Data protection terms",
+                                        }
+                                    }
+                                ]
+                            },
+                            tags=[],
+                            last_updated=None,
+                        ),
+                        SearchResult(
+                            id="urn:li:glossaryTerm:022b9b68-c211-47ae-aef0-2db13acfeca8",
+                            result_type="GLOSSARY_TERM",
+                            name="Other term",
+                            description="Term description to test groupings work",
+                            matches={},
+                            metadata={
+                                "parentNodes": [
+                                    {
+                                        "properties": {
+                                            "name": "Data protection terms",
+                                            "description": "Data protection terms",
+                                        }
+                                    }
+                                ]
+                            },
+                            tags=[],
+                            last_updated=None,
+                        ),
+                    ],
+                    "description": "Data protection terms",
+                },
+                {
+                    "name": "Unsorted",
+                    "members": [
+                        SearchResult(
+                            id="urn:li:glossaryTerm:0eb7af28-62b4-4149-a6fa-72a8f1fea1e6",
+                            result_type="GLOSSARY_TERM",
+                            name="Security classification",
+                            description="Only data that is 'official'",
+                            matches={},
+                            metadata={"parentNodes": []},
+                            tags=[],
+                            last_updated=None,
+                        )
+                    ],
+                    "description": "",
+                },
+            ]
+        }
+
+        assert expected_context == glossary_context.context
+
+
 @pytest.mark.parametrize(
     "domain, expected_subdomains",
     [

tests/test_views.py

@@ -54,3 +54,8 @@ def test_details_data_product_not_found(self, client, mock_catalogue):
             )
         )
         assert response.status_code == 404
+
+class TestGlossaryView:
+    def test_details(self, client):
+        response = client.get(reverse("home:glossary"))
+        assert response.status_code == 200