From 891776f9c073a789c3392564c6a0143ff0bb34c3 Mon Sep 17 00:00:00 2001
From: Valentin Deniaud <vdeniaud@entrouvert.com>
Date: Wed, 3 Nov 2021 17:44:01 +0100
Subject: [PATCH] search: index external links data independantly (#58269)

---
 combo/apps/search/utils.py | 32 +++++++++++---------------------
 tests/test_search.py       | 29 +++++++++++++++++++++++++++++
 2 files changed, 40 insertions(+), 21 deletions(-)

diff --git a/combo/apps/search/utils.py b/combo/apps/search/utils.py
index faa7783c..2a2f22ee 100644
--- a/combo/apps/search/utils.py
+++ b/combo/apps/search/utils.py
@@ -48,7 +48,6 @@ def index_site():
     # populate ContentType cache
     ContentType.objects.get_for_models(*cell_classes)
     IndexedCell.objects.all().delete()
-    external_urls = {}
     validity_info_list = list(ValidityInfo.objects.select_related('content_type'))
     pages_by_pk = {
         p.pk: p for p in (Page.objects.prefetch_related(Prefetch('groups', to_attr='prefetched_groups')))
@@ -90,26 +89,17 @@ def index_site():
 
             for link_data in cell.get_external_links_data():
                 # index external links
-                indexed_cell = external_urls.get(link_data.get('url'))
-                if indexed_cell is None:
-                    # create an entry for that link.
-                    indexed_cell = IndexedCell(
-                        cell_type=cell_type,
-                        cell_pk=cell.id,
-                        page_id=cell.page_id,
-                        public_access=bool(cell.page.public and cell.public),
-                        url=link_data['url'],
-                        title=link_data['title'],
-                        indexed_text=link_data.get('text') or '',
-                    )
-                    indexed_cell.save()
-                    set_cell_groups(indexed_cell, cell)
-                    external_urls[indexed_cell.url] = indexed_cell
-                else:
-                    # if that link already exists, add detailed texts
-                    indexed_cell.indexed_text += ' ' + link_data['title']
-                    indexed_cell.indexed_text += ' ' + (link_data.get('text') or '')
-                    indexed_cell.save()
+                indexed_cell = IndexedCell(
+                    cell_type=cell_type,
+                    cell_pk=cell.id,
+                    page_id=cell.page_id,
+                    public_access=bool(cell.page.public and cell.public),
+                    url=link_data['url'],
+                    title=link_data['title'],
+                    indexed_text=link_data.get('text') or '',
+                )
+                indexed_cell.save()
+                set_cell_groups(indexed_cell, cell)
 
 
 def search_site(request, query, pages=None, with_description=None):
diff --git a/tests/test_search.py b/tests/test_search.py
index 3cd0fc4c..e8d8b2f5 100644
--- a/tests/test_search.py
+++ b/tests/test_search.py
@@ -1430,3 +1430,32 @@ def test_search_by_page_title(app):
     assert hits[0]['text'] == page_of_interest.title
     assert hits[0]['url'] == f'/{page_of_interest.slug}/'
     assert hits[0]['rank'] > hits[1]['rank']
+
+
+def test_search_same_link_multiple_pages(settings, app):
+    settings.KNOWN_SERVICES = {}
+
+    page = Page.objects.create(title='first page', slug='one')
+    LinkCell.objects.create(
+        title='foobar', url='http://example.net', page=page, placeholder='content', order=0
+    )
+    second_page = Page.objects.create(title='second page', slug='two')
+    LinkCell.objects.create(
+        title='barfoo', url='http://example.net', page=second_page, placeholder='content', order=0
+    )
+    index_site()
+
+    cell = SearchCell.objects.create(
+        page=page, placeholder='content', _search_services={'data': ['_text_page_one']}, order=1
+    )
+    resp = app.get('/ajax/search/%s/_text_page_one/?q=foobar' % cell.pk, status=200)
+    assert resp.text.count('<li') == 1
+    resp = app.get('/ajax/search/%s/_text_page_one/?q=barfoo' % cell.pk, status=200)
+    assert resp.text.count('<li') == 0
+
+    cell._search_services = {'data': ['_text_page_two']}
+    cell.save()
+    resp = app.get('/ajax/search/%s/_text_page_two/?q=foobar' % cell.pk, status=200)
+    assert resp.text.count('<li') == 0
+    resp = app.get('/ajax/search/%s/_text_page_two/?q=barfoo' % cell.pk, status=200)
+    assert resp.text.count('<li') == 1
-- 
2.30.2