From 077a17b093d9a6331e14e5ef4b2234fd268df6f6 Mon Sep 17 00:00:00 2001
From: Eliot Berriot <contact@eliotberriot.com>
Date: Thu, 12 Apr 2018 18:41:43 +0200
Subject: [PATCH] Scan task can now handle until

---
 api/funkwhale_api/federation/tasks.py | 27 ++++++++++++---
 api/tests/federation/test_tasks.py    | 49 +++++++++++++++++++++++++--
 2 files changed, 69 insertions(+), 7 deletions(-)

diff --git a/api/funkwhale_api/federation/tasks.py b/api/funkwhale_api/federation/tasks.py
index fd03d3290..03124d16b 100644
--- a/api/funkwhale_api/federation/tasks.py
+++ b/api/funkwhale_api/federation/tasks.py
@@ -1,27 +1,44 @@
+from requests.exceptions import RequestException
+
 from funkwhale_api.taskapp import celery
 
 from . import library as lb
 from . import models
 
 
-@celery.app.task(name='federation.scan_library')
+@celery.app.task(
+    name='federation.scan_library',
+    autoretry_for=[RequestException],
+    retry_backoff=30,
+    max_retries=5)
 @celery.require_instance(models.Library, 'library')
-def scan_library(library):
+def scan_library(library, until=None):
     if not library.federation_enabled:
         return
 
     data = lb.get_library_data(library.url)
     scan_library_page.delay(
-        library_id=library.id, page_url=data['first'])
+        library_id=library.id, page_url=data['first'], until=until)
 
 
-@celery.app.task(name='federation.scan_library_page')
+@celery.app.task(
+    name='federation.scan_library_page',
+    autoretry_for=[RequestException],
+    retry_backoff=30,
+    max_retries=5)
 @celery.require_instance(models.Library, 'library')
-def scan_library_page(library, page_url):
+def scan_library_page(library, page_url, until=None):
     if not library.federation_enabled:
         return
 
     data = lb.get_library_page(library, page_url)
     lts = []
     for item_serializer in data['items']:
+        item_date = item_serializer.validated_data['published']
+        if until and item_date < until:
+            return
         lts.append(item_serializer.save())
+
+    next_page = data.get('next')
+    if next_page and next_page != page_url:
+        scan_library_page.delay(library_id=library.id, page_url=next_page)
diff --git a/api/tests/federation/test_tasks.py b/api/tests/federation/test_tasks.py
index 85684c1a3..80f2365b6 100644
--- a/api/tests/federation/test_tasks.py
+++ b/api/tests/federation/test_tasks.py
@@ -37,6 +37,7 @@ def test_scan_library_fetches_page_and_calls_scan_page(
     scan_page.assert_called_once_with(
         library_id=library.id,
         page_url=collection.data['first'],
+        until=None,
     )
 
 
@@ -51,11 +52,55 @@ def test_scan_page_fetches_page_and_creates_tracks(
         'item_serializer': serializers.AudioSerializer,
     }
     page = serializers.CollectionPageSerializer(page_conf)
-    #scan_page = mocker.patch(
-    #    'funkwhale_api.federation.tasks.scan_library_page.delay')
     r_mock.get(page.data['id'], json=page.data)
 
     tasks.scan_library_page(library_id=library.pk, page_url=page.data['id'])
 
     lts = list(library.tracks.all().order_by('-published_date'))
     assert len(lts) == 5
+
+
+def test_scan_page_trigger_next_page_scan_skip_if_same(
+        mocker, factories, r_mock):
+    patched_scan = mocker.patch(
+        'funkwhale_api.federation.tasks.scan_library_page.delay'
+    )
+    library = factories['federation.Library'](federation_enabled=True)
+    tfs = factories['music.TrackFile'].create_batch(size=1)
+    page_conf = {
+        'actor': library.actor,
+        'id': library.url,
+        'page': Paginator(tfs, 3).page(1),
+        'item_serializer': serializers.AudioSerializer,
+    }
+    page = serializers.CollectionPageSerializer(page_conf)
+    data = page.data
+    data['next'] = data['id']
+    r_mock.get(page.data['id'], json=data)
+
+    tasks.scan_library_page(library_id=library.pk, page_url=data['id'])
+    patched_scan.assert_not_called()
+
+
+def test_scan_page_stops_once_until_is_reached(
+        mocker, factories, r_mock):
+    library = factories['federation.Library'](federation_enabled=True)
+    tfs = list(reversed(factories['music.TrackFile'].create_batch(size=5)))
+    page_conf = {
+        'actor': library.actor,
+        'id': library.url,
+        'page': Paginator(tfs, 3).page(1),
+        'item_serializer': serializers.AudioSerializer,
+    }
+    page = serializers.CollectionPageSerializer(page_conf)
+    r_mock.get(page.data['id'], json=page.data)
+
+    tasks.scan_library_page(
+        library_id=library.pk,
+        page_url=page.data['id'],
+        until=tfs[1].creation_date)
+
+    lts = list(library.tracks.all().order_by('-published_date'))
+    assert len(lts) == 2
+    for i, tf in enumerate(tfs[:1]):
+        assert tf.creation_date == lts[i].published_date