Merge pull request cc-archive#488 from creativecommons/outloudvi-proxy

Outloudvi thumbnail proxy

cccatalog-api/cccatalog/api/controllers/search_controller.py

@@ -6,10 +6,11 @@
 from elasticsearch_dsl.query import Query
 from cccatalog import settings
 from django.core.cache import cache
+from django.urls import reverse
 import cccatalog.api.models as models
 import logging as log
 from rest_framework import serializers
-from cccatalog.settings import THUMBNAIL_PROXY_URL, PROXY_THUMBS
+from cccatalog.settings import PROXY_THUMBS
 from cccatalog.api.utils.validate_images import validate_images
 from cccatalog.api.utils.dead_link_mask import get_query_mask, get_query_hash
 from itertools import accumulate
@@ -21,7 +22,6 @@
 DEAD_LINK_RATIO = 1 / 2
 THUMBNAIL = 'thumbnail'
 URL = 'url'
-THUMBNAIL_WIDTH_PX = 600
 PROVIDER = 'provider'
 DEEP_PAGINATION_ERROR = 'Deep pagination is not allowed.'
 QUERY_SPECIAL_CHARACTER_ERROR = 'Unescaped special characters are not allowed.'
@@ -128,10 +128,12 @@ def _post_process_results(s, start, end, page_size, search_results,
             else:
                 to_proxy = URL
             original = res[to_proxy]
-            proxied = '{proxy_url}/{width}/{original}'.format(
-                proxy_url=THUMBNAIL_PROXY_URL,
-                width=THUMBNAIL_WIDTH_PX,
-                original=original
+            ext = res["url"].split(".")[-1]
+            proxied = "http://{}{}".format(
+                request.get_host(),
+                reverse('thumbs', kwargs={
+                    'identifier': "{}.{}".format(res["identifier"], ext)
+                })
             )
             res[THUMBNAIL] = proxied
         results.append(res)

cccatalog-api/cccatalog/api/views/site_views.py

@@ -1,6 +1,7 @@
 import logging as log
 import secrets
 import smtplib
+from urllib.request import urlopen
 from django.core.mail import send_mail
 from rest_framework.response import Response
 from rest_framework.reverse import reverse
@@ -10,11 +11,13 @@
 from cccatalog.api.serializers.oauth2_serializers import\
     OAuth2RegistrationSerializer, OAuth2RegistrationSuccessful, OAuth2KeyInfo
 from drf_yasg.utils import swagger_auto_schema
-from cccatalog.api.models import ContentProvider
+from cccatalog.api.models import ContentProvider, Image
 from cccatalog.api.models import ThrottledApplication, OAuth2Verification
 from cccatalog.api.utils.throttle import TenPerDay, OnePerSecond
 from cccatalog.api.utils.oauth2_helper import get_token_info
+from cccatalog.settings import THUMBNAIL_PROXY_URL, THUMBNAIL_WIDTH_PX
 from django.core.cache import cache
+from django.http import HttpResponse
 
 IDENTIFIER = 'provider_identifier'
 NAME = 'provider_name'
@@ -285,3 +288,50 @@ def get(self, request, format=None):
             'verified': verified
         }
         return Response(status=200, data=response_data)
+
+
+class Thumbs(APIView):
+    """
+    Return the thumb of an image.
+    """
+
+    lookup_field = 'identifier'
+    queryset = Image.objects.all()
+
+    @swagger_auto_schema(operation_id="thumb_lookup",
+                         responses={
+                             200: 'The thumb of an image',
+                             400: 'Bad Request',
+                             404: 'Not Found'
+                         })
+    def get(self, request, identifier, format=None):
+        path_element = identifier.split(".")
+        identifier = path_element[0]
+        extname = ""
+        if len(path_element) == 2:
+            extname = path_element[1]
+        elif len(path_element) > 2:
+            return Response(status=400)
+        try:
+            image = Image.objects.get(identifier=identifier)
+            if extname and image.url.split(".")[-1] != extname:
+                return Response(status=404, data='Not Found')
+        except Image.DoesNotExist:
+            return Response(status=404, data='Not Found')
+
+        upstream_url = '{proxy_url}/{width}/{original}'.format(
+            proxy_url=THUMBNAIL_PROXY_URL,
+            width=THUMBNAIL_WIDTH_PX,
+            original=image.url
+        )
+        upstream_response = urlopen(upstream_url)
+        status = upstream_response.status
+        content_type = upstream_response.headers.get('Content-Type')
+
+        response = HttpResponse(
+            upstream_response.read(),
+            status=status,
+            content_type=content_type
+        )
+
+        return response

cccatalog-api/cccatalog/settings.py

@@ -156,11 +156,13 @@
 }
 
 # Produce CC-hosted thumbnails dynamically through a proxy.
-PROXY_THUMBS = bool(os.environ.get('PROXY_THUMBS', False))
+PROXY_THUMBS = bool(os.environ.get('PROXY_THUMBS', True))
 THUMBNAIL_PROXY_URL = os.environ.get(
-    'THUMBNAIL_PROXY_URL', 'https://localhost:8222'
+    'THUMBNAIL_PROXY_URL', 'http://localhost:8222'
 )
 
+THUMBNAIL_WIDTH_PX = 600
+
 AUTHENTICATION_BACKENDS = (
     'oauth2_provider.backends.OAuth2Backend',
     'django.contrib.auth.backends.ModelBackend',

cccatalog-api/cccatalog/urls.py

@@ -20,7 +20,7 @@
 from cccatalog.api.views.image_views import SearchImages, ImageDetail,\
     Watermark, RelatedImage, OembedView, ReportImageView
 from cccatalog.api.views.site_views import HealthCheck, ImageStats, Register, \
-    CheckRates, VerifyEmail
+    CheckRates, VerifyEmail, Thumbs
 from cccatalog.api.views.link_views import CreateShortenedLink, \
     ResolveShortenedLink
 from cccatalog.settings import API_VERSION, WATERMARK_ENABLED
@@ -110,6 +110,7 @@
     ),
     path('link', CreateShortenedLink.as_view(), name='make-link'),
     path('link/<str:path>', ResolveShortenedLink.as_view(), name='resolve'),
+    path('thumbs/<str:identifier>', Thumbs.as_view(), name='thumbs'),
     path('oembed', OembedView.as_view(), name='oembed')
 ]
 if WATERMARK_ENABLED:

cccatalog-api/test/v1_integration_test.py

@@ -108,6 +108,12 @@ def test_image_delete(search_fixture):
     deleted_response = requests.get(f'{API_URL}/image/{test_id}')
     assert deleted_response.status_code == 404
 
+@pytest.fixture
+def test_image_thumb(search_fixture):
+    thumbnail_url = search_fixture['results'][0]['thumbnail']
+    thumbnail_response = requests.get(thumbnail_url)
+    assert thumbnail_response.status_code == 200
+    assert thumbnail_response.headers["Content-Type"].startswith("image/")
 
 @pytest.fixture
 def link_shortener_fixture(search_fixture):

docker-compose.yml

@@ -74,6 +74,7 @@ services:
       ELASTICSEARCH_PORT: "9200"
       DISABLE_GLOBAL_THROTTLING: "True"
       ROOT_SHORTENING_URL: "localhost:8000"
+      THUMBNAIL_PROXY_URL: "http://thumbs:8222"
     stdin_open: true
     tty: true