Creates image proxy view (closes #104). (#142)

* Creates image proxy view (closes #104).

* Passes .env to environment when creating locally.

* Use image proxy URLs for served images.

conf/web.sh

@@ -1,3 +1,3 @@
 #!/usr/bin/env bash
 
-exec uwsgi --http :${PORT:-8000} --wsgi-file /app/recommendation/wsgi.py --master
+exec uwsgi --wsgi-disable-file-wrapper --http :${PORT:-8000} --wsgi-file /app/recommendation/wsgi.py --master

recommendation/conf.py

@@ -5,9 +5,11 @@
 TESTING = env.get('RECOMMENDATION_TESTING', None) == 'true'
 
 KEY_PREFIX = env.get('RECOMMENDATION_KEY_PREFIX', 'query_')
+SERVER_NAME = env.get('RECOMMENDATION_SERVER_NAME', 'universal-search.dev')
 
 CACHE_TTL = int(env.get('RECOMMENDATION_CACHE_TTL', 7 * 24 * 60 * 60))
 MEMCACHED_TTL = int(env.get('RECOMMENDATION_MEMCACHED_TTL', CACHE_TTL))
+IMAGEPROXY_TTL = int(env.get('IMAGEPROXY_TTL', CACHE_TTL))
 
 BING_ACCOUNT_KEY = env.get('BING_ACCOUNT_KEY', '')
 EMBEDLY_API_KEY = env.get('EMBEDLY_API_KEY', '')

recommendation/factory.py

@@ -10,6 +10,7 @@
 from recommendation.mozlog.middleware import request_timer, request_summary
 from recommendation.views.debug import debug
 from recommendation.views.dummy import dummy
+from recommendation.views.images import images
 from recommendation.views.main import main
 from recommendation.views.static import static
 from recommendation.views.status import status
@@ -24,6 +25,7 @@ def create_app():
     # Register views.
     app.register_blueprint(main)
     app.register_blueprint(debug)
+    app.register_blueprint(images)
     app.register_blueprint(static)
     app.register_blueprint(status)
 
@@ -45,7 +47,8 @@ def create_app():
 
     app.config.update(
         CELERY_BROKER_URL=conf.CELERY_BROKER_URL,
-        DEBUG=conf.DEBUG
+        DEBUG=conf.DEBUG,
+        SERVER_NAME=conf.SERVER_NAME
     )
     return app
 

recommendation/mozlog/middleware.py

@@ -12,7 +12,8 @@
     '/__heartbeat__',
     '/__lbheartbeat__',
     '/nginx_status',
-    '/robots.txt'
+    '/robots.txt',
+    '/images'
 ]
 
 
@@ -42,20 +43,20 @@ def request_summary(response):
 
     log = {}
     query = request.args.get('q')
-    data = response.get_data(as_text=True)
-    try:
-        body = json.loads(data)
-    except json.decoder.JSONDecodeError:
-        body = {}
 
     log['agent'] = request.headers.get('User-Agent')
     log['errno'] = 0 if response.status_code < 400 else response.status_code
     log['lang'] = request.headers.get('Accept-Language')
     log['method'] = request.method
     log['path'] = request.path
-    log['t'] = (request.finish_time - request.start_time) * 1000 # in ms
+    log['t'] = (request.finish_time - request.start_time) * 1000  # in ms
 
     if query:
+        data = response.get_data(as_text=True)
+        try:
+            body = json.loads(data)
+        except json.decoder.JSONDecodeError:
+            body = {}
         query = query.lower()
         log['predicates.query_length'] = len(query) > 20
         log['predicates.is_protocol'] = (re.match(IS_PROTOCOL, query) is not

recommendation/search/classification/embedly.py

@@ -5,6 +5,7 @@
 from recommendation import conf
 from recommendation.memorize import memorize
 from recommendation.search.classification.base import BaseClassifier
+from recommendation.util import image_url
 
 
 class BaseEmbedlyClassifier(BaseClassifier):
@@ -54,7 +55,7 @@ def enhance(self):
             return {}
         return {
             'color': self._get_color(api_data),
-            'url': favicon_url,
+            'url': image_url(favicon_url, width=32, height=32),
         }
 
 
@@ -128,8 +129,10 @@ def enhance(self):
         try:
             image_data = self._get_image(api_data)
             image = {k: image_data.get(k) for k in ['url', 'height', 'width']}
+            image['url'] = image_url(image['url'])
         except (KeyError, IndexError):
             image = {}
+
         return {
             'image': image,
             'title': self._get_title(api_data),

recommendation/search/classification/movies.py

@@ -5,6 +5,7 @@
 
 from recommendation.memorize import memorize
 from recommendation.search.classification.base import BaseClassifier
+from recommendation.util import image_url
 
 
 class MovieClassifier(BaseClassifier):
@@ -93,7 +94,7 @@ def enhance(self):
             'title': data.get('Title'),
             'year': data.get('Year'),
             'plot': data.get('Plot'),
-            'poster': data.get('Poster'),
+            'poster': image_url(data.get('Poster')),
             'rating': {
                 'imdb': self._score(data.get('imdbRating'), 10),
                 'metacritic': self._score(data.get('Metascore'), 100)

recommendation/search/classification/tests/test_embedly.py

@@ -1,4 +1,3 @@
-from unittest import TestCase
 from unittest.mock import patch
 from urllib.parse import parse_qs, urlparse
 
@@ -8,6 +7,8 @@
 from recommendation.search.classification.embedly import (
     BaseEmbedlyClassifier, FaviconClassifier, WikipediaClassifier)
 from recommendation.tests.memcached import mock_memcached
+from recommendation.tests.util import AppTestCase
+from recommendation.util import image_url
 
 
 MOCK_API_KEY = '0123456789abcdef'
@@ -80,7 +81,7 @@
 }
 
 
-class TestBaseEmbedlyClassifier(TestCase):
+class TestBaseEmbedlyClassifier(AppTestCase):
     classifier_class = BaseEmbedlyClassifier
 
     def tearDown(self):
@@ -149,7 +150,8 @@ def test_enhance(self, mock_api_url):
                       status=200)
         enhanced = self._classifier(MOCK_RESULT_URL).enhance()
         eq_(enhanced['color'], MOCK_RESPONSE['favicon_colors'][0]['color'])
-        eq_(enhanced['url'], MOCK_RESPONSE['favicon_url'])
+        eq_(enhanced['url'], image_url(
+            MOCK_RESPONSE['favicon_url'], width=32, height=32))
 
     @patch('recommendation.search.classification.embedly.FaviconClassifier'
            '._api_response')
@@ -223,6 +225,8 @@ def test_enhance(self, mock_api_url):
         responses.add(responses.GET, MOCK_API_URL, json=MOCK_RESPONSE,
                       status=200)
         enhanced = self._classifier(MOCK_WIKIPEDIA_URL).enhance()
+        MOCK_WIKIPEDIA_RESPONSE['image']['url'] = (
+            image_url(MOCK_WIKIPEDIA_RESPONSE['image']['url']))
         eq_(enhanced, MOCK_WIKIPEDIA_RESPONSE)
 
     @patch('recommendation.search.classification.embedly.WikipediaClassifier'

recommendation/search/classification/tests/test_movies.py

@@ -1,5 +1,4 @@
 from copy import copy
-from unittest import TestCase
 from unittest.mock import patch
 from urllib.parse import parse_qs, urlparse
 
@@ -8,6 +7,8 @@
 
 from recommendation.search.classification.movies import MovieClassifier
 from recommendation.tests.memcached import mock_memcached
+from recommendation.tests.util import AppTestCase
+from recommendation.util import image_url
 
 
 IMDB_ID = 'tt0116756'
@@ -54,7 +55,7 @@
 }
 
 
-class TestMovieClassifier(TestCase):
+class TestMovieClassifier(AppTestCase):
     def setUp(self):
         self.classifier = MovieClassifier(RESULT_IMDB, [])
 
@@ -125,7 +126,7 @@ def test_enhance(self, mock_api_response):
         eq_(enhanced['title'], MOCK_RESPONSE['Title'])
         eq_(enhanced['year'], MOCK_RESPONSE['Year'])
         eq_(enhanced['plot'], MOCK_RESPONSE['Plot'])
-        eq_(enhanced['poster'], MOCK_RESPONSE['Poster'])
+        eq_(enhanced['poster'], image_url(MOCK_RESPONSE['Poster']))
         eq_(enhanced['rating']['imdb']['stars'], 1.4)
         eq_(enhanced['rating']['imdb']['raw'], 2.8)
         eq_(enhanced['rating']['metacritic']['stars'], 1.2)

recommendation/search/classification/tests/test_tld.py

@@ -1,19 +1,20 @@
-from unittest import TestCase
 from unittest.mock import patch
 from urllib.parse import ParseResult
 
 import responses
 from nose.tools import eq_, ok_
 
 from recommendation.search.classification.tld import TLDClassifier
+from recommendation.tests.util import AppTestCase
+from recommendation.util import image_url
 
 
 DOMAIN = 'www.mozilla.com'
 URL = 'http://%s/' % DOMAIN
 LOGO = 'https://logo.clearbit.com/%s' % DOMAIN
 
 
-class TestTLDClassifier(TestCase):
+class TestTLDClassifier(AppTestCase):
     def _result(self, url):
         return {
             'url': url
@@ -71,4 +72,4 @@ def test_enhance(self, mock_logo_exists):
         mock_logo_exists.return_value = False
         eq_(self._enhance(URL), None)
         mock_logo_exists.return_value = True
-        eq_(self._enhance(URL), LOGO)
+        eq_(self._enhance(URL), image_url(LOGO, width=64, height=64))

recommendation/search/classification/tld.py

@@ -1,6 +1,7 @@
 import requests
 
 from recommendation.search.classification.base import BaseClassifier
+from recommendation.util import image_url
 
 
 class TLDClassifier(BaseClassifier):
@@ -24,4 +25,4 @@ def enhance(self):
         logo = self._get_logo()
         if not self._logo_exists(logo):
             return None
-        return logo
+        return image_url(logo, width=64, height=64)

recommendation/search/tests/test_recommendation.py

@@ -1,4 +1,3 @@
-from unittest import TestCase
 from unittest.mock import patch
 
 from nose.tools import eq_, ok_
@@ -16,6 +15,7 @@
 from recommendation.search.query.tests.test_yahoo import (
     QUERY as YAHOO_QUERY, MOCK_RESPONSE as YAHOO_RESPONSE)
 from recommendation.tests.memcached import mock_memcached
+from recommendation.tests.util import AppTestCase
 
 
 QUERY = 'Cubs'
@@ -25,7 +25,7 @@
 SUGGESTIONS = ['a', 'b', 'c']
 
 
-class TestSearchRecommendation(TestCase):
+class TestSearchRecommendation(AppTestCase):
     def setUp(self):
         self.instance = SearchRecommendation('')
 

recommendation/tests/test_util.py

@@ -0,0 +1,47 @@
+from urllib.parse import parse_qs, quote, urlparse
+
+from flask import current_app
+from nose.tools import eq_
+
+from recommendation.tests.util import AppTestCase
+from recommendation.util import image_url
+
+
+DIMENSION = '64'
+IMAGE = 'https://foo.bar/image.jpg'
+EMBEDLY_BASE = 'https://i.embed.ly/'
+EMBEDLY_IMAGE = '{}?url={}'.format(EMBEDLY_BASE, quote(IMAGE))
+
+
+class TestImageUrl(AppTestCase):
+    def _image_url(self, url, **kwargs):
+        with current_app.app_context():
+            url = image_url(url, **kwargs)
+            parsed = urlparse(url) if url else None
+            qs = parse_qs(parsed.query) if parsed else None
+            return url, parsed, qs
+
+    def test_none(self):
+        url, parsed, qs = self._image_url(None)
+        eq_(url, None)
+
+    def test_formed(self):
+        url, parsed, qs = self._image_url(IMAGE, width=DIMENSION,
+                                          height=DIMENSION)
+        eq_(IMAGE, qs['url'][0])
+        eq_(DIMENSION, qs['width'][0])
+        eq_(DIMENSION, qs['height'][0])
+
+    def test_embedly(self):
+        url = self._image_url(IMAGE)
+        embedly_url = self._image_url(EMBEDLY_IMAGE)
+        eq_(url, embedly_url)
+
+    def test_embedly_no_url(self):
+        url, parsed, qs = self._image_url(EMBEDLY_BASE)
+        eq_(qs['url'][0], EMBEDLY_BASE)
+
+    def test_embedly_empty_url(self):
+        URL = '{}?url='.format(EMBEDLY_BASE)
+        url, parsed, qs = self._image_url(URL)
+        eq_(qs['url'][0], URL)

recommendation/util.py

@@ -0,0 +1,22 @@
+from urllib.parse import parse_qs, urlparse
+
+from flask import current_app, url_for
+
+
+def image_url(url, **kwargs):
+    if not url:
+        return
+    kwargs['url'] = url
+    parsed = urlparse(url)
+
+    # If the image is already being proxied by Embedly, pull the `url`
+    # querystring param out and use that instead to prevent double-billing.
+    if parsed.netloc == 'i.embed.ly':
+        qs = parse_qs(parsed.query)
+        try:
+            kwargs['url'] = qs['url'][0]
+        except (IndexError, KeyError):
+            pass
+
+    with current_app.app_context():
+        return url_for('images.proxy', **kwargs)

recommendation/views/images.py

@@ -0,0 +1,44 @@
+from urllib.parse import urlencode
+
+import requests
+from flask import abort, Blueprint, request, Response, stream_with_context
+
+from recommendation import conf
+
+EMBEDLY_RESIZE = 'https://i.embed.ly/1/display/resize'
+
+
+images = Blueprint('images', __name__)
+
+
+def make_embedly_url(url, **kwargs):
+    """
+    Passed the URL to an image, returns a string to the Embedly resize URL for
+    that image. Accepts optional `width` and `height` keyword arguments.
+    """
+    qs = {}
+    for param in ['width', 'height']:
+        if param in kwargs:
+            qs[param] = kwargs[param][0]
+    qs['animate'] = 'false'
+    qs['compresspng'] = 'true'
+    qs['key'] = conf.EMBEDLY_API_KEY
+    return '{}?{}'.format(EMBEDLY_RESIZE, urlencode(qs))
+
+
+@images.route('/images')
+def proxy():
+    try:
+        url = make_embedly_url(**request.args)
+    except TypeError:
+        abort(400)
+    try:
+        req = requests.get(url, stream=True, timeout=10)
+    except requests.RequestException:
+        abort(400)
+    if req.status_code != 200:
+        abort(400)
+    response = Response(stream_with_context(req.iter_content()),
+                        content_type=req.headers['content-type'])
+    response.headers['Cache-Control'] = 'max-age=%d' % conf.IMAGEPROXY_TTL
+    return response