[NTP Snippets] Fetcher: only download snippets that have a thumbnail

components/ntp_snippets/ntp_snippets_fetcher.cc

Index: components/ntp_snippets/ntp_snippets_fetcher.cc
diff --git a/components/ntp_snippets/ntp_snippets_fetcher.cc b/components/ntp_snippets/ntp_snippets_fetcher.cc
index b0f81d16fb41e4a14d702d4420cb18d21822fab6..3a34ac9cd21c34151c76f7d4c2dbb1a65d61c54e 100644
--- a/components/ntp_snippets/ntp_snippets_fetcher.cc
+++ b/components/ntp_snippets/ntp_snippets_fetcher.cc
@@ -38,6 +38,10 @@ const char kRequestParameterFormat[] =
     "    \"local_scoring_params\": {"
     "      \"content_params\": {"
     "        \"only_return_personalized_results\": false"
+    "      },"
+    "      \"content_restricts\": {"
+    "        \"type\": \"METADATA\","
+    "        \"value\": \"ALL\""

[May, 2016/04/06 13:39:20]

Does ALL mean title, snippet, thumbnail, URL? Or would it also remove snippets
that don't have a favicon or canonical publisher name. A lot of the data I've
seen returned from ChromeReader don't have these two (though I think from
Patrick's previous talks with Newsstand we can't use that data anyway).

[Marc Treib, 2016/04/06 13:52:54]

ALL means title, snippet, and thumbnail. (Presumably URL is always there
anyway). Those are all the possible values; I guess if new metadata restricts
are added in the future, then those would also be applied. So maybe it would be
better to explicitly specify title, snippet, and thumbnail - WDYT?

[May, 2016/04/07 16:05:12]

Yeah, I think so. It would insulate us better from changes that ChromeReader
makes to their API.

[Marc Treib, 2016/04/07 16:15:52]

Done.

     "      }"
     "%s"
     "    },"