Add primary_genre to avoid duplication of podcasts across category sh…

…owcases
eemp · Mar 2, 2019 · 85d9a1a · 85d9a1a
1 parent 3d34293
commit 85d9a1a
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 0 deletions.
diff --git a/packages/podcasts-fetcher/etc/mappings.json b/packages/podcasts-fetcher/etc/mappings.json
@@ -58,6 +58,19 @@
       "popularity": {
         "type": "long"
       },
+      "primary_genre": {
+        "type": "object",
+        "properties": {
+          "id": {
+            "type": "keyword",
+            "ignore_above": 256
+          },
+          "name": {
+            "type": "keyword",
+            "ignore_above": 256
+          }
+        }
+      },
       "release_date": {
         "type": "date"
       }

diff --git a/packages/podcasts-fetcher/index-itunes-data.js b/packages/podcasts-fetcher/index-itunes-data.js
@@ -73,6 +73,10 @@ function getRecordFromResult(scanOffset, result, idx) {
     id: result.trackId,
     name: result.trackName,
     popularity: scanOffset + idx,
+    primary_genre: {
+      id: _.get(result, 'genreIds[0]'),
+      name: _.get(result, 'genres[0]',
+    },
     release_date: result.releaseDate,
     type: PODCAST_MEDIA,
   };

diff --git a/packages/podcasts-fetcher/scripts/reindex-docs.sh b/packages/podcasts-fetcher/scripts/reindex-docs.sh
@@ -4,3 +4,4 @@
 [ -z "$ES_INDEX" ] && echo "Need to set ENV.ES_INDEX" && exit 1
 
 curl -XPOST $ES_SERVER/$ES_INDEX/podcast/_update_by_query?conflicts=proceed -H 'Content-Type: application/json' -d '{ "query": { "match_all": {} } }'
+#curl -XPOST $ES_SERVER/$ES_INDEX/podcast/_update_by_query?conflicts=proceed -H 'Content-Type: application/json' -d '{ "query": { "match_all": {} }, "script": { "source": "ctx._source['"'"'primary_genre'"'"'] = ctx._source['"'"'genres'"'"'][0]" } }'