Clean family before re-run its analysis

Dockerfile

@@ -3,5 +3,5 @@ FROM diogok/java8:zulu
 WORKDIR /opt
 CMD ["java","-server","-XX:+UseConcMarkSweepGC","-XX:+UseCompressedOops","-XX:+DoEscapeAnalysis","-jar","rrapp-idx.jar"]
 
-ADD target/rrapp-idx-0.0.3-standalone.jar /opt/rrapp-idx.jar
+ADD target/rrapp-idx-0.0.4-standalone.jar /opt/rrapp-idx.jar
 

project.clj

@@ -1,4 +1,4 @@
-(defproject rrapp-idx "0.0.3"
+(defproject rrapp-idx "0.0.4"
   :description "Run analysis on DarwinCore taxa and occurrences aggregated"
   :url "http://github.com/biodivdev/rrapp-idx"
   :license {:name "MIT"}

src/rrapp_idx/core.clj

@@ -17,12 +17,14 @@
 (defn run-all
   [] 
   (let [families    (chan 2)
+        families-d  (chan 2)
         species     (chan 5)
         names       (chan 5)
         occurrences (chan 2)
         results     (chan 5)]
 
-  (pipe families get-species species)
+  (pipe families clean-family families-d)
+  (pipe families-d get-species species)
   (pipe species get-names names)
   (pipe names get-occurrences occurrences)
   (pipe occurrences get-results results)

src/rrapp_idx/db.clj

@@ -70,6 +70,15 @@
     :hits
     (map :_source)))
 
+(defn clean-family
+  [family out] 
+  (go
+    (log/info "Clean species of" family)
+    (post-json (str (config :elasticsearch) "/" (config :index) "/analysis/_delete_by_query")
+      {:query {:term {:family.keyword family}}})
+    (>! out family)
+    (close! out)))
+
 (defn get-species
   [family out] 
   (go