Make sample data generation a separate lein task

.gitignore

@@ -22,3 +22,4 @@ profiles.clj
 /node_modules/
 /.babel_cache
 /coverage
+/resources/sample-dataset-*

project.clj

@@ -5,7 +5,8 @@
   :description "Metabase Community Edition"
   :url "http://metabase.com/"
   :min-lein-version "2.5.0"
-  :aliases {"test" ["with-profile" "+expectations" "expectations"]}
+  :aliases {"test" ["with-profile" "+expectations" "expectations"]
+            "generate-sample-dataset" ["with-profile" "+generate-sample-dataset" "run"]}
   :dependencies [[org.clojure/clojure "1.7.0"]
                  [org.clojure/core.logic "0.8.10"]
                  [org.clojure/core.match "0.3.0-alpha4"]              ; optimized pattern matching library for Clojure
@@ -30,9 +31,7 @@
                  [com.novemberain/monger "3.0.0"]                     ; MongoDB Driver
                  [compojure "1.4.0"]                                  ; HTTP Routing library built on Ring
                  [environ "1.0.0"]                                    ; easy environment management
-                 [faker "0.2.2"]                                      ; Fake data generator -- port of Perl/Ruby
                  [hiccup "1.0.5"]                                     ; HTML templating
-                 [incanter/incanter-core "1.5.6"]                     ; Satistical functions like normal distibutions
                  [korma "0.4.2"]                                      ; SQL lib
                  [log4j/log4j "1.2.17"
                   :exclusions [javax.mail/mail
@@ -92,4 +91,9 @@
                                        "-Dmb.api.key=test-api-key"
                                        "-Xverify:none"]}              ; disable bytecode verification when running tests so they start slightly faster
              :uberjar {:aot :all
-                       :prep-tasks ^:replace ["npm" "webpack" "javac" "compile"]}})
+                       :prep-tasks ^:replace ["npm" "webpack" "generate-sample-dataset" "javac" "compile"]}
+             :generate-sample-dataset {:dependencies [[faker "0.2.2"]                   ; Fake data generator -- port of Perl/Ruby
+                                                      [incanter/incanter-core "1.5.6"]] ; Satistical functions like normal distibutions}})
+                                       :source-paths ["sample_dataset"]
+                                       :global-vars {*warn-on-reflection* false}
+                                       :main ^:skip-aot metabase.sample-dataset.generate}})

src/metabase/sample_data/generate.clj → sample_dataset/metabase/sample_dataset/generate.clj

-(ns metabase.sample-data.generate
+(ns metabase.sample-dataset.generate
   (:require [clojure.math.numeric-tower :as math]
             [clojure.string :as s]
             (faker [address :as address]
@@ -11,6 +11,11 @@
                    [db :as kdb]))
   (:import java.util.Date))
 
+(def ^:private ^:const sample-dataset-version "1.0.0")
+
+(def ^:private ^:const sample-dataset-filename
+  (str (System/getProperty "user.dir") (format "/resources/sample-dataset-%s.db" sample-dataset-version)))
+
 (defn- normal-distribution-rand [mean median]
   (dist/draw (dist/normal-distribution mean median)))
 
@@ -345,7 +350,7 @@
 
      ;; Insert the data
      (println "Inserting data...")
-     (metabase.util/pdoseq [[table rows] (seq data)]
+     (doseq [[table rows] (seq data)]
        (assert (keyword? table))
        (assert (sequential? rows))
        (let [entity (-> (k/create-entity (s/upper-case (name table)))
@@ -362,3 +367,8 @@
        (k/exec-raw db (format "GRANT SELECT ON %s TO GUEST;" (s/upper-case (name table)))))
 
      (println "Done."))))
+
+(defn -main [& [filename]]
+  (let [filename (or filename sample-dataset-filename)]
+    (println (format "Writing sample dataset to %s..." filename))
+    (create-h2-db filename)))