Refactoring for processing feeds with data.zip

project.clj

@@ -3,4 +3,5 @@
   :url "http://github.com/eprunier/ariane"
   :license {:name "Eclipse Public License"
             :url "http://www.eclipse.org/legal/epl-v10.html"}
-  :dependencies [[org.clojure/clojure "1.5.1"]])
+  :dependencies [[org.clojure/clojure "1.5.1"]
+                 [org.clojure/data.zip "0.1.1"]])

src/ariane/atom.clj

@@ -1,16 +1,54 @@
 (ns ariane.atom
-  "Namespace for parsing Atom feeds.")
+  "Namespace for parsing Atom feeds."
+  (:require [clojure.zip :as zip]
+            [clojure.data.zip.xml :as zip-xml]))
+
+(defn- add-map-entry
+  "Add key/value entry to map only if value is not nil."
+  [to [key value]]
+  (if value
+    (assoc to key value)
+    to))
+
+(defn- links
+  [links]
+  (vec (for [link links]
+         (let [link-spec {:href 
+                          (first (zip-xml/xml-> link (zip-xml/attr :href)))}]
+           (if-let [rel (first (zip-xml/xml-> link (zip-xml/attr :rel)))]
+             (assoc link-spec :rel rel)
+             link-spec)))))
+
+(defn- infos
+  [root]
+  (-> {}
+      (add-map-entry [:title (first (zip-xml/xml-> root :title zip-xml/text))])
+      (add-map-entry [:description (first (zip-xml/xml-> root :summary zip-xml/text))])
+      (add-map-entry [:updated (first (zip-xml/xml-> root :updated zip-xml/text))])
+      (add-map-entry [:links (links (zip-xml/xml-> root :link))])))
+
+(defn- authors
+  [authors]
+  (vec (for [author authors]
+         {:name (first (zip-xml/xml-> author :name zip-xml/text))
+          :uri (first (zip-xml/xml-> author :uri zip-xml/text))})))
+
+(defn- entries 
+  [root]
+  (for [entry (zip-xml/xml-> root :entry)]
+    (-> {}
+        (add-map-entry [:id (first (zip-xml/xml-> entry :id zip-xml/text))])
+        (add-map-entry [:title (first (zip-xml/xml-> entry :title zip-xml/text))])
+        (add-map-entry [:links (links (zip-xml/xml-> entry :link))])
+        (add-map-entry [:updated (first (zip-xml/xml-> entry :updated zip-xml/text))])
+        (add-map-entry [:authors (authors (zip-xml/xml-> entry :author))])
+        (add-map-entry [:description 
+                        {:type (first (zip-xml/xml-> entry :content (zip-xml/attr :type)))
+                         :content (first (zip-xml/xml-> entry :content zip-xml/text))}]))))
 
 (defn parse-atom
-  "Parse Atom feed generated by ariane.core/parse"
+  "Parse Atom feed generated by ariane.core/parse. "
   [feed]
-  (for [tag feed]
-    (let [tag-name (:tag tag)
-          tag-content (:content tag)]
-      (cond
-       (= tag-name :subtitle) [:description (first tag-content)]
-       (= tag-name :summary) [:description (first tag-content)]
-       (= tag-name :link) [:link (get-in tag [:attrs :href])]
-       (= tag-name :author) [:author (parse-atom (:content tag))]
-       (= tag-name :entry) [:item (parse-atom (:content tag))]
-       :else [tag-name (first tag-content)]))))
+  (let [root (zip/xml-zip feed)]
+    {:infos (infos root) 
+     :entries (entries root)}))

src/ariane/core.clj

@@ -17,20 +17,6 @@
   [source]
   (let [feed (xml/parse source)]
     (cond 
-     (rss-feed? feed) (rss/parse-rss (:content feed))
-     (atom-feed? feed) (atom/parse-atom (:content feed))
-     :else (println "Unkown format"))))
-
-(defn infos
-  "Returns feed informations."
-  [data]
-  (filter (fn [[name _]]
-            (not= :item name))
-          data))
-
-(defn items
-  "Returns feed items."
-  [data]
-  (filter (fn [[name _]]
-            (= :item name)) 
-          data))
+     (rss-feed? feed) (rss/parse-rss feed)
+     (atom-feed? feed) (atom/parse-atom feed)
+     :else (println "Unsupported format"))))

src/ariane/rss.clj

@@ -1,26 +1,55 @@
 (ns ariane.rss
-  "Namespace for parsing RSS feeds.")
+  "Namespace for parsing RSS feeds."
+  (:require [clojure.zip :as zip]
+            [clojure.data.zip.xml :as zip-xml]))
 
-(defn- channel
-  [feed]
-  (:content (first feed)))
+(defn- add-map-entry
+  "Add key/value entry to map only if value is not nil."
+  [to [key value]]
+  (if (and value
+           (not-empty value))
+    (assoc to key value)
+    to))
 
-(defn- parse-feed
-  [feed]
-  (for [tag feed]
-    (let [tag-name (:tag tag)
-          tag-content (:content tag)]
-      (cond
-       (= tag-name :guid) [:id (first tag-content)]
-       (= tag-name :lastBuildDate) [:updated (first tag-content)]
-       (= tag-name :pubDate) [:published (first tag-content)]
-       (= tag-name :author) [:author (parse-feed (:content tag))]
-       (= tag-name :item) [:item (parse-feed (:content tag))]
-       :else [tag-name (first tag-content)]))))
+(defn- links
+  [links]
+  (vec (for [link links]
+         (let [link-spec {:href 
+                          (first (zip-xml/xml-> link zip-xml/text))}]
+           (if-let [rel (first (zip-xml/xml-> link (zip-xml/attr :rel)))]
+             (assoc link-spec :rel rel)
+             link-spec)))))
+
+(defn- infos
+  [root]
+  (-> {}
+      (add-map-entry [:title (first (zip-xml/xml-> root :title zip-xml/text))])
+      (add-map-entry [:description (first (zip-xml/xml-> root :description zip-xml/text))])
+      (add-map-entry [:updated (first (zip-xml/xml-> root :lastBuildDate zip-xml/text))])
+      (add-map-entry [:links (links (zip-xml/xml-> root :link))])))
+
+(defn- authors
+  [authors]
+  (vec (for [author authors]
+         {:name (first (zip-xml/xml-> author :name zip-xml/text))
+          :uri (first (zip-xml/xml-> author :uri zip-xml/text))})))
+
+(defn- entries 
+  [root]
+  (for [entry (zip-xml/xml-> root :item)]
+    (-> {}
+        (add-map-entry [:id (first (zip-xml/xml-> entry :guid zip-xml/text))])
+        (add-map-entry [:title (first (zip-xml/xml-> entry :title zip-xml/text))])
+        (add-map-entry [:links (links (zip-xml/xml-> entry :link))])
+        (add-map-entry [:updated (first (zip-xml/xml-> entry :pubDate zip-xml/text))])
+        (add-map-entry [:authors (authors (zip-xml/xml-> entry :author))])
+        (add-map-entry [:description 
+                        {:content (first (zip-xml/xml-> entry :description zip-xml/text))}]))))
 
 (defn parse-rss
-  "Parse RSS feed generated by ariane.core/parse"
+  "Parse Atom feed generated by ariane.core/parse. "
   [feed]
-  (-> feed 
-      channel
-      parse-feed))
+  (let [root (zip/xml-zip feed)
+        channel (first (zip-xml/xml-> root :channel))]
+    {:infos (infos channel) 
+     :entries (entries channel)}))