document sample literals

src/data_readers.clj

@@ -0,0 +1 @@
+{opennlp/sample opennlp.sample/read-document-sample}

src/opennlp/sample.clj

@@ -0,0 +1,46 @@
+(ns opennlp.sample
+  (:require [clojure.java.io :as io])
+  (:import (opennlp.tools.doccat DocumentSample)
+           (opennlp.tools.util ObjectStream)))
+
+(defn print-sample [sample ^java.io.Writer w]
+  (.write w "#opennlp/sample {")
+  (.write w ":category ")
+  (binding [*out* w]
+    (prn (.getCategory sample)))
+  (.write w " :text ")
+  (binding [*out* w]
+    (prn (vec (.getText sample))))
+  (.write w "}"))
+
+(defmethod print-method DocumentSample
+  [sample w]
+  (print-sample sample w))
+
+(defmethod print-dup DocumentSample
+  [sample w]
+  (print-sample sample w))
+
+(defn read-document-sample [{:keys [category text]}]
+  (DocumentSample. category (into-array String text)))
+
+(defn clojure-document-sample-stream [in]
+  (let [i (java.io.PushbackReader. (io/reader in))
+        buf (atom [])
+        pos (atom 0)]
+    (reify
+      ObjectStream
+      (read [_]
+        (if (= @pos (count @buf))
+          (when-let [obj (read i false nil)]
+            (swap! buf conj obj)
+            (swap! pos inc)
+            obj)
+          (let [p @pos]
+            (swap! pos inc)
+            (nth @buf p))))
+      (close [_]
+        (.close i)
+        (.close in))
+      (reset [_]
+        (reset! pos 0)))))

test/opennlp/test/sample.clj

@@ -0,0 +1,24 @@
+(ns opennlp.test.sample
+  (:require [clojure.test :refer :all]
+            [opennlp.sample :refer [clojure-document-sample-stream]])
+  (:import (opennlp.tools.doccat DocumentSample)))
+
+(deftest test-samples-round-trip
+  (let [d #opennlp/sample {:category "foo" :text ["bar"]}]
+    (is (= d (read-string (pr-str d))))))
+
+(deftest test-clojure-document-sample-stream
+  (let [d #opennlp/sample {:category "foo" :text ["bar"]}
+        x (java.io.ByteArrayInputStream.
+           (.getBytes
+            (with-out-str
+              (prn d)
+              (prn d))))
+        s (clojure-document-sample-stream x)]
+    (is (= (.read s) d))
+    (is (= (.read s) d))
+    (is (nil? (.read s)))
+    (.reset s)
+    (is (= (.read s) d))
+    (is (= (.read s) d))
+    (is (nil? (.read s)))))