From 0fcb078b95a224f167968bd54b96e8fc7c10cdbb Mon Sep 17 00:00:00 2001
From: Justin <justin2004@hotmail.com>
Date: Thu, 2 Mar 2023 15:34:01 -0500
Subject: [PATCH] jaro

---
 project.clj            |  3 ++-
 src/datething/jaro.clj | 38 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 40 insertions(+), 1 deletion(-)
 create mode 100644 src/datething/jaro.clj

diff --git a/project.clj b/project.clj
index 4ac5514..b398b7e 100644
--- a/project.clj
+++ b/project.clj
@@ -3,9 +3,10 @@
   :url "https://github.com/justin2004/datething"
   :dependencies [[org.clojure/clojure "1.10.0"]
                  [wit/duckling "0.4.24"]
+                 [info.debatty/java-string-similarity "2.0.0"]
                  [org.apache.jena/apache-jena-libs "4.2.0" :extension "pom"]]
   :repl-options {:init-ns datething.parse}
   :profile {:uberjar {:aot :all
                       :debug true}}
-  :main datething.parse) ; i think i just need this to get AOT when i run `lein uberjar`
+  :main datething.jaro) ; i think i just need this to get AOT when i run `lein uberjar`
                          ; TODO  just do AOT instead
diff --git a/src/datething/jaro.clj b/src/datething/jaro.clj
new file mode 100644
index 0000000..a48d110
--- /dev/null
+++ b/src/datething/jaro.clj
@@ -0,0 +1,38 @@
+(ns datething.jaro
+  (:gen-class
+   :extends org.apache.jena.sparql.function.FunctionBase2)
+  (:require [clojure.pprint :as pp]))
+
+; (def jw (new info.debatty.java.stringsimilarity.JaroWinkler))
+; (.distance jw "atlanta, ga" "atlanta")
+
+(defn -exec ^org.apache.jena.sparql.expr.NodeValue [^org.apache.jena.sparql.expr.NodeValue & v ]
+  (do 
+    ; (.println java.lang.System/err "you passed in") 
+    ;   (.println java.lang.System/err v)
+      (let [asked0 (.asVar (.get (.getList (nth v 2)) ; asked is the Var passed into the function
+                                0))
+            asked1 (.asVar (.get (.getList (nth v 2)) ; asked is the Var passed into the function
+                                1)) 
+            looked0 (str (.getLiteralValue (.get (nth v 1) ; looked is the Var's value as found (looked up) in the bindings
+                                                asked0)))
+            looked1 (str (.getLiteralValue (.get (nth v 1) ; looked is the Var's value as found (looked up) in the bindings
+                                                asked1)))
+            jaroDistance (.distance (new info.debatty.java.stringsimilarity.JaroWinkler)
+                                    looked0 looked1) ]
+
+      ; (.println java.lang.System/err "asked0")
+      ; (.println java.lang.System/err looked0)
+      ; (.println java.lang.System/err "asked1")
+      ; (.println java.lang.System/err looked1)
+      (new org.apache.jena.sparql.expr.nodevalue.NodeValueFloat jaroDistance))))
+
+(comment  (defn -exec ^org.apache.jena.sparql.expr.NodeValue [^org.apache.jena.sparql.expr.NodeValue & v]
+  (do (let [asked (.asVar (.get (.getList (nth v 2)) ; asked is the Var passed into the function
+                                0))
+            looked (str (.getLiteralValue (.get (nth v 1) ; looked is the Var's value as found (looked up) in the bindings
+                                                asked)))
+            parsed (str (:value (:value (first (duck/parse :en$core    ; parsed is the parsed Var's value
+                                                           looked 
+                                                           [:time])))))] 
+        (new org.apache.jena.sparql.expr.nodevalue.NodeValueString parsed)))))