added excel.

src/clj/auto_ap/parse/excel.clj

@@ -0,0 +1,61 @@
+(ns auto-ap.parse.excel
+  (:import [org.apache.poi.ss.util CellAddress])
+  (:require [dk.ative.docjure.spreadsheet :as d]
+            
+            [clojure.string :as str]))
+
+(def templates
+  [{:vendor "Isp Productions"
+    :keywords [#"ISP PRODUCTIONS"]
+    :extract {:customer-identifier [#"SERVICES PROVIDED TO" 1 0]
+              :total [#"PAY THIS" -1 0]
+              :date [#"INVOICE DATE" 0 1]
+              :invoice-number [#"INVOICE NUMBER" 0 1]}}
+   {:vendor "Southern Wine Online"
+    :keywords [#"Please note that the total invoice amount may"]
+    :extract {:customer-identifier [#"Customer #" 1 0]
+              :total [#"Total Invoice" 0 5]
+              :date [#"Date" 0 0 #"Date: (.*)"]
+              :invoice-number [#"Invoice #" 0 0  #"Invoice #: (.*)"]}}])
+
+(defn template-applies? [text {:keys [keywords]}]
+  (every? #(re-find % text) keywords))
+
+(defn extract [wb {:keys [extract vendor]}]
+  (println extract)
+  
+  (reduce-kv
+   (fn [invoice k [regex offset-row offset-column extract-regex]]
+     (assoc invoice k
+            (->> wb
+                 (d/sheet-seq)
+                 first
+                 (d/cell-seq)
+                 (filter (fn [cell]
+                           (re-find regex (str (d/read-cell cell)))))
+                 (map (fn [cell]
+                        (let [address (.getAddress cell)
+                              cell-value (str (d/read-cell (d/select-cell  (.toString (CellAddress. (+ offset-row (.getRow address)) (+ offset-column (.getColumn address)) ))
+                                                            (first (d/sheet-seq wb)))))] 
+                          (if extract-regex
+                            (second (re-find extract-regex cell-value))
+                            
+                            cell-value))))
+                 first)))
+   {:vendor vendor}
+   extract))
+
+(defn parse-file
+  [file filename]
+  [(let [wb (d/load-workbook file)
+         text (->> wb
+                   (d/sheet-seq)
+                   first
+                   (d/cell-seq)
+                   (map d/read-cell)
+                   (str/join " "))]
+     (->> templates
+          (filter (partial template-applies? text))
+          first
+          (extract wb)
+          ))])