first version of sample parser.

src/auto_ap/handler.clj

@@ -0,0 +1,29 @@
+(ns auto-ap.handler
+  (:require [compojure.core :refer :all]
+            [compojure.route :as route]
+            [clojure.java.io :as io]
+            [auto-ap.parse :as parse]
+            [ring.middleware.multipart-params :as mp]
+            [ring.util.response :as response]
+            [ring.middleware.defaults :refer [wrap-defaults site-defaults]]
+            [ring.middleware.json :refer [wrap-json-response]]))
+
+(defroutes app-routes
+  (GET "/" [] (response/resource-response "index.html" {:root "public"}))
+  (POST "/pdf-upload"
+        {{  files "file"} :params :as params}
+        (let [{:keys [filename tempfile]} (second files)]
+          (io/copy tempfile (io/file "resources" "public" filename))
+          (for [{:keys [total date invoice-number customer-identifier]} (parse/parse-file (str "resources/public/" filename))]
+            (do
+              (println (str "An invoice #" invoice-number " on " date " for " total))
+              (str "An invoice for customer " customer-identifier " #" invoice-number " on " date " for " total )))))
+  (route/resources "/")
+  (route/not-found "Not Found"))
+
+#_(defroutes routes
+  (GET "/" [] (resource-response "index.html" {:root "public"}))
+  (resources "/"))
+
+(def app
+  (wrap-json-response (mp/wrap-multipart-params app-routes)))

src/auto_ap/parse.clj

@@ -0,0 +1,54 @@
+(ns auto-ap.parse
+  (:require [clojure.java.io :as io]
+            [clojure.string :as str]
+            [clojure.java.shell :as sh]))
+
+(def templates
+  [{:keywords [#"CHEF'S WAREHOUSE"]
+    :extract {:total #"2 WKS C\.C\.\s+([\d.,]+)"
+              :customer-identifier #"\n([A-Z][A-Z ]+)\s{2,}"
+              :date #"\s+([0-9]+/[0-9]+/[0-9]+)"
+              :invoice-number #"\s+[0-9]+/[0-9]+/[0-9]+\s+([0-9]+)"}}
+
+   {:keywords [#"Golden Gate Meat"]
+    :extract {:total #"Invoice Total\:\s+\$([\d.,]+)"
+              :customer-identifier #"Bill To\s*:\s*([\w ]+)\s{2,}"
+              :date #"Printed:\s+([0-9]+/[0-9]+/[0-9]+)"
+              :invoice-number #"Invoice\s+[^\n]+\n[^\n]+\n\s+([0-9]+)"}}
+
+   {:keywords [#"CINTAS CORPORATION"]
+     :extract {:invoice-number #"INVOICE\s#\s+([\d.,]+)"
+               :customer-identifier #"BILL TO\s*:\s{2,}([\w ]+)\s{2,}"
+               :date #"INVOICE DATE\s*\n.*\s+([0-9]+/[0-9]+/[0-9]+)"
+               :total #"INVOICE TOTAL\s+([0-9.]+)"}
+    :multi #"\f\f"}])
+
+(defn template-applies? [text {:keys [keywords]}]
+  (every? #(re-find % text) keywords))
+
+(defn extract-template [text template]
+  (if (:multi template)
+    (mapcat
+     #(extract-template % (dissoc template :multi))
+     (str/split text (:multi template)))
+
+    [(->> template
+           :extract
+           (reduce-kv
+            (fn [result k v]
+              (assoc result k (some-> (first (map second (re-seq v text)))
+                                      str/trim )))
+            {}))]))
+
+(defn parse [text]
+  (->> templates
+       (filter (partial template-applies? text))
+       first
+       (extract-template text)))
+
+(defn parse-file
+  [file]
+  (->  (sh/sh "pdftotext" "-layout" file "-")
+       :out
+       (doto println)
+       parse))