company identification.

src/clj/auto_ap/parse/excel.clj

@@ -1,29 +1,15 @@
 (ns auto-ap.parse.excel
   (:import [org.apache.poi.ss.util CellAddress])
   (:require [dk.ative.docjure.spreadsheet :as d]
-            
-            [clojure.string :as str]))
+            [clojure.string :as str]
+            [auto-ap.parse.templates :as t]))
+
 
-(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
@@ -54,7 +40,7 @@
                    (d/cell-seq)
                    (map d/read-cell)
                    (str/join " "))]
-     (->> templates
+     (->> t/excel-templates
           (filter (partial template-applies? text))
           first
           (extract wb)

src/clj/auto_ap/parse/templates.clj

@@ -0,0 +1,38 @@
+(ns auto-ap.parse.templates)
+
+(def pdf-templates
+  [{:vendor "CHFW"
+    :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]+)"}}
+
+   {:vendor "GGM"
+    :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]+)"}}
+
+   {:vendor "CINTAS"
+    :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"}])
+
+(def excel-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 #: (.*)"]}}])