DVW statements

This commit is contained in:
Bryce Covert
2019-10-26 18:00:40 -07:00
parent f0e93ec3cf
commit 332eae2f58
5 changed files with 73 additions and 35 deletions

View File

@@ -10,38 +10,40 @@
(every? #(re-find % text) keywords))
(defn extract [wb {:keys [extract vendor]}]
(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-code vendor}
extract))
(if (fn? extract)
(extract wb vendor)
[(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-code 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 " "))]
(->> t/excel-templates
(filter (partial template-applies? text))
first
(extract wb)
))])
(let [wb (d/load-workbook file)
text (->> wb
(d/sheet-seq)
first
(d/cell-seq)
(map d/read-cell)
(str/join " "))]
(->> t/excel-templates
(filter (partial template-applies? text))
first
(extract wb)
)))

View File

@@ -1,4 +1,7 @@
(ns auto-ap.parse.templates)
(ns auto-ap.parse.templates
(:require [dk.ative.docjure.spreadsheet :as d]
[clojure.string :as str])
(:import (org.apache.poi.ss.util CellAddress)))
(def pdf-templates
@@ -80,6 +83,9 @@
:total #"\s{2,}INVOICE\s{2,}.*?(?=TOTAL)TOTAL\s+([0-9.]+)"}
:parser {:date [:clj-time "MM/dd/yyyy"]}}])
(defn offset [c x y]
(.toString (CellAddress. (+ y (.getRow (.getAddress c))) (+ x (.getColumn (.getAddress c))) )))
(def excel-templates
[{:vendor "Isp Productions"
:keywords [#"ISP PRODUCTIONS"]
@@ -92,4 +98,27 @@
:extract {:customer-identifier [#"Customer #" 1 0]
:total [#"Total Invoice" 0 5]
:date [#"Date" 0 0 #"Date: (.*)"]
:invoice-number [#"Invoice #" 0 0 #"Invoice #: (.*)"]}}])
:invoice-number [#"Invoice #" 0 0 #"Invoice #: (.*)"]}}
{:vendor "DVW Commercial"
:keywords [#"Thank you!!!"]
:extract (fn [wb vendor]
(let [[sheet] (d/sheet-seq wb)]
(transduce (comp (filter (fn [c]
(re-find #"Invoice" (str (d/read-cell c)))))
(map (fn [c]
(let [customer-identifier (d/read-cell (->> (d/select-cell (offset c -3 0) sheet)
(iterate (fn [c]
(d/select-cell (offset c 0 -1) sheet)))
(filter (fn [c]
(not (str/blank? (d/read-cell c)))))
first))]
{:customer-identifier customer-identifier
:text customer-identifier
:full-text customer-identifier
:date (d/read-cell (d/select-cell (offset c 2 0) sheet))
:invoice-number (d/read-cell (d/select-cell (offset c 4 0) sheet))
:total (str (d/read-cell (d/select-cell (offset c 8 0) sheet)))
:vendor-code vendor}))))
conj
[]
(d/cell-seq sheet))))}])