Removes docjure, freeing space

This commit is contained in:
2023-05-10 10:39:58 -07:00
parent 05f1f009fd
commit 8bd73b8a98
5 changed files with 60 additions and 125 deletions

View File

@@ -2,18 +2,23 @@
(:require [auto-ap.parse.templates :as t]
[auto-ap.parse.util :as u]
[clojure.string :as str]
[dk.ative.docjure.spreadsheet :as d])
(:import (org.apache.poi.ss.util CellAddress)))
[amazonica.aws.lambda :as lambda]
[clojure.data.json :as json]
[config.core :refer [env]]
[clojure.java.io :as io]
[amazonica.aws.s3 :as s3])
)
(defn template-applies? [text {:keys [keywords]}]
(every? #(re-find % text) keywords))
(defn extract [wb {:keys [extract vendor parser]}]
(if (fn? extract)
(extract wb vendor)
[(reduce-kv
#_[(reduce-kv
(fn [invoice k [regex offset-row offset-column extract-regex]]
(assoc invoice k
(->> wb
@@ -38,17 +43,34 @@
{:vendor-code vendor}
extract)]))
(defn extract-sheet-details [bucket object]
(-> (lambda/invoke {:function-name "xls-extractor" :payload
(json/write-str
{"s3_url" object "s3_bucket" bucket})})
:payload
slurp
json/read-str))
(defn parse-file
[file _]
(let [wb (d/load-workbook file)
text (->> wb
(d/sheet-seq)
first
(d/cell-seq)
(map d/read-cell)
(str/join " "))]
(let [tmp-key (str "xls-invoice/import/" (java.util.UUID/randomUUID))
_ (with-open [f (io/input-stream file)]
(s3/put-object {:bucket-name (:data-bucket env)
:key tmp-key
:input-stream f}))
sheet (extract-sheet-details (:data-bucket env) tmp-key)
text (str/join " " (mapcat seq sheet))]
(->> t/excel-templates
(filter (partial template-applies? text))
first
(extract wb)
)))
(extract sheet))))
(defn xls-date->date [f]
(when (not-empty f)
(let [f (Double/parseDouble f)
unix-days (- f 25569.0)
unix-secs (* unix-days 86400.0)]
(java.util.Date. (long (Math/round (* 1000.0 unix-secs)))))))