notid/integreat
1
0
Fork 0
Code Issues Pull Requests 2 Actions Packages Projects Releases Wiki Activity

Adds Cheetah

Browse Source
This commit is contained in:
Bryce Covert
2022-01-11 11:11:36 -08:00
parent 318ae3024c
commit 55450a3c32
3 changed files with 82 additions and 80 deletions
Download Patch File Download Diff File Expand all files Collapse all files

44
src/clj/auto_ap/parse.clj
View File

@@ -1,16 +1,15 @@
(ns auto-ap.parse
(:require [auto-ap.parse.csv :as csv]
[auto-ap.parse.excel :as excel]
[auto-ap.parse.templates :as t]
[auto-ap.parse.util :as u]
[clj-fuzzy.metrics :as m]
[clj-time.core :as time]
[clj-time.format :as f]
[clojure.java.shell :as sh]
[clojure.set :as set]
[clojure.string :as str]
[clojure.tools.logging :as log]
[auto-ap.logging :refer [info-event]]))
(:require
[auto-ap.logging :refer [info-event]]
[auto-ap.parse.csv :as csv]
[auto-ap.parse.excel :as excel]
[auto-ap.parse.templates :as t]
[auto-ap.parse.util :as u]
[clj-fuzzy.metrics :as m]
[clojure.java.shell :as sh]
[clojure.set :as set]
[clojure.string :as str]
[clojure.tools.logging :as log]))
(def last-text (atom nil))
@@ -57,11 +56,11 @@
(extract-template text)))
(defmulti parse-file (fn [file filename] (.toLowerCase (last (str/split filename #"\." )))))
(defmulti parse-file (fn [_ filename] (.toLowerCase (last (str/split filename #"\." )))))
(defmethod parse-file
"pdf"
[file filename]
[file _]
(-> (sh/sh "pdftotext" "-layout" file "-")
:out
parse))
@@ -87,7 +86,7 @@
(best-match clients invoice-client-name 0.25))
([clients invoice-client-name threshold]
(let [fuzzy-match (->> clients
(mapcat (fn [{:keys [:db/id :client/matches :client/name] :as client :or {matches []}}]
(mapcat (fn [{:keys [:client/matches :client/name] :as client :or {matches []}}]
(map (fn [m]
[client (m/jaccard (.toLowerCase invoice-client-name) (.toLowerCase m))])
(conj matches name))))
@@ -98,7 +97,7 @@
word-set (set (filter (complement str/blank?) (str/split (.toLowerCase invoice-client-name) #"[\s:\-]" )))
client-word-match (->> clients
(map
(fn [{:keys [:db/id :client/matches :client/name] :as client :or {matches []}}]
(fn [{:keys [:client/matches :client/name] :as client :or {matches []}}]
(let [client-words (-> #{}
(into
(mapcat
@@ -119,7 +118,7 @@
(mapcat (fn [{:keys [:location-match/location :location-match/matches]}]
(map (fn [match] [location match]) matches)))
(filter (fn [[location match]]
(filter (fn [[_ match]]
(re-find (re-pattern (str "(?i)" match)) text)) )
first
first)
@@ -127,15 +126,14 @@
:client/location-matches
(mapcat (fn [{:keys [:location-match/location :location-match/matches]}]
(map (fn [match] [location match]) matches)))
(filter (fn [[location match]] (re-find (re-pattern (str "(?i)" match)) full-text)) )
(filter (fn [[_ match]] (re-find (re-pattern (str "(?i)" match)) full-text)) )
first
first)
(:client/default-location client)
(first (:client/locations client))))
#_{:clj-kondo/ignore [:clojure-lsp/unused-public-var]}
(defn dbg-parse [v]
(doto
(map
(fn [x] (dissoc x :full-text :text))
(parse v))
clojure.pprint/pprint ))
(map
(fn [x] (dissoc x :full-text :text))
(parse v)))