Disallows chatgpt for emailed invoices.

This commit is contained in:
2025-03-25 21:26:40 -07:00
parent 581033001b
commit 01329bbdf5
5 changed files with 30 additions and 24 deletions

File diff suppressed because one or more lines are too long

View File

@@ -62,7 +62,7 @@
:input-stream (io/input-stream filename) :input-stream (io/input-stream filename)
:metadata {:content-type "application/pdf" :metadata {:content-type "application/pdf"
:content-length (.length (io/file filename))}) :content-length (.length (io/file filename))})
(let [imports (->> (parse/parse-file filename filename) (let [imports (->> (parse/parse-file filename filename :allow-glimpse? false)
(map #(assoc % (map #(assoc %
:source-url (str "https://" (:data-bucket env) :source-url (str "https://" (:data-bucket env)
"/" "/"

View File

@@ -61,14 +61,20 @@
(extract-template text))) (extract-template text)))
(defmulti parse-file (fn [_ filename] (.toLowerCase (last (str/split filename #"\." )))))
(defmulti parse-file
"Parses a file based on its extension. Accepts options as additional arguments.
Options:
- :allow-glimpse? (default false) - If true, allows parsing a glimpse of the file."
(fn [_ filename & opts]
(.toLowerCase (last (str/split filename #"\.")))))
(defn invoke-glimpse2 [f] (defn invoke-glimpse2 [f]
(let [result (slurp (:payload (lambda/invoke {:client-config {:request-timeout 120000 (let [result (slurp (:payload (lambda/invoke {:client-config {:request-timeout 120000
:socket-timeout 120000}} {:function-name "glimpse2" :payload :socket-timeout 120000}} {:function-name "glimpse2" :payload
(json/write-str (json/write-str
(alog/peek ::x {"url" (str "https://" "data.prod.app.integreatconsult.com" "/" f ) })) (alog/peek ::x {"url" (str "https://" "data.prod.app.integreatconsult.com" "/" f)}))})))]
})))]
(alog/info ::glimpse2-payload :payload result) (alog/info ::glimpse2-payload :payload result)
(-> result (-> result
@@ -92,35 +98,35 @@
(get "label")) (get "label"))
:total (get i "total") :total (get i "total")
:invoice-number (get i "invoice_number") :invoice-number (get i "invoice_number")
:template "None found - defaulting to ChatGPT"}) :template "None found - defaulting to ChatGPT"}))
)
(catch Exception e (catch Exception e
(alog/warn ::glimpse2-not-work :error e) (alog/warn ::glimpse2-not-work :error e)
nil))) nil)))
(defmethod parse-file (defmethod parse-file
"pdf" "pdf"
[file _] [file _ & {:keys [allow-glimpse?] :or {allow-glimpse? false}}]
(or (or
(-> (sh/sh "pdftotext" "-layout" file "-") (-> (sh/sh "pdftotext" "-layout" file "-")
:out :out
parse) parse)
(alog/peek ::glimpse2-result (glimpse2 file)))) (and allow-glimpse? (alog/peek ::glimpse2-result (glimpse2 file)))))
(defmethod parse-file (defmethod parse-file
"csv" "csv"
[file filename] [file filename & _]
(csv/parse-file file filename)) (csv/parse-file file filename))
(defmethod parse-file (defmethod parse-file
"xls" "xls"
[file filename] [file filename & _]
(excel/parse-file file filename)) (excel/parse-file file filename))
(defmethod parse-file (defmethod parse-file
"xlsx" "xlsx"
[file filename] [file filename & _]
(excel/parse-file file filename)) (excel/parse-file file filename))
(defn best-match (defn best-match

View File

@@ -278,7 +278,7 @@
"text/csv" "text/csv"
"application/pdf") "application/pdf")
:content-length (.length tempfile)}) :content-length (.length tempfile)})
imports (->> (parse/parse-file (.getPath tempfile) filename) imports (->> (parse/parse-file (.getPath tempfile) filename :allow-glimpse? true)
(map #(assoc % (map #(assoc %
:client-override client :client-override client
:location-override location :location-override location

View File

@@ -716,7 +716,7 @@
:content-length (.length tempfile)}) :content-length (.length tempfile)})
imports (->> (if force-chatgpt imports (->> (if force-chatgpt
(parse/glimpse2 (.getPath tempfile)) (parse/glimpse2 (.getPath tempfile))
(parse/parse-file (.getPath tempfile) filename)) (parse/parse-file (.getPath tempfile) filename :allow-glimpse? true))
(map #(assoc % (map #(assoc %
:client-override force-client :client-override force-client
:location-override force-location :location-override force-location