From 7b4403c84ee18862bd4aed3e0d7cc54bc5d5001a Mon Sep 17 00:00:00 2001
From: Martin Winkler <efelon@winklerfamilie.eu>
Date: Mon, 26 Dec 2022 16:19:46 +0100
Subject: [PATCH] paperless-venv - Step to install document classification
 model (nltk)

---
 seqs/paperless-venv.sh | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/seqs/paperless-venv.sh b/seqs/paperless-venv.sh
index 0deb66d..cd9bc85 100755
--- a/seqs/paperless-venv.sh
+++ b/seqs/paperless-venv.sh
@@ -189,6 +189,21 @@ step_17() {
   exe "${sc_paperlessVenv}/bin/python3" manage.py migrate
 }
 
+step_19_info() { echo "Install document classification model (nltk)"; }
+step_19_alias() { echo 'nltk'; }
+step_19() {
+  info 'Installing nltk'
+  exe "${sc_paperlessVenv}/bin/pip3" install -U nltk
+  endReturn -o $? "Failed to install nlkt"
+
+  exe cd "${sc_paperlessHome}/data"
+  exe mkdir -p "nltk"
+  exe "${sc_paperlessVenv}/bin/python3" -m nltk.downloader -d ./nltk/ snowball_data stopwords punkt
+  endReturn -o $? "Failed to install nlkt modules snowball stopwords punkt"
+
+  step service restart
+}
+
 step_30_info() { 
   echo "Retag existing documents"
   echoinfo "OPTIONS"