From: Paul Hänsch <paul@plutz.net>
Date: Tue, 5 Mar 2024 00:42:07 +0000 (+0100)
Subject: use temp files when unzipping documents
X-Git-Url: http://git.plutz.net/?a=commitdiff_plain;h=092bfad64ae6b2022991f6193f042d6d53bf10c2;p=rigidfind

use temp files when unzipping documents
---

diff --git a/index.cgi b/index.cgi
index 9ba696a..0078a2f 100755
--- a/index.cgi
+++ b/index.cgi
@@ -9,7 +9,7 @@
 debug "$REQUEST_METHOD	$REQUEST_URI	$SERVER_PROTOCOL	$_DATE"
 
 ingest() {
-  local J="$1"
+  local J="$1" ztmp="${TMP:-/tmp}/zipfile_$$.zip"
 
   # json_get "$J" title
   # json_get "$J" parts.comments
@@ -39,24 +39,28 @@ ingest() {
       | sed 's;<[^>]*>;;g'
       ;;
     *.docx)
-      printf %s "$content" |base64 -d \
-      | unzip -qc /dev/stdin word/document.xml \
+      printf %s "$content" |base64 -d >"$ztmp"
+      unzip -qc "$ztmp" word/document.xml \
       | head -c 128M | sed 's;<[^>]*>;;g'
+      rm -- "$ztmp"
       ;;
     *.xlsx)
-      printf %s "$content" |base64 -d \
-      | unzip -qc /dev/stdin xl/sharedStrings.xml \
+      printf %s "$content" |base64 -d >"$ztmp"
+      unzip -qc "$ztmp" xl/sharedStrings.xml \
       | head -c 128M | sed 's;<[^>]*>; ;g'
+      rm -- "$ztmp"
       ;;
     *.odt)
-      printf %s "$content" |base64 -d \
-      | unzip -qc /dev/stdin content.xml \
+      printf %s "$content" |base64 -d >"$ztmp"
+      unzip -qc "$ztmp" content.xml \
       | head -c 128M | sed 's;<[^>]*>;;g'
+      rm -- "$ztmp"
       ;;
     *.ods|*.odp)
-      printf %s "$content" |base64 -d \
-      | unzip -qc /dev/stdin content.xml \
+      printf %s "$content" |base64 -d >"$ztmp"
+      unzip -qc "$ztmp" content.xml \
       | head -c 128M | sed 's;<[^>]*>; ;g'
+      rm -- "$ztmp"
       ;;
     *):;;
   esac