X-Git-Url: https://git.plutz.net//?a=blobdiff_plain;f=index.cgi;h=5351b2b27e3d42ed17a4124b79647b2c78e294ad;hb=ed8107a661e4737beb99ebb6abd0512b5e965bfc;hp=9ba696abe827ad71aeafd48f8bc178e96905d575;hpb=1a5612ef66703f28969500e8ff7df258747c7169;p=rigidfind

diff --git a/index.cgi b/index.cgi
index 9ba696a..5351b2b 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
@@ -105,6 +109,23 @@ search() {
 _INDEX="${PATH_INFO#/}" _INDEX="${_INDEX%%/*}"
 _records="${_DATA}/${_INDEX}/_0_DOCS"
 
+if [ "${INDEX}" -a ! -d "${_DATA}/${_INDEX}" ]; then
+  printf '%s\r\n' "Status: 404 Not Found" ""
+  exit 0
+elif authlist="$(DBM "${_DATA}/auth.db" get "${_INDEX}" )"; then
+  auth="$(HEADER Authorization)" auth="${auth#Basic }"
+  for a in $authlist deny; do
+    [ "$auth" = "$a" ] && break
+  done
+  if [ "$a" = "deny" -o ! "$auth" ]; then
+    printf '%s\r\n' "Status: 401 Unauthorized" \
+      "WWW-Authenticate: Basic realm=\"Rigid Find\"" "" \
+    | debug
+    exit 0
+  fi
+  unset a auth authlist
+fi
+
 if   [ "$REQUEST_METHOD" = "PUT" ]; then
   _doc="${PATH_INFO#"/${_INDEX}/_doc"}"
 
@@ -190,7 +211,7 @@ elif [ "$REQUEST_METHOD" = "DELETE" ]; then
   exit 0
 
 elif [ "$REQUEST_METHOD" = "POST" ]; then
-  J="$(json_load "$(head -c "${CONTENT_LENGTH:-0}")")"
+  J="$(json_load "$(head -c "${CONTENT_LENGTH:-0}" |debug)")"
   J="$(json_get "$J" query.bool.must.bool.should)"
 
   words="$(
@@ -198,10 +219,12 @@ elif [ "$REQUEST_METHOD" = "POST" ]; then
       json_get "$(UNSTRING "$j")" match_phrase_prefix.content
     done 2>/dev/null
   )"
+  debug "Search words: $words"
 
   results="@	$(
     search "${_DATA}/${_INDEX}" $words \
     | while read -r score id source; do
+      debug "Hit: $id	$score"
       S="$(DB2   "" set _index  str:"${_INDEX}")"
       S="$(DB2 "$S" set _id     str:"$(UNSTRING "${id#/}")")"
       S="$(DB2 "$S" set _score  num:"$score")"
@@ -233,7 +256,7 @@ elif [ "$REQUEST_METHOD" = "HEAD" ]; then
   && ctype="${accept}" || ctype="application/json"
 
   cat <<-EOF
-	HTTP/1.1 200 OK
+	Status: 200 OK
 	X-elastic-product: Elasticsearch
 	content-type: ${ctype}
 	
@@ -246,7 +269,7 @@ elif [ "$REQUEST_METHOD" = "GET" ]; then
   && ctype="${accept}" || ctype="application/json"
 
   cat <<-EOF
-	HTTP/1.1 200 OK
+	Status: 200 OK
 	X-elastic-product: Elasticsearch
 	content-type: ${ctype}