implemented HTTP Basic authorization

[rigidfind] / index.cgi

diff --git a/index.cgi b/index.cgi
index d4f4cec6f8ed573ce27032ccdf9b0467798d0fee..5351b2b27e3d42ed17a4124b79647b2c78e294ad 100755 (executable)
--- a/index.cgi
+++ b/index.cgi
@@ -6,10 +6,10 @@
 
 [ "$_DATE" ] || _DATE="$(date +%s)"
 
-debug "$REQUEST_METHOD $REQUEST_URI    $SERVER_PROTOCOL"
+debug "$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"}"
 
@@ -122,15 +143,23 @@ if   [ "$REQUEST_METHOD" = "PUT" ]; then
     s;,[ \t\r\n]*"content"[ \t\r\n]*:[ \t\r\n]*"[^"]*";;
   ')"
   J="$(json_load "${J}")"
-  
-  ingest "$J" "$content"\
-  | "${_EXEC}/concordance.sh" \
-    "$_DATA/$_INDEX/" "$(STRING "$_doc")       $_DATE"
+
+  debug "Content: ${#content} bytes"
+  debug "$(json_dump "$J")"
+
+  if [ "${#content}" -gt 0 ]; then
+    ingest "$J" "$content"\
+    | "${_EXEC}/concordance.sh" \
+      "$_DATA/$_INDEX/" "$(STRING "$_doc")     $_DATE"
+  fi
 
   J="${J#obj:}"
   J="$(DB2 "$J" set _indexdate num:"$_DATE")"
 
-  if   DBM "$_records" insert "$_doc" "$J"; then
+  if [ "${#content}" -eq 0 ]; then
+    printf '%s: %s\r\n' "Status" "200 OK"
+    result="updated"
+  elif DBM "$_records" insert "$_doc" "$J"; then
     printf '%s: %s\r\n' "Status" "201 Created" "Location" "/${_INDEX}/_doc/$(URL "$_doc")" \
     result="created"
   elif DBM "$_records" update "$_doc" "$J"; then
@@ -141,10 +170,10 @@ if   [ "$REQUEST_METHOD" = "PUT" ]; then
     exit 0
   fi
 
-  sed 's;$;\r;' <<-EOF
-       X-elastic-product: Elasticsearch
-       content-type: application/vnd.elasticsearch+json;compatible-with=8
-
+  cat <<-EOF
+       X-elastic-product: Elasticsearch\r
+       content-type: application/vnd.elasticsearch+json;compatible-with=8\r
+       \r
        { "_index": $(json_dump str:"${_INDEX}"),
          "_id": $(json_dump str:"$_doc"),
          "result": "$result",
@@ -169,10 +198,10 @@ elif [ "$REQUEST_METHOD" = "DELETE" ]; then
     result="not_found"
   fi
 
-  sed 's;$;\r;' <<-EOF
-       X-elastic-product: Elasticsearch
-       content-type: application/vnd.elasticsearch+json;compatible-with=8
-
+  cat <<-EOF
+       X-elastic-product: Elasticsearch\r
+       content-type: application/vnd.elasticsearch+json;compatible-with=8\r
+       \r
        { "_index": $(json_dump str:"${_INDEX}"),
          "_id": $(json_dump str:"$_doc"),
          "result": "$result",
@@ -182,7 +211,7 @@ elif [ "$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="$(
@@ -190,10 +219,12 @@ elif [ "$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")"
@@ -203,11 +234,11 @@ elif [ "$REQUEST_METHOD" = "POST" ]; then
   )"
   results="${results%  }"
 
-  sed 's;$;\r;' <<-EOF
-       Status: 200 OK
-       X-elastic-product: Elasticsearch
-       Content-Type: application/vnd.elasticsearch+json;compatible-with=8
-
+  cat <<-EOF
+       Status: 200 OK\r
+       X-elastic-product: Elasticsearch\r
+       Content-Type: application/vnd.elasticsearch+json;compatible-with=8\r
+       \r
        { "took":0,
          "timed_out":false,
          "_shards":{"total":1,"successful":1,"skipped":0,"failed":0},
@@ -224,11 +255,11 @@ elif [ "$REQUEST_METHOD" = "HEAD" ]; then
   [ ! "${accept#*"vnd.elasticsearch+json"*}" ] \
   && ctype="${accept}" || ctype="application/json"
 
-  sed 's;$;\r;' <<-EOF
-       HTTP/1.1 200 OK
-       X-elastic-product: Elasticsearch
-       content-type: ${ctype}
-
+  cat <<-EOF
+       Status: 200 OK\r
+       X-elastic-product: Elasticsearch\r
+       content-type: ${ctype}\r
+       \r
        EOF
   exit 0
 
@@ -237,15 +268,15 @@ elif [ "$REQUEST_METHOD" = "GET" ]; then
   [ ! "${accept#*"vnd.elasticsearch+json"*}" ] \
   && ctype="${accept}" || ctype="application/json"
 
-  sed 's;$;\r;' <<-EOF
-       HTTP/1.1 200 OK
-       X-elastic-product: Elasticsearch
-       content-type: ${ctype}
-
+  cat <<-EOF
+       Status: 200 OK\r
+       X-elastic-product: Elasticsearch\r
+       content-type: ${ctype}\r
+       \r
        EOF
        
   if [ "$PATH_INFO" = "/${_INDEX}/" ]; then
-  sed 's;$;\r;' <<-EOF
+  cat <<-EOF
        { $(json_dump str:"${_INDEX}"): {
            "aliases":{},
            "mappings": {
@@ -276,7 +307,7 @@ elif [ "$REQUEST_METHOD" = "GET" ]; then
        }
        EOF
   else
-    sed 's;$;\r;' <<-EOF
+    cat <<-EOF
        { "name" : "head",
          "cluster_name" : "elasticsearch",
          "version" : {