From 1a5612ef66703f28969500e8ff7df258747c7169 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Paul=20H=C3=A4nsch?= <paul@plutz.net>
Date: Tue, 5 Mar 2024 01:29:52 +0100
Subject: [PATCH] avoid indexing empty content

---
 index.cgi | 70 +++++++++++++++++++++++++++++++------------------------
 1 file changed, 39 insertions(+), 31 deletions(-)

diff --git a/index.cgi b/index.cgi
index d4f4cec..9ba696a 100755
--- a/index.cgi
+++ b/index.cgi
@@ -6,7 +6,7 @@
 
 [ "$_DATE" ] || _DATE="$(date +%s)"
 
-debug "$REQUEST_METHOD	$REQUEST_URI	$SERVER_PROTOCOL"
+debug "$REQUEST_METHOD	$REQUEST_URI	$SERVER_PROTOCOL	$_DATE"
 
 ingest() {
   local J="$1"
@@ -122,15 +122,23 @@ if   [ "$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" "$J"; then
@@ -141,10 +149,10 @@ if   [ "$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
+	content-type: application/vnd.elasticsearch+json;compatible-with=8
+	
 	{ "_index": $(json_dump str:"${_INDEX}"),
 	  "_id": $(json_dump str:"$_doc"),
 	  "result": "$result",
@@ -169,10 +177,10 @@ elif [ "$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
+	content-type: application/vnd.elasticsearch+json;compatible-with=8
+	
 	{ "_index": $(json_dump str:"${_INDEX}"),
 	  "_id": $(json_dump str:"$_doc"),
 	  "result": "$result",
@@ -203,11 +211,11 @@ elif [ "$REQUEST_METHOD" = "POST" ]; then
   )"
   results="${results%	}"
 
-  sed 's;$;\r;' <<-EOF
-	Status: 200 OK
-	X-elastic-product: Elasticsearch
-	Content-Type: application/vnd.elasticsearch+json;compatible-with=8
-
+  cat <<-EOF
+	Status: 200 OK
+	X-elastic-product: Elasticsearch
+	Content-Type: application/vnd.elasticsearch+json;compatible-with=8
+	
 	{ "took":0,
 	  "timed_out":false,
 	  "_shards":{"total":1,"successful":1,"skipped":0,"failed":0},
@@ -224,11 +232,11 @@ elif [ "$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
+	HTTP/1.1 200 OK
+	X-elastic-product: Elasticsearch
+	content-type: ${ctype}
+	
 	EOF
   exit 0
 
@@ -237,15 +245,15 @@ 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
+	HTTP/1.1 200 OK
+	X-elastic-product: Elasticsearch
+	content-type: ${ctype}
+	
 	EOF
 	
   if [ "$PATH_INFO" = "/${_INDEX}/" ]; then
-  sed 's;$;\r;' <<-EOF
+  cat <<-EOF
 	{ $(json_dump str:"${_INDEX}"): {
 	    "aliases":{},
 	    "mappings": {
@@ -276,7 +284,7 @@ elif [ "$REQUEST_METHOD" = "GET" ]; then
 	}
 	EOF
   else
-    sed 's;$;\r;' <<-EOF
+    cat <<-EOF
 	{ "name" : "head",
 	  "cluster_name" : "elasticsearch",
 	  "version" : {
-- 
2.39.2