Specific GROK filter for multi-line Postgresql log

Elastic Stack Logstash
8

I expect the next:

event 1: ...[137-1] and [137-2]...
event 2: ...[2953-1]...
event 3: ...[3779-1]...
event 4: ...[138-1] and [138-2]...
event 5: ...[3780-1]...
event 6: ...[139-1]...
event 7: ...[2954-1] and [2954-2]...

Task_id in this case may be [24977] or 137.

My friend help me with ruby code and I made such a configuration:

filter {
grok {
match => [ "message", "%{SYSLOGTIMESTAMP:timestamp} %{SYSLOGHOST:logsource} %{SYSLOGPROG}: [%{INT:line}-%{INT:part_of_line}] %{GREEDYDATA:ostatok}" ]
}

aggregate {
        task_id => "%{line}"
         code =>  "map['full_message'] ||= '' ; map['full_message'] += event['ostatok'] ; event['full_message'] ||= '' ; event['full_message'] = map['full_message']"
       }

}
But I have next event on elasticsearch:

event 1: ...[137-1]...
event 2: ...[137-1] and [137-2]...
event 3: ...[2953-1]...
event 4: ...[3779-1]...
event 5: ...[138-1]...
event 6: ...[138-1] and [138-2]...
event 7: ...[3780-1]...
event 8: ...[139-1]...
event 9: ...[2954-1]...
event 10: ...[2954-1] and [2954-2]...
How I can clean my events from event 1: ...[137-1]..., event 5: ...[138-1]..., event 7: ...[2954-1]... ?