Using logstash to strip XSSI prefix from JSON response - json

Question

I have a fairly simple problem but it's confusing to me. I'm trying to use Logstash to get Gerrit data via rest api. I'm using http_poller and I get a right response with my configuration, so I'm almost there.
Now I need to strip the XSSI prefix )]}' from the start of Gerrits JSON response. The question is, how? How to strip or split or mutate it, or how should I proceed?
My input configuration:
input {
http_poller {
urls => {
gerrit_projects => {
method => get
url => "http://url.to/gerrit/a/projects/"
headers => { Accept => "application/json" }
auth => { user => "userid" password => "supresecret" }
}
}
target => "http_poller_data"
metadata_target => "http_poller_metadata"
request_timeout => 60
interval => 60
}
}
filter {
if [http_poller_metadata] {
mutate {
add_field => {
"http_poller_host" => "%{http_poller_metadata[host]}"
"http_poller" => "%{http_poller_metadata[name]}"
}
}
}
if [http_poller_metadata][runtime_seconds] and [http_poller_metadata][runtime_seconds] > 0.5 {
mutate { add_tag => "slow_request" }
}
if [http_request_failure] or [http_poller_metadata][code] != 200 {
mutate { add_tag => "bad_request" }
}
}
output {
stdout { codec => rubydebug }
}
And parts of the response:
Pipeline main started
JSON parse failure. Falling back to plain-text {:error=>#<LogStash::Json::ParserError: Unexpected character (')' (code 41)): expected a valid value (number, String, array, object, 'true', 'false' or 'null')
at ... (bunch of lines)...
{
"http_poller_data" => {
"message" => ")]}'\n{\"All-Users\":{\"id\":\"All-Users\",....(more valid JSON)...",
"tags" => [
[0] "_jsonparsefailure"
],
"#version" => "1",
"#timestamp" => "2016-12-13T09:48:25.397Z"
},
"#version" => "1",
"#timestamp" => "2016-12-13T09:48:25.397Z",
"http_poller_metadata" => { ... }
This is my first question to StackOverflow. Thank you for being kind with your answers!

Answer 1

You can use the mutate filter with the gsub option (link) to remove the )]}
mutate {
gsub => [
"message", "\)]}'", ""
]
}
But the gsub replace all occurences of a regex, so you have to be sure that the pattern only appears once.

Answer 2

I use "sed 1d" to remove the ")]}'" prefix and "jq" to process the JSON output. For example, to get the state of a Gerrit project I execute:
curl -s --header 'Content-Type:application/json' --request GET --netrc https://<GERRIT-SERVER>/a/projects/?r=<GERRIT-PROJECT> | sed 1d | jq --raw-output ".[] | .state"
ACTIVE