feat(perl): add script commands trigger

[Abhishek84313]

What changes are being made and why?

Adds ScriptTrigger and CommandsTrigger to plugin-script-perl, so a flow can be triggered when a Perl script or set of Perl commands completes, with an exit condition deciding whether the trigger fires.

plugin-script-perl was the last of the scripting submodules missing both. This follows the existing implementations in plugin-script-go, plugin-script-node, plugin-script-python, plugin-script-ruby, plugin-script-shell and plugin-script-r — same package as the task classes, no nested subpackage, and the same property set (containerImage, script/commands, exitCondition, interval, edge).

Both triggers implement PollingTriggerInterface and emit an output carrying the timestamp, the rendered condition, the exit code and any structured vars; execution id, namespace and flow id come from TriggerService.generateExecution. exitCondition accepts either exit N (compared against the exit code) or a regex, with a substring fallback when the regex is invalid, matched against emitted vars and failure logs. edge mode (default true) emits only on a transition from not-matching to matching, so a persistently failing script does not fire on every poll.

Part of #313. Closes #429.

One deviation from the issue worth flagging for review: #429 proposes commandPattern / argumentsPattern / exitCondition, but every existing submodule uses the containerImage / script-or-commands / exitCondition / interval / edge shape, where the trigger runs the script rather than matching a pattern against externally observed executions. I followed the established convention for consistency. If the schema in the issue is the intended direction, it is a change across all submodules and is probably better settled separately.

---

How the changes have been QAed?

ScriptTrigger — fires when the script exits non-zero:

id: perl_script_trigger
namespace: company.team

triggers:
  - id: on_perl_fail
    type: io.kestra.plugin.scripts.perl.ScriptTrigger
    interval: PT10S
    exitCondition: "exit 1"
    edge: true
    containerImage: perl
    script: |
      print "about to fail\n";
      exit 1;

tasks:
  - id: log
    type: io.kestra.plugin.core.log.Log
    message: "Triggered with exitCode={{ trigger.exitCode }} (condition={{ trigger.condition }})"
id: perl_commands_trigger
namespace: company.team

triggers:
  - id: on_perl_commands
    type: io.kestra.plugin.scripts.perl.CommandsTrigger
    interval: PT10S
    exitCondition: "toto"
    edge: true
    containerImage: perl
    commands:
      - echo '::{"outputs":{"listing":"toto"}}::'

tasks:
  - id: log
    type: io.kestra.plugin.core.log.Log
    message: "Triggered with vars={{ trigger.vars }}"

[jymaire]

thanks for the PR, it seems you also did #432 in the same time? could you remove the commit and open its own PR? (and comment on issue so I can assign you)

[jymaire]

Hi @Abhishek84313 I dropped your modification on R script, as they already are on another PR

[jymaire]

Two inline findings from review: a blocking ReDoS gap in matchesCondition and a note on non-persisted edge state.

[jymaire]

QA report — Perl ScriptTrigger / CommandsTrigger

Edition: OSS | Docker tag: v1.3.39 | PR head: 0ba7de7
Instance: http://gmknr.kestra.docker.localhost:1355/ (local, left running for manual review — docker rm -f kestra-gmknr to stop)
Screenshots: https://claude.ai/artifact/ANisDPSWMPGSZX9krfS9do

Summary

#	Flow	Covers	Result
1	script_trigger	ScriptTrigger @Example, exit 1, edge mode	✅ 1 execution, no re-fire after restart
2	commands_trigger	CommandsTrigger @Example (as fixed in f613113), edge mode	✅ 1 execution, no re-fire after restart
3	commands_trigger_vars	CommandsTrigger substring match on emitted vars	✅ 1 execution, vars={"listing":"toto"}
4	script_trigger_redos	Catastrophic regex (.*a){20}$	✅ never fires, scheduler keeps polling ⚠️ see note
5	perl_script_nonreg	Non-regression: Perl Script task	✅ SUCCESS, answer=42

Edge mode, before vs after 0ba7de7 (interval: PT10S, condition always true):

• f613113 (in-memory lastMatched): flows 1 and 3 fired on every poll — 22 executions in ~4.5 min each. The original @Example for flow 2 (perl missing.pl) never fired; fixed in f613113.
• 0ba7de7 (namespace KV store): with the edge keys cleared and flows 1–3 enabled, each fired once in ~90 s. After docker restart and a further ~90 s of polling, no new executions. The KV keys trigger-edge-<len>-<flow>-<trigger> kept updating on every poll.

Flow 1: script_trigger (✅)

Flow YAML

id: script_trigger
namespace: company.team

triggers:
  - id: script_failure
    type: io.kestra.plugin.scripts.perl.ScriptTrigger
    interval: PT10S
    exitCondition: "exit 1"
    edge: true
    containerImage: perl
    script: |
      # This fails with a non-zero exit code.
      exit 1;

tasks:
  - id: log
    type: io.kestra.plugin.core.log.Log
    message: "Triggered with exitCode={{ trigger.exitCode }} (condition={{ trigger.condition }})"

Executions (screenshot): 1 execution (SUCCESS, 0.49 s) after enabling on 0ba7de7, none after restart. Earlier rows are the f613113 build firing every ~10 s.

Logs synthesis: Triggered with exitCode=1 (condition=exit 1).

Trigger outputs: exitCode: 1, condition: exit 1, timestamp, no vars (failed run).

Flow 2: commands_trigger (✅)

Flow YAML

id: commands_trigger
namespace: company.team

triggers:
  - id: commands_failure
    type: io.kestra.plugin.scripts.perl.CommandsTrigger
    interval: PT10S
    exitCondition: "exit 1"
    edge: true
    containerImage: perl
    commands:
      - perl -e 'exit 1'

tasks:
  - id: log
    type: io.kestra.plugin.core.log.Log
    message: "Triggered with exitCode={{ trigger.exitCode }} (condition={{ trigger.condition }})"

Executions (screenshot): 1 execution (SUCCESS, 0.53 s) on 0ba7de7, none after restart.

Trigger outputs: exitCode: 1, condition: exit 1.

Flow 3: commands_trigger_vars (✅)

Flow YAML

id: commands_trigger_vars
namespace: company.team

triggers:
  - id: on_perl_commands
    type: io.kestra.plugin.scripts.perl.CommandsTrigger
    interval: PT10S
    exitCondition: "toto"
    edge: true
    containerImage: perl
    commands:
      - echo '::{"outputs":{"listing":"toto"}}::'

tasks:
  - id: log
    type: io.kestra.plugin.core.log.Log
    message: "Triggered with vars={{ trigger.vars }}"

Executions (screenshot): 1 execution on 0ba7de7, none after restart.

Logs synthesis (screenshot): Triggered with vars={"listing":"toto"}.

Trigger outputs: exitCode: 0, condition: toto, vars: {listing: toto}.

Edge state (screenshot): one KV key per trigger in company.team, updated after the restart.

Flow 4: script_trigger_redos (✅ ⚠️)

Flow YAML

id: script_trigger_redos
namespace: company.team

triggers:
  - id: redos
    type: io.kestra.plugin.scripts.perl.ScriptTrigger
    interval: PT10S
    exitCondition: "(.*a){20}$"
    edge: true
    containerImage: perl
    script: |
      print '::{"outputs":{"k":"' . ('a' x 40) . '!"}}::' . "\n";

tasks:
  - id: log
    type: io.kestra.plugin.core.log.Log
    message: "SHOULD NOT FIRE vars={{ trigger.vars }}"

Executions (screenshot): none — the 5 s timeout fires and the substring fallback does not match, so the scheduler is not blocked.

⚠️ CompletableFuture.get(5s) does not stop the matcher: each poll leaves a runaway ForkJoinPool.commonPool thread (container CPU climbed to ~650 % within minutes). Same behaviour as the Ruby guard on main; tracked family-wide in #450. Tested on f613113 — the regex guard is unchanged in 0ba7de7.

Flow 5: perl_script_nonreg (✅)

Flow YAML

id: perl_script_nonreg
namespace: company.team

tasks:
  - id: perl
    type: io.kestra.plugin.scripts.perl.Script
    containerImage: perl
    script: |
      print "hello from perl $]\n";
      print '::{"outputs":{"answer":42}}::' . "\n";

Gantt (screenshot)

Task	Status	Duration
perl	SUCCESS	0.58s
Total	SUCCESS	0.58s

Logs synthesis: hello from perl 5.044000, Command succeed with exit code 0, no WARN/ERROR.

Outputs synthesis (screenshot): vars.answer: 42, exitCode: 0.

Timeouts

None.

Tests

./gradlew :plugin-script-perl:test --rerun-tasks on 0ba7de7: 53 passed, 0 failed.

[jymaire]

I edited PR to apply fixes so the PR doesn't stay stale. Thanks for your contribution!

[Abhishek84313]

@jymaire Thank you