feat(runs): state machine, idempotent resume, reaper extension, retention

Pipeline v2 Phase 1, commit 4/4. Run_Stage_State persists per-run
per-stage state keyed run_id + stage + agent_role + item_key (item_key
scopes articles because one run_id spans a whole batch; agent_role is
the Phase-2 quorum dimension, schema-only). Done stages snapshot output
and are reused on re-entry — writer stages via
Content_Generator::set_run_item, the review via its stored verdict, a
published item skipped outright — never re-run, never re-charged. Post
creation is keyed by _prautoblogger_run_id + new _prautoblogger_idea_hash
(check-before-insert in Publisher) so retries cannot duplicate posts.
Run-level research/analysis checkpoints + done/failed lifecycle marks
ride Pipeline_Runner/Executor (incl. orphan-abort + budget-abort paths).

Run_Reaper rides the EXISTING #19 daily cron: stages stuck running >2x
expected stage wall-clock and runs silent >2x expected run wall-clock ->
failed (filterable), open stages reaped, queryable as 'incomplete'.
Retention: request_json + stage output payloads NULLed after R days, R
from the new Audit Payload Retention setting (default constant 14,
0 = keep forever, no literal at point of use). Audit_Writer records
editorial verdicts (run_decisions) and per-article kept sources
(run_sources) for new runs.

Under-300: status-transient/summary helpers extracted verbatim to
Pipeline_Status (Pipeline_Runner 271; removed the worker's duplicate).

Agent-Session: pipeline-v2-phase1-engineer-20260612
This commit is contained in:
peptiderepo
2026-06-11 04:31:55 +00:00
parent 04f09f5f24
commit ed8ec9425f
17 changed files with 1358 additions and 102 deletions

View File

@@ -19,6 +19,12 @@ class PRAutoBlogger_Content_Generator {
private PRAutoBlogger_LLM_Provider_Interface $llm;
private PRAutoBlogger_Cost_Tracker $cost_tracker;
/** @var string|null Run UUID for stage-state checkpointing (null = off). */
private ?string $run_id = null;
/** @var string|null Stage item key scoping this article within the run. */
private ?string $item_key = null;
public function __construct(
PRAutoBlogger_LLM_Provider_Interface $llm,
PRAutoBlogger_Cost_Tracker $cost_tracker
@@ -27,6 +33,25 @@ class PRAutoBlogger_Content_Generator {
$this->cost_tracker = $cost_tracker;
}
/**
* Enable per-stage state checkpointing for one article item (v0.18.0).
*
* When set, every LLM stage (outline/draft/polish or the single-pass
* draft) records pending->running->done state and snapshots its output;
* a re-entered done stage returns the snapshot WITHOUT a new LLM call
* (never re-charged). When unset (default), behavior is exactly
* pre-v0.18.0 — used by callers outside a run (e.g. eval mode).
*
* @param string|null $run_id Run UUID, or null to disable.
* @param string|null $item_key Item key from Run_Stage_State::item_key_for_idea().
* @return $this
*/
public function set_run_item( ?string $run_id, ?string $item_key ): self {
$this->run_id = $run_id;
$this->item_key = $item_key;
return $this;
}
/**
* Generate a blog post from an article idea.
*
@@ -159,6 +184,19 @@ class PRAutoBlogger_Content_Generator {
array $options,
?string $prompt_key = null
): string {
// Idempotent resume: a done stage returns its snapshot, no LLM call.
if ( null !== $this->run_id && null !== $this->item_key ) {
$cached = PRAutoBlogger_Run_Stage_State::get_output( $this->run_id, $stage, '', $this->item_key );
if ( null !== $cached ) {
PRAutoBlogger_Logger::instance()->info(
sprintf( 'Reusing completed %s stage output (idempotent resume).', $stage ),
'pipeline'
);
return $cached;
}
PRAutoBlogger_Run_Stage_State::start( $this->run_id, $stage, '', $this->item_key );
}
$ctx = PRAutoBlogger_Opik_Trace_Context::current();
$span_id = $ctx->start_span(
@@ -209,6 +247,17 @@ class PRAutoBlogger_Content_Generator {
$prompt_key
);
if ( null !== $this->run_id && null !== $this->item_key ) {
PRAutoBlogger_Run_Stage_State::done(
$this->run_id,
$stage,
'',
$this->item_key,
(string) $response['content'],
$this->llm->estimate_cost( $response['model'], (int) $response['prompt_tokens'], (int) $response['completion_tokens'] )
);
}
return $response['content'];
}