use std::collections::HashMap;
use std::sync::Arc;
use tokio::sync::RwLock;
use tokio_cron_scheduler::{Job, JobScheduler};
use uuid::Uuid;
use crate::pipeline::engines::BasicPipelineEngine;
use crate::pipeline::interface::PipelineEngine;
use crate::pipeline::model::PipelineContext;
use crate::platform::adapters::data::DataAdapter;
use crate::platform::model::PipelineInvocationEntry;
use crate::platform::services::pipeline_hits::PipelineHitsService;
use crate::platform::services::pipeline_runtime::PipelineRuntimeService;
use crate::platform::services::project_config::ProjectConfigurationService;
pub struct PipelineScheduler {
sched: Arc<JobScheduler>,
runtime: Arc<PipelineRuntimeService>,
engine: Arc<BasicPipelineEngine>,
hits: Arc<PipelineHitsService>,
data: Arc<dyn DataAdapter>,
zebflow_cfg: Arc<ProjectConfigurationService>,
jobs: Arc<RwLock<HashMap<String, Uuid>>>,
}
impl PipelineScheduler {
pub async fn start(
runtime: Arc<PipelineRuntimeService>,
engine: Arc<BasicPipelineEngine>,
hits: Arc<PipelineHitsService>,
data: Arc<dyn DataAdapter>,
zebflow_cfg: Arc<ProjectConfigurationService>,
) -> Result<Arc<Self>, Box<dyn std::error::Error + Send + Sync>> {
let sched = JobScheduler::new().await?;
sched.start().await?;
let scheduler = Arc::new(Self {
sched: Arc::new(sched),
runtime,
engine,
hits,
data,
zebflow_cfg,
jobs: Arc::new(RwLock::new(HashMap::new())),
});
scheduler.register_all().await;
Ok(scheduler)
}
pub async fn register_all(&self) {
for pipeline in self.runtime.list_all() {
for trigger in &pipeline.schedule_triggers {
self.register_job(
&pipeline.owner,
&pipeline.project,
&pipeline.file_rel_path,
&pipeline.graph.id,
&trigger.node_id,
&trigger.cron,
)
.await;
}
}
}
pub async fn sync_pipeline(&self, owner: &str, project: &str, file_rel_path: &str) {
let key_prefix = format!("{}/{}/{}", owner, project, file_rel_path);
let stale: Vec<(String, Uuid)> = {
let jobs = self.jobs.read().await;
jobs.iter()
.filter(|(k, _)| k.starts_with(&key_prefix))
.map(|(k, v)| (k.clone(), *v))
.collect()
};
for (key, uuid) in stale {
let _ = self.sched.remove(&uuid).await;
self.jobs.write().await.remove(&key);
println!("Scheduler: removed job {}", key);
}
if let Some(pipeline) = self.runtime.get(owner, project, file_rel_path) {
for trigger in &pipeline.schedule_triggers {
self.register_job(
owner,
project,
file_rel_path,
&pipeline.graph.id,
&trigger.node_id,
&trigger.cron,
)
.await;
}
}
}
async fn register_job(
&self,
owner: &str,
project: &str,
file_rel_path: &str,
graph_id: &str,
node_id: &str,
cron: &str,
) {
if cron.trim().is_empty() {
eprintln!(
"Scheduler: skipping {}/{}/{} — empty cron expression",
owner, project, file_rel_path
);
return;
}
let job_key = format!("{}/{}/{}:{}", owner, project, file_rel_path, node_id);
let runtime = self.runtime.clone();
let engine = self.engine.clone();
let hits = self.hits.clone();
let data = self.data.clone();
let zebflow_cfg_arc = self.zebflow_cfg.clone();
let owner_s = owner.to_string();
let project_s = project.to_string();
let file_rel_path_s = file_rel_path.to_string();
let graph_id_s = graph_id.to_string();
let node_id_s = node_id.to_string();
let job_result = Job::new_async(cron, move |_uuid, _sched| {
let runtime = runtime.clone();
let engine = engine.clone();
let hits = hits.clone();
let data = data.clone();
let zebflow_cfg = zebflow_cfg_arc.clone();
let owner = owner_s.clone();
let project = project_s.clone();
let file_rel_path = file_rel_path_s.clone();
let graph_id = graph_id_s.clone();
let node_id = node_id_s.clone();
Box::pin(async move {
println!("⏰ Schedule tick: {}/{}/{}", owner, project, file_rel_path);
let Some(compiled) = runtime.get(&owner, &project, &file_rel_path) else {
eprintln!(
"Scheduler: pipeline no longer active — {}/{}/{}",
owner, project, file_rel_path
);
return;
};
let project_cfg = match zebflow_cfg.read_or_default(&owner, &project) {
Ok(config) => config,
Err(err) => {
eprintln!(
"Scheduler: invalid project configuration for {}/{}: {}",
owner, project, err
);
return;
}
};
let pipeline_retention = compiled
.graph
.metadata
.as_ref()
.and_then(|metadata| metadata.settings.invocation_retention.as_ref());
let log_max_n = pipeline_retention
.and_then(|retention| retention.max_invocations)
.map(|value| value.max(1) as usize)
.unwrap_or_else(|| {
project_cfg
.configs
.pipelines
.logging
.effective_max_invocations()
});
let max_age_secs = pipeline_retention
.and_then(|retention| retention.max_age_secs)
.map(|value| value.max(1) as i64);
let fired_at = chrono::Utc::now();
let ctx = PipelineContext {
owner: owner.clone(),
project: project.clone(),
pipeline: graph_id.clone(),
request_id: format!("schedule-{}", Uuid::new_v4()),
route: String::new(),
input: serde_json::json!({
"trigger": "schedule",
"fired_at": fired_at.to_rfc3339(),
"node_id": node_id,
}),
trigger: None,
placeholder: None,
};
let exec_start = std::time::Instant::now();
match engine.execute_async(&compiled.graph, &ctx).await {
Ok(output) => {
let duration_ms = exec_start.elapsed().as_millis() as u64;
hits.record_success(&owner, &project, &file_rel_path);
let _ = data.log_pipeline_invocation(
&owner,
&project,
&file_rel_path,
&PipelineInvocationEntry {
run_id: ctx.request_id.clone(),
at: fired_at.timestamp(),
duration_ms,
status: "ok".to_string(),
trigger: "schedule".to_string(),
error: None,
trace: output.node_trace,
},
log_max_n,
max_age_secs,
);
println!("✅ Schedule OK: {}/{}/{}", owner, project, file_rel_path);
}
Err(e) => {
let duration_ms = exec_start.elapsed().as_millis() as u64;
hits.record_failure(
&owner,
&project,
&file_rel_path,
"schedule",
e.code,
&e.message,
);
let _ = data.log_pipeline_invocation(
&owner,
&project,
&file_rel_path,
&PipelineInvocationEntry {
run_id: ctx.request_id.clone(),
at: fired_at.timestamp(),
duration_ms,
status: "error".to_string(),
trigger: "schedule".to_string(),
error: Some(e.message.clone()),
trace: e.node_trace.clone(),
},
log_max_n,
max_age_secs,
);
eprintln!(
"❌ Schedule failed {}/{}/{}: {}",
owner, project, file_rel_path, e
);
}
}
})
});
match job_result {
Ok(job) => match self.sched.add(job).await {
Ok(uuid) => {
self.jobs.write().await.insert(job_key.clone(), uuid);
println!("✅ Scheduler: registered job {} cron={}", job_key, cron);
}
Err(e) => eprintln!("Scheduler: failed to add job {}: {}", job_key, e),
},
Err(e) => eprintln!("Scheduler: invalid cron '{}' for {}: {}", cron, job_key, e),
}
}
}