k-photos/crates/adapters/postgres/src/processing/mod.rs

use crate::helpers::{MapDomainError, pg_repo};
use async_trait::async_trait;
use chrono::{DateTime, Utc};
use domain::{
    entities::{
        BatchStatus, Job, JobBatch, JobStatus, JobType, PipelineStep, Plugin, PluginType,
        ProcessingPipeline,
    },
    errors::DomainError,
    ports::{JobBatchRepository, JobRepository, PipelineRepository, PluginRepository},
    value_objects::{DateTimeStamp, MetadataValue, StructuredData, SystemId},
};
use uuid::Uuid;

// ──────────────────────────────────────────────
// Job
// ──────────────────────────────────────────────

#[derive(sqlx::FromRow)]
struct JobRow {
    job_id: Uuid,
    job_type: String,
    target_asset_id: Option<Uuid>,
    batch_id: Option<Uuid>,
    status: String,
    priority: i32,
    payload: serde_json::Value,
    result_data: Option<serde_json::Value>,
    retry_count: i32,
    max_retries: i32,
    created_at: DateTime<Utc>,
    started_at: Option<DateTime<Utc>>,
    completed_at: Option<DateTime<Utc>>,
    error_message: Option<String>,
}

fn job_type_from_str(s: &str) -> JobType {
    match s {
        "scan_directory" => JobType::ScanDirectory,
        "extract_metadata" => JobType::ExtractMetadata,
        "generate_derivative" => JobType::GenerateDerivative,
        "sync_sidecar" => JobType::SyncSidecar,
        "detect_duplicates" => JobType::DetectDuplicates,
        other => JobType::Custom(other.to_string()),
    }
}

fn job_type_to_str(t: &JobType) -> String {
    match t {
        JobType::ScanDirectory => "scan_directory".to_string(),
        JobType::ExtractMetadata => "extract_metadata".to_string(),
        JobType::GenerateDerivative => "generate_derivative".to_string(),
        JobType::SyncSidecar => "sync_sidecar".to_string(),
        JobType::DetectDuplicates => "detect_duplicates".to_string(),
        JobType::Custom(s) => s.clone(),
    }
}

fn job_status_from_str(s: &str) -> JobStatus {
    match s {
        "queued" => JobStatus::Queued,
        "processing" => JobStatus::Processing,
        "completed" => JobStatus::Completed,
        "failed" => JobStatus::Failed,
        "cancelled" => JobStatus::Cancelled,
        _ => JobStatus::Queued,
    }
}

fn job_status_to_str(s: &JobStatus) -> &'static str {
    match s {
        JobStatus::Queued => "queued",
        JobStatus::Processing => "processing",
        JobStatus::Completed => "completed",
        JobStatus::Failed => "failed",
        JobStatus::Cancelled => "cancelled",
    }
}

fn structured_from_json(v: serde_json::Value) -> StructuredData {
    if let serde_json::Value::Object(map) = v {
        let mut sd = StructuredData::new();
        for (k, val) in map {
            sd.insert(k, MetadataValue::from(val));
        }
        sd
    } else {
        StructuredData::new()
    }
}

fn structured_to_json(sd: &StructuredData) -> serde_json::Value {
    let map: serde_json::Map<String, serde_json::Value> = sd
        .inner()
        .iter()
        .map(|(k, v)| (k.clone(), serde_json::Value::from(v)))
        .collect();
    serde_json::Value::Object(map)
}

impl From<JobRow> for Job {
    fn from(r: JobRow) -> Self {
        Self {
            job_id: SystemId::from_uuid(r.job_id),
            job_type: job_type_from_str(&r.job_type),
            target_asset_id: r.target_asset_id.map(SystemId::from_uuid),
            batch_id: r.batch_id.map(SystemId::from_uuid),
            status: job_status_from_str(&r.status),
            priority: r.priority as u32,
            payload: structured_from_json(r.payload),
            result_data: r.result_data.map(structured_from_json),
            retry_count: r.retry_count as u32,
            max_retries: r.max_retries as u32,
            created_at: DateTimeStamp::from_datetime(r.created_at),
            started_at: r.started_at.map(DateTimeStamp::from_datetime),
            completed_at: r.completed_at.map(DateTimeStamp::from_datetime),
            error_message: r.error_message,
        }
    }
}

pg_repo!(PostgresJobRepository);

#[async_trait]
impl JobRepository for PostgresJobRepository {
    async fn find_by_id(&self, id: &SystemId) -> Result<Option<Job>, DomainError> {
        let row = sqlx::query_as::<_, JobRow>(
            "SELECT job_id, job_type, target_asset_id, batch_id, status, priority,
                    payload, result_data, retry_count, max_retries, created_at,
                    started_at, completed_at, error_message
             FROM jobs WHERE job_id = $1",
        )
        .bind(*id.as_uuid())
        .fetch_optional(&self.pool)
        .await
        .map_pg()?;

        Ok(row.map(Into::into))
    }

    async fn find_next_queued(&self) -> Result<Option<Job>, DomainError> {
        let row = sqlx::query_as::<_, JobRow>(
            "SELECT job_id, job_type, target_asset_id, batch_id, status, priority,
                    payload, result_data, retry_count, max_retries, created_at,
                    started_at, completed_at, error_message
             FROM jobs WHERE status = 'queued'
             ORDER BY priority DESC, created_at ASC
             LIMIT 1
             FOR UPDATE SKIP LOCKED",
        )
        .fetch_optional(&self.pool)
        .await
        .map_pg()?;

        Ok(row.map(Into::into))
    }

    async fn claim_next(&self) -> Result<Option<Job>, DomainError> {
        let row = sqlx::query_as::<_, JobRow>(
            "UPDATE jobs SET status = 'processing', started_at = NOW()
             WHERE job_id = (
                 SELECT job_id FROM jobs
                 WHERE status = 'queued'
                 ORDER BY priority DESC, created_at ASC
                 LIMIT 1
                 FOR UPDATE SKIP LOCKED
             )
             RETURNING job_id, job_type, target_asset_id, batch_id, status, priority,
                       payload, result_data, retry_count, max_retries, created_at,
                       started_at, completed_at, error_message",
        )
        .fetch_optional(&self.pool)
        .await
        .map_pg()?;

        Ok(row.map(Into::into))
    }

    async fn find_all(
        &self,
        status: Option<&str>,
        limit: u32,
        offset: u32,
    ) -> Result<Vec<Job>, DomainError> {
        let rows = match status {
            Some(s) => sqlx::query_as::<_, JobRow>(
                "SELECT job_id, job_type, target_asset_id, batch_id, status, priority,
                            payload, result_data, retry_count, max_retries, created_at,
                            started_at, completed_at, error_message
                     FROM jobs WHERE status = $1
                     ORDER BY created_at DESC
                     LIMIT $2 OFFSET $3",
            )
            .bind(s)
            .bind(limit as i64)
            .bind(offset as i64)
            .fetch_all(&self.pool)
            .await
            .map_pg()?,
            None => sqlx::query_as::<_, JobRow>(
                "SELECT job_id, job_type, target_asset_id, batch_id, status, priority,
                            payload, result_data, retry_count, max_retries, created_at,
                            started_at, completed_at, error_message
                     FROM jobs
                     ORDER BY created_at DESC
                     LIMIT $1 OFFSET $2",
            )
            .bind(limit as i64)
            .bind(offset as i64)
            .fetch_all(&self.pool)
            .await
            .map_pg()?,
        };
        Ok(rows.into_iter().map(Into::into).collect())
    }

    async fn count(&self, status: Option<&str>) -> Result<u64, DomainError> {
        let count: (i64,) = match status {
            Some(s) => sqlx::query_as("SELECT COUNT(*) FROM jobs WHERE status = $1")
                .bind(s)
                .fetch_one(&self.pool)
                .await
                .map_pg()?,
            None => sqlx::query_as("SELECT COUNT(*) FROM jobs")
                .fetch_one(&self.pool)
                .await
                .map_pg()?,
        };
        Ok(count.0 as u64)
    }

    async fn find_by_batch(&self, batch_id: &SystemId) -> Result<Vec<Job>, DomainError> {
        let rows = sqlx::query_as::<_, JobRow>(
            "SELECT job_id, job_type, target_asset_id, batch_id, status, priority,
                    payload, result_data, retry_count, max_retries, created_at,
                    started_at, completed_at, error_message
             FROM jobs WHERE batch_id = $1
             ORDER BY created_at ASC",
        )
        .bind(*batch_id.as_uuid())
        .fetch_all(&self.pool)
        .await
        .map_pg()?;

        Ok(rows.into_iter().map(Into::into).collect())
    }

    async fn save(&self, job: &Job) -> Result<(), DomainError> {
        sqlx::query(
            "INSERT INTO jobs (job_id, job_type, target_asset_id, batch_id, status, priority,
                    payload, result_data, retry_count, max_retries, created_at,
                    started_at, completed_at, error_message)
             VALUES ($1, $2, $3, $4, $5, $6, $7, $8, $9, $10, $11, $12, $13, $14)
             ON CONFLICT (job_id) DO UPDATE SET
               status = EXCLUDED.status,
               priority = EXCLUDED.priority,
               payload = EXCLUDED.payload,
               result_data = EXCLUDED.result_data,
               retry_count = EXCLUDED.retry_count,
               started_at = EXCLUDED.started_at,
               completed_at = EXCLUDED.completed_at,
               error_message = EXCLUDED.error_message",
        )
        .bind(*job.job_id.as_uuid())
        .bind(job_type_to_str(&job.job_type))
        .bind(job.target_asset_id.as_ref().map(|id| *id.as_uuid()))
        .bind(job.batch_id.as_ref().map(|id| *id.as_uuid()))
        .bind(job_status_to_str(&job.status))
        .bind(job.priority as i32)
        .bind(structured_to_json(&job.payload))
        .bind(job.result_data.as_ref().map(structured_to_json))
        .bind(job.retry_count as i32)
        .bind(job.max_retries as i32)
        .bind(job.created_at.as_datetime())
        .bind(job.started_at.as_ref().map(|d| d.as_datetime()))
        .bind(job.completed_at.as_ref().map(|d| d.as_datetime()))
        .bind(&job.error_message)
        .execute(&self.pool)
        .await
        .map_pg()?;
        Ok(())
    }
}

// ──────────────────────────────────────────────
// JobBatch
// ──────────────────────────────────────────────

#[derive(sqlx::FromRow)]
struct BatchRow {
    batch_id: Uuid,
    batch_type: String,
    total_jobs: i32,
    completed_count: i32,
    failed_count: i32,
    status: String,
}

fn batch_status_from_str(s: &str) -> BatchStatus {
    match s {
        "in_progress" => BatchStatus::InProgress,
        "completed_with_errors" => BatchStatus::CompletedWithErrors,
        "completed" => BatchStatus::Completed,
        "cancelled" => BatchStatus::Cancelled,
        _ => BatchStatus::InProgress,
    }
}

fn batch_status_to_str(s: &BatchStatus) -> &'static str {
    match s {
        BatchStatus::InProgress => "in_progress",
        BatchStatus::CompletedWithErrors => "completed_with_errors",
        BatchStatus::Completed => "completed",
        BatchStatus::Cancelled => "cancelled",
    }
}

impl From<BatchRow> for JobBatch {
    fn from(r: BatchRow) -> Self {
        Self {
            batch_id: SystemId::from_uuid(r.batch_id),
            batch_type: r.batch_type,
            total_jobs: r.total_jobs as u32,
            completed_count: r.completed_count as u32,
            failed_count: r.failed_count as u32,
            status: batch_status_from_str(&r.status),
        }
    }
}

pg_repo!(PostgresJobBatchRepository);

#[async_trait]
impl JobBatchRepository for PostgresJobBatchRepository {
    async fn find_by_id(&self, id: &SystemId) -> Result<Option<JobBatch>, DomainError> {
        let row = sqlx::query_as::<_, BatchRow>(
            "SELECT batch_id, batch_type, total_jobs, completed_count, failed_count, status
             FROM job_batches WHERE batch_id = $1",
        )
        .bind(*id.as_uuid())
        .fetch_optional(&self.pool)
        .await
        .map_pg()?;

        Ok(row.map(Into::into))
    }

    async fn save(&self, batch: &JobBatch) -> Result<(), DomainError> {
        sqlx::query(
            "INSERT INTO job_batches (batch_id, batch_type, total_jobs, completed_count, failed_count, status)
             VALUES ($1, $2, $3, $4, $5, $6)
             ON CONFLICT (batch_id) DO UPDATE SET
               total_jobs = EXCLUDED.total_jobs,
               completed_count = EXCLUDED.completed_count,
               failed_count = EXCLUDED.failed_count,
               status = EXCLUDED.status",
        )
        .bind(*batch.batch_id.as_uuid())
        .bind(&batch.batch_type)
        .bind(batch.total_jobs as i32)
        .bind(batch.completed_count as i32)
        .bind(batch.failed_count as i32)
        .bind(batch_status_to_str(&batch.status))
        .execute(&self.pool)
        .await
        .map_pg()?;
        Ok(())
    }
}

// ──────────────────────────────────────────────
// Plugin
// ──────────────────────────────────────────────

#[derive(sqlx::FromRow)]
struct PluginRow {
    plugin_id: Uuid,
    name: String,
    plugin_type: String,
    is_enabled: bool,
    configuration: serde_json::Value,
}

fn plugin_type_from_str(s: &str) -> PluginType {
    match s {
        "media_processor" => PluginType::MediaProcessor,
        "scheduled_task" => PluginType::ScheduledTask,
        "sidecar_writer" => PluginType::SidecarWriter,
        _ => PluginType::MediaProcessor,
    }
}

fn plugin_type_to_str(t: &PluginType) -> &'static str {
    match t {
        PluginType::MediaProcessor => "media_processor",
        PluginType::ScheduledTask => "scheduled_task",
        PluginType::SidecarWriter => "sidecar_writer",
    }
}

impl From<PluginRow> for Plugin {
    fn from(r: PluginRow) -> Self {
        Self {
            plugin_id: SystemId::from_uuid(r.plugin_id),
            name: r.name,
            plugin_type: plugin_type_from_str(&r.plugin_type),
            is_enabled: r.is_enabled,
            configuration: structured_from_json(r.configuration),
        }
    }
}

pg_repo!(PostgresPluginRepository);

#[async_trait]
impl PluginRepository for PostgresPluginRepository {
    async fn find_by_id(&self, id: &SystemId) -> Result<Option<Plugin>, DomainError> {
        let row = sqlx::query_as::<_, PluginRow>(
            "SELECT plugin_id, name, plugin_type, is_enabled, configuration
             FROM plugins WHERE plugin_id = $1",
        )
        .bind(*id.as_uuid())
        .fetch_optional(&self.pool)
        .await
        .map_pg()?;

        Ok(row.map(Into::into))
    }

    async fn find_all(&self) -> Result<Vec<Plugin>, DomainError> {
        let rows = sqlx::query_as::<_, PluginRow>(
            "SELECT plugin_id, name, plugin_type, is_enabled, configuration FROM plugins",
        )
        .fetch_all(&self.pool)
        .await
        .map_pg()?;

        Ok(rows.into_iter().map(Into::into).collect())
    }

    async fn find_enabled(&self) -> Result<Vec<Plugin>, DomainError> {
        let rows = sqlx::query_as::<_, PluginRow>(
            "SELECT plugin_id, name, plugin_type, is_enabled, configuration
             FROM plugins WHERE is_enabled = true",
        )
        .fetch_all(&self.pool)
        .await
        .map_pg()?;

        Ok(rows.into_iter().map(Into::into).collect())
    }

    async fn save(&self, plugin: &Plugin) -> Result<(), DomainError> {
        sqlx::query(
            "INSERT INTO plugins (plugin_id, name, plugin_type, is_enabled, configuration)
             VALUES ($1, $2, $3, $4, $5)
             ON CONFLICT (plugin_id) DO UPDATE SET
               name = EXCLUDED.name,
               plugin_type = EXCLUDED.plugin_type,
               is_enabled = EXCLUDED.is_enabled,
               configuration = EXCLUDED.configuration",
        )
        .bind(*plugin.plugin_id.as_uuid())
        .bind(&plugin.name)
        .bind(plugin_type_to_str(&plugin.plugin_type))
        .bind(plugin.is_enabled)
        .bind(structured_to_json(&plugin.configuration))
        .execute(&self.pool)
        .await
        .map_pg()?;
        Ok(())
    }
}

// ──────────────────────────────────────────────
// Pipeline
// ──────────────────────────────────────────────

#[derive(sqlx::FromRow)]
struct PipelineRow {
    pipeline_id: Uuid,
    trigger_event: String,
    steps: serde_json::Value,
}

#[derive(serde::Serialize, serde::Deserialize)]
struct StepJson {
    plugin_id: Uuid,
    step_order: u32,
    configuration: serde_json::Map<String, serde_json::Value>,
}

fn steps_from_json(v: serde_json::Value) -> Vec<PipelineStep> {
    let arr: Vec<StepJson> = serde_json::from_value(v).unwrap_or_default();
    arr.into_iter()
        .map(|s| {
            let mut config = StructuredData::new();
            for (k, val) in s.configuration {
                config.insert(k, MetadataValue::from(val));
            }
            PipelineStep {
                plugin_id: SystemId::from_uuid(s.plugin_id),
                step_order: s.step_order,
                configuration: config,
            }
        })
        .collect()
}

fn steps_to_json(steps: &[PipelineStep]) -> serde_json::Value {
    let arr: Vec<StepJson> = steps
        .iter()
        .map(|s| {
            let config: serde_json::Map<String, serde_json::Value> = s
                .configuration
                .inner()
                .iter()
                .map(|(k, v)| (k.clone(), serde_json::Value::from(v)))
                .collect();
            StepJson {
                plugin_id: *s.plugin_id.as_uuid(),
                step_order: s.step_order,
                configuration: config,
            }
        })
        .collect();
    serde_json::to_value(arr).unwrap_or(serde_json::Value::Array(vec![]))
}

impl From<PipelineRow> for ProcessingPipeline {
    fn from(r: PipelineRow) -> Self {
        Self {
            pipeline_id: SystemId::from_uuid(r.pipeline_id),
            trigger_event: r.trigger_event,
            steps: steps_from_json(r.steps),
        }
    }
}

pg_repo!(PostgresPipelineRepository);

#[async_trait]
impl PipelineRepository for PostgresPipelineRepository {
    async fn find_by_id(&self, id: &SystemId) -> Result<Option<ProcessingPipeline>, DomainError> {
        let row = sqlx::query_as::<_, PipelineRow>(
            "SELECT pipeline_id, trigger_event, steps
             FROM processing_pipelines WHERE pipeline_id = $1",
        )
        .bind(*id.as_uuid())
        .fetch_optional(&self.pool)
        .await
        .map_pg()?;

        Ok(row.map(Into::into))
    }

    async fn find_all(&self) -> Result<Vec<ProcessingPipeline>, DomainError> {
        let rows = sqlx::query_as::<_, PipelineRow>(
            "SELECT pipeline_id, trigger_event, steps FROM processing_pipelines",
        )
        .fetch_all(&self.pool)
        .await
        .map_pg()?;

        Ok(rows.into_iter().map(Into::into).collect())
    }

    async fn find_by_trigger(&self, event: &str) -> Result<Vec<ProcessingPipeline>, DomainError> {
        let rows = sqlx::query_as::<_, PipelineRow>(
            "SELECT pipeline_id, trigger_event, steps
             FROM processing_pipelines WHERE trigger_event = $1",
        )
        .bind(event)
        .fetch_all(&self.pool)
        .await
        .map_pg()?;

        Ok(rows.into_iter().map(Into::into).collect())
    }

    async fn save(&self, pipeline: &ProcessingPipeline) -> Result<(), DomainError> {
        sqlx::query(
            "INSERT INTO processing_pipelines (pipeline_id, trigger_event, steps)
             VALUES ($1, $2, $3)
             ON CONFLICT (pipeline_id) DO UPDATE SET
               trigger_event = EXCLUDED.trigger_event,
               steps = EXCLUDED.steps",
        )
        .bind(*pipeline.pipeline_id.as_uuid())
        .bind(&pipeline.trigger_event)
        .bind(steps_to_json(&pipeline.steps))
        .execute(&self.pool)
        .await
        .map_pg()?;
        Ok(())
    }
}