| 1 | //! Durable cleanup, committed in the same transaction as identity deletion. |
| 2 | //! Each target is idempotent: a timeout after success may call it again. |
| 3 | use g1t_contracts::events::{NewEvent, Publish}; |
| 4 | use g1t_contracts::time::rfc3339; |
| 5 | use g1t_kit::now_ms; |
| 6 | use serde::{Deserialize, Serialize}; |
| 7 | use serde_json::{Value, json}; |
| 8 | use worker::{D1PreparedStatement, Result}; |
| 9 | |
| 10 | use crate::Identity; |
| 11 | |
| 12 | #[derive(Deserialize, Serialize)] |
| 13 | pub(crate) struct PurgeJob { |
| 14 | subject_id: String, |
| 15 | kind: String, |
| 16 | target: String, |
| 17 | payload: String, |
| 18 | created_at: String, |
| 19 | completed_at: Option<String>, |
| 20 | attempts: u32, |
| 21 | retry_at: String, |
| 22 | last_error: Option<String>, |
| 23 | } |
| 24 | |
| 25 | impl Identity { |
| 26 | pub(crate) fn purge_job_statements( |
| 27 | &self, |
| 28 | kind: &str, |
| 29 | id: &str, |
| 30 | payload: Value, |
| 31 | ) -> Result<Vec<D1PreparedStatement>> { |
| 32 | let now = rfc3339(now_ms()); |
| 33 | let payload = payload.to_string(); |
| 34 | ["AVATAR", "CHAT", "NOTIFY", "ARTIFACTS", "EVENTS"] |
| 35 | .into_iter() |
| 36 | .map(|target| { |
| 37 | self.db.prepare(include_str!("purge_job.sql")).bind(&[ |
| 38 | id.into(), |
| 39 | kind.into(), |
| 40 | target.into(), |
| 41 | payload.as_str().into(), |
| 42 | now.as_str().into(), |
| 43 | ]) |
| 44 | }) |
| 45 | .collect() |
| 46 | } |
| 47 | |
| 48 | pub(crate) async fn run_purge_jobs(&self) -> Result<()> { |
| 49 | let jobs = self.db.prepare("SELECT * FROM purge_jobs WHERE completed_at IS NULL AND retry_at <= ? ORDER BY retry_at LIMIT 25") |
| 50 | .bind(&[rfc3339(now_ms()).into()])?.all().await?.results::<PurgeJob>()?; |
| 51 | for job in jobs { |
| 52 | let result = self.run_purge_job(&job).await; |
| 53 | // Do not persist remote response bodies: they may contain personal data or secrets. |
| 54 | let now = rfc3339(now_ms()); |
| 55 | if matches!(result, Ok(true)) { |
| 56 | self.db.prepare("UPDATE purge_jobs SET completed_at = ?3, attempts = attempts + 1, last_error = NULL, payload = '{}' WHERE subject_id = ?1 AND target = ?2 AND completed_at IS NULL") |
| 57 | .bind(&[job.subject_id.as_str().into(), job.target.as_str().into(), now.as_str().into()])?.run().await?; |
| 58 | } else if matches!(result, Ok(false)) { |
| 59 | self.db.prepare("UPDATE purge_jobs SET attempts = attempts + 1, retry_at = ?3, last_error = NULL WHERE subject_id = ?1 AND target = ?2 AND completed_at IS NULL") |
| 60 | .bind(&[job.subject_id.as_str().into(), job.target.as_str().into(), rfc3339(now_ms() + 900_000).into()])?.run().await?; |
| 61 | } else { |
| 62 | let retry = rfc3339(now_ms() + 900_000 * (1_u64 << job.attempts.min(6))); |
| 63 | self.db.prepare("UPDATE purge_jobs SET attempts = attempts + 1, retry_at = ?3, last_error = 'Cleanup failed; retry pending' WHERE subject_id = ?1 AND target = ?2 AND completed_at IS NULL") |
| 64 | .bind(&[job.subject_id.as_str().into(), job.target.as_str().into(), retry.into()])?.run().await?; |
| 65 | worker::console_error!( |
| 66 | "purge target {} failed for {}; retry pending", |
| 67 | job.target, |
| 68 | job.subject_id |
| 69 | ); |
| 70 | } |
| 71 | } |
| 72 | Ok(()) |
| 73 | } |
| 74 | |
| 75 | async fn run_purge_job(&self, job: &PurgeJob) -> Result<bool> { |
| 76 | let payload: Value = serde_json::from_str(&job.payload)?; |
| 77 | if job.target == "AVATAR" { |
| 78 | self.forget_avatar( |
| 79 | payload |
| 80 | .get("avatar") |
| 81 | .and_then(Value::as_str) |
| 82 | .map(str::to_owned), |
| 83 | ) |
| 84 | .await?; |
| 85 | return Ok(true); |
| 86 | } |
| 87 | let service = self.env.service(&job.target)?; |
| 88 | if job.target == "EVENTS" { |
| 89 | let (kind, data) = if job.kind == "user" { |
| 90 | ( |
| 91 | "user.deleted", |
| 92 | json!({"userId": job.subject_id, "username": payload["username"]}), |
| 93 | ) |
| 94 | } else { |
| 95 | ( |
| 96 | "workspace.deleted", |
| 97 | json!({"workspaceId": job.subject_id, "slug": payload["slug"]}), |
| 98 | ) |
| 99 | }; |
| 100 | let _: Value = g1t_kit::call( |
| 101 | &service, |
| 102 | "publish", |
| 103 | &Publish { |
| 104 | events: vec![NewEvent { |
| 105 | kind, |
| 106 | source: "identity", |
| 107 | repo_id: None, |
| 108 | actor: None, |
| 109 | data, |
| 110 | }], |
| 111 | }, |
| 112 | ) |
| 113 | .await?; |
| 114 | } else { |
| 115 | let result: Value = g1t_kit::call(&service, "purge_personal_data", &json!({ |
| 116 | "kind": job.kind, "id": job.subject_id, "username": payload["username"], "slug": payload["slug"], |
| 117 | "old_slugs": payload["old_slugs"], "members": payload["members"], |
| 118 | })).await?; |
| 119 | if result.get("pending") == Some(&Value::Bool(true)) { |
| 120 | return Ok(false); |
| 121 | } |
| 122 | if result.get("ok") != Some(&Value::Bool(true)) { |
| 123 | return Err(worker::Error::RustError( |
| 124 | "Cleanup did not confirm success".into(), |
| 125 | )); |
| 126 | } |
| 127 | } |
| 128 | Ok(true) |
| 129 | } |
| 130 | |
| 131 | pub(crate) async fn pending_purge_jobs(&self) -> Result<Vec<PurgeJob>> { |
| 132 | self.db.prepare("SELECT subject_id, kind, target, '{}' AS payload, created_at, completed_at, attempts, retry_at, last_error FROM purge_jobs WHERE completed_at IS NULL ORDER BY created_at LIMIT 500") |
| 133 | .all().await?.results() |
| 134 | } |
| 135 | } |