Merge pull request #120 from Portabase/dev

Dev

docker-compose.yml

@@ -21,7 +21,7 @@ services:
       LOG: debug
       TZ: "Europe/Paris"
       #      TMPDIR: /scratch
-      EDGE_KEY: "eyJzZXJ2ZXJVcmwiOiJodHRwOi8vbG9jYWxob3N0Ojg4ODciLCJhZ2VudElkIjoiMjcyMWFiNGQtZDkzNi00NmFkLWE1MWEtOWVhYTk3ZjJhMjAwIiwibWFzdGVyS2V5QjY0IjoiQlhWM1hvbEM2NTZTVjdkTmdjV1BHUWxrKytycExJNmxHRGk3Q1BCNWllbz0ifQ=="
+      EDGE_KEY: "eyJzZXJ2ZXJVcmwiOiJodHRwOi8vbG9jYWxob3N0Ojg4ODciLCJhZ2VudElkIjoiN2ZjMGViODMtYTY0YS00NzI3LTg5ZjktNzFhZWJmZDQ3NmEyIiwibWFzdGVyS2V5QjY0IjoiQlhWM1hvbEM2NTZTVjdkTmdjV1BHUWxrKytycExJNmxHRGk3Q1BCNWllbz0ifQ=="
       #CHUNK_SIZE_MB: "1"
       #POOLING: 1
       #RETRY_ATTEMPTS: 3

src/domain/postgres/backup.rs

@@ -34,6 +34,14 @@ pub async fn run(
         let pg_dump = select_pg_path(&version).join("pg_dump");
         logger.log("debug", format!("Using pg_dump at {:?}", pg_dump));
 
+        let (excludes, invalid) = exclude_args(&cfg);
+        for key in invalid {
+            logger.log("warn", format!("Ignoring options.{} for {}: expected a list of patterns", key, cfg.name));
+        }
+        if !excludes.is_empty() {
+            logger.log("info", format!("Excluding from dump of {}: {}", cfg.name, excludes.join(" ")));
+        }
+
         match format {
             PostgresDumpFormat::Fc => {
                 logger.log("info", format!("Running FC backup for {}", cfg.name));
@@ -46,6 +54,7 @@ pub async fn run(
                     .arg("--port").arg(cfg.port.to_string())
                     .arg("--username").arg(&cfg.username)
                     .arg("--dbname").arg(&cfg.database)
+                    .args(&excludes)
                     .arg("-Fc")
                     .arg("-f").arg(&file_path)
                     .arg("-v")
@@ -97,6 +106,7 @@ pub async fn run(
                     .arg("--port").arg(cfg.port.to_string())
                     .arg("--username").arg(&cfg.username)
                     .arg("--dbname").arg(&cfg.database)
+                    .args(&excludes)
                     .arg("-Fd")
                     .arg("-j").arg("4")
                     .arg("-f").arg(&dump_dir)
@@ -151,3 +161,23 @@ pub async fn run(
     })
     .await?
 }
+
+pub(crate) fn exclude_args(cfg: &DatabaseConfig) -> (Vec<String>, Vec<&'static str>) {
+    let mut args = Vec::new();
+    let mut invalid = Vec::new();
+    for (key, flag) in [("exclude_tables", "--exclude-table"), ("exclude_schemas", "--exclude-schema")] {
+        match cfg.options.get(key).map(|v| v.as_array()) {
+            None => {}
+            Some(Some(items)) => args.extend(
+                items
+                    .iter()
+                    .filter_map(|v| v.as_str())
+                    .map(str::trim)
+                    .filter(|p| !p.is_empty())
+                    .map(|p| format!("{flag}={p}")),
+            ),
+            Some(None) => invalid.push(key),
+        }
+    }
+    (args, invalid)
+}

src/tests/domain/postgres.rs

@@ -316,6 +316,62 @@ async fn restore_run_unified_fc_roundtrip() {
     assert!(result.is_ok(), "restore::run failed: {:?}", result);
 }
 
+#[tokio::test]
+async fn backup_as_non_owner_succeeds_only_with_exclude_tables() {
+    init_tracing_for_test();
+
+    let (_container, admin) = create_config().await;
+    crate::domain::postgres::connection::connect(&admin)
+        .await
+        .unwrap()
+        .batch_execute(
+            "CREATE ROLE backup_ro LOGIN PASSWORD 'x';
+             CREATE TABLE public.keep(id int);
+             CREATE TABLE public.search_index__x(id int);
+             GRANT SELECT ON public.keep TO backup_ro;",
+        )
+        .await
+        .unwrap();
+
+    let mut ro = admin.clone();
+    ro.username = "backup_ro".into();
+    ro.password = "x".into();
+
+    let mut excluded = ro.clone();
+    excluded
+        .options
+        .insert("exclude_tables".into(), serde_json::json!(["public.search_index__*"]));
+
+    for format in [PostgresDumpFormat::Fc, PostgresDumpFormat::Fd] {
+        let dir = TempDir::new().unwrap();
+        let denied = crate::domain::postgres::backup::run(
+            ro.clone(), format, dir.path().to_path_buf(), pg_dump_env(&ro), Arc::new(JobLogger::new()),
+        )
+        .await;
+        assert!(denied.is_err(), "{format:?}: non-owner dump must fail without exclusions");
+
+        let dir = TempDir::new().unwrap();
+        let logger = Arc::new(JobLogger::new());
+        let dump = crate::domain::postgres::backup::run(
+            excluded.clone(), format, dir.path().to_path_buf(), pg_dump_env(&excluded), logger.clone(),
+        )
+        .await
+        .unwrap_or_else(|e| panic!("{format:?}: dump with exclusions failed: {e:?}"));
+
+        let entries = Arc::try_unwrap(logger).expect("single owner").into_entries();
+        assert!(entries.iter().any(|e| e.message
+            == format!("Excluding from dump of {}: --exclude-table=public.search_index__*", excluded.name)));
+
+        if format == PostgresDumpFormat::Fc {
+            let version = server_version(&admin).await.unwrap();
+            let pg_restore = select_pg_path(&version).join(pg_restore_binary_name());
+            let prepared = prepare_archive(format, &dump, &pg_restore, &JobLogger::new()).unwrap();
+            assert!(prepared.toc().contains("TABLE public keep "));
+            assert!(!prepared.toc().contains("search_index__x"));
+        }
+    }
+}
+
 #[tokio::test]
 async fn drop_all_schemas_removes_user_schema() {
     init_tracing_for_test();
@@ -924,6 +980,61 @@ mod clean_mode_tests {
     }
 }
 
+mod exclude_args_tests {
+    use crate::domain::postgres::backup::exclude_args;
+    use crate::services::config::DatabaseConfig;
+    use serde_json::json;
+
+    fn cfg(options: serde_json::Value) -> DatabaseConfig {
+        serde_json::from_value(json!({
+            "name": "t", "database": "testdb", "type": "postgresql",
+            "username": "u", "password": "p", "port": 5432, "host": "localhost",
+            "generated_id": "00000000-0000-0000-0000-000000000000",
+            "path": "", "max_packet_size": "", "volume_name": "",
+            "options": options
+        }))
+        .unwrap()
+    }
+
+    #[test]
+    fn no_options_yields_nothing() {
+        assert_eq!(exclude_args(&cfg(json!({}))), (vec![], vec![]));
+        assert_eq!(
+            exclude_args(&cfg(json!({"exclude_tables": [], "exclude_schemas": []}))),
+            (vec![], vec![])
+        );
+    }
+
+    #[test]
+    fn lists_map_to_single_arg_flags_in_order() {
+        let (args, invalid) = exclude_args(&cfg(json!({
+            "exclude_schemas": ["metabase_tmp"],
+            "exclude_tables": ["public.search_index__*", "  audit.tmp_*  ", "", "   ", 42, "-f/tmp/x", "\"MyTable\""]
+        })));
+        assert_eq!(
+            args,
+            vec![
+                "--exclude-table=public.search_index__*",
+                "--exclude-table=audit.tmp_*",
+                "--exclude-table=-f/tmp/x",
+                "--exclude-table=\"MyTable\"",
+                "--exclude-schema=metabase_tmp",
+            ]
+        );
+        assert!(invalid.is_empty());
+    }
+
+    #[test]
+    fn non_list_values_are_reported_and_ignored() {
+        let (args, invalid) = exclude_args(&cfg(json!({
+            "exclude_tables": "public.search_index__*",
+            "exclude_schemas": {"a": 1}
+        })));
+        assert!(args.is_empty());
+        assert_eq!(invalid, vec!["exclude_tables", "exclude_schemas"]);
+    }
+}
+
 mod toc_tests {
     use crate::domain::postgres::restore::toc_creates_public_schema;