refactor: add get_storage_path() and get_catalog_and_schema() (#4397)

refactor: add get_storage_path() and get_catalog_and_schema()

src/store-api/Cargo.toml

@@ -30,4 +30,5 @@ tokio.workspace = true
 
 [dev-dependencies]
 async-stream.workspace = true
+common-meta.workspace = true
 serde_json.workspace = true

src/store-api/src/path_utils.rs

@@ -45,10 +45,48 @@ pub fn region_dir(path: &str, region_id: RegionId) -> String {
     )
 }
 
+/// get_storage_path returns the storage path from the region_dir. It will always return the storage path if the region_dir is valid, otherwise None.
+/// The storage path is constructed from the catalog and schema, which are generated by `common_meta::ddl::utils::region_storage_path`.
+/// We can extract the catalog and schema from the region_dir by following example:
+/// ```
+/// use common_meta::ddl::utils::get_catalog_and_schema;
+///
+/// fn catalog_and_schema(region_dir: &str, region_id: RegionId) -> Option<(String, String)> {
+///     get_catalog_and_schema(&get_storage_path(region_dir, region_id)?)
+/// }
+/// ```
+pub fn get_storage_path(region_dir: &str, region_id: RegionId) -> Option<String> {
+    if !region_dir.starts_with(DATA_DIR) {
+        return None;
+    }
+
+    // For example, if region_dir is "data/my_catalog/my_schema/42/42_0000000001/", the parts will be '42/42_0000000001'.
+    let parts = format!(
+        "{}/{}",
+        region_id.table_id(),
+        region_name(region_id.table_id(), region_id.region_number())
+    );
+
+    // Ignore the last '/'. The original path will be like "${DATA_DIR}${catalog}/${schema}".
+    let pos = region_dir.rfind(&parts)? - 1;
+
+    if pos < DATA_DIR.len() {
+        return None;
+    }
+
+    Some(region_dir[DATA_DIR.len()..pos].to_string())
+}
+
 #[cfg(test)]
 mod tests {
+    use common_meta::ddl::utils::{get_catalog_and_schema, region_storage_path};
+
     use super::*;
 
+    fn catalog_and_schema(region_dir: &str, region_id: RegionId) -> Option<(String, String)> {
+        get_catalog_and_schema(&get_storage_path(region_dir, region_id)?)
+    }
+
     #[test]
     fn test_region_dir() {
         let region_id = RegionId::new(42, 1);
@@ -57,4 +95,32 @@ mod tests {
             "data/my_catalog/my_schema/42/42_0000000001/"
         );
     }
+
+    #[test]
+    fn test_get_catalog_and_schema_from_region_dir() {
+        let tests = [
+            (RegionId::new(42, 1), "my_catalog", "my_schema"),
+            (RegionId::new(1234, 1), "my_catalog_1234", "my_schema_1234"),
+            (RegionId::new(5678, 1), "my_catalog_5678", "my_schema"),
+            (RegionId::new(5678, 1), "my_catalog", "my_schema_5678"),
+        ];
+
+        for (region_id, test_catalog, test_schema) in tests.iter() {
+            let region_dir = region_dir(
+                region_storage_path(test_catalog, test_schema).as_str(),
+                *region_id,
+            );
+            let (catalog, schema) = catalog_and_schema(&region_dir, *region_id).unwrap();
+            assert_eq!(catalog, *test_catalog);
+            assert_eq!(schema, *test_schema);
+        }
+    }
+
+    #[test]
+    fn test_get_catalog_and_schema_from_invalid_region_dir() {
+        assert_eq!(
+            catalog_and_schema("invalid_data", RegionId::new(42, 1)),
+            None
+        );
+    }
 }