import migrations and don't require diesel_cli for admins (#555)

* import migrations via macro

* panic on database not to the latest migration

* add subcommand to plm

* create migration that run tantivy index creation

* remove diesel_cli from places it was

* use our migration system for tests

* create table __diesel_schema_migrations if needed

src/main.rs

@@ -41,6 +41,7 @@ extern crate webfinger;
 use diesel::r2d2::ConnectionManager;
 use plume_models::{
     db_conn::{DbPool, PragmaForeignKey},
+    migrations::IMPORTED_MIGRATIONS,
     search::{Searcher as UnmanagedSearcher, SearcherError},
     Connection, Error, CONFIG,
 };
@@ -83,6 +84,22 @@ fn init_pool() -> Option<DbPool> {
 
 fn main() {
     let dbpool = init_pool().expect("main: database pool initialization error");
+    if IMPORTED_MIGRATIONS
+        .is_pending(&dbpool.get().unwrap())
+        .unwrap_or(true)
+    {
+        panic!(
+            r#"
+It appear your database migration does not run the migration required
+by this version of Plume. To fix this, you can run migrations via
+this command:
+
+    plm migration run
+
+Then try to restart Plume.
+"#
+        )
+    }
     let workpool = ScheduledThreadPool::with_name("worker {}", num_cpus::get());
     // we want a fast exit here, so
     #[allow(clippy::match_wild_err_arm)]