feat: one transaction, build in memory and vacuum out

src/main.rs

@@ -45,11 +45,13 @@ fn run(
     );
 
     log::info!("opening connection");
-    let mut conn = Connection::open(output_file)?;
+    let mut conn = Connection::open_in_memory()?;
 
     create_tables(&mut conn)?;
     insert_stats(&mut conn, &stats)?;
 
+    conn.execute("VACUUM INTO ?1", (output_file,))?;
+
     Ok(())
 }
 
@@ -58,8 +60,9 @@ fn run(
 struct Config {
     #[arg()]
     input_file: PathBuf,
+    // TODO: non-utf-8 (the problem is saving the database)
     #[arg(short, long = "output")]
-    output_file: PathBuf,
+    output_file: String,
 }
 
 #[derive(Error, Debug)]
@@ -125,12 +128,12 @@ fn insert_stats(conn: &mut Connection, stats: &Stats) -> rusqlite::Result<()> {
     let mon_count = stats.data.len();
     let mon_count_digits = mon_count.to_string().len();
 
+    let tx = conn.transaction()?;
     for (i, (mon, data)) in stats.data.iter().enumerate() {
         log::debug!(
             "Processing mon {:mon_count_digits$}/{mon_count} ({mon})",
             i + 1
         );
-        let tx = conn.transaction()?;
 
         // normalizing with mon_count gives us data that is much easier to work
         // with. for example, if pikachu has 10k count (weighted) and thunderbolt
@@ -196,9 +199,8 @@ fn insert_stats(conn: &mut Connection, stats: &Stats) -> rusqlite::Result<()> {
                 (mon, opp, percentage, stddev),
             )?;
         }
-
-        tx.commit()?;
     }
+    tx.commit()?;
     Ok(())
 }