rebalance skeleton (but is it correct?)

1 files changed, 37 insertions, 0 deletions

diff --git a/src/main.rs b/src/main.rs
index a283891..0459bf0 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -25,6 +25,10 @@ enum Cmd {
         count: usize,
     },
     Simplify,
+    Rebalance {
+        #[structopt(default_value = "5")]
+        start: usize,
+    },
     Format,
 }
 
@@ -89,6 +93,23 @@ fn main() {
             )
             .expect("save");
         }
+        Cmd::Rebalance { start } => {
+            let mut batches: Vec<Batch> = fs::read("data/batches.json")
+                .map_err(anyhow::Error::from)
+                .and_then(|x| Ok(serde_json::from_slice(&x)?))
+                .unwrap_or_default();
+            let kanji_levels = read_kanji_levels().expect("read_kanji_levels");
+            for (level, _) in kanji_levels.iter() {
+                rebalance_level(level, &mut batches[start..]);
+            }
+            fs::write(
+                "data/batches.json",
+                serde_json::to_string_pretty(&batches)
+                    .expect("serialize")
+                    .as_bytes(),
+            )
+            .expect("save");
+        }
         Cmd::Format => {
             let jmdict = fs::read_to_string("data/JMdict_e.xml").expect("read_jmdict");
             let jmdict = roxmltree::Document::parse_with_options(
@@ -472,6 +493,22 @@ fn simplify_batch(batch: &mut Batch) {
     }
 }
 
+fn rebalance_level(level: &str, batches: &mut [Batch]) {
+    let mut i_batch = vec![];
+    let mut n_ex = 0;
+    for (i, b) in batches.iter().enumerate() {
+        if b.level == level {
+            i_batch.push(i);
+            n_ex += b.examples.len();
+        }
+    }
+    if i_batch.len() < 2 {
+        return;
+    }
+    println!("Level {}: {} batches, {} examples, avg {:.2}", level, i_batch.len(), n_ex, n_ex as f32 / i_batch.len() as f32);
+    todo!()
+}
+
 // =====================================================================
 //                          FORMATTING TO HTML
 // =====================================================================