From e0f80b10d7874a8317ea0ae4a621fc3556eac491 Mon Sep 17 00:00:00 2001
From: Alex Auvolat <alex@adnab.me>
Date: Sun, 24 Sep 2023 18:27:18 +0200
Subject: merge elementary school kanji levels

---
 src/main.rs | 25 ++++++++++++++++++++++---
 1 file changed, 22 insertions(+), 3 deletions(-)

(limited to 'src')

diff --git a/src/main.rs b/src/main.rs
index 0459bf0..126f4de 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -23,6 +23,8 @@ enum Cmd {
     New {
         #[structopt(default_value = "10")]
         count: usize,
+        #[structopt(long = "truncate")]
+        truncate: Option<usize>,
     },
     Simplify,
     Rebalance {
@@ -42,7 +44,7 @@ fn main() {
                 println!("{}: {}", level, chars.to_string());
             }
         }
-        Cmd::New { count } => {
+        Cmd::New { truncate, count } => {
             let kanji_levels = read_kanji_levels().expect("read_kanji_levels");
             let all_kanji = Charset::new(
                 kanji_levels
@@ -61,6 +63,9 @@ fn main() {
                 .map_err(anyhow::Error::from)
                 .and_then(|x| Ok(serde_json::from_slice(&x)?))
                 .unwrap_or_default();
+            if let Some(t) = truncate {
+                batches.truncate(t);
+            }
             println!("---- starting after {} batches ----", batches.len());
             for _ in 0..count {
                 let batch = gen_batch(&batches, &kanji_levels, &ex).expect("gen_batch");
@@ -196,6 +201,10 @@ fn parse_kanjidic() -> Result<Vec<(String, Charset)>> {
                 }
             }
         }
+        match grade {
+            Some(i) if i <= 6 => grade = Some(7),
+            _ => (),
+        }
         if let Some(lit) = literal {
             levels
                 .entry((jlpt, grade))
@@ -216,8 +225,12 @@ fn parse_kanjidic() -> Result<Vec<(String, Charset)>> {
     let mut pc = Charset::default();
     for ((j, g), chars) in levels.into_iter() {
         let name = match (j, g) {
+            (Some(j), Some(7)) => format!("N{}a", j),
+            (Some(j), Some(8)) => format!("N{}b", j),
             (Some(j), Some(g)) => format!("N{}-{}", j, g),
             (Some(j), None) => format!("N{}+", j),
+            (None, Some(7)) => format!("N0a"),
+            (None, Some(8)) => format!("N0b"),
             (None, Some(g)) => format!("N0-{}", g),
             (None, None) => format!("N0+"),
         };
@@ -505,8 +518,14 @@ fn rebalance_level(level: &str, batches: &mut [Batch]) {
     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!()
+    println!(
+        "Level {}: {} batches, {} examples, avg {:.2}",
+        level,
+        i_batch.len(),
+        n_ex,
+        n_ex as f32 / i_batch.len() as f32
+    );
+    //todo!()
 }
 
 // =====================================================================
-- 
cgit v1.2.3