wip

1 files changed, 34 insertions, 0 deletions

diff --git a/plugins/tokenize-ja/src/tokenize.rs b/plugins/tokenize-ja/src/tokenize.rs
new file mode 100644
index 0000000..b5c13db
--- /dev/null
+++ b/plugins/tokenize-ja/src/tokenize.rs
@@ -0,0 +1,34 @@
+use anyhow::anyhow;
+use anyhow::Context;
+use lindera::dictionary::load_dictionary;
+use lindera::mode::Mode;
+use lindera::segmenter::Segmenter;
+use lindera::tokenizer::Tokenizer;
+use std::sync::OnceLock;
+
+static TOKENIZER: OnceLock<Tokenizer> = OnceLock::new();
+
+pub fn init() -> Result<(), anyhow::Error> {
+    let dictionary = load_dictionary("embedded://ipadic").context("failed to load dictionary")?;
+    let segmenter = Segmenter::new(Mode::Normal, dictionary, None).keep_whitespace(true);
+    let tokenizer = Tokenizer::new(segmenter);
+    TOKENIZER
+        .set(tokenizer)
+        .map_err(|_| anyhow!("failed to initialize tokenizer"))?;
+    Ok(())
+}
+
+pub fn tokenize(text: &str) -> Result<Vec<(String, Option<String>)>, anyhow::Error> {
+    let tokenizer = TOKENIZER.get().context("call load() first")?;
+    let mut tokens = tokenizer
+        .tokenize(text)
+        .context("failed to tokenize text")?;
+    Ok(tokens
+        .iter_mut()
+        .map(|t| {
+            let surface: String = t.surface.to_string();
+            let detail = t.get_detail(0).map(|s| s.into());
+            (surface, detail)
+        })
+        .collect())
+}