From 8f8134dbf1397fa9d48495d3faef63f1d4e86a03 Mon Sep 17 00:00:00 2001
From: Anatolii Smolianinov <zarkonesmall@gmail.com>
Date: Tue, 10 Dec 2024 13:45:41 +0100
Subject: [PATCH] use phf and lazy_static for lang map

---
 rust/Cargo.toml                  |  3 +-
 rust/src/tantivy_util/stemmer.rs | 53 ++++++++++++++++++--------------
 2 files changed, 32 insertions(+), 24 deletions(-)

diff --git a/rust/Cargo.toml b/rust/Cargo.toml
index 23e2502..9db1562 100644
--- a/rust/Cargo.toml
+++ b/rust/Cargo.toml
@@ -21,7 +21,8 @@ unicode-segmentation = "1.11.0"
 logcall = "0.1"
 lazy_static = "1.5.0"
 tantivy-jieba = { version = "0.11.0", optional = true }
+phf = { version = "0.11.2", features = ["macros"] }
 
 [features]
 default = ["jieba"]
-jieba = ["dep:tantivy-jieba"]
\ No newline at end of file
+jieba = ["dep:tantivy-jieba"]
diff --git a/rust/src/tantivy_util/stemmer.rs b/rust/src/tantivy_util/stemmer.rs
index cea0f0d..0ef6841 100644
--- a/rust/src/tantivy_util/stemmer.rs
+++ b/rust/src/tantivy_util/stemmer.rs
@@ -1,28 +1,35 @@
-use tantivy::tokenizer::{Language, Stemmer};
 use crate::tantivy_util::TantivyGoError;
+use lazy_static::lazy_static;
+use phf::phf_map;
+use tantivy::tokenizer::{Language, Stemmer};
+
+lazy_static! {
+    pub static ref LANGUAGES: phf::Map<&'static str, Language> = phf_map! {
+        "ar" => Language::Arabic,
+        "da" => Language::Danish,
+        "nl" => Language::Dutch,
+        "en" => Language::English,
+        "fi" => Language::Finnish,
+        "fr" => Language::French,
+        "de" => Language::German,
+        "el" => Language::Greek,
+        "hu" => Language::Hungarian,
+        "it" => Language::Italian,
+        "no" => Language::Norwegian,
+        "pt" => Language::Portuguese,
+        "ro" => Language::Romanian,
+        "ru" => Language::Russian,
+        "es" => Language::Spanish,
+        "sv" => Language::Swedish,
+        "ta" => Language::Tamil,
+        "tr" => Language::Turkish,
+    };
+}
 
 pub fn create_stemmer(lang: &str) -> Result<Stemmer, TantivyGoError> {
-    let stemmer_language = match lang {
-        "ar" => Ok(Language::Arabic),
-        "da" => Ok(Language::Danish),
-        "nl" => Ok(Language::Dutch),
-        "en" => Ok(Language::English),
-        "fi" => Ok(Language::Finnish),
-        "fr" => Ok(Language::French),
-        "de" => Ok(Language::German),
-        "el" => Ok(Language::Greek),
-        "hu" => Ok(Language::Hungarian),
-        "it" => Ok(Language::Italian),
-        "no" => Ok(Language::Norwegian),
-        "pt" => Ok(Language::Portuguese),
-        "ro" => Ok(Language::Romanian),
-        "ru" => Ok(Language::Russian),
-        "es" => Ok(Language::Spanish),
-        "sv" => Ok(Language::Swedish),
-        "ta" => Ok(Language::Tamil),
-        "tr" => Ok(Language::Turkish),
-        _ => Err(TantivyGoError(format!("{} is an unsupported language", lang))),
-    }?;
+    let stemmer_language = LANGUAGES
+        .get(lang)
+        .ok_or_else(|| TantivyGoError(format!("{lang} is an unsupported language")))?;
 
-    Ok(Stemmer::new(stemmer_language))
+    Ok(Stemmer::new(stemmer_language.to_owned()))
 }