Auto-detect post language (fixes #2870)

Cargo.toml

@@ -120,6 +120,7 @@ reqwest = { version = "0.12.7", default-features = false, features = [
   "gzip",
   "rustls-tls",
 ] }
+cfg-if = "1"
 reqwest-middleware = "0.3.3"
 reqwest-tracing = "0.5.3"
 clokwerk = "0.4.0"

crates/db_schema/Cargo.toml

@@ -38,7 +38,7 @@ full = [
   "rustls",
   "i-love-jesus",
   "tuplex",
-  "diesel-bind-if-some",
+  "diesel-bind-if-some","lingua"
 ]
 
 [dependencies]
@@ -82,6 +82,8 @@ diesel-bind-if-some = { workspace = true, optional = true }
 moka.workspace = true
 derive-new.workspace = true
 tuplex = { workspace = true, optional = true }
+lingua = { version = "1.6.2", optional = true }
+cfg-if.workspace =true
 
 [dev-dependencies]
 serial_test = { workspace = true }

crates/db_schema/src/detect_language.rs

@@ -0,0 +1,84 @@
+use crate::{newtypes::LanguageId, source::language::Language, utils::DbPool};
+use lemmy_utils::error::LemmyResult;
+use lingua::{IsoCode639_1, Language as LinguaLanguage, LanguageDetectorBuilder};
+
+pub async fn detect_language(input: &str, pool: &mut DbPool<'_>) -> LemmyResult<LanguageId> {
+  // TODO: should only detect languages which are allowed in community
+  let detector = LanguageDetectorBuilder::from_iso_codes_639_1(&[
+    IsoCode639_1::EN,
+    IsoCode639_1::ES,
+    IsoCode639_1::DE,
+  ])
+  .build();
+
+  let lang: Option<LinguaLanguage> = detector.detect_language_of(input);
+  let Some(lang) = lang else {
+    return Ok(LanguageId(0));
+  };
+  let confidence = detector.compute_language_confidence("languages are awesome", lang);
+  let lang = lang.iso_code_639_1().to_string().to_lowercase();
+  dbg!(&lang, &confidence);
+  if confidence < 0.4 {
+    return Ok(LanguageId(0));
+  }
+
+  Ok(Language::read_id_from_code(pool, &lang).await?)
+}
+
+#[cfg(test)]
+#[expect(clippy::indexing_slicing)]
+mod tests {
+
+  use super::*;
+  use crate::utils::build_db_pool_for_tests;
+  use pretty_assertions::assert_eq;
+  use serial_test::serial;
+
+  #[tokio::test]
+  #[serial]
+  async fn test_detect_language() -> LemmyResult<()> {
+    let pool = &build_db_pool_for_tests();
+    let pool = &mut pool.into();
+
+    // some easy comments
+    assert_eq!(
+      LanguageId(37),
+      detect_language(
+        "I don't think it's supposed to be taken seriously. It's just a throwaway meme.
+",
+        pool
+      )
+      .await?
+    );
+    assert_eq!(
+      LanguageId(39),
+      detect_language(
+        "Oh! Mencion casual de la mejor pelicula navideña… Die hard!
+",
+        pool
+      )
+      .await?
+    );
+    assert_eq!(
+      LanguageId(32),
+      detect_language(
+        "Die Forderung finde ich nutzlos.
+",
+        pool
+      )
+      .await?
+    );
+
+    // different languages
+    assert_eq!(
+      LanguageId(0),
+      detect_language(
+        "Die Forderung finde ich nutzlos. It's just a throwaway meme.
+",
+        pool
+      )
+      .await?
+    );
+    Ok(())
+  }
+}

crates/db_schema/src/lib.rs

@@ -1,42 +1,37 @@
 #![recursion_limit = "256"]
+use cfg_if::cfg_if;
 
-#[cfg(feature = "full")]
-#[macro_use]
-extern crate diesel;
-#[cfg(feature = "full")]
-#[macro_use]
-extern crate diesel_derive_newtype;
-
-#[cfg(feature = "full")]
-#[macro_use]
-extern crate diesel_derive_enum;
-
-// this is used in tests
-#[cfg(feature = "full")]
-#[macro_use]
-extern crate diesel_migrations;
-
-#[cfg(feature = "full")]
-#[macro_use]
-extern crate async_trait;
-
-pub mod aggregates;
-#[cfg(feature = "full")]
-pub mod impls;
-pub mod newtypes;
-pub mod sensitive;
-#[cfg(feature = "full")]
-#[rustfmt::skip]
-pub mod schema;
-#[cfg(feature = "full")]
-pub mod aliases {
+cfg_if! {
+  if #[cfg(feature = "full")] {
+    #[macro_use]
+    extern crate diesel;
+    #[macro_use]
+    extern crate diesel_derive_newtype;
+    #[macro_use]
+    extern crate diesel_derive_enum;
+    // this is used in tests
+    #[macro_use]
+    extern crate diesel_migrations;
+    #[macro_use]
+    extern crate async_trait;
+    pub mod impls;
+    #[rustfmt::skip]
+    pub mod schema;
+    pub mod detect_language;
+    pub mod aliases {
       use crate::schema::{community_actions, person};
       diesel::alias!(
         community_actions as creator_community_actions: CreatorCommunityActions,
         person as person1: Person1,
         person as person2: Person2,
       );
+    }
+  }
 }
+
+pub mod aggregates;
+pub mod newtypes;
+pub mod sensitive;
 pub mod source;
 #[cfg(feature = "full")]
 pub mod traits;

crates/utils/Cargo.toml

@@ -83,7 +83,7 @@ lettre = { version = "0.11.10", default-features = false, features = [
 markdown-it = { version = "0.6.1", optional = true }
 ts-rs = { workspace = true, optional = true }
 enum-map = { workspace = true, optional = true }
-cfg-if = "1"
+cfg-if.workspace = true
 clearurls = { version = "0.0.4", features = ["linkify"] }
 markdown-it-block-spoiler = "1.0.0"
 markdown-it-sub = "1.0.0"