Rename Python confidence functions and add docs and tests

Author: ZJaume

src/identifier.rs

@@ -59,13 +59,23 @@ impl Identifier {
             ignore_confidence: ignore_confidence,
         }
     }
+    /// Disable use of confidence thresholds
+    pub fn disable_confidence(&mut self) {
+        self.ignore_confidence = true;
+    }
 
-    /// Enable use of confidence thresholds
-    pub fn disable_confidence(&mut self) -> &mut Self {
+    /// Disable use of confidence thresholds
+    pub fn without_confidence(&mut self) -> &mut Self {
         self.ignore_confidence = true;
         self
     }
 
+    /// Enable use of confidence thresholds
+    pub fn with_confidence(&mut self) -> &mut Self {
+        self.ignore_confidence = false;
+        self
+    }
+
     /// Get the most probable language according to the current language scores
     fn pick_winner(&mut self) -> (Lang, f32) {
         // if only one lang is requested, just search for the minimum score (winner)
@@ -436,4 +446,18 @@ mod tests {
                 "expected  = {:?}\npredict = {:?}", pred, expected);
         }
     }
+
+    #[test_log::test]
+    fn test_confidence() {
+        pyo3::prepare_freethreaded_python();
+        let identifier = Identifier::load(
+            &python::module_path().expect("Python module needs to be installed"),
+            None,
+        ).expect("Could not load model, please run 'heliport bianrize' if you haven't")
+            .disable_confidence();
+
+        let pred = identifier.identify("hello");
+        assert!(pred.0 == Lang::sah);
+    }
+
 }

src/python.rs

@@ -23,7 +23,6 @@ pub fn module_path() -> PyResult<PathBuf> {
 }
 
 /// Bindings to Python
-/// //TODO support returning both lang+score
 /// //TODO support parallel identification
 /// //TODO support loading relevant languages from text
 #[pymethods]
@@ -39,19 +38,24 @@ impl Identifier {
         Ok(identifier)
     }
 
+    /// Identify the language of a string
     #[pyo3(name = "identify")]
     fn py_identify(&mut self, text: &str) -> String {
         self.identify(text).0.to_string()
     }
 
-    #[pyo3(name = "identify_with_confidence")]
-    fn py_identify_with_confidence(&mut self, text: &str) -> (String, f32) {
+    /// Identify the top-k most probable languages of a string and return the prediction scores.
+    /// This score is the confidence score (difference with the 2nd best)
+    /// or the raw score if ignore_confidence is enabled.
+    #[pyo3(name = "identify_with_score")]
+    fn py_identify_with_score(&mut self, text: &str) -> (String, f32) {
         let pred = self.identify(text);
         (pred.0.to_string(), pred.1)
     }
 
-    #[pyo3(name = "identify_topk_with_confidence")]
-    fn py_identify_topk_with_confidence(&mut self, text: &str, k: usize) -> Vec<(String, f32)> {
+    /// Identify the language of a string and return the raw prediction score.
+    #[pyo3(name = "identify_topk_with_score")]
+    fn py_identify_topk_with_score(&mut self, text: &str, k: usize) -> Vec<(String, f32)> {
         let preds = self.identify_topk(text, k);
         let mut out = Vec::<_>::with_capacity(preds.len());
         for (pred, conf) in preds {