Add more crates to accuracy comparison.

Author: finnbear

Cargo.toml

@@ -75,8 +75,10 @@ serde = {version = "1", features=["derive"], optional = true}
 [dev-dependencies]
 rand = "0.8"
 csv = "1.1"
-censor_crate = {package = "censor", version = "0.3.0"}
-rustrict_old = {package = "rustrict", version = "0.7.24"}
+censor_crate = { package = "censor", version = "0.3.0" }
+rustrict_old = { package = "rustrict", version = "0.7.24" }
 serial_test = "0.5"
+stfu_crate = { package = "stfu", version = "0.1.0" }
+profane_rs_crate = { package = "profane-rs", version = "0.0.4" }
 bincode = "1.3.3"
 serde_json = "1"

Makefile

@@ -24,6 +24,9 @@ test:
 compare:
 	COMPARE=1 make test
 
+table:
+	cargo test --release -- accuracy --nocapture
+
 # Skips accuracy analysis so finishes faster.
 test_debug:
 	cargo test --features pii -- --nocapture

README.md

@@ -179,6 +179,8 @@ is used as a dataset. Positive accuracy is the percentage of profanity detected
 |-------|----------|-------------------|-------------------|------|
 | [rustrict](https://crates.io/crates/rustrict) | 79.83%   | 94.00%            | 76.30%            | 9s   |
 | [censor](https://crates.io/crates/censor) | 76.16%   | 72.76%            | 77.01%            | 23s  |
+| [stfu](https://crates.io/crates/stfu) | 91.74% | 77.69% | 95.25% | 45s |
+| [profane-rs](https://crates.io/crates/profane-rs) | 80.47% | 73.79% | 82.14% | 52s |
 
 ## Development
 

src/censor.rs

@@ -1163,6 +1163,18 @@ mod tests {
             filter.check(s)
         }
 
+        let mut stfu_filter = stfu_crate::types::OwnedFilter::default();
+        use stfu_crate::word_lists::severity::{MILD, SEVERE, STRONG};
+        stfu_filter.add_slice(&MILD);
+        stfu_filter.add_slice(&STRONG);
+        stfu_filter.add_slice(&SEVERE);
+
+        let stfu = |s: &str| -> bool { stfu_filter.filter_string(s).is_some() };
+
+        fn profane_rs(s: &str) -> bool {
+            profane_rs_crate::contains_profanity(s, false)
+        }
+
         println!("| Crate | Accuracy | Positive Accuracy | Negative Accuracy | Time |");
         println!("|-------|----------|-------------------|-------------------|------|");
         print_accuracy(
@@ -1172,12 +1184,19 @@ mod tests {
             Some(rustrict_old as fn(&str) -> bool).filter(|_| std::env::var("COMPARE").is_ok()),
         );
         print_accuracy("https://crates.io/crates/censor", censor, false, None);
+        print_accuracy("https://crates.io/crates/stfu", stfu, false, None);
+        print_accuracy(
+            "https://crates.io/crates/profane-rs",
+            profane_rs,
+            false,
+            None,
+        );
     }
 
     #[allow(dead_code)]
     fn print_accuracy(
         link: &str,
-        checker: fn(&str) -> bool,
+        checker: impl Fn(&str) -> bool,
         find_detections: bool,
         compare_to: Option<fn(&str) -> bool>,
     ) {
@@ -1196,7 +1215,7 @@ mod tests {
 
     #[allow(dead_code)]
     fn accuracy_of(
-        checker: fn(&str) -> bool,
+        checker: impl Fn(&str) -> bool,
         find_detections: bool,
         compare_to: Option<fn(&str) -> bool>,
     ) -> (f32, f32, f32) {

src/lib.rs

@@ -53,13 +53,13 @@ pub(crate) type Set<V> = rustc_hash::FxHashSet<V>;
 #[allow(deprecated)]
 pub use censor::add_word;
 
+#[cfg(all(feature = "context", feature = "width"))]
+pub use context::ContextWordBreakOptions;
 #[cfg(feature = "context")]
 pub use context::{
     BlockReason, Context, ContextProcessingOptions, ContextRateLimitOptions,
     ContextRepetitionLimitOptions,
 };
-#[cfg(all(feature = "context", feature = "width"))]
-pub use context::ContextWordBreakOptions;
 
 #[cfg(feature = "pii")]
 pub use pii::censor_and_analyze_pii;