opmph

nekolib-src/seq/Cargo.toml

@@ -11,3 +11,4 @@ inversion = { path = "inversion" }
 lis = { path = "lis" }
 suffix_array = { path = "suffix_array" }
 majority_vote = { path = "majority_vote" }
+opmph = { path = "opmph" }

nekolib-src/seq/opmph/Cargo.toml

@@ -0,0 +1,9 @@
+[package]
+name = "opmph"
+version = "0.1.0"
+edition = "2021"
+
+[dependencies]
+
+[dev-dependencies]
+maplit = "1.0.2"

nekolib-src/seq/opmph/src/lib.rs

@@ -0,0 +1,35 @@
+use std::collections::BTreeSet;
+
+// order-preserving minimal perfect hashing
+pub trait Opmph<B> {
+    fn opmph(&self) -> B;
+}
+
+impl<I, T, B> Opmph<B> for I
+where
+    for<'a> &'a I: IntoIterator<Item = &'a T>,
+    T: Clone + Ord,
+    B: FromIterator<(T, usize)>,
+{
+    fn opmph(&self) -> B {
+        let seen: BTreeSet<_> = self.into_iter().cloned().collect();
+        seen.into_iter().zip(0..).collect()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::collections::BTreeMap;
+
+    use maplit::btreemap;
+
+    use super::*;
+
+    #[test]
+    fn sanity_check() {
+        let a = vec![3, 5, 1, 2, 5];
+        let enc: BTreeMap<_, _> = a.opmph();
+
+        assert_eq!(enc, btreemap! { 1 => 0, 2 => 1, 3 => 2, 5 => 3 });
+    }
+}

nekolib-src/seq/src/lib.rs

@@ -4,5 +4,6 @@ doc_inline_reexport! {
     inversion,
     lis,
     majority_vote,
+    opmph,
     suffix_array,
 }