progress

.gitignore

@@ -3,3 +3,8 @@ tags
 target
 *.lock
 tmp
+*.csv
+*.fst
+*-got
+*.csv.idx
+words

Cargo.toml

@@ -17,6 +17,12 @@ license = "Unlicense/MIT"
 name = "fst"
 path = "src/bin/main.rs"
 
+[[bench]]
+name = "build"
+path = "./benches/build.rs"
+test = false
+bench = true
+
 [dependencies]
 byteorder = "*"
 

benches/build.rs

@@ -0,0 +1,57 @@
+#![feature(test)]
+
+extern crate fst;
+extern crate test;
+
+use std::collections::HashSet;
+
+use fst::{Builder, Fst};
+use test::Bencher;
+
+const WORDS: &'static str = include_str!("./../data/words-10000");
+
+fn get_words() -> Vec<String> {
+    WORDS.lines().map(|s| s.to_owned()).collect()
+}
+
+fn get_words_outputs() -> Vec<(String, u64)> {
+    WORDS.lines().map(|s| (s.to_owned(), s.len() as u64)).collect()
+}
+
+#[bench]
+fn build_fst_set(b: &mut Bencher) {
+    let words = get_words();
+    b.bytes = WORDS.len() as u64;
+    b.iter(|| {
+        let mut bfst = Builder::memory();
+        for word in &words {
+            bfst.add(word).unwrap();
+        }
+        Fst::new(bfst.into_inner().unwrap()).unwrap();
+    });
+}
+
+#[bench]
+fn build_fst_map(b: &mut Bencher) {
+    let words = get_words_outputs();
+    b.bytes = WORDS.len() as u64;
+    b.iter(|| {
+        let mut bfst = Builder::memory();
+        for &(ref word, len) in &words {
+            bfst.insert(word, len).unwrap();
+        }
+        Fst::new(bfst.into_inner().unwrap()).unwrap();
+    });
+}
+
+#[bench]
+fn build_hash_set(b: &mut Bencher) {
+    let words = get_words();
+    b.bytes = WORDS.len() as u64;
+    b.iter(|| {
+        let mut set = HashSet::new();
+        for word in &words {
+            set.insert(word);
+        }
+    });
+}