fix byte compression

microsoft · Dec 30, 2024 · a72d469 · a72d469
1 parent 9e5a36c
commit a72d469
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 20 deletions.
diff --git a/src/bytecompress.rs b/src/bytecompress.rs
@@ -81,6 +81,13 @@ impl ByteCompressor {
         self.map_cache[&e]
     }
 
+    fn add_single_byte(&mut self, b: u8) {
+        if self.mapping[b as usize] == INVALID_MAPPING {
+            self.mapping[b as usize] = self.alphabet_size as u8;
+            self.alphabet_size += 1;
+        }
+    }
+
     pub fn compress(&mut self, exprset: &ExprSet, rx_list: &[ExprRef]) -> (ExprSet, Vec<ExprRef>) {
         self.mapping = vec![INVALID_MAPPING; exprset.alphabet_size()];
 
@@ -93,23 +100,11 @@ impl ByteCompressor {
             visited[e.as_usize()] = true;
             todo.extend_from_slice(exprset.get_args(e));
             match exprset.get(e) {
-                Expr::Byte(b) => {
-                    assert!(
-                        self.mapping[b as usize] == INVALID_MAPPING,
-                        "visiting the same byte the second time"
-                    );
-                    self.mapping[b as usize] = self.alphabet_size as u8;
-                    self.alphabet_size += 1;
-                }
-                Expr::ByteSet(bs) => {
-                    self.bytesets.push(bs.to_vec());
-                }
+                Expr::Byte(b) => self.add_single_byte(b),
+                Expr::ByteSet(bs) => self.bytesets.push(bs.to_vec()),
                 Expr::RemainderIs(_, _) => {
                     for b in exprset.digits {
-                        if self.mapping[b as usize] == INVALID_MAPPING {
-                            self.mapping[b as usize] = self.alphabet_size as u8;
-                            self.alphabet_size += 1;
-                        }
+                        self.add_single_byte(b);
                     }
                 }
                 _ => {}

diff --git a/tests/emptiness.rs b/tests/emptiness.rs
@@ -226,16 +226,13 @@ fn remainder_is_check(should_be_empty: bool, d: u32, other_rx: &str) {
     let mut bld = RegexBuilder::new();
     let id = bld
         .mk(&RegexAst::And(vec![
-            RegexAst::MultipleOf(d),
             RegexAst::Regex(other_rx.to_string()),
+            RegexAst::MultipleOf(d),
         ]))
         .unwrap();
     let mut rx = bld.to_regex(id);
     if rx.always_empty() != should_be_empty {
-        panic!(
-            "empty({} % & {:?}) != {}",
-            d, other_rx, should_be_empty
-        );
+        panic!("empty({} % & {:?}) != {}", d, other_rx, should_be_empty);
     }
 }