Implement custom TinyStr

TinyStr produces terrible code for arbitrary lengths

make-tzdb/src/main.rs

@@ -136,10 +136,9 @@ pub fn main() -> anyhow::Result<()> {
 // OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 use once_cell::race::OnceBox;
-use tinystr::TinyAsciiStr;
 use tz::TimeZone;
 
-use crate::DbTimeZone;
+use crate::{{DbTimeZone, Lower32}};
 "#
     )?;
 
@@ -169,13 +168,12 @@ use crate::DbTimeZone;
         f,
         "pub(crate) fn tz_by_name(s: &str) -> Option<&'static DbTimeZone> {{"
     )?;
+    let max_len = names_and_indices.iter().map(|t| t.1.len()).max().unwrap();
+    assert!(max_len <= 32);
     writeln!(
         f,
-        "    let s: TinyAsciiStr<{}> = s.parse().ok()?;",
-        names_and_indices.iter().map(|t| t.1.len()).max().unwrap(),
+        "    Some(*TIME_ZONES_BY_NAME.get(Lower32([0u64; 4]).for_str(s)?)?)"
     )?;
-    writeln!(f, "    let s = s.to_ascii_lowercase();")?;
-    writeln!(f, "    Some(*TIME_ZONES_BY_NAME.get(&s)?)")?;
     writeln!(f, "}}")?;
     writeln!(f)?;
 

tzdb/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "tzdb"
-version = "0.0.1"
+version = "0.0.3"
 edition = "2021"
 authors = ["René Kijewski <[email protected]>"]
 repository = "https://github.com/Kijewski/tzdb"
@@ -11,7 +11,7 @@ categories = ["date-and-time"]
 readme = "README.md"
 
 [dependencies]
+byte-slice-cast = "1.2.0"
 once_cell = { version = "1.9.0", features = ["parking_lot", "race"] }
 phf = "0.10.1"
-tinystr = "0.5.0"
 tz-rs = "0.5.3"

tzdb/src/lib.rs

@@ -62,6 +62,7 @@ mod generated;
 use std::fmt;
 use std::ops::Deref;
 
+use byte_slice_cast::{AsByteSlice, AsMutByteSlice};
 use once_cell::race::OnceBox;
 use tz::TimeZone;
 
@@ -137,6 +138,28 @@ pub trait TimeZoneExt {
 
 impl TimeZoneExt for TimeZone {}
 
+#[repr(align(32))]
+struct Lower32([u64; 4]);
+
+impl Lower32 {
+    #[inline]
+    fn for_str<'a>(&'a mut self, s: &str) -> Option<&'a str> {
+        self.0
+            .as_mut_byte_slice()
+            .get_mut(..s.len())?
+            .copy_from_slice(s.as_bytes());
+
+        self.0[0] |= 0x2020_2020_2020_2020_u64;
+        self.0[1] |= 0x2020_2020_2020_2020_u64;
+        self.0[2] |= 0x2020_2020_2020_2020_u64;
+        self.0[3] |= 0x2020_2020_2020_2020_u64;
+
+        std::str::from_utf8(self.0.as_byte_slice())
+            .ok()?
+            .get(..s.len())
+    }
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;